Determination of Fluorescent Cyanobenz[ f Jisoindole Derivatives of Dopamine and Norepinephrine Using High Performance Liquid Chromatography with Chemiluminescence Detection Takao Kawasaki? Center for Bioanalytical Research, University of Kansas, Lawrence, Kansas 66045, USA

Kazuhiro Imai Branch Hospital Pharmacy, University of Tokyo, Tokyo 112, Japan

Takeru HiguchiPII and Osborne S. WongSP Oread Laboratories, Inc., 1 501 Wakarusa Dr. Lawrence, Kansas 66046, 913-749-0034, USA

Dopamine (DA), norepinephrine (NE) and 3,4-dihydroxybenzylamine (DHBA) are converted to highly fluorescent cyanobenz[f]isoindole (CBI) derivatives by reacting these amines with naphthalene-2,3-dicarboxaldehyde in the presence of cyanide ion. Femtomole amounts of these CBI derivatives separated by reverse-phase high performance liquid chromatography can be detected by a post-column chemiluminescence system utilizing bis(2,4-dinitrophenyI) oxalate and hydrogen peroxide. Linear detection response was observed between 1 to 600 fmol, giving a signal to noise ratio of approximately 15 at the 1 fmol level. An assay procedure was developed for the determination of DA and NE in 20 p L of urine sample using DHBA as the internal standard.

INTRODUCTION Chemically induced light emission, chemiluminescence (CL), generated by the reaction of fluorescent compounds with aryl oxalate (AO) and hydrogen peroxide (H202) has been established as a sensitive detection method for high performance liquid chromatography (HPLC) analysis (Honda et al., 1985, 1986; and Imai et aZ., 1985, 1986, 1987a,b; Sigvardson et al., 1984). A number of HPLC-CL assays, including AO+ H 2 0 2and luminol+ H z 0 2 and lucigenin systems, reported sensitivity unattainable by routine fluorescence and electrochemical methods (Kobayashi and Imai, 1980; Kawasaki et al., 1985; Miyaguchi et al., 1984; Honda et aZ., 1985a,b; Veazey and Nieman, 1980). However, many analytes of biomedical interest are nonfluorescent; therefore, chemical derivatization using the appropriate fluorophore is often necessary in order to obtain analyte derivatives which are efficient energy acceptors in the A 0 + H,Oz CL reaction (Givens et al., 1989). + RNH, + CN'

__t

CN

OH

OH

C BI doparnine (DA)

H

R1=

OH,

R,

R1=

OH,

R P = OH

=

norepinephrine ( N E )

&N-R

\ NDA

Fluorescent cyanobenzrf lisoindole (CBI) derivatives generated from the reaction of primary amines (RNH2) with naphthalene-2,3-dicarboxaldehyde(NDA) in the presence of cyanide ion (CN) (Carlson et aZ., 1986; de Montigny et al., 1987; Matuszewski et al., 1987) is found to be a highly efficient energy acceptor in the A 0 + H 2 0 2 CL reaction (Givens et al., 1989). The mechanism of the A O + H , 0 2 CL reaction has been examined in detail (Mohan, 1985; McCapra, 1973) and a series of potential intermediates has been reported by Alvarez et al. (1986).

\ CBI

(a)

t Current address: Sankyo Co. Ltd., Shinagawa, IOKYO 140, Japan. $- Author to whom correspondence should be addressed. 8 Also affiliated at the Department of Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas 66045, USA. I' Also affiliated at the Center for Bioanalytical Research, University of Kansas.

(b)

In an earlier report, we described an HPLC assay for the determination of two catecholamines (CAs) dopamine (DA) and norepinephrine (NE) in urine using the pre-separation NDA + C N derivatization method with fluorescence detection (Kawasaki et al., 1989). In this work, we explore the detection capability and limitations of CBI derivatives in HPLC assays using a post-separation CL detection system involving

CCC-0269-3879/90/0113-0118$3.00 @ Heyden & Son Limited, 1990

BIOMEDICAL CHROMATOGRAPHY, VOL. 4,

NO. 3, 1990 113

T. KAWASAKI

bis(2,4)dinitrophenyI) (DNPO) oxalate and H 2 0 2 . Specifically, an assay method was developed for the determination of DA and NE in micro samples of urine.

Urine 0 . 5 m EDTA ~ 0.2 M Phosphate buffer; pH 8.0 1 p~ DHBA (1,s.) Alumina

EXPERIMENTAL

20 pL 1 mL 1 mL 20 pL 10 mg

mixed for 3 min aspirated washed with H 2 0

i

Reagents. Bis(2,4-dinitrophenyI) oxalate (DNPO) was obtained from Wako Chemicals USA, Inc. (Dallas, TX, USA). Naphthalene-2,3-dicarboxaldehyde(NDA) was supplied by the organic synthesis group of the Center for Bioanalytical Research at the University of Kansas. DA, NE, and 3,4dihydroxybenzylamine (DHBA) were acquired from Sigma Chemical Company (St. Louis, MO, USA). Sodium cyanide, boric acid, sodium tetraborate, phosphoric acid and potassium hydrogen phosphate, all ACS grade, were purchased from Aldrich Chemical Company (Milwaukee, W1, USA) and were used as supplied. HPLC grade acetonitrile, methanol and tetrahydrofuran were acquired from Fisher Chemical Company (St. Louis, MO. USA). Water was purified by distillation and stored in glass. Apparatus. A Farrand (Valhalla, NY, USA) System 3 scanning spectrofluorometer equipped with a 10 pL flow cell was used for HPLC fluorescence detection. An Atto model AC-2220 Biomonitor (Tokyo, Japan) with a 60 p L spiral flow cell was used to measure the CL intensity. DNPO (0.5 mM) and H20, (12.5 mM) in MeCN were delivered separately using two lsco model 314 syringe pumps (Lincoln, NE, USA) at a flow rate of 1.0 and 2.0 mL/min, respectively. The isocratic HPLC system consists of an LKB (Gaithersburg, MD, USA) model 2150 dual-piston pump equipped with a Rheodyne (Cotati, CA, USA) model 7125 sample injector with a 5 p L sample loop. A schematic diagram of the HPLC-CL system is depicted in Fig. 1. Stainless steel tubings of 0.5 mm internal diameter were used to connect M1 to M2 and M1 to LM. An Orion (Cambridge, MA, USA) model 811 pH meter was used for all pH measurements. Chromatographic conditions. CBI derivatives of catecholamine standards were separated using a TSK ODS-120T column (4.6 x 150 mm, 5 pm, Tosoh Company, Japan). The flow rate was 1.0 mL/min. Time constarit for the CL detector was 5.0 s. For the analysis of DA and NE in urine samples, the mobile phase for isocratic elution was 36% MeCN+4% THF+60% 10 mM potassium phthalate buffer, pH 2.5 (v:v:v). Derivatization procedure for amine standards. Into a 6 mL amber vial containing 0.920 mL borate buffer (pH 9) was added 2 0 p L of an aqueous mixture containing 0 . 1 0 - 1 0 . 0 p ~ of amines in 0.10 M borate buffer (pH 9.0), 20 p L of 2 p M DHBA as an internal standard and 20 pL 0.5 mM NaCN in borate buffer (0.4 M, pH 9.0). The reaction was initiated by the addition of 20 p L of 0.5 mM NDA in methanol. Final concentrations were [NDA] = [CN] = 10 pM, and [DHBA] = 40 nM.

Ana1.C.

Figure 1. Schematic flow diagram of the chemiluminescence HPLC system. P, pump (Pl, LKB 2150; P,, Pa, ISCO 314); I, injection valve (Rheodyne 7125); LM, detector (ATTO AC-2220 Luminomonitor 1); Rec, recorder (Linear 1200); M, mixing device; G.C. guard column (RP-18, 15x3.2 mm, 7 pm); Anal. C., column (TSK ODS-l20T, 150 x 4.6 mm, 5 p n ) ; E, eluent (10 mM potassium hydrogen phthalate, pH 2.5+MeCN, 50:50) 1 mL/min; R, ,0.5 mM DNPO +MeCN, 1 mL/min; R,, 12.5 mM H,O,+MeCN. 2mL/min.

114 BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO 3, 1990

Alumina

0.1 m L x 2

0 1 M Phosphoric acid

Alumina eluate

I

200 pL 50 pL 50 pL

0 2 M Borate buffer. pH 9 0 100 W M NaCN lOObM NDA stood for 1 h a t non temp

HPLC analysis 20 KL injected into HPLC

Figure 2. Urine sample preparation method.

After mixing, the reaction solution was allowed to stand at room temperature for 30min. The CBI derivatives of the amines were separated by HPLC and quantified by CL detection. Analysis of dopamine and norepinephrine in urine samples. Figure 2 summarizes the procedure used for the analysis of DA and NE in urine samples. Briefly, 2 0 p L urine, 2 m L distilled water, 1 mL 0.5 mM EDTA in 1 mM HC1, 1mL 0.2 M phosphate buffer (pH KO), 20 p L 1.0 p~ DHBA, and 10 mg alumina were mixed in a screw-capped glass test tube and mixed with a vortex mixer for 3 min. After mixing, the supernatant was removed by an aspirator, and the alumina was then washed 3 times with 5 mL portions of distilled water. Desorption of catecholamines from the alumina was achieved by washing (30 s) the alumina twice with 100 p L 0.10 M phosphoric acid. The combined acidic supernatant was collected in a 6 mL amber screw-capped vial. Derivatization was carried out by adding to the supernatant 0.20 mL of borate buffer (0.4 M, pH 9.0), 50 pL 100 mM NaCN, and 50 p L 0.10 mM NDA in methanol. The reaction solution was thoroughly mixed and allowed to react at room temperature for 1 h. Reaction solutions were either analyzed by HPLC immediately, or were stored at 4 "C until analysis was performed. ~

~~~

~

RESULTS AND DISCUSSION

DA, NE and DHBA were readily converted to the corresponding fluorescent CBI derivatives under mild conditions (room temperature, pH 9). The physico-chemical properties of these derivatives were described in an earlier report (Kawasaki et al., 1989). The design of the post-separation chemiluminescence detection system shown in Fig. 1 is well known from the work of Imai and others (Kobayashi and Imai, 1980; Imai, 1986). The, separation of catecholamine CBI derivatives using reversed phase HPLC requires the use of an acidic eluent (Kawasaki et al., 1989). Because bis(2,4-dinitrophenyl) oxalate (DNPO) was reported to be an effective CL reagent in acidic media (Honda et al., 1985a), it was selected as the reagent of choice for this CL system. However, DNPO exhibits limited stability in the presence of H 2 0 2 (Honda et al., 1985a), therefore, DNPO and H 2 0 2 were delivered separately using a @ Heyden & Son Limited, 1990

F L U O R E S C E N T DERIVATIVES OF D A A N D N E U S I N G H P L C A N D C L

dual-syringe pump configuration and the two reagents were mixed prior to the final mixing with the chromatographic eluent. Effects of [DNPO] and [H,O,] on the HPLC-CL performance

Optimum conditions were determined for the CL reaction using DA CBI derivative in a flow injection mode. Figure 3 shows the effects of H z 0 2 concentration (in MeCN) on the peak height and the signal-to-noise (S/ N) ratio of DA+ CBI. The concentration of DNPO was kept constant at 0.5 mM. Under these conditions the peak height showed a nonlinear increase, but the S /N ratio showed a maximum response at 12.5 mM H 2 0 2 . Similar experiments were conducted by varying the concentration of DNPO (0.25-2.0mM in MeCN) at constant H 2 0 2concentration (12.5 mM). As depicted in Fig. 4, there was a sharp increase in peak height between 0.25 and 0.50mM DNPO, but no improvement in response was observed beyond 0.50 mM. Moreover, maximal S/N ratio was recorded at 0.50 mM DNPO. CL reagent and CL reaction mixing time

As indicated above, DNPO and H,O, were mixed in the reagent delivery system prior to final mixing with the chromatographic eluent to generate the CL emission. The mixing time of the two reagents can be controlled by using various tubing lengths to connect the two mixers (M1 and M2). In this study the inner diameter of the stainless sieel tubing was constant (0.5 mm). Mixing time was varied by changing the tubing length. Figure 5 shows the effects of mixing time on both peak height and S/N ratio. Optimal performance was realized at 12 s using a three mete; tubing. The A 0 + H 2 0 2CL reaction is known to generate a characteristic rise and fall in CL emission intensity time profile (Honda et al., 1985a; Alvarez et al., 1986; Givens et al., 1989). The overall detection sensitivity of any HPLC-CL system is believed to be determined by the reaction time which, in a post-column HPLC application, is controlled by the tubing dimensions (diameter and length) between mixer M1 and the CL monitor (see Fig. 1). The peak height of DA+CBI showed a rapid decrease as the mixing time of the eluent and CL reagents

I 0:s

1.0 DNPO;

2.0

mM

Figure 4. Effect of DNPO concentration on CL: Eluent, 10mM potassium hydrogen phthalate pH 2.5 +MeCN 50:50,v/v, 1 mL/rnin; 12.5 mM +MeCN, DNPO. 0.25-2.0 mM +MeCN, 1 rnL/rnin; H,O, 2 mL/min.

increases (see Fig. 6), but the S /N ratio reaches a maximum at 0.6 s . Beyond 0.6 s, the baseline noise width remains fairly constant and the S/N ratio follows the peak height profile. These observations can be largely attributed to the CL-emission time profile considerations proposed by Givens et al. (1989). The origin of the noise dependence has not been explored in detail, but the nixi: g dy.iami%s of w-ganic and aqueous solutions with different viscosities and refractive indexes may give rise to the results observed here. It is interesting to note that phosphate buffer gives a chromatographic baseline noise width which is five times larger than that of the phthalate buffer (data not shown). This may be due to the low solubility of phosphate buffer in acetonitrile which may lead to precipitation of micro crystals. Detection linearity and limits

This HPLC-CL system gives good linear response for DA and NE CBI derivatives from 1 to 600 fmol of DA and NE CBI derivatives injected on column (Fig. 7(a, b)). The reproducibilities of this assay method for multiple derivatization of DA and NE ( n = 5) range from 0.46% at 250 fmol to 2.85% at 63 fmol. Figure 8

Mixing Time 5 10

(5)

15

2(

I E .-ma

f x

W

a

1

H 2 0 2 ; mM

Figure 3. Effect of hydrogen peroxide concentration on CL: Eluent, 10 rnM potassium hydrogen phthalate pH 2.5+MeCN 50:50 v/v, 1 mL/min; DNPO,0.5mM+MeCN, 1 mL/rnin; H,0,,3.125-200mM+ MeCN, 2 mL/min.

6 Heyden & Son Limited,

1990

2 3 Tube Length (rn)

4

5

Figure 5. Effect of tube length (reagent mixing) on CL: Eluent, 10 mM potassium hydrogen phthalate pH 2.5 +MeCN 50:50, v/v, 1 mL/min; DNPO, 0.5 rnM+MeCN, 1 mL/rnin; H,O, 12.5 mM +MeCN, 2 rnL/min.

BIOMEDICAL CHROMATOGRAPHY, VOL 4, NO. 3, 1990 115

T. KAWASAKI Mixing T i m e 0.5 1

(s)

2 1.5

1

3 t 0

lb

2b 30 Tube Length

60 (cm)

Figure 6. Effect of tube length (CL reaction) on CL: Eluent, 10 mM potassium hydrogen phthalate pH 2.5 +MeCN 50:50, v/v, 1 rnL/min; DNPO, 0.5 mM/MeCN, 1 mL/min; H,02, 12.5 mM/MeCN, 2 rnL/min.

J

Amine

(f mol)

Figure 8. Chromatogram of cyanobenzisoindole derivatives of amines. HPLC conditions: Pump, LK6 2150 HPLC Pump, ISCO 314; Detector, ATTO AC-2220 Luminomonitor 1; Column, TSK gel ODS120T. 150 ~ 4 . 6mm, 5 km; Eluent, 10 mM potassium hydrogen phthalate, pH 2.5+MeCN 50:50, v/v; Flow rate, 1 rnL/rnin; Reagent, 0.5 mM DNPO +MeCN, 1 mL/min; 12.5 mM H2O,+MeCN,2mL/min. Peaks: 1, norepinephrine; 2, 3,4-dihydroxybenzylarnine (IS.); 3, dopamine. (each peak 1 frnol).

shows a tracing of a chromatographic run containing 1 fmol of DA, NE and DHBA CBI derivatives. The S/N ratio at this level is approximately 15. Table 1 shows a representative comparison of the detection limits of several catecholamine assays using various luminescence detection methods. It is clear that CL detection offers enhanced detection limits. In the detection of CBI derivatives, the CL method developed here gives approximately 100 times greater sensitivity than the conventional fluorescence method. Roach and Harmony (1987) were able to detect sub-femtomole levels of amino acid CBI derivatives using an argon-ion laser, but detection limits at these levels are not available to analytical chemists using commercial fluorescence detectors. Determination of DA and NE in micro sample of urine

Highly sensitive analytical methods are useful when the analyte exists in very low concentration or when the ~~

~

Table 1. Comparison of the detection limits of dopamine and norepinephrine by various HPLC-fluorescence assay methods Detection method

Amine

(f mol)

Figure 7. Standard curves for amines with CL detection: B, Norepinephrine, r=0.999; 0, Dopamine, r=1.000.

116 BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO. 3, 1990

Fluorescence NDA+CN Natural fluorescence OPA (post column) OPA (laser induced) Trihydroxyindole Fluorescamine Ethylenediamine Chemiluminescence NDA+CN Fluorescamine Dansyl chloride a

Dopamine

0.02 2 0.8 0.1

Reported detection limits' (pmol) Ref Norepinephrine

0.5 0.3

0.02 1.8 0.4 0.03 0.006 NA 0.12

Kawasaki (1989) Jackman (1980) Yui (1981) Todoriki (1983) Yui (1979) Kobayashi (1981 Mori (1985)

0.001 0.025 NA

0.001 NA 0.016

This work Kobayashi (1981) Melbin (1983)

-

NA, not available.

@ Heyden & Son Limited, 1990

FLUORESCENT DERIVATIVES OF D A A N D N E USING HPLC A N D C L

Table 2. Reproducibility and recovery of dopamine and norepinephrine in human urine samples"

Sample

1 (Day 1)

2 (Day 2)

a

Figure 9. Chromatograms of human urine and spiked urine: (a) human urine, (b) spiked urine. Peaks: 1, norepinephrine; 2, 3.4dihydroxybenzylamine; 3, dopamine. Mobile phase: 10 mM potassium hydrogen phthalate pH 2.5 +CH,CN +THF, 60:36:4 v/v.

sample size is limited. Since DA and NE are present in urine at levels readily quantifiable by existing methods, we elected to challenge the usefulness of this HPLC-CL assay by measuring the amount of DA and NE in microliter volumes of urine. Two urine samples (approximately 200 mL each) were collected from a healthy male subject over a 2-day period. A 20 FL volume of these samples was extracted and derivatized following the procedure described in Fig. 2. Standard addition of DA and NE was also performed to assist in peak identification and recovery evaluation. Figure 9(a, b) shows two chromatograms obtained from an original and spiked urine samples. Table 2 presents a summary of the amounts of the CA found, reproducibilities and recoveries. These findings demonstrate (i) the high detection sensitivity of CL method when coupled with HPLC and (ii) the effectiveness of CBI derivatives as a chemiluminescent fluorophore in the determination of primary amines. However, because of the high sensitivity afforded by CL methods, interferences from trace quantities of fluorescent impurities and side products from the derivatization reaction can be a problem because the signals or responses from these substances are also magnified in the CL detection mode. Careful handling of samples and solutions and the use of high purity buffers and reagents are often required in order to control background interferences from these materials.

CA added pmol/ZO pL NE DA

none

none

20

20

none

none

40

40

CA found

prnol/20 pL NE

2.20 (6.1%) 21.5 (2.2%) 9.22 (4.2%) 52.3 (2.7%)

1%

SD) DA

Recovery % NE DA

23.9 (5.4%) 44.4 (2.0%) 38.0 (4.5%)

-

-

93

101

_

_

77.5

108

99

(44%)

n =5 for all samples.

The selection of the proper aryl oxalate, in this case DNPO, is aided by the findings and discussions provided by Honda et al. ( 1985a). Bis(2,4,6-trichlorophenyl)oxalate (TCPO) is another useful popular oxalate ester used in AO+H,O, CL detection. Since TCPO is relatively stable in the presence of H 2 0 2 (Imaizumi et al., 1989; Hanaoka, 1989), the two reagents can be premixed and the post-column CL system would require only one reagent pump, simplifying the operation of the HPLC system. However, TCPO is not an effective CL reagent when the pH of the chromatographic eluent is below 6. (Honda et al., 1985a; Weinberger, 1984). This precludes the use of TCPO in this assay. Recently, Irnai et al. (1987a) developed a new oxalate, bis[4-nitro-2-(3,6,9trioxadecyloxycarbonyl)phenyl] oxalate (TDPO), which gives improved aqueous solubility and chemical stability in the presence of H,02 relative to TCPO. This may be a potentially useful reagent to explore in future experiments because it would allow the use of a single post-column reagent delivery pump, and higher oxalate concentration. The use of two CL reagent pumps obviously adds to the complexity of the operation of the chromatographic system, but it is manageable. The long-term durability and reproducibility of HPLC-CL assays have not been explored in detail. Currently, we are applying this HPLC-CL assay to the determination of a peptide CBI derivative using a post-column CL system utilizing single CL reagent pump by pre-mixing TCPO or TDPO with H 2 0 2 .The results of this work will be presented in the near future. Acknowledgements This work was partly supported by Oread Laboratories, Inc. and grants from the National Institute on Aging (5R44AG06271-02) and the Kansas Advanced Technology Commission. The authors wish to thank the Tosoh Company, Japan for supplying the HPLC columns for this project.

REFERENCES Alvarez. F., Parekh, N. J.. Matuszewski, 6..Givens, R. S., Higuchi. T. and Schowen, R. L. (1986) J. Am. Chem. SOC.108,6435. Carlson, R. G., Srinivasachar, K., Givens, R. S. and Matuszewski, 6. (1986) J. Org. Chem. 51. 3978. de Montigny. P., Stobaugh, J. F., Givens, R. S., Carlson, R. G., Srinivasachar, K., Sternson, L. A., and Higuchi, T. (1987) Anal. Chem., 59, 1096.

@ Heyden & Son Limited, 1990

Givens, R. S.. Schowen, R . L.,Stobaugh, J. F., Kuwana. T., Alvarez, F.,Parekh, N.. Matuszewshi, 6..Kawasaki, T., Wong, 0.. Orlovic. M., Chokshi, H. and Nakashima, K. (1989) American Chemical Society Symposium Series 383,Ch. 8, ed by M . C. Goldberg. Hanaoka, N., (1989) Anal. Chem. 61, 1289. Honda, K., Miyaguchi, K. and Imai. K. (1985) Anal. Chim. Acta 177, 103.

BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO. 3, 1990 117

T. KAWASAKI Honda, K., Miyaguchi. K. and Irnai, K. (1985b) Anal. Chim. Acta 177, 111. Honda, K., Miyaguchi, K., Nishino. H., Tanaka, H., Yao, T., and Irnai, K. (1986) Anal. Biochem. 153, 50. Irnai, K. and Weinberger, R. (1985) Trends Anal. Chem. 4, 170. Irnai, K. (1986) Methods in Enzymology 133, 435. Irnai. K., Nawa, H.,Tanaka, M., and Ogata, H. (1986) Analyst 111,209, Irnai, K., Matsunaga, Y.. Tsukarnoto. Y., and Nishitani, A. (1987a) J. chromarogr. 400, 169. Irnai, K., Nishitani, A.. andTsukarnoto,Y. (1987b) J. Chromatographia 24, 77. Irnaizurni, N., Kayakawa, K., Miyazaki, M., and Irnai. K. (1989) Analyst 114, 161. Jackrnan, G. P., (1980) J. Chromatogr. 182, 277. Kawasaki,T.. Maeda, M. and Tsuji, A. (1985) J . Chromatogr. 328,121. Kawasaki, T., Irnai, K., Higuchi. T. and Wong, 0. S., (1989) Anal. Biochen~.180, 279. Kobayashi, S., Sekino, J., Honda. K., and Irnai, K., (1981) Anal. Biochem. 112, 99. Kobayashi, S. and Irnai, K. (1980) Anal. Chem. 52, 424.

118 BIOMEDICAL CHROMATOGRAPHY, VOL. 4, NO. 3, 1990

Kobayashi, S. (1981) Anal. Biochem. 112, 99. Matuszewski. B.. Givens, R. G., Srinivasachar, K., Carlson, R. G. and Higuchi. T. (1987) Anal. Chem., 59, 1102. McCapra, F. (1973) Prog. Org. Chem. 8, 231. Melbin, G. (1983) J. Liq. Chromatogr. 6, 1603. Miyaguchi, K., Honda, K. and Irnai, K. (1984) J. Chromarogr. 303,173. Mohan, A. G. (1985) in Chemi- and Bioiominescence, ed by J. G. Burr, ch 5, Marcel Dekker, New York. Mori, K. and Irnai, K. (1985) Anal. Biochem. 246, 283. Roach, M . C. and Harmony, M . D. (1987) Anal. Chem., 59,411. Sigvardson, K., Kennish, J. and Birks, J. (1984) Anal. Chsm. 56,1096. Todoriki, H., Hayashi, T., Naruse. H. and Hirakawa, A. (1983) J. Chromatogr. 276, 45. Veazey, R. L. and Niernan, T. A. (1980) J. Chromatogr. 200, 153. Weinberger. R. (1984) J. Chromatogr. 314, 155. Yui, Y., Kirnura, M., Itokawa, Y. and Kawai, C. (1979) J. Chromatogr. 177, 376. Yui, Y. and Kawai, C. (1981) J . Chromatogr. 206, 586. Received 8 July 1989; accepted 9 July 1989

0 Heyden & Son Limited,

1990