`Hydrogen Peroxide in Aqueous Solution
`
`DAYONG LUO,1 SCOTT W. SMITH,2 BRADLEY D. ANDERSON1
`
`1Department of Pharmaceutical Sciences, University of Kentucky, Lexington, Kentucky 40506
`
`2Pfizer Global Research and Development, San Diego, California 92121
`
`Received 10 August 2004; revised 22 September 2004; accepted 22 September 2004
`
`Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20253
`
`ABSTRACT: The oxidation of thiol-containing small molecules, peptides, and proteins in
`the presence of peroxides is of increasing biological and pharmaceutical interest.
`Although such reactions have been widely studied there does not appear to be a consensus
`in the literature as to the reaction products formed under various conditions, the reaction
`stoichiometry, and the reaction mechanisms that may be involved. This study examines
`the reaction kinetics of cysteine (CSH) with hydrogen peroxide (H2O2) in aqueous buffers
`(in the absence of metal ions) over a wide range of pH (pH 4–13) and at varying ratios of
`initial reactant concentrations to explore the range of conditions in which a two-step
`nucleophilic model describes the kinetics. The disappearance of CSH and H2O2 and
`appearance of cystine (CSSC) versus time were monitored by reverse-phase high-
`performance liquid chromatography (HPLC). The effects of oxygen, metal ions (Cu2þ), pH
`(4–13), ionic strength, buffer concentration, and temperature were evaluated. Data
`obtained at [H2O2]0/[CSH]0 ratios from 0.01–2.3 demonstrate that the reaction of CSH
`with H2O2 in the absence of metal ions is quantitatively consistent with a two-step
`nucleophilic reaction mechanism involving rate-determining nucleophilic attack of
`thiolate anion on the unionized H2O2 to generate cysteine sulfenic acid (CSOH) as an
`intermediate. Second-order rate constants for both reaction steps were generated
`through model fitting. At [H2O2]0/[CSH]0 > 10, the % CSSC formed as a product of the
`reaction declines due to the increased importance of alternative competing pathways for
`consumption of CSOH. A thorough understanding of the mechanism in aqueous solution
`will provide valuable background information for current studies aimed at elucidating
`the influence of such factors on thiol oxidation in solid-state formulations.
`ß 2004 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 94:304–316, 2005
`Keywords:
`cysteine; hydrogen peroxide; thiol oxidation; peptide stability; protein
`stability; nucleophilic substitution; disulfide formation; sulfenic acid; sulfinic acid;
`sulfonic acid
`
`INTRODUCTION
`
`The oxidation of peptide and protein pharma-
`ceutical products has become an increasingly
`important problem in drug development as more
`biotechnology derived products progress toward
`clinical studies and commercialization.1–3 Pro-
`
`Correspondence to: Bradley D. Anderson (Telephone: 859-
`257-2300, ext. 235; Fax: 859-257-2489;
`E-mail: bande2@email.uky.edu)
`
`Journal of Pharmaceutical Sciences, Vol. 94, 304–316 (2005)
`ß 2004 Wiley-Liss, Inc. and the American Pharmacists Association
`
`teins containing one or more free cysteine resi-
`dues (e.g., human serum albumin,4 recombinant
`human a1-antitrypsin,5 the superfamily of pro-
`tein tyrosine phosphatases,6,7 and a variety of
`others8), as well as thiol-containing peptides (e.g.,
`glutathione9,10 the angiotensin-converting enzyme
`inhibitor captopril11 and the prostate-specific
`antigen peptides12) are particularly susceptible
`to thiol-oxidation through a variety of mechan-
`isms. In vitro, these mechanisms may involve
`free-radical scavenging of various reactive oxy-
`gen species such as superoxide9,10 or hydroxyl
`
`304
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`Eton Ex. 1056
`1 of 13
`
`
`
`REACTION OF CYSTEINE AND HYDROGEN PEROXIDE IN AQUEOUS SOLUTION
`
`305
`
`radical,13 resulting in thiol loss and the consump-
`tion of oxygen. Alternatively, direct reaction of
`thiols via nucleophilic substitution with certain
`reactive oxygen species such as hydrogen per-
`oxide may occur.4,5,7,14,15
`Because of their reactivity as scavengers of
`reactive oxygen species, thiol-containing com-
`pounds can be highly effective as antioxidants for
`stabilizing pharmaceuticals.13,16–20 However, this
`propensity toward oxidation makes formulation of
`proteins and peptides containing free cysteine
`residues more difficult. In pharmaceutical formu-
`lations, thiol autoxidation is generally thought to
`involve a series of complex reaction processes
`catalyzed by traces of transitional metal
`ions
`(e.g., CuII and FeIII).11,21–23 For this reason, metal
`ion chelators such as EDTA have proven to be
`effective stabilizers of thiol containing peptides
`and proteins.24–26 Considering, however, that
`peroxides are also found as impurities in some
`common formulation excipients (e.g., polyvinyl-
`pyrrolidone, polyethylene glycol, polysorbate 80,
`etc.),27–34 and that H2O2 is produced as a bypro-
`duct of metal-catalyzed oxidations,35–37 an under-
`standing of the mechanism of the thiol–H2O2
`reaction is a prerequisite for understanding the
`general case of thiol oxidation in formulations.
`Our interest in stabilizing thiol-containing
`peptides stems from recent attempts to prepare
`lyophilized prototype formulations of several pros-
`tate-specific antigen (PSA) peptides under con-
`sideration by the National Cancer Institute for
`vaccine therapy for human prostate cancer,12,38
`including the cys ras peptide (KLVVVGAC-
`GVGKS),25 PSA-3 (VISNDVCAQV),23,25 and
`PSA-OP (FLTPKKLQCVDLHVISNDVCAQVH-
`PQKVTK).26 Although EDTA was found to be
`highly effective in preventing metal-catalyzed
`disulfide formation in these lyophiles, the poten-
`tial for the reaction of cysteine residues with trace
`peroxide impurities in excipients remains even in
`the presence of EDTA.
`Several studies of the reaction between various
`low molecular weight thiols (e.g., cysteine, N-
`acetylcysteine, glutathione, and others) and hy-
`drogen peroxide (H2O2) or other peroxides (e.g.,
`benzoyl peroxide, lipid hydroperoxides, etc.) have
`been reported, but there does not appear to be a
`consensus as to the reaction products formed
`under various conditions,10,14,15,39 –42 the mechan-
`ism(s) by which these products form,10,14,15,41 –44 or
`the rate constants for various reaction steps.
`This study examines the reaction kinetics of
`cysteine with H2O2 in aqueous buffers (in the
`
`absence of metal ions) over a wide range of pH
`(pH 4–13) and at varying ratios of initial reactant
`concentrations to explore the range of conditions
`over which a two-step nucleophilic model as
`depicted in Scheme I accounts for the reaction
`kinetics and stoichiometry. In aqueous solution
`the first step in Scheme I is rate-determining, but
`it may be possible to estimate rate constants for
`each process, depending on the extent to which the
`two reaction steps differ.
`
`Scheme I.
`
`A thorough understanding of the mechanism of
`reaction between low molecular weight thiols and
`H2O2 in aqueous solution should be useful in
`rationalizing the fate of thiol-containing proteins
`undergoing reaction with hydrogen peroxide,
`where the buildup of the sulfenic acid intermediate
`(RSOH) suggests that the second reaction step
`may become rate-limiting in some circumstances.
`This may reflect the increased importance of
`reactant mobility in the second reaction step.
`Similarly, the reaction of the RSOH intermediate
`with a thiol residue in a second molecule of a thiol-
`containing peptide or protein may also become rate
`limiting in amorphous (e.g., lyophilized) pharma-
`ceutical formulations that contain peroxide impu-
`rities. Evidence that the disulfide may not be the
`sole reaction product under certain conditions led
`to an exploration of the possible factors influencing
`the fate of the sulfenic acid intermediate (RSOH)
`and the reaction products ultimately formed when
`RSOH reacts with H2O2 or another molecule of
`RSOH.
`
`EXPERIMENTAL
`
`Reagents
`
`Cysteine (>98% purity by TLC), cystine (Sigma-
`Ultra, >99% purity by TLC), cysteine sulfinic acid
`(water content 1 mol/mol), cysteic acid (mono-
`hydrate), and ethylenediaminetetraacetic acid
`(EDTA) were purchased from Sigma Chemical
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`Eton Ex. 1056
`2 of 13
`
`
`
`306
`
`LUO, SMITH, AND ANDERSON
`
`Co. (St. Louis, MO). H2O2 (30% aqueous solution)
`and cupric sulfate (CuSO4 5H2O) were obtained
`from Mallinckrodt Baker Inc. (Phillipsburg, NJ).
`All reagents were analytical grade and used as
`supplied. Deionized ultra-filtered (DIUF1) water
`used to prepare CSH and H2O2 solutions and
`HPLC grade acetonitrile were purchased from
`Fisher Scientific Co. (Pittsburgh, PA). Water used
`in mobile phase was deionized and further
`purified through a Milli-Q1 UV Plus Ultrapure
`Water System, Millipore Ltd. (Billerica, MA).
`
`High-Performance Liquid Chromatography (HPLC)
`
`The analyses of CSH, CSSC, and H2O2 were
`carried out by reverse-phase high-performance
`liquid chromatography (HPLC). The system con-
`sisted of a Beckman 110A Solvent Delivery
`Module (Beckman Coulter, Fullerton, CA), a
`WatersTM 717 plus Autosampler (Waters Corp.,
`Milford, MA), and a Hewlett-Packard 1040M
`Series II HPLC detector (Hewlett-Packard Com-
`pany, Palo Alto, CA) operating at 214 nm, which
`was connected to a PC for data acquisition and
`analysis using HP LC/MSD ChemStation soft-
`ware. A Supelcosil ABZþPlus, 5-mm (250
`4.6 mm)
`(Supelco, Bellefonte, PA) analytical
`column and 5 mm (2 cm 2.1 mm) guard column
`were employed. Elutions were performed isocra-
`tically using a mobile phase consisting of 50%
`(v/v) of an aqueous solution of 50 mM phosphoric
`acid (solution A) and 50% (v/v) of a solution
`containing 5 mM sodium 1-nonanesulphonate
`(99%, Lancaster Synthesis, Inc., Windham, NH),
`50 mM phosphoric acid, and 5% (v/v) acetonitrile
`in water (solution B) at a flow rate 1.5 mL/min.
`Solution B was also used as the quench solution.
`The pH of the mobile phase was adjusted to 2.5
`with NaOH. The retention times for H2O2, CSH,
`and CSSC were 2 min, 4.8 min, and 6.5 min,
`respectively. Cysteine sulfinic acid (CSO2H) and
`cysteic acid (CSO3H) eluted prior to H2O2 at
`1.8 min, but could not be adequately resolved
`for quantitation purposes under these conditions.
`Separations of CSO2H from CSO3H required two
`Supelcosil ABZþPlus columns connected in series
`and a flow rate of 0.5 mL/min. Under these condi-
`tions, retention times for CSO2H and CSO3H were
`10.4 and 11 min, respectively.
`
`General Procedure for Kinetic Studies
`
`Fresh stock solutions of CSH were prepared by
`weighing 20 mg CSH into a 1.5-mL plastic
`microcentrifuge tube (VWR Scientific Products
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`Co., Buffalo Grove, IL) and dissolving in 1 mL
`phosphate buffer prepared from phosphoric acid
`solution (50 mM) adjusted to the desired pH with
`sodium hydroxide. Stock solutions of H2O2 were
`prepared by combining 0.1 mL 30% aqueous H2O2
`solution and 0.9 mL of the same phosphate buffer
`(50 mM) in a 1.5-mL plastic microcentrifuge tube.
`Reactions were initiated after first diluting the
`above stock solutions with the same buffer to a
`concentration twice that in the final reaction,
`equilibrating to the desired reaction temperature
`in a water bath set at 258C. Equal volumes of each
`diluted solution were then combined by adding the
`diluted H2O2 solution to the CSH solution and
`rapidly mixing in a 15 mL plastic centrifuge tube
`(Falcon1, Becton Dickinson Co., Franklin Lakes,
`NJ). Reaction solutions were stored at 258C and
`aliquots (0.1 mL) were taken at predetermined
`time intervals and transferred into 1-mL HPLC
`vials. An aliquot of 0.1-mL solution B was added to
`quench the reaction by lowering the pH and the
`total volume was adjusted to 1 mL with the mobile
`phase. Samples were analyzed immediately by
`HPLC.
`Kinetic studies to obtain a preliminary assess-
`ment of reaction order, metal ion effects, and the
`influence of oxygen utilized the initial rate method
`whereby the formation of cystine was monitored at
`several time points in the very early stages of the
`reaction (first minute) at varying initial concen-
`trations of CSH and H2O2. More comprehensive
`studies for use in fitting mathematical models to
`the data (see Data Analyses), generating pH-rate
`profiles, and examining the effects of buffer
`concentration and ionic strength involved the
`determination of complete reactant (i.e., CSH
`and H2O2) and product concentration versus time
`profiles.
`
`Metal Ion and Oxygen Effects
`
`Solutions containing equal (5 mM) concentrations
`of CSH and H2O2 and varying in EDTA concen-
`tration (0, 25, and 50 mM) were prepared by
`combining stock solutions in pH 6.0 (50 mM)
`sodium phosphate buffer either with or without
`addition of 50 mM cupric sulfate (CuSO4). Buffers
`employed in the absence of added cupric ion were
`prepared using either Milli-Q treated deionized
`water or DIUF1 water (see Reagents).
`The influence of oxygen was explored using
`initial rate studies both in air-equilibrated sam-
`ples and samples prepared and stored under a
`nitrogen atmosphere (in a glove box). Sample
`
`Eton Ex. 1056
`3 of 13
`
`
`
`REACTION OF CYSTEINE AND HYDROGEN PEROXIDE IN AQUEOUS SOLUTION
`
`307
`
`equation to obtain an estimate for the energy of
`activation.
`
`Effect of [CSH]0/[H2O2]0 Ratio on Kinetics and
`Reaction Products
`
`To further explore the range of applicability of
`the simple nucleophilic mechanism depicted in
`Scheme I, additional experiments were carried
`out in which the ratio of starting concentrations of
`CSH and H2O2 ([CSH]0/[H2O2]0) was varied from
`100 to 0.001. A ratio of [CSH]0:[H2O2]0¼ 100
`using 50 mM CSH and 500 mM H2O2 at pH 6.0
`was employed to assess the effect of a large excess
`of CSH on the rate of disappearance of H2O2. The
`effects of [CSH]0:[H2O2]0 ratios 1.0 (i.e., 1 to
`0.001) were evaluated at pH 5.0, 6.0, and 7.0 to
`explore the possible generation of new reaction
`products in the presence of excess H2O2. The
`concentration of [CSH]0 in these experiments was
`3 mM at [CSH]0:[H2O2]0 ratios of 1 and 0.5 and
`1 mM at ratios between 0.2 and 0.001. Solution
`aliquots were taken after the reactions were
`complete (72 h at pH 7.0, 96 h at pH 6.0, and
`120 h at pH 5.0) and analyzed by HPLC for the
`CSSC product formed during the reaction. Ana-
`lyses of some of the samples at later time points
`verified that the CSSC concentrations did not
`change after completion of the reaction.
`
`DATA ANALYSIS
`
`temperatures were controlled by placing samples
`in a water jacketed container at 258C. An OM-4
`oxygen meter (Microelectrodes, Inc., Bedford, NH)
`was used to verify the removal of oxygen from
`diluted stock solutions that were bubbled with
`nitrogen prior to being combined to start a given
`reaction. The oxygen meter was calibrated by
`adjusting the reading to 0% in buffer solutions
`treated by bubbling with a N2 stream for more than
`30 min (only 20 min was necessary to achieve a
`constant reading) and to 100% in oxygen saturated
`buffer solutions. Experiments were at pH 6.0 and
`at a fixed starting concentration of CSH (25 mM)
`while H2O2 starting concentrations were varied
`(i.e., 5, 10, 25, and 50 mM) or a fixed starting
`concentration of H2O2 (10 mM) while CSH
`starting concentrations were varied (i.e., 2, 5, 10,
`and 25 mM).
`
`Buffer, Ionic Strength, pH, and
`Temperature Effects
`
`Phosphate buffers (pH 7) were prepared at dif-
`ferent buffer concentrations (10, 30, and 50 mM)
`and at a constant ionic strength of 0.5 M (adjusted
`with sodium chloride (NaCl)). The kinetics of
`reaction in solutions containing 4 mM CSH and
`4 mM H2O2 were monitored as a function of
`buffer concentration and second-order reaction
`rate constants generated through model fitting
`were compared.
`The effect of ionic strength was explored by
`comparing the kinetics of reaction between 4 mM
`CSH and 4 mM H2O2 in 50 mM sodium phosphate
`buffers at pH 5.0, 7.0, 10.0, and 13.0 either in the
`absence of added NaCl or in buffers adjusted to an
`ionic strength of 0.5 M with NaCl. Second-order
`rate constants generated through model fitting of
`the experimental data were then compared.
`The influence of pH on the reaction between
`4 mM CSH and 4 mM H2O2 was monitored in
`buffers ranging from pH 4–13 at 258C. Buffers
`were prepared with 50 mM phosphoric acid
`solution and sodium hydroxide (NaOH). Ionic
`strength was not adjusted. The observed reaction
`rate constants (kobs) obtained through the model
`fitting of the experimental data were employed to
`construct the pH-rate profile.
`Reactions between CSH and H2O2 were con-
`ducted at different temperatures (0, 25, and 508C)
`in pH 6.0 buffer solution using an ice bath for 08C, a
`water bath at 258C and an oven at 508C. Rate
`constants (k1) generated from model fitting (see
`Data Analysis) were then fit to the Arrhenius
`
`Concentration versus time curves for the disap-
`pearance of CSH and H2O2 and appearance of
`CSSC were fit to a mathematical model derived
`from Scheme I using nonlinear least-squares
`regression analysis (SCIENTIST1, Micromath
`Inc., Salt Lake City, UT) to obtain estimates for
`the two second-order rate constants, k1 and k2.
`The following differential equations can be gen-
`erated from Scheme I:
` d½CS
`dt
`
`¼ k1 CS
`½
`
`
`
` H2O2½
`
` þ k2 CS
`½
`
`
`
` CSOH½
`
`
`
`ð1Þ
`
`
`
`½
`d CSOH
`dt
`
` d HOOH½
`
`
`dt
`
`
`
`¼ k1 CS
`½
`
`
`
` H2O2½
`
`
`
`ð2Þ
`
`¼ k1 CS
`½
`
`
`
` H2O2½
`
` k2 CS
`½
`
`
`
` CSOH½
`
` ð3Þ
`
`½
`
`d CSSC
`dt
`
`¼ k2 CS
`½
`
`
`
` CSOH½
`
`
`
`ð4Þ
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`Eton Ex. 1056
`4 of 13
`
`
`
`308
`
`LUO, SMITH, AND ANDERSON
`
`where [CS ], [H2O2], [CSOH], and [CSSC] are the
`concentrations of cysteine thiolate anion, hydro-
`gen peroxide, the cysteine sulfenic acid inter-
`mediate, and cystine, respectively, at time t.
`
`RESULTS AND DISCUSSION
`
`Simultaneous analysis of CSH, H2O2,
`and CSSC by HPLC
`
`The reaction between CSH and H2O2 has been
`previously studied by several other groups.10,14,43,45
`Normally in these studies, the concentrations of
`thiol and H2O2 were measured by colorimetric
`methods and CSSC concentration was not ana-
`lyzed. Although HPLC methods to separate CSH
`and CSSC46 and to separately determine H2O2
`concentration in various media34,47 have been
`reported, we are not aware of attempts to analyze
`all three reaction components in a single HPLC
`method. Figure 1 displays typical HPLC chroma-
`tograms obtained during the reaction of CSH with
`H2O2 leading to the formation of CSSC. Response
`factors were verified to be linear for all three
`compounds over concentration ranges of 0.025–
`2.5 mM for CSH, 0.1–10 mM for H2O2, and 3–
`400 mM for CSSC (solubility limited the range for
`CSSC) with intraday precision of <1% for all
`analytes. Interday precision in response factors
`was 1.6, 1.3, and 6.6% for CSH, H2O2, and CSSC,
`respectively. Detection limits at 214 nm estimated
`from three times the standard deviation for the
`lowest concentrations analyzed were 1.4 mM
`(0.14 nmol), 6.5 mM (.65 nmol), and 0.3 mM
`(0.03 nmol), respectively, for CSH, H2O2, and
`CSSC. These detection limits indicate that HPLC
`detection at 214 nm is not as sensitive for H2O2 as
`methods using electrochemical detection34 nor as
`sensitive for CSH and CSSC as HPLC with cou-
`lometric detection.46 However, Vignaud et al.46
`have observed that UV detection is more stable
`and easier to handle.
`
`Elimination of Metal Ions and Oxygen
`as Variables in the Kinetic Studies
`
`Because autoxidation of CSH (i.e., the direct
`reaction between CSH and molecular oxygen) is
`a spin-forbidden process, this reaction was not
`expected to contribute to the kinetics in this
`study.48 However, the presence of trace of transi-
`tion metal ions will significantly accelerate thiol
`autoxidation. Transition metal ions such as Cu2þ
`and Fe3þ not only catalyze autoxidation, but also
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`Figure 1. HPLC chromatograms (214 nm) at various
`times during the reaction of CSH with H2O2. Conditions:
`[CSH]0¼ 4 mM, [H2O2]0¼ 2 mM in pH 6.0 (50 mM)
`phosphate buffer at 258C. Times: (A) 15 s; (B) 30 min; (C)
`90 min; (D) 900 min. [Color figure can be seen in the
`online version of this article, available on the website,
`www.interscience.wiley.com.]
`
`act as direct oxidants in thiol oxidation.22,36,37,49 –52
`Given the above, it is clear that transition metal
`ions had to be avoided to minimize the contribu-
`tion of alternate routes of thiol oxidation.
`The ability of the metal ion chelator EDTA to
`eliminate the effect of metal ions on the reaction
`between CSH and H2O2 was investigated in pH 6.0
`solutions containing equal (5 mM) concentrations
`of CSH and H2O2 and varying in EDTA concentra-
`tion (0, 25, and 50 mM) either with or without the
`addition of 50 mM cupric sulfate (CuSO4). Initial
`rates of formation of CSSC, shown in Figure 2,
`
`Eton Ex. 1056
`5 of 13
`
`
`
`REACTION OF CYSTEINE AND HYDROGEN PEROXIDE IN AQUEOUS SOLUTION
`
`309
`
`quantitative modeling of the reaction, a prelimin-
`ary study designed to ascertain the reaction order
`with respect to CSH and H2O2 was conducted.
`Initial rates of formation of CSSC during the first
`minute after mixing at 258C were obtained for
`solutions containing a fixed starting concentra-
`tion of CSH (25 mM) and varying H2O2 concen-
`tration (i.e., 5, 10, 25, and 50 mM) or fixed H2O2
`concentration (10 mM) and varying CSH concen-
`tration (i.e., 2, 5, 10, and 25 mM). Plots of CSSC
`concentration versus time (data not shown) were
`linear for every combination, but in some cases,
`particularly at high reactant concentrations, the
`extent of the reaction as determined by either
`CSSC formation or loss of CSH exceeded 20%.
`Therefore, the slopes of the CSSC concentration
`versus time plots (d[CSSC]/dt) were normalized
`by dividing by the average concentration of the
`fixed reactant to adjust for the fact that mean
`reactant concentrations were less than 100% even
`during the first minute of the reaction. The slopes
`obtained appeared to be proportional to the total
`concentrations of each reactant in all ionization
`states, [CSH]t and [H2O2]t, indicating that the
`following rate law may be applicable:
`t
`=dt ¼ kobs CSH½
`
`t H2O2½
`½
`d CSSC
`Slopes of log–log plots of the normalized rates of
`CSSC formation versus the variable reactant con-
`centration were 0.96 (0.01) versus 0.94 (0.03) for
`varying H2O2 concentrations under nitrogen ver-
`sus air, respectively, and 0.92 (0.01) versus 0.92
`(0.02), for varying CSH concentrations under
`nitrogen versus air, respectively. None of these
`slopes were significantly different from 1.0 as
`determined by the overlap of their 95% confidence
`intervals with 1.0, indicating that the reaction
`appears to be first-order in terms of each reactant.
`A more comprehensive set of experiments was
`then performed in which the concentrations of CSH,
`H2O2, and CSSC were monitored versus time and fit
`to the model represented by eqs 1–4 and depicted in
`Scheme I. Initially, a set of five experiments at a fixed
`pH (6.0) and CSH concentration (4 mM) and varying
`concentrations of H2O2 (2, 4, 6, 8, and 9.2 mM) were
`conducted at 258C so that pH effects could be
`neglected. Panels representing three of these experi-
`ments are shown in Figure 3.
`Evident from all three plots, exactly 2 mol of
`CSH are consumed per mol of H2O2, consistent
`with Scheme I and the overall stoichiometry for the
`reaction shown below:
`2CSH þ H2O2 ! CSSC þ H2O
`
`ð5Þ
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`Figure 2.
`Initial rates of CSSC formation in solutions
`containing equal (5 mM) concentrations of CSH and
`H2O2 versus EDTA concentration (0, 25, and 50 mM) in
`pH 6.0 (50 mM) sodium phosphate buffer either with (&)
`or without (~, }) addition of 50 mM cupric sulfate
`(CuSO4). Buffers employed in the absence of added cu-
`pric ion were prepared using either Milli-Q treated
`deionized water (~) or commercially obtained deionized
`ultrafiltered (DIUF1) water (}).
`
`indicate that the addition of cupric ion (Cu2þ)
`dramatically increases the reaction rate in the
`absence of EDTA, but that addition of EDTA
`suppresses this effect. In the presence of 25 mM
`EDTA, the rate of CSSC formation with the
`addition of Cu2þ is identical to the rate in deionized
`water containing no added cupric ion. EDTA addi-
`tion had no apparent effect itself on the reaction
`rate in the absence of added metal ion but did
`interfere with the HPLC analysis of hydrogen
`peroxide. Because the reaction rate was the same
`with or without EDTA when a high purity com-
`mercially available deionized ultrafiltered water
`(DIUF1, see Reagents) was employed, subsequent
`kinetic studies in which H2O2 was monitored were
`conducted without EDTA addition.
`Initial rates of CSSC formation were monitored
`in solutions at pH 6.0 at a fixed starting concen-
`tration of CSH (25 mM), while H2O2 starting
`concentrations were varied or at a fixed start-
`ing concentration of H2O2 (10 mM) while CSH
`starting concentrations were varied (see rate law
`studies). Replicate experiments were performed in
`air-equilibrated samples and in samples prepared
`and stored under a nitrogen atmosphere. No signi-
`ficant differences were observed in any of the re-
`actant solutions, indicating that air oxygen is not
`involved in the reaction between CSH and H2O2.
`
`Rate Law, Stoichiometry, and Reaction Mechanism
`
`Prior to generating complete reactant and product
`concentration versus time profiles for use in
`
`Eton Ex. 1056
`6 of 13
`
`
`
`310
`
`LUO, SMITH, AND ANDERSON
`
`Figure 3. Concentration versus time plots of cysteine
`(&) cystine (~), and hydrogen peroxide (*) during the
`reaction of CSH ([CSH]0¼ 4 mM) with varying concen-
`trations of H2O2 in 50 mM pH 6.0 phosphate buffer at
`258C. (A) [H2O2]0¼ 2 mM; (B) [H2O2]0¼ 4 mM; (C).
`[H2O2]0¼ 9.2 mM. The curves are nonlinear least-
`squares best fits using the kinetic model depicted in
`Scheme I (eqs. 1–4).
`
`The solid curves in Figure 3A–C represent
`simultaneous nonlinear least-squares fits of the
`data to the model depicted in Scheme I. Assuming
`a pKa for the cysteine sulfhydryl ionization of 8.44
`(see later discussion), and correcting the pKa for
`ionic strength, values for the two second-order
`reaction rate constants at pH 6.0 and 258C
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 2, FEBRUARY 2005
`
`generated through model fitting are k1¼ 15.2
`0.1 M 1s 1, k2¼ 720 70 M 1s 1. The excellent fit
`of the data strongly supports the nucleophilic
`mechanism that Scheme I describes.
`The apparent first-order dependence on CSH
`and H2O2 concentrations determined in the initial
`rate studies can only be consistent with Scheme I if
`the first step in the reaction, the nucleophilic
`attack of thiol (or thiolate anion) on H2O2 to
`generate a highly reactive sulfenic acid intermedi-
`ate is the rate-determining step. The relative
`values obtained for k1 and k2 (k1 k2) are consis-
`tent with the first step being rate-determining. To
`our knowledge, this is the first attempt to obtain a
`quantitative estimate for the bimolecular rate con-
`stant for thiolate anion attack on cysteine sulfenic
`acid to generate cystine.
`Among those studies in the literature that have
`concluded that hydrogen peroxide reacts with
`thiols to produce exclusively the corresponding
`disulfides there is disagreement as to the reaction
`stoichiometry and mechanism. Although Darkwa
`et al.43 suggested that the oxidation of cysteine by
`H2O2 is predominantly free radical-mediated,
`most groups have generated results consistent
`with a two-step nucleophilic substitution mechan-
`ism as described in Scheme I.10,14,15,44 As demon-
`strated in Figure 3A–C,
`the overall molar
`stoichiometry (H2O2:RSH) of the nucleophilic
`mechanism is 1:2. However, Abedinzadeh et al.
`have argued that this stoichiometry depends on
`the reactant ratios, such that, at initial reactant
`concentration ratios ([RSH0]/[H2O2]) > 2.5 the
`expected 1:2 stoichiometry is observed while at
`ratios <2.5 (i.e., excess peroxide) the stoichiometry
`is 1:1.41,42 The study described in Figure 3C
`contained excess peroxide yet the overall stoichio-
`metry remained 1:2.
`Abedinzadeh et al.41,42,53 also reported that the
`concentration of H2O2 undergoes abrupt decreases
`in the initial stage of reactions between glu-
`tathione or N-acetylcysteine, which they attribu-
`ted to the formation of complexes between RSH
`and H2O2. It is evident in Figure 3A–C that there
`is no abrupt decrease in hydrogen peroxide con-
`centration in the early stage of this reaction.
`Moreover, the excellent fit of the data to the model
`represented in Scheme I, which does not consider
`complex formation, suggests that a more compli-
`cated mechanism is unnecessary. However, apart
`from the fact that different thiols were the subject
`of their investigations, Abedinzadeh et al. mon-
`itored their reactions using absorbance measure-
`ments, whereas these studies employed HPLC.
`
`Eton Ex. 1056
`7 of 13
`
`
`
`REACTION OF CYSTEINE AND HYDROGEN PEROXIDE IN AQUEOUS SOLUTION
`
`311
`
`Further studies are necessary to rationalize these
`apparently disparate results.
`
`Buffer, Ionic Strength, pH, and
`Temperature Effects
`
`Parallel experiments were conducted in 50 mM
`phosphate buffers at different pH values (pH 5, 7,
`10, and 13) with or without control of
`ionic
`strength. In the absence of ionic strength adjust-
`ment, 50 mM phosphate buffers have an ionic
`strength ranging from 0.05 at pH 4 to 0.43 and
`pH 13. In the ionic strength-controlled buffer
`solutions, ionic strength was adjusted to 0.5 M
`with addition of NaCl. The experimental results
`in both groups of reactions were similar after
`correcting for ionic strength effects on pKa values
`(i.e., no statistically significant difference was
`observed). The absence of an ionic strength effect
`on the reaction between CSH and H2O2 is
`consistent with a previous study of the reaction
`of 2-mercaptoethanol with H2O2 as a function of
`ionic strength over the range of 0.05 M <I <0.4 M
`reported by Leung et al.15
`The kinetics of reaction in solutions containing
`4 mM CSH and 4 mM H2O2 were monitored as a
`function of buffer concentration in phosphate
`buffers (pH 7) varying in concentration (10, 30,
`and 50 mM) and at a constant ionic strength of
`0.5 M [adjusted with sodium chloride (NaCl)].
`Fitted values of k1 were 14.0, 17.4, and 20.5 M 1s 1
`in 10, 30, and 50 mM buffer, respectively. Although
`the 95% confidence intervals at 10 and 50 mM did
`not overlap, indicating that the differences were
`significant, the dominant contribution to k1 is the
`nonbuffer-catalyzed reaction.
`The influence of pH on the reaction between
`4 mM CSH and 4 mM H2O2 was monitored at 258C
`in buffers ranging in pH from 4–13. Concentration
`versus time curves such as those shown in Figure 3
`were generated at each pH and fit simultaneously
`to obtain estimates for the thermodynamic pKa
`values of cysteine and hydrogen peroxide, and
`values for k1 and k2 reflecting the entire pH range.
`The pKa value found for cysteine, 8.44 0.02, is in
`reasonable agreement with literature estimates of
`8.3354 and 8.53,55 while the pKa obtained for H2O2,
`11.51 0.02, also agrees favorably reported values
`of 11.4515 and 11.6.56 The values found for k1
`(¼14.7 0.35 M 1s 1) and k2 (¼570 170 M 1s 1)
`are not significantly different from those obtained
`at pH 6.0 in which five groups of kinetics experi-
`ments varying in CSH and H2O2 concentration
`were analyzed. Shown in Figure 4 is the pH
`
`Figure 4. pH dependence of the rate constant for the
`nucleophilic reaction between cysteine and hydrogen
`peroxide, k1, at 258C.
`
`rate profile generated from the pH 4–13 data,
`where kobs¼ k1*fCS *fHOOH and fCS and fHOOH
`are the fractions of thiolate anion and the neutral
`species of hydrogen peroxide, determined from
`their pKa values. The solid line (which was not
`fitted to the points) in Figure 4 represents the k1
`(¼14.9 M 1s 1) obtained from averaging the fits of
`the pH 4–13 and pH 6 concentration versus time
`data and the pKa values found from those previous
`fits (i.e., 8.44 for cysteine and 11.51 for H2O2).
`Our estimate of k1 (¼14.9 M 1s 1) at 258C
`appears to be in reasonable accord with literature
`estimates at other temperatures. Barton et al.14
`obtained a value of 12.4 M 1s 1 at room tempera-
`ture, while Winterbourn and Metodiewa10 esti-
`mated k1 to be 26 M 1s 1 at 378C. Radi et al.57
`estimated a k1 of 17.1 M 1s 1 at 378C.
`Values of k1 generated in this study through
`model fitting at different temperatures from 0–
`508C are well described by the Arrhenius equation
`(Fig. 5) with an activation energy (Ea) of 70.96 kJ/
`mol (16.96 kcal/mol). Alternatively, use of the
`Eyring equation yielded an enthalpy of activation,
`DH{, of 16.4 0.3 kcal/mol and entropy of acti-
`vation, DS{, of 1.7 1.1 cal K 1mol 1. These
`activation parameters result in an estimated k1
`of 42.6 M 1s 1 at 378C.
`The pH profile in Figure 4 covering a range of pH
`from 4–13 and the kinetic model used to fit these
`data are consistent with the results of Leung
`et al.,15 who examined the r