Antibiotic Susceptibility Testing at a Screen-Printed Carbon Electrode

Jan 9, 2008 - Screen-printed carbon electrode arrays were treated to allow ... data is made available by participants in Crossref's Cited-by Linking s...
4 downloads 0 Views 164KB Size
Download PDF
Anal. Chem. 2008, 80, 843-848

Antibiotic Susceptibility Testing at a Screen-Printed Carbon Electrode Array Thomas S. Mann† and Susan R. Mikkelsen*

Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1

Screen-printed carbon electrode arrays were treated to allow respiratory activity-based measurement of antibiotic susceptibility with Escherichia coli JM105. Carbon working electrodes were examined for antibiotic adsorption and were pretreated with various electrochemical and chemical protocols to minimize antibiotic adsorption. Treatment by voltammetry in basic solution or by chemical modification with poly-L-lysine or chitosan were found to be effective methods for the elimination of adsorption of the studied group of 17 antibiotics, which comprised several classes and modes of action. Measurements consisted of two-electrode amperometry of the bacterial suspension after 10 min of incubation with antibiotic followed by addition of an oxidative cocktail of ferricyanide and dichlorophenolindophenol for a further 10 min; response currents, which indicate the extent of reduction of ferricyanide to ferrocyanide by cellular respiratory activity, decrease with increasing concentration of antibiotic present in the initial 10 min incubation. IC50 values obtained for chloramphenicol with these electrode modification methods are consistent at 2.0 ( 0.2 mM, in approximate agreement with previously reported respiration-based results for this organism but significantly higher than values reported for growth-based antibiotic susceptibility testing methods. Microorganisms are diverse and plentiful life forms that are known to be both beneficial and harmful to other forms of life.1-4 In the human digestive tract, for example, the normal and beneficial microbial flora can be overwhelmed by the growth of pathogenic bacteria that exclude other species, secrete toxins and ultimately cause serious and life-threatening infections.5,6 Such conditions are treated by antibiotic chemotherapy, initially on an empirical basis, and are followed by adjustments to the identity or dose of the antibiotic based on laboratory results of antibiotic * Corresponding author. E-mail: [email protected]. Fax: 519-746-0435. † Current address: Rapid Laboratory Microsystems Inc., Freeport Health Centre, P.O. Box 9056, 3570 King Street East, Kitchener, Ontario, Canada N2P 2G5. (1) Dethlefsen, L.; Eckburg, P. B.; Bik, E. M.; Relman, D. A. Trends Ecol. Evol. 2006, 21, 517-523. (2) Zoetendal, E. G.; Vaughan, E. E.; de Vos, W. M. Mol. Microbiol. 2006, 59, 1639-1650. (3) Flint, H. J.; Duncan, S. H.; Scott, K. P.; Louis, P. Environ. Microbiol. 2007, 9, 1101-1111. (4) Jean-Baptiste, E. J. Intensive Care Med. 2007, 22, 63-72. (5) Johnson, A. P. J. Antimicrob. Chemother. 1994, 33, 1083-1089. (6) Versalovic, J. Am. J. Clin. Pathol. 2003, 119, 403-412. 10.1021/ac701829c CCC: $40.75 Published on Web 01/09/2008

© 2008 American Chemical Society

susceptibility tests (ASTs) performed on purified cultures of the causative organisms.7 Clinical microbiology laboratories rely on growth-based phenotyping methods as the so-called “gold standard” for determining the identities and AST profiles of pathogenic organisms.8 These methods involve individual incubations of the isolated organism with an established range of potential metabolic substrates and a chosen series of antibiotics, for a time determined by the growth rate of the organism, to allow observation of the presence or absence of growth. The predominant commercial technologies in current use are the Vitek, MicroScan, and Phoenix systems, which have different growth media, detection systems, and automation levels.9 Over the past decade, great interest has surrounded genetic testing methods for both identification and antibiotic resistance testing of microorganisms, because of their abilities to detect unique nucleic acid sequences that can identify organisms and also detect the presence of genes that code for known antibiotic resistance-related proteins such as β-lactamase and multidrug efflux pumps.2,10 To date, however, these methods have found limited use in clinical practice. A recent review considers obstacles that have prevented their implementation in routine testing.11 Antibiotic chemotherapy is based on the rational choice of a chemical agent that will affect the offending microbes by preventing proliferation (bacteriostatic agents) or by lethal action (bactericidal agents). Growth-based AST exploits one of several possible definitions of microbial viability; it requires cellular proliferation, measured as growth in the absence or presence of the drug in question. The growth rates of individual species limit the speed of these methods to a range of several hours to days or even weeks for particularly slow-growing organisms. Alternative methods to measure viability (the ratio of living cells to total cells in a culture) have been put forward by researchers: the presence of an intact membrane,12 the presence of localized enzyme activities,13,14 cell elongation observed in the presence of a celldivision inhibitor,15 and the presence of respiration.16 (7) Baron, S., Ed. Medical Microbiology, 4th ed.; The University of Texas Medical Branch at Galveston: Galveston, TX, 1996. (8) Pfaller, M. A.; Jones, R. N.; Microbiology Resource Committee, College of American Pathologists. Arch. Pathol. Lab. Med. 2006, 130, 767778. (9) Wiegand, I.; Geiss, H. K.; Mack, D.; Stuerenburg, E.; Seifert, H. J. Clin. Microbiol. 2007, 45, 1167-1174. (10) Jannes, G.; de Vos, D. Methods Mol. Biol. 2006, 345, 1-21. (11) Mothershed, E. A.; Whitney, A. M. Clin. Chim. Acta 2006, 363, 206-220. (12) Mason, D. J.; Lopez-Amoros, R.; Allman, R.; Stark, J. M.; Lloyd, D. J. Appl. Bacteriol. 1995, 78, 309-315.

Analytical Chemistry, Vol. 80, No. 3, February 1, 2008 843

Previous work in this laboratory has shown that both aerobic and anaerobic microbial respiration can be measured using amperometry, in the presence of a mediator cocktail consisting of the oxidant ferricyanide, alone (for Gram-negative organisms) or in combination with a hydrophobic oxidant such as dichlorophenolindophenol (DCIP, for Gram-positive organisms and fungi).17,18 Further, we demonstrated that Pt working electrodes (as opposed to C or Au) best minimized adsorptive effects from the bacteria and the antibiotics studied. Antibiotic susceptibility tests were conducted using respiratory activity measurements on cell suspensions that had been growth-arrested (by cooling) and were complete within 30 min. Research in other groups has shown that disposable, goldplated, screen-printed carbon electrodes can be used in a microplate-compatible electrochemical array for the measurement of oxygen depletion with time in suspensions of aerobic bacteria and mammalian cells and that these measurements can be applied to antibiotic19 and anticancer20 drug susceptibility tests as well as bacterial classification tests.21,22 In these studies, oxygen depletion was correlated with viable cell numbers in the suspensions, and the measurements were used after long incubation periods to indicate whether or not cellular proliferation (growth) occurred. Incubations of inoculated growth media were performed in the presence of various antibiotic and isoflavanoid anticancer compounds for different times: 16.65,19 24-72,20 14.33,21 and 10 h,22 prior to the measurement of oxygen consumption as an indicator of growth. In the present work, disposable screen-printed carbon working electrode arrays, prepared by printing inexpensive carbon-based ink onto flexible polycarbonate supports, have been examined for application to AST measurements with Escherichia coli JM105. A microwell structure has been devised so that the screen-printed circular working electrodes form the bottoms of the individual test wells, allowing rapid assembly and disassembly for replacement of the working electrode arrays. Electrochemical and chemical pretreatment methods were examined for the carbon surfaces to prevent fouling by antibiotics and bacteria. Oxidative electrochemical pretreatment in basic solution was used to generate a more hydrophilic electrode surface due to the introduction of phenolic, quinoid, and carboxylate functional groups.23 Modification of the carbon surfaces with positively charged films of chitosan24 and poly-L-lysine,25 both of which have been used previously in biosensor applications, was also examined. Nega(13) Nyren, P.; Edwin, V. Anal. Biochem. 1994, 220, 39-45. (14) Breeuwer, P.; Drocourt, J.-L.; Rombouts, F. M.; Abee, T. Appl. Environ. Microbiol. 1994, 60, 1467-1472. (15) Kogure, K.; Simidu, U.; Taga, N. Can. J. Microbiol. 1979, 25, 415-420. (16) Hadjipetrou, L. P.; Gray-Young, T.; Lilly, M. D. J. Gen. Microbiol. 1966, 45, 479-488. (17) Ertl, P.; Unterladstaetter, B.; Bayer, K.; Mikkelsen, S. R. Anal. Chem. 2000, 72, 4949-4956. (18) Ertl, P.; Robello, E.; Battaglini, F.; Mikkelsen, S. R. Anal. Chem. 2000, 72, 4957-4964. (19) Kitahara, T.; Koyama, N.; Matsuda, J.; Hirakata, Y.; Kamihira, S.; Kohno, S.; Nakashima, M.; Sasaki, H. Biol. Pharm. Bull. 2003, 26, 1229-1234. (20) Andreescu, S.; Sadik, O. A.; McGee, D. W.; Suye, S. Anal. Chem. 2004, 76, 2321-2330. (21) Karasinski, J.; Andreescu, S.; Sadik, O. A.; Lavine, B.; Vora, M. N. Anal. Chem. 2005, 77, 7941-7949. (22) Karasinski, J.; White, L.; Zhang, Y.; Wang, E.; Andreescu, S.; Sadik, O. A.; Lavine, B. K.; Vora, M. Biosens. Bioelectron. 2007, 22, 2643-2649. (23) Chen, P.; McCreery, R. L. Anal. Chem. 1996, 68, 3958-3965. (24) Cruz, J.; Kawasaki, M.; Gorski, W. Anal. Chem. 2000, 72, 680-686.

844

Analytical Chemistry, Vol. 80, No. 3, February 1, 2008

tively charged film coatings, such as Nafion, were not examined due to their electrostatic rejection of the ferri-ferrocyanide redox couple,26 which is detected in our amperometric susceptibility assay. Results obtained with these screen-printed electrode arrays are in agreement with previously obtained respiratory activity results for this strain of E. coli, obtained with commercially available platinum working electrodes, but IC50 values obtained from respiratory activity inhibition are significantly larger than those obtained from growth inhibition measurements, which require much longer incubation times prior to measurement. EXPERIMENTAL SECTION Materials and Instrumentation. Chemicals were obtained in the best available quality and were used as received. Dipotassium hydrogen phosphate trihydrate, potassium dihydrogen phosphate, ammonium sulfate, ammonium chloride, potassium ferricyanide, tryptone, calcium chloride dihydrate, R-D-glucose, poly-L-lysine hydrobromide, chitosan, succinic acid, cyclohexanone, sodium perchlorate, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC), silver wire (1 mm × 50 cm), and all antibiotics were obtained from Sigma-Aldrich. Magnesium sulfate heptahydrate, trisodium citrate dihydrate, potassium chloride, sodium hydroxide, sulfuric acid, hydrochloric acid, and sodium hydrogen carbonate were supplied by BDH. Potassium ferrocyanide was obtained from Fisher Scientific. EM Science supplied ethanol and potassium nitrate. Pierce supplied N-hydroxysulfosuccinimide (NHS), whereas yeast extract powder was obtained from ICN Biomedical. E. coli JM105 was obtained from the strain collection of the Department of Biology, University of Waterloo. The Sylgard 184 silicone elastomer kit was supplied by Dow Corning. All solutions and media were prepared using distilled, deionized water (Barnstead NanoPure). Electrochemical measurements were performed using either a Princeton Applied Research potentiostat/galvanostat model 263A, or a CH Instruments model CHI650A potentiostat. All electrochemical measurements were made on solutions prepared in a pH 6.8 phosphate buffer containing 2.88 g/L KH2PO4, 5.76 g/L K2HPO4‚3H2O, 1.2 g/L trisodium citrate, 0.48 g/L MgSO4‚ 7H2O, 0.048 g/L CaCl2‚2H2O, 1.63 g/L (NH4)2SO4, and 1.34 g/L NH4Cl. A commercial Ag/AgCl reference electrode (Bioanalytical Systems) was used to prepare the homemade Ag/AgCl wires that were used as counter electrodes with the electrochemical microwell arrays. Silver wires (1 mm × 5 cm) were anodized at +0.4 V for 60 s in a 1 M KCl solution, using a stainless steel auxiliary electrode. When used as reference electrodes for cyclic voltammetry of 1 mM ferricyanide in phosphate buffer (pH 6.8), with a commercial glassy carbon working electrode (Bioanalytical Systems) and a Pt wire auxiliary electrode (Alfa Aesar) a formal potential of 106 ( 2 mV was obtained, and this value was stable over 3 months of intermittent use for array measurements. Fermentations were carried out in baffled shake flasks at 37 °C and 250 rpm, using a water bath shaker (New Brunswick Scientific). Optical density measurements were made at 600 nm using a Cary 1 double-beam UV-vis spectrophotometer. (25) Johnson, D.; Norman, S.; Tuckey, R. C.; Martin, L. L. Bioelectrochemistry 2003, 59, 41-47. (26) Shi, Y.; Seliskar, C. J.; Heineman, W. R. Anal. Chem. 1997, 69, 48194827.

Figure 1. Electrochemical microwell array design: (A) top view of carbon electrodes and conducting tracks printed on polycarbonate substrate; (B) assembly of screen-printed electrode array into holder to produce 16 sealed microwells.

Methods. Cultivation of E. coli JM105. A stock solution of trace elements for growth media was prepared in 5 N HCl (Merck) and contained the following compounds: FeSO4‚7H2O (40 g/L), MnSO4‚H2O (10 g/L), AlCl3‚6H2O (10 g/L), CoCl2‚6H2O (4 g/L), ZnSO4‚7H2O (2 g/L), Na2MoO4‚2H2O (2 g/L), CuCl2‚2H2O (1 g/L), and H3BO3 (0.5 g/L). The growth medium contained the following components: KH2PO4 (2.88 g/L), K2HPO4‚3H2O (5.76 g/L), tryptone (2.4 g/L), yeast extract powder (1.2 g/L), trisodium citrate dihydrate (1.2 g/L), MgSO4‚7H2O (0.48 g/L), CaCl2 2H2O (0.048 g/L), (NH4)2SO4 (1.63 g/L), NH4Cl (1.34 g/L), glucose (13.2 g/L), and 240 µL of the trace element stock solution per liter of medium. The glucose was prepared as a concentrated aqueous solution, sterilized separately, and added to the growth medium after the solutions reached room temperature. Inocula were previously prepared as 1 mL aliquots of 1:1 glycerol/cell culture mixture (stored at -80 °C) and were added to 50 mL of sterile growth medium in a baffled shake flask at 37 °C; cultivation proceeded at this temperature in an incubator-shaker at 200 rpm. Determination of Optical Density. Optical density was measured in triplicate. A measured aliquot of the cell suspension was centrifuged at 10 000 rpm for 10 min (Beckmann Microfuge E). The supernatant was removed, and the pellet was reconstituted in 1.00 mL of water. Sample dilution factors were chosen such that the measured optical density was between 0.1 and 0.8 au, using distilled deionized water as a reference. Preparation of Electrode Arrays. Arrays were prepared on polycarbonate substrates (250 µm thick, Cadillac Plastic Ltd., Ontario) using a DEK 248 semiautomatic screen printer and a stainless steel emulsion-coated mesh screen (Hybrid Integrated Services Inc., Ontario). The array design is shown in Figure 1A. Polycarbonate sheets (35 × 35 cm) were rinsed with ethanol and preheated for 1 h at 80 °C prior to printing, to prevent later shrinkage. Following printing with graphite ink (E-8203, Coates Screen), arrays were cured at 85 °C for 15 min and were stored at room temperature (22 ( 2 °C). Immediately prior to use, arrays were sonicated in ethanol for 2 min, rinsed with deionized water, and blotted dry with Kimwipes.

Assembly of Microwell Arrays. The screen-printed arrays were sandwiched between insulating layers to produce microwells as shown in Figure 1B. Each array consisted of 16 working electrodes and rested on a foam pad stabilized by a Teflon block. Electrodes were exposed through holes created using punch pliers in a poly(dimethylsiloxane) (PDMS) insulating layer, about 2 mm thick, cast in a mold using the silicone elastomer Sylgard 184, with a 1:10 (w/w) curing agent/elastomer ratio, cured at room temperature for 4 h, followed by 4 h at 65 °C. A second Teflon block with aligned holes was used as an upper layer to allow the structure to be clamped together, to secure the PDMS layer and create 16 sealed microwells for individual electrochemical measurements. The total volume of each microwell is approximately 120 µL. Previous work using untreated electrodes in the multiwell array shown in Figure 1 involved calibration for electrode area using chronoamperometry of a 1.0 mM ferrocyanide solution in 0.10 M KCl. Average area was found to be 0.037 cm2, with a relative standard deviation of 8%.27 Pretreatment of Working Electrodes. Electrochemical pretreatment consisted of cyclic voltammetry in 0.5 M NaOH at 50 mV/s from 0.6 to 2.0 V versus Ag/AgCl for 5 min, followed by rinsing with distilled water; voltammograms obtained following this modification showed increased background current and the presence of quinine/hydroquinone redox activity centered at about -100 mV versus Ag/AgCl in 0.5 M phosphate buffer, pH 6.88, as expected.23 Chitosan modification of electrochemically pretreated electrodes involved overnight evaporation of 10 µL of a 0.01% chitosan solution prepared in hot 0.05 M HCl and filtered through a 0.45 µM Acrodisc syringe filter (PALL Gelman Inc.); electrodes were then exposed to 100 µL of 0.1 M NaOH for 10 min, followed by 100 µL of 0.5 M phosphate (pH 6.88) for 1 h prior to use. PolyL-lysine modification used electrodes that had been previously electrochemically activated as described above: 40 µL of a solution containing 24 mM EDC and 5 mM NHS were added to each well and incubated 20 min; this was followed by addition of 30 µL of an aqueous poly-L-lysine solution (4 mM as lysine); the resulting reaction mixture was allowed to evaporate overnight, and the electrodes were then rinsed with distilled water. Antibiotic Susceptibility Assay. These studies were conducted with pure cultures of E. coli JM105, as are all standard AST methods,9 due to differing patterns of susceptibility for different organisms. Aliquots of bacteria (1.00 mL) were harvested by centrifugation (10 000 rpm, 10 min), resuspended in 1.00 mL of buffer (consisting of growth medium without glucose, tryptone, yeast extract, and trace elements), and stored in an ice bath for 20 min prior to use. Aliquots of this suspension (20 µL) were combined with 60 µL of buffer prepared in the absence or presence of antibiotic and incubated 10 min at 37 °C. Reagent solution (20 µL) containing 250 mM potassium ferricyanide and 50 mM succinate was then added, and incubation proceeded at 37 °C for 10 min. The resulting solution was transferred to one of the array wells for measurement. Chronocoulometry was performed at +0.500 V versus Ag/AgCl for 120 s. RESULTS AND DISCUSSION For later comparison, preliminary results of the effects of antibiotics on ferricyanide electron transfer were obtained with (27) Ertl, P.; Wagner, M.; Corton, E.; Mikkelsen, S. R. Biosens. Bioelectron. 2003, 18, 907-916.

Analytical Chemistry, Vol. 80, No. 3, February 1, 2008

845

Table 1. Effect of Pretreatment of Screen-Printed Carbon Electrodes on Electrode Fouling by Antibiotics cathodic peak current ratioa antibiotic (concn, mM)

nothingb

ECc

following modification with chitosanc

poly-L-lysinec

none

0.94 ( 0.09 0.98 ( 0.03 1.01 ( 0.05

1.04 ( 0.03

bacitracin (4.77) D-cycloserine (10.0) erythromycin (4.66) geneticin (4.97) hygromycin (5.06) kanamycin (9.85) neomycin (9.97) paromomycin (2.45) rifampicin (0.52) streptomycin (5.00) trimethoprim (2.75) vancomycin (4.99)

d 0.89 ( 0.06 0.88 ( 0.06 0.90 ( 0.08 1.06 ( 0.06 0.97 ( 0.02 1.05 ( 0.08 1.05 ( 0.04 0.39 ( 0.08 1.02 ( 0.11 1.31 ( 0.11 0.75 ( 0.09

0.69 ( 0.04 0.94 ( 0.03 1.00 ( 0.06 1.17 ( 0.04 1.20 ( 0.05 1.22 ( 0.10 1.25 ( 0.08 1.28 ( 0.07 0.82 ( 0.04 1.24 ( 0.18 0.99 ( 0.02 0.88 ( 0.03

0.80 ( 0.06 0.90 ( 0.06 0.71 ( 0.11 1.00 ( 0.06 0.80 ( 0.09 1.05 ( 0.17 0.98 ( 0.05 1.20 ( 0.07 0.76 ( 0.07 1.04 ( 0.04 0.87 ( 0.06 0.99 ( 0.05

0.96 ( 0.14 1.08 ( 0.06 0.83 ( 0.12 0.96 ( 0.15 1.03 ( 0.06 0.99 ( 0.03 0.80 ( 0.08 0.95 ( 0.09 0.93 ( 0.01 0.87 ( 0.06 0.95 ( 0.06 0.78 ( 0.05

chloramphenicol (5.04) nystatin (0.31)

0.70 ( 0.07

1.08 ( 0.06

1.01 ( 0.04

0.65 ( 0.02 1.02 ( 0.06 0.96 ( 0.17

0.96 ( 0.05

carbenicillin (2.65) 0.73 ( 0.05 0.93 ( 0.12 0.91 ( 0.25 cefotaxime (2.16) d 0.93 ( 0.22 0.91 ( 0.05 nalidixic acid (1.18) 0.89 ( 0.04 1.04 ( 0.12 1.02 ( 0.06

1.07 ( 0.09 0.94 ( 0.05 0.83 ( 0.04

a Measured in 1.0 mM potassium ferricyanide in phosphate buffer (pH 6.8) as the ratio of cathodic peaks obtained at 50 mV/s in the presence to the absence of antibiotic, with an incubation time of 30 min at open circuit; uncertainty is 1 s. b n ) 4. c n ) 3. d Cathodic peak was not observed following exposure to antibiotic. See the Experimental Section for modification conditions.

unmodified screen-printed carbon electrode arrays. Cyclic voltammetry of 1.0 mM ferricyanide in phosphate buffer (pH 6.8) was performed at 50 mV/s, in the absence followed by the presence of antibiotic. Control experiments were performed in the absence of antibiotics, to determine whether the first voltammetric scan conditions the carbon working electrode surface and changes the magnitude of the cathodic peak. Results are expressed as the ratio of cathodic peak currents observed in the presence to the absence of antibiotic, and are given in the first column of Table 1. The antibiotics listed in Table 1 have different charge, solubility, and mechanistic properties. The first 12 compounds in Table 1 are positively charged at neutral pH, while the following two have no charge, and the final three carry negative charge. The least soluble of the tested antibiotics are rifampicin, nystatin, and nalidixic acid; where possible, adsorption was tested at concentrations in excess of 2 mM. Comparison of peak current ratios with control measurements show that, of the 17 antibiotic compounds tested, 9 have no apparent effect on the cathodic peak current ratio for ferricyanide. For these compounds, if adsorption to the carbon surfaces occurs in the time frame of these experiments (30 min of exposure at open circuit), the adsorption neither inhibits nor enhances electron transfer. Of the remaining eight antibiotics, only one promotes ferricyanide reduction (trimethoprim), while the other seven cause significant inhibition of the cathodic process. Two of the antibiotics (bacitracin and cefotaxime) inhibit the process completely, since no cathodic peak was observed following exposure to these compounds. These effects are not likely to be caused by interactions between the antibiotics and the ferricyanide, since the antibiotics with high nominal positive charge (neomycin (+6), paromomycin (+5), geneticin (+4), kanamycin (+4), hygromycin 846 Analytical Chemistry, Vol. 80, No. 3, February 1, 2008

Figure 2. Cyclic voltammograms of 1.00 mM ferricyanide in phosphate buffer (pH 6.8) at unmodified screen-printed carbon electrodes before (solid lines) and after (dotted lines) 30 min of incubation with (A) paromomycin, (B) trimethoprim, and (C) bacitracin.

(+3), streptomycin (+3)) neither inhibit nor enhance electron transfer, whereas those that do inhibit ferricyanide reduction are positively charged (bacitracin, rifampicin, vancomycin, all +1), neutral (chloramphenicol, nystatin), and negative (cefotaxime (-1) and carbenicillin (-2)). Trimethoprim, which promotes electron transfer, carries a single positive charge. The inhibitory effect is not linked to structure in any obvious manner; for example, bacitracin contains no aromatic moieties and several charged amino acid residues, whereas rifampicin contains substituted naphthyl and aliphatic hydrocarbon components. Representative voltammograms are shown for paromomycin, trimethoprim, and bacitracin in Figure 2. The same experiment was performed following various electrode modifications to determine whether fouling by antibiotics could be reduced or eliminated by either electrochemical (cyclic voltammetry in basic solution) or chemical pretreatment (chitosan

or poly-L-lysine modification) as described in the Experimental Section. Results for these experiments are also shown in Table 1, alongside the values obtained with the unmodified screen-printed carbon electrode arrays. Figure 3 shows the effects of the different electrode pretreatments on voltammograms obtained in the absence of antibiotics after 15 min of exposure to the ferricyanide solution. Little effect was observed from electrochemical and chitosan treatment; however, poly-L-lysine modification significantly changed the peak shapes, positions, and magnitudes for the ferri-ferrocyanide redox couple. Preconcentration of the negatively charged redox-active species within the poly-L-lysine layer at the electrode surface leads to more Gaussian-shaped peaks with less separation and the characteristic appearance of voltammograms for surface-confined Faradaic reactions.28 No changes were observed in any of the voltammograms following longer (30 min) exposure to the ferricyanide solution. Pretreatment of the carbon screen-printed electrodes has a significant effect on fouling by antibiotics, as indicated by the peak current ratios shown in Table 1. Comparison of each ratio obtained for an antibiotic in Table 1 with results for the control experiment (first row of Table 1) allows selection of the most appropriate electrode pretreatment conditions for the elimination of electrode fouling by that antibiotic. Fouling observed by neutral and negatively charged antibiotics (the last five entries in Table 1) was eliminated by all three pretreatment methods, with the single exception of nalidixic acid at poly-L-lysine-modified electrodes. For the positively charged antibiotics, more diversity was observed. For example, the fouling observed at unmodified electrodes by bacitracin and rifampicin is minimal after poly-L-lysine modification, whereas fouling by vancomycin is removed following chitosan modification. The signal enhancement observed at unmodified electrodes following trimethoprim exposure is eliminated by electrochemical or poly-L-lysine pretreatment. Application of these modified electrodes to antibiotic susceptibility testing of bacterial suspensions requires a linear electrochemical response to ferrocyanide concentration, since bacterial respiratory activity causes reduction of ferricyanide to ferrocyanide, which is then detected by its electrochemical oxidation. Calibration data (not shown) obtained by chronocoulometry of ferrocyanide at +0.50 V versus Ag/AgCl (integration between 60 and 120 s) at electrodes modified with chitosan or poly-L-lysine showed excellent linearity (R2 g 0.994) over the 10 µM to 2.00 mM concentration range, in both the absence and presence of 50.0 mM ferricyanide. Chloramphenicol was selected to demonstrate quantitative susceptibility testing of E. coli JM105 suspensions at electrodes pretreated electrochemically and with the chemical pretreatments, since fouling of the electrodes by chloramphenicol was eliminated by all three methods (see Table 1). Thus, if fouling occurred, it could be attributed to the bacteria; however, no fouling was observed under the conditions of these experiments. Aliquots of bacterial suspensions were incubated 10 min in phosphate buffer, with or without chloramphenicol, and a further 10 min after addition of succinate and ferricyanide to final concentra-

Figure 3. Cyclic voltammograms of 1.00 mM ferricyanide in phosphate buffer (pH 6.8) after (A) no modification, (B) electrochemical pretreatment, (C) chitosan modification, and (D) poly-L-lysine modification. Scans were performed at 50 mV/s immediately upon exposure to ferricyanide solution (solid lines) and after 15 min of incubation (dotted lines).

(28) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd ed.; Wiley: New York, 2001; pp 590-593.

tions of 10 mM and 50 mM, respectively, and were then transferred to the electrode array wells for chronocoulometric Analytical Chemistry, Vol. 80, No. 3, February 1, 2008

847

Figure 4. Respiratory activity results obtained at chitosan-modified screen-printed carbon electrodes for suspensions of E. coli JM105 following 10 min of incubation with antibiotic followed by 10 min of incubation with 50 mM ferricyanide: (A) raw chronocoulometric traces for (a) no chloramphenicol, (b) 0.62 and (c) 5.0 mM chloramphenicol, and (d) no chloramphenicol and no bacteria; (B) percent respiratory activity vs log[chloramphenicol] for chitosan-modified electrodes. See text for details.

oxidation of the ferrocyanide produced by bacterial respiratory activity. Figure 4A shows typical chronocoulometric traces obtained at chitosan-modified electrodes, including a control obtained in the absence of antibiotic and a second control obtained in the absence of bacteria. These experiments were performed using a range of chloramphenicol concentrations, and the difference in chronocoulometric charge between 60 and 120 s was calculated for each trace. Percent respiratory activity was then obtained at each chloramphenicol concentration by dividing each value by the value obtained with no chloramphenicol present (Figure 4A,

848 Analytical Chemistry, Vol. 80, No. 3, February 1, 2008

trace a). The resulting plot of percent activity against log[chloramphenicol] is shown in Figure 4B, where the Hill equation29 has been fit to the experimental data. The Hill equation parameters include the IC50, which is the concentration at which 50% of the maximum inhibition level is reached; this value is 2.3 mM (log[IC50] ) 3.2) for the curve shown in Figure 4B. Similar results were obtained using an identical procedure at screen-printed electrode arrays that had been modified electrochemically (IC50 ) 2.0 mM) and chemically with poly-L-lysine (IC50 ) 1.7 mM). Finally, the need for separate incubation and transfer steps prior to electrochemical measurement was examined at chitosanmodified electrode arrays. In these experiments, the two 10 min incubation steps were carried out in the array wells, and the transfer step was eliminated. The resulting IC50 value for chloramphenicol, 2.1 mM, is well within the range obtained using separate incubations followed by a transfer to the modified electrode array wells just prior to measurement. Values obtained in this work for the chloramphenicol IC50 with E. coli JM105 are about twice the value of 0.91 mM obtained previously with this organism using commercially available Pt working electrodes in a similar respiratory activity assay, and all respiratory values are more than 2 orders of magnitude higher than the value of 3.2 µM obtained using a standard 10 h growthbased dilution method.18 We attribute the large discrepancy between methods to the kinetics of uptake and action of the drug; the concentrations in the external media are known, but the intracellular concentrations and the speed with which the drug is able to affect respiratory activity have not yet been determined. In such a rapid assay, it is perhaps reasonable to expect that the apparent IC50 values would be quite large. Further investigations are underway involving antibiotics with differing mechanisms of action as well as a wide range of organisms with known human pathogenicity. ACKNOWLEDGMENT Financial support from the Natural Science and Engineering Research Council of Canada is gratefully acknowledged. Received for review August 30, 2007. Accepted November 3, 2007. AC701829C (29) Voet, G.; Voet, J. G. Biochemistry, 3rd ed.; Wiley: New York, 2004; pp 323324.