The invention relates to biocompatible oxygen demand (BOD) assays and in particular to the use of ferrocene carboxylic acid (FcA) based BOD assays for measuring oxygen consumption.

Introduction

Oxygen is essential for aerobic life, acting as the final electron acceptor in the respiratory pathway. Therefore, the rate of oxygen consumption is proportionally linked to cell numbers, viability, growth rate and the phase of cell cycle that an individual cell is in ^[1] . In the past decade, biocompatible oxygen demand assays (BOD) were developed, specifically for the food and waste water processing industry, as a mean to determine and quantitate the presence of aerobic microorganisms in food and water sources. The current BOD methodologies can be sub-divided into two main approaches, fluorescence-based and electrochemical.

Fluorescence-based BODs have been based upon using dual-dye systems, whereby, two fluorophores are used simultaneously ^{1 1} . Dual-dye system was introduced into biological studies to remove inconsistencies produced by dye loading, leakage, bleaching and cell movement artefacts ^[2][3][4][5] . There are several major disadvantages with using fluorophores in BOD systems. Fluorophores are commonly cytotoxic themselves ^[7] . Additionally, fluorophores are extremely sensitive to light and are prone to photobleaching leading to the loss of active fluorophores over time. Fluorescence-based BODs are only ratiometrically quantitative and not absolutely quantitative ^[2][6][8] . The ratiometric quantification method is only accurate to the number of active fluorophores present in the system and limited by the visibility of the solution. In addition, ratiometric quantification is not very sensitive and can only be useful for comparing general experimental variables, such as the drop in relative signal intensity. Moreover, ratiometric quantification results have large data errors due to many inter and intra experimental variables, such as controlled cellular uptake of fluorophores, photobleaching and visibility of the solution ^[9] . Finally, fluorescence-based BODs are expensive and can only be used once.

On the other hand, electrochemical sensors are highly sensitive, because they are capable of detecting electron flow from one molecule ^[10] . This unique sensitivity allows electrochemical systems to overcome many of the problems that optical techniques may encounter, such as optimising dye loading concentration, photobleaching, reactive oxygen species concentration outside the limited range of detection. In addition, electrochemical sensors are faster, more simple to use, and operate at a low cost compared to optical techniques' ^11] . Due to these advantages, electrochemistry systems have gained popularity in many biological applications, such as BODs' ^12] , fruit fly brain signal studies' ¹³¹ and glucose' ^14] , hydrogen peroxide ^[15] and nitric oxide' ^16] sensing.

The current electrochemical BOD systems include use of Clark-type electrode' ^12b] , microchannel' ^12c] and ferricyanide systems' ^12a " ^17] . The BODs that use the Clark-type electrode system are based on using platinum electrodes pre-coated with a Teflon membrane, allowing oxygen to diffuse through and measured in real-time. The major setback with this system is that the Teflon membrane limits the amount of oxygen that can diffuse to the electrode, reducing the sensitivity of the sensor. Additionally, formation of platinum oxide means that the system is not fully biocompatible for live cell studies due to its toxicity' ^12b] . The microchannel system is highly sensitive, capable of monitoring single cell metabolism in a real time assay. However, it is relatively expensive and time consuming to fabricate the microchannels ^{[1 c]} . The ferricyanide system encompasses incubating the cells with ferricyanide for one hour in deoxygenated environment, then measuring the bulk concentration of ferricyanide being reduced by the cells after one hour. This is a non-real time assay, which means key cellular events can be missed. For example some molecules are toxic early on but with subsequent passage of time cells can recover. Therefore, these early events can be missed with the standard plat viability assay including presently using BOD ^[12a][17] . This is important in circumstances such as drug development as this early toxicity is to be avoided.

The aim of the Applicant's study is to develop a new electrochemical BOD system, which can monitor the respiration rate, down to a single cell level and in real-time. Additionally, a system that is easy to use, with high sensitivity and accuracy, and cost-effective. The primary process to achieving this was to conduct a series of experiments to identify a biocompatible mediator based on ferrocene derivatives for the specific detection of oxygen. Ferrocene derivatives are commonly used as electrochemical mediators as they have rapid rates of electron transfer, making it ideal for real-time monitoring' ^18] . Moreover, ferrocene has an electrochemically reversible one-electron redox behaviour that displays pH-independent redox potentials and non-auto-oxidation of the reduced form ^[19] . These unique properties of ferrocene have made it a popular choice as a mediator used in a wide range of electrochemical studies, including in the sensing of glucose ^[18][20] and development of dehydrogenase- ^[21] , oxidases- and other oxidoreductase enzyme-based biosensors ^[22] .

Although ferrocene derivatives have been shown to be capable of react with superoxides generated in aqueous solution, to the best of our knowledge there are no known reports of ferrocene being used to sense aerobic cellular respiration. Thus they investigated the electrochemical ability of FcA and ferrocene-methanol (FcMeOH) to report on the presence of oxygen. They then perform scan rate studies to investigate the kinetics to establish the optimal mediator to use in cellular studies. The best mediator was then used to report on the oxygen concentration in solution in the presence of cells.

The best mediator was identified as ferrocene carboxylic acid (FcA).

Ferrocene carboxylic acid is also known as carboxyferrocene, carboxyl ferrocene, ferrocenoic acid and cyclopentadiencarboxylic acid. It has CAS Number 1271-42-7 and is commercially available from, for example, Sigma- Aldrich Limited. It can exist in an oxidised (Fc ³⁺ ) form and reduced (Fc ²⁺ ) form.

FcA was observed to be capable of being highly sensitive and capable of measuring oxygen consumption down to a single cell level, in substantially real time (a matter of minutes) compared to prior art systems. An advantage of FcA compared to, for example, ferrocene methanol is that a separate oxygen specific peak was observed with FcA. The unique and separate FcA peaks obtained for FcA considerably simplifies the sensor because there is no need to conduct, for example, two cyclic voltammetry in the presence and absence of oxygen. This makes the sensor easier to use, for example, in the field, away from the laboratory.

The invention provides a biocompatible oxygen demand (BOD) sensor comprising an electrode and ferrocene carboxylic acid (FcA), wherein the sensor is adapted to measure an oxygen specific current peak from the FcA in the presence of oxygen.

The term "oxygen specific" is intended to mean that the peak is seen in the presence of oxygen and is lower or absent in the absence of oxygen. Typically the peak is substantially proportional to the amount of oxygen present in the environment in which the sensor is placed.

Typically the oxygen specific current peak is measured by cyclic voltammetry. Hence the sensor may be adapted to measure a cyclic voltammogram for FcA and the oxygen specific current peak is detectable on a cyclic voltammogram.

It typically comprises a reference electrode, a working electrode and a counter electrode. Potentially any suitable electrode may be used. Typically the working electrode is glassy carbon, platinum or gold. The counter, or "secondary", electrode may be any material which conducts easily and typically will not react with the environment in which the sensor is placed, such as platinum. The reference electrode may be a silver/silver chloride (Ag/AgCl) reference electrode of the type generally known in the art.

Cyclic voltammograms compare the potential between a reference electrode and the working electrode, where the electrode potential is ramped versus time. The current observed is plotted against the potential. FcA produces an oxygen dependent peak which may be detected by the sensor and used to calculate oxygen concentration.

Typically the sensor comprises a computer readable memory comprising one or more predetermined values representing the current at the peak compared to oxygen concentration. This may be in the form of a look up table representing a calibration curve or an algorithm to allow the conversion of current measured in a sample into a value representing oxygen concentration. It may compare a reading of earlier level of oxygen with a present level to indicate a level of oxygen consumption or oxygen production. This may be displayed, for example, on a read out on the sensor.

At slow scan rates two oxidation peaks are observed. The peak that occurs at more positive potentials (termed the 2 ^nd peak and represents a catalytic event) is the catalytic peak of interest and is typically found in cyclic voltammograms between approximately 380 and 410 mV, especially 390-400 or 395 mV. Typical peak is for 2 mM FcA, using an Ag/AgCl reference electrode, glassy carbon working electrode and platinum counter electrode, for example, as defined herein. Typically the FcA is in a solution, such as an aqueous solution, and can be used at nanomolar to millimolar concentrations. Typically the concentration of FcA in a solution is 0.1 to 20 mM, 1 to 10 mM or 2 mM.

The FcA may also be immobilised on the electrode, such as the working electrode. For example, it may be entrapped in a polymer such as a printed screen ink or polypyrrole.

The sensor may comprise a receptacle in which the sample is mixed with FcA and then a current is measured using the electrode(s).

One or more cells may be immobilised on a surface of an electrode or elsewhere on the sensor. This arrangement would be especially useful for the detection of the effects of a known compound on oxygen consumption by the cell, thereby showing, for example, effect of the health of the cell. Alternatively, it may be used to show the presence of toxic compounds in a sample of, for example, water from a contaminated water source.

Methods of measuring BOD are also provided, comprising providing an electrode and an FcA mediator, adding a sample to the sensor and measuring a current from the FcA mediator. Typically an oxygen specific peak is measured as described above. The sensor may be as defined above.

The sample may potentially be any sample, but is typically an aqueous sample. The sample may, for example, be a water sample to allow the presence or absence of contaminants or the presence of organisms such as bacteria or algae to be measured. The sample may be a growth medium into which a chemical is added to determine the effects of that chemical on cells attached to the sensor or otherwise within fluid communication with the sensor to allow the oxygen consumption by the cells to be measured.

Oxygen consumption or oxygen production may be measured. Oxygen levels may be detected at different times to allow oxygen consumption or production to be assessed. The amount of oxygen consumption per cell may be determined using Equation 1 (i). The concentration of oxygen from each cycle can be calculated using the Randles-Sevcik equation

1 1 1

for an irreversible system \ _P =W ^{} x} 1 ^{β5 α} n _a y-ACD ² \- ² equation l(ii). Equation 1

(i)

(moles of O] ₂ from 1st cycle x A')— moles o f O _z from 2nd cycle

0 ₂ consumption {moles cell ¹ s ^l j

time (s) x mrtber of cells

X = percentage of current drop between 2 cycles due to the limitation of oxygen diffusion to the electrode. This percentage drop varies with different electrolytes, and is obtained by doing a non-living organism control e.g. filtering the samples.

Where: i _p = the peak current (A); n = the number of electrons; a = the transfer coefficient; n _a

= the number of electrons in the rate limiting step; A = the surface area of the electrode

(cm ² ); D = the diffusion coefficient (cmV ¹ ) calculated from the non-electrocatalytic FcA peak (Figure 1 01) using the Randles-Sevcik equation for reversible system; v = the scan rate (Vs ^"1 ).

The organism may be an aerobic bacterium, fungus, insect, crustacean, bird, mammal, algae or higher plant. Higher plants are typically eukaryotic and comprise chloroplasts in at least some of their cells.

The invention will now be described by way of example only with reference to the following figures:

Figure 1 show typical cyclic voltammograms obtained for PBS solutions containing 2 mM

FcA or 2 mM FcMeOH in the absence of air (deoxygenated) and FcMeOH and the presence of air (Oxygenated).

Figure 2 shows a current functional plot of i _pa and i _pc for FcA and FcMeOH; peak currents obtained from cyclic voltammograms of FcMeOH-oxygenated♦, FcMeOH-deoxygenated +, non-electrocatalytic FcA-oxygenated (Figure 1 02) ▲, FcA-oxygenated 0 ₂ electrocatalytic peak * (Figure 1 03), Difference between peak currents obtained for FcMeOH-oxygenated and FcMeOH-deoxygenated which termed the electrocatalytic 0 ₂ peak • < 20 mVs ^"1 ). Insets represent current values at lower scan rates.

Figure 3 shows (i) typical CVs obtained for the first of the cycles for solutions containing different E. coli concentration of 0, 0.25, 0.5, 0.75, 1, 1.25, 1.5, 1.75, and 2 O.D./ml with 2 mM FcA. (ii) shows a plot of ODeoomn vs. electrocatalytic O2 peak current obtained from cyclic voltammograms in (i) All CVs were performed at a scan rate of 5 mVs ^"1 . (n=8, ±1SE).

Figure 4 shows a histogram of the magnitude of the mean electrocatalytic O2 peak current obtained from the cyclic voltammogram on the first cycle and second cycle (0, 0.25, 0.5, and 0.75 O.D./ml). (ii) summarizes the difference between the first and second electrocatalytic O2 peak (ipa) minus 18% (negative control - cyclic voltammetry studies conducted in the absence of E. coli), calculated as the limitation in the rate of oxygen dissolving into the system, (n=5, ±1SE).

Figure 5 (i) summarises the magnitude of the mean electrocatlytic O2 peak current difference between the first and second cyclic voltammetric cycles for solutions of cells at 4 O.D./ml at varying concentrations of ethanol (0, 2.5, 5, 10, or 12.5% v/v). A cyclic voltammogram was conducted in the absence of E. coli which acted as a negative control (n=5, ±1SE). (ii) summarise the mean colony counts post-ethanol incubation of same batch of cells used in the cyclic voltammetry study (n=14-16, ±1SE). (iii) summarises the mean optical density measured by UV spectrophotometer at 600 nm post-ethanol incubation of the same batch of cells used for cyclic voltammetry studies and agar plate bacterial growth assay, (n=5, ±1SE).

Figure 6 shows oxidation peak potential versus logarithm of scan rate measured for FcMeOH-oxygenated, FcA-oxygenated.

Figure 7 is a plot of oxidation and reduction peak currents versus square root of scan rate from CVs obtained for FcA and FcMeOH at varying scan rates.

Figure 8 (i) shows a histogram of the magnitude of the mean electrocatalytic O2 peak current obtained from the cyclic voltammograms on the first cycle and second cycle (Filtered and raw Vale/canal water samples (n=3)).Figure 8 (ii) shows a histogram of the mean percentage electrocatalytic 0 ₂ peak current drop between first cycle and second cycle in the cyclic voltammogram (Water control (n=2), filtered and raw Vale/canal water samples (n=3)) (±1SD).

Figures 9(i) - (iii) show cyclic voltammagrams of a cellulose acetate membrane wherein ferrocene carboxylic acid has been trapped on the surface of the membrane. All cyclic voltammagrams were performed at 5mV sec ^"1 .

Figures 10(i) - (iii) show cyclic voltammagrams of a cellulose acetate membrane wherein ferrocene carboxylic acid has been trapped on the surface of the membrane. All cyclic voltammagrams were performed at 20mV sec ^"1 for 25 CVs.

Material and Methods Chemicals

LB agar plates, ferrocene carboxylic acid and glucose were purchased from Sigma- Aldrich. LB broth was purchased from Fisher and ferrocene methanol was purchased from Acros Organics.

E. coli DH5-a culture

E. coli DH5-a strain was stored on LB agar plates at 4°C. E. coli were sub-cultured in liquid LB broth medium the day before the experiment was performed and allowed to grow for 18 hours at 37°C on a shaker at 200 rpm in baffled flasks. Bacteria were sub-cultured at 1 :21 dilution in fresh LB broth in baffled flasks and incubated at 37°C on a shaker for 2 hours. Bacteria were subsequently harvested by centrifuging at 3261 x g and washed twice in 10 ml of sterile phosphate buffer solution (PBS) (50 mM PBS, pH 7.3 containing 0.1 M KCl) before re-suspending in PBS. Cell density measurements were performed at Οϋβοο using a Cecil CE1020 UV spectrometer. The cell suspension was kept on ice and for up to 5 hours until cells were needed for assaying.

Electrochemical measurements

A Ag/AgCl reference electrode, 3 mm diameter glassy carbon working electrode and platinum counter electrode were used in the cyclic voltammetric (CV) studies. A Gamry 600 potentiostat with data acquisition software was used for electrochemistry experiments. The glassy carbon electrode was polished with 50 nm alumina powder for 5 minutes prior to each cyclic voltammogram (CV) being performed. Cyclic voltammetric studies were performed from starting potential of 0 V with a switching potential of 0.6 V and an end potential of 0 V for solutions of 2 mM of ferrocene carboxylic acid (FcA) and ferrocene methanol (FcMeOH) in PBS. CVs in the absence of E. coli were performed at scan rates from 5 to 2000 mVs ^"1 for FcA and FcMeOH in the presence of air (oxygenated). Additionally, CVs were performed on solutions containing FcMeOH and FcA that were purged with oxygen free nitrogen for removal of air (deoxygenated) at scan rates of 5-40 mVs ^"1 and 5 mVs ^"1 , respectively.

Electrochemical determination of bacterial cell numbers

Before each assay, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4 OD/ml of E. coli were pre-incubated with 10 mM of glucose in PBS at 37 °C on a shaker in conical flasks for 30 minutes. These solutions were then diluted by half. FcA was then added to give a final concentration of 2 mM and CVs were performed at 5 mVs ^"1 .

Cytotoxicity measurements

A 5 ml stock solution containing E. coli (4 ODeoonm) was prepared in PBS, containing 10 mM glucose as the growth substrate and HPLC grade ethanol at final concentrations of 0, 2.5, 5, 10 and 12.5 %. This E. coli suspension was then incubated in a shaking incubator at 37 °C for 1 hour. OD measurements were taken before and after incubation to verify if any cellular growth occurred during the 1 hour incubation. For electrochemical interrogation a 3.5 ml sample of the incubated E. coli suspension was added to 3.5 ml of 4 mM FcA in PBS. This gave a final working assay concentration used in cyclic voltammetric studies of E. coli at an OD of 2 and FcA at a concentration of 2 mM. Cyclic voltammetry was then performed at a scan rate of 5 mV s ^"1 . For agar plate growth assays, the E. coli suspension was diluted to 1 :1000, 1 : 100,000, 1 :500,000 and 1 :1 million prior to seeding onto the agar plates. These agar plates were then incubated for 20 hours at 37°C followed by colony counts after the incubation period.

Results and Discussion

Comparison of the ability of FcMeOH and FcA to electrocatalyse the oxidation of reactive oxygen species

Understanding the electrochemical properties of the mediators is essential in the decision making process for choosing an appropriate mediator for electrochemical-cell studies. Cyclic voltammetry was used to investigate the electrochemical characteristics of FcA and FcMeOH in order to establish the most suitable mediator for probing bacterial aerobic respiration. In preparation for our respiration studies of cells, it was important to ascertain the effect of oxygen on the generated cyclic voltammograms for the mediators FcA and FcMeOH.

Therefore, cyclic voltammograms were performed with PBS containing 2 mM FcA or FcMeOH exposed to oxygen and solutions degassed with oxygen- free nitrogen for the removal of oxygen from the electrolyte and typical CVs obtained are shown in Figure 1. Importantly, there is a major difference between the CVs obtained for FcA and FcMeOH in the presence of air. For cyclic voltammograms obtained in the presence of oxygen for FcMeOH only one oxidation peak at a mean peak potential of 257 mV was observed (Figure 1 01). However, for FcA there are two oxidation peaks obtained in the presence of oxygen (Figure 1 02 and 03) observed at mean peak potentials of 352 mV and 394. mV. Diffusion is the rate limiting step governing the magnitude of the oxidation currents (Figure 1 01 and 02) for the simple 1 electron oxidation in which there is no contribution from catalytic current for FcA and FcMeOH (Figure 7) as demonstrated by a plot of peak current versus the square root of scan rate.

Figure 1 also indicates that the magnitude of peak current for FcMeOH (Figure 1 01) and FcA (Figure 1 03) is dependent on oxygen's presence. This is shown by comparing the oxidation peak currents obtained for solutions of FcMeOH in the presence (Figure 1 01- oxygenated, mean=6.75 μΑ, ±1SD=0.037) and absence (Figure 1 01-deoxygenated), mean=6.48 μΑ, ±1SD=0.053) of oxygen and indicates that in its presence an increase in the 01 peak of approximately 200 nA and an equivalent decrease in the reduction peak occurred (Figure 1). The magnitude of the oxidation peak current obtained in the presence of FcMeOH which can be attributed to oxygen is equal to the difference between the peak current obtained in the presence of oxygen and its absence. We conclude any current generated that is associated with the presence of O2 is convoluted with the normal FcMeOH electrochemistry. Additionally, results from the current function plots for FcMeOH (Figure 2) also supports this proposition as the correlation coefficients obtained at low scan rates prior to subtracting the deoxygenated peak (R ² 0.965) lies between the deoxygenated correlation coefficient value of R ² 0.752 and O2 electrocatalytic current correlation coefficient value of R ² 0.983.

However, when CVs on solutions of FcA were performed in deoxygenated solution the 03 peak current associated with presence of oxygen (Figure 1 03) was no longer observed. We conclude that this is the de-convo luted current which arises from the presence of oxygen. For simplicity the current associated with oxygen presence is termed the O2 electrocatalytic current.

The proposed mechanism for O2 electrocatalytic current with FcA and FcMeOH is attributed to the oxygen which is electrochemically reduced at 0 V forming an O2 and is supported by Cassidy ^[23] . This conclusion is elucidated by the fact that on removal of the oxygen there is a decrease in current observed with FcMeOH (Figure 1) and removal of the electrocatalytic O2 peak in CVs obtained for FcA. The O2 subsequently chemically oxidises water forming the reactive oxygen species hydroxyperoxyl (H02 ^~ ) ^an <A a hydroxyl ion (OH ). As we scan through the voltammogram in a forward direction the Fc is electrochemically oxidised to Fc ⁺ . We suggest the OH and HO2 generated from spontaneously chemically reduces Fc ⁺ to Fc and at the same time the chemically reduced mediator would be once again electrochemically oxidised. This leads to a catalytic enhancement in the magnitude of the oxidation peak current as the concentration of reduced Fc is increased in the presence of oxygen. Moreover, it is well understood that peak current is proportional to concentration of redox molecules under investigation and explains why in deoxygenated solutions we see a decrease in the electrocatalytic peak oxidation current for FcMeOH and FcA. The mechanism we propose is an electrochemical-chemical-electrochemical (E-C-E) system. It can be presumed that the chemical steps must be relatively slow because with increasing scan rate there is no increase in the observed O2 electrocatalytic current above 20 mVs ^"1 for FcMeOH, and 40 mV s ^"1 for FcA. We suggest that the reason for this difference is because the rate at which the chemical step occurs is slower with the FcMeOH than the FcA. This is also supported by Figure 6 in which we show the actual charge transfer coefficient for the electrocatalytic oxidation is fast for the FcA when compared to FcMeOH. Consequently, FcA produces a larger electrocatalytic current under the same condition.

We tentatively suggest that the difference in behavior observed for FcA and FcMeOH in terms of the position of peak potentials is caused by the different functional groups. FcA contains a carbonyl and FcMeOH a hydroxyl. The electron withdrawing effect of the carbonyl group adjacent to the cyclop entadienyl ring would lead to lowering the electron density around the iron (Fe ²⁺ ) center meaning the ferrocene requires a larger over-potential to be oxidized and evidence for this behavior has been reported by others for different ferrocene derivatives. This would decrease the ability of the Fe ²⁺ to lose electrons, therefore, as observed, a higher potential is required to electrochemically oxidize FcA than FcMeOH, resulting in FcMeOH having a lower redox potential then FcA (Figure 1). This also explains why the 0 ₂ electrocatalytic peak for FcA is de-convo luted, whilst for FcMeOH it is convoluted, and occurs at higher potentials than the simple non-electrocatalytic current for the FcA. This is explained by the fact that after the Fc is oxidized into Fc ⁺ , the Fe ³⁺ center interacts with the surrounding OH and HO2 . We propose the Fe ⁺ centre is instantaneously reduced by the reactive oxygen species forming an adduct, we hypothesize that the carbonyl group on the FcA makes the adduct relatively stable. This causes a further lowering of the electron density of the iron centre when compared to FcA alone. As a result, an even higher oxidation peak potential is needed to oxidise FcA resulting in the separation of the O2 electrocatalytic peak. On the other hand, the FcMeOH adduct is not stabilized and instantaneously oxidised. Therefore the catalytic signal seen for FcMeOH is not separate from the normal FcMeOH signal when compared with the FcA.

In order to understand the mechanistic control and provide further evidence that electrocatalysm was occurring via the mechanism proposed. Investigations into the kinetics of the reaction were performed using cyclic voltammograms of the FcA and FcMeOH at varying scan rates in both oxygenated and deoxygenated solutions. Peak currents obtained from cyclic voltammograms performed at varying scan rates were then plotted into a current functional plot (Figure 2). At higher scan rates there is no relationship with current function (Figure 2 iv). This lack of relationship at relatively fast scan rates is indicative of a diffusion limited process and supports the data obtained in Figure 7. However, when we analyze the plot at slower scan rates there is a deviation from this behavior when oxygen is present both for FcA and FcMeOH (Figure 2 i, ii, and iii) and is indicative of an electrocatalytic process.

In the absence of oxygen the current function plot for FcMeOH (<20 mV s ^"1 ) yields a correlation coefficient value of ² = 0.752 (Figure 2 i) suggesting the current is diffusion limited as expected for a simple 1 electron transfer event. On the other hand in the presence of oxygen the electrocatalytic current obtained for FcMeOH (attained by subtracting the peak current obtained in the presence of oxygen minus the peak current obtained in the absence of oxygen) is directly correlated to scan rate in the current function plot with a correlation coefficient value of R ² 0.983 indicative of a non-diffusion limited process (Figure 2 ii). This relative large correlation provides supporting evidence that the electrocatalysm is occurring via the proposed E-C-E mechanism and we attribute this deviation to a slow chemical step which is rate limiting.

The correlation coefficient values obtained for current function plot for the non- electrocatalytic FcA peak (Figure 1 02) in the presence of oxygen yields a relatively low correlation coefficient value of R ² =0.872 at low scan rates (Figure 2 i). This indicates that the peak current even at slower scan rates (<40mV s ^"1 ) is under diffusion control as expected for a simple 1 electron transfer redox process. Moreover, the current function values obtained for the electrocatalytic O2 peak for FcA (Figure 2 ii) yield a correlation coefficient of R ² =0.985 indicating the current is under non-diffusion control and we suggest this arises due to this peak representing the electrocatalytic oxidation. This deviation from diffusion controlled behavior is attributed to the chemical step being rate limiting which is similar behavior to that observed with FcMeOH. Consequently, oxidation peak 02 (Figure 1) is largely non-catalytic current arising from the redox events of the FcA alone whereas the electrocatalytic O2 peak (Figure 1-03) represents the electrocatalytic oxidation of OH

Combining the data discussed for FcA and FcMeOH, FcA was specifically chosen for the next part of our study for the following reasons: Even though FcMeOH has the lowest E _pa making it the most favourable mediator in a biological system, the slower chemical electron transfer kinetics in the reduction of FcMeOH in the presence of oxygen would make the system less sensitive. The unique and separate electrocatalytic O2 peak obtained for FcA would simplify the system of study because there is no need to conduct two cyclic voltammetry studies in the presence of and absence of oxygen in order to obtain the electrocatalytic signal.

Using FcA-mediated system to detect E. coli concentration

E. coli (strain DH5 ) was used to characterise the ability of FcA to monitor cellular respiration via oxygen concentration. Optical density measurements were calibrated by determining the E. coli cell numbers, using a haemocytometer, in a solution that had a 1 ODeoonm per ml. This was followed by serial dilution of the stock E. coli solution to the appropriate concentrations in PBS. The cell solution was subsequently pre-incubated with 10 mM of glucose then mixed with final concentration of 2 mM FcA. Two consecutive cyclic voltammograms were performed on solutions containing varying numbers of cells and Figure 3 summarises the magnitude of the FcA electrocatalytic O2 peak measured from the first of the two consecutive cyclic voltammograms. As expected, an increase in E. coli concentration in the system leads to an increase in oxygen consumption, and therefore, a decrease in the electrocatalytic O2 peak current. In addition, the decrease in the electrocatalytic O2 peak current is directly proportional to the increase in E. coli cell numbers.

The E. coli in the system were continuously consuming oxygen, and this could be monitored in real-time by measuring the change in the catalytic peak current between two consecutive cycles. The electrocatalytic O2 peak current generated on the first cycle and the second cycle is exactly one cyclic voltammetry cycle apart at a fixed scan rate of 5 mV s ^"1 . A plot of the electrocatalytic 02 peak current from the two cycles at the varying cell concentrations can be observed in Figure 4 i. Assay solutions containing higher cell numbers results in a larger difference in the magnitude of the current between the cycles. It is also worth mentioning that there is always an 18% drop in the magnitude of the peak current for the electrocatalytic O2 peak between the first and second cyclic voltammetric cycles in the control study (cyclic voltammetry study conducted in the absence of E. coli), possibly due to the limitation in how fast oxygen dissolves into the system. Therefore, any current decrease measured that is greater than 18% between the first and second electrocatalytic O2 peak can be assigned to oxygen consumption by the E. coli in the system. Moreover, the decrease in peak current obtained for the electrocatalytic O2 peak between the two cyclic voltammetry cycles is directly proportional to the increase in the number of E. coli in the system (Figure 4 ii). By using Equation 1, the oxygen consumption rate down to a single cell level (cells per second) was calculated to be approximately 1.122 x 10 ^"17 moles s ^"1 cell ^"1 , which is similar to the values published in the literature (4.31 x 10 ^"20 mole s ^"1 cell ^"1 ) for E. coli strain K21. The difference in oxygen consumption rate observed in our study using E. coli strain DH5- and the K21 strain could possibly be due to the difference in metabolic demands between the two E. coli strains. More importantly, in this study, we demonstrated for the first time the simplicity and the accuracy and sensitivity of the FcA-mediated system.

Electrochemical cytotoxicity assay

To confirm FcA-mediated system can accurately report on the metabolic rate of cells and to demonstrate the wide capabilities of the developed assay the FcA was used to detect the toxicity of a model toxin ethanol. Cyclic voltammetry studies were conducted using a fixed number of E. coli (4 O.D./ml) and incubated with different concentrations of ethanol (0, 2.5, 5, 10, and 12.5% v/v). Ethanol was used because it is cytotoxic to E. coli through the disruption of plasma membrane. An increase in ethanol concentration should lead to a decrease in the number of viable cells, leading to the decrease in oxygen consumption and therefore an increase in the electroctalytic O2 peak current. In addition, the optical density of E. coli was measured post-incubation with ethanol to ensure the same number of cells are still present prior to performing cyclic voltammetric studies to confirm any difference was not due to bulk changes in cells concentration. Moreover, our electrochemistry assay results for analyzing the toxic effect of cells was compared to a standard viable agar plate method of toxicity testing to enable validation of our system.

The cyclic voltammetry studies and electrocatalytic O2 peak currents generated in the presence of ethanol and cells are summarized in Figure 5 i. In the presence of relatively high ethanol concentrations a decrease in oxygen consumption and consequently increase in the electrocatalytic O2 peak current was observed for increased ethanol concentrations. As expected the magnitude of the peak current inversely proportional to increased ethanol. A complete loss in oxygen consumption was determined at 12% v/v ethanol, suggesting the system had no viable cells since the electrocatalytic O2 peak current of the negative control (cyclic voltammetry studies in the absence of cells) is approximately the same as cell incubated with 12% v/v ethanol. In addition, the number of viable cells was confirmed by the growth assay (Figure 5 ii) with the increase in ethanol concentration, there is a decrease in viable cells and a complete lost in cell viability at 12.5% v/v ethanol concentration matching the electrochemistry results. Moreover, the optical density study confirms the number of cells pre- and post- ethanol incubation was consistent, see Figure 5 iii. This further demonstrated the biological compatibility and sensitivity of the FcA-mediated system and shows advantageous toxicity assay behavior as our system was quick when compared to the standard plate viability assay which takes 24 hours to perform compared to seconds with the our electrochemical methods. In addition our system reports sub-lethal toxicity which is missed by the plate viability assay as we see that there is no significant difference at concentrations of ethanol obtained equal to and below 10%. This is because over the 24 hour period the cells can recover from the lower concentration and therefore the sub-lethal toxicity is missed which is reported by the electrochemistry assay. This is important and demonstrates a key advantage of our toxicity assay as pharmaceutical companies are interested in avoiding sub-lethal toxicity. The developed electrochemical method (Figure 5 i) is also more accurate than the plate viability assay as indicated by the much larger standard error bars (Figure 5 ii).

Conclusions

In summary, we report on unique behaviour of FcA to electrocatalytically report on oxygen concentration via its interaction with OFT and H0 ₂ and provide a detailed mechanistic insight into the reaction. FcAs ability in monitoring oxygen was taken advantage of to develop a rapid cellular respiration assay which could be monitored in near real-time. This assay was shown to be able to report accurately on the cell numbers present and was adapted to be used as a rapid toxicity assay. It is envisaged that the developed assay has potential to impact on fields and industries ranging from environmental toxicology through to pharmaceutical and agrochemical industries as demonstrated by the shortening of a commercially available bioxygen demand assay down to minutes rather than 5 days.

Supplementary Material

Electrochemical characterisation of FcA and FcMeOH

All cyclic voltammetric studies described were conducted in PBS with a final concentration of 2 mM FcA or 2 mM FcMeOH. A typical CVs obtained for FcA and FcMeOH conducted on the open bench (oxygenated) at a scan rate (SR) of 5 mVs ^"1 is shown in Figure 1. The mean separation in peak potential (ΔΕ _Ρ ) for FcA and FcMeOH is 64.94 mV (±1SD=0.075) and 70.33 mV (±1SD=4.163), respectively. The values for the mean oxidation peak current over reduction peak current (i _pa /i _P c) are 0.993 (±1SD=0.022) and 0.997 (±1SD=0.007), respectively. These values indicate that both FcA and FcMeOH are displaying a quasi- reversible electrochemical behavior ^[27] . The mean peak potential (Ep) values for FcMeOH are 257.5 mV (±1SD=2.098) for the oxidation peak and 187.2 mV (±1SD=2.066) for the reduction peak, whereas in the case of FcA, they are 354.6 mV (±1SD=1.141) for the oxidation peak and 289.6 mV (±1SD=1.216) for the reduction peak ^[27a] . These values are similar to the values reported in the literature. Additionally a second oxidation peak is observed at approximately 395 mV (±1SD=0.199) which is slightly higher than 317mV reported in Cassidy's study and is likely due to the use of different electro lyte ^[28] . This study showed FcMeOH requires much less energy to trigger the electrochemical redox event than FcA (See Figure 6) as highlighted by the lower peak potential values.

A scan rate study was performed with solutions of FcMeOH and FcA and typical peak current obtained at the varying scan rates are plotted in Figure 7. The peak currents for both FcA (A) and FcMeOH (♦) are proportional to the square root of the scan rate. This well- known behavior indicates that the peak current is under diffusion control. We also noted that at higher scan rates the electrochemical O2 peak (Figure 1 peak 03) observed for FcA is no longer present. We were interested in ascertaining the rate of diffusion of the mediators as this can influence the sensitivity of the system. Using the Randies- Sevcik equation for a reversible system (i _p =(2.69xl0 ⁵ )n ^{3 2} AD ^1/2 Cv ^1/2 ), the diffusion coefficient (D) of FcMeOH and FcA were calculated from 5 mVs ^"1 oxidation peak data (Peak 01 and 03 from Figure 1 i). The diffusion coefficient for FcMeOH (D=7.87 x 10 ^~7 cmV ¹ ) is similar to that reported in the literature (D=2.5 x 10 ^"7 cm ² s ^_1 ) ^{[ 0]} . Whereas, the diffusion coefficient for FcA (D=7.08 x 10 ^"7 cmV ¹ ) is a magnitude smaller than reported in the literature (D=4.3 x 10 ^"6 cm ² s ^"1 ) ^[27b It is not a surprise that the diffusion coefficient for FcMeOH and FcA obtained in our study are not identical to those reported in the literature, since the viscosity of the room temperature ionic liquids used would greatly influence the diffusion coefficient of the mediator ^[30] . We show FcMeOH has a faster diffusion coefficient than FcA in our study, which could be explained by FcMeOH (Mw=216.06) being a smaller molecule than FcA (Mw=274.05).

The oxidation peak potential of FcMeOH (Figure 1 01 ) and FcA (Figure 1 02 and 03) obtained from cyclic voltammograms performed in the presence of oxygen were plotted against logarithm of scan rate (Figure 6). As seen from the figure, there were no changes in oxidation peak potential for scan rates up to about 100 mVs ^"1 in the case of FcMeOH (Figure 6 FcMeOH 01 peak) and FcA non-electrocatalytic peak (Figure 6 FcA 02 peak). For scan rates beyond 100 mVs ^"1 , both peak potentials mentioned changed linearly with log of scan rate with correlation coefficient values of R ² 0.937 and R ² 0.955 for FcA-oxygenated non- catalytic peak (Figure 6 FcA 02 peak) and FcMeOH-oxygenated (Figure 6 FcMeOH 01 peak). On the other hand, the FcA electrocatalytic peak (Figure 6 FcA 03 peak) yielded a correlation coefficient value of R ² = 0.964 from very low scan rates (5 mVs ^"1 to 40 mVs ^"1 ). These observations indicated FcMeOH (Figure 6 FcMeOH 01 peak) and FcA- non- electrocatalytic peak (Figure 6 FcA 02 peak) were quasi-reversible over the scan rate range of 5 to 100 mVs ^"1 and irreversible beyond 100 mVs ^"1 . Whereas in the case of the FcA electrocatalytic process (Figure 6 FcA 03 peak) is irreversible from 5-40 mV s ^_1 .

By comparing the graphical lines for the irreversible behavior we can ascertain that the electrocatalytic current for FcMeOH is convoluted with the non-electrocatalytic current. We can calculate the charge transfer coefficients (ana) for the various peaks by using the following equation in which, AEpa/Alogv is equivalent to the gradient of the plots obtained in figure 6.

_{Equation 2}

In which Epa (V) is the is oxidation peak potential, v is scan rate in mV s ¹ , ru is the number of electrons in the rate determining step and a is the charge transfer coefficient in which with increasing faster charge transfer process. R is the standard gas constant, T is the standard temperature in Kelvin at 25°C and F is the Faraday constant (96485 C mol ^"1 ).

The charge transfer coefficient value of 0.28 was obtained for the FcA non-electrocatalytic peak. In contrast, the FcA O2 electrocatalytic peak (Figure 6 FcA 03 peak) has an a value of 0.77 for one electron. This indicates two electrons are involved in this oxidation. However, due to the nature of FcMeOH electrocatalytic peak and non- electrocatalytic peak being convoluted, the FcMeOH electrocatalytic peak potential just simply cannot be extract for the a value calculation. Water Quality Testing

SI.2 Fresh water sample testing

Method and results

To investigate further the applicability of the developed FcA mediated BOD system to more complex samples, experiments were performed with fresh water samples collected from nearby natural water resources. Investigations were subsequently performed using the BOD system developed to calculate the oxygen levels. None living organism controls (acellular controls) were prepared by filtering the water samples through a 0.5 μιη filter. 3.5 ml of each sample was mixed with 3.5 ml of 4 mM FcA giving a final concentration of 2 mM FcA. Cyclic voltammetry was then performed at a scan rate of 5 mV s ^"1 .

Algae rich fresh water samples were collected from two different water sources, namely from a canal and from a stream (Vale water). These water sources were filtered to remove all living organisms and decaying matter, serving as acellular controls. This allowed us to obtain the standard percentage current drop due to the limitation of oxygen diffusion in different electrolyte environments. From simple observation, the water sample collected from the canal appears greener than the water sample from the Vale, suggesting higher algae content in the canal water. After filtration, both water samples became clear. CVs were performed on acellular controls and raw water samples from the canal and Vale stream and the 02 electrocatalytic peak currents were measured. A summary of the current changes and percentage current changes between the two CV cycles for both raw samples and control samples are plotted in Figure S3. Using Equation 1 part ii (from the main text), the electrocatalytic 02 peak current of the first cycle can be used to calculate the overall oxygen contained in the sample. For the Vale and canal water samples a total oxygen concentration of 341.6 μΜ and 351.3 μΜ was calculated, respectively.The results, with regards to the dynamic change in oxygen concentration with time, from algae rich fresh water samples are completely opposite from the E. coli samples. In the case of E. coli samples, the cells were consuming oxygen in the solution therefore the decrease of the electrocatalytic 02 peak at the second cycle is greater when comparing to the acellular controls. That is if one calculates the percentage change in current between the first cycle from the second cycle of acellular control, and then subtract the equivalent percentage change calculated from the E. coli working sample at OD 0.75 a relative negative change when compared to the acellular control of approximately of -32% is calculated, representing oxygen consumption (18%- 1%=-32% values obtained from figure 3(ii)). For the case of the raw water samples obtained from the Vale and canal an approximate relative change, when working samples were compared to acellular controls, of +14% and +18% were calculated, respectively. The larger value obtained for the canal sample of 18% is indicative a high algae content as previously observed by a deeper green colour of the sample. Algae are capable of photosynthesising and producing oxygen and consequently this experimentally proves a higher algae content on the canal than the Vale water. In this study, the electrocatalytic 02 peak obtained for the second CVs for raw algae rich water samples shows a relatively much smaller decrease in the current when comparing it to the acellular controls. This indicates that there is an increase in oxygen content over time for the raw water samples. By inputting the data into Equation SI, an increase in oxygen level of 2.522 x 10 ^"9 moles sec-1 and 3.422 x 10 ^~9 moles sec ^"1 was obtained from the vale and canal water, respectively. These experiments demonstrated that FcA mediated BOD system can also be used for detecting increases in oxygen content in solution over time, not just designed for detecting oxygen consumption.

Equation SI

... Mmoles of O!-, from ^" 1st cycle X X)— moles of Oi [rem ind cycle 0 ₂ increase (moles s ) = ^{^} "

time s)

Embedding of ferrocene carboxylic acid on printed carbon electrodes

Method:

50 mM of ferrocene carboxylic acid (FcA) was dissolved in ethanol. 0.1% to 1% cellulose acetate (CA) was diluted in 90% acetone and 10 % water. FcA was dropped on top of the communicably printed carbon electrode and allow it to dry. Drop 0.1% to 1% of CA on FcA coated surfaces and allow to dry (at room temperature (RT) or 65 °C). Repeat CA dropcoating on some of the surfaces.

Results:

Drying at 65 °C was resulting in peeling off of CA membrane; therefore drying in at room temperature was used for the rest of the study. Thickness of the CA was affecting the oxygen diffusion therefore affecting the oxygen catalytic peak height. Therefore, different concentrations of CA was tested. With all concentration of CA, 0.3% was giving the best result regarding the oxygen catalytic peak location and height. The cyclic voltammetry studies illustrate that FcA is trapped on the surface as the peak is <59 mV. Although further optimisation can continue to be carry out (e.g. improvement on the non-dissolution of the CA membrane), the studies performed so far indicate that the extra peak that we observe for FcA in solution is also observed when FcA is embedded in a solid support.

References

[I] a)D. S. Martin, J Gen Physiol 1932, 15, 691; b)K. C. Fisher, F. H. Armstrong, Fed Proc 1946, 5, 27.

[2] X.-H. Wang, H.-S. Peng, H. Ding, F.-T. You, S.-H. Huang, F. Teng, B. Dong, H.-W.

Song, J Maters Chem 2012, 22, 16066.

[3] M. C. Ashby, A. V. Tepikin, Physiol Rev 2002, 82, 701.

[4] R. B. Sekar, A. Periasamy, J Cell Biol 2003, 160, 629.

[5] J. T. Bradshaw, T. Knaide, Artel 2010, 1.

[6] D. P. O'Neal, M. A. Meledeo, J. R. Davis, B. L. Ibey, V. A. Gant, M. V. Pishko, G. L.

Cote, IEEE Sens J 2004, 4, 728.

[7] R. Alford, H. M. Simpson, J. Duberman, G. C. Hill, M. Ogawa, C. Regino, H.

Kobayashi, P. L. Choyke, Mo I Imaging 2009, 8, 341.

[8] F. M. Cabrerizo, J. Arnbjerg, M. P. Denofrio, R. Erra-Balsells, P. R. Ogilby, Chem

Phys Chem 2010, 11, 796.

[9] K. M. Hanson, M. J. Behne, N. P. Barry, T. M. Mauro, E. Gratton, R. M. Clegg,

Biophys J 2002, 83, 1682.

[10] a)A. Nitzan, M. A. Ratner, Science 2003, 300, 1384; b)X. Shan, I. Diez-Perez, L.

Wang, P. Wiktor, Y. Gu, L. Zhang, W. Wang, J. Lu, S. Wang, Q. Gong, Nature

Nanotech 2012.

[I I] J. Wang, Biosens Bioelectron 2006, 21, 1887.

[12] a)M. A. Jordan, D. T. Welsh, R. John, K. Catterall, P. R. Teasdale, Water Res 2013,

47, 841; b)H. Suzuki, T. Hirakawa, I. Watanabe, Y. ikuchi, Anal Chim Acta 2001, 431, 249; c)T. Saito, C.-C. Wu, H. Shiku, T. Yasukawa, M. Yokoo, T. Ito-Sasaki, H. Abe, H. Hoshi, T. Matsue, Analyst 2006, 131, 1006.

[13] M. A. Makos, .-A. Han, M. L. Heien, A. G. Ewing, ACS Chem Neurosci 2009, 1, 74.

[14] Z. Nie, C. A. Nijhuis, J. Gong, X. Chen, A. Kumachev, A. W. Martinez, M.

Narovlyansky, G. M. Whitesides, Lab Chip 2010, 10, All.

[15] S. Chen, R. Yuan, Y. Chai, F. Hu, Microchim Acta 2013, 180, 15.

[16] S. Griveau, F. Bedioui, Anal Bioanal Chem 2013, 1.

[17] N. Pasco, K. Baronian, C. Jeffries, J. Webber, J. Hay, Biosens Bioelectron 2004, 20, 524.

[18] A. E. G. Cass, G. Davis, G. D. Francis, H. A. O. Hill, W. J. Aston, I. J. Higgins, E. V.

Plotkin, L. D. L. Scott, A. P. F. Turner, Anal Chem 1984, 56, 667.

[19] A. E. G. Cass, G. Davis, M. J. Green, H. A. O. Hill, J Electroanal Chem 1985, 190,

111.

[20] a)W. Schuhmann, Biosens Bioelectron 1995, 10, 181; b)W. Schuhmann, Biosens

Bioelectron 1993, 8, 191.

[21] S. Serban, N. El Murr, Biosens Bioelectron 2004, 20, 161.

[22] J. M. Dicks, W. J. Aston, G. Davis, A. P. F. Turner, Anal Chim Acta 1986, 182, 103.

[23] J. Cassidy, J. O'Gorman, M. Ronane, E. Howard, Electrochem Commun 1999, 1, 69.

[24] a)X. Wu, E. R. T. Tiekink, I. Kostetski, N. Kocherginsky, A. L. C. Tan, S. B. Khoo,

P. Wilairat, M.-L. Go, Eur J Pharm Sci 2006, 27, 175; b)M. Hocek, P. Stepnicka, J.

Ludvik, I. Cisafova, I. Votruba, D. Reha, P. Hobza, Chem - Eur J 2004, 10, 2058.

[25] R. S. Nicholson, I. Shain, Anal Chem 1965, 37, 178.

[26] K. M. Dombek, L. Ingram, J Bacteriol 1984, 157, 233.

[27] (a) Laschi, S.; Palchetti, I.; Marrazza, G.; Mascini, M., Enzyme-amplified electrochemical hybridization assay based on PNA, LNA and DNA probe-modified micro -magnetic beads. Bioelectrochemistry 2009, 76 (1), 214-220; (b) Nie, Z.; Nijhuis, C. A.; Gong, J.; Chen, X.; Kumachev, A.; Martinez, A. W.; Narovlyansky, M.; Whitesides, G. M., Electrochemical sensing in paper-based microfluidic devices.

Lab on a Chip 2010, 10 (4), 477-483.

[28] Cassidy, J.; O'Gorman, J.; Ronane, M.; Howard, E., Note on the voltammetry of ferrocene carboxylate in aqueous solution. Electrochem Commun 1999, / (2), 69-71.

[29] Haberl, S.; Miklavcic, D.; Sersa, G.; Frey, W.; Rubinsky, B., Cell membrane electroporation-Part 2: the applications. Electrical Insulation Magazine, IEEE 2013,

29 (1), 29-37.

[30] Lovelock, K. R.; Ejigu, A.; Loh, S. F.; Men, S.; Licence, P.; Walsh, D. A., On the diffusion of ferrocenemethanol in room-temperature ionic liquids: an electrochemical study. Physical chemistry chemical physics : PCCP 2011, 13 (21), 10155-64.