Bioluminescent Assay for Heroin and Its Metabolites - American

Illicit heroin is trafficked as a solid particulate drug, while heroin abuse is monitored by testing urine samples for its principal metabolites, morp...
0 downloads 0 Views 213KB Size
Anal. Chem. 1996, 68, 1877-1882

Bioluminescent Assay for Heroin and Its Metabolites Peter-John Holt, Neil C. Bruce, and Christopher R. Lowe*

Institute of Biotechnology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QT, United Kingdom

Illicit heroin is trafficked as a solid particulate drug, while heroin abuse is monitored by testing urine samples for its principal metabolites, morphine and morphine-3glucuronide. Two novel bacterial enzymes were used in the development of a linked-enzyme assay for heroin and its metabolites: heroin esterase, which converts heroin to morphine, and morphine dehydrogenase, which oxidizes morphine to morphinone with the concomitant reduction of NADP+. A bioluminescent assay involving heroin esterase, morphine dehydrogenase, and the bacterial luciferase from Vibrio harveyi was developed and shown to be sensitive to 89 ng/mL heroin and 2.0 ng/ mL morphine. Excellent correlation with the results from 83 authentic samples submitted for urine drug screening at a hospital laboratory was obtained. The bioluminescent assay exhibited greater specificity and speed than current immunological screening methods. A novel format of the bioluminescent assay involving immobilized enzymes was sensitive to 101 ng (250 pmol) of heroin and responded well to particulate heroin. This form of the test was sensitive enough to respond to one or two typical particles of illicit heroin. Possession and use of many drugs, including heroin, is illegal in Britain and defined by the Misuse of Drugs Act (1971). Heroin enjoys a similar status throughout most of Europe and in many other countries, including the U.S.A.1 Consequently, numerous agencies world-wide routinely and frequently need to detect and identify heroin, both in the form of particulate drug preparations per se and as metabolites in biological fluids.1,2 Illicit heroin is trafficked, sold, and prepared for consumption in many forms, although mainly as solid particulate matter of greatly varying appearance and chemical composition.1,3,4 Law enforcement agencies, therefore, need to be able to detect and identify solid particulate heroin.1 The Marquis test provides a relatively fast screening method for unidentified particulate matter once isolated but does not provide a facility for the general screening of cargo, baggage, and clothing for concealed heroin.1,2,5 Several highly sophisticated attempts are being made to provide instrumentation to perform this task, but none is currently deemed adequate for routine use.1 The availability of particulate drug for collection by a vacuum or suction device and subsequent delivery (1) Gough, T. A. The Analysis of Drugs of Abuse, 1st ed.; Wiley: Chichester, England, 1991. (2) Moffat, A. C. Clarke’s Isolation and Identification of Drugs of Abuse, 2nd ed.; The Pharmaceutical Press: London, England, 1986. (3) O’Neil, P. J.; Pitts, J. E. J. Pharm. Pharmacol. 1992, 44, 1-6. (4) White, P. C.; Jane, I.; Scott, A.; Connett, B. E. J. Chromatogr. 1983, 265, 293-300. (5) Masoud, A. N. J. Pharm. Sci. 1975, 64, 841-844. S0003-2700(95)01207-8 CCC: $12.00

© 1996 American Chemical Society

to an analytical instrument has been demonstrated,1 while, more recently, analysis of illicit heroin samples has shown that a typical particle contains ∼140 pmol of heroin, a value which can be used as a guide to the desired detection capability of any method designed to detect illicit heroin.6 Drug abuse is indicated by the detection of the major urinary metabolites of heroin, namely morphine and morphine-3-glucuronide.1,2,7,8 Corporate and professional screening of urine samples in the form of preemployment, regular on-going, or random testing is an expanding area of testing for illicit drug use.9,10 Detection of urinary metabolites can also be used to identify people attempting to traffick the drug as an internal body concealment.1,11 Thin-layer chromatography (TLC) and the Syva Co. Ltd. EMIT opiate dau assay (EMIT) are currently the most extensively used urine drug screening methods.1,2,7-11 For most of the circumstances under which tests for heroin or its metabolites need to be carried out, an optimum system is not yet available. Although this is especially true for the detection of concealed illicit heroin, improvements could also be made in some of the other tests. For example, the detection limits of the most widely employed urine screening methods, TLC and EMIT, do not reach the minimum levels currently accepted as positive in urine analysis.1,12,13 The National Institute on Drug Abuse (NIDA) in the U.S.A. defines this value to be 0.1 µg mL-1 morphine,1 although the view exists that this could be lowered to 0.05 µg mL-1 morphine.1 Much effort has been devoted to developing biosensors for numerous drugs, although no system for heroin or morphine has yet been reported.14 However, several novel enzymes involved in the microbial degradation of alkaloids have been identified in this laboratory.15-21 An acetylmorphine carboxyesterase (heroin esterase) capable of hydrolyzing heroin and 6-acetylmorphine to (6) Holt, P.-J. In press. (7) Hassan, F. M. In Methods in Clinical Chemistry, 1st ed.; Pesce, A. J., Kaplan, L. A., Eds.; The C. V. Mosby Co.: St. Louis, MO, 1987; pp 368-383. (8) Coyle, D. E. In Methods in Clinical Chemistry, 1st ed.; Pesce, A. J., Kaplan, L. A., Eds.; The C. V. Mosby Co.: St. Louis, MO, 1987; pp 409-416. (9) Alexander, I. Med. Lab. World 1992, October, 19-20. (10) Wu, A. H.; Wong, S. S.; Johnson, K. G.; Callies, J.; Shu, D. X.; Dunn, W. E.; Wong, S. H. J. Anal. Toxicol. 1993, 17, 241-245. (11) McCleave, N. R. Med. J. Aust. 1992, 159, 750-754. (12) Gough, T. A.; Baker, P. B. J. Chromatogr. Sci. 1982, 20, 289-329. (13) Syva EMIT st and dau Product Information, Syva International, Maidenhead, England, 1992. (14) Guilbault, G. G.; Schmid, R. D. Biotechnol. Appl. Biochem. 1991, 14, 133145. (15) Britt, A. J.; Bruce, N. C.; Lowe, C. R. J. Bacteriol. 1992, 174, 2087-2094. (16) Lister., D. L.; Sproule, R. F.; Britt, A. J.; Lowe, C. R.; Bruce, N. C. Appl. Environ. Microbiol. 1996. 62, 94-99. (17) Cameron, W. W.; Jordan, K. N.; Holt, P.-J.; Baker, P. B.; Lowe, C. R.; Bruce, N. C. Appl. Environ. Microbiol. 1994, 60, 3881-3883. (18) Holt, P.-J. Development of Enzyme Assays for Heroin and Morphine. Ph.D. Thesis, University of Cambridge, England, 1995; Chapter 4.

Analytical Chemistry, Vol. 68, No. 11, June 1, 1996 1877

Figure 1. Schematic of the bioluminescent enzyme assay for heroin and its metabolites.

morphine has been purified and characterized from a Rhodococcus sp. H1.17,18 Furthermore, a highly specific NADP+-dependent morphine dehydrogenase capable of oxidizing morphine to morphinone has been purified and characterized from Pseudomonas putida M10.19-21 Cloning of the structural genes for both morphine dehydrogenase, morA, and heroin esterase, her, with subsequent overexpression of the recombinant enzymes in strains of Escherichia coli, has also been achieved.22-24 A recently reported electrochemical assay based on the enzymes heroin esterase and morphine dehydrogenase represents the first enzymatic method for determining heroin and morphine.25 Unfortunately, however, this method was not able to attain the sensitivity required for either a heroin detection system or a urine screening test.25 The aim of the present work was to develop a more sensitive enzymatic assay for heroin and its metabolites that could not only offer some advantages over established methods but also detect particulate heroin. The bioluminescent approach described in this paper exploits the enzymes heroin esterase and morphine dehydrogenase linked to bacterial luciferase from Vibrio harveyi (Figure 1). EXPERIMENTAL SECTION Enzymes and Reagents. Morphine dehydrogenase (morphine:NADP+ oxidoreductase, MDH) with a specific activity of 80 units/mg was prepared from a recombinant strain of E. coli.22 Acetylmorphine carboxyesterase (heroin esterase) with a specific activity of 50-60 units/mg was prepared from Rhodococcus sp. (19) Bruce, N. C.; Wilmot, C. J.; Jordan, K. N.; Trebilcock, A. E.; Gray Stephens, L. D.; Lowe, C. R. Arch. Microbiol. 1990, 154, 465-470. (20) Bruce, N. C.; Wilmot, C. J.; Jordan, K. N.; Gray Stephens, L. D.; Lowe, C. R. Biochem. J. 1991, 274, 875-880. (21) Bruce, N. C.; Willey, D. L.; Coulson, F. W.; Jeffery, J. Biochem. J. 1993, 299, 805-811. (22) Willey, D. L.; Caswell, D. A.; Lowe, C. R.; Bruce, N. C. Biochem. J. 1993, 290, 539-544. (23) Rathbone, D. A.; Holt, P.-J.; Bruce, N. C.; Lowe, C. R. Ann. N. Y. Acad. Sci. In press. (24) Rathbone, D. A.; Holt, P.-J.; Lowe, C. R.; Bruce, N. C. Submitted to Biochem. J. (25) Holt, P.-J.; Gray Stephens, L. D.; Bruce N. C.; Lowe, C. R. Biosens. Bioelectron. 1995, 10, 517-526.

1878

Analytical Chemistry, Vol. 68, No. 11, June 1, 1996

H1.17,18 Purified morphine dehydrogenase and heroin esterase were dialyzed against ammonium acetate (25 mM; pH 7.0), freezedried, and stored at -20 °C. The enzymes were reconstituted in potassium phosphate buffer (100 mM; pH 7.8) at 4 °C prior to use. β-D-Glucuronidase (β-D-glucuronide glucuronosohydrolase; EC 3.2.1.31, 1870 units/mg) type VII-A from E. coli, and partially purified luciferase (EC 3.14.14.3) from V. harveyi containing NAD(P)H:FMN oxidoreductases, with no defined activities, were obtained from Sigma (Poole, Dorset, UK). Diacetylmorphine hydrochloride (heroin), morphine hydrochloride, and codeine hydrochloride were obtained from Macfarlan Smith Ltd. (Edinburgh, UK). 6-Acetylmorphine hydrochloride was kindly provided by Dr. M. McPherson, Macfarlan Smith Ltd. All other chemical reagents were the highest grade obtainable from Sigma, Aldrich (Gillingham, UK), or BDH (Poole, Dorset, UK). All aqueous buffers and solutions were prepared using ELGA Elgastat UHP purified water (dH2O). All solutions for high-performance liquid chromatography were passed through a 0.2 µm filter under vacuum and degassed before use. Spectrophotometry. All spectrophotometric work was performed using a Shimadzu UV-2101PC UV-visible spectrophotometer including a CPS-260 six-cell holder with Peltier control of the cell temperature to (0.1 °C (VA Howe, Banbury, Oxon, UK). Luminometry. Luminescence measurements were made using a BioOrbit 1250 luminometer and display unit (Labsystems Group UK Ltd., Basingstoke, UK) with cell temperature control. Protein Assay. Protein was determined using an adaptation of the method of Bradford,27 using bovine serum albumin (BSA) solutions of known concentrations as standards and Coomassie Blue G-250 dye reagent (Pierce Ltd., Rockford, IL). Heroin Esterase Assay. Activity against 6-acetylmorphine, heroin, or morphine-3-glucuronide was determined by incubating 20 µL of esterase preparation (∼0.4 unit/mL) with 1 µmol of substrate in bicine buffer (50 mM; pH 8.0) of 1.0 mL total volume for 10 min at 30 °C. Reactions were stopped by the addition of (26) Poochikian, G. K.; Cradock, J. C. J. Chromatogr. 1979, 171, 371-376. (27) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.

2.0 M trichloroacetic acid (10 µL), centrifuged to remove protein precipitate, and assayed by high-performance liquid chromatography (HPLC). Separations of heroin, 6-acetylmorphine, morphine3-glucuronide, and morphine were performed using a Waters 600 HPLC system (Waters Millipore UK Ltd.), comprising a 600E system controller, a 484 UV absorbance detector set to 235 nm with 0-1 V fsd, and a WISP 712 autosampler set to inject 50 µL per sample. The column used was a 4.6 mm × 250 mm C18 5 µm Spherisorb-ODS column (Anachem Ltd., Luton, UK), with a 4.6 mm × 40 mm guard column containing the same packing. Data were processed using Maxima 820 software (Waters Millipore). Samples were diluted 1:9 (v/v) with mobile phase and assayed in triplicate using the method of Poochikian and Cradock.26 Quantitation of heroin, 6-acetylmorphine, and morphine was by reference to peak area standard curves over the concentration range 0-1.0 mM for each compound. Samples containing opiate concentrations above 1.0 mM were diluted to within the range of the calibration curves prior to assay. A unit of heroin esterase activity is defined as the amount of enzyme required to convert 1.0 µmol of 6-acetylmorphine to morphine per minute at pH 8.5 and 30 °C.17 β-D-Glucuronidase Assay. β-D-Glucuronidase activity was determined by a modification of the method of Fishman,28,29 as described by Sigma.30 A “modified Fishman” unit of glucuronidase activity is defined as the amount of enzyme required to liberate 1 µg of phenolphthalein from phenolphthalein glucuronide per hour at 40 °C. The activity of β-D-glucuronidase against morphine-3glucuronide, over a range of pH values, temperatures, and times, was determined in the same manner as heroin esterase activity against morphine-3-glucuronide. Morphine Dehydrogenase Assay. Morphine dehydrogenase activity was measured using an adaptation of a previously reported method.20,21 An aliquot of enzyme (∼10 µL; 30-40 units/mL) was added to 2.5 mM NADP+ with 2.5 mM morphine-HCl in Bis-Tris propane buffer (50 mM; pH 9.5) to a total volume of 1.0 mL at 30 °C, and the increase in absorbance at 340 nm was monitored. A unit of morphine dehydrogenase activity is defined as the amount of enzyme required to reduce 1.0 µmol of NADP+ per minute at pH 9.5 and 30 °C.20,21 Analysis of Urine Samples. Samples of 83 urine specimens (2-3 mL) submitted to the Norfolk & Norwich Hospital Department of Chemical Pathology for drug screening were kindly provided by Dr. Christine Dawson. These samples were identified by sequential numbering only. Results from TLC and EMIT analyses of the samples performed at the Department of Chemical Pathology were also provided. The pH of each sample was adjusted to 7.5-8.5 by measuring the sample pH with color test strips calibrated to 0.5 pH unit (either Sigma pH Test Strips, range pH 4.5-10.0, or BDH Indicator Strips, range pH 5-10.0) and adding either phosphoric acid or potassium hydroxide dropwise until a fresh test strip recorded a pH of 7.5-8.5. Urine samples were assayed in each of two ways, using the optimal assay conditions determined for morphine and morphine-3-glucuronide, respectively. For morphine, each sample (20 µL) was included in the assay solution, and morphine dehydrogenase was added after stabilization of the background signal (∼1-1.5 min; 0-0.2 (28) Fishman, W. H. J. Biol. Chem. 1948, 173, 449-456. (29) Fishman, W. H. Methods Biochem. Anal. 1967, 15, 77-145. (30) Sigma product information sheet, product no. G-7646; β-Glucuronidase (βD-glucuronide glucuronosohydrolase; EC 3.2.1.31, 1870 units/mg) type VII-A from Escherichia coli.

mV). For morphine-3-glucuronide, β-D-glucuronidase hydrolysis was used initially to convert any morphine-3-glucuronide to free morphine: Sample (50 µL) was incubated with β-D-glucuronidase (50 µL; 10 000 units/mL) in potassium phosphate buffer (200 mM; pH 6.8) for 20 min at 37 °C. Incubated sample (20 µL) was then included in the assay solution, and morphine dehydrogenase was added after stabilization of the background signal (∼1-1.5 min; 0-0.2 mV). Enzyme Immobilization. Crystal polystyrene strips 10 mm × 40 mm × 1.2 mm were cut from a 100 mm × 100 mm × 1.2 mm square (Goodfellows Ltd., Cambridge, UK) using a heated cutting wire, and the edges were filed smooth. Sections (10 mm2) of Hybond-N+ membrane (Amersham International plc, Bucks, UK) were cut and placed over one end of each of the polystyrene strips. For a morphine test, an enzyme solution comprising 10 mg of Sigma luciferase preparation (∼5 mg of protein), 200 µL of FMN (10 mM), 200 µL of morphine dehydrogenase (25 units/ mL), and 200 µL of NADP+ (50 mM) in a total volume of 1.2 mL was prepared. All the stock solutions used were made up in potassium phosphate buffer (100 mM; pH 7.8) containing 1% (w/ v) BSA, 10 mM EDTA, and 5% (v/v) glycerol. Twenty of the square membrane sections were immersed in this volume of the solution at the bottom of a concave evaporating dish, resting on ice, for 1 h. The membrane squares were then moved to the edge of the dish, allowed to dry for 2 h, and placed at one end of a polystyrene strip, and then 6 µmol of decanal (10 µL of decanal diluted 1:10 (v/v) with ethanol) was applied to the surface of each membrane square. This procedure bonded the membrane to the polystyrene substrate within seconds and evaporated to dryness in a few minutes. For a heroin test, 200 µL of heroin esterase solution (50 units/mL) was included in the enzyme solution. The morphine-3-glucuronide test included 200 µL of β-D-glucuronidase solution (10 000 units/mL) instead of heroin esterase. RESULTS Optimization of the Bioluminescent Assay. The concentrations of the reagents proved critical to the overall performance of the assay. Optimization of the reagents for the assay was conducted in reverse order, i.e., the luminescent stage first, followed by the inclusion of morphine dehydrogenase and NADP+ for morphine assay, the inclusion of β-D-glucuronidase for morphine3-glucuronide assay, and, finally, the inclusion of heroin esterase for heroin determination (Figure 1). Both heroin esterase and morphine dehydrogenase exhibit poor thermal stability at temperatures above 30 °C and impaired activity at temperatures e25 °C;18 hence, a temperature of 30 °C was selected for all luminescence assays. A potassium phosphate buffer (0.1 M) was selected because of its stabilizing effects on luciferase at this concentrations31 and known compatibility with heroin esterase, morphine dehydrogenase, and β-D-glucuronidase.18,30 The optimum concentration of each component of the luminescent reaction was determined by varying each component of the assay in turn and measuring the response to 0.1 mM NADPH. These proved to be 1.6 mg of luciferase/oxidoreductase preparation and 50 µM FMN in the potassium phosphate buffer (100 mM; pH 7.8) containing decanal (1 µL/L; 6 µM) in a 1.0 mL total volume. The morphine assay required the addition of 750 µM NADP+ and 0.5 unit of morphine dehydrogenase to the optimized luminescence assay in a total volume of 1.0 mL. (31) Hastings, J. W.; Baldwin, T. O.; Nicoli, M. Z. Methods in Enzymology Vol. LVII; Academic Press: London, England, 1978; pp 135-152.

Analytical Chemistry, Vol. 68, No. 11, June 1, 1996

1879

Figure 2. Time course profile for the luminescent opiate assays. Morphine, codeine, and heroin (1 µM) were assayed using the optimized assays described. A suitably diluted sample of the relevant opiate (50 µL) was included in the assay solution, and morphine dehydrogenase (0.5 unit) was added to initiate the reaction after stabilization of the background signal (∼1-1.5 min; 0-0.2 mV). Similar responses were observed for each compound.

Incorporation of β-D-glucuronidase directly with the other reagents to assay morphine-3-glucuronide did not prove viable: The optimal conditions of pH 6.8 and 37 °C for β-D-glucuronidase activity30 were incompatible with the other reagents. It was also found that inclusion of sufficient β-D-glucuronidase activity to cause rapid hydrolysis of morphine-3-glucuronide directly in the luminescent assay severely reduced the observed signal. Consequently, in order to assay samples containing morphine-3glucuronide effectively, samples (50 µL) were preincubated with β-D-glucuronidase (50 µL; 10 000 units/mL) in potassium phosphate buffer (200 mM; pH 6.8) for 20 min at 37 °C prior to assay for morphine. Under these conditions, ∼60-70% of the morphine3-glucuronide was converted to morphine, and a valid calibration curve or detection limit for the luminescent assay of morphine3-glucuronide could not be determined. Kinetic data for heroin esterase highlighted this enzyme as the rate-limiting step of the assay,18 and excess esterase was required. For the assay with 0.5 unit of morphine dehydrogenase, the minimum amount of heroin esterase observed to give a fast response to 0.1 mM heroin was 2.0 units of the enzyme (7.8 units/ mg). Performance of the Solution Assay. Figure 2 shows a typical response curve when a suitably diluted sample of the relevant opiate (50 µL) was included in the assay solution and morphine dehydrogenase was added to initiate the reaction after stabilization of the background signal (∼1-1.5 min; 0-0.2 mV). The brief spike in the initial stabilization period was observed for blank, control, and test samples alike and was assumed to be a mixing effect only. Peak responses were recorded 2-3 min after the addition of morphine dehydrogenase (Figure 2), and these values were used to prepare the response curves to morphine, codeine, and heroin (Figure 3) over the final concentration range 0-1 µM using the optimized assay described above. Blank and background signals were not observed. Codeine and morphine could be quantitated up to 1.0 µM and heroin up to 15 µM. Detection limits of 2S above zero were calculated, using values of s for all the data points, where 1S ) (∑s2/2n)1/2, with s representing the standard deviation of each triplicate assay and n 1880 Analytical Chemistry, Vol. 68, No. 11, June 1, 1996

Figure 3. Calibration curves for codeine, morphine, and heroin. Various opiate concentrations were added to the reagents optimized for morphine assay, and peak luminometer responses were recorded after the addition of the relevant opiate. Each opiate concentration was assayed in triplicate, and the calibration curves obtained for (a) morphine (9) and codeine (b) and (b) heroin (2) are shown. Error bars represent 1 standard error, although these are small and can be obscured by the plot symbols.

the number of assays.32 The detection limits calculated on this basis were 1.05 ng/mL (3.5 nM) codeine, 1.7 ng/mL (6 nM) morphine, and 89 ng/mL (240 nM) heroin. Coefficients of variation for the assays (n ) 3) were as follows: codeine, 1% at 500 nM and 10% at 10 nM; morphine, 3% at 500 nM and 9% at 10 nM; heroin, 9% at 10 µM and 11% at 1.0 µM. Analysis of Urine Samples. The urine samples were tested using the protocols for morphine and morphine-3-glucuronide. Each assay required 1-2 min of stabilization before addition of the morphine dehydrogenase. The values recorded were the maximum responses (in mV) reached 3 min after addition of morphine dehydrogenase to trigger the reaction. Resolution of the observed signal was excellent: a rise of 0.1 mV above background represented a clear positive signal. All the samples were tested in triplicate, and the values are quoted (1 standard error. Of the urine samples tested with the luminescent morphine assay, many specimens that contained morphine were identifiable without recourse to the β-D-glucuronidase hydrolysis step. Some specimens, however, were identifiable only after hydrolysis or gave a significantly stronger signal following hydrolysis. A urine drug (32) Fraser, C. G.; Wilde, C. E. Commun. Lab. Med. 1986, 2, 1-5.

screening service will regularly encounter samples that contain either very low total drug concentrations or drug that approaches 100% conjugation. Specimens obtained more than 72 h after the last heroin/morphine dose was taken are especially likely to fall into these categories.2 A step such as β-D-glucuronidase hydrolysis will always be required to determine morphine-3-glucuronide with a morphine dehydrogenase-based method, as morphine dehydrogenase has no activity against conjugated morphine.33 Codeine-containing samples were detected more readily than morphine-containing samples, partly because codeine is excreted in urine primarily as free drug, rather than the glucuronide,2 and partly because morphine dehydrogenase has an apparent Km for codeine ∼10-fold less than that for morphine.21 Correlation between the luminescence assay and the TLC and EMIT results from the pathology laboratory is shown in Figure 4. Although little conclusive information can be determined from such plots, some general trends are apparent: negative EMIT results correlate with negative luminescent results, high EMIT results largely correlate with high luminescence results, and luminescence results of g30 mV are observed only for samples containing codeine. Of the 37 specimens that did not respond to EMIT or reveal any opiates by TLC, all tested negative with the luminescence assay. Of the 23 samples showing morphine or codeine by TLC, all gave a response to the luminescence assay, while of the 16 samples that responded to EMIT but showed no opiates by TLC, 12 also gave a response to the luminescence assay. Furthermore, of the 7 samples that showed dihydrocodeine as the only opiate on TLC and consequently responded to EMIT, 6 gave no response to the luminescence assay. It is unclear why one sample produced a signal with the luminescence assay. As the luminescent assay is extremely sensitive to codeine (1.5 ng/ mL; 3.5 nM), it is possible that low levels of codeine were present in the sample which remained undetected or unresolved by TLC and which would not be distinguished from dihydrocodeine in the positive EMIT response. Immobilized Enzyme Tests. The prepared enzyme test strips were retained at 4 °C for a maximum of 3 h prior to use. The strips were slotted inside a luminometer cuvette with the exposed side of the membrane facing outward, such that 5 µL of a suitably diluted opiate solution could be applied to the membrane. Light produced would be emitted on the distal side of the membrane, through the polystyrene to reach the photomultiplier tube detector of the luminometer. Three measurements of each of a range of opiate concentrations were made using fresh test strips for each measurement, and calibration curves were obtained (Figure 5). Morphine and morphine-3-glucuronide could be quantitated up to 10 µmol and heroin up to 1.0 µmol. The 2S detection limit for morphine was 10 pmol. A 2S limit is valid in the case of morphine, as a blank value of zero was obtained.32 However, for heroin and morphine-3-glucuronide, mean blank values of 5.0 and 3.5 mV, respectively were observed. When a blank signal is obtained, the detection limit is defined as mean blank signal +2.8S,32 which gives detection limits of ∼250 pmol of heroin and morphine-3-glucuronide, respectively. It is not valid to quote the sensitivity of such a test directly in terms of a concentration, but if the relevant amounts of opiates were dissolved in 1.0 mL, the concentrations would be 3 µg/mL (10 nM) morphine, 92 µg/mL (250 nM) heroin, and 115 µg/mL (250 (33) Bruce, N. C.; Caswell, D. A.; French, C. E.; Hailes, A. M.; Long, M. T.; Willey, D. L. Ann. N. Y. Acad. Sci. 1994, 721, 85-99.

Figure 4. Correlation between EMIT and bioluminescent methods for opiates. The bioluminescence test responses to the 83 urine screening samples tested are shown plotted against the EMIT AU values for the samples. An EMIT AU value g300 is regarded as positive. All samples are represented in (a), which shows that samples generating a luminescent response of g30 mV all contain codeine. The expanded section (b) shows bioluminescent responses e2.5 mV and illustrates the distinction between opiate-negative and opiatepositive bioluminescence responses. The samples are classified according to their EMIT and TLC results and plotted with different symbols: +, no drugs by TLC, EMIT negative; ×, morphine by TLC, EMIT positive; O, dihydrocodeine by TLC, EMIT positive; 2, codeine by TLC, EMIT positive; ], no drugs by TLC, EMIT positive.

nM) morphine-3-glucuronide. Coefficients of variation for the assays (n ) 3) were as follows: heroin, 4% at 250 µmol and 9% at 250 pmol; morphine, 3% at 250 nmol and 9% at 25 pmol; morphine3-glucuronide, 9% at 250 nmol and 11% at 25 pmol. No method for delivering a measured amount of particles containing nanomole to picomole amounts of drug has been described. Weighed amounts of drug can be used, but for the range 1-50 particles of ∼45 µm diameter, the weights concerned were too low to be determined accurately in this laboratory (∼5 × 10-8-2 × 10-6 g). However, using a microspatula, amounts of heroin barely visible to the naked eye were delivered to the surface of several test strips. With test strips that were still dry, little or no response was seen. Using test strips that had been moistened with 10 µL of dH2O, strong responses were observed. Dry test strips stored at 4 °C overnight displayed up to 60% of the original activity after being moistened. Longer term storage conditions have not been investigated. Since the amounts of drug delivered Analytical Chemistry, Vol. 68, No. 11, June 1, 1996

1881

Figure 5. Calibration curves for the immobilised enzyme opiate assay. Heroin (0), morphine (]), and morphine-3-glucuronide (O) were assayed with immobilized enzyme test strips. Each opiate concentration was assayed in triplicate, with error bars representing 1 standard error plotted, although these are small and obscured by the plot symbols.

were not identical in each case, absolute values for these responses were not recorded, and no estimates of error or reproducibility could be made. DISCUSSION The detection limit for morphine (1.7 ng/mL) of the luminescent test is substantially lower than the lowest level of morphine currently accepted as positive in a urine sample (0.1 µg/mL) and is lower than the limit attainable with either TLC (1.0 µg/mL) or EMIT 0.3 µg/mL). Incorporation of β-D-glucuronidase directly into the assay to enable the detection of morphine-3-glucuronide proved unviable as it also was in the amperometric enzyme assay of this compound.25 Preincubation of glucuronidase with samples containing morphine-3-glucuronide and subsequent assaying for morphine proved viable, although irreproducible. Reported attempts to use glucuronidase hydrolysis to liberate morphine from its conjugated form have sometimes required harsh conditions34,35 or have also proven unsuccessful.36 However, incorporation of heroin esterase in the assay established a method sensitive to 89 ng/mL (240 nM) heroin. The luminescence assay is highly specific for a narrow spectrum of morphine alkaloids. For example, the specificity attainable with the luminescence method compared to that possile with EMIT has been demonstrated with respect to the presence of dihydrocodeine in clinical samples. Twelve compounds described by Syva as “being members of a specific assay class” give (34) Johansen, M.; Rasmussen, K. E.; Christophosen, A. S. J. Chromatogr. 1990, 532, 277-284. (35) Chen, Z. R.; Reynolds, G.; Bochner, F.; Somogyi, A. J. Chromatogr. 1989, 493, 313-324. (36) Bertholf, R. L.; Adamson, D. T.; Vellom, D. C.; Rosen, F.; Epstein, L.; Seegmiller, J. E. Anal. Biochem. 1991, 197, 266-272. (37) Singh, A. K.; Granley, K.; Mirsha, U.; Naeem, K.; White, T.; Jiang, Y. Forensic Sci. Int. 1992, 54, 9-22. (38) Mule, S. J.; Whitlock, E.; Jukofsky, D. Clin. Chem. 1974, 20, 243-248. (39) Weast, R. C., Ed. CRC Handbook of Chemistry and Physics, 67th ed.; CRC Press Inc.: Boca Raton, FL, 1987. (40) Haggerty, C.; Jablonski, E.; Stav, L.; DeLuca, M. Anal. Biochem. 1978, 88, 162-173. (41) Ugarova, N. N.; Lebedeva, O. V.; Frumkina, I. G. Anal. Biochem. 1988, 173, 221-227. (42) Gautier, S. M.; Blum, L. J.; Coulet, P. R. J. Biolum. Chemilum. 1990, 5, 57-63. (43) Blum, L. J.; Gautier, S. M.; Coulet, P. R. J. Biotechnol. 1993, 31, 357-368.

1882 Analytical Chemistry, Vol. 68, No. 11, June 1, 1996

a positive response to the EMIT dau opiate test,13 although sensitivity to each compound is different. The specificity of the luminescence assay is conferred by the enzyme morphine dehydrogenase, which has activity against morphine, codeine, codeine N-oxide, nalorphine, and ethylmorphine.33 Codeine N-oxide and ethylmorphine are exceedingly rare, and nalorphine is a narcotic antagonist.2 These compounds are, therefore, highly unlikely to be encountered in a urine drug screening laboratory. This leaves morphine, the principal heroin metabolite, and codeine. Codeine has proved to be a cross-reacting bane for many an analytical technique, including EMIT.1,2,7,8,13 Furthermore, several substances interfere with the EMIT assay in a manner that can apparently give a positive response and include hordenine.37 Syva reports13 that dextromethorphan at 226 µg/mL can also give positive results.13 The detection limits for the immobilized-enzyme test are very similar to those obtained for the solution assay: limits of 250 pmol of heroin and 10 pmol of morphine represent 250 nM heroin and 10 nM morphine in 1.0 mL, the volume used for the solution assay, compared to 240 nM heroin and 6 nM morphine. This form of the test was designed as a method for detecting particulate heroin, although the results observed suggest that testing solution samples for both morphine and morphine-3-glucuronide with an immobilized-enzyme system may also be viable. The number of heroin particles of various sizes required to provide at least 250 pmol of drug were calculated, based on heroin at a minimum density39 of 1.56 g cm-3, and compared to the reported amperometric enzyme method.25 It has been reported that an average particle of illicit heroin contains 137 pmol of the drug,6 whence the luminescent immobilized-enzyme assay would be sensitive to two particles. Furthermore, immobilized bacterial luciferase systems for other analytes have been reported,40-42 including methods sensitive to 0.1 pmol of NADH.43 Thus, the high specificity and sensitivity of the luminescent enzyme method shows excellent potential for the future development of a heroin biosensor. ACKNOWLEDGMENT The authors would like to thank Her Majesty’s Customs and Excise (HMCE) for funding the work, Dr. Peter Baker at the Laboratory of the Government Chemist, both Macfarlan Smith Ltd., Edinburgh, and HMCE for kindly providing various chemicals, and Keith Jordan at the Institute of Biotechnology for growing the bacterial cultures from which heroin esterase and morphine dehydrogenase were purified. Samples of urine specimens, with the corresponding results from TLC and EMIT analyses performed by the Pathology Laboratory, were generously provided for this study by Dr. Christine Dawson of the Norfolk & Norwich Hospital Department of Chemical Pathology. The authors gratefully acknowledge this contribution, as well as comments and advice from Dr. Dawson. Received for review December 13, 1995. Accepted March 2, 1996.X AC951207R X

Abstract published in Advance ACS Abstracts, April 15, 1996.