Anal. Chem. 1903, 65, 2030-2039
2036
Detection and Removal of Escherichia coli Using Fluorescein Isothiocyanate Conjugated Monoclonal Antibody Immobilized on Bacterial Magnetic Particles Noriyuki Nakamura, J. Grant Burgess, Kaoru Yagiuda, Satoko Kudo, Toshifumi Sakaguchi, and Tadashi Matsunaga' Department of Biotechnology, Tokyo University of Agriculture and Technology, Koganei, Tokyo 184, Japan
A novel fluoroimmunoassay method using bacterial magnetic particles for the highly sensitive detection of bacteria has been developed. Fluorescein isothiocyanate (FITC) conjugated monoclonal anti-Escherichia coli antibody was immobilized onto bacterial magnetic particles (BMPs) using a heterobifunctionalreagent,N-succinimdyl 3-(2-pyridyldithio)propionate(SPDP). E. colicells were reacted with FITC-antibody-BMP conjugates for 15 min in an inhomogeneous magnetic field which enhanced aggregation. The cell/BMP complexes sedimented, causing relative fluorescence intensity of the solution to decrease with increasing microbial cell concentration. A linear relationship was obtained between the relative fluorescence intensity and cell concentration in the range of 10L106 cells/mL. Selectivity of this detection system was satisfactory. Monoclonal antibody immobilized on BMPs was also applied to the specific removal of E. colifrom the bacterial suspension.
INTRODUCTION The detection of microbial cells is important in the clinical, environmental, and food industrial fields. Methods based on colony formation have been employed for the detection of viable cells in microbiology.1 However, these methods are time consuming and demand complicated procedures. Enzyme-linked immunosorbent assay (ELISA) and radioimmunoassay (RIA) systems performed either in microtiter plates, on nitrocellulose membranes, or in tubes are known to be higly sensitive, whereas counterimmunoelectrophoresis and latex agglutination or coagglutination methods can provide more rapid resulta.2 Recently, a magnetic separation technique was introduced for tissue typing, cell sorting, subcellular organelle fractionation, and DNA separation.= Bacteria were captured by specific monoclonal antibodies bound to magnetic beads coated with covalently bound goat antimouse immunoglobulin and detected in an ATP-depen(1)Postage, J. R. In Methods in Microbiology;Norrie, J. R., Ribbons, D. W., Eds.; Academic Press: New York, 1969; Vol. 1, pp 611-621. (2) Fung, J. C.;Tilton, R. C. In Manual of Clinical Microbiology, 4th ed.; Lennette, E. H., Balowe, A., Hausler, W., Jr., Shadomy, H. J., Eds.; American Society for Microbiology: Washington DC, 1985; pp 883-890. (3) Funderud, S.;Nustad, K.; Lea, T.; Vartdal, F.; Gaudernack, G.; Stenstad, P.; Ugelstad, J. Lymphocytes: A Practical Approach; Klaus, G. G. B., Ed.; IRL Press: Oxford, UK, 1987; pp 55-65. (4) Lea,T.;Vartdal,F.; Nustad, K.; Funderud, S.;Berge, A.; Ellingsen, T.; Schmid, R.; Stenstad, P.; Ugelstad, J. J. Mol. Recogn. 1988,1,9-18. (5) Uhlen, M. Nature 1989,340,733-734. 0003-2700/93/0385-2036$04.00/0
dent luciferase-luciferin enzyme system.* The system was evaluated with monoclonal antibodies specific for Klebsiella, Enterobacter, and Proteus. In this system, the detectable sensitivity was 10s CFU/mL. Another immunomagnetic technique to detect and identify Salmonella serogroup C1 has been developed.' Monoclonal antibodies specific for 0 antigen 6,7 of Salmonella lipopolysaccharide coupled to magnetic beads (Dynabeads, M-450) were used. Captured bacteria were identified by acridine orange staining and measured by enzyme immunoassays with a conjugate antilipopolysaccharidemonoclonal antibody asthe detector probe. However, the whole detection process required 2-3 h and the sensitivity was 109-104cells/mL. The diameter of bead is 4.5 pm. The conditions of the antigen-antibody reaction depend on diffusion, and the time for the formation of the immunocomplexes is influenced by the viscosity of the reaction medium and bead size. Sensitivity and time required for measurement were attributed to the bead size. Therefore, small particles with good dispersion properties are required for more sensitive and rapid detection of bacteria. Magnetic bacteria which orient and swim along geomagnetic fields may be isolated from fresh and marine sedimenta.8lB The bacteria contain magnetite particles with controlled size (50-100 nm) and disperse very well because they are covered with a stable lipid membrane.10 We have succeeded in the pure culture of magnetic bacteria capable of growing under aerobic conditions" and have used purified bacterial magnetic particles (BMPs) for a number of clinical applications. Enzymes and antibodies have been immobilized on BMPs using both bifunctional reagents and glutaraldehyde and been found to have higher activities than those immobilized onto artificial magnetic particles.'* Recently, we have developed a novel solid-phase fluoroimmunoassay using antibody immobilized onto BMPs for the rapid detection of mouse immunoglobulin G (IgG).U In this paper, fluorescein isothiocyanate (FITC) conjugated monoclonal anti-Escherichia coli antibody was immobilized onto BMPs using a heterobifunctional reagent,N-succinimdyl 3-(2-pyridyldithio)propionate(SPDP). This system using BMPs was applied to the rapid and highly sensitive detection and removal of microbial cells. (6) Torensma, R.; Vieser, M. J. C.; Amman, C. J. M.;GroebbbHeij, A.; Poppelier, M. J. J. G.; Van Beurden, R.; Fluit, A. C.; Verhoef, J. J. MicrobioI. Methods 1992, 15, 135-142. (7) Luk, J. M. C.;Lindberg, A. A. J.Zmmunol.Methods 1991,137,l-8. (8) Blakemore, R. P. Science 1975,190,377-379. (9) Kirschvink, J. L. J. Exp. Biol. 1980,86, 345-365. (10) Gorby, Y. A.; Beveridge, T. J.;Blakemore, R. P. J. Bacteriol. 1988, 170,834-841. (11) Mataunaga, T.;Sakaguchi, T.; Tadokoro, F. Appl. Microbiol. Biotechnol. 1991,35, 651-655. (12) Mataunaga, T.;Kamiya, S. Appl. Microbiol. Biotechnol. 1987,26, 328-332. (13) Nakamura, N.; Hashimoto, K.; Mataunaga,T. A d . Chem. 1991, 63,268-272. 0 1993 American Chemical Soclely
4ALYTICAL CHEMISTRY, VOL. 65, NO. 15, AUGUST 1, 1993
EXPERIMENTAL SECTION Materials. Fluorescein isothiocyanate (FITC)was purchased from Sigma Chemical Co. (St. Louis, MO). N-Succinimdyl3-(2pyridy1dithio)propionate (SPDP) was obtained from Pierce Chemical Co. (Rockford, IL). Mouse anti-E. coli monoclonal antibody was purchased from Chemicon Intemational Inc., (Temecula, CA). Other reagents were commercially available analytical reagents or laboratory grade materials. Deionizeddistilled water was in all procedures. Bacterial Strains and Cultivation Conditions. E. coli (JM109) was cultured in 3 mL of L broth at 37 OC for 12 h. Pseudomonas putida and Pseudomonas aeruginosa were cultured in 3 mL of nutrient broth at 37 OC for 12 h. Serratia marcescens was cultured in 3 mL of medium containing 1% polypeptone, 0.2% yeast extract, and 0.1% MgSOd7H20 (pH 7.0) at 30 OC for 12 h. The grown cells were washed twice and diluted (10-10' cells/mL) with phosphate-buffered saline (PBS; pH 7.4). Magnetic bacteria were isolated from sludge obtained from pondsinTokyoandwere ~~IturedinMSGMmediumaspreviously described.11J4 The pH of the medium was adjusted to 6.75 with NaOH solution before sterilization. The cells were cultured in 5 L of medium at 25 "C for 7-10 days. Magnetic bacteria grown to stationary phase (about 2 X 108cells/mL) were centrifuged at 5000g for 10 min. The collected cells were washed with 10 mM acid] HEPES [1-(2-hydroxyethyl)piperazine-4-2-ethanesdfonic buffer (pH 7.4). Preparation of BMPs. BMPs were isolated from magnetic bacteria by the following method. Approximately 10l2 cells suspended in 20 mL of HEPES buffer were disrupted by three passes through a French pressure cell at 1300 kgf/cm2(Ohtake Works Co. Ltd., Tokyo, Japan; No.5615). Disrupted cells were treated in the ultrasonic disrupter (Tomy Seiko Co. Ltd., Tokyo, Japan; UR-200P) five times, each operated for 5 min at 0 "C. BMPs were collected from the sonicated cell fraction by using a samarium-cobalt magnet (18 X 11X 14mm) that produced an inhomogeneousmagnetic field (0.4Ton the surface of the magnet and an average gradient of 0.2 T/cm). BMPs were collected at the bottom of the tube due to the presence of the magnet, and the supematant was removed. The collectedBMPs were washed with HEPES buffer at least five times and kept at 4 "C in phosphate-buffered saline (pH 7.4) containing 0.01 % sodium azide before use. The particle size of the BMPs was measured with a particle size analyzer (Shimadzu Co. Ltd., Kyoto, Japan; SA-CP3). When magnetic bacteria at the stationary phase of growth (about 2 X 108 cells/mL) are used to prepare magnetic particles, the yield is about 0.4 mg/L of culture. The yield from 1012 cells (5 L) is usually about 2 mg. Immobilization of Antibody onto BMPs. Figure 1shows the structure of SPDPand schematic diagram for immobilization of antibody onto BMPs. A 100-pLaliquot of 20 mM dithiothreitol (DTT) solution in phosphate buffer containing 0.1 M NaCl and 5 mM EDTA (pH 6.0) was added to 100pL of anti-E. coli antibody solution (40mg/mL) for the reduction of antibody. The mixture was incubated for 2 hat roomtemperature. After the incubation, the sample was purified using a NAP-10 column, eluting with phosphate buffer according to the manufacturer's instructions (Pharmacia, Uppsala, Sweden). Reduced anti-E. coli antibody was directly immobilized onto BMPs by modifying the organic membrane with SPDP.ls A 2.5-pL aliquot of acetone containing 0.06 mg of SPDP was added to 1 mL of BMP suspension (500 ccp/mL). Then the suspension was dispersed by sonication and incubated for 2 h at room temperature. The modified BMPs were washed with PBS and incubated with reduced anti-&. coli antibody solution for 12 h at 4 "C. Antibody-coupled BMPs were washed with PBS several times to remove excess antibody. The concentration of antibody in the solution was determined by the Lowry methodl6before and after immobilization, and the quantities of antibody immobilized on BMPs were calculated. Conjugation of FITC was carried out by a modification of the (14) Blakemore, R.P.;Maratea, D.; Wolfe, R. S . J. Bacteriol. 1979, 140, 720-729. (15) Carleson,J.; Drevin, H.; Axen, R.Biochem. J. 1978,173,723-737.
(16)Lowry,0.H.; Roaebrough,N. J.; Farr, A. L.;Randall, R.I. J. Biol. Chem. 1951,193, 266-276.
2037
N-Succinimidyl3-(2-pyridyldithio)propionate (SPDP)
Bacterial magnetic particles
SPDP
* Dithiothreitol
p-DTT*
U
Flguro 1. Structure of SPDP and schematic diagram showing immobilization of antlbody onto BMPs using SPDP.
method of The and Fe1tkamp.l' Antibody-coupled BMPs (500 pg) were suspended in 500 mL of borate buffer (0.5 M, pH 9.2). A 100-pg sample of FITC was added to the solution and the resultant mixture was incubated for 2 h with stirring. After the incubation, FITC antibody-conjugated BMPs were washed with PBS and collected by using a Sm-Co magnet. FITC was also conjugated to nontreated (not antibody coupled) BMPe in the same procedure. Fluoroimmunodetectionof E. coli by Using FITC Conjugated Antibody Immobilized onto BMPs. Figure 2 shows the procedure for fluoroimmunodetection of E. coli. A 50-pg sample of antibody coupled with BMPs and 100 pL of sample containing E. coli cella were mixed and incubated in a test tube (16.5-mm0.d.) for 15minat 37 OC (Figure 2.1). The agglutination reaction was enhanced by applying an inhomogeneousmagnetic field (Figure 2.2) (the test tube was placed on a Sm-Co magnet), which increased aggregation of FITC-antibody-BMPs when these conjugates reacted with the bacteria (Figure 2.3). The strength of the applied magnetic field was adjusted by changing the position of the magnet and measured by using a flux meter (Shimadzu, GK-3). Then the mixture was added to 2 mL of Tris-HC1 buffer (pH 9.0) in a 5-mL test tube, mixed for a few seconds, and transferred to a 10 x 10 mm quartz glass cuvette (Figure 2.1). After being left at 25 "C for 15min, the fluorescence intensity of nonprecipitated FITC-labeled BMPs remaining in the liquid was measured using a fluorescence spectrophotometer (Shimadzu, RF-5000) with the excitation wavelength set at 492 nm and the emission wavelength at 513 nm. Removal of E. coli from Bacterial Suspension. By adding the antibody coupled with BMPs to 1-mL test sample (fiial cell concentration of E. coli was adjusted to 5 X l(r cells/mL and the other bacteria to 5 X 104cells/mL). The suspensionwas incubated at 25 OC for 15min with gentle mixing. BMPs were collected at the bottom of the tube due to the presence of the magnet. After the removal of supematant, BMPs were washed three times briefly with PBS and finally resuspended in 2 mL of PBS. E. coli removal was estimated from the fluorescence intensity. (17) The, T. H.; Feltkamp, T. E. W. Immunology 1970,18,88&873.
2038
ANALYTICAL CHEMISTRY, VOL. 65, NO. 15, AUGUST 1, 1993 E. coli
(0
I t 1
I
FITC conjugated antibody bacterial magnetic particle
-
EX. 491nm
Enhanced agglutination under applied magnetic field
1.
2.
Dispersion
in 2ml Tris-HCI buffer
3.
4.
Sedimentation
5.
Figure 2. Schematic diagram showing the principle of bacterial detectlon using FITC-antibody-bacterial magnetic particles. The assay is based on the aggiutlnation of bacterial magnetic particles covalently linked to FITC-labeled antCE. coliantlbodiesand bacterial cells present in the sample. For full explanation, see text.
R E S U L T S AND DISCUSSION Preparation of BMPs. The median diameter of BMPs was 0.12 pm, and BMPs purified in this way retain their magnetosome membrane, which consists mainly of phoepholipid as described.1l On the other hand, BMPs, which were washed with chloroform-methanol (2:1, v/v) after ultrasonication, were not covered with a magnetosome membrane. BMPs without a magnetosome membrane formed larger aggregates (the median diameter was 12.5 fim). The magnetosome membrane consisted of 98% lipids and 2% other compounds including proteins. Phospholipids comprised 58% of the total lipid, and phosphatidylethanolamine accounted for 50% of the total phospholipids present. These results show that the magnetosome membrane containing amino residues covers each BMP. Therefore, proteins may be immobilized onto the membrane after activation using SPDP. This reagent reacts with primary amines via an N-hydroxysuccinimide ester to install a flexible spacer chain terminating with a maleimide for conjugation with thiolcontaining proteins. Therefore, antibody can be immobilized onto the membrane activated by SPDP. Immobilization of Anti-E. coli Antibody onto BMPs. Reduced anti-E. coli antibody was immobilized on BMPs with SPDP. The extent of antibody coupling with BMPs without a magnetosome membrane, i.e., treated with chloroform-methanol (2:l) was 46 pg/mg of particles, whereas coupling with BMPa with a magnetosome membrane was 150 pg/mg of particles. BMPs were covered with a stable lipid membrane. There was slight aggregation in each particle as a result of its own magnetic properties. Therefore, BMPs with the membrane were superior in dispersion to BMPs without the magnetosomemembrane. The extent of antibody coupling with BMPs was about 3-fold higher than that achieved when BMPs treated with organic solvent were used. In the present study, we used bacterial magnetic particles which possessed a lipid membrane. These particles bind large amounts of capture antibody and have low nonspecific adsorption of antibody. The surface of the particles was modified to incorporate a cross-linker molecule terminating with a maleimide. This functionality was used to couple the
antibody through thiols obtained by reduction with dithiothreitol. Then, FITC was coupled to antibody as described in the Experimental Section and these modified particles were reacted with bacteria. Immunoreaction of FITC-Anti-E. coli Antibody-BMP Conjugates. As shown in Figure 2, the immunoassaysystem for detection of bacteria consists of two manual steps: an agglutination reaction, followed by measurement of fluorescence. The assay mechanism occurs in five distinct phases: (1)FITC-anti-E. coli antibody-BMP conjugates are mixed with E. coli cells. (2) BMPs bind to the E. coli cella via the FITC-labeled antibody and the cells are covered with magnetite. The magnetite-labeled cella form aggregates in the presence of an inhomogeneous magnetic field. The magnetic field causes the particles to align and brings them closer together, enhancing agglutination. (3) Many particles to bind to a singleE. colicell and the cell magnetite complexes sediment. Particle size was measured before and after the immune reaction using a Shimadzu particle size analyzer as previously described and a similar increase in particle size was detected.13 (4) In order to release unbound BMPs which cosedimented with the cell/magnetite complexes,the sample was redispersed during the transfer to a cuvette. (5) After 15 min, the ceU/magnetite complexes sediment and the residual fluorescenceof FJTC bound to free magneticparticles is determined. The extent of the decrease in fluorescence compared to the initial value allows determination of the bacterial concentration. Fluorescence intensity (arbitrary unit 97.5) decreased by 78% in the presence of E. coli in the range of 10s cells/mL and 60% in the range lo" cells/ml. On the other hand, the fluorescence intensity did not decrease after 15-min incubation in the absence of E. coli. Detection a n d h m o v a l of E. coliby FITC-Anti-E. coli Antibody-BMP Conjugates. The cell concentration of E. coli was measured by using FITC-anti-E. coli antibody-BMP conjugates. Figure 3 shows the relationship between the relative fluorescence intensity and cell concentration. The experiments were carried out using BMPs only conjugated with FITC. In this case, the fluorescence intensity scarcely
ANALYTICAL CHEMISTRY, VOL. 65, NO. 15, AUQUST 1, 1993
€
x e .(0
m ._
80
Cell concentration (cells per ml)
Fbure5. Relationshipbetweenthe relatlve fluorescence intenstfyand cell concentration: (0)FITC conJugatedBMPs, (0)FITC-antl-E. coll antibody-BMPs; (-) In the presence of an Inhomogeneous magnetic fleld, (- -) no magnetic field.
-
~
Table I. Specificity of the Detection of Bacteria Using FITC-Anti-E. coli Antibody-BMP Conjugates relative fluorescence intensity (% ) bacterial strain
none Escherichia coli Pseudomonas putida Pseudomonas aeruginosa Serratia marcescens
100 50
96 98 101
decreased in spite of the increase in cell concentration. This result indicates the decrease in the fluorescence intensity was not caused by adsorption of E. coli to the BMPs. The relative fluorescence intensity decreased with increasing cell concentration using FITC-anti-E. coli antibody-BMP conjugates. The fluorescence intensity decreased to 73% in the range of 106 cells/mL without a magnetic field. An inhomogeneous magnetic field (0.4T o n the surface of the magnet and an average gradient of 0.2 T/cm) applied using a Sm-Co magnet during incubation period enhanced the aggregation of FITC-anti-E. coli antibody-BMP conjugates, and the fluorescence intensity decreased to 50% in the range of l o 5 cells/mL. Aggregation of FITC-anti-E. coli antibody-BMP conjugates was enhanced and the relative fluorescence intensity decreased by applying a magnetic field during incubation. Fluorescence intensity was reproducible with an average relative error of 3% when a sample containing 104 cells/mL was measured five times. Table I shows the specificity of the detection of bacteria using FITC-anti-E. coli antibody-BMP conjugates. The cell concentration of each bacteria was adjusted to lo5cells/mL. Except for E. coli, the fluorescence intensity scarcely decreased. The fluorescence intensity only decreased in the presence of E. coli. These results suggested that antibodyspecific aggregation based on the antigen-antibody reaction can be detected from the fluorecence intensity decrease. Removal of E. coli from bacterial suspension waa also performed (Figure 4). When the BMPs were removed from the cell suspension containing E. coli and P. putida, the relative fluorescence intensity of BMP reached 58%. In the
2039
0
* a none
E coli
d P pulrda
E coli P putrda d 8 S marcescens S marcescens
Flgure 4. Selectlve removal of E. cdiusingFITC conjugated antibody Immoblllzed on bacterial magnetic particles. Total cell concentration was adjusted to lo6 cells/mL.
case of the suspension containing E. coli and S. marcescens, the relative fluorescence intensity was 60%. On the other hand, for suspensions not containing E. coli, the relative fluorescence intensity did not decrease. This is due to the aggregationof FITC-anti-E. coli antibody-BMP conjugates. In this way, monoclonal antibody immobilized on BMPs was also applied to the selective removal of E. coli from the bacterial mixture suspension. The aim of the present study was to show whether it is possible to detect bacteria by using FITC-anti-E. coli antibody-BMP conjugates. The assay described here is based on the aggregation of FITC-anti-E. coli antibody-BMP conjugates and E. coli in the sample due to the immunoreaction. The aggregation of FITC-anti-E. coli antibody-BMP conjugates was enhanced during the incubation period by applying an inhomogeneous field. This resulted in a shortening of the required time. The reaction was completedwithin 30 min. This method is also highly sensitive; the detectable limit of E. coli was lo2cells/mL. The novel concept described here gives rise to a rapid and highly sensitive system for the selective detection of bacteria. Furthermore, this method may be generally applicable to the detection of any suitable antigen. Bacterial magnetic particles may also be obtained using a high-density culture procedure which gives 5 g of particles from 90 g wet weight of cells (unpublished data). Thus commercialapplication of immunoassaysusing bacterial magnetic particles may be expected in the near future.
ACKNOWLEDGMENT This work was partially supported by a Grant-in-Aid for Special Project Research 03205032 from the Ministry of Education, Science and Culture.
RECEIVED for review November 22, 1993.
13, 1992. Accepted April
