586
Anal. Chem. 1991, 63,586-595
High-Performance Immunoaffinity Chromatography and Chemiluminescent Detection in the Automation of a Parathyroid Hormone Sandwich Immunoassay David S. Hage*pl a n d P a i C. Kao Department of Laboratory Medicine, Mayo CliniclFoundation, Rochester, Minnesota 55905
An automated sandwich immunoassay was developed based on high-performance lmmunoafflnlty chromatography and chemllumlnescent detection, using the determination of parathyroid hormone (PTH) in plasma as a model system. I n thls method, Injectionsof plasma and acrldlnium ester-labeled antl-( 1-34 PTH) antibodies were made onto a column containing Immobilized anti-( 44-68 PTH) antibodies. Upon eiutlon, PTH and Its assoclated labeled antlbody were combined with an alkaline peroxide postcolumn reagent, and the resulting light production was measured. Factors considered in optimizing thls system included the column’s dissociation propertles, the rate of llght production In the postcolumn reactor, and the use of sequential vs simultaneous injectlon of sample and labeled antibody. The flnal system developed requlred 6 min per plasma Injection, fdlowlng a 141lncubatlon of sample wlth labeled antibody. The response was linear over 2-3 orders of magnltude and the lower llmlt of detection for a 66-pL plasma sample was only 16 amol, or 2.4 X M. Overall, thls method had a preclslon and response similar to those of manual PTH methods but requlred 24-fold less time to perform. By uslng different Immobilized and labeled antlbodies, thls method could easlly be adapted for use wlth other analytes.
INTRODUCTION The immunoassay is an important tool in bioanalytical chemistry. The basis of this method is the use of biochemical reagents known as antibodies for the selective determination of sample components. Antibodies, or immunoglobulins, are a diverse group of glycoproteins that make up part of the body’s immune system. Of the 106-108types of antibodies in the body, each has the ability to selectively bind a particular foreign agent, or antigen ( I ) . The variety of antibodies available in nature, plus the development of techniques to raise and harvest antibodies binding a given agent, has long made them popular reagents in clinical and biochemical analyses. Examples of their use in immunoassays include methods for the determination of drugs, hormones, enzymes, nucleic acids, whole cells, and viruses ( 2 ) . Of the various immunoassay formats available, one of the most sensitive is the sandwich technique. In this method, two antibodies binding the analyte of interest are used: one immobilized onto a solid support and the other free in solution, but labeled with some easily detectable chemical compound. Examples of chemical labels that may be used for the second antibody include radioisotopes, fluorescent compounds, and enzyme systems which generate colored or electrochemically active products ( 2 ) . When samples containing analyte are placed in this system, the analyte binds to both the immo-
* Author to whom correspondence should be addressed.
’
Current address: Department of Chemistry, University of Nebraska, Lincoln, NE 68588-0304.
bilized antibody and labeled antibody. The result is a “sandwich” immune complex on the support’s surface. The analyte is detected by washing away nonbound sample components add excess labeled antibody and measuring the amount of labeled antibody complexed to analyte on the support’s surface. This method is not only highly specific, as a result of the selective nature of the antibody-analyte interactions, but is also very sensitive, due to the use of labels with good limits of detection ( 2 ) . The combination of these characteristics makes the sandwich immunoassay a powerful tool for the ultratrace determination of many biologically related compounds. For some analytes, such as various peptide hormones, this technique may be the only method available by which the compound of interest can be easily measured ( 2 ) . But despite the selectivity and sensitivity of sandwich immunoassays, they suffer from several disadvantages. One is that they have traditionally been manual methods. This is due to the multiple reagent additions, washing steps, or centrifugation steps often involved in their use. Another disadvantage is that these assays usually require prolonged incubation times. As a result, analysis times of hours or days are common. To overcome these disadvantages in other types of immunoassays, a number of automated systems have been developed. Examples include instruments based on fluorescence polarization, rate nephelometry, particle counting, and EMIT centrifugal analysis (3). These automated techniques, however, have important limitations that prevent their use in sandwich immunoassays. For example, all of these systems are based on homogeneous methods, in which simple solution reactions between antibody and analyte are used for compound detection ( 3 , 4 ) .As a result, they are not designed to carry out the more complex washing and separation steps required in a sandwich immunoassay. Another problem is that M these methods typically have limits of detection of or greater (3,4).In comparison, some compounds determined by sandwich immunoassays may have sample concentrations M, requiring much lower detection limits of less than (5). One way in which sandwich immunoassays could potentially be automated is by using high-performance immunoaffinity chromatography (HPIAC). HPIAC is a highly selective technique based on the reversible interactions between antibodies and their antigens. In this method, antibodies capable of binding the compound of interest are immobilized onto a small, rigid support, such as diol-bonded-silica, and placed in a column. When sample is applied to this column, any analyte present will bind to the immobilized antibody while other components will elute nonretained. After washing away nonretained components, the mobile-phase conditions are changed to disrupt the antibody-analyte bond, eluting the analyte for quantitation or collection for further use. The column is then regenerated by going back to the initial mobile-phase conditions, and the process is repeated (6). In the determination of several major serum proteins using HPIAC and direct UV detection, it has been shown that this technique
0003-2700/9 1 /0363-0586$02.50/0C3 1991 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 63,NO. 6, MARCH 15, 1991
587
CHJ'
I
R
wCHI'
I
0
i e
R
Ant1.(44.68 PTH! Anl1.(1.34 PTH!'
CHr
I
+hu
0
Figure 1. Chemiluminescent reaction of acridinium esters. is not only fast,often taking only minutes to perform, but also has good correlation vs reference methods ( 7 , B ) . By proper selection of the antibody and elution conditions, it has also been found that these columns can be quite stable. In a number of studies, several hundred injections per column have been reported (7-10). All of these characteristics make HPIAC an attractive approach for the automation of a sandwich immunoassay. However, little research has been done in this field. One exception is work by de Alwis and Wilson, in which HPIAC and enzyme-linked electrochemical detection were used to automate a sandwich assay for immunoglobulin G (9). In this system, sample and antibodies labeled with glucose oxidase were injected sequentially onto an appropriate HPIAC column. Several injections of glucose substrate were then made, and the hydrogen peroxide generated by labeled antibody on the column was determined amperometrically. By use of this approach, a total analysis time of 30 min per injection was obtained. The limits of detection for serum and standards were about 1 pmol and 0.2 fmol, respectively (i.e., 10-8-10-11 M for a 3 0 - ~ Lsample) (9). Although these results are a clear indication of the potential of HPIAC in automated sandwich immunoassays, the limits of detection reported are comparable to those already obtained with current automated systems (3, 4). For the determination of lower concentration analytes, this study will examine the use of HPIAC with an alternative means of detection, namely chemiluminescence. Chemiluminescence may be defined as the production of light as a result of a chemical process. This is a relatively common phenomenon in nature and has also been demonstrated in a number of synthetic systems (11). One area in which this is of particular interest is in the development of chemiluminescent labels for immunoassays (12). A number of compounds have been used for this ( 1 0 , but one group that has shown particular promise is the acridinium esters (12). The chemiluminescent reaction of acridinium esters is shown in Figure 1. These compounds are derivatives of 9-phenyl acridinium ester (see Figure la), with a side chain (R) attached for coupling them to antibodies or other molecules. They are useful as labels in immunoassays since they emit an intense flash of light in the presence of an alkaline peroxide solution, allowing them to be easily detected. In this reaction, hydrogen peroxide first undergoes dissociation to form its conjugate base, a hydroperoxyl anion (HO;). This anion then attacks the acridinium ring structure (Figure la), causing a concerted cleavage of multiple bonds that produces an excited molecule of N-methylacridone (11, 13). Once produced, the excited acridone quickly returns to its ground state (Figure Ib) by emission of light a t about 430 nm, allowing the label to be detected (12). One potential advantage of using acridinium esters for detection in HPIAC is that their chemiluminescence is very
try Figure 2. PTH sandwich immunoassay based on HPIACXL. The asterisk (*) represents the acridinium ester label. rapid, usually on the order of seconds. These compounds have also been shown in manual immunoassays to have excellent limits of detection, typically 10-17-10-18mol (i.e., 10-13-10-14 M for a 1OO-pL sample) (11). Such characteristics should allow their use in the development of detection systems for HPIAC that are sensitive and yet do not require a prolonged period of time for analyte detection. The need for only simple chemical reagents to trigger light production is also appealing since this could potentially be performed by the use of a standard postcolumn reactor. To test the use of HPIAC with chemiluminescent detection (HPIAC/CL) in the automation of sandwich immunoassays, this work used the determination of parathyroid hormone (PTH) in plasma as a model system. PTH is an 84 amino acid peptide hormone important in the regulation of calcium metabolism. Normal plasma concentrations of the intact to 6 X M, with both hormone range from 1 X elevated and decreased levels being of clinical interest (14). State-of-the-art determinations of P T H are currently performed with manual sandwich immunoassays with either chemiluminescent or radioactive labels (5, 15). Sandwich immunoassays are necessary not only to detect the low concentrations of P T H but also to avoid interferences from the large number of P T H fragments present in the circulation (5, 15). To do this, immobilized antibodies are used that bind one end of PTH, such as its 1-34 N-terminal region, and labeled antibodies are used that bind a second portion, such as PTH's 44-68 midmolecule region. In this way, only the intact P T H molecule will react with both the immobilized and labeled antibodies and will be detected (5, 14, 15). Like many other trace biological compounds, no system has previously been described for the automated determination of P T H (3). In this work, such a system was developed using an HPIAC column containing immobilized anti-(4448 PTH) antibodies for extraction of P T H from samples and acridinium ester-labeled anti-( 1-34 PTH) antibodies for the hormone's detection. The scheme used is shown in Figure 2. To initiate the chemiluminescenceof the labeled antibodies as they eluted with P T H from the column, a postcolumn reactor was used. This paper will consider various items important in the optimization of this HPIAC/CL system. These items will include the dissociation properties of the HPIAC column and methods for controlling the rate of the chemiluminescent reaction. The use of sequential vs simultaneous injection of sample and labeled antibody will also be discussed. The automated immunoassay system will then be compared to manual P T H methods in terms of its analysis time, precision, linearity, and limit of detection. The advantages and disadvantages of this system will also be discussed, as well as its potential use in the determination of other biological compounds.
588
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
Pump A
PF Appli
AC
I
Waste
Elution buffer Pump C
1-Y\
I
I
CD
1 Waste Postcolumn reagents
Figure 3. HPIACICL system for a sandwich immunoassay. Abbreviations: S, saturator column; PD, pulse dampener; MT, mixing tee; CD, chemiluminescence detector; Inj, injector/autosampler; V 1 and V2, switching valves 1 and 2; PF, precolumn filter; AC, affinity column.
EXPERIMENTAL SECTION Reagents. The 1-34, 1-44, and 44-68 fragments of human PTH and the 1-84 whole molecule were obtained from Peninsula (Belmont, CA). The cyanogen bromide activated Sepharose 4B, goat immunoglobulin G (IgG), human serum albumin (HSA), bovine serum albumin (BSA, RIA grade), Fruend's complete adjuvant, and 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide (EDC) were from Sigma (St. Louis, MO). Triton X-100, electrophoresis grade, was from Fisher Scientific (Fair Lawn, NJ). The 4-[2-[ (succinimidooxy)carbonyl]ethyl]phenyl 10-methylacridinium-9-carboxylate fluorosulfonate (i.e., acridinium ester) was from London Diagnostics (Eden Prairie, MN). Nucleosil Si-1000 (7-pm particle diameter, 1000-8, pore size) was obtained from Alltech (Deerfield,IL). Sephadex G-25 was from Pharmacia (Piscataway, NJ), and the reagents for the bicinchoninic acid (BCA) protein assay were from Pierce (Rockford, IL). The 0.25-in. polymeric beads used for the manual PTH immunoassay were from Micromembranes, Inc. (Newark, NJ). All solutions were prepared with water from a Milli-Q water system (Millipore, Bedford, MA). Instrumentation. Figure 3 is a schematic diagram of the final system used for the PTH determinations. This consisted of a Hitachi L-6200 ternary pump (Tokyo, Japan) for mixing and pumping postcolumn reagent. Two Hitachi L-6000 isocratic pumps were used for delivering the application and elution buffers to the affinity column. Samples were injected by using a Hitachi 655A-40 autosampler. Two 4.0-mm-i.d. X 2.0-cm saturator columns containing Nucleosil Si-lo00 diol-bonded silica were placed immediately after the isocratic pumps. Components leaving the affinity column in the elution buffer were detected by combining the elution buffer and postcolumn reagents with a standard mixing tee and passing them through an 825-CL chemiluminescence detector (Jasco, Easton, MD). Nonretained components eluting in the application buffer were monitored a t 280 nm by using a Hitachi 665A UV detector. The application and elution buffers were alternately applied to the affinity column by using a column switching system, as shown in Figure 4. This consisted of a Rheodyne 5701 tandem enrichment valve and a Vici DVI actuator (Chromtech, Apple Valley, MN) controlled from the L-6200 pump. By placing a precolumn filter (Upchurch, Oak Harbor, WA) on the first valve and the affinity column on the second, this switching system allowed particulate matter to be removed from the sample before reaching the affinity column and for both the filter and column to be backflushed during the elution step. This approach not only helped decrease the total cycle time for analysis but also minimized increases in the back pressure with sample application, helping to increase the column lifetime. Antiserum Production. Anti-(4448 PTH) antiserum (Mayo 25G 10/20/88) was prepared by conjugating the 44-68 fragment of PTH to BSA with EDC (16). The conjugate was purified by overnight dialysis vs water using as 12000-14000 MW cutoff membrane. A total of 100 pg of conjugate was placed in 0.9% sodium chloride, 0.10 M potassium phosphate buffer, pH 7.4,
Application buffer
t
Detector
I
Elution buffer
Figure 4. Valve switching scheme for the HPIACXL system. Abbreviations: PF, precolumn filter; AC, affinity column. The injection and elution positions of the valves are indicated by (-) and (---), respectively.
mixed 1:l with Freund's complete adjuvant and injected subcutaneously into goats on a monthly basis. Anti-(1-44 PTH) antiserum (Mayo 950G 11/5/84) was similarly prepared with BSA-conjugated 1-44 PTH being used as the initial immunogen, followed later by injections of unconjugated 1-44 PTH. Antibody Purification. Antibodies were purified by using affinity columns containing 1-34,l-44, or 44-68 PTH fragments immobilized onto cyanogen bromide activated Sepharose 4B. One milligram of PTH fragments per gram of Sepharose 4B was immobilized according to the manufacturer's instructions. The 1-34, 1-44, and 44-68 PTH supports were then poured into separate 0.7-cm i.d. X 15-cm glass columns and equilibrated with 0.10 M potassium phosphate buffer, pH 7.4. Anti-(44-68 PTH) antibodies were purified by centrifuging Mayo 25G 10/20/88 antiserum at 3000 rpm for 5 min and applying 2 mL of the supernatant to the 44-68 PTH Sepharose 4B column. The column was then washed with 0.10 M potassium phosphate buffer a t pH's of 7.4 and 5.0. Aliquots of 1 mL were collected as 0.10 M potassium phosphate buffer, pH 3.0, was applied. After collection,the fractions were immediately adjusted to pH 6.0 and stored a t 4 "C for 24 h. The remaining antibody on the column was eluted with 0.10 M potassium phosphate buffer, pH 2.0, and the column was regenerated by applying the pH 7.4 phosphate buffer. Anti-(1-34 PTH) antibodies were purified on the 1-34 PTH Sepharose 4B column by applying 2 mL of centrifuged Mayo 950G 11/5/84 anti-(1-44 PTH) antisera. This column next was washed with 0.9% sodium chloride and 0.1% Triton X-100 in the pH 7.4 phosphate buffer, followed by elution of the antibodies with 0.001 M hydrochloric acid. The antibodies collected were dialyzed overnight vs pH 7.4 phosphate buffer, lyophilized, and stored at -20 "C. Antibody Immobilization. Diol-bonded Nucleosil was prepared as described previously (17). The diol coverage of the Nucleosil prior to activation was 90 f 4 (1 SEM) pmollg of silica, as determined in duplicate by the periodate oxidation method (18, 19). Anti-(44-68 PTH) antibodies were immobilized onto the diol-bonded Nucleosil by using the Schiff base method (20), with the silica being sonicated under vacuum for 15 min a t the beginning of the activation and immobilization steps (21). After sonication in the immobilization step, the antibodies were added to the silica suspension, and the resulting mixture was shaken at 4 "C for 3 days. Also added to the silica suspension was 37.5 mg of sodium cyanoborohydride/g of silica. This reagent was used to reduce any Schiff bases that formed between antibodies and the support. This resulted in the production of a stable secondary amine linkage (20). After the immobilization reaction, any aldehyde groups remaining on the silica were reduced by adding 25 mg of sodium borohydride/g of silica (20) and shaking the mixture overnight at 4 "C. The silica was then centrifuged, washed with 2 M sodium chloride and 0.10 M phosphate buffer, pH 7.4, and stored at 4 "C. Part of the silica was further washed with water and vacuum dried a t room temperature. The dried silica was digested (20) and the supernatant assayed for protein content by using the BCA method (22),with goat IgG as the standard and diol-bonded silica as the blank.
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
Anti44448 PTH) antibodies were immobilized onto the 0.25-in. polymeric beads according to the manufacturer's instructions. A total of 0.25 pg of antibodies per bead was used in this procedure. After preparation, the beads were washed with 0.10 M phosphate buffer, pH 7.0, and stored at 4 "C. Chemiluminescent Labeled Antibodies. The anti-(1-34 PTH) antibodies were labeled with acridinium ester as described earlier (12),using 4 pug of acridinium ester/mg of antibody. The labeled antibodies were initially purified over a 0.7-cm4.d. X 25-cm Sephadex G-25 column and eluted with 0.9% sodium chloride, 0.1 70 BSA, and 0.01% sodium azide in 0.10 M potassium phosphate buffer, pH 6.3. Label was detected by collecting 1-mL fractions of the eluent, preparing aliquots in 0.10 M phosphate buffer, pH 3.0, and measuring the chemiluminescence of the aliquots on a Ciba Corning Magic Lite analyzer (Medfield, MA) using reagents supplied by the manufacturer. The labeled antibody fractions were then pooled and dialyzed for 24 h vs 0.10 M phosphate buffer, pH 7.4, to remove any acridinium ester noncovalently bound to antibody. Inactive antibodies were removed by applying the dialysate to the 0.7-cm-i.d. x 15-cm 1-44 PTH Sepharose 4B affinity column and eluting the retained and nonretained fractions with the procedure already described for the 1-34 PTH column. The labeled antibodies were stored at -20 "C and were diluted prior to use with 0.10 M phosphate buffer, pH 7.4, containing 0.1% BSA, 0.1% Triton X-100, and 0.1% goat IgG. A 1:50 dilution was arbitrarily set as being equal to one unit of the labeled antibody. Chromatography. The anti-(44-68 PTH) Nucleosil support was downward slurry packed at 3500 psi into either 2.0-mm or 4.0-mm4.d. X 2.0-cm columns from Upchurch. The application buffers for these columns were made with 0.10 M potassium phosphate buffer, pH 7.4, and the elution buffers were made with 0.10 M potassium phosphate buffer, pH 3.0. All chromatography was performed at room temperature. PTH plasma samples used in the chromatographic studies were prepared by spiking 1-84 human PTH into EDTA plasma collected from patients with no endogenous PTH production. Plasma samples were applied to the affinity column in pH 7.4 application buffer containing 0.1% Triton X-100 and 0.1% BSA. Retained PTH was removed from the column by using pH 3.0 elution buffer containing 0.1% Trixon X-100. Prior to injection, all samples were filtered by using 0.45-pm ACRO LC13 disposable filters (Gelman Sciences, Ann Arbor, MI). The binding capacity and stability of the anti-(44-68 PTH) support were determined by frontal analysis. This involved continuously applying 5 pg/mL 44-68 PTH in pH 7.4 phosphate buffer, 0.10 M, to the 2.0-mm-i.d. X 2.0-cm column at 0.025 mL/min. The amount of 44-68 PTH needed to saturate the column was then determined by integration of the resulting breakthrough curves (23) and correcting for the system void volume. Frontal analysis with HSA was similarly performed, with 10 pg/mL HSA being used in place of 44-68 PTH in the application buffer. Potential kinetic effects in the binding capacity measurements (24,25)were minimized by using a relatively long column residence time (Le., 2.0 min). The magnitude of such effects under these conditions was estimated by using procedures described in ref 7 . This was done by injecting 100 pL of a 2:l mixture made up of 0-250 pmol/L PTH plasma and one unit of the labeled antibody. These injections were made onto the 4.0-mm4.d. X 2.0-cm HPIAC column at flow rates of 0.33-2.0 mL/min. At each flow rate, the relative amount of the PTH-labeled antibody complex retained by the column was determined by using the postcolumn reactor system described in the Results and Discussion section. From these data and equations given in ref 7 , the apparent first-order adsorption rate constant for the binding of PTH to the column was determined. The estimated value of this rate constant was greater than 0.5 s-'. Although this value would be expected to decrease with sample size (251,it did show that binding of PTH to the column was quite rapid and indicated that the kinetic effects reported in the literature (24,25)were probably not significant under the conditions used in the binding capacity measurements. Nonspecific adsorption of 1-84 PTH to the diol-bonded support was tested by injecting 100 pL of 1000 pg/mL PTH plasma onto a 4.0-mm-i.d. x 2.0-cm column containing diol-bonded silica.
589
Fractions of 0.2 mL were collected in tubes containing 100 p L of PTH hypoplasma. The amount of PTH in each fraction was determined by using a PTH immunochemiluminometric assay (ICMA) ( 5 ) . The total PTH injected was determined by both calculation and collecting fractions immediately after the injector for measurement on the Ciba Corning analyzer. Nonspecific adsorption of other plasma components and acridinium ester-labeled antibodies to both the diol-bonded silica and immunoaffinity support was tested by injecting 100 p L of plasma or labeled antibodies onto 4.0-mm-i.d. X 2.0-cm columns containing the desired support. Retention of plasma components, particularly proteins and peptides, was determined by monitoring the eluent at 280 nm. Binding of the labeled antibodies was determined by collecting fractions and measuring their chemiluminescence on the Ciba Corning analyzer. The postcolumn reagent was optimized by diluting the labeled anti-(1-34 PTH) antibodies in pH 3.0 elution buffer containing 0.1% Triton X-100 and injecting 25 pL of this solution into the same buffer flowing at 1.0 mL/min. This was combined directly with postcolumn reagent, also flowing at 1.0 mL/min, and immediately passed through the chemiluminescence detector. As injections of labeled antibody were made, the relative amounts of sodium hydroxide, hydrogen peroxide, and Triton X-100 in the postcolumn reagent were varied while the size of the resulting peak was monitored. The reagent concentrations needed for optimum detection were then determined by using both direct analysis of the peak height data and simplex optimization (26).
RESULTS AND DISCUSSION Binding Properties of the HPIAC Support. The immobilized antibody support is a key factor in the successful use of HPIAC (6, 7). One desirable characteristic for such a support is that it have low nonspecific adsorption for labeled antibodies, the analyte, and other injected components. This is required to produce a selective HPIAC system with good limits of detection. For use in a sandwich immunoassay, the support must also contain an excess of immobilized antibodies vs the amount of analyte to be injected. This is needed for rapid and complete binding of analyte as it passes through the HPIAC column. Lastly, it is desired that the immobilized antibodies be highly active and be easily accessible to the analyte for binding. This allows only a minimal amount of antibody to be used for the extraction of analyte from samples. In this work, each of these characteristics was considered in developing HPIAC columns for an automated PTH sandwich immunoassay. The support used in this study was based on diol-bonded silica. This material is often used in HPIAC since it has little or no binding to most biological molecules and yet can be easily chemically modified for antibody attachment (6,20). In this work, the nonspecific binding of such a support was tested by placing it into a column and injecting 100 p L of solutions containing acridinium ester labeled antibodies or P T H plasma samples. A 100-pL sample size was chosen for these experiments since it represented the maximum injection volume desired for use in the final HPIAC system. Initial experiments with on-line detection at 280 nm revealed that less than 0.5% of the proteins and peptides in the plasma samples were adsorbed by the diol-bonded silica under these conditions. However, off-line chemiluminescent detection showed that a significant amount of the labeled antibodies did bind to this support. It was later found that this binding could be greatly reduced by the addition of 0.1% BSA and 0.1% Triton X-100 to the application buffer (27). When these additives were used, less than 0.5% of the labeled antibodies adsorbed nonspecifically to the column. Under the same conditions, nonspecific adsorption of PTH was also quite low, representing less than 5% of the total amount injected. This latter value is particularly impressive since P T H is known to strongly adsorb to a variety of substances, including silica and glass ( 2 8 ) . Overall, these levels of nonspecific binding were similar to those observed in manual immunoassays and were
590
ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991
Table I. Properties of the Anti-PTH Immunoaffinity Support
value
property
antibody
18 & 4 nmol antibody/g silica
immobilized”
antibody coverageb binding capacity specific activity
20
t
0.007 monolayer 12.6 f 0.1 nmol 44-68 PTH/g silica 0.70 f 0.05 mol 44-68 PTH/mol antibody 0.034
“Determined by using a molecular weight for goat IgG of 150000 g/mol. bDetermined by using a surface area of 25 m2/g for Nucleosil Si-1000 and a Stoke’s diameter of 100 A for IgG (29). considered satisfactory for use in an HPIAC-automated system. Once nonspecific adsorption onto the diol-bonded silica had been reduced to acceptable levels, the support was then used to immobilize anti-PTH antibodies for use in an HPIAC column. The properties of the resulting immobilized antibody support are summarized in Table I. As indicated in Table I, the amount of antibody immobilized was relatively small, with a value of only 18 nmol or 2.7 mg of antibody/g of silica. This gave an antibody coverage on the support of about 0.03 monolayer. The use of such a low-coverage support was possible due to the small amount of PTH to be measured. For example, it was calculated that 100 pL of even a very elevated PTH sample (i.e., 100 pmol/L) would require only 1pg of this support in order to provide enough immobilized antibodies for quantitative binding. As will be discussed later, this antibody coverage was also sufficient to provide rapid binding of analyte to the column, with greater than 99% of the P T H being extracted from a sample in as little as 6 s. Thus, this coverage and antibody excess was considered adequate for use in the HPIAC system. To study the activity of the immobilized antibodies, the binding capacity of the immunoaffinity support was determined by frontal analysis using 44-68 PTH. On the basis of nine successive measurements, the total binding capacity was determined to be 12.6 nmol of 44-68 P T H / g of silica. By combining this result with the antibody coverage of the support, the specific activity of the immobilized antibodies could also be determined. As shown in Table I, this specific activity was found to be 0.70 mol of 44-68 PTH/mol of antibody. This represents about 35% of the antibody’s initial activity, assuming a maximum binding of two 44-68 PTH molecules per antibody. Although this result is somewhat lower than those obtained in other studies (7), it still indicates that a significant fraction of the antibodies retained their activity throughout the immobilization process and was accessible to analyte molecules for binding. Nonspecific binding of the immobilized antibodies to plasma components and to the labeled antibodies was tested in the same manner as used for the diol-bonded silica. In these experiments, no significant change was noted from the results observed with the diol-bonded support (Le.,
