Liquid-Phase Binding Assay of α-Fetoprotein Using a Sulfated

Michio Sata , Fuminori Moriyasu , Hiroshi Imamura , Seiji Kawasaki , Namiki ... Satoshi Moriya , Masataka Taguri , Satoshi Morita , Miki Konno , A...
0 downloads 0 Views 58KB Size
Anal. Chem. 1998, 70, 954-957

Liquid-Phase Binding Assay of r-Fetoprotein Using a Sulfated Antibody for Bound/Free Separation Kenji Nakamura,* Nobuko Imajo, Yukari Yamagata, Hideo Katoh, Kazunari Fujio, Takumi Tanaka, Shinji Satomura, and Shuji Matsuura

Osaka Research Laboratories, Wako Pure Chemical Industries, Ltd., 6-1 Takada cho, Amagasaki, Hyogo, Japan 661

A rapid immunoassay using a sulfated antibody for bound/free separation in a liquid-phase binding assay is described. A first anti-r-fetoprotein monoclonal antibody was labeled with peroxidase (Fab′-POD) and a second monoclonal antibody was conjugated with polysulfated tyrosine peptide (Fab′-YS). The monoclonal antibodies and r-fetoprotein (AFP) were mixed, incubated, and analyzed directly by anion-exchange column chromatography. The amount of POD activity in the column effluent was determined fluorophotometrically. The bound (Fab′POD + AFP + Fab′-YS) and free (Fab′-POD) forms of the conjugate were clearly and easily separated by ionic charge, and the free sulfated antibody (Fab′-YS) was not detectable fluorophotometrically. The elution position of the bound conjugate was adjusted by varying the length of the polysulfated tyrosine peptide. This method is convenient for antigen measurement because (1) only two modified antibodies are used in a buffer solution, (2) the concentration of antibodies and other assay conditions are easily set, (3) no solid phase is required, and (4) no washing is necessary. Immunoassay is widely used to detect a great variety of environmental, clinical, and biochemical substances. The heterogeneous immunoassay methods reported thus far require immobilization of antibody or antigen onto a solid phase. Accordingly it is difficult to adjust the concentration of antibody or antigen. Unfortunately, setting these concentrations is very important to assay performances. Furthermore this limitation causes the lack of stoichiometric reaction and nonspecific bindings that are the greatest problems in solid-phase assays. We have developed a new method, formed “liquid-phase binding assay” (LBA) that uses a liquid-phase binding reaction between antigen and antibody and that separates bound and free forms by high-performance liquid chromatography (HPLC) without the need for a solid phase.1-3 The LBA method allows the user to set a sufficient concentration of antibody in the solution phase such that all analyte molecules are bound completely and * Corresponding author: (fax) +81-6-4991524; (e-mail) [email protected]. (1) Nakamura, K.; Satomura, S.; Tanaka, T.; Matsuura, S. Anal. Sci. 1992, 8, 157-160. (2) Nakamura, K.; Satomura, S.; Matsuura, S. Anal. Chem. 1993, 65, 613616. (3) Hara, T.; Nakamura, K.; Satomura, S.; Matsuura, S. Anal. Chem. 1994, 66, 351-354.

954 Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

in a stoichiometric ratio. Furthermore, the immune reaction is finished within a few seconds and a wide dynamic range of analyte concentration can be obtained. One limitation to this technique, however, is that when a large amount of enzyme-labeled antibody is used in the reaction mixture, the separation of bound and free forms of the conjugate by gel filtration becomes difficult. The present study overcomes this limitation by using both polysulfated tyrosine-conjugated and enzyme-labeled antibodies, followed by ion-exchange chromatography for separation. In this study, the LBA method was applied to the assay of R-fetoprotein. EXPERIMENTAL SECTION Materials. A commercially available R-fetoprotein (AFP) assay kit was obtained from Abbott Laboratories (Abbott Park). Poly(L-aspartic acid) of average molecular weight 50 000 was obtained from Sigma-Aldrich Japan K. K. (Tokyo, Japan). AFP and other reagents were manufactured by Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Apparatus. A model LC-9A HPLC pump (Shimadzu Co., Kyoto, Japan) and model 232-401 automatic sample processor (Gilson Medical Electronics, Inc.) fitted with a postcolumn flowthrough coil (0.25 mm i.d. × 20 m) were used. An anion-exchange column (POROS, diethylaminoethyl type, 4.5 mm × 10 mm, Perseptive Biosystems, Tokyo, Japan) was used with a gradient of sodium chloride between 0 and 3000 mM in 50 mM tris(hydroxymethyl)aminomethane (Tris) hydrochloride buffer (pH 8.0) at a 2.0 mL/min flow rate. After separation, a fluorogenic substrate solution consisting of 90 mM 4-N-(4-carbobutyryl)aminophenol4 and 20 mM H2O2 in 10 mM citric acid buffer, pH 5.5, was added to the effluent at a flow rate of 0.2 mL/min and mixed by an on-line mixer. Then the mixture was incubated in a 20-m-long flow-through coil at 55 °C for 25 s to allow the peroxidase (POD) enzyme reaction. The activity of POD in the effluent was determined fluorophotometrically using an excitation wavelength of 328 nm and an emission wavelength of 432 nm. Figure 1 shows a scheme of the experimental apparatus. Synthesis of Polysulfated Tyrosine Peptides. A Fmoc-βAla-Alko resin (100-200 mesh, Watanabe Chemical Industries, Ltd., Osaka, Japan) was used as a starting material to synthesis polytyrosine-containing peptides by a solid-phase technique with BOP/HOBT and Fmoc-tyrosine.5 After the elongation reaction, trifluoroacetic acid was added to the resin and Fmoc-polytyrosine (4) Siga, M.; Yakata, K.; Aoyama, M.; Sasamoto, K.; Takagi, M.; Ueno, K. Anal. Sci. 1995, 11, 195-201. (5) Merrifield, R. B. J. Am. Chem. Soc. 1963, 85, 2149-2154. S0003-2700(97)01149-9 CCC: $15.00

© 1998 American Chemical Society Published on Web 02/03/1998

Figure 1. Schematic diagram of the HPLC system. Table 1. Synthetic Polysulfated Peptides

peptide was obtained. Fmoc-polytyrosine peptide was dissolved in a N,N-dimethylformamide and dimethylformamide-SO3 solution [a mixture of N,N-dimethylformamide-SO3 (Fluka ChemikaBiochemika, Tokyo, Japan) and N,N-dimethylformamide-pyridine] was added, followed by an overnight reaction at 4 °C. After piperidine treatment of the Fmoc-polysulfated tyrosine peptide to remove a Fmoc residue at the N-terminal, polysulfated tyrosine peptide (YS) was precipitated as a barium salt. Sulfosuccinimidyl4-(p-maleimidophenyl) butyrate (Pierce, Rockford, IL) was reacted with polysulfated tyrosine peptide to give maleimidyl polysulfated tyrosine peptides having 1, 3, 4, 5, 7, 8, and 10 sulfated tyrosine residues which were purified by reversed-phase HPLC (5C18, Wako). Maleimidyl polysulfated serine peptides also were synthesized by the same method (Table 1). Caution: For sample handling, eyes and skin should be protected with safety goggles and gloves, respectively. Antibodies. Two anti-AFP monoclonal antibodies (clone no. WA-1 and A4-4) that recognize different AFP epitopes were selected from a panel of antibodies obtained from Wako. The antibodies were digested with pepsin and reduced to Fab′. The single binding site fragment moiety (Fab′, WA-1) was conjugated to POD (Toyobo, Osaka, Japan) with N-(8-maleimidocapryloxy)sulfosuccinimide (Pierce). The Fab′-POD conjugate obtained was homogeneous as judged by gel filtration HPLC using a Wakosil-200 column (8 mm i.d. × 300 mm, Wako). Fab′-POD concentrations were determined from POD enzyme activities. The second Fab′(A4-4) was conjugated with various lengths of maleimidyl polysulfated tyrosine peptides or maleimidyl polysulfated serine peptides. Poly(L-aspartic acid) was introduced to the second Fab′ (A4-4) with N-(8-maleimidocapryloxy)sulfosuccinimide (Pierce). Elution Positions of Immune Complex. The effects of multiple sulfated tyrosine residues conjugated to Fab′ were investigated: 100 µL of 200 nM Fab′-YS and 100 nM Fab′-POD solution was mixed with 10 µL of AFP solution (100 ng/mL), and a reaction was carried out at 8 °C for 10 min. Portions (20 µL) of reaction mixture were analyzed by HPLC using the above conditions. Assay of Serum AFP. For serum AFP concentration determinations, 100 µL of 140 nM Fab′-POD and 100 nM Fab′-YS8 (see Table 1) solution was mixed with 3 µL of serum and the mixture incubated at 8 °C for 90 s. Then, 20-µL portions of the reaction mixtures were injected into HPLC and resolved by an

no.

formula

no.

formula

YS1A YS3A YS4A YS5A YS7A YS8A YS10A

Ala-[Tyr(SO3H)]1-β-Ala Ala-[Tyr(SO3H)]3-β-Ala Ala-[Tyr(SO3H)]4-β-Ala Ala-[Tyr(SO3H)]5-β-Ala Ala-[Tyr(SO3H)]7-β-Ala Ala-[Tyr(SO3H)]8-β-Ala Ala-[Tyr(SO3H)]10-β-Ala

YS4 YS5 YS7 YS8 SS3A SS5A SS8A

Ala-[Tyr(SO3H)]4 Ala-[Tyr(SO3H)]5 Ala-[Tyr(SO3H)]7 Ala-[Tyr(SO3H)]8 Ala-[Ser(SO3H)]3-β-Ala Ala-[Ser(SO3H)]5-β-Ala Ala-[Ser(SO3H)]8-β-Ala

anion-exchange column. The free form Fab′-POD conjugate was eluted quickly at a high flow rate (9 mL/min) initially and in a second step, the bound form (Fab′-POD + AFP + Fab′-YS8) conjugate was eluted at a low flow rate (2 mL/min) using 3 M sodium chloride in Tris-HCl buffer pH 8.0. RESULTS Elution Positions from Anion-Exchange HPLC. The relationship between the number of anion residues in the immune complex and its elution position was investigated for 14 sulfated peptides and poly(L-aspartic acid) introduced Fab′. The free Fab′POD and the immune complex consisting of Fab′-POD and AFP eluted in a linear gradient of from 0.1 to 0.25 M sodium chloride, as shown in Figure 2A. By adding Fab′-YS8 to the reaction mixture, the first immune complex (Fab′-POD + AFP, Figure 2A) vanished entirely and a new peak at 1.4 M sodium chloride emerged, as shown in Figure 2B. The second peak should be an immune complex of Fab′-POD + AFP + Fab′-YS8 and was completely and easily separated from free Fab′-POD. Figure 3 shows the relationship between the elution position of immune complexes obtained with sulfated antibodies and sodium chloride concentration. As seen in this figure, increasing the number of sulfated peptide residues in an immune complex caused a shift in the peak elution to an increased sodium chloride concentration. The elution position was not effected by the presence of a β-alanine residue at the C-terminal of the sulfated peptides. However, the elution position of immune complexes comprising poly(L-aspartic acid) (average number of carboxyl residues 450) was similar to that of YS8 (number of sulfate residues 8), the immune complex of polysulfated serine peptides were eluted earlier than the corresponding complex comprising the same number of polysulfated tyrosine residues. All serum components were eluted at a sodium chloride concentration below 0.3 M (data not shown). Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

955

Figure 2. Chromatographic pattern of the reaction product using a liner gradient: (A) reaction mixture of Fab′-POD and AFP; (B) reaction mixture of Fab′-POD, AFP, and Fab′-YS8.

Figure 3. Elution position of immune complex that contains sulfate residues of varying length. Sodium chloride concentration at the elution peak (b); range of sodium chloride concentration from start to end of the peak (s).

Elution Program. The analysis time of anion-exchange chromatography was shortened by using a switching valve and stepwise gradient techniques. Free Fab′-POD was washed out completely by an initial solution of 50 mM Tris-HCl buffer, pH 8.0, containing 0.4 M sodium chloride for 45 s at a flow rate of 9 mL/min, because of low back pressure. The free conjugate was eluted into a drain directly through a valve without passing through the on-line coil and detector. After the valve was switched from drain to coil and detector, the immune complex were eluted with 50 mM Tris-HCl buffer, pH 8.0, containing 3.0 M sodium chloride at a flow rate of 2 mL/min. A typical chromatogram and elution program is shown in Figure 4. The immune complex was eluted as a sharper peak when a step gradient was used than when a linear gradient was used, reflecting a decreased influence of free Fab′-POD. This system allowed the complete separation of immune complex from a more than 7 order magnitude higher concentration of free Fab′-POD (data not shown). Use of a step gradient allowed shortening of the analysis time to 2.8 min/test. 956 Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

Figure 4. Elution program and chromatogram: Fab′-POD (- - -); reaction mixture of Fab′-POD, AFP (200 ng/mL), and Fab′-YS8 (s); flow rate program (- - -); sodium chloride concentration (- -).

Figure 5. Sequential chromatograms: reaction mixture of various concentrations of AFP injected continuously.

Figure 5 shows sequential chromatograms of this system at varied AFP concentrations; within-run-assay coefficients of variation ranged from 1.5 to 7.0% (n ) 5). Stoichiometry of the Immune Reaction. The time course for formation of the immune complex was examined. Various concentrations of AFP (50, 200, 1000, 3000, and 4000 ng/mL) were sampled, and added to 140 nM Fab′-POD and 100 nM Fab′YS8, and analyzed (Figure 6). All immunoreactions reached equilibrium within 50 s. A Scatchard plot of bound/free vs bound was fitted a straight line (data not shown) and the number of Fab′-POD and Fab′-YS8 molecules bound per AFP molecule was calculated to be 0.95 and 0.96, respectively. All antigens were bound to antibody at short times upon addition of a sufficient amount of antibodies. Thus, one AFP molecule could be measured via the activity of one POD molecule. Assay of AFP in Human Serum. The method was applied to the measurement of human serum AFP. The amount of bound AFP form was calculated from its peak height on the chromatogram. Figure 7 shows a dose response curve. This figure shows a linear relationship between peak height and AFP concentration

Figure 6. Immunoreaction time course. The time course of immune complex formation was examined using various concentrations of AFP (ng/mL): b, 50; O, 200; 4, 1000; 9, 3000; 0, 4000.

Figure 7. Dose response of AFP: AFP concentration 0-800 ng/ mL, emission wavelength 432 nm (b); AFP concentration 0-12 000 ng/mL, emission wavelength 550 nm (O).

up to 800 ng/mL using an emission wavelength of 432 nm. The linearity of response was seen up to 10 000 ng/mL AFP when an emission wavelength of 550 nm was used. The AFP detection limit in the sample was 0.5 ng/mL. We measured various concentrations of serum AFP using both the present method and a commercially available EIA kit. A good correlation was found (109 samples, r ) 0.988). DISCUSSION In general, immobilization techniques have been used for separating bound and free antibody conjugates in immunoassays. On the other hand, a liquid-phase immune reaction coupled with column separation of bound and free forms has been shown to accurately measure antigen. Ion-exchange chromatography revealed a sufficient and fast separation of a bound conjugate (Fab′POD + AFP + Fab′-YS) from free conjugate (Fab′-POD) using as a second antibody conjugate that contained polysulfated peptide. The bound form, having sulfate residues, was adsorbed quickly and quantitatively by an anion-exchange column. The adsorption of bound conjugate was easily controlled by altering the number of sulfate residues or the type of amino acid used. The difference in adsorption ability between serine and tyrosine residues was

partly a result of hydrophobic interaction with the anion-exchange resin used. The adsorption ability of sulfate residues was higher than that of carboxyl residues. The former residue is useful for establishing an optimized assay condition because the number of sulfate residues needed is only from 3 to 10 in order to complete bound and free antibody conjugate separation. And four or more sulfated residues are needed to remove serum interference and obtain accurate AFP measurements. Fab′-YS8 was selected for AFP measurement because the peak width of bound antibody conjugate using Fab′-YS10 is wider than that of YS8. Moreover, use of a sulfated conjugate allows two different antigens to be measured simultaneously because two different immune complexes can form and be separated. Antibody binding reactions in the LBA assay reach completion in less than 1 min because these reactions occur in a liquid phase and optimum assay conditions such as antibody concentration are easily selected. However, use of a higher concentration of labeled antibody often increases background signals in a conventional immunoassay because of nonspecific binding to the solid phase. On the other hand, a prolonged assay time occurs with the LBA method because a long time is needed to elute the free and excess labeled antibody. Reaction and separation conditions were optimized for speed by using sulfated antibody and switching valve technologies. Furthermore, because of the requirement for only two labeled antibodies, this method can shorten the development time periods for new diagnostic assays. The use of Fab′ fragments of monoclonal antibodies also simplifies the assay procedure because antibody binds to antigen in a one-to-one ratio. Accordingly, one antigen molecule can bind one labeled molecule stoichiometrically to form a linear signal response. The results from the present study in fact showed a linear dose response relationship that passes through the origin. Using the present method, a detection limit of 0.5 ng/mL (20 fmol/test) and a dynamic range of 0.5-10 000 ng/mL was achieved for the AFP. In contrast, conventional immunoassay methods yield narrower dynamic ranges that typically are sigmoid or a logarithmic because of the small quantities of antibody used on the solid phase and the consequent lack of stoichiometry. The LBA system uses HPLC, which is well suited for automation analysis. The major advantages of the LBA system shown by this study are stoichiometric binding of labeled conjugate, low detection limit, higher dynamic range with a linear dose response curve, shortened reaction and analysis time, and high potential for automation. ACKNOWLEDGMENT We thank Dr. M. Motsenbocker for his kind advice and help.

Received for review December 9, 1997.

October

17,

1997.

Accepted

AC9711495

Analytical Chemistry, Vol. 70, No. 5, March 1, 1998

957