Influence of applied currents on the viability of cells close to microelectrodes

Abstract

Electrodes have become more and more popular in biomedical and bioengineering applications, where they are used mostly to apply/measure potentials or currents to/from biological systems. Under such conditions, electrochemical reactions commonly occur at the electrode surface. With the aim to better describe these processes we applied constant currents using transparent indium tin oxide microelectrodes to induce a local change in pH, associated with electrolysis. The pH change was monitored optically within the first lateral 170 μm vicinity using microscopy and a pH sensitive fluorescent dye combination as indicator. The data were then fitted with a simple diffusion model. The effect of such an induced pH change was also assessed by measuring the desorption of a cationic polyelectrolyte (poly(L-lysine)-grafted-poly(ethylene glycol)) as a function of the local pH. Since this polymer interacts electrostatically with surfaces in a pH dependent manner, we could show a strong effect in unbuffered solutions while buffered solutions restricted the electrochemically induced pH change below the optical resolution of the microscope. The effect of applied current on the behavior of cells was also studied on myoblasts cultured directly on the microelectrodes. We have found that current densities larger than 0.57 A m⁻² induced cell death within 2 min of exposure. Based on our model we could attribute this to the change in local pH although the effect of other electrochemically created reactive molecules could not be excluded.

---

Insight, innovation, integration

Electrodes in biology or biomedicine are used to apply/measure potentials or currents to/from biological systems such as skin, muscle or single cells. It is known that electrochemical reactions occur at the electrode surface but very little is known about the pH values close to the electrode and their effect on tissue or single cells. We measured the extent of the pH change optically in the micrometre range and fitted the data to a simple diffusion model, which allowed an estimation of the pH value between the electrode and a living cell grown on it. Furthermore, the effect of this local pH change was assessed with living cells and the critical current density was established.

Introduction

Electrodes have long been used in electrophysiology¹ for recording and stimulation of cells or tissues. The most advanced systems nowadays are auditory and visual implants^2,3 which stimulate the neurons of the nervous system directly.

Other devices are used in neurorehabilitation for neuromuscular electrical stimulation⁴ or arrays of microelectrodes to study neuronal activity.⁵ Bioelectrodes are used to apply drugs through the skin by electroporation.⁶ Improved wound healing⁷ and tumor treatment⁸ have also been reported by direct electrical stimulation.

Electrically controlled transfection of cells has been shown by applying short pulsed voltages on conductive microelectrode arrays (MEA).^9,10

A growing field of bioelectrode applications in biotechnology is the dynamic control over the adsorption/desorption of polyelectrolytes^11–14 or polyelectrolyte multilayers.¹⁵ Especially interesting for such purposes are transparent electrodes made of non-toxic indium tin oxide (ITO). They are well suited for biological experiments where a parallel optical measurement or transmitted light microscopy is performed e.g. microarrays for electrophysiology,¹⁶ or cell sheet engineering with electronic control.¹⁷

In all these applications, a potential is applied in an aqueous environment and water electrolysis occurs close to the electrode surface.^18–20 Electrolysis of water is well-known, however, except for a theoretical attempt to describe short pulsed currents,²¹ we found no complete model that could predict the pH changes close to electrode surfaces.²² Nevertheless the effect itself is already used for thin film deposition²³ or peptide synthesis.²⁴

There were also attempts to measure and model the diffusion of the induced pH in gels²⁵ or to measure the pH change close to electrodes using pH sensitive beads.²⁶

In this paper we present a high-resolution method to measure the pH change due to electrochemical reactions close to an ITO electrode using confocal laser scanning microscopy. This further allowed us to study and model pH changes in physiological solutions where the extent of the effect is within the hundred micrometre range. Visualizing such local pH changes is highly interesting for thin film applications and electrochemistry. Based on the predictions of our model we were able to connect the electrochemically induced pH changes with the observed behavior of cultured myoblasts grown directly on the electrode surface. Current densities larger than 0.57 A m⁻² lead to propidium iodide uptake, indicating membrane pore formation or cell death, which was also seen when exposing cells to low pH conditions.

Materials and methods

Direct measurement of electrochemically induced pH change with fluorescent dyes

Electrolysis in aqueous solution. If a potential is applied between two electrodes in an aqueous solution, not only must the electrolysis of water be taken into account, but also the reaction of solvated molecules or ions (Na⁺, Cl⁻, etc.) in the solution.²⁷ When electrolysis of water occurs, a local pH change at the anode or cathode takes place and diffusion of the generated species is expected. (See also ESI.† )

NaCl is the most prevalent electrolyte besides other inorganic salts, amino acids and vitamins in cell culture media as well as in animal body fluids. The following experiments were all performed in physiological aqueous solution either with 150 mM NaCl added to lower the ohmic resistance or 107 mM NaCl already contained in the commercially available culture medium for cell experiments.

Production of transparent ITO microelectrodes. Common microscopy cover slides (0.17 mm thickness) were sputter-coated with a 50 nm ITO layer at the Institute of Microtechnology, University of Neuchâtel, Switzerland. Electrodes were then produced (by a common photolithography technique) with a positive photoresist Shipley S1805 (Rohm & Haas, Germany) followed by wet etching of the excessive ITO in 3 M HCl for 20 s. The remaining photoresist was removed by immersion in pure acetone for 5 min, followed by rinsing in isopropanol, ethanol and H₂O.

Flow cell design and electronic setup. The pH measurements were performed in a custom-built polycarbonate flow cell with a chamber size of 18 × 12 × 0.6 mm (l, w, h) (Fig. 1A). The bottom part was sealed by rubber O-rings pressing against the substrate. The ITO electrodes were connected to the potentiostat by spring contacts outside the O-ring seal. All experiments were carried out under constant current controlled by an Autolab potentiostat PGSTAT302N (Metrohm Ag, Switzerland) with a 3-electrode setup, whereby the ITO was set as the working electrode (WE), a platinum wire served as the counter electrode (CE) and a chlorinated silver wire as the reference electrode (RE).

Fig. 1 Schematics of the flow cells used for pH and desorption measurements. The transparent substrate contains the indium tin oxide working electrode (WE) while the reference (RE) and counter electrodes (CE) are integrated in the top part. (A) Direct pH measurement: the flow cell is filled with a pH indicator solution and the focus of the CLSM is set to 10 μm above the substrate surface. Changes in fluorescence intensity provide information about the local pH. (B) Indirect pH measurement: the substrate is fully coated with fluorescently labeled PLL-g-PEG (red). The CLSM focus is in-plane with the PLL-g-PEG monolayer. The measured fluorescent PLL-g-PEG desorption indicates the local pH changes.

Preparation of pH indicator containing solutions. The pH sensitivity of fluorescent dyes is known and can be used to visualize local pH changes.²⁸ The following protocol has been optimized for measurements in the range between pH 2–7. An indicator stock solution was made from 20 μmol 5(6)-carboxyeosin and 20 μmol fluorescein sodium salt in a mixture of 500 μl H₂O and 500 μl dimethyl sulfoxide, all purchased from Sigma-Aldrich (Switzerland). The stock solution was diluted 1 : 2000 to a final concentration of 10 μM carboxyeosin and 10 μM fluorescein for all measurements. The electrochemically induced pH change was measured only in H₂O, whereby the ohmic resistance was lowered by adding NaCl to a final concentration of 150 mM. All solutions were injected into the flow cell using syringes.

Phosphate citrate buffered pH standard solutions were made according to the mixing table proposed by McIlvaine²⁹ in order to obtain a standard, pH vs. fluorescence intensity curve. (See ESI.† )

Optical pH measurement with confocal laser scanning microscopy. A Zeiss LSM 510 confocal laser scanning microscope (CLSM) with a 63×/1.4 NA oil M27 plan-apochromat objective was used for all measurements. The dyes were excited simultaneously at 488 and 514 nm and their fluorescence emission was detected using a band-pass filter 530–600 nm.

The flow cell was then filled with water containing the indicator mixture and 150 mM NaCl at pH 6.6. Then a current of 4 A m⁻² was applied to the electrode while imaging the xy-plane with the CLSM (Fig. 1A). The measured fluorescence intensity of the indicator could then be related, using a standard curve (ESI† ), to the local pH with a temporal resolution of 250 ms per scanned line.

Modeling. The used ITO electrode had a length of 8 mm and width of only 30 μm and as such can be considered as a line source in an infinite volume. Hence, the perpendicular cross section of this symmetric system is equivalent to a point source in an infinite plane. Eqn (1) gives the concentration at a distance r from a point source.³⁰ In our experiments a constant current was applied, producing diffusing substances continuously in a semi-infinite plane due to the perpetual electrochemical reactions on the electrode surface. Such an experimental system can be described by eqn (2), which is deduced by integrating eqn (1) with respect to time and replacing the amount of diffusing substance M by a flux Φ.

C(r,t): [mol m⁻³] is the concentration of substance

M: [mol m⁻¹] amount of diffusing substance deposited initially at the point source

D: [m² s⁻¹] diffusion coefficient of the substance in the solution

r: [m] distance from the point source

t: [s] time

C ₀, C₁: [mol m⁻³] is the initial concentration of substance

Φ: [mol s⁻¹ m⁻¹] flux of substance per unit length for a semi-infinite plane

I: [A] current

F: [C mol⁻¹] Faraday constant

l: [m] length of the electrode

E ₁: exponential integral

The measured pH values (Fig. 2B) correspond to the concentration C(r,t) in eqn (2). We fitted all curves with the same parameters I and D at different time points (t) and distances from the electrode (r) by a least squares method to obtain the parameters I_fit and D_fit.

Fig. 2 (A) Confocal laser scanning microscopy image of the solution adjacent to the ITO electrode is shown. The focus was set to 10 μm above the surface. About 8 s after the scanning of the xy plane was initiated, a current of 4 A m⁻² was applied and the progression of the electrically induced pH change was monitored by the darkening of the indicator dyes. (B) The corresponding pH progression (symbols) in H₂Ovs. electrode distance in steps of 1.5 s along with the fitted data (lines). The scaling along the x-axis is equivalent for A and B.

Indirect measurement of electrochemically induced pH change with an indicator molecule on the surface

In this experiment an electrostatically adsorbed polyelectrolyte is used as an indicator for the pH change in the vicinity of the electrode. 0.1 mg ml⁻¹ rhodamine labeled poly-(L-lysine)-grafted-poly(ethylene glycol) (PLL-g-PEG) (Surface Solutions, Switzerland) was dissolved in different concentrations of McIlvaine buffer solutions pH 6.6 1 : 1 (145.5 mM Na₂HPO₄, 27.3 mM citric acid, 150 mM NaCl) and 1 : 10 (14.55 mM Na₂HPO₄, 2.73 mM citric acid, 150 mM NaCl), respectively. The substrate and ITO electrodes were cleaned with pure ethanol, rinsed with H₂O and blow dried with N₂. The flow cell described above was assembled and O₂ plasma treated for 2 min (Harrick Plasma, USA). The labeled polyelectrolyte solution was then injected immediately into the chamber and incubated for 30 min followed by rinsing with the buffer solution selected for the measurement. The rhodamine labeled PLL-g-PEG was excited with a 561 nm laser and detected using a 575 nm long pass filter.

Data processing. A rectangular reference region was photobleached in the rhodamine labeled PLL-g-PEG monolayer (Fig. 3A) and an image was taken before applying the current. A second image was taken after the current I = 4 A m⁻² was applied for 30 s. The relative intensities of the PLL-g-PEG monolayer were compared between these two images (by subtracting the average intensity value of the bleached area from the average intensity of 100 lines in the y-direction) (Fig. 3B). The experiment was performed with different media: H₂O, McIlvaine buffer 1 : 10 and McIlvaine buffer 1 : 1 for testing the influence of the buffer concentration.

Fig. 3 (A) Rhodamine labeled PLL-g-PEG monolayer with a bleached region as reference before (top) and after (bottom) a current of 4 A m⁻² was applied to the ITO working electrode (WE) for 30 s. (B) Amount of labeled PLL-g-PEG measured by the normalized fluorescence intensity as a function of distance from the electrode in H₂O, McIlvaine’s buffer 1 : 10 and McIlvaine’s buffer 1 : 1, respectively. (C) Modeled pH in H₂Ovs. distance to the electrode with the fitted parameters D_fit = 3.3 × 10⁻¹⁰ m² s⁻¹, I_fit = 0.18 A m⁻² and theoretical parameters D = 9 × 10⁻⁹ m² s⁻¹, I = 4 A m⁻² for the time-range between 10 and 30 s.

Effects of applied current to viable cells

The following experiments were conducted to evaluate the effect of an applied current on the behavior of cells grown directly on the electrode.

Cell culture. C2C12 myoblasts were cultured on MEA60 200 biochips consisting of ITO microelectrodes on glass purchased from Ayanda Biosystems^TM, Switzerland. The chip was rinsed with pure ethanol and MilliQ water, blow dried with N₂ and plasma treated for 2 min prior to the experiment. After cleaning, the chip was coated with 0.1 mg ml⁻¹ poly-L-lysine (wt > 300 kDa, Sigma-Aldrich, Switzerland) in phosphate buffered saline (PBS) for 30 min and rinsed with PBS. After trypsinizing myoblasts, the suspended cells were transferred to the chip and formed a confluent layer within 24 h. The myoblasts were cultured in culture medium containing Dulbecco modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin solution all purchased from Invitrogen (Switzerland). The experiments and the cell culturing were performed in the incubator at 37 °C and 7% CO₂.

Current experiments. The culture medium was exchanged after 1 day in vitro to either PBS (pH 7.4) (Invitrogen, Switzerland) or PBS with 2.5 μg propidium iodide (PBS–PI) (Promega, Switzerland). The MEA60 chip was mounted to an MEA1060 interface (without amplifiers) purchased from Multi Channel Systems, Germany. A simple platinum wire dipped into the solution served as a counter electrode and a chlorinated silver wire as a reference electrode. The whole setup was put in a cell culture incubator and connected to the potentiostat. Different currents up to 2 A m⁻² were applied for 2 min on 5 individual electrodes to cells in PBS or PBS–PI, respectively. (Currents higher than 8 A m⁻² resulted in electrode breakdown in our setup.)

Cells treated in PBS–PI remained in PBS–PI for 20 min before fixation in 4% formaldehyde in PBS. The cells in PBS had the chance to recover from the treatment by incubating in growth medium for 30 min, before the growth medium was changed to PBS–PI for 20 min followed by the fixation. In order to calculate the percentage of PI positive cells, all cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (Promega, Switzerland).

Results

Direct measurement of electrochemically induced pH change with fluorescent dyes

The electrochemically induced pH change close to the electrode was measured by relating the fluorescence intensity of an indicator solution to a pH value with a standard curve. In this section, the measured results and the corresponding fits are presented and compared to the theoretically expected pH.

Acquiring a typical CLSM image using our settings takes about 20 s by scanning the laser line by line from the top towards the bottom of the image. Thus Fig. 2A represents the time evolution since our sample was homogenous in the y direction (note secondary axis of Fig. 2A). The induced pH progression is then visualized by bright pixels indicating high pH and darker pixels indicating low pH regions. About 8 s after the scanning started (indicated by the arrow in Fig. 2A) a current of 4 A m⁻² was applied. The brightness values were converted to the corresponding pH values (Fig. 2B, symbols) using the standard curve described in the ESI.† Already 1.5 s after the current was applied the regions close to the electrode became darker immediately, indicating a local drop in the pH to 4.5 within 50 μm distance from the electrode due to electrochemical proton generation. The local pH at the electrode becomes stable afterwards and its value depends only on the applied current. Due to the diffusion of the H⁺ ions the pH change gradually extends into the solution. The slopes of the pH curves decrease with time, reaching after 6 s a quasi stable pH value, at least within a distance of 170 μm from the electrode (Fig. 2B). After 6 s the curves do not change significantly (data not shown).

A least square method was used to fit the presented five pH progression curves using eqn (2) with two uniform fitting parameters: current I_fit and diffusion coefficient D_fit. Although a good qualitative agreement was found, the obtained I_fit value was about 44 times smaller than the applied current I = 4 A m⁻² and D_fit about 27 times smaller than the theoretical value D = 9.3 × 10⁻⁹ m² s⁻¹³¹ for protons in water at infinite dilution.

Indirect measurement of electrochemically induced pH change with an indicator molecule on the surface

In the previous section, an indicator solution was used to measure the electrochemically induced pH change close to an electrode. In this section, a polyelectrolyte adsorbed on a SiO₂ surface is used as an indirect indicator for induced pH change. Fluorescently labeled PLL-g-PEG was adsorbed on the substrate as described above, then a current was applied for 30 s and the amount of desorbed PLL-g-PEG was measured with CLSM.

The substrate with the ITO electrode was coated with rhodamine labeled PLL-g-PEG (Fig. 1B) then a background reference area was bleached into the adsorbed monolayer. The same current density I = 4 A m⁻² as in the previous experiment was applied to the electrode. After 30 s a dark region close to the electrode was observed, indicating desorption of the electrostatically adsorbed PLL-g-PEG (Fig. 3A bottom) in the darkened region close to the electrode. The amount of PLL-g-PEG desorbed within 30 s gradually decreased from the electrode: from about 50% observed close to the electrode to no desorption at about 20 μm distance from the electrode in H₂O. 1 : 10 McIlvaine buffer reduced the extent of pH change to 5–10 μm, whereas no change in the image was observed in the undiluted McIlvaine buffer, indicating that the extent of detectable pH change is below the resolution of confocal laser scanning microscopy (Fig. 3B).

The pH in the vicinity of the electrode was calculated with eqn (2) and also with a finite element method (FEM) and analysis software (COMSOL, Multiphysics®) using the same parameters. Both methods showed a similar pH range in H₂Ovs. distance to the electrode, however, only the results of the FEM method are shown in Fig. 3C. In the previous section, we found a large difference between the expected³¹ and the fitted values for I and D. Therefore, we decided to use both parameter pairs to estimate the pH changes in this arrangement. The pH range reached in the time between 10–30 s is shown in Fig. 3C for the fitted parameters (grey) D_fit = 3.3 × 10⁻¹⁰ m² s⁻¹I_fit = 0.09 A m⁻² and the theoretical parameters (turquoise) D = 9.3 × 10⁻⁹ m² s⁻¹, I = 4 A m⁻². Despite the difference between the two parameter sets, the calculated pH values and slopes were very similar (ΔpH ~ 0.4) close to the electrode.

Effects of electrochemically induced pH change on viable cells

The previous sections have shown the possibility of measuring and modeling the pH change close to electrodes. The physiological effects of such a pH change close to viable cells were studied for myoblasts cultured directly on transparent electrodes. Different current densities were therefore applied to the electrodes for 2 min and the amount of dead cells was visualized by propidium iodide staining as described in the Materials and Methods section.

Fig. 4A₁–B₂ show differential interference contrast (DIC) microscopy images of a confluent myoblast layer on the microelectrodes (Ayanda MEA 60 chip) superimposed on the corresponding PI fluorescence signal (red). The ITO electrodes appear a bit darker than the underlying glass substrate and are labeled with the applied current density. During the current exposure the cells in Fig. 4A₁ and A₂ were immersed in PBS whereas the cells in Fig. 4B₁ and B₂ were immersed in PBS–PI. The cells immersed in PBS showed no substantial cell death up to the current density of 0.57 A m⁻², while the cells in PBS–PI already showed an increased amount (40%) of PI positive cells at this current. (See middle electrode in Fig. 4B₂.) Higher current density increased the amount of PI positive cells in both cases. The percentage of PI positive cells was calculated with respect to the DAPI stained nuclei and is plotted versus the applied current density in Fig. 4C.

Fig. 4 The effect of an applied current on the viability of C2C12 myoblasts. DIC microscopy images superimposed on the corresponding propidium iodide fluorescence signals: 0.57 A m⁻² did not result in substantial cell death in PBS (A₁) while about 40% of the cells stained positive in PBS–PI. (B₁) Higher currents resulted in an increase of the PI positive cells in both cases (A₂ and B₂). The quantitative assessment is summarized in plot C. (D) Magnified region of a typical cyclic voltammogram indicating the current range (red zone) where cell death occurs. (The inset shows the full CV scan.)

A voltammogram of the electrode versus an Ag|AgCl reference electrode is presented in Fig. 4D. The magnified plot shows the onset of the curve and the red zone indicates the current range where the cells were killed in larger amounts. The whole cyclic voltammogram is provided as an inset.

Discussion

Direct measurement of electrochemically induced pH change with fluorescent dyes

We overcame the difficulties in fast pH measurement on the small scale by using an indicator solution and the detection of its pH dependent fluorescence with confocal laser scanning microscopy. The indicator solution contained carboxyeosin and fluorescein. The two dyes have similar excitation and emission maxima but different pK_a values, so that their change in fluorescence intensity occurs at different pH values. Using both dyes together in the same solution, a broader pH range, pH 2–7, could be measured compared to using only a single dye. The experimental setup is visualized in Fig. 1A and shows the flow cell filled with the indicator solution. The protons diffuse immediately into the solution after the current is applied (Fig. 2A). The pH curves shown in Fig. 2B were all fitted with the same parameters at different times and distances with eqn (2), which delivered a single value for the current I_fit that is 44 times smaller than the current applied in the experiment. We attribute this unexpectedly low efficiency of the electrochemical reaction to other reactions occurring at the electrode: Since NaCl is present in the solution, Cl₂ is also produced at the anode. (See Reaction SIV in the ESI.† ) The Cl₂ dissociates into protons, chloride ions and hypochloric acid according to Reaction I but is only moderately soluble in H₂O.

Cl₂ + H₂O ⇌ H⁺ + Cl⁻ + HClO (I)

The equilibrium of Reaction I is pH dependent. At low pH the reaction towards the left side is more pronounced. Gas bubble formation has not been observed in our experiments with current densities ≤4 A m⁻², which is consistent with the reported gas bubble formation at current densities ≥10–50 A m⁻².²² Hence, the majority of the Cl₂ might be physically dissolved in the buffers and only a small amount dissociated to produce detectable protons. Another reason for the low current efficiency could be the relatively large amount of indicatordyes (40 μM) present in the solution acting itself as a buffer or the buffering effect of the inevitable dissolution of CO₂ from the environment. A 27 times smaller proton diffusion coefficient than expected was found by fitting the pH values in dilute unbuffered solution. The proton current must be therefore restricted by chemical reactions e.g. with the indicatordye. The fitted parameters (D_fit, I_fit) as well as the theoretical parameters (D_proton, I) were used to calculate the prevalent pH value close to the electrode. The calculated pH values showed only a marginal difference and so we believe our model describes the real pH value at the electrode with reasonably good (i.e. ~0.4 pH unit) accuracy.

The slope of the pH curve slowly decreases also after the 6 s time interval investigated; a stable pH value is reached due to the effect of OH⁻ ions diffusing from the counter electrode (ESI† , Reaction SI). The neutralization reaction between H⁺ and OH⁻ ions then leads to a quasi stable pH gradient between the two electrodes.

Indirect measurement of electrochemically induced pH change with an indicator molecule on the surface

To assess the local pH change the amount of a fluorescent polyelectrolyte adsorbed on a surface was measured using CLSM. The polyelectrolyte used, PLL-g-PEG, is highly cationic at physiological pH and adsorbs electrostatically onto negatively charged surfaces. The strength of its interaction with the surface depends on the pH or the ionic strength of the solution which makes the molecule an indirect indicator for pH changes. The PLL-g-PEG adsorption occurs between the isoelectric point of the SiO₂ surface and the pK_a of the poly(L-lysine), showing the strongest binding at pH 8.³² However, as soon as the local pH is unfavorable, i.e. lower than pH 5, or the local ionic strength is higher than physiological, we expect desorption of PLL-g-PEG from the SiO₂.³³ Thus, the observed loss of PLL-g-PEG at the calculated 3.8–4.2 pH range is in accordance with the literature. However, this change in pH and local ionic strength might not be the only explanation for the observed desorption. If we consider the finding from the previous experiment that a large amount of Cl₂ might be produced at the electrode, some hypochloric acid is produced according to Reaction I. The produced HClO can react with the amines of the lysine side chains to form monochloramines (–RNHCl derivatives) as described by Grisham et al. (Reactions II and III).³⁴

RNH₃⁺ ⇌ RNH₂ + H⁺ (II)

HClO + RNH₂ ⇌ H₂O + RNHCl (III)

Chloramines decompose to nitrogen-centered radicals or carbonyl groups. The formation of carbonyl groups accounts for ≪10% of the added HClO. Such a process may be biologically significant, since carbonyls can cross-link with free amine groupsviaSchiff-base formation,³⁵ a process that might also account for the recent finding of Ngankam and Van Tassel on the continuous adsorption of poly(L-lysine) under an applied potential.³⁶ The relevance of this reaction to our system is the following: PLL-g-PEG side groups could intra- or intermolecularly cross-link and, therefore, lose their charge. Consequently their ability to electrostatically adsorb on a surface is lost.

Therefore, we consider the electrochemically induced pH change and the formation of HClO to be the main reasons for the PLL-g-PEG desorption. The effect is not limited to the PLL-g-PEG molecule. Other electrostatically adsorbed polyelectrolytes, polyelectrolyte multilayers, or proteins show similar behavior.^15,37

Effects of applied current to viable cells

The PI staining experiment presented at the end of the Results section revealed PI positive cells at current densities higher than 0.57 A m⁻² (see Fig. 4B). This indicates either that PI becomes cytotoxic or cell membrane permeant by electrochemical modification or, which is more likely, cell membrane pore formation or membrane rupture occurs due to the change in local pH and the production of reactive molecules in the gap between the electrode and the cells.

In order to investigate whether the local pH could be responsible for the observed effect we have modeled the situation by a 2D finite element method described in the ESI.† We assumed a hypothetical surface attached cell (rectangular box 30 × 10 μm) and a flat electrode of the same size (30 μm). The gap between the cell and the electrode was assumed to be the same as the distance (100 nm) between astrocytes and a silicon chip measured by fluorescence interference contrast microscopy.³⁸

The calculation gave similar pH values for both parameter sets: pH 2.9 (I_fit = 0.032 A m⁻², D_fit = 3.310⁻¹⁰ m s⁻²) and pH 3.2 (I_real = 0.64 A m⁻², D_Proton = 9.310⁻⁹ m s⁻²). These pH values were reached already after 0.5 s and did not change during the simulated time of 2 min.

These calculated values were compared to a control experiment where we checked the pH resistance of myoblasts towards different pH solutions. There, an increasing amount of PI positive cells at pH ≤ 4 was observed. All cells were PI positive at pH ≤ 2. (See Fig. S2 in the ESI.† ) This means that the electrochemically induced pH change in the gap alone could explain the observed cell killing on the electrode.

Besides the local pH the cytotoxic effect of the electrically induced species, such as >50 μM HClO could also account for the observed cell killing.³⁴ Therefore, although we found a current threshold that induces cell death or the opening of membrane pores which leads to PI uptake within 2 min, the reason (i.e. either pH change or the formation of reactive molecules) remains unclear.

Conclusion and outlook

In this paper we have presented two methods that can be used to visualize the electrochemically induced pH changes close to a microelectrode. The first method is based on pH dependent fluorescent dyes in solution, while the second uses the induced desorption of an electrostatically adsorbed polyelectrolyte for the quantification of the process. These methods allowed us to monitor the pH progression close to the electrode and a simple model could be fitted to the data.

The effect of a locally applied current on the behavior of myoblasts has also been studied. We have found that currents larger than 0.57 A m⁻² induced cell death or propidium iodide uptake due to membrane pore formation within 2 min of exposure. We could attribute this to the change in local pH although the effect of other electrochemically created reactive molecules could not be excluded.

This presented approach could be used in different areas of applications where locally restricted pH change is important. For example, the system can be used for the local synthesis of short peptides ²⁴ or sol –gel thin film deposition.²³

Another advantage of this system is that the extent of the local pH change can be restricted also to the sub-micrometre range using buffered solutions. This is highly important in applications where cells are in close contact with the electrodes, such as cultures on electrodes or implants that trigger cellular responses with electric currents (pacemakers, neural implants). In addition, intelligent cell culture systems can be envisioned where the cell growth and motility is controlled by microelectrode arrays e.g. through providing unfavorable local environments (i.e. pH) in selected directions around the cells. The same effect could also be used to release drugs from pH sensitive polyelectrolyte multilayers or gels and facilitate the uptake of drugs into cells at the same time.^10,15

Acknowledgements

The authors would like to thank Prof. P. Rouxhet, Universite Catholique de Louvain, Belgium, for discussions on proton diffusion, Dr N. Wyrsch from University Neuchatel, Switzerland, for ITO coating and our technical staff Mr P. Lüthi and Mr M. Lanz at the Swiss Federal Institute of Technology (ETH), Switzerland, for the production of the flow cell and the microfabrication of the ITO electrode substrates.

References

1. R. Schmukler, Ann. Biomed. Eng., 1992, 20(3), 265–268.
2. P. Heiduschka and S. Thanos, Prog. Neurobiol. (Amsterdam, Neth.), 1998, 55(5), 433–461.
3. D. D. Zhou and R. J. Greenberg, Front. Biosci., 2005, 10, 166–179.
4. L. R. Sheffler and J. Chae, Muscle Nerve, 2007, 35(5), 562–590.
5. F. O. Morin, Y. Takamura and E. Tamiya, J. Biosci. Bioeng., 2005, 100(2), 131–143.
6. A.-R. Denet, R. Vanbever and V. Preat, Adv. Drug Delivery Rev., 2004, 56(5), 659–674.
7. J. C. Ojingwa and R. R. Isseroff, J. Invest. Dermatol., 2003, 121(1), 1–12.
8. E. Nilsson, H. von Euler, J. Berendson, A. Thorne, P. Wersall, I. Naslund, A. S. Lagerstedt, K. Narfstrom and J. M. Olsson, Bioelectrochemistry, 2000, 51(1), 1–11.
9. F. Yamauchi, K. Kato and H. Iwata, Nucleic Acids Res., 2004, 32(22), e187.
10. F. Yamauchi, K. Kato and H. Iwata, Langmuir, 2005, 21(18), 8360–8367.
11. J. P. Bearinger, J. Vörös, J. A. Hubbell and M. Textor, Biotechnol. Bioeng., 2003, 82(4), 465–473.
12. C. S. Tang, M. Dusseiller, S. Makohliso, M. Heuschkel, S. Sharma, B. Keller and J. Voros, Anal. Chem., 2006, 78(3), 711–717.
13. C. Tang, L. Feller, P. Rossbach, B. Keller, J. Voros, S. Tosatti and M. Textor, Surf. Sci., 2006, 600(7), 1510–1517.
14. C. S. Tang, P. Schmutz, S. Petronis, M. Textor, B. Keller and J. Vörös, Biotechnol. Bioeng., 2005, 91(3), 285–295.
15. F. Boulmedais, C. S. Tang, B. Keller and J. Voros, Adv. Funct. Mater., 2006, 16(1), 63–70.
16. G. W. Gross, W. Y. Wen and J. W. Lin, J. Neurosci. Methods, 1985, 15(3), 243–252.
17. O. Guillaume-Gentil, Y. Akiyama, M. Schuler, C. Tang, M. Textor, M. Yamato, T. Okano and J. Vörös, Adv. Mater., 2008, 20(3), 560–565.
18. R. B. Beard, B. N. Hung and R. Schmukler, Ann. Biomed. Eng., 1992, 20(3), 395–410.
19. C. Q. Huang, P. M. Carter and R. K. Shepherd, Ann. Biomed. Eng., 2001, 29(9), 791–802.
20. J. S. Guffey, M. J. Rutherford, W. Payne and C. Phillips, J. Orthop. Sports Phys. Ther., 1999, 29(11), 656–660.
21. C. L. Ballestrasse, R. T. Ruggeri and T. R. Beck, Ann. Biomed. Eng., 1985, 13(5), 405–424.
22. A. T. Kuhn and C. Y. Chan, J. Appl. Electrochem., 1983, 13(2), 189–207.
23. R. Shacham, D. Mandler and D. Avnir, Chem.–Eur. J., 2004, 10(8), 1936–1943.
24. K. Maurer, A. McShea, M. Strathmann and K. Dill, J. Comb. Chem., 2005, 7(5), 637–640.
25. Y. Hirose, G. Giannetti, J. Marquardt and T. Tanaka, J. Phys. Soc. Jpn., 1992, 61(11), 4085–4097.
26. N. Klauke, P. Monaghan, G. Sinclair, M. Padgett and J. Cooper, Lab Chip, 2006, 6(6), 788–793.
27. T. W. Swaddle, Inorganic Chemistry an Industrial and Environmental Perspective, Academic Press, San Diego, 1997, XVI, p. 482.
28. H. E. Posch, M. J. P. Leiner and O. S. Wolfbeis, Fresenius’ Z. Anal. Chem., 1989, 334(2), 162–165.
29. T. C. McIlvaine, J. Biol. Chem., 1921, 49(1), 183–186.
30. J. Crank, The Mathematics of Diffusion, Clarendon Press, Oxford, 1998, VIII, p. 414.
31. D. R. Lide, CRC Handbook of Chemistry and Physics: a Ready-Reference Book of Chemical and Physical Data, CRC, Boca Raton, 2007.
32. G. L. Kenausis, J. Voros, D. L. Elbert, N. P. Huang, R. Hofer, L. Ruiz-Taylor, M. Textor, J. A. Hubbell and N. D. Spencer, J. Phys. Chem. B, 2000, 104(14), 3298–3309.
33. T. M. Blattler, S. Pasche, M. Textor and H. J. Griesser, Langmuir, 2006, 22(13), 5760–5769.
34. M. B. Grisham, M. M. Jefferson, D. F. Melton and E. L. Thomas, J. Biol. Chem., 1984, 259(16), 404–413.
35. C. L. Hawkins, D. I. Pattison and M. J. Davies, Amino Acids, 2003, 25(3), 259–274.
36. A. P. Ngankam and P. R. Van Tassel, Proc. Natl. Acad. Sci. U. S. A., 2007, 104(4), 1140–1145.
37. M. A. Brusatori and P. R. Van Tassel, Biosens. Bioelectron., 2003, 18(10), 1269–1277.
38. P. Fromherz, Neuroelectronic Interfacing: Semiconductor Chips with Ion Channels, Nerve Cells, and Brain, in Nanoelectronics and Information Technology, ed. R. Waser, Wiley-VCH, Berlin, 2003, p. 781–810.

---

Footnote

† Electronic supplementary information (ESI) available: Electrolysis details, fluorescent dye standard curve, details of the control experiment exposing cells to different pH solutions and pH calculation in the gap between cell and electrode. See DOI: 10.1039/b814237h

---