Fully Automated Chemiluminescence Immunoassay of Insulin Using

Anal. Chem. 2000, 72, 3518-3522

Fully Automated Chemiluminescence Immunoassay of Insulin Using Antibody-Protein A-Bacterial Magnetic Particle Complexes Tsuyoshi Tanaka and Tadashi Matsunaga*

Department of Biotechnology, Tokyo University of Agriculture and Technology, Koganei, Tokyo 184-8588, Japan

We report a fully automated sandwich immunoassay for the determination of human insulin using antibodyprotein A-bacterial magnetic particle (BMP) complexes and an alkaline phosphatase-conjugated secondary antibody. BMPs bearing protein A-MagA inserted on the external surface of the membrane were prepared in the Magnetospirillum sp. AMB-1 transconjugant for a protein A-magA fusion gene. MagA protein was used as an anchor to attach protein A onto the membrane. Protein A-BMP complexes harvested from transconjugant AMB-1 were subsequently complexed with anti-human insulin antibodies by specific binding between the Z domain of protein A and the Fc component of IgG to form the antibody-protein A-BMP complexes. The complexes were quite monodisperse after the binding of the antibody. The BMPs’ monodispersity resulted in high signal and low noise in the immunoassay. The luminescence intensity ((kilocounts/s)/µg of antibody) from antibody-protein A-BMP complexes after immunoreaction was higher than that from BMPs chemically conjugated to an antibody. This was explained by a difference in dispersion. The fully automated sandwich immunoassay system using antibodyprotein A-BMP complexes made possible precise assays of human insulin in serum. The increase in the incidence of diabetes has led to an increased demand for simple and automated diagnostic systems. Current diagnostic approaches include evaluation of pancreatic β-cell responses to oral glucose by assay of insulin and C-peptide and evaluation of plasma glucose by assay of glycated proteins1-4 and glycation end products.5,6 The development of an automated technique for general-purpose immunoassay of such diagnostic markers with high throughput has become necessary. * Corresponding author. Tel: +81-42-388-7020. Fax: +81-42-385-7713. E-mail: [email protected]. (1) Lim, S.-J.; Kim, C.-K. Anal. Biochem. 1997, 247, 89-95. (2) Weykamp, C. W.; Penders, T. J.; Baadenhuijsen, H.; Muskiet, F. A.; Martina, W.; van der Slik, W. Clin. Chem. 1995, 41, 713-6. (3) Ikeda, K.; Sakamoto, Y.; Kawasaki, Y.; Miyake, T.; Tanaka, K.; Urata, T.; Katayama, Y.; Ueda, S.; Horiuchi, S. Clin. Chem. 1998, 44, 256-63. (4) Shiba, T.; Yano, M.; Maehata, E.; Kiyose, H.; Kotaki, Y.; Fukuzawa, N.; Hagura, R. Diabetes Res. Clin. Pract. 1996, 32, 175-82. (5) Meng, J.; Sakata, N.; Takebayashi, S.; Asano, T.; Futata, T.; Araki, N.; Horiuchi, S. Diabetes 1996, 45, 1037-43. (6) Wu, J. T.; Tu, M. C.; Zhung, P. J. Clin. Lab. Anal. 1996, 10, 21-34.

3518 Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

Magnetic bacteria7-10 synthesize intracellular magnetite particles. Each particle is small (50-100 nm) and is covered with a stable lipid bilayer11 containing 98% lipid and 2% other compounds. The lipid fraction consists of phospholipids (58%) and other lipids (42%). Therefore, the bacterial magnetic particles (BMPs) have good dispersion. BMPs without a membrane were reported to form larger aggregates (the mean diameter was 12.5 µm) compared with those formed by BMPs containing a membrane (the mean diameter was 0.12 µm).12 This observation is consistent with the hypothesis that the BMP membrane is a significant determinant of the aqueous dispersion characteristics of BMPs. The amount of antibody coupling with BMPs was about 4-fold higher than that with artificial magnetite particles of the same size because of the superior dispersion of BMPs compared with that of artificial magnetite particles of the same size in aqueous solution.13 Furthermore, BMPs permit the development of highly sensitive chemiluminescence enzyme immunoassays14 by chemical coupling of foreign proteins to their surfaces and of drug delivery systems using magnetoliposomes with high capture volumes15 based on good dispersion. Recently, transposon mutagenesis in Magnetospirillum sp. AMB-1 has led to the isolation of magA.16 This gene has been completely sequenced.17 The magA gene was used to encode an integral iron translocating protein of a BMP membrane.18 The MagA protein may therefore be used as an anchor for site-specific expression of foreign proteins on BMP membranes by gene fusion, thus obviating the need for immobilization by covalent (7) Matsunaga, T.; Tadokoro, F.; Nakamura, N. IEEE Trans. Magn. 1990, 26, 1557-9. (8) Blakemore, R. P. Science 1975, 190, 377-9. (9) Matsunaga, T.; Sakaguchi, T.; Tadokoro, F. Appl. Microbiol. Biotechnol. 1991, 35, 651-5. (10) Sakaguchi, T.; Burgess, J. G.; Matsunaga, T. Nature (London) 1993, 365, 47-9. (11) Balkwill, D. L.; Maratea, D.; Blakemore, R. P. J. Bacteriol. 1980, 141, 1399408. (12) Nakamura, N.; Matsumaga, T. Anal. Chim. Acta 1993, 281, 585-589. (13) Nakamura, N.; Hashimoto, K.; Matsunaga, T. Anal. Chem. 1991, 63, 26872. (14) Matsunaga, T.; Kawasaki, M.; Yu, X.; Tsujimura, N.; Nakamura, N. Anal. Chem. 1996, 68, 3551-4. (15) Matsunaga, T.; Higashi, Y.; Tsujimura, N. Cell. Eng. 1997, 2, 7-11. (16) Matsunaga, T.; Nakamura, C.; Burgess, J. G.; Sode, K. J. Bacteriol. 1992, 174, 2748-53. (17) Nakamura, C.; Burgess, J. G.; Sode, K.; Matsunaga, T. J. Biol. Chem. 1995, 270, 28392-6. (18) Nakamura, C.; Kikuchi, T.; Burgess, J. G.; Matsunaga, T. J. Biochem. 1995, 118, 23-7. 10.1021/ac9912505 CCC: $19.00

© 2000 American Chemical Society Published on Web 06/24/2000

chemical cross-linking. We have cloned a proteinA-magA fusion gene into Magnetospirillum sp. AMB-1 and obtained BMPs bearing protein A immobilized on the BMP membrane surface (protein A-BMP complexes).19 The development of a fully automated sandwich immunoassay for determination of human insulin in human serum samples using anti-insulin-antibody-binding protein A-BMP complexes is presented. This successful pilot demonstration paves the way for refinement of a system that may be used to rapidly generate specific immunoreagents for a fully automated, rapid, sensitive, and high-throughput immunoassay of any medical diagnostic analyte. EXPERIMENTAL SECTION Materials. Purified human insulin was purchased from Cosmo Bio (Tokyo). Enzymun Test insulin and standard serum containing human insulin were obtained from Roche Diagnostics (Tokyo). A mouse anti-human insulin monoclonal antibody was purchased from Nippon Bio Test (Tokyo), and a rabbit anti-human-insulin IgG fraction (polyclonal) was obtained from Austral Biologicals (San Ramon, CA). Lumi-Phos 530, which includes lumigen PPD, 4-methoxy-4-[3-(phosphonooxy)phenyl]spiro[1,2-dioxetane-3,2′adamantane], disodium salt (3.3 × 10-4 M), as the luminescence substrate for alkaline phosphatase, was obtained from Wako Pure Chemical Industries (Osaka, Japan). Sulfosuccinimidyl 6-[3-(pyridyldithio)propionamido]hexanoate (Sulfo-LC-SPDP) and sulfosuccinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate (Sulfo SMCC) were obtained from Pierce Chemical Co. (Rockford, IL). Other reagents were commercially available analytical reagents or laboratory grade materials. Deionized-distilled water was used in all procedures. Expression of Fusion Proteins and Preparation of Antibody-Protein A-BMP Complexes. Expression plasmid pRZM (12.8 kbp), containing the synthetic protein A gene EZZ and the magA fusion gene with the magA promoter, was constructed as described previously.19 The plasmid pUM5A, containing the magA gene isolated from a wild-type AMB-1 gene bank, was used as a template for PCR amplification of the magA gene. The magA structural gene was ligated to the 3′-end of the Z region of the proteinA gene in pEZZ18 (Amersham Pharmacia Biotech), containing the gene sequence for the Z region, which is the IgG antibody-binding domain in protein A isolated from Staphylococcus aureus. The amplified fusion gene fragment was cloned into pNEP, consisting of the magA promoter and pRK415 (Tcr, lacZ, mob+),20 to yield pRZM. Constructed plasmids were transferred into a wildtype strain of AMB-1 by transconjugation.16 Transconjugants were selected in a magnetic spirillum growth medium (MSGM)21 containing 2.5 µg/mL tetracycline at 25 °C. Batch culture of AMB-1 was done in a flask with a working volume of 4 L × 4. Wet cells (approximately 1.4 mg) suspended in PBS (10 mM phosphate buffered saline, pH 7.4) were disrupted by three passes through a French press cell at 1500 kg/cm2 (Ohtake Works Co. Ltd., Tokyo). Protein A-BMP complexes were collected from the disrupted cell fraction using a columnar neodymium-boron (Nd(19) Matsunaga, T.; Sato, R.; Kamiya, S.; Tanaka, T.; Takeyama, H. J. Magn. Magn. Mater. 1999, 194, 126-31. (20) Keen, N. T.; Tamaki, S.; Kobayashi, D.; Trollinger, D. Gene 1988, 70, 1917. (21) Blakemore, R. P.; Maratea, D.; Wolfe, R. S. J. Bacteriol. 1979, 140, 720-9.

B) magnet (diameter 22.5 mm, height 12.5 mm), producing an inhomogeneous magnetic field (0.5 T at the surface). Protein A-BMP complexes were pelleted at the bottom of the tube using the magnet, and the supernate was removed. The collected protein A-BMP complexes were washed at least three times with PBS and stored at 4 °C in PBS containing 0.01% (w/v) sodium azide until they were required. Antibody-protein A-BMP complexes were prepared by adding the anti-human insulin antibody (100 µg) to protein A-BMP complexes (500 µg/mL PBS). The complexes were magnetically separated and washed three times with PBS. The resuspension ratio of BMPs in PBS was estimated from optical density measurements at 660 nm (OD660) acquired on a spectrophotometer (UV-2200, Shimadzu, Kyoto, Japan); 1.0 × OD660 corresponded to 172 µg of BMPs/mL by dry weight. Preparation of BMPs Chemically Conjugated to Antibodies. The immobilization of antibodies onto BMPs isolated from wild-type AMB-1 was carried out using heterobifunctional coupling reagents, Sulfo-LC-SPDP and Sulfo-SMCC, as described previously.14 Sulfo-SMCC (1 mg) was added to 500 µL of anti-human insulin monoclonal antibody solution (300 µg/mL 50 mM borate buffer, pH 7.6), and the mixture was incubated for 0.5 h at 30 °C with continuous shaking. The reaction product was desalted by chromatography on an NAP-5 column (Amersham Pharmacia Biotech) using 1 mL of TBS (0.1 M Tris/HCl, 1 mM MgCl2, 0.1 mM ZnCl2, pH 7.0) as the eluent according to the manufacturer’s instructions. Sulfo-LC-SPDP was added to a final concentration of 10 mM to 1 mL of a suspension of BMPs (1 mg/mL in PBS). The suspension was mixed by sonication and then incubated for 30 min at room temperature with pulsed sonication (1 min pulses at 5 min intervals). Derivatized BMPs were magnetically separated from the reaction mixture using an Nd-B magnet and washed three times with 1.0 mL of PBS. Derivatized BMPs were dispersed in 1 mL of acetate-buffered dithiothreitol (100 mM sodium acetate, 100 mM NaCl, 20 mM dithiothreitol, pH 4.5), and the mixture was incubated for 30 min at room temperature with pulsed sonication (1 min pulses at 5 min intervals) for activation of the thiol group. Derivatized BMPs were incubated in aqueous solution for 20 h at 4 °C with anti-human insulin antibodies prederivatized with Sulfo-LC-SMCC. BMPs chemically conjugated to antibodies were washed three times with PBS to remove unconjugated antibodies. Sulfo-SMCC and Sulfo-LC-SPDP were also used for conjugation of alkaline phosphatase (ALP) to anti-human insulin IgG antibodies. The concentration of the alkaline phosphatase-conjugated antibodies (ALP-antibodies) was determined by a protein assay according to the Lowry method.22 Measurement of BMP Size Distributions. To evaluate the diffusion of BMP solutions, the size distributions of BMPs were analyzed at various pHs. Size distributions of BMPs (25 µg/mL) were measured after dispersion using an ultrasonic bath. The experiments were performed using a laser particle analyzer (PAR-III, Otsuka Electronics, Osaka, Japan). BMPs in the 40160 nm size range were defined as monodisperse particles. Assay of the Binding of Protein A-BMPs to ALP-Antibodies. Protein A-BMP complexes (500 µg) were resuspended in a solution (1 mL) of ALP-anti-human insulin antibodies (10 µg/mL) (22) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 245, 4508.

Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

3519

Figure 1. Experimental procedure for the sandwich immunoassay using the automated immunoassay system and antibody-protein A-BMP complexes.

in PBS, and the suspensions was incubated for 20 min at room temperature. The BMPs were washed with 1 mL of PBS to remove excess antibodies. They were then mixed with Lumi-Phos 530 (30 µL) for 30 min, and the luminescence intensity was measured using a luminometer (Lucy-2, Aloka, Tokyo). Chemiluminescence Enzyme Immunoassay of Human Insulin Using an Automated Sandwich Immunoassay System. The automated sandwich immunoassay system consists of a reaction station, a tip rack, and an automated eight-way pipet bearing a retractable magnet mounted close to the pipet tip and correspondent to the microtiter plate (96 wells). The microtiter plate may be mounted in the reaction station. There is one rack to hold 8 × 3 tips for reaction. The procedure for chemiluminescence enzyme immunoassay of human insulin using the automatic 3520

Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

system and antibody-protein A-BMP complexes is summarized in Figure 1. A human insulin solution (50 µL) in PBS or serum was mixed with a suspension of antibody-protein A-BMP complexes (50 µL; 25 µg/mL in PBS). The mixture was dispersed by pipet action and then incubated for 20 min at room temperature. Antigen and antibody-protein A-BMP complexes were magnetically separated from the mixture using an Nd-B magnet and then washed four times by resuspension (20 cycles of pipet action) in 100 µL of PBS containing 0.1% (w/v) BSA and 0.05% (v/v) Tween 20. ALPantibodies (100 µL; 5 µg/mL in PBS) were added to the washed antigen-antibody-protein A-BMP complexes and dispersed by automated pipet action (20 cycles). The mixture was incubated for 20 min at room temperature. The complexes were then

magnetically separated from the reaction mixture and washed four times by automated pipet action (20 cycles) in PBS. Finally, the complexes were suspended in PBS and the luminescence intensity was determined. RESULTS AND DISCUSSION Expression of Protein A on BMPs. The expression of protein A at the surfaces of the BMPs was assessed using ALPantibodies. BMPs isolated from wild-type AMB-1 yielded little luminescence upon treatment with ALP-antibodies and a luminescent reagent. In contrast, BMPs isolated from the AMB-1 transconjugant for pRZM yielded a luminescence intensity of 146.3 kilocounts/s, which is approximately 20 times higher than that obtained with native BMPs from wild-type AMB-1. These data are consistent with successful expression of functional protein A on the BMP membrane in the AMB-1 transconjugant for pRZM. Size Distributions of Protein A-BMP Complexes. The dispersion of BMPs increased with increasing pH. The ζ potential for the BMPs was calculated from electrophoretic mobilities determined using an electrophoretic light-scattering spectrophotometer (ELS-6000, Otsuka Electronics, Osaka, Japan). The ionic strength was held constant at 1.0. The ζ potential for the BMPs changed from -2.5 to -25.0 mV upon changing from pH 4.0 to 7.0. The positive correlation between increasing surface negativity of the BMP membrane surface and pH may be explained by the presence of polarizable primary amine groups on the BMP membrane surface as a result of the high percentage content (50% of the total phospholipids) of phosphatidylethanolamine. Phosphatidylethanolamine is negatively charged at high pH but neutral at low pH. This is due to the effect of pH on the polarization of the primary amine group. We propose that the external surface of the BMP lipid membrane is more negative in charge under conditions of low ionic strength and high pH in an aqueous suspension. Elevation of external pH increases the surface negativity of the BMP membrane, leading to improved particulate dispersion. Neutral pH solutions containing 1.0% NaCl were used as buffers in the following assays under monodisperse conditions. The most effective amount of antibody-protein A-BMP complex was 25 µg/mL in this assay without the influence of blocking the luminescence by the BMPs during the measurements. The investigation of the size distributions of BMPs was performed using 25 µg/mL BMP samples in the following experiments. Figure 2 shows the particle size distribution of the BMPs after the attachment of anti-human insulin antibodies. Attachment of the antibodies by chemical conjugation yielded a significant decrease in the monodispersity of the BMPs to less than 40% (w/ w). We propose that this is due to the presence of cross-linkers on the BMP surface. In contrast, the immobilization of anti-human antibodies by complexation with membrane-bound protein A yielded no significant change in the particle size distribution profile for the BMPs. Optimum Conditions for Immunoreactions of AntibodyProtein A-BMP Complexes and ALP-Antibodies Using an Automated System. The efficiency of automated magnet separations of BMPs from suspensions was optimized (Table 1) because robotic magnetic separations of BMP-containing immunoreaction complexes from small volumes of suspensions with excellent recoveries are essential for reproducible and accurate automated immunoassay formats using derivatized BMPs. A single automated

Figure 2. Size distributions of BMPs chemically conjugated to antibodies and antibody-protein A-BMP complexes from recombinant AMB-1. Table 1. Automated Magnet Separation Efficiencies of BMPs for the Fully Automated magnet sepn efficiency of BMPs (%) conditions

first step

final step

no coating coatinga coatinga + BSA + Tween 20c

77 88 ( 2b 90 ( 2b

51 69 ( 5b 83 ( 4b

a Dimethyldichlorosilane coating. Tween 20 were added.

b

n ) 8. c 0.1% BSA and 0.05%

sandwich immunoassay requires the separation of BMP-containing immunoreaction complexes from suspensions in 10 different stages: (1) separation of analyte-(antibody-protein A-BMP) complexes following the primary immunoreaction; (2-5) four washes of the primary immunoreaction product; (6) separation of complexes following the secondary immunoreaction, (7-10) four washes of the secondary immunoreaction product. Nonspecific adsorption to the inner surface of the polypropylene pipet tips used in the automated system is a significant contributor to loss in the recovery of BMP-containing complexes. When an uncoated tip was used, the magnet efficiency for the separation of the BMPs was 77%, and finally half the amount of the BMPs was lost in the final step. Adsorption of the BMPs to the inner tip surface was observed in the first step. This suggests that enhancement of magnet separation efficiency can be achieved by decreasing the adsorption of the BMPs to the surface of the tip in the first step. Nonspecific adsorption of the BMPs was reduced by increasing the hydrophobicity of the polypropylene surface through derivatization with 0.2% dimethyldichlorosilane. Furthermore, the magnet separation efficiency in the final step was maintained at 83 ( 4% by adding 0.1% (v/w) BSA as a blocking reagent and 0.05% (v/v) Tween 20 as a detergent. ALP-antibody concentration is a significant parameter affecting the amount of ALP-antibodies bound and the decrease of nonspecific adsorption to the BMPs’ membrane. The luminescence intensity due to bound ALP-antibodies increased with increasing ALP-antibody concentration and was saturated at 5 µg/mL. The assay background also increased with increasing ALP-antibody concentration (at more than 10 µg/mL). The difference in Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

3521

luminescence intensities was a maximum at 5 µg/mL. These results suggest that the optimal ALP-antibody concentration is 5 µg/mL for the greatest sensitivity and precision. To optimize the immunoreaction time, the time courses of the antigen-antibody reactions of human insulin and antibodyprotein A-BMP complexes were examined. Human insulin standard samples (500 µL, 254 µU/mL) were mixed with antibodyprotein A-BMP complexes, and the mixtures were incubated at room temperature. In each case, after three washings, the time course of the luminescence intensity based on the antigenantibody reaction of ALP-antibody and antibody-protein A-BMP complexes was investigated. The luminescence intensity showed that the immunoreaction continued for more than 30 min. In parallel, the percentage of monodisperse BMPs was also investigated. The concentration of BMPs was more than 95% till 20 min after dispersion; however, it significantly changed at 30 min after dispersion. The concentration of monodisperse BMPs decreased to less than 75%, and the percentage of aggregated BMPs (400-900 nm) sufficient to cause sedimentation increased. On the basis of these results, although extending the immunoreaction makes this assay sensitive, immunoreaction times were fixed to 20 min in the following experiments for effective immunoreactions using monodisperse BMPs. Determination of Human Insulin by Chemiluminescence Enzyme Immunoassay Using and Automated System. When a chemiluminescence enzyme immunoassay was carried out using antibody-protein A-BMP complexes and BMPs chemically conjugated to antibodies, the antigen-binding activity per microgram of antibodies on antibody-protein A-BMP complexes was 2 times higher than that for antibody-BMP conjugates prepared by chemical coupling. This observation could be explained by a difference in dispersion between chemically modified BMPs and antibody-protein A-BMP complexes. Furthermore, the BMPs’ monodispersity resulted in high signal and low noise in the immunoassay because the background signal (insulin concentration ) zero) in the immunoassay using aggregated BMPs was 2 times higher. Human insulin concentrations in PBS and blood serum were measured by a fully automated sandwich immunoassay system using antibody-protein A-BMP complexes and ALP-antibodies as primary and secondary immunoreactants, respectively. Doseresponse curves were obtained from the luminescence intensity for human insulin concentrations in the range 2-254 µU/mL and are shown in Figure 3. A detection limit of 2 µU/mL was observed, which is comparable to that of a conventional enzyme immunoassay.1 A linear correlation between signal and concentration was apparent over the range 19-254 µU/mL. The luminescence intensity and the slope were lower when human insulin in serum was used as a sample, although the dynamic ranges were not significant. This suggests that the antigen-antibody interaction was inhibited by certain conditions in the serum such as the displacement of antibodies binding to protein A. Furthermore, the nonspecific adsorption on antibody-protein A-BMP complexes was lower than that on untreated BMPs from wild type of AMB-1 (data not shown). Antibodies on protein A-BMP complexes prevent secondary antibodies from adsorbing onto BMP mem-

3522 Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

Figure 3. Relationships between luminescence intensities and insulin concentrations in (A) PBS and (B) serum. Luminescence intensities were measured after 30 min incubations. Table 2. Analytical Precisions of Sandwich Immunoassays Using Antibody-Protein A-BMP Complexes assay method

correln coeff

slope

intercept

error (%)

automatic procedure manual procedure

0.999 0.989

0.99 1.01

3.4 -4.2

4.2a 6.3a

a

Insulin concentration was 254 µU/mL.

branes. This result suggests that nonspecific adsorption may be decreased by overexpression of protein A on BMP complexes. The correlation of results for insulin in serum with results obtained for commercially available insulin (Enzymun Test Insulin) was investigated (Table 2). Linear regression analysis yielded a correlation coefficient (r) of 0.999 and a slope of 0.99 for methods using automatic procedures, while r ) 0.989 and slope ) 1.01 were obtained for manual procedures. Our technique allowed insulin concentrations in the range 19-254 µU/mL to be determined with accuracy and precision (coefficients of variation from 6 to 12%). The intercept was not significantly different from zero for both automatic and manual procedures. In conclusion, a fully automated technique for chemiluminescence enzyme sandwich immunoassay of insulin using antibodyprotein A-BMP complexes was developed. Use of the fully automated immunoassay system afforded precise assays of human insulin in serum. ACKNOWLEDGMENT This work was funded in part by a Grant-in-Aid for Scientific Research on Priority Areas (No. 10145102) and a Grant-in-Aid for Scientific Research (No. 10555285) from the Ministry of Education, Science, Sports and Culture of Japan. We thank Shinji Kamiya and Rika Sato of TDK Ltd. for providing pRZM. Received for review November 1, 1999. Accepted February 15, 2000. AC9912505