Journal oflmmunologicalMethods, 134 (1990) 71-79
71
Elsevier JIM05733
A silicon sensor-based filtration immunoassay using biotin-mediated capture John D. Olson, Peter R. Panfili, Richard Armenta, Mary Beth Femmel, Holly Merrick, Jenny Gumperz, Marion Goltz and Robert F. Zuk Molecular Devices Corporation, 4700 Bohannon Drive, Menlo Park, CA 94025, U.S.A.
(Received 26 June 1990; accepted 10 July 1990)
A sensitive sandwich immunoassay for human chorionic gonadotropin (hCG) was developed with biotin-mediated filtration capture and silicon sensor detection. A high density of biotin on the membrane assured efficient capture of complexes containing streptavidin and analyte. Capture efficiency was not affected over a wide range of filtration flow rates or biotin concentrations. The assay utilized the pH sensing ability of the light addressable potentiometric sensor (LAPS) for the detection of urease-antibody conjugates. A LAPS reader was constructed which allowed the enzyme conjugate to be detected in - 1/tl volumes. Effects from variations in detection volume were studied. 10 pg of h C G could be detected in an assay time of 20 rain with four standard deviations separation from background. Comparison to a commercial R I A was made. Key words." Sensor; Biotin; Immunoassay;Streptavidin; Light addressable potentiometric sensor; Filtration
Introduction
Semiconductor sensing devices offer several intrinsic advantages for biochemical measurements, including adaptability for solid phase and particulate samples, potential for miniaturization and very high sensitivity, and multiplicity of measurement sites (Karube, 1987). Immunoassay is one
Correspondence to: J.D. Olson, Molecular Dj~vicesCorporation, 4700 Bohannon Drive, Menlo Park, CA 94025, U.S.A. Abbreviations: B-BSA, biotin-bovine serum albumin; CHEMFET, chemically sensitive field effect transistor; hCG, human chorionic gonadotropin; LAPS, light addressable potentiometric sensor; PBS, phosphate-buffered saline; RIA, radioimmunoassay; SMCC, succinimidyl 4-(N-maleimidomethyl)cyclohexane-l-carboxylate; SPDP, N-succinimidyl 3(2-pyridydithio)propionate; TNBS, 2,4,6-trinitrobenzene sulfonic acid.
area of analytical biochemistry where sensors can have immediate utility. Janata (1975) proposed a chemically sensitive field effect transistor ( C H E M F E T ) in which chemical specificity was endowed by immobilization of an antibody in the gate region of the transistor. A direct reading immunochemical C H E M F E T was not successful because the antibody binding site was located too far from the gate region to alter interfacial charge density (Janata and Blackburn, 1984). Demonstration of enzyme-based C H E M F E T S (Blackburn, 1987) was viewed as an advance in the development of a sensor-based immunoassay because of the signal amplification the enzyme offers. The most promising practical approach employs p H sensing with a silicon nitride insulator which monitors the change in proton concentration caused by en-
0022-1759/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
72 zymes such as urease, glucose oxidase, penicillinase, and acetylcholinesterase. The utility of CHEMFETs is limited by the difficulty in reliably encapsulating micro-semiconductor devices which function in aqueous environments. A light addressable potentiometric sensor (LAPS) (Hafeman et al., 1988) is a simple semiconductor device that functions much like a CHEMFET. The LAPS device contains a monolithic sificon structure with a planar sensing surface with no microcircuitry that needs to be encapsulated. Sensor planarity facilitates formation of reproducible microvolumes. In an assay based on pH change caused by enzyme activity, decreasing the assay volume to concentrate enzyme increases sensitivity. Another salient feature of the LAPS is the capability for multiple potentiometric measurements. A single LAPS semiconductor device can detect multiple concurrent chemical reactions by monitoring the photocurrent induced through illumination of discrete sensing sites. Simultaneous determinations of several samples, controls, and calibrators improves assay throughput and accuracy. A solid-phase enzyme immunoassay was developed, incorporating microvolume detection with LAPS. The assay format entails formation of specific immune complexes in liquid phase followed by membrane filtration. Streptavidin-biotin (Wilchek and Bayer, 1988) mediates the capture of the immune complex on the membrane. Membrane-bound enzyme is precisely measured in a LAPS reader designed to accept membranes. Use of microporous membranes to capture immune complexes is compatible with LAPS. A variety of membrane based immunofiltration methods have rapid, efficient wash protocols and the capability for parallel processing of multiple samples (IJsselmuiden et al., 1989). The membrane interstitial volume can be used as microvolume reaction chambers to increase sensitivity in a pH sensing assay. We describe the filtration immunoassay using the LAPS and major parameters for optimizing the filtration capture step and the detection of enzyme conjugates in microvolumes. A quantitative immunoassay with picogram sensitivity in biological samples requiring a total assay time of 20 rain is described.
Materials and methods
Chemical and biochemical reagents Bovine serum albumin (BSA) (A-7888), urease type VII (U-0376), dithiothreitol (DTT) (D-0632), diaminobiotin-agarose (D-1395), 2,4,6-trinitrobenzene sulfonic acid (TNBS) (P-3402), 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (E-7750) and succinic anhydride (S-7626) were purchased from Sigma Chemical Company, St. Louis, MO. Streptavidin (S-1214) was purchased from Scripps Laboratories, San Diego, CA. lzsI-labeled streptavidin (IM 236) was purchased from Amersham, Arlington Heights, IL. Radioiodination kit (NEZ151) was purchased from New England Nuclear, Boston, MA. Urease labeled streptavidin (44000U) was purchased from JD Biologics, Downsview, Ont., Canada. Anti-human chorionic gonadotropin (anti-hCG) monoclonal antibodies were obtained from Dupont, Wilmington, DE. Human chorionic gonadotropin (hCG) was purchased from Immunosearch, Emeryville, CA. Tandem-R hCG immunoradiometric assay was purchased from Hybritech, San Diego, CA. Succinimidyl 6-(biotinamido)hexanoate (S-1582), succinimidyl 4(N-maleimidomethyl)cyclohexane-l-carboxylate (SMCC) (S-1534), and N-succinimidyl 3-(2-pyridyldithio)propionate (SPDP) (S-1531) were purchased from Molecular Probes, Eugene, OR. Dimethyl formamide (DMF) (50117) was purchased from Pierce Chemical Co., Rockford, IL. 2,2'-oxy(bisethylamine) (17,609-5) was purchased from Aldrich Chemical Company, Milwaukee, WI. Biogel A5M was purchased from Biorad, Hercules, CA. Sephadex G100 (17-0060-1) and Sephadex G-25 columns (PD-10) were purchased from Pharmacia, Piscataway, NJ. Preparation of anti-hCG-urease conjugates Urease was rehydrated in 50 mM (Na2H / NaH2)PO4, 150 mM NaC1 pH 7.0 (phosphate buffer) with 10 mM DTT, and incubated for 1 h to disaggregate the enzyme. The enzyme was separated from the DTT by using a Sephadex G-25 column and reacted with antibody within 1 h. Urease was conjugated to anti-hCG by using a modification of the method described by Ishikawa et al. (1983). Maleimide groups were introduced onto purified monoclonal antibodies specific for
73 the fl subunit of h C G by reaction with SMCC. Maleimidyl-antibody was incubated with urease for 3 h at 4°C. The conjugates were purified using a Biogel A5M column (2.5 x 90 cm) equilibrated in phosphate buffer plus 100 m M Na2SO3, 1 m M EDTA, 2 m M DTT, 0.05% N a N 3 (pH 7.0). Column fractions containing the conjugates were pooled and stored under N 2 at 4°C.
Preparation of anti-hCG-streptavidin conjugates Streptavidin was conjugated to anti-hCG by using a modification of the methods described by Carlsson et al. (1978) and by Ishikawa et al. (1983). Maleimide groups were introduced onto purified monoclonal antibodies specific for the ct subunit of h C G by reaction with SMCC. Sulfhydryl groups were introduced onto streptavidin by reaction with SPDP followed by DTT. The sulfhydryl-streptavidin was incubated with maleimidyl-antibody overnight at room temperature. Conjugate was purified using a diaminobiotina g a r o s e / S e p h a d e x G-100 column. 5 ml of diaminobiotin-agarose was packed on the top of Sephadex G-100 in a 2.5 x 30 cm column equilibrated in phosphate buffer. The conjugate reaction mixture was applied to the column and the column was washed thoroughly. To elute the bound streptavidin conjugates, 10 ml of 0.1 M sodium citrate buffer (pH 3.5) were added. Phosphate buffer was then used to chromatograph the antibody-streptavidin conjugates from free streptavidin. The fractions containing conjugate were pooled and stored in phosphate buffer at 4°C.
Preparation of biotin-BSA BSA at 40 m g / m l was dialyzed against 100 m M N a H C O s, 150 m M NaC1 p H 8.2 (bicarbonate buffer). A 200-fold molar excess of KzCO 3 followed by a 200-fold molar excess of succinic anhydride (40 m g / m l in D M F ) was added to the BSA. After 1 h at room temperature.the mixture was dialyzed against bicarbonate buffer. The succinyl-BSA was concentrated to 40 m g / m l by tangential flow ultrafiltration (Minitan, Millipore, Bedford, MA) and 2,2'-oxy(bisethylamine) was added to a final concentration of 0.1 M. While stirring, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide was added at a molar ratio to BSA of 220 : 1. After 1 h at room temperature the mixture
was exhaustively dialyzed against bicarbonate buffer. The aminated BSA was adjusted to 15 m g / m l . Succinimidyl 6-(biotinamido)hexanoate (40 m g / m l in D M F ) was added at a molar ratio of 20 : 1. After 1 h at room temperature the reaction mixture was dialyzed against phosphate buffer. The number of biotin groups on the BSA was estimated by the consumption of free amino groups by using the TNBS assay (Fields, 1972) and determined to be approximately 10 per BSA molecule.
Preparation of biotin-BSA coated nitrocellulose An immobilization solution containing biotinBSA (B-BSA) at 5 m g / m l in phosphate buffer was prepared. Nitrocellulose m e m b r a n e (BA-85, Schleicher and Schuell, Keene, N H ) was saturated with the solution and placed on the b o t t o m of a glass dish for 45 min. The m e m b r a n e was washed with phosphate buffer then crosslinked with 0.1% glutaraldehyde in phosphate buffer for 2 h at room temperature. The m e m b r a n e was washed twice with phosphate buffer for 15 min each followed by deionized H 2 0 for 15 min then dried in a convection oven at 65°C. To determine the amount of biotin-BSA coated onto the m e m b r a n e a trace amount of biotin-125I-BSA (radioiodinated using the NEN kit) was spiked into the B-BSA solution. Coating density was determined by testing 0.5 cm 2 samples of m e m b r a n e in a g a m m a counter (Beckman no. 5500). B-BSA density was approximately 6 0 / . t g / c m 2. For some experiments, nitrocellulose membrane was coated with varying amounts of B-BSA, maintaining total protein concentration of the coating solution at 5 m g / m l by the addition of native BSA.
Filtration capture assay The membrane was mounted with adhesive transfer tape (Y-9460, 3M Corp., Minneapolis, MN) over a window cut in a vinyl acetate stick (Transvy, Transilwrap West Corp., So. San Francisco, CA). Membrane sticks were placed into filter units which directed the flow of solution through the membrane. Filtration rate was controlled by a regulated vacuum unit (Threshold, Molecular Devices Corp., Menlo Park, CA). Streptavidin was diluted to 100 n g / m l in 10 m M ( N a 2 H / N a H 2 ) P O 4 , 150 m M NaC1 p H 7 (PBS)
74
with 0.1% BSA, 10 mM Na2SO3, 1 mM EDTA, 0.05% NAN3, 1 m M D T F (conjugate dilution buffer). 125I-Streptavidin was added to the streptavidin solution to a specific activity of 227 c p m / n g streptavidin. 91% of the cpm were precipitated by 10% trichloroacetic acid (TCA). The specific activity of active streptavidin therefore was taken to be 207 cpm/ng. The streptavidin solution (1 ml) was filtered (n = 3) through the membrane and then washed with 500 /L1 of PBS, 0.1% BSA, 0.05% Tween 20 (wash buffer) at a flow rate of 500 /~ml/min. The filtration area of the membrane was removed from the stick and analyzed on a Beckman 5500 gamma counter to estimate streptavidin capture.
I Stick VariableBiasVoltage
Enz)
hCG assayprotocol Fig. 1 depicts the sandwich immunoassay for hCG. hCG was diluted in pooled human serum collected from healthy male donors, and values assigned by repeated testing using the Hybritech RIA. Anti-hCG-urease conjugate was mixed with anti-hCG-streptavidin conjugate in conjugate dilution buffer, hCG serum calibrator or patient sample (10 #1) was added to 300 /~1 of conjugate solution and incubated for 15 rain at 37°C. The mixture was filtered through B-BSA membrane at a rate of 100 ~ l / m i n . The membrane was washed with 500/~1 wash buffer at a rate of approximately 500 #l/rain. The area of filtration was 0.14 cm 2.
1. Incubate
El--X Anli-I~ Urease
~,
EF-x hCG
2. Filter,Wash
>-x f" Urea
3. Read
/////////~//////////////////////////// Figure 1. Schematic of hCG immunoassay.
Anti~ Streptavidin
Biotinylated Membrane
/"~'~/
Fig. 2. Schematic of LAPS reader.
The wash step removed unbound conjugate and equilibrated the membrane at the appropriate pH and buffer concentration for subsequent potentiometric measurement of bound enzyme. The LAPS reader, illustrated in Fig. 2, had a vertical sensing surface with a p H sensitive silicon nitride insulator contacting the electrolyte, an enzyme substrate solution (100 mM urea in wash buffer). The membrane stick with bound conjugate was inserted into the reader. When placed in the LAPS reader, the filtration area aligned precisely with the illuminated area on the sensor. A plunger compressed the membrane against the sensor reducing the assay volume to as low as 1/zl. A dial indicator positioned on the plunger mechanism provided an indirect measurement of the plunger stop distance from the sensor surface. The dial indicator was calibrated to 1 t~m intervals. A threaded plunger and lock nut enabled adjustment of the plunger stop. With the membrane against the sensor, data collection was started, Urease-catalyzed hydrolysis of urea to ammonia and CO 2 induced a p H change. This altered the surface potential at the electrolyte-silicon nitride interface. The potential at the
75
interface was monitored through a transient photocurrent produced by illumination with a light emitting diode behind the silicon surface opposite the filtration area on the membrane. Modulating the illumination intensity creates an alternating photocurrent in the external circuit. The magnitude of the detected photocurrent depends on the sum of the applied bias voltage, the surface potential at the electrolyte-insulator interface and the reference electrode potential. Measurements of surface potential were made every second for 1 min. The results were collected and analyzed using an IBM PS2/30 computer with specially designed software. The rate of change of pH was determined by the slope of the plot of change in surface potential vs time in /~V/second. A change of 1 pH unit was approximately equivalent to 50,000 ~V.
Results
Biotin mediated separation step The hCG sandwich procedure yields a liquid phase immune complex containing streptavidin and urease. The separation step employed filtration capture of the immune complex to a biotinylated nitrocellulose membrane. Initial studies characterized the filtration capture of streptavidin as a function of B-BSA density and flow rate (Fig. 3). At a flow rate of 170 /~l/min the streptavidin capture was > 95% at all B-BSA densities tested (>_ 1.8 /~g/cm2). At the higher flow rate of 1000 /~l/min streptavidin binding was reduced below
40
20 0~ 0
5
10
B - B S A on M e m b r a n e
15
20
(p.g/cm 2)
Fig. 3. Effect of m e m b r a n e b i o t i n density a n d flow rate u p o n c a p t u r e of strcptavidin. The flow rate was adjusted to 170 / L l / m i n (D) or 1000 p l / m i n (O). E r r o r bars indicate + 1 s t a n d a r d deviation.
30OO
;~ 2000
0 ~ 0
100
± 200
Filtration Flow Rate
300
":~''" 400
(~tL/minute)
Fig. 4. Capture of h C G complexes is independent of flow rate. The standard h C G assay was run with 0 pg ([3) or 1000 pg ( , ) h C G (n = 4). The filtration rate was varied during the i m m u n e complex separation step. Error bars represent + 1 standard deviation.
90% only when the B-BSA density was 6 / ~ g / c m 2 or less. The BSA was labeled with biotin at a molar ratio of 10 per BSA. A similar dependence of filtration flow rate and biotin density was observed when the biotin substitution on BSA was varied. When the amount of BSA immobilized was held constant, streptavidin capture efficiency was reduced as the biotin labeling approached a molar ratio of 1 : 1 (data not shown). These results suggested a means of achieving reproducible filtration capture of streptavidin-containing immune complexes. Conditions favoring maximal capture were selected: slow filtration flow rates, high biotin substitution on BSA and B-BSA immobilization levels greater than 2 0 / ~ g / c m 2. Fig. 4 illustrates a LAPS filtration immunoassay for hCG using nitrocellulose membrane coated with B-BSA at a density of 6 0 / ~ g / c m 2. B-BSA with a molar ratio of 10:1 was used for the membrane coating. Immune complexes were formed by incubation of the urease and streptavidin anti-hCG conjugates with 1.0 ng hCG. The mixtures were then filtered onto the biotinylated membrane at different flow rates to test if the immune complex sterically hindered capture. To within experimental error the enzymatic signal was identical with filtration rates from 25 to 350 /~l/min indicating robust, reproducible capture of the hCG immune complexes.
Optimization of microoolume p H detection In the pH measurement of enzyme activity, consideration of the major elements contributing to the total buffer capacity of the system is neces-
76 400
700
•
"
,
.
,
.
,
600 300
=~ 500 400
~ 2o0
300
d- 200 D. 100
r~
0
I
I
I
I
20
40
60
80
0 50
10O
Conc. of Phosphate (mM)
I
I
I
100
150
200
Measured Distance (l.tm)
Fig. 5. Effect of b u f f e r c o n c e n t r a t i o n on assay signal. S a m p l e s were filtered at 100 # l / m i n ( n = 4). T h e m e m b r a n e was w a s h e d
Fig. 6. T h e effect of assay v o l u m e u p o n assay signal. T h e slope of the r e s p o n s e was d e t e r m i n e d for distances _< the thickness of
with p H 7.0 b u f f e r c o n t a i n i n g the indicated p h o s p h a t e con-
the m e m b r a n e ( - 150 um). E r r o r b a r s i n d i c a t e + 1 s t a n d a r d deviation.
centration. T h e L A P S r e a d e r was filled with the s a m e w a s h b u f f e r plus 100 m M urea. E r r o r bars indicate + 1 deviation.
standard
sary. The effect of phosphate buffer concentration in enzyme substrate used in the LAPS reader is shown in Fig. 5. Urease was bound to the membrane by mixing the urease and streptavidin antih C G conjugates with 0 or 200 pg h C G followed by filtration through the B-BSA membrane• The enzyme assay was carried out in substrate formulated with phosphate buffer ranging from 10 to 100 mM. For both 0 and 200 pg hCG, the observed enzymatic rate of change of potential diminished as the buffer concentration increased• For the standard h C G assay 10 m M phosphate was chosen as substrate buffer• Although higher signals were obtained at phosphate concentrations lower than 10 mM, signal variation due to dissolved CO 2 and membrane buffer effects were observed. Reducing the volume of the reaction chamber should increase sensiti{,ity. The closure of the plunger compresses the membrane against the LAPS sensing surface increasing the local concentration of enzyme and reducing the phosphate buffering capacity• The impact of modulating the reaction microvolume is shown in Fig. 6. The standard h C G assay was run with 0 and 200 pg hCG. The LAPS measurement of captured urease employed the plunger stop set 70-210 /~m from the sensor surface• The thickness of the hydrated nitrocellulose membrane before compression by the plunger was determined to be about 150 /~m. The estimated reaction volumes ranged from less than 1•1-2.4/~1 for plunger stop distances from 70 to 150/~m respectively. The slope of the response
curve as the plunger approached the silicon was used to determine that the enzyme rate increased 4 /~V/s//~m at 200 pg hCG.
Filtration immunoassay performance This filtration immunoassay combined the elements of liquid phase formation of an immune complex, the biotin mediated filtration capture, and the microvolume urease activity measurement by the LAPS reader• For the measurement of membrane-bound urease the LAPS reader had the plunger stop set at 100 /~m from the sensing surface. Urea substrate was formulated with 10 m M phosphate• Total assay time was less than 20 rain. Fig. 7 is a typical standard curve. Table 1 shows data from the assay of various levels of h C G spiked into pooled h u m a n serum. The limit of detection for the assay was below 5 m I U / m l (10 pg/assay), the lowest concentration tested• The 5 m I U / m l level was statistically different from the blank at the 99.8% confidence limit by Student's one-tailed t test. Coefficients of varia-
200
•
,
•
,
•
,
•
,
•
,
"
150 :=L
Z loo ¢¢
50 zo*0
0: 0
i 10
i
i
i
i
20
30
40
50
60
[hCG] (pg) Fig. 7. h C G s t a n d a r d curve. C a l i b r a t o r s were r u n in replicates of 4. E r r o r bars r e p r e s e n t + 1 s t a n d a r d deviation.
77 TABLE I SPIKE RECOVERY Precision and accuracy. Samples of hCG were added into pooled serum and assayed (n = 20). Values were calculated from a standard curve and % recovery and quantitation CV were determined. hCGadded (mIU/ml)
hCGrecovered (mIU/ml)
Recovery (%)
Quant. CV (%)
5 25 100 500
5.8 26 102 558
116 104 102 112
13 5 7 4
tion were 13% at the 5 m l U / m l detection limit and substantially less than 10% at hCG levels from 25 to 1000 m I U / m l . Potential for accurate quantitation by the LAPS system was established by a comparison with a commercial RIA. This filtration immunoassay quantitated hCG up to levels of 1000 m I U / m l . The highest calibrator in the RIA was 400 m I U / m l ; samples with greater than 400 m I U / m l were diluted 1/10 and rerun in the RIA. Fig. 8 illustrates the results of quantitating 31 sera. Linear regression analysis yielded a slope of 0.92 and a r 2 value of 0.957.
Discussion
We report a sensitive immunoassay integrating features of the LAPS reader and membrane filtration. The flat sensing surface of the monolithic silicon facilitates microvolume measurements of membrane-bound enzyme with high precision and 2
looo
y = 0'.92x'+ 36.0 "r^2" 0.957 ' j
~ 80o
71
Elsevier JIM05733
A silicon sensor-based filtration immunoassay using biotin-mediated capture John D. Olson, Peter R. Panfili, Richard Armenta, Mary Beth Femmel, Holly Merrick, Jenny Gumperz, Marion Goltz and Robert F. Zuk Molecular Devices Corporation, 4700 Bohannon Drive, Menlo Park, CA 94025, U.S.A.
(Received 26 June 1990; accepted 10 July 1990)
A sensitive sandwich immunoassay for human chorionic gonadotropin (hCG) was developed with biotin-mediated filtration capture and silicon sensor detection. A high density of biotin on the membrane assured efficient capture of complexes containing streptavidin and analyte. Capture efficiency was not affected over a wide range of filtration flow rates or biotin concentrations. The assay utilized the pH sensing ability of the light addressable potentiometric sensor (LAPS) for the detection of urease-antibody conjugates. A LAPS reader was constructed which allowed the enzyme conjugate to be detected in - 1/tl volumes. Effects from variations in detection volume were studied. 10 pg of h C G could be detected in an assay time of 20 rain with four standard deviations separation from background. Comparison to a commercial R I A was made. Key words." Sensor; Biotin; Immunoassay;Streptavidin; Light addressable potentiometric sensor; Filtration
Introduction
Semiconductor sensing devices offer several intrinsic advantages for biochemical measurements, including adaptability for solid phase and particulate samples, potential for miniaturization and very high sensitivity, and multiplicity of measurement sites (Karube, 1987). Immunoassay is one
Correspondence to: J.D. Olson, Molecular Dj~vicesCorporation, 4700 Bohannon Drive, Menlo Park, CA 94025, U.S.A. Abbreviations: B-BSA, biotin-bovine serum albumin; CHEMFET, chemically sensitive field effect transistor; hCG, human chorionic gonadotropin; LAPS, light addressable potentiometric sensor; PBS, phosphate-buffered saline; RIA, radioimmunoassay; SMCC, succinimidyl 4-(N-maleimidomethyl)cyclohexane-l-carboxylate; SPDP, N-succinimidyl 3(2-pyridydithio)propionate; TNBS, 2,4,6-trinitrobenzene sulfonic acid.
area of analytical biochemistry where sensors can have immediate utility. Janata (1975) proposed a chemically sensitive field effect transistor ( C H E M F E T ) in which chemical specificity was endowed by immobilization of an antibody in the gate region of the transistor. A direct reading immunochemical C H E M F E T was not successful because the antibody binding site was located too far from the gate region to alter interfacial charge density (Janata and Blackburn, 1984). Demonstration of enzyme-based C H E M F E T S (Blackburn, 1987) was viewed as an advance in the development of a sensor-based immunoassay because of the signal amplification the enzyme offers. The most promising practical approach employs p H sensing with a silicon nitride insulator which monitors the change in proton concentration caused by en-
0022-1759/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)
72 zymes such as urease, glucose oxidase, penicillinase, and acetylcholinesterase. The utility of CHEMFETs is limited by the difficulty in reliably encapsulating micro-semiconductor devices which function in aqueous environments. A light addressable potentiometric sensor (LAPS) (Hafeman et al., 1988) is a simple semiconductor device that functions much like a CHEMFET. The LAPS device contains a monolithic sificon structure with a planar sensing surface with no microcircuitry that needs to be encapsulated. Sensor planarity facilitates formation of reproducible microvolumes. In an assay based on pH change caused by enzyme activity, decreasing the assay volume to concentrate enzyme increases sensitivity. Another salient feature of the LAPS is the capability for multiple potentiometric measurements. A single LAPS semiconductor device can detect multiple concurrent chemical reactions by monitoring the photocurrent induced through illumination of discrete sensing sites. Simultaneous determinations of several samples, controls, and calibrators improves assay throughput and accuracy. A solid-phase enzyme immunoassay was developed, incorporating microvolume detection with LAPS. The assay format entails formation of specific immune complexes in liquid phase followed by membrane filtration. Streptavidin-biotin (Wilchek and Bayer, 1988) mediates the capture of the immune complex on the membrane. Membrane-bound enzyme is precisely measured in a LAPS reader designed to accept membranes. Use of microporous membranes to capture immune complexes is compatible with LAPS. A variety of membrane based immunofiltration methods have rapid, efficient wash protocols and the capability for parallel processing of multiple samples (IJsselmuiden et al., 1989). The membrane interstitial volume can be used as microvolume reaction chambers to increase sensitivity in a pH sensing assay. We describe the filtration immunoassay using the LAPS and major parameters for optimizing the filtration capture step and the detection of enzyme conjugates in microvolumes. A quantitative immunoassay with picogram sensitivity in biological samples requiring a total assay time of 20 rain is described.
Materials and methods
Chemical and biochemical reagents Bovine serum albumin (BSA) (A-7888), urease type VII (U-0376), dithiothreitol (DTT) (D-0632), diaminobiotin-agarose (D-1395), 2,4,6-trinitrobenzene sulfonic acid (TNBS) (P-3402), 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (E-7750) and succinic anhydride (S-7626) were purchased from Sigma Chemical Company, St. Louis, MO. Streptavidin (S-1214) was purchased from Scripps Laboratories, San Diego, CA. lzsI-labeled streptavidin (IM 236) was purchased from Amersham, Arlington Heights, IL. Radioiodination kit (NEZ151) was purchased from New England Nuclear, Boston, MA. Urease labeled streptavidin (44000U) was purchased from JD Biologics, Downsview, Ont., Canada. Anti-human chorionic gonadotropin (anti-hCG) monoclonal antibodies were obtained from Dupont, Wilmington, DE. Human chorionic gonadotropin (hCG) was purchased from Immunosearch, Emeryville, CA. Tandem-R hCG immunoradiometric assay was purchased from Hybritech, San Diego, CA. Succinimidyl 6-(biotinamido)hexanoate (S-1582), succinimidyl 4(N-maleimidomethyl)cyclohexane-l-carboxylate (SMCC) (S-1534), and N-succinimidyl 3-(2-pyridyldithio)propionate (SPDP) (S-1531) were purchased from Molecular Probes, Eugene, OR. Dimethyl formamide (DMF) (50117) was purchased from Pierce Chemical Co., Rockford, IL. 2,2'-oxy(bisethylamine) (17,609-5) was purchased from Aldrich Chemical Company, Milwaukee, WI. Biogel A5M was purchased from Biorad, Hercules, CA. Sephadex G100 (17-0060-1) and Sephadex G-25 columns (PD-10) were purchased from Pharmacia, Piscataway, NJ. Preparation of anti-hCG-urease conjugates Urease was rehydrated in 50 mM (Na2H / NaH2)PO4, 150 mM NaC1 pH 7.0 (phosphate buffer) with 10 mM DTT, and incubated for 1 h to disaggregate the enzyme. The enzyme was separated from the DTT by using a Sephadex G-25 column and reacted with antibody within 1 h. Urease was conjugated to anti-hCG by using a modification of the method described by Ishikawa et al. (1983). Maleimide groups were introduced onto purified monoclonal antibodies specific for
73 the fl subunit of h C G by reaction with SMCC. Maleimidyl-antibody was incubated with urease for 3 h at 4°C. The conjugates were purified using a Biogel A5M column (2.5 x 90 cm) equilibrated in phosphate buffer plus 100 m M Na2SO3, 1 m M EDTA, 2 m M DTT, 0.05% N a N 3 (pH 7.0). Column fractions containing the conjugates were pooled and stored under N 2 at 4°C.
Preparation of anti-hCG-streptavidin conjugates Streptavidin was conjugated to anti-hCG by using a modification of the methods described by Carlsson et al. (1978) and by Ishikawa et al. (1983). Maleimide groups were introduced onto purified monoclonal antibodies specific for the ct subunit of h C G by reaction with SMCC. Sulfhydryl groups were introduced onto streptavidin by reaction with SPDP followed by DTT. The sulfhydryl-streptavidin was incubated with maleimidyl-antibody overnight at room temperature. Conjugate was purified using a diaminobiotina g a r o s e / S e p h a d e x G-100 column. 5 ml of diaminobiotin-agarose was packed on the top of Sephadex G-100 in a 2.5 x 30 cm column equilibrated in phosphate buffer. The conjugate reaction mixture was applied to the column and the column was washed thoroughly. To elute the bound streptavidin conjugates, 10 ml of 0.1 M sodium citrate buffer (pH 3.5) were added. Phosphate buffer was then used to chromatograph the antibody-streptavidin conjugates from free streptavidin. The fractions containing conjugate were pooled and stored in phosphate buffer at 4°C.
Preparation of biotin-BSA BSA at 40 m g / m l was dialyzed against 100 m M N a H C O s, 150 m M NaC1 p H 8.2 (bicarbonate buffer). A 200-fold molar excess of KzCO 3 followed by a 200-fold molar excess of succinic anhydride (40 m g / m l in D M F ) was added to the BSA. After 1 h at room temperature.the mixture was dialyzed against bicarbonate buffer. The succinyl-BSA was concentrated to 40 m g / m l by tangential flow ultrafiltration (Minitan, Millipore, Bedford, MA) and 2,2'-oxy(bisethylamine) was added to a final concentration of 0.1 M. While stirring, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide was added at a molar ratio to BSA of 220 : 1. After 1 h at room temperature the mixture
was exhaustively dialyzed against bicarbonate buffer. The aminated BSA was adjusted to 15 m g / m l . Succinimidyl 6-(biotinamido)hexanoate (40 m g / m l in D M F ) was added at a molar ratio of 20 : 1. After 1 h at room temperature the reaction mixture was dialyzed against phosphate buffer. The number of biotin groups on the BSA was estimated by the consumption of free amino groups by using the TNBS assay (Fields, 1972) and determined to be approximately 10 per BSA molecule.
Preparation of biotin-BSA coated nitrocellulose An immobilization solution containing biotinBSA (B-BSA) at 5 m g / m l in phosphate buffer was prepared. Nitrocellulose m e m b r a n e (BA-85, Schleicher and Schuell, Keene, N H ) was saturated with the solution and placed on the b o t t o m of a glass dish for 45 min. The m e m b r a n e was washed with phosphate buffer then crosslinked with 0.1% glutaraldehyde in phosphate buffer for 2 h at room temperature. The m e m b r a n e was washed twice with phosphate buffer for 15 min each followed by deionized H 2 0 for 15 min then dried in a convection oven at 65°C. To determine the amount of biotin-BSA coated onto the m e m b r a n e a trace amount of biotin-125I-BSA (radioiodinated using the NEN kit) was spiked into the B-BSA solution. Coating density was determined by testing 0.5 cm 2 samples of m e m b r a n e in a g a m m a counter (Beckman no. 5500). B-BSA density was approximately 6 0 / . t g / c m 2. For some experiments, nitrocellulose membrane was coated with varying amounts of B-BSA, maintaining total protein concentration of the coating solution at 5 m g / m l by the addition of native BSA.
Filtration capture assay The membrane was mounted with adhesive transfer tape (Y-9460, 3M Corp., Minneapolis, MN) over a window cut in a vinyl acetate stick (Transvy, Transilwrap West Corp., So. San Francisco, CA). Membrane sticks were placed into filter units which directed the flow of solution through the membrane. Filtration rate was controlled by a regulated vacuum unit (Threshold, Molecular Devices Corp., Menlo Park, CA). Streptavidin was diluted to 100 n g / m l in 10 m M ( N a 2 H / N a H 2 ) P O 4 , 150 m M NaC1 p H 7 (PBS)
74
with 0.1% BSA, 10 mM Na2SO3, 1 mM EDTA, 0.05% NAN3, 1 m M D T F (conjugate dilution buffer). 125I-Streptavidin was added to the streptavidin solution to a specific activity of 227 c p m / n g streptavidin. 91% of the cpm were precipitated by 10% trichloroacetic acid (TCA). The specific activity of active streptavidin therefore was taken to be 207 cpm/ng. The streptavidin solution (1 ml) was filtered (n = 3) through the membrane and then washed with 500 /L1 of PBS, 0.1% BSA, 0.05% Tween 20 (wash buffer) at a flow rate of 500 /~ml/min. The filtration area of the membrane was removed from the stick and analyzed on a Beckman 5500 gamma counter to estimate streptavidin capture.
I Stick VariableBiasVoltage
Enz)
hCG assayprotocol Fig. 1 depicts the sandwich immunoassay for hCG. hCG was diluted in pooled human serum collected from healthy male donors, and values assigned by repeated testing using the Hybritech RIA. Anti-hCG-urease conjugate was mixed with anti-hCG-streptavidin conjugate in conjugate dilution buffer, hCG serum calibrator or patient sample (10 #1) was added to 300 /~1 of conjugate solution and incubated for 15 rain at 37°C. The mixture was filtered through B-BSA membrane at a rate of 100 ~ l / m i n . The membrane was washed with 500/~1 wash buffer at a rate of approximately 500 #l/rain. The area of filtration was 0.14 cm 2.
1. Incubate
El--X Anli-I~ Urease
~,
EF-x hCG
2. Filter,Wash
>-x f" Urea
3. Read
/////////~//////////////////////////// Figure 1. Schematic of hCG immunoassay.
Anti~ Streptavidin
Biotinylated Membrane
/"~'~/
Fig. 2. Schematic of LAPS reader.
The wash step removed unbound conjugate and equilibrated the membrane at the appropriate pH and buffer concentration for subsequent potentiometric measurement of bound enzyme. The LAPS reader, illustrated in Fig. 2, had a vertical sensing surface with a p H sensitive silicon nitride insulator contacting the electrolyte, an enzyme substrate solution (100 mM urea in wash buffer). The membrane stick with bound conjugate was inserted into the reader. When placed in the LAPS reader, the filtration area aligned precisely with the illuminated area on the sensor. A plunger compressed the membrane against the sensor reducing the assay volume to as low as 1/zl. A dial indicator positioned on the plunger mechanism provided an indirect measurement of the plunger stop distance from the sensor surface. The dial indicator was calibrated to 1 t~m intervals. A threaded plunger and lock nut enabled adjustment of the plunger stop. With the membrane against the sensor, data collection was started, Urease-catalyzed hydrolysis of urea to ammonia and CO 2 induced a p H change. This altered the surface potential at the electrolyte-silicon nitride interface. The potential at the
75
interface was monitored through a transient photocurrent produced by illumination with a light emitting diode behind the silicon surface opposite the filtration area on the membrane. Modulating the illumination intensity creates an alternating photocurrent in the external circuit. The magnitude of the detected photocurrent depends on the sum of the applied bias voltage, the surface potential at the electrolyte-insulator interface and the reference electrode potential. Measurements of surface potential were made every second for 1 min. The results were collected and analyzed using an IBM PS2/30 computer with specially designed software. The rate of change of pH was determined by the slope of the plot of change in surface potential vs time in /~V/second. A change of 1 pH unit was approximately equivalent to 50,000 ~V.
Results
Biotin mediated separation step The hCG sandwich procedure yields a liquid phase immune complex containing streptavidin and urease. The separation step employed filtration capture of the immune complex to a biotinylated nitrocellulose membrane. Initial studies characterized the filtration capture of streptavidin as a function of B-BSA density and flow rate (Fig. 3). At a flow rate of 170 /~l/min the streptavidin capture was > 95% at all B-BSA densities tested (>_ 1.8 /~g/cm2). At the higher flow rate of 1000 /~l/min streptavidin binding was reduced below
40
20 0~ 0
5
10
B - B S A on M e m b r a n e
15
20
(p.g/cm 2)
Fig. 3. Effect of m e m b r a n e b i o t i n density a n d flow rate u p o n c a p t u r e of strcptavidin. The flow rate was adjusted to 170 / L l / m i n (D) or 1000 p l / m i n (O). E r r o r bars indicate + 1 s t a n d a r d deviation.
30OO
;~ 2000
0 ~ 0
100
± 200
Filtration Flow Rate
300
":~''" 400
(~tL/minute)
Fig. 4. Capture of h C G complexes is independent of flow rate. The standard h C G assay was run with 0 pg ([3) or 1000 pg ( , ) h C G (n = 4). The filtration rate was varied during the i m m u n e complex separation step. Error bars represent + 1 standard deviation.
90% only when the B-BSA density was 6 / ~ g / c m 2 or less. The BSA was labeled with biotin at a molar ratio of 10 per BSA. A similar dependence of filtration flow rate and biotin density was observed when the biotin substitution on BSA was varied. When the amount of BSA immobilized was held constant, streptavidin capture efficiency was reduced as the biotin labeling approached a molar ratio of 1 : 1 (data not shown). These results suggested a means of achieving reproducible filtration capture of streptavidin-containing immune complexes. Conditions favoring maximal capture were selected: slow filtration flow rates, high biotin substitution on BSA and B-BSA immobilization levels greater than 2 0 / ~ g / c m 2. Fig. 4 illustrates a LAPS filtration immunoassay for hCG using nitrocellulose membrane coated with B-BSA at a density of 6 0 / ~ g / c m 2. B-BSA with a molar ratio of 10:1 was used for the membrane coating. Immune complexes were formed by incubation of the urease and streptavidin anti-hCG conjugates with 1.0 ng hCG. The mixtures were then filtered onto the biotinylated membrane at different flow rates to test if the immune complex sterically hindered capture. To within experimental error the enzymatic signal was identical with filtration rates from 25 to 350 /~l/min indicating robust, reproducible capture of the hCG immune complexes.
Optimization of microoolume p H detection In the pH measurement of enzyme activity, consideration of the major elements contributing to the total buffer capacity of the system is neces-
76 400
700
•
"
,
.
,
.
,
600 300
=~ 500 400
~ 2o0
300
d- 200 D. 100
r~
0
I
I
I
I
20
40
60
80
0 50
10O
Conc. of Phosphate (mM)
I
I
I
100
150
200
Measured Distance (l.tm)
Fig. 5. Effect of b u f f e r c o n c e n t r a t i o n on assay signal. S a m p l e s were filtered at 100 # l / m i n ( n = 4). T h e m e m b r a n e was w a s h e d
Fig. 6. T h e effect of assay v o l u m e u p o n assay signal. T h e slope of the r e s p o n s e was d e t e r m i n e d for distances _< the thickness of
with p H 7.0 b u f f e r c o n t a i n i n g the indicated p h o s p h a t e con-
the m e m b r a n e ( - 150 um). E r r o r b a r s i n d i c a t e + 1 s t a n d a r d deviation.
centration. T h e L A P S r e a d e r was filled with the s a m e w a s h b u f f e r plus 100 m M urea. E r r o r bars indicate + 1 deviation.
standard
sary. The effect of phosphate buffer concentration in enzyme substrate used in the LAPS reader is shown in Fig. 5. Urease was bound to the membrane by mixing the urease and streptavidin antih C G conjugates with 0 or 200 pg h C G followed by filtration through the B-BSA membrane• The enzyme assay was carried out in substrate formulated with phosphate buffer ranging from 10 to 100 mM. For both 0 and 200 pg hCG, the observed enzymatic rate of change of potential diminished as the buffer concentration increased• For the standard h C G assay 10 m M phosphate was chosen as substrate buffer• Although higher signals were obtained at phosphate concentrations lower than 10 mM, signal variation due to dissolved CO 2 and membrane buffer effects were observed. Reducing the volume of the reaction chamber should increase sensiti{,ity. The closure of the plunger compresses the membrane against the LAPS sensing surface increasing the local concentration of enzyme and reducing the phosphate buffering capacity• The impact of modulating the reaction microvolume is shown in Fig. 6. The standard h C G assay was run with 0 and 200 pg hCG. The LAPS measurement of captured urease employed the plunger stop set 70-210 /~m from the sensor surface• The thickness of the hydrated nitrocellulose membrane before compression by the plunger was determined to be about 150 /~m. The estimated reaction volumes ranged from less than 1•1-2.4/~1 for plunger stop distances from 70 to 150/~m respectively. The slope of the response
curve as the plunger approached the silicon was used to determine that the enzyme rate increased 4 /~V/s//~m at 200 pg hCG.
Filtration immunoassay performance This filtration immunoassay combined the elements of liquid phase formation of an immune complex, the biotin mediated filtration capture, and the microvolume urease activity measurement by the LAPS reader• For the measurement of membrane-bound urease the LAPS reader had the plunger stop set at 100 /~m from the sensing surface. Urea substrate was formulated with 10 m M phosphate• Total assay time was less than 20 rain. Fig. 7 is a typical standard curve. Table 1 shows data from the assay of various levels of h C G spiked into pooled h u m a n serum. The limit of detection for the assay was below 5 m I U / m l (10 pg/assay), the lowest concentration tested• The 5 m I U / m l level was statistically different from the blank at the 99.8% confidence limit by Student's one-tailed t test. Coefficients of varia-
200
•
,
•
,
•
,
•
,
•
,
"
150 :=L
Z loo ¢¢
50 zo*0
0: 0
i 10
i
i
i
i
20
30
40
50
60
[hCG] (pg) Fig. 7. h C G s t a n d a r d curve. C a l i b r a t o r s were r u n in replicates of 4. E r r o r bars r e p r e s e n t + 1 s t a n d a r d deviation.
77 TABLE I SPIKE RECOVERY Precision and accuracy. Samples of hCG were added into pooled serum and assayed (n = 20). Values were calculated from a standard curve and % recovery and quantitation CV were determined. hCGadded (mIU/ml)
hCGrecovered (mIU/ml)
Recovery (%)
Quant. CV (%)
5 25 100 500
5.8 26 102 558
116 104 102 112
13 5 7 4
tion were 13% at the 5 m l U / m l detection limit and substantially less than 10% at hCG levels from 25 to 1000 m I U / m l . Potential for accurate quantitation by the LAPS system was established by a comparison with a commercial RIA. This filtration immunoassay quantitated hCG up to levels of 1000 m I U / m l . The highest calibrator in the RIA was 400 m I U / m l ; samples with greater than 400 m I U / m l were diluted 1/10 and rerun in the RIA. Fig. 8 illustrates the results of quantitating 31 sera. Linear regression analysis yielded a slope of 0.92 and a r 2 value of 0.957.
Discussion
We report a sensitive immunoassay integrating features of the LAPS reader and membrane filtration. The flat sensing surface of the monolithic silicon facilitates microvolume measurements of membrane-bound enzyme with high precision and 2
looo
y = 0'.92x'+ 36.0 "r^2" 0.957 ' j
~ 80o
