Electrochemical Quantification of Extracellular Local H₂O₂ Kinetics Originating from Single Cells

Abstract

Aims: H₂O₂ is produced by all eukaryotic cells under physiological and pathological conditions. Due to its enormous relevance for cell signaling at low concentrations and antipathogenic function at high concentrations, precise quantification of extracellular local hydrogen peroxide concentrations ([H₂O₂]) originating from single cells is required.

Results: Using a scanning electrochemical microscope and bare platinum disk ultramicroelectrodes, we established sensitive long-term measurements of extracellular [H₂O₂] kinetics originating from single primary human monocytes (MCs) ex vivo. For the electrochemical techniques square wave voltammetry, cyclic and linear scan voltammetry, and chronoamperometry, detection limits for [H₂O₂] were determined to be 5, 50, and 500 nM, respectively. Following phorbol ester stimulation, local [H₂O₂] 5–8 μm above a single MC increased by 3.4 nM/s within the first 10 min before reaching a plateau. After extracellular addition of H₂O₂ to an unstimulated MC, the local [H₂O₂] decreased on average by 4.2 nM/s due to degradation processes of the cell. Using the scanning mode of the setup, we found that H₂O₂ is evenly distributed around the producing cell and can still be detected up to 30 μm away from the cell. The electrochemical single-cell measurements were validated in MC populations using electron spin resonance spectroscopy and the Amplex^® UltraRed assay.

Innovation and Conclusion: We demonstrate a highly sensitive, spatially, and temporally resolved electrochemical approach to monitor dynamics of production and degradation processes for H₂O₂ separately. Local extracellular [H₂O₂] kinetics originating from single cells is quantified in real time. Antioxid. Redox Signal. 29, 501–517.

Introduction

Among reactive oxygen species (ROS) relevant for cells, H₂O₂ is probably the most important molecule. It is produced by intra- and extracellular reactions, including the respiratory chain of mitochondria, xanthine oxidases, and NADPH oxidases (NOXs) (16, 24, 52). As major sources of ROS, plasma membrane-residing NOXs produce superoxide radicals (^●O₂⁻) outside the cell by transferring one electron from NADPH, thereby reducing O₂. Due to its charge, ^●O₂⁻ does not rapidly diffuse through the plasma membrane into the cytosol, but is almost quantitatively dismutated into H₂O₂ and O₂ (2 ^●O₂⁻ + 2 H⁺ → H₂O₂ + O₂). Dismutation occurs spontaneously and is enhanced by the superoxide dismutase SOD3 (ecSOD) in the extracellular environment (34). Caused by the short ^●O₂⁻ half-life time of around 2 μs (75), direct determination of the ^●O₂⁻ production rate is rather difficult (70). As an indirect quantitative measure of ^●O₂⁻, the more stable H₂O₂ is monitored, which is considered to be the main transmitter of redox signaling (20, 40). To maintain a well-defined redox state of cells, which is an important prerequisite for cellular signaling, a variety of H₂O₂ degradation systems antagonize its production (6, 73). Supported by aquaporin channels, H₂O₂ can diffuse back into the producing cell (5, 11, 27, 36) and, depending on local hydrogen peroxide concentration ([H₂O₂]) and cell density, it can act on neighboring cells. Considering its relevance for many biological processes through oxidation of protein cysteine residues and of other targets (71), quantification of local extracellular H₂O₂ kinetics originating from single cells and separation of H₂O₂ production and degradation are needed to fully explore its physiological roles. Taking intra- and extracellular sources of H₂O₂ into account, sensitive, quantitative, and dynamic measurements of this species in the cell and outside the plasma membrane are required.

Innovation.

Electrochemical methods are tested for quantification of extracellular local concentration of hydrogen peroxide [H₂O₂] kinetics originating from a single cell. With a scanning electrochemical microscope and bare disk platinum ultramicroelectrodes of 10 μm diameter (vs. Ag/AgCl), 5-500 nM up to 1 mM [H₂O₂] could be resolved depending on the method. Using chronoamperometry, we performed stable long-term (>3 h) measurements with high temporal (>1 Hz) and spatial (<1 μm) resolution from individual primary human monocytes ex vivo. Careful calibrations allow quantification of local extracellular [H₂O₂] in the vicinity of single cells and the separation of H₂O₂ production and degradation by the same cell.

To monitor intracellular H₂O₂ on the single-cell level, genetically encoded fluorescent protein sensors such as HyPer or roGFP2-Orp1 (9, 20, 24, 57, 68) and roGFP2-Tsa2ΔCPΔCR (49) can be applied. These sensors provide fascinating insights into intracellular H₂O₂ and its spatial distribution in subcellular compartments. However, there are also drawbacks: problematic calibration, often a limited dynamic range, and interference by pH and by endogenous reductants, as well as light-induced ROS generation (12, 72).

To monitor extracellular H₂O₂ and ^●O₂⁻, Amplex^® Red/Amplex^® UltraRed assays (41, 57) and electron spin resonance (ESR) spectroscopy (19, 63), respectively, are frequently used. They offer sensitivity in the low nanomolar range, but do not allow single-cell measurements. Extracellular determinations of [H₂O₂] in populations of 10,000 to 100,000 cells per mL have revealed nano- up to low millimolar values (24, 32). However, considering the complex H₂O₂ production and degradation kinetics, these values cannot easily be converted to single cells.

Since a couple of years, sophisticated electrochemical methods have been developed and adapted to selectively determine H₂O₂ from cell populations and single cells (38, 39). Thereby, excellent results for certain individual parameters such as sensitivity or specificity have been obtained [e.g., (2)]; comprehensive measurements of parameters being relevant for the analysis of interrelated physiological processes on the single-cell level, however, have not been reported.

In the present study, we introduce a highly sensitive, spatially, and temporally resolved electrochemical approach to monitor kinetics of H₂O₂ production and degradation from a single cell. Measurements were validated in respective monocyte (MC) populations by a recently modified ESR spectroscopic method (63) and an adapted fluorescence-based Amplex UltraRed assay (47).

Results

Quantification of [H₂O₂] with the scanning electrochemical microscope

The ElProScan ELP 3 scanning electrochemical microscope (SECM) used for the present study is specified in the Materials and Methods section. For all measurements, ultramicroelectrodes (UMEs) with a bare platinum wire of 10 μm diameter (vs. Ag/AgCl) were used, whose electrochemical properties substantially differ from electrodes of larger scales (3, 7, 26). Experiments were performed in phosphate-buffered saline (PBS-I) in ambient air to meet oxygen requirement of living cells. Electrochemically, H₂O₂ can be determined by its reductive (H₂O₂ + 2H⁺ + 2e⁻ → 2H₂O) or oxidative (H₂O₂ → O₂ + 2H⁺ + 2e⁻) electrode reaction. Since reduction of H₂O₂ in a buffer containing O₂ interferes with the oxygen reduction reaction (ORR), leading to additional H₂O₂ (1, 22, 59, 75), anodic oxidation was chosen for H₂O₂ determination.

[H₂O₂] calibration by cyclic voltammetry and linear scan voltammetry

To define the respective redox potential under our experimental conditions, cyclic voltammetry (CV) was analyzed in a potential window of -100 mV up to +1.0 V at a scan rate of 50 mV/s in the absence (blank current; black traces in Fig. 1A) or the presence of 20 μM [H₂O₂] (red trace). Current values specifically increased with increasing [H₂O₂] at a potential range of 500 to 650 mV, which is in agreement with the peak potential for anodic oxidation of H₂O₂ at 650 mV (39, 67). Consequently, in CV and linear scan voltammetry (LSV) (Fig. 1A, B), current values were sampled at 650 mV (I₆₅₀) and chronoamperometric (CA, see the [H₂O₂] calibration by chronoamperometry section) experiments were conducted at 650 mV (Fig. 1C). Current values (I₆₅₀) linearly increased by 3.3 ± 0.06 pA for CV (Fig. 1D, red line) and by 3.7 ± 0.11 pA for LSV per μmol of H₂O₂ (Fig. 2E, red line). As explained in the Materials and Methods section, the starting potential of the CV and LSV determines the slope of the linear regression. Starting at +100 mV revealed almost no current change with increasing [H₂O₂] (blue lines in Fig. 1D, E), whereas starting at -1.0 V was followed by the steepest slope (black lines in Fig. 1D, E). This can be explained by additional electrochemical production of H₂O₂ due to the ORR (see the Materials and Methods section). Red-colored traces in Figure 1D and E were obtained with a starting potential at -100 mV where cleaning of the UME is expected (compare the Materials and Methods section).

FIG. 1.

[H₂O₂] calibration with different electrochemical techniques. Experiments were conducted in PBS-I buffer, using a 10 μm Pt-UME versus Ag/AgCl. (A) CVs with 0 (black) and 20 μM H₂O₂ (red). Potential window from -0.1 to +1.0 V; scan rate 50 mV/s. In the inset (potential window of 0.6–0.7 V at a larger scale), the lower detection limit for [H₂O₂] of 50 nM (red trace; black trace 0 H₂O₂) at a scan rate of 20 mV/s is exemplified. Green arrows indicate forward and backward directions of cycling. (B) LSVs with 0 (black) and 35 μM H₂O₂ (red). Forward scanning was conducted from -0.1 to +1.0 V at a scan rate of 50 mV/s. Then, potential was reduced in one quick step to -0.1 V, clamped for 3 s, before forward scanning started again. (C) Representative CA measurement at 650 mV under pulsed additions of H₂O₂ (5 μM steps; arrows) to cell-free buffer (example for n = 10 similar experiments). (D) Current–dose relationship for H₂O₂ measured in CV at different potential windows described in the box (scan rate 50 mV/s). ΔI values represent current values sampled at 650 mV subtracted by respective basal current values of the CV before addition of H₂O₂. For each experiment with different potential window and/or different [H₂O₂] current values (ΔI₆₅₀) of sweep numbers 2–10 were averaged. Mean ± SEM (n = 5) is shown. (E) Current–dose relationship for H₂O₂ measured in LSV at different potential windows described in the box [scan rate 50 mV/s; protocol as in (B)]. Current values (ΔI₆₅₀) were analyzed as for CVs in (D). Mean ± SEM (n = 4) is shown. (F) Current–dose relationship measure in CA for [H₂O₂] from 500 nM to 1 mM (cell free, n = 10; red inset n = 3 for [H₂O₂] of 0–5 μM). For calculations, current values (I₆₅₀) of the respective approximate linear steady-state phase after each H₂O₂ addition were averaged. Mean ± SEM is shown. CA, chronoamperometric; CV, cyclic voltammogram or cyclic voltammetry; [H₂O₂], local hydrogen peroxide concentration; LSV, linear scan voltammogram; UME, ultramicroelectrode.

FIG. 2.

H₂O₂ determination by SWV. Experiments were performed in PBS-I buffer, using a 10 μm Pt-UME versus Ag/AgCl. (A) SWV is a repetitive double-step chronoamperometric technique. For each double-step, two equal, but oppositely directed, potential pulses (±75 mV) were generated, the duration of which was set to τ = 2 s. In a potential window from -0.1 up to +0.9 V mean (nominal) voltage was increased over time by an increment of 10 mV set on top of each positively directed potential step (scan rate = 5 mV/s). Values of the anodic forward (1) and cathodic reverse current (2) were sampled and averaged from the last 10% of the respective plateau phase (red bars). (B) Current traces of a representative SWV experiment in blank buffer. For the net current (I_difference; red traces), I_reverse (measured during the negatively directed potential steps; blue trace) was subtracted from I_forward (measured during the positively directed steps; black trace). The inset with a potential window from 0.1 to 0.9 V shows the characteristic change in the current trace (I_difference) at a mean potential between 470 and 500 mV after H₂O₂ to a final concentration of 5 nM had been added (dotted line). (C) For calibrating current changes to [H₂O₂], maxima of I_difference were determined for different [H₂O₂] (in μM: 990: e, dark blue; 690: d, yellow; 390: c, green; 190: b, light blue; 0: a, black). After definition of a baseline (solid red line), Δcurrent (I_peak - I_baseline) was calculated. Alternatively, the size of the area under the curve (gray) was determined. As an example, analysis for the dark blue curve (e) is shown. (D) Δcurrent and area under the curve as determined in (C) were normalized to each other and plotted in arbitrary units against [H₂O₂]. The inset shows a magnification at lower [H₂O₂]. SWV, square wave voltammetry.

To summarize, in CV and LSV calibration experiments, [H₂O₂] from 500 nM up to 1 mM could be resolved. At lower scan rates of 20, 10, or 5 mV/s, but with reduced temporal resolution, [H₂O₂] down to 50 nM was detectable (the scan rate of 20 mV/s shown in the inset of Fig. 1A, red trace).

[H₂O₂] calibration by chronoamperometry

Steady-state conditions, where the current measured with UMEs is purely faradaic with no capacitive component, are ideally met during CA experiments at a fixed potential. For calibration, H₂O₂ was added to the PBS-I buffer to reach final concentrations of 5 μM or multiples of it (Fig. 1C). Calibrations were performed in the presence of attached human MCs at the end of each experiment. The UME was fixed at least 1 mm above the cell layer to avoid local effects from the cells. Under these conditions, currents increased on average by 224 ± 12.4 fA per μmol of H₂O₂ (n = 25). In cell-free PBS-I, the resulting calibration factor of 233.6 ± 20 fA per μmol of H₂O₂ (n = 10; Fig. 1F) was not significantly different. To determine the lowest [H₂O₂] measurable with CA, [H₂O₂] of a cell-free PBS-I buffer was increased stepwise by 100 nM. As shown in the red inset of Fig. 1F, 500 nM [H₂O₂] could be distinguished from blank buffer. For [H₂O₂] higher than 1 mM, calibrations did not provide reliable data (not shown), in agreement with findings of Urbach and Bowen (65) and Zhang and Wilson (74).

The CA calibration for [H₂O₂] was linear over more than three orders of magnitude from 500 nM up to 1 mM (Fig. 1F). Considering the signal-to-noise ratio of a typical UME and the calibration value of 224 ± 12.4 fA per μmol of H₂O₂, a concentration of 500 nM can thus be reliably measured in CA.

[H₂O₂] calibration by square wave voltammetry

Finally, square wave voltammetry (SWV) was tested for its applicability to measure [H₂O₂] in the nano- to millimolar range (Fig. 2). This technique combines chronoamperometric with pulsed voltammetric components (44–46). SWV was reported to be superior to other voltammetric methods, mainly due to enhanced sensitivity for analytical determinations, but has not been applied to measure low [H₂O₂] with UMEs (44). The principle of SWV is described in detail elsewhere (44). In brief, anodic oxidation of H₂O₂ was used for its determination within a potential window from -100 up to +900 mV. Oppositely directed potential steps of 1 s duration each, along a staircase ramp, were applied repetitively to values 75 mV above and below the mean potential line and thus reached anodic oxidation of H₂O₂ already at a mean potential between 470 and 500 mV. Potential was increased over time by an increment of 10 mV, set on top of each positively directed potential step (scan rate = 5 mV/s). Figure 2A shows a current trace resulting from one potential step pair. The initial peaks of charging current quickly decline to reach the respective stable plateau. Values of the anodic forward (red-colored 1) and cathodic reverse current (red-colored 2) were sampled and averaged from the last 10% of the respective plateau phase (Fig. 2A, red bars).

In Figure 2B, current traces from a representative SWV experiment are depicted. For the net current (I_difference; red traces), values from I_reverse (cathodic current; blue trace) were subtracted from those of I_forward (anodic current; black trace) according to I_difference = I_forward - I_reverse. For analysis of the current–dose relationship for H₂O₂, values of I_difference were compared at a potential interval from 0 to +900 mV, where H₂O₂ oxidation-induced current changes are expected. The inset in Figure 2B reveals the characteristic change of I_difference at a mean potential between 470 and 500 mV after H₂O₂ had been added to the buffer. At the given scan rate of 5 mV/s, [H₂O₂] of 5 nM could be resolved (Fig. 2B, inset).

Methods to analyze SWV experiments (Fig. 2C, D) are detailed in the Materials and Methods section. In brief, first the baseline of a peak current curve (I_difference) is defined (solid red line in Fig. 2C). Then, the area under the curve (gray) is determined by integration or, as an alternative, ΔI is calculated from the baseline to the peak of the curve (Fig. 2C). Both approaches turned out to be equally valid for SWV calibration of [H₂O₂] (Fig. 2D).

In summary, with SWV, [H₂O₂] from 5 nM up to 1 mM could be measured, making it the most sensitive method with the largest concentration range of six orders of magnitude.

Buffering capacity of PBS-I – pH measurements

H₂O₂ is known to chemically act as a weak acid and, by pH changes, may modify electrochemical properties of the measuring solution. To test the buffering capacity of PBS-I used in all electrochemical experiments, [H₂O₂] was increased stepwise to reach final concentrations of 1 up to 5 mM in the stirred buffer solution. Simultaneous pH measurements revealed stability of the starting pH value at 7.37 over the whole experiment and a wide range of [H₂O₂] (see Fig. 4A).

FIG. 4.

Control experiments. (A) pH measurements in PBS-I before and after addition of 1 to 5 mM H₂O₂ (n = 3). (B) and (C) Single-cell chronoamperometric experiments (at 650 mV) with a 10 μm Pt-UME versus Ag/AgCl in PBS-I with the electrode fixed 5–8 μm above the MC. (B) Addition of SOD (50 U/mL) to a TPA-stimulated MC (example for n = 3). (C, D) TPA stimulation of MCs pretreated either with the pan NOX inhibitor VAS2870 (20 μM in DMSO; 30 min; red trace) or with DMSO (black trace; examples for n = 5 per condition). (E, F) Comparison of normalized ESR spectra from experiments with MC suspensions (depicted in Fig. 7D, E) to test for NO^● formation by TPA-stimulated MCs. Experimental conditions and settings of the ESR spectrometer are detailed in the Materials and Methods section. Black traces are from DMSO treated (control; n = 3) and red traces from cells about 35 min after TPA application (n = 3). Spectra in (E) and the enlarged cut out in (F), corresponding to the magnetic field interval marked by the red bar in (E), show complete overlay of the normalized spectra. Because of the low intensity, the signal-to-noise ratio of the spectra from unstimulated MCs is higher than from stimulated cells. ESR, electron spin resonance; SOD, superoxide dismutase.

Effect of glucose in the buffer

Since glucose is required as a nutrient for living cells and as a prerequisite of H₂O₂ production in human MCs (51), we compared the H₂O₂ measurements in the absence and presence of 5 mM glucose in cell-free PBS-I. Neither with the voltammetric techniques (CV, LSV, or SWV) nor during CA measurements did glucose show any electrochemical effect [not shown; (35)].

Quantification of H₂O₂ generated by single human MCs

MCs are white blood cells (leukocytes) that fulfill important functions for innate immunity. They can differentiate into macrophages (MPhs) that eliminate bacteria and other pathogens by phagocytosis (50, 60, 64). MCs and MPhs are known to produce high amounts of ROS intra- and extracellularly. The major sources for external ROS are NADPH oxidases (NOXs), which are located in the plasma membrane (37, 52). In human MCs, NOX2 is the main ^●O₂⁻- and thus H₂O₂-generating enzyme complex (57).

Artificial NOX activation by the phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA) causes MCs to produce ^●O₂⁻, which is dismutated to H₂O₂. Using CA at 650 mV, current was measured with the UME placed 5–8 μm above a single MC (Fig. 3C). Using this technique, only local current changes (ΔI₆₅₀) within an active volume around the cell and the UME are monitored, which might be completely different from the electrochemical situation in the unstirred bulk solution. Increasing the gap between the UME and the H₂O₂-producing cell is followed by a decrease in ΔI reaching ΔI = 0 at a distance of ≥1 mm (not shown).

FIG. 3.

H₂O₂ measurements from single MCs after stimulation with the phorbol ester TPA. CA experiments were conducted at 650 mV in PBS-I buffer. A 10 μm Pt-UME (vs. Ag/AgCl) was placed 5–8 μm above different individual MCs in the experiments shown in (A) and (D). (A) CA measurement of a single MC after stimulation with TPA (red trace). The solvent DMSO (0.2%) was used in a control experiment with the same cell, but preceding TPA stimulation (black trace). (B) Average total current increase (±SEM) of single cells as shown in (A) after application of DMSO (161 ± 33.6 fA; n = 25) or TPA stimulation (639.8 ± 108.8 fA; n = 23). (C) Bright-field image of an MC and the tip of the UME. (D) Three consecutive 2D scans after TPA stimulation of the MC shown in (C) whose position is marked by dashed white circles. Numbers in the right upper corners refer to the time points indicated in (A) where another exemplary MC was measured. 2D scan images (D) are representative for n = 17 different scanning experiments with individual cells. Settings: constant height; scanning rate x = 2 μm/s, 50-nm resolution; y = 2 μm, alternate mode; raw data, no compression. MC, monocyte; TPA, 12-O-tetradecanoylphorbol-13-acetate.

As shown in Figure 3A, the solvent for TPA, dimethyl sulfoxide (DMSO), caused a small and transient electrochemical artifact. Following TPA application, a significant rise in current was measured above the MC reflecting a rise in [H₂O₂].

To test, if in addition to H₂O₂ other reactive oxygen or nitrogen species are also released by a TPA-stimulated MC, species which possibly would modify the H₂O₂ determination at the given potential (2), we performed several control experiments (Fig. 4). First, superoxide dismutase (SOD) was applied on top of TPA after H₂O₂ production of the cell had reached the plateau phase (Fig. 4B). Since no change of current was observed, we conclude that ^●O₂⁻ is not determined at 650 mV and, in addition, ^●O₂⁻ produced at the plasma membrane is quickly and quantitatively dismutated to H₂O₂. Furthermore, MCs were pretreated for 30 min with VAS2870 in DMSO, a cell-permeable pan NOX inhibitor (Fig. 4C, D, red). This treatment completely prevented the TPA-induced current rise (only the small DMSO artifact was left). Application of external H₂O₂ (5 μM), however, resulted in the usually observed current increase. Control cells of the same batch, but pretreated with DMSO instead of VAS2870, responded to TPA stimulation with H₂O₂ production (Fig. 4C, black trace; Fig. 4D, gray column). We conclude that the current increase exclusively reflects an increase in [H₂O₂] due to NOX enzyme activity.

To test for production of other radicals besides ^●O₂⁻, ESR spectroscopy was applied (Fig. 4E, F). We analyzed signals from TPA-stimulated MCs (red traces in Fig. 4E, F) and signals of unstimulated controls (black traces in Fig. 4E, F). Because of ^●O₂⁻ production, signal intensity from TPA-stimulated MCs was higher than signal intensity of control cells. Following normalization of different intensities, ESR spectra could be plotted on top of each other (Fig. 4E, F). Highly diffusible radicals (such as NO^● and NO₂^●) should modify the line width of the spectra. As this was not the case, we conclude that measurable amounts of additional radicals (e.g., NO^●) were not produced. If minimal quantities of NO^● would be generated constitutively or after TPA stimulation, this species would effectively react with ^●O₂⁻ to form ONOO^- (peroxynitrite). It cannot be excluded that ONOO^- is also trapped at a potential of 650 mV (2). However, since NO^● could not be detected, [ONOO⁻] should be much lower than [^●O₂⁻] and [H₂O₂] if present at all. Thus, ONOO⁻ signals should not compromise H₂O₂ measurements. Collectively, our experiments indicate that current changes (ΔI₆₅₀) induced by TPA-stimulated MCs reflect changes in [H₂O₂].

The extracellular distribution of H₂O₂ released from a single cell is highly relevant not only for the producing cell but also to its neighboring cells. Therefore, we used the scanning option of the setup to monitor the spatial distribution of H₂O₂ being continuously delivered by an MC into its immediate vicinity. Only those MCs were scanned that had no neighboring cells in the periphery to exclude the interference by other MCs. For scanning purposes, the position of the UME was fixed (constant height mode) and the table with the cell dish was automatically moved following a predefined protocol. Three consecutive 2D scans from the MC pictured in Figure 3C are presented in Figure 3D corresponding to the time points indicated in Figure 3A. The activity profiles show that H₂O₂ is rather evenly distributed all over the cell, resembling similar levels as shown in Figure 3A (from a different MC). Due to diffusion, H₂O₂ forms a concentration gradient around the cell and is detectable at least 20 μm away from it in this example. Thus, H₂O₂ generated by one cell could easily influence neighboring cells.

H₂O₂ quantification

TPA-stimulated single MCs as the one depicted in Figure 3A produced a total current increase (ΔI₆₅₀) of 639.8 ± 108.8 fA (n = 23) compared with 161.0 ± 33.6 fA by the solvent DMSO (n = 25; Fig. 3B). Maximal current values were reached 23.5 ± 4.2 min (n = 17) after TPA application. Subtracting the DMSO artifact, maximal ΔI of 639.8 fA – 161.0 fA = 478.8 fA was measured in the 23 cells treated as the one in Figure 3A. Since current changes in single-cell experiments as well as in calibration experiments are measured in the same volume of solution around the UME tip, current–dose calibrations for H₂O₂ can be applied to determine its concentration in cell experiments. Using the calibration factor for CA experiments of 224.0 ± 12.4 fA/μmol of H₂O₂, an increase of ΔI by 478.8 fA reflects a maximal [H₂O₂] of 2.1 μM in the immediate vicinity of the producing MC. Considering that this concentration is reached on average 23.5 min after TPA application, the net increase in [H₂O₂] caused by a single cell per second is approximately 1.5 nM in the local volume at the UME tip.

Applying the same conditions as in the CA measurements (Fig. 3A), we repeated the experiments with CV. For this purpose, the UME was positioned 5–8 μm above an MC. Cycling in a potential window from -100 mV up to +1.0 V at a scan rate of 20 mV/s was repeated for >1 h. After 10 min of equilibration in blank buffer, cells were challenged with TPA (1 μM). Current changes at 650 mV (ΔI₆₅₀) within the first 30 min after TPA stimulation were quantified and the respective calibration for CV of 3.3 ± 0.06 pA per μM H₂O₂ (see Fig. 1) was applied. We found an average local [H₂O₂] increase of 1.3 nM per second (n = 7; not shown) in very good agreement with the CA experiments. We conclude from the CV and CA experiments that a TPA-stimulated human MC increased extracellular local [H₂O₂] on average by 1.4 nM per second.

H₂O₂ production of MCs stimulated by Escherichia coli bacteria - respiratory burst

TPA is a commonly used nonphysiological stimulus for human MCs. To mimic physiological conditions, MCs were exposed to a cell-free supernatant from cultured Escherichia coli bacteria. The supernatant contains bacterial signal peptides that activate formyl peptide receptors in the plasma membrane of the MC (17). Thereby, an intracellular signaling cascade is initiated, which results in the so-called respiratory burst. The respiratory burst is an important part of the complex innate immune response to bacterial pathogens and includes high consumption rate of oxygen coupled to H₂O₂ production (23, 30, 43). Other components of the bacterial supernatant may also directly activate NOX2 (54).

An aliquot of the supernatant (1:100 vol/vol; final [protein] of 110 μg/mL) was applied to cells in the measuring buffer and five consecutive 2D scans were taken from the individual MC depicted in Figure 5A. Each scan comprises 20 scan lines, one line in the positive, alternating with one line in the negative x-direction, but always separated by 2.5 μm in the y-direction. Scanning of the area of 50 × 50 μm at a velocity of 2 μm/s required 8.5 min per scan. Starting from low basal activity (Fig. 5B), the cell responded with a respiratory burst characterized by a fast and huge increase of current, which gradually declined (Fig. 5C). The series of five 2D scan images shows the example of one individual MC. The decline of current with time can be explained by a combination of H₂O₂ diffusion, degradation in- and outside the cell (see the Quantification of H₂O₂ degraded by single human MCs section), and fast desensitization of the formyl peptide receptors (17).

FIG. 5.

Respiratory burst of an MC after stimulation with Escherichia coli. 2D scans (B, C) of the cell in (A) and control experiments (E) were performed chronoamperometrically at 650 mV in PBS-I buffer using a 10 μm Pt-UME versus Ag/AgCl. (B, C). With the UME placed 5–8 μm above, the MC shown in (A) was measured in five consecutive 2D scans before (B) and after addition (C) of a cell-free E. coli supernatant in LB medium (Bac). Dashed white circles mark the position of the MC (A) during the scanning procedure. Scan settings: constant height; scanning rate x = 2 μm/s, 50-nm resolution; y = 2.5 μm, alternate mode; raw data, no compression. To complete one 2D scan, 8.5 min were needed. (D) Dotted blue traces quantify current values of the five 2D scan images in (B) and (C); numbers of waves 1–5 refer to the corresponding 2D scan numbers. Each 2D scan comprises 20 scan lines alternating in the x-direction, separated by 2.5 μm in the y-direction (meander scan). The mean current value of each scan line of the five 2D scans (5*20 scan lines = 100 mean values) is represented by one blue dot (D). (E) Control experiments (representative of n = 8 per condition). Addition of bacterial supernatant [Bac; same amount as in (C)] to blank PBS-I (black trace) and to PBS-I containing H₂O₂ (10 μM; red trace).

The traces in Figure 5D average the current from five 2D scans shown in Figure 5B and C. Each blue dot in Figure 5D represents the mean current of one scan line; the numbers of the five waves in Figure 5D refer to the respective scan numbers in Figure 5B and C. The traces reflect the characteristic dynamics of [H₂O₂] changes during the respiratory burst of a single MC.

In control experiments, addition of the appropriate amount of bacterial supernatant (Bac) or pure LB medium to cell-free PBS-I buffer did not change the current flow (Fig. 5E, black trace). In the presence of 10 μM H₂O₂, however, the supernatant (Bac) as well as the LB medium caused reduction of current by the same amount (Fig. 5E, red trace). As a consequence, in experiments applying bacterial supernatant in LB medium, current values of the respiratory burst are underestimated and [H₂O₂] concentrations should be even higher than reported here.

Quantification of H₂O₂ degraded by single human MCs

As depicted in Figure 3A for a single TPA-stimulated MC, the initial local [H₂O₂] increase (measured 5–8 μm above the MC as a rise in ΔI) can be higher than the mean increase while approaching a steady-state level. Within the first 10 min after TPA application, the average local [H₂O₂] change was 3.4 nM per second (n = 13). Such a change in production kinetics was observed in most cells and subsequently [H₂O₂] declined, although the stimulus was still present. It can be excluded that the decline is caused by NADPH- or O₂ depletion of the cell since the experiments were conducted in a glucose-containing buffer and in the presence of O₂. However, multiple extra- and intracellular mechanisms are known to be responsible for H₂O₂ breakdown (5, 21, 66).

To monitor H₂O₂ degradation by single cells ex vivo, a bolus of H₂O₂ was applied to unstimulated MCs. The CA measurement from a single cell revealed a decay of [H₂O₂] from 20 μM to baseline in about 50 min in the example shown (Fig. 6A). In this case, degradation was quite fast; some other MCs showed slower response (compare, for instance, Fig. 7B). This variability was found to be specific for individual cells and not influenced by surrounding cells. Making use of the scanning mode of the setup, another MC (Fig. 6B), having been treated similarly as the one in Figure 6A, was scanned 10 min after H₂O₂ application. Since our setup scans a selected area from the lower right to the upper left corner, we have turned the scan image in Figure 6C by 180° to visualize the [H₂O₂] decrease over time corresponding to Figure 6A. Figure 6C shows that [H₂O₂] above and in close vicinity of the cell (blue area) was markedly reduced within a few minutes. It is assumed that extracellular H₂O₂ diffuses into the cell and is effectively degraded intracellularly. Results from both, Figure 6A and C, impressively visualize how MCs can function as a sink for H₂O₂.

FIG. 6.

Degradation of extracellular applied H₂O₂ by single MCs. CA experiments were conducted at 650 mV in PBS-I buffer using a 10 μm Pt-UME versus Ag/AgCl. The UME was placed 5–8 μm above the MC. (A) An MC was challenged first with the solvent DMSO (0.2%) and 30 min later with H₂O₂ (20 μM). (B) A different MC was treated in the same way as described in (A), and then a 2D scan was started 5 min after H₂O₂ application [C, green square corresponds to the time point in (A)]. The dashed white circle marks the position of the MC (B) during scanning procedure. Scan settings were the same as in Figure 3. (D) A cell-free control CA experiment over more than 4 h in the absence or presence of H₂O₂ (n = 3).

FIG. 7.

H₂O₂ production and degradation by MCs: validation of single-cell measurements using the ElProScan with cell population measurements by ESR and an AUR assay. (A, B) Chronoamperometric (CA) measurements of single MCs (n = 5–11) with a 10 μm Pt-UME versus Ag/AgCl. (C) Control in cell-free PBS-I (n = 4). (D, E) Cell population measurements with ESR (n = 3). (F) Control in cell-free PBS-I (n = 3). (G, H) Cell population measurements with the fluorescence-based AUR assay (n = 3). (I) Control in cell-free PBS-I (n = 11). For each approach in (A–I), one representative experiment is shown; results for ESR and AUR are expressed in arbitrary/fluorescence units (AU and FU, respectively). While CA measurements were conducted continuously (1 Hz sampling rate), in ESR- and AUR experiments, H₂O₂ was determined at selected time points for a pool of 25,000 cells each. Protocol for all experiments: 1 μM of the phorbol ester TPA (or its solvent DMSO, 0.2%) was applied 30 min after the initial equilibration phase. Thirty to 40 min later, either H₂O₂ (10 μM) or A. dest. was added. The same protocol was followed for cell-free control experiments shown in C, F, I, with TPA being replaced by DMSO. (A, D, G) TPA-stimulated cells with (red) and without (black) further treatment with H₂O₂. (B, E, H) Unstimulated cells challenged with H₂O₂ (red) or A. dest. (black). (C, F, I) Cell-free control experiments with DMSO and H₂O₂ (red) or A. dest. (black). AUR control experiments were terminated by addition of catalase (100 U/mL; arrow) to test the assay for selectivity to H₂O₂. AUR, Amplex^® UltraRed.

In control experiments (Fig. 6D; n = 3) performed in cell-free PBS-I, current values were stable over 4 h in the absence as well as in the presence of 10 μM H₂O₂, indicating also the chemical stability of H₂O₂ under the present experimental conditions. Thus, H₂O₂ degradation was completely dependent on cellular mechanisms. Comparing the experiments shown in Figure 6A and C with those in Figures 3 and 5, we conclude that production and degradation of H₂O₂ are working in parallel. Cellular degradation of H₂O₂ seems to be triggered solely by critical [H₂O₂] and is obviously not directly stimulated by TPA or formyl peptides. The capacity of an MC to eliminate H₂O₂ with a high rate and covering a much larger area than the cell itself (Fig. 6C) may be of great importance to control local [H₂O₂] relevant for autocrine and paracrine signaling in neighboring cells.

Quantification of local [H₂O₂] decrease outside of 10 MCs similar to the ones shown in Figure 6A and C revealed that the [H₂O₂] declined with rates between 1.7 and 6.6 nM per second. If degradation would not occur, single MCs would increase local extracellular [H₂O₂] not only by 1.4 nM (taken from CA and CV experiments) but by 3.1 up to 8.0 nM per second. Similarly, the initial local [H₂O₂] increase by 3.4 nM per second after TPA stimulation could even reflect initial local [H₂O₂] increases of 10 nM per second.

Validation of single-cell H₂O₂ determinations using bulk measurements with other techniques

To validate single-cell extracellular H₂O₂ measurements performed with the ElProScan (Fig. 7A, B; n = 5–11), we determined H₂O₂ in cell populations using the fluorescence-based Amplex UltraRed assay (AUR; Fig. 7G, H) and an ESR-based method recently established in our laboratory (63) (Fig. 7D, E). For comparison of the respective techniques side by side, modifications of the ESR and AUR protocols were required. Signal changes in the AUR assay and the here applied spectroscopic method are irreversible, thus not allowing dynamic measurements. To circumvent this disadvantage, single time point determinations of H₂O₂ were performed (47) (see the Materials and Methods section for details), which allow conclusions about the reaction kinetics, but with reduced temporal resolution.

H₂O₂ measurements with all three approaches were conducted in parallel using MCs from the same blood donors. While the electrochemical recordings with the ElProScan were continuous (recording of 1 Hz) and always from one single MC, ESR- and AUR-based H₂O₂ determinations reflect values at selected time points, representing about 25,000 MCs per point. When stimulated with TPA, cells in suspension as well as the electrochemically measured single cells responded with H₂O₂ production, revealing similar kinetics (black traces in Fig. 7A, D, G), which are not well time resolved in case of ESR and AUR. Application of H₂O₂ (10 μM) following TPA stimulation caused the signals to increase and subsequently to decline (red traces in Fig. 7A, D, G). The drop of the signals was triggered by H₂O₂ degradation processes of MCs and not by spontaneous chemical breakdown of H₂O₂ because cell-free experiments depicted in Figure 7 (red traces in C, F, I) revealed that [H₂O₂] was quite stable over 70 min under the experimental conditions used here.

Similar to TPA-stimulated MCs, unstimulated cells efficiently removed external H₂O₂ from the buffer (Fig. 7, red traces in B, E, H). The single-cell responses (Fig. 7: compare A and B and also Fig. 6A) demonstrate the individual characteristics of selected cells.

Addition of H₂O₂ to single cells or to MCs in suspension initially caused signal peaks, which, however, never reached the height of the peaks in the absence of cells. This was true for all three approaches (Fig. 7: compare A and B with C; D and E with F; and G and H with I). Two reasons may contribute to explain this effect: cells function as very effective sinks to absorb H₂O₂ and degrade it intracellularly, and in addition, H₂O₂-degrading enzymes (such as catalase) may be active in the extracellular space. Together, this explains reduced extracellular [H₂O₂] compared with cell-free buffer solution. Addition of catalase at the end of the cell-free AUR experiments completely blocked H₂O₂ signals, thereby confirming the specificity of the method (Fig. 7I).

In conclusion, cell population experiments validated the electrochemical single-cell measurements.

Discussion

Comprehensive understanding of inter-related physiological processes of auto- and paracrine H₂O₂ signaling (14, 21) requires real-time quantification of extracellular local [H₂O₂] kinetics originating from a single cell as presented here.

In recent studies, H₂O₂ has been predominantly determined in the extracellular space of cell populations with the fluorescence-based Amplex Red/Amplex UltraRed assays (25, 28, 41, 55, 62). Although it is possible to monitor H₂O₂ quite selectively in the subnanomolar range (47) and with some dynamics when using single time point determination (compare Fig. 7), the method has significant disadvantages (24, 68, 72). Most importantly, single-cell measurements and high temporal resolution are impossible and quantification is difficult.

Inside cells, genetically encoded fluorescent sensors such as HyPer variants (12, 13, 20, 57, 69), Orp1-roGFP2 (41, 48, 58), and roGFP2-Tsa2ΔCPΔCR (49) allow determinations of very low [H₂O₂], in part, with subcellular resolution. However, several problems are difficult to resolve: (i) Light can induce ROS generation. (ii) For measurement of [H₂O₂] dynamics, oxidized sensors (by H₂O₂) have to be reduced (recycled), for example, by intracellular glutathione and thioredoxins. Activities of these reductant mechanisms vary between cells and compartments and additionally depend on the metabolic state. Thus, precise quantification of [H₂O₂] would require quantification of endogenous redox capabilities of individual cells. (iii) Sensors may change their properties inside cells compared with calibration conditions outside. (iv) Extracellular application of defined [H₂O₂] is not appropriate for intracellular calibration because H₂O₂ transport into cells is highly regulated and can establish outside:inside gradients varying from 7:1 up to 650:1 (5, 27). In addition, H₂O₂ is effectively degraded in- and outside of cells [compare Figs. 6 and 7 and (66)]. (v) It cannot be excluded that transfection of cells using any vector may disturb physiological functioning of the cell, even when not directly visible.

For electrochemical H₂O₂ determination in the extracellular space of single cells, some excellent studies have been conducted using either nanotube arrays (31) or UMEs in the bare state or coated with different materials. A scanning electron microscope, very similar to our ElProScan, was used in these studies to monitor ROS and reactive nitrogen species (RNS) from single cells [e.g., (2, 3)]. Applying different voltammetric and amperometric methods, including biological scanning electrochemical microscopy, currents elicited by low [H₂O₂] have been measured with considerable temporal and spatial resolution (10, 30, 42, 56). However, quantification of real-time dynamic changes in [H₂O₂] due to physiologically relevant production and degradation by single primary cells of human origin (ex vivo) has not been performed.

Combining a bare disk platinum UME with different voltammetric and amperometric techniques, we found chronoamperometry to be the method of choice to achieve the following goals: specificity for H₂O₂, quantification of dynamic [H₂O₂] changes in the physiologically relevant nanomolar range and stable (without UME fouling) long-term monitoring of H₂O₂ production and removal in the extracellular space right at the plasma membrane of a single cell. Our local measurements report real-time [H₂O₂] changes above the intact whole cell body and can be easily calibrated for each measuring point at a temporal resolution of at least 1 Hz. Furthermore, production of H₂O₂ by a single human MC can be separated from concomitant degradation of H₂O₂ by the same cell.

Covering three up to six orders of magnitude of [H₂O₂], our electrochemical methods include the full physiological concentration range compared with usually only one order of magnitude reported for the genetically encoded fluorescent sensors. Chronoamperometry (CA) offers the best temporal resolution (≥1 Hz) over several hours, whereas with SWV, the best sensitivity can be achieved down to 5 nM [H₂O₂]. With the here applied electrochemical techniques, [H₂O₂] changes over time are determined about 5–8 μm above a single cell as a change in current (ΔI₆₅₀) within the active volume around the UME tip. These local changes are not detectable tens of micrometers distant from the cell in the bulk solution. Therefore, increasing and decreasing current values at the UME tip directly reflect local [H₂O₂] changes caused by a single cell. Using CA to measure single human MCs, we determined an average local [H₂O₂] increase of 3.4 nM per second in the first ( = maximal) H₂O₂ production phase after TPA stimulation. If concomitant H₂O₂ degradation was taken into account, local [H₂O₂] increases as high as 10 nM per second were estimated.

In addition to single cells, the technology is, in principle, also applicable for extracellular [H₂O₂] recordings in intact tissues or isolated organs if accessible by the electrode.

However, there are also disadvantages of the electrochemical H₂O₂ measurements presented here: only the least sensitive CA method provides a sufficient time resolution ≥1 Hz. While the more sensitive SWV, CV, and LSV can resolve [H₂O₂] as low as 5 or 50 nM, respectively, their time resolution is more limited. In addition, SWV experiments, when conducted over ≥1 h, caused marked desensitization and finally the ruin of the electrode. Whether precoverage of the platinum wire with protective material, as described by other groups (2, 8) or other applications, could avoid these problems has to be elaborated in future. Finally, scanning electrochemical microscopy, as presented here, enables only measurements of extracellular H₂O₂, which could, however, be considered an advantage in case effects of H₂O₂ on receptors, ion channels (15), or intercellular signaling are to be studied. It is, however, a disadvantage if intracellular [H₂O₂] measurements are required. Already in 2012, Wang et al. (67) described an electrochemical technique to determine ROS and RNS also inside cultured cells with nanoelectrodes, a method that was also applied more recently by Actis et al., 2014 (1). However, our UME of 10 μm ø cannot be used for intracellular measurements because it is too large to be impaled into MCs or cells of comparable size.

While temporal resolution of CA is very high, spatial resolution exemplarily shown in 2D scannings of single MCs (Figs. 3, 5, and 6) is in the low micrometer range and thus not as good as for genetically encoded fluorescent sensors, which are only limited by optical resolution.

To understand the redox physiology in more detail, especially in immune cells, a variety of ROS and RNS (such as ONOO⁻) as well as other metabolites (such as H₂S) have to be considered in future (4). Further development of the setup and measuring protocols are necessary to electrochemically determine the abovementioned species with similar standard as for H₂O₂.

In summary, SWV, CV, LSV, and CA allow quantification of extracellular local [H₂O₂] kinetics originating from single cells. SWV offers the highest sensitivity, but is compromised by relatively low temporal resolution and irreversible UME destruction, which exclude reliable cell measurements over extended times. CV and LSV have a slightly reduced sensitivity and also a relatively low temporal resolution, but due to electrode cleaning procedures, UMEs can be used over extended times. In contrast to voltammetric methods, CA shows the lowest sensitivity for H₂O₂. However, it is sufficient for quantification of physiologically relevant extracellular local [H₂O₂] kinetics originating from single cells. In addition, CA offers stable long-term measurements with the best temporal resolution.

Materials and Methods

Materials were purchased from ^ASigma-Aldrich; ^BGibco; ^CLife Technologies/Molecular Probes; ^DNoxygen; ^EPromocell; ^FMerck; and ^GBecton-Dickinson BD (all in Germany).

Solutions

Dulbecco's PBS-I^A (PBS-I), containing (mM) NaCl: 136.9; KCl: 2.68; CaCl₂: 0.905; MgCl₂: 0.49; KH₂PO₄: 1.47; and Na₂HPO₄: 8.1/enriched with glucose, 5 mM, pH 7.4 ± 0.03, 288 ± 1 mOsm; LB Medium^G (for culture of E. coli) containing (g/L) tryptone: 10; NaCl: 10; and yeast extract: 5; and RPMI 1640-Medium^B (+ 10% FCS +1% antibiotics pen/strep) were used.

Chemicals

Agarose^A, Amplex UltraRed Reagent^C (dissolved in DMSO), catalase^A (from bovine liver, 2000–5000 U/mg protein, freshly dissolved in PBS-I), CMH^D (1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine hydrochloride; dissolved in anoxic A. dest. +25 μM DF +5 μM DETC), DETC^D (Na-diethyldithiocarbamate, dissolved in A. dest.), DF^D (deferoxamine methanesulfonate salt, dissolved in A. dest.), DMSO^A (dimethyl sulfoxide), FCS^B (fetal calf serum), ferrocene methanol^A (1 mM in 100 mM KCl), glucose^F, HRP^A (horseradish peroxidase, Type VI-A, dissolved in A. dest.), and hydrogen peroxide^A (30% w/w H₂O₂ stock solution; 9.8 M at 25°C; controlled by titration with KMnO₄) were used. From this stock, aliquots diluted with A. dest. were freshly prepared and stored in dark vials. KCl, KMnO₄^A, pen/strep^B (penicillin 10 kU/mL/streptomycin 10 mg/mL), SOD^A (superoxide dismutase, bovine, dissolved in PBS-I), TEMPOL^A (4-hydroxy-TEMPO = 4-hydroxy-2,2,6,6-tetramethylpiperidine 1-oxyl; chemical oxidant; dissolved in A. dest.), TPA^A (12-O-tetradecanoylphorbol-13-acetate, dissolved in DMSO), and VAS2870^A (dissolved in DMSO) were used.

Other materials

Ultralow Attachment Surface Plates (U-LASPs, Corning/Costar) were used to keep cells in suspension. Dish used for electrochemical measurements was FluoroDish, sterile and with 35.5 mm × 10 mm/23 mm glass bottom (WPI). UMEs (disk platinum electrode, 10 μm ø) and electrode polishing device MHK1 (HEKA Elektronik Dr. Schulze GmbH) were used. Glass capillaries for ESR measurements were Ringcaps Pyrex (Hirschmann). Microplates for AUR measurements were black, 96-well, and with clear flat bottom (BD Falcon).

Preparation and culture of human MCs

The present research carried out with ex vivo leukocytes from anonymized healthy human blood donors (male/female) has been authorized by the local ethics committee in the declaration from April 16th, 2015 (84/15; Prof. Dr. Rettig-Stürmer). Out of use leukocyte-reducing system chambers were kindly provided by the local blood bank in the Department of Haemostaseology at Saarland University.

For preparation of human MCs, peripheral blood mononuclear cells were isolated, from which the MC population was separated following a selective adhesion protocol (57). MCs were cultured in RPMI medium (+ antibiotics) suspended in U-LASPs at 37°C in a humidified 5% CO₂ atmosphere for up to 25 days. For electrochemical experiments, cells (in RPMI medium) were plated at a density of 10⁵ to 5 × 10⁵ cells per dish at least 1 day before use and kept under culture conditions. Cells attach to the glass bottom within 1 h and eventually grow and differentiate to macrophages. Immediately before starting the experiment, attached MCs were washed twice with PBS-I and were measured in 2 mL of PBS-I. Dissolved effector substances were pipetted into the measuring buffer, followed by gentle mixing.

Electrochemical H₂O₂ measurements with the electrochemical probe scanner (ElProScan) ELP 3

The ElProScan system (HEKA Elektronik Dr. Schulze GmbH) working as a scanning electrochemical microscope (SECM) consists of three main components: (i) the controller ESC 3 to control electrode and specimen positions; (ii) the amplifier EPC 10 USB (Rev Y) to generate voltage and measure current, connected to a 3-electrode head stage; and (iii) the positioning system with inverted microscope optics and CCD camera (ProgRes, CapturePro 2.8.8, MFcool; Jenoptik, Germany) for monochrome bright-field pictures.

The systems to drive specimen and UME positions, respectively, consist of a high-resolution 3-axis part (closed loop, linear optical encoder with 10-nm resolution; Applied Scientific Instrumentation) and an additional piezo translation stage on the z-axis (closed loop, linear optical encoder with 1.5-nm resolution; Physik Instrumente GmbH & Co. KG).

The inverted microscope optics is equipped with a tube lens (200 mm focus length), a motorized focus control (closed-loop rotary encoders with 5.5-nm resolution, bidirectional repeatability, SD <250 nm; Applied Scientific Instrumentation), and two objectives mounted on a nosepiece revolver: LUCPLFLN 40 × , 0.6 N.A., WD 2.7-4 mm and PLCN 4 × , 0.1 N.A. (both Olympus).

The equipment is mounted on a vibration-dampening table (TMC-Technical Manufacturing Corporation; sold by Qioptiq Photonics GmbH & Co. KG) and shielded in a Faraday cage (homemade), the front part of which is made up by a metal curtain (Ecologa Europe GmbH).

The 3-electrode setup consisted of the working (WE), the pseudoreference (RE; silver wire coated with AgCl), and the counter (CE; platinum wire) electrodes. For low-current measurements in the fA range, as needed for single-cell experiments, the 2-electrode mode (WE + RE) turned out to be more favorable than the 3-electrode mode (+CE). An interface of polymerized agarose (5%) in KCl (3 M) to bridge the RE to the measuring buffer was tested: neither did it reduce current noise levels nor did it affect stability of long-time experiments (and was therefore not used).

A disk UME with a bare platinum wire of 10 μm ø and RG values of 1.6 - 3 [nomenclature for UME (7)] served as WE. RG is defined as the ratio between the insulator thickness (glass) and the radius of the platinum wire (here 5 μm). Electrochemical quality of a UME was routinely tested before starting a measurement. For that, cyclic voltammograms (CVs) in a potential window of -100 up to +400 mV (scan rate 50 mV/s) were recorded in ferrocene methanol (1 mM in KCl, 100 mM; vs. Ag/AgCl) solution. A UME was considered to be of high quality if a stable CV of low-noise current and a sigmoidal shape without significant hysteresis, but with long plateau phases at negative and higher positive potentials, were achieved (18, 33). Organic material adhering at the tip or other contaminations of the surface as well as partial damage of the tip are reflected by a worsening of the CV. Ultrasonic (for ≤10 s in ethanol, 70% followed by A. dest.; ultrasonic device Sanitas; Hans Dinslage GmbH) or mechanical cleaning of the tip with the MHK1 electrode polisher frequently helped to improve the properties of the UME. For preconditioning of the UME before starting the measurement (and thereby shortening the run-in period) cleaning pulses were performed, which consisted of three repetitive potential steps (for 1 s, each) from 0 to +200 mV, then to -500 mV, and back to +200 mV.

Measurements were conducted at room temperature (RT) under constant white light and ambient air. Photographs of measured cells were routinely stored together with the experimental data. Single MCs attached to the bottom of the dish were approached by the electrode from above. The electrode positioning system of the setup measures the exact z-position of the UME tip with a resolution of <1 μm. The vertical distance between the tip and the cell is measured by the focus levels in the following way: after having approached the electrode about 20–30 μm above the cell, the height of the UME tip in focus and then the upper cell surface in focus are determined. Thereafter, the UME is slowly lowered and fixed 5–8 μm above the cell. Because of the irregular surface (microvilli etc.) of the cell, it is not possible to give a more precise value.

With an external standard device, the instrumental setup was routinely adjusted for amplification, capacitance, filters, and other parameters (defined in the ElProScan manual).

For data acquisition and analysis, the PATCHMASTER software (versions 2x74 and 2x90.2) with ELPROSCAN and IMAGING extensions (HEKA) was used. The system allows amperometric 2D line and matrix scanning of a probe following a freely programmable protocol. Scanning electrochemical microscopy (SECM) uses chronoamperometric (at 650 mV) constant height 2D scans of steady velocity with a speed of 1-10 μm/s in alternate mode (meander scan). Under these conditions, the position of the UME is fixed and the table with the cell dish is moved. According to a predefined setting, a scan starts in the lower right corner of the selected scan area, and the current is measured every x to x+Δx nm while scanning along the x-axis; scan lines are separated by Δy nm in the y-direction.

The usual instrumental setting for chronoamperometry (CA) was current gain 50 mV/pA; filter (I) 15 Hz Bessel; filter (II) 100 Hz Bessel; and current sampling frequency was 1 kHz. Each second, 1000 sampled values were averaged, resulting in recordings with a time resolution of 1 Hz. For transient voltammetric techniques such as CV, LSV, and SWV, parameters were adapted as described for the respective experiments. CA measurements of H₂O₂ were performed at 650 mV; in CV and LSV, currents were sampled in selected potential windows and with varying scan rates, depending on the aim of the experiment. For SWV experiments, parameters were set as follows: start and end potential of -100 mV and +900 mV; amplitude of the steps was ±75 mV and duration of one step pair was τ = 2 s. At a scan rate of 5 mV/s, a frequency of 0.5 Hz, and a potential increment of 10 mV on top of each positively directed step, current values were sampled from the last 10% of the plateau phase of each up- and downward peak. SWV experiments were analyzed in two alternative ways: after having defined the baseline of a peak curve of I_difference (I_difference = I_forward - I_reverse; see Fig. 2B), integration of the area of the curve or, alternatively, Δcurrent (baseline to peak) was used to calibrate SWV (compare Fig. 2C, D). Both methods gave similar results. However, definition of the baseline is problematic when trace curves are not clearly developed (especially at low [H₂O₂]) or when more than one peak appears.

In SWV experiments, so far unused UMEs showed high sensitivity for low nanomolar [H₂O₂]; however, they lost sensitivity during continued measuring, irrespective of whether or not H₂O₂ was added. For higher [H₂O₂] of 50 μM up to 1 mM, continued loss in sensitivity did not impair the measurements. Decline of UME sensitivity for H₂O₂ with time, even under repetitive electrochemical cleaning, made all attempts impossible to calibrate SWV measurements. Obviously, adsorption of material at the surface of the UME tip and modifications of the platinum wire in even deeper layers are of particular complexity when SWV is used (53). Thus, restoration of a desensitized UME was not possible with ultrasonic or other established cleaning methods. Occasionally, after polishing with abrasive paper (using the MHK-1 device), the electrode was usable again.

UME polarization, limits of detection and quantification

Upon applying any potential between the electrodes in PBS-I measuring buffer, polarization (2) of the UME along the so-called run-in period can be observed. Components of the buffer adsorb at the UME tip, which modify the active area of the UME and may influence charge transfer. Upon starting CA measurements at 650 mV, the run-in phase characterized by decreasing current values lasts 5–20 min before a steady-state current flow is approached. The time required for current stabilization differs for each individual UME and depends on its preconditioning. In 72 different CA experiments (with 10 different UMEs), current values (I₆₅₀) of the respective approximate linear steady-state phases (averaged over 5 min) were calculated to be 507.2 ± 22.5 fA (mean ± SEM; SD = 190.8 fA). The limits of detection and of quantification, both calculated on the basis of standard deviation of the steady-state blank current (I_blank = 507.2 ± 190.8 fA), are 572.4 fA (3 × SD_blank) and 954 fA (5 × SD_blank), respectively (61).

Starting potential effect on sensitivity of voltammetric H₂O₂ determinations

Along with calibration experiments for H₂O₂ using CV, LSV, or SWV, it turned out that the starting potential was of crucial importance for the sensitivity of the UME. End potential was always +1.0 V. The current, sampled at 650 mV (I₆₅₀; where anodic oxidation of H₂O₂ occurs), gradually increased as the starting potential (beginning at +30 mV) moved into the negative direction. The effect reached its optimal plateau at potentials between -50 mV and -100 mV. The same was true for LSV experiments. In LSV, however, after jumping back from the end to the starting potential, the latter had to be clamped for at least 2 s before starting a new sweep. Obviously, within the above described critical potential interval, effective electrochemical cleaning of the electrode surface occurs due to desorption phenomena, thus yielding enhanced sensitivity toward H₂O₂ detection. As a consequence, for voltammetric experiments such as CV, LSV, and SWV, the starting potential was set to -100 mV (compare Figs. 1A, B, and 2B).

Generation of H₂O₂ at the UME along the ORR

As shown in Figure 1, voltammetric (CV, LSV) calibration experiments for H₂O₂ could be complicated when starting from -1.0 V. Current values (I₆₅₀) were even higher than expected for the deliberately added H₂O₂ to the buffer solution (black lines in Fig. 1D, E). At potentials more negative than -100 mV, the electrode reduction of dissolved oxygen is forming measurable amounts of H₂O₂ (due to O₂ + 2H⁺ + 2e⁻ → H₂O₂) (22, 59, 75). In an unstirred solution, this increases [H₂O₂] near the electrode above average values in the bulk. Consequently, current values (I₆₅₀) rise over time in the course of a repetitive CV experiment within the potential window from -1.0 V to +1.0 V. Adding H₂O₂ (10 μM) to the buffer resulted in a further increase of current. Conducting similar CV experiments in the presence of the enzyme catalase (∼400 U/mL), which causes dismutation of H₂O₂, no increase of current was seen neither before nor after extra application of H₂O₂. If those long-term CV measurements, however, were started at a potential of -100 mV (up to +1.0 V), no H₂O₂ production was measurable. Upon addition of H₂O₂, however, the expected current jump was observed (not shown).

Although oxygen can be reduced at Pt electrodes (vs. Ag/AgCl; neutral pH) at negative potentials starting from 0 V already (unpublished observations), this effect was not visible in our CV experiments. We suggest the following explanations for this discrepancy: The amount of H₂O₂ being produced at the slightly negative potential range between 0 and -100 mV cannot be high since this range is quickly changed toward positive potentials in CV. In addition, for production (at −100 mV) and measurement (at 650 mV) of H₂O₂ within one CV, the same UME is used. For measurement, H₂O₂ is oxidized at the electrode, thus preventing accumulation of H₂O₂ at the tip (59). Furthermore, due to the duration of one CV (at 50 mV/s) of >20 s, parts of H₂O₂ produced at the UME may have diffused away. Fundamental details about ORR are provided elsewhere (29).

Nevertheless, our results support the conclusion that at -100 mV, repetitive electrochemical cleaning of the electrode surface in CV (and LSV) is the reason for increased sensitivity rather than electrochemical formation of additional H₂O₂ due to ORR. The cleaning pulses for preconditioning of the UME, which are mentioned above, hint in the same direction.

H₂O₂ measurements with ESR spectroscopy

An ESP300 spectrometer (Bruker) equipped with a TMH8603 cavity was used. ESR samples were transferred to 50-μl glass capillaries and measured at RT. For H₂O₂ determinations in MC suspensions, the hydroxylamine spin probe CMH was used, which served as an electron donor for the horseradish peroxidase-catalyzed reduction of H₂O₂, forming two CM radicals accessible by ESR (63). For the measurements, suspended MCs (≥5 days after isolation) were washed in PBS-I, followed by centrifugation (RT, 220 g, 10 min). Then, four samples, each containing about 0.5 Mio cells/mL PBS-I, were differently treated following the protocol described in Figure 7. At selected time points, a 150-μL aliquot of each sample was taken and cells were spun down for 1 min at 14,200 g. From the supernatant, 50 μL was added to 50 μL measuring solution containing horseradish peroxidase (HRP) and CMH at final concentrations of 100 U/mL and 1 mM, respectively. Fifty microliters of this mixture was transferred to a glass capillary, and ESR spectra were measured. ESR parameters were set to a microwave power of 20 mW, a frequency of 9.5 GHz, a magnetic flux density of 0.3445 T, a sweep width of 6 mT, and modulation amplitude of 0.1 mT. The duration of one scan was 1 min with a time constant of 20 ms. TEMPOL (100 μM) served as a reference substance. The peak-to-peak height of the signals is proportional to the CM-radical concentration and hence to the [H₂O₂]. To ensure that reduction of H₂O₂ by HRP was complete at each time point of its determination, follow-up scans were checked for termination of the reaction. Experiments without cells (Fig. 7F) were conducted in the same way. Additional control experiments in the presence and absence of HRP, respectively, confirmed that exclusively H₂O₂, but not ^●O₂⁻, was measured. For calculations, the intensities of the ESR spectra were analyzed.

H₂O₂ measurements with the AUR assay

Measurements were conducted at RT in a microplate reader (Infinite M 200 PRO; Tecan, Germany). For the fluorometric determination of extracellular H₂O₂, MCs were used ≥5 days after isolation. Oxidation of the colorless nonfluorescent AUR to a fluorescent resorufin derivative by H₂O₂ is not reversible, thus allowing only endpoint determinations of [H₂O₂] in the solution. To follow the kinetics of the H₂O₂-producing and -degrading pathways of cells, the recommended measuring protocol for AUR was modified [(47) with description below]. MCs were prepared and treated as for ESR measurements and following the protocol described in Figure 7. At selected time points, 50 μL (in duplicate) of each sample (including about 25,000 cells) was added to 50 μL of measuring solution containing AUR and HRP at final concentrations of 30 μM and 0.1 U/mL, respectively (light-sensitive!). After having collected all samples of one respective time point in the microplate, it was immediately measured (bottom reading; gain 55; 9 reads per well; excitation at 535 nm/emission at 590 nm). The same procedure was applied for each time point of treatment as well as for the cell-free control samples.

Presentation and analyses of data; statistics

Cyclic and linear scan voltammogram presentations follow the IUPAC voltammogram convention with positive potential and anodic current in the positive x-axis and positive y-axis, respectively. For analysis of the experimental raw data, the standard accessory PATCHMASTER software (versions 2x74 and 2x90.2) of the measuring system was used. Further processing of data, statistics, and graphical presentations were carried out using OriginPro 8.5G and CorelDRAW X4. Values are mean ± SEM (or ± SD if explicitly stated).

Abbreviations Used

AUR

Amplex^® UltraRed

CA

chronoamperometry

CMH

1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidinehydrochloride

CV

cyclic voltammetry, cyclic voltammogram

DMSO

dimethyl sulfoxide

ElProScan

electrochemical probe scanner

ESR

electron spin resonance

[H₂O₂]

local concentration of hydrogen peroxide

HRP

horseradish peroxidase

LSV

linear scan voltammetry, linear scan voltammogram

MPh

macrophage

MC

monocyte

NO^●

nitric oxide

NOX

NADPH oxidase

^●O₂⁻

superoxide

ORR

oxygen reduction reaction

RNS

reactive nitrogen species

ROS

reactive oxygen species

SECM

scanning electrochemical microscope (microscopy)

SOD

superoxide dismutase

SWV

square wave voltammetry

TPA

12-O-tetradecanoylphorbol-13-acetate

U-LASP

ultralow attachment surface plate

UME

ultramicroelectrode

Acknowledgments

The authors thank Stephanie Saul, Carmen Hässig, and Cora Hoxha for preparation of the cells. For discussions in the electrochemical field, the authors are grateful to Rubin Gulaboski, Nguyen Nhat Bui, and Katerina Stankoska. The material submitted for publication has not been previously reported and is not under consideration for publication elsewhere. The authors acknowledge funding by the Deutsche Forschungsgemeinschaft: SFB 894 (project A1), SFB 1027 (projects A2 and C4), and SPP 1710 (projects HO2190/4-1 and BO3643/3-1). In addition, the BMBF project ZIM KF2152005AK1 is gratefully acknowledged.

Author Disclosure Statement

C.H. is employed at HEKA Elektronik Dr. Schulze GmbH, 67466 Lambrecht, Germany. All other authors declare no conflicting financial interests.