NMR Characterization of H₂O₂ Hydrogen Exchange

Abstract

Quantification of H₂O₂ concentration in aqueous solutions is of interest in many fields. It usually is based on indirect methods that rely on oxidation reactions that turn on/off fluorescent probes. Such methods can suffer from reaction incompleteness and interfering chemical species. We describe optimization of NMR detection that enables direct quantification of H₂O₂ down to the nanomolar range. Taking advantage of fast hydrogen exchange (HX) between H₂O₂ and water permits the use of very short interscan delays, greatly increasing sensitivity. The specific acid-, base- and water-catalyzed HX rates at 2 °C were measured to be 2.1 × 10⁷, 6.1 × 10⁹, and 1.4 × 10⁻¹ M⁻¹s⁻¹, respectively, which result in a minimum HX rate at pH 6.2. Furthermore, the exchange is accelerated by general acid/base catalysis. MES and phosphate buffers catalyze HX strongest in their unprotonated forms. For imidazole, only the unprotonated form catalyzes HX, which contrasts with acetic acid where only the protonated state catalyzes exchange. Inorganic salts such as sodium chloride and azide have negligible effect on HX. We present optimal conditions for accurate measurement of H₂O₂ concentrations as low as 40 nM in aqueous samples in a few hours.

Graphical Abstract

Introduction

Hydrogen peroxide (H₂O₂) is found in biological systems [1, 2] as well as ambient air, rain and the upper atmosphere [3]. H₂O₂ in respiratory fluids has been reported [4] and may be implicated in the chemical degradation of airborne virus, which is known to occur faster at a relative humidity where the respiratory droplet does not fully dehydrate [5]. Thus, reliable and accurate quantification of H₂O₂ is important in many areas, ranging from atmospheric science to respiratory viral disease. Recent work has shown that H₂O₂ forms spontaneously when water condenses to form microdroplets [6]. In related work it was shown that H₂O₂ generation associated with ultrasonic cavitation of water, used in commercial humidifiers, varies strongly with atmospheric humidity, a finding with potentially important implications for the duration airborne respiratory virus remains viable [7].

H₂O₂ is commonly quantified via indirect methods that rely on its chemical reaction with fluorogenic probes that increase in quantum yield upon oxidation [8]. Consequently, these methods are susceptible to chemical interference in complex biological samples, as well as reaction incompleteness resulting from the slow kinetics of catalyst-free H₂O₂ oxidations. For example, the conversion of aryl boronate compounds to fluorescent phenolic derivatives upon oxidation by H₂O₂ is the most utilized reaction among the indirect methods. However, this reaction is sluggish (typical rate of 1–2 M⁻¹s⁻¹) [9, 10] and considerably slower than that of other interfering reactive species such as ONOO⁻ (10⁶ M⁻¹s⁻¹) and HOCl (10⁴ M⁻¹s⁻¹), thus posing a problem in quantitative analysis [11].

Here, we demonstrate that by optimization of experimental parameters, direct NMR detection of H₂O₂ in water is readily possible down to the nanomolar range, thereby providing direct quantitative information on its concentration. In water, the NMR spectrum of H₂O₂ features an exchange-broadened line with a unique chemical shift of around 11 ppm, that previously has been used for H₂O₂ quantification down to a concentration of 1 mM [12]. Subsequently, Tsiafoulis and Gerothanassis showed that 6-fold dilution of aqueous H₂O₂ solution slows down hydrogen exchange (HX) and allows cooling the sample to −13 °C, thereby sharpening the resonance and achieving a detection limit as low as 20 μM (corresponding to 120 μM in the original aqueous solution) [13]. This method has been used for quantification of H₂O₂ in plant extracts [14] and cosmetic products [15]. More recently, Ryoo et al. exploited the chemical exchange saturation transfer (CEST) technique in combination with the ultrafast Z-spectroscopic method for highly sensitive detection of H₂O₂ in aqueous solutions, and demonstrated detection in the low millimolar range in as little as 2 seconds [16]. This method takes advantage of the rapid HX between H₂O₂ and H₂O protons and measures the effect of saturating the weak H₂O₂ resonance on the intense H₂O signal, but quantitative analysis requires accurate knowledge of the H₂O₂ HX rate. In related work, Buljubasich et al. measured the effect of H₂O₂ on the water transverse relaxation rate as a function of pH and concentration [17]. In an elegant application, they then used the extracted parameters to follow the kinetics of heterogeneously catalyzed H₂O₂ decomposition [18].

In contrast to previous efforts that relied on organic co-solvents to slow the HX rate and thereby sharpen the H₂O₂ resonance, our approach to measuring H₂O₂ takes advantage of the rapid exchange of H₂O₂ protons with water, similar to CEST [19]. Combined with selective excitation of the downfield region of the spectrum, which leaves the water magnetization unperturbed, the H₂O₂ z magnetization then returns to its Boltzmann equilibrium value at the rate of HX, which is much faster than its longitudinal relaxation rate. This rapid recovery of z magnetization permits the use of short interscan delays, only limited by the time needed to collect the free induction decay.

Here, we evaluate some of the most important factors that impact HX rates, including pH, temperature, and buffer concentration. Anbar et al. have previously studied the pH dependence of the H₂O₂ HX rate [20]. However, their measurements applied to the molar concentration regime and their pH of minimum HX rate differs substantially from our results at dilute concentrations. We also investigate the effect of common pH buffers, some of which share functional groups with amino acid sidechains and are shown to catalyze HX. Furthermore, we evaluate the effect of NaCl, which is usually present in biological fluids at high concentration.

Experimental Section

NMR sample preparation.

In a typical sample, MES stock solution (10 μL, 50 mM in D₂O) was added to the analyte (490 μL) followed by adjusting the pH to within ± 0.05 pH units of the desired value using a glass electrode. Regular 5 mm NMR tubes and a 600 MHz Avance III Bruker NMR spectrometer equipped with a non-cryogenic probe was used for all measurements unless otherwise mentioned.

Commercial H₂O₂ standardization.

The concentration of our commercial source of H₂O₂ (30% w/w; Fischer Scientific) was validated from the relative intensities of the H₂O₂ and H₂O resonances, observed following a small flip angle pulse.

R₁ measurement.

Inversion recovery T₁ measurements were carried out with the standard Bruker pulse sequence but replacing the non-selective 180° and 90° pulses by Gaussian shaped pulses of 3.5 ms duration. Variable recovery delays of 0.1, 2, 5, 10, 15, 20, 30, 50 and 120 ms were used. H₂O₂ intensities were fitted to I(t)/I₀ = 1−A₁exp(−R₁t), in which the fitted parameters, A₁ and R₁, are the preexponential factor and longitudinal relaxation rate, respectively; I(t) represents the spectral intensity for a recovery delay of duration t; and I₀ corresponds to the intensity obtained in the absence of the 180° pulse. Interscan delays of 300 ms were used.

R₂ measurements.

An interleaved Hahn echo experiment was used for H₂O₂ transverse relaxation rate measurements, with the non-selective 90° and 180° pulses replaced by Gaussian shaped pulses of 3.5 ms duration, and using Hahn echo delay durations of (0.1, 2, 5, 10, 15, 20, 30, 50 and 120 ms. A standard EXORCYCLE phase cycling scheme [21] was used to remove the effect of 180° pulse imperfections. The data were fitted to I(t) = A₂exp(−R₂t), in which the fitted parameters, A₂ and R₂, are the preexponential factor and the transverse relaxation rate, respectively. Interscan delays of 300 ms were used.

HX rate measurement.

Considering that R₁ ≈ R₂ and that both rates are multiple orders of magnitude larger than the natural R₁ and R₂ rates for pure water (assumed to be comparable to pure H₂O₂), measured R₁ and R₂ rates are effectively equivalent to HX rates. Therefore, the Hahn echo R₂ experiment was used for measuring HX rates, neglecting the very small contribution of natural transverse relaxation. Typically, Hahn echo delay durations of 0.1, 2, 5, 10, 15, 20, 30, 50 and 120 ms were used but for fast exchanging samples, durations of 0.1, 0.3, 2, 3, 5, 7, 10, 17 and 60 ms were used.

Results and Discussion

At room temperature, a 1 mM H₂O₂ sample in pure water, containing 2% D₂O, yields a weak exchange-broadened resonance at 11.3 ppm and a 55,000 times stronger water signal at the center of the spectrum, consistent with previous reports [12]. This large dynamic range adversely impacts the sensitivity attainable for the H₂O₂ signal. Due to fast exchange of H₂O₂ protons with those of H₂O, presaturation of the water signal obliterates the H₂O₂ signal and selective excitation with a shaped pulse, centered at 11.3 ppm, is used instead.

As previously reported [13], the HX rate is strongly temperature dependent and the H₂O₂ resonance narrows considerably upon cooling the sample to 2 °C (Fig. 1). Therefore, this temperature is used for all subsequent measurements.

Figure 1.

NMR spectra of 1 mM H₂O₂ in water, highlighting the temperature-dependent HX-related broadening of the signal. The sample contains 2% D₂O and spectra were recorded at 600 MHz using a non-cryogenic probe, with 256 scans for a total measurement time of ca 15 s. The signal resulted from excitation with a 3.2-ms Gaussian shaped 90° excitation pulse, centered at the H₂O₂ resonance. The chemical shift scale is referenced to internal (CH₃)₃SiCH₂CH₂CH₂SO₃Na (DSS).

Upon addition of 50 mM phosphate buffer, used to evaluate the pH dependence of the H₂O₂ resonance line width, we observed that phosphate strongly broadens the H₂O₂ resonance across acidic, neutral and basic pH values, making it difficult to detect and indicating that phosphate catalyzes HX. For comparison, a sample containing only 1 mM of 2-(N-morpholino) ethanesulfonic acid (MES) buffer showed a fairly sharp resonance with a line width of ca15 Hz in the 5.5–6.0 pH range, consistent with results of Ryoo et al. [16].

To further investigate HX, we used H₂O₂-selective pulses to measure the selective longitudinal and transverse (R₂) relaxation rates, R₁ and R₂ (Fig. 2). At pH 5.9, measured R₁ and R₂ rates were the same within error of the measurement (45.0 ± 0.4 vs. 45.3 ± 0.4 s⁻¹, respectively). This result indicates that the loss of longitudinal and transverse magnetization is dominated by the exchange of H₂O₂ hydrogens with water. Thus, to a good approximation, the R₁ and R₂ measurements correspond to the HX rate. As shown in Fig. 2, the excited H₂O₂ magnetization decays to ca 10% of its initial value in as little as 50 ms, defining this duration as a close to the approximate optimal acquisition time for recording spectra with adequate sensitivity and resolution. Together with a short interscan delay of 1 ms, this allows for recording ca18 scans per second, thereby increasing the signal to noise ratio per unit of time compared to recording other small molecule NMR spectra.

Figure 2.

R₁ and R₂ measurements of the H₂O₂ signal at 11.3 ppm on a sample at 2 °C containing 1 mM H₂O₂ buffered at pH 5.9 using 1 mM MES.

A plot of the HX rates, derived from an interleaved Hahn echo experiment, against pH shows a well-defined minimum at pH 6.2 (Fig. 3). The plot is similar to those of peptide backbone amides except that the latter has a minimum near pH 4 [22]. Similar to proteins [23, 24], the first order apparent exchange rate (k_ex) can be written as

$$k_{ex}=k_{H\mathit{2}O}\left[H_{\mathit{2}}O\right]+k_{MES}[MES]+k_H\times {\mathit{10}}^{-pH}+k_{OH}\times {\mathit{10}}^{-pOH}$$ (1)

in which k_H2O, k_MES, k_H and k_OH are water, MES, acid and base catalyzed rate constants, respectively, and pH and pOH are the negative of the base 10 logarithm of the H⁺ and OH⁻ ion concentrations, respectively. Fitting the data to eq. 1 (dashed line in Fig. 3) yields values of 12 ± 1 s⁻¹ for k_H2O [H₂O] + k_MES [MES]; k_H = (2.1 ± 0.1) × 10⁷ M⁻¹s⁻¹ and k_OH = (6.1 ± 0.2) × 10⁹ M⁻¹s⁻¹. The change of pH upon cooling of the sample from 20 to 2 °C was determined by linear extrapolation of glass electrode measurements at 20 and 5 °C [25]. For calculating pOH at 2 °C, a water dissociation constant of pK_w = 14.86 was used [26]. By varying the MES buffer concentration while keeping the sample at pH 6.2 and 2 °C, k_MES and subsequently k_H2O were extracted (Fig. 4; Table 1).

Figure 3.

The pH dependence of the HX rate of H₂O₂ protons with water. The pH of the samples was adjusted at 20 °C and the rates were measured at 2 °C. Samples contain 1 mM MES and 1 mM H₂O₂. The dashed line represents the best fit to equation (1).

Figure 4.

Dependence of the H₂O₂ HX rate on the MES concentration at pH 6.2, 2 °C.

Table 1.

Acid, base, water, and MES HX rate constants at 2 °C.

Catalyst	kex (M−1s−1)
H3O+	(2.1 ± 0.1) × 107
OH−	(6.1 ± 0.2) × 109
H2O	(1.4 ± 0.4) × 10−1
MESpH 6.2	(4.2 ± 0.1) × 103

The contribution of the unprotonated and protonated MES species to the exchange catalysis was dissected by varying the pH. Upon increasing the pH from 5.6 to 6.8 (at 2 °C), the mole fraction of the unprotonated MES increases from 0.14 to 0.71 (calculated using MES ionization enthalpy of +3.54 kcal/mol for pKa temperature correction). This pH increase resulted in an increase in the slope of the HX rates versus MES concentration (Fig. 5A), indicating that the unprotonated form of MES catalyzes the exchange more than its protonated counterpart. The MES contribution to the exchange rate from the protonated and unprotonated forms can be written as

$$k_{ex,MES}=C+\left(k_a{X}_a+k_b{X}_b\right)[MES]$$ (2)

in which k_a and k_b are the general acid- and base-catalyzed rate constants, respectively, and X_a and X_b denote the corresponding mole fractions. The term C is a constant accounting for water as well as specific acid- and base-catalyzed contributions, which vary with pH. By minimizing the residuals function,

$${\chi }^2={\sum }_i{\left\{k_{ex}^{obs(i)}-\left(C_i+\left(k_a{X}_a^i+k_a{X}_a^i\right)[MES]\right)\right\}}^2$$ (3)

k_a and k_b values can then be obtained, where i runs over all the experimental pH values, and the mole fractions at each pH are calculated using the literature buffer pK_a values and their temperature dependence [25]. The unprotonated form of MES is found to catalyze the exchange about eight times stronger than its protonated form (Table 2), presumably resulting from the involvement of the MES tertiary amine in the HX catalysis. Similarly, the free nitrogen of the unprotonated imidazole buffer catalyzes exchange nearly 100-fold stronger than MES, while its protonated form has no detectable effect on HX. Analogous to the above cases, the more basic phosphate species, ${\text{HPO}}_4^2{}^{-}$, catalyzes HX about five-fold more than ${\text{H}}_2{\text{PO}}_4{}^{-}$. In contrast, acetic acid catalyzes the exchange only in its protonated form, as is reflected in the larger slope of the line at pH 5.5 compared to pH 6.8 (Fig. 4D). Monovalent inorganic salts have only small effects on the exchange: Addition of 50 mM sodium chloride or azide increases the HX rates from 38.1 ± 0.3 M⁻¹ s⁻¹, in the absence of salt, to 44.8 ± 0.7 M⁻¹ s⁻¹ and 58.9 ± 0.2 M⁻¹ s⁻¹, respectively (Fig. 6).

Figure 5.

Concentration dependence of H₂O₂ HX rates for (A) MES (B) imidazole (C) phosphate and (D) acetate at two pH values, 2 °C.

Table 2.

Contribution of protonated and unprotonated forms of several buffers and salts to the H₂O₂ HX rate.

	kex (M−1s−1)
Compound	protonated	unprotonated
MES	(1.3 ± 0.4) × 103	(1.0 ± 0.1) × 104
Imidazole	< 103	(8.0 ± 0.1) × 105
Phosphate	(1.3 ± 0.1) × 104	(6.6 ± 0.5) × 104
Acetic acid	(8.7 ± 0.4) × 104	< 102
NaCl	---	(1.3 ± 0.2) × 102
NaN3	---	(4.2 ± 0.1) × 102

Figure 6.

Measurement of HX rates for samples containing 50 mM NaCl and NaN₃ compared with rates of the salt-free sample (1 mM MES).

To test the sensitivity of the NMR detection method, a 40 nM H₂O₂ sample was prepared with the spectrum recorded on a 600 MHz instrument equipped with a Bruker TCI cryoprobe (Fig. 7). A signal to noise (S/N) of 3.8 was achieved after 100K scans (97 minutes). This S/N corresponds to an error of only ~25% in the quantification of a 40-nM H₂O₂ sample and a detection limit of 32 nM. Based on typical concentrations reported in the literature [2, 3, 15], this sensitivity suffices for many applications, including quantification of H₂O₂ in exhaled breath condensate, air, rain, blood and other biological fluids.

Figure 7.

NMR spectrum of a 40-nM H₂O₂ sample in water containing 2 % D₂O and 1 mM MES. The pH was adjusted to 6.0 at 20 °C, and the spectrum was obtained at 2 °C using a 600-MHz Bruker NEO spectrometer equipped with a TCI cryoprobe. To further reduce excitation of the water resonance for this very dilute sample, selectivity of the excitation pulse was increased by extending the duration of the Gaussian shaped excitation pulse to 6 ms. An acquisition time of 50 ms and an interscan delay of 1 ms, with 102400 scans for a total measurement time of ca 1.6 h were used.

Conclusions

In the absence of pH-dependent HX catalysts, at dilute H₂O₂ concentrations the narrowest resonance is obtained at pH values of 6.0 at 20 °C (or 6.2 at 2 °C). The strong temperature dependence of the HX rate results in narrowing of the resonance when cooling samples to just above the freezing point of water. Unlike peptide backbone amides, water-catalyzed HX of H₂O₂ is significant near the pH value where the combined acid- and base-catalyzed HX is at a minimum. Our measurements show that while phosphate species and common functional groups in proteins such as imidazole rings and carboxylic acids can catalyze HX, sodium chloride or azide do not affect the exchange significantly. The present characterization of the H₂O₂ HX rates serves as the basis for preparing the sample conditions needed for quantifying the presence of small quantities of H₂O₂ in a wide range of biological fluids and other aqueous samples. Taking advantage of the rapid HX of water, selective excitation of the H₂O₂ resonance permits the use of short delays between scans, yielding a sensitivity on a 600 MHz instrument equipped with a triple resonance cryogenic probe of ca 25:1 per 6 minutes for a 1-μM sample. Single-channel cryogenic probeheads, optimized solely for ¹H detection, could further improve the sensitivity of H₂O₂ detection by NMR.

Highlights.

• Acid- and base-catalysed hydrogen exchange (HX) rates were measured for H₂O₂

• Water significantly catalyses HX of H₂O₂

• Common buffers such as phosphate and MES catalyse HX of H₂O₂

• HX catalysis by acetic acid and imidazole is strongly pH dependent

• H₂O₂ concentrations as low as 40 nM in water are readily measured