Subfemtomole enzyme immunoassay for human growth hormone using affinity chromatography and enzyme amplified detection

ANALYTICAL

BIOCHEMISTRY

189,

217-222

(1990)

Subfemtomole Enzyme Immunoassay for Human Growth Hormone Using Affinity Chromatography and Enzyme Amplified Detection R. Lejeune,* L. Thunus,*

F. Gomez,? F. Frankenne,?

J.-L. Cloux,$ and G. Hennent

*Service de Chimie analytique, Universite’ de Lb&e, Institut de Pharmacie, rue Fusch, 3, B-4000 LiGge, Belgium; TService d%ndocrinologie expbrimentale et clinique, Universite’ de Litige, Centre hospitalier universitaire B 23, Sart-Tilman, B-4000 Liege, Belgium; and SBiocode, Passage Lemmonier, 34, B-4000 LiGge, Belgium

Received

December

29, 1989

An immunometric assay is described which allows fast detection of attomole amounts of an antigen. The sensitivity is 100 to 1000 times better than that of classical sandwich immunometric assays. Our system allowed the measurement of human growth hormone in the range of 0.1 amol to 100 fmol in a 4-h time period overall. A chromatography column is sequentially filled with two immunoaffinity resins: SP-MI--E,-Ab, in the upper half and SP-M,--E,-Ab, in the lower half, where Ab, and Ab, represent complementary antibodies reacting with the antigen to be assayed, E, and E, represent enzymes, MI and M, represent substances reacting reversibly with E, and E,, respectively, and SP represents the chromatographic solid phase; the sign - represents covalent linkages and the sign -- reversible linkages. The sample solution is passed through the column, resulting in binding of the antigen to the first encountered antibody, yielding the immobilized complex SPMI--E,-Ab,--Ag. The M, bound is then destabilized by washing with solution of agonist to M,. The freed complex is immediately trapped by the second antibody in the lower part of the column, resulting in the entity SPM,--E,-Abz--Ag--Ab,-E, . After a washing step, an amplified detection allows the measurement of the antigen through the activity of the enzyme E, . The antigen-antibody reactions occur in the presence of a very large excess of antibody. The continuous equilibrium displacement due to the chromatographic procedure enhances the yield of complex formation. These factors explain the extremely low levels (subattomole) capable of being detected with this original technique. o 1990 Academic

Press,

Inc.

The sandwich assay, remarkable for its high sensitivity and good precision (l), necessitates prolonged incu0003-2697190 $3.00 Copyright 0 1990 by Academic Press, All right,s of reproduction in any form

bation and critical manipulations. Especially when assays are carried out in antibody-coated plastic tubes, slow protein leakage as well as irreproducible adsorption behavior is often encountered (2,3). An advantageous alternative is to perform immunological reactions on the solid phase of an affinity chromatography column (4), which results in significantly reduced incubation times and improved reproducibility. However, the advantage of the chromatographic process appears only at the immunoassay first step. Consequently, several additions of the labeled antibody are needed for reproducible binding (5). Competitive immunoassays have recently been improved by using a reversible capture of the specific antibody onto an electrode membrane which is used as a detector (6). This procedure allows the use of only one detector for assaying different antigens but does not improve the sensitivity and the speed of the assay. The method proposed here uses affinity chromatography and reversible capture of enzyme antibody conjugates in an original way to obtain a significant increase in noncompetitive immunoassay efficiency. The originality of the method is to preserve the efficiency of affinity chromatography during both steps of immunoassays. Affinity chromatography and reversible capture of antibodies being the main characteristic of the new method, we have tentatively named our method reversible antibodies capture immunoassay (RACIA).’ The measurement of human growth hormone concentration has been used as a model to check the performances of the RACIA in terms of speed, reproducibility, and sensitivity. ’ Abbreviations noassay.

used:

RACIA,

reversible

antibodies

captive

immu-

217 Inc. reserved.

218

LEJEUNE

+O-

AL.

order to get the highest sensitivity possible. During the elution step, the affinity column goes back to a native state and is again ready to receive antibody-enzyme conjugates directed at the same antigen or another antigen provided that specific antibodies are conjugated to identical enzymes.

.

HPO;

ET

MATERIALS

Ethanal

NADP+

)<~~

Acetaldehyde

Alc$+ogm .

MAD"

+

H'

6 Diaphorase Formazan

7

: Alkaline

]

: 6.0

phasphatase galactosldase

< <

INT

: Ant1 :

FIG. 1.

METHOD

-

Anti

CH w

Assay

I

antdmdy

antibody

1

: P,,ospho",c

: ATPG

Q:

"m&y Antigen

procedure.

DESCRIPTION

RACIA is described in Fig. 1. The affinity column is packed with two superposed gels. The first one is labeled with a phosphonic group (ligand of phosphatase) and the second one with aminophenyl thiogalactoside (P-D-galactosidase ligand). By passing through the column solutions containing complementary anti-human growth hormone antibodies conjugated to @-D-gdaCtOsidase or phosphatase, the solid phases become two immunoaffinity gels directed at different epitopes of human growth hormone. The sample is then passed through the column, which results in the binding of human growth hormone in the first part of the column containing the phosphatase antibody conjugate. Further addition of an excess of mineral phosphates displaces the enzymatically labeled antibody-antigen complex, which is immediately trapped in the second part of the column by the complementary anti-human growth hormone antibody. After a washing step, the whole complex, detected by uv monitoring at 280 nm, is eluted by an alkaline buffer. It is collected and the phosphatase activity measured using an amplification technique in

AND

METHODS

Materials. Mouse monoclonal antibodies were prepared and purified at the Laboratoire d’Endocrinologie experimentale et clinique (7). Mouse ascites fluids were taken and antibodies were precipitated with 50% ammonium sulfate. Proteins were chromatographed on a weak base ion exchange gel (diethylaminoethyl) equilibrated with ammonium hydrogen carbonate buffer (0.01 M, pH 7.5). Protein elution was carried out by a linear gradient of ammonium hydrogen carbonate (0.01 to 0.5 M). The immunoglobulin fraction was lyophilized. ‘251-labeled tracers were prepared using the methods of the Laboratoire d’Endocrinologie experimentale et clinique (8). All chemicals used were of analytical grade (or similar). Preparation of mouse antibody-enzyme conjugates. The conjugates were prepared as described by Yoshitake and co-workers (9). The first antibody was coupled with alkaline phosphatase and the second was coupled with fl-D-galactosidase in a high molar ratio (5:l antibody:enzyme). Preparation of affinity microcolumns (immunoreactar). Fractogel TSK Merck (HW 65/F) was activated by butanediol diglycidoxy ether in sodium hydroxyde as described by Dean and co-workers (10). Activated gel was reacted either with antibodies or with aminophenyl thiogalactoside or with histidine (11). Reactions were performed at room temperature and reaction kinetics were followed by measuring the uv absorption of the supernatant. The reaction was stopped when the activation level rose to l%o (antibody) or 100 pmol/g (ligands). The histidine-containing gel was activated by phosphonic groups as described by Landt and coworkers (12). Affinity gels were packed into 3 X 30 mm columns. Apparatus. The apparatus designed for the assay is described in Fig. 2. It comprises four buffer bottles connected with a low-pressure liquid chromatographic system consisting of a pump, an immunoreactor, an uv detector at 280 nm, and a fraction collector. Standards and samples. The standards were prepared using human growth hormone 22K isolated and purified at the Laboratoire d’Endocrinologie experimentale et clinique (13). The protein was dissolved directly in a Tris-acetic acid buffer, 0.05 M, pH 7.4, containing either 0.5% or 6% bovine serum albumin. Assay procedure. In a first step, the two affinity columns (phosphonic and aminophenyl thiogalactoside)

ENZYME

IMMUNOASSAY

VIA

AFFINITY

219

CHROMATOGRAPHY

rranfer Auffer

-

Wash suffer

_

FIG.

2.

Experimental

setup

for the two-step

were converted into an immunoreactor by washing with a Tris-acetic acid buffer, 0.05 M, pH 7.4, containing 0.5% bovine serum albumin and a mixed solution of flD-galactosidaseand phosphatase-antibody conjugates (SP-M,--E,-Ab, and SP-M,--E,-Ab,). Human growth hormone was introduced in the immunoreactor and bound to the first part of the column owing to its high affinity for the antibody-coated solid phase (SP-M,-E,-Ab,--Ag). Elution of human growth hormone-antibody-phosphatase complex (E,Ab,--Ag) was carried out by applying to the column a Tris-acetic acid solution, 0.05 M, pH 7.4, containing 0.5% bovine serum albumin and phosphate, 10 mM. The effluent was directed to the second part of the immunoreactor where the complex was strongly retained by the complementary anti-human growth hormone antibody. The column was exhaustively washed by a Tris-acetic acid, 0.05 M, pH 7.4, buffer to eliminate the bovine serum albumin. Thorough elution of the entire antigen-enzyme-antibody complex (SP-M,--E,-Ab2--Ag--Ab,-E,) was obtained by washing the immunoreactor with a diethanolamine buffer, 100 mM, pH 9.8, containing ethanol (4%), magnesium chloride (1 mM), zinc chloride (0.1 mM), and sodium azide (15 mM). The protein fraction (E,-Ab,--Ag-Ab,-E, + E,-Ab,) was monitored with an uv detector, collected, and reacted with an equal volume of a solution containing INT violet (1.1 mM), NADP (40 PM), magnesium chloride (1 mM), zinc chloride (0.1 mM), ethanol (4%), sodium azide (15 mM), alcohol dehydrogenase (40 IU/ml), diaphorase (4 IU/ml), and bovine serum albumin (1%). The absorption was measured kinetically at 492 nm (14).

affinity

chromatography

immunoassay.

first optimized: bound antibody concentration on the solid phase, antibody affinity, sample dilution, and flow rate. Figure 3 shows the extend of antigen binding as a function of antibody-antigen molar ratio. Two antibodies of different affinities were tested. Data were collected after injecting antigen onto columns filled with different antibodies at different concentrations. Although antigen binding normally depends on both antibody concentration and antibody affinity, it appeared that antigen binding rose to 95% in all cases when the antibody-antigen molar ratio exceeded 104. In further experiments, this value was multiplied by a factor of 100 to account for potential loss of activity during storage and repetitive use (0.2 mg conjugate/g gel).

L

bound 100

_----------

90

-

so

_

70

-

-1

-__-.__-____-______________

0

1

_ _____-___

2

I

4

5

loq

R

RESULTS

Preliminary studies. Several parameters potentially affecting antigen binding in an affinity column were

FIG. 3. Effect of the antibody-antigen ing. (-) Antibody affinity constant; body affinity constant; (0) 10’ liters

molar ratio on antigen bind(0) 10” liters. mol-‘. (- - -) Anti* mol-‘.

220 :

LEJEUNE

bound

100

I

90

----0

00

i

----___ 0

Q--Q--A-

70 60

1

50 . 40 30 20 . 10

. . 0

FIG.

1

4.

2

Dilution

4

I

5

Log

v (111)

effect on the antigen binding.

The binding levels of antigen are given as a function of dilution in Fig. 4. Small amounts of radiolabeled antigen were diluted in a wide range of buffer volumes and injected onto the affinity columns. By direct determination of the bound radioactivity, no significant decrease of the binding was detected. The flow-rate effect on the binding efficiency was studied in the range of 10 to 1000 pl/min. The 100 times factor increase in the flow rate elicited a 20% decrease in the signal intensity but its effect was compensated by a significant noise decrease. In a first set of experiments, we performed assays in which only the first step of the reaction scheme was carried out with affinity chromatography. The performances of these assays in terms of second antibody binding efficiency are summarized in Table 1. Two series of identical columns were used. On the first one, increasing amounts of radiolabeled hormones were injected, while on the second one, identical amounts of cold hormone were injected followed by a determined amount of radiolabeledcomplementary antibody. The boundradioactivity was measured. By comparing the recorded data, it could be assumed that binding efficiency fell to a few

TABLE

1

Efficiency of an Immunoassay Where Only the First Step Takes Advantage of an Affinity Chromatography Process Human growth hormone (pg)

Radiolabeled antibody (ng)

Efficiency (%)

900

6.5 6.5 6.5 6.5 0.65

96 86

300 100 50 50

n Calculated.

12 2 2

Signal-tonoise ratio

r

33

10 0.5

0.180 1.8

0.9989

ET AL.

percent for low amounts of antigen. It is worth noting that the noise level was high especially when the binding efficiency was low and that inefficiency did not result in a loss of linearity. Consequently, the usual criteria of evaluation appear inadequate. RACIA development, The application of RACIA gave the following results. By measuring the rate of formazan production, it was possible to assay the antigen in a wide concentration range (10 amol to 100 fmol). The detection limit was lower than 100 amol and the regression coefficient, calculated on a calibration curve between 100 amol and 10 fmol, was always better than 0.997 (Fig. 5). The standard deviation, as determined from 10 measurements of 1 fmol, did not exceed 10%. Matrix effects were estimated by using bovine serum albumin as an interfering protein. No modifications of the results were observed when the analyte solution was prepared in buffers containing 0.5 or 6% bovine serum albumin. Chemicals and colored compounds did not interfere with the assay except for known inhibitors of capture and revelation enzymes. The highest sensitivity was obtained with specific conditions for the revelation step. The protein fraction was collected and mixed with a solution containing NADP only. After a long incubation time (3 h), amplification enzymes and other reagents were added. In this way, 0.1 amol of human growth hormone was detected. This result was obtained without any increase in incubation times. Only the revelation step was prolonged. The data collected after a series of 10 measurements are given in Fig. 6. DISCUSSION

In affinity chromatography, complete retention of low amounts of proteins by immobilized ligands forming poorly stable complexes (coenzymes) is well known, provided the column has a sufficiently high number of plates and that the ligand is in large excess. Mathematically, the agonist binding by a sufficient excess of ligand can reach 100% irrespective of the agonist concentration or ligand affinity. Actually, the chromatographic process improves the reaction rate between macromolecules. This advantage led other authors to propose running immunoassays on chromatography columns (4,5). However, in these previous studies, only the first step of the noncompetitive immunoassay was improved by the use of a chromatographic procedure (ligand binding by a large excess of antibody solid phase), whereas the second step was always run with a lack of ‘stationary phase excess (the stationary phase level controlled by bound antigen). This results in the loss of the chromatography advantages at the second step of the assay. Since that second step has a limiting character, it hindered the efficiency of the whole assay. As shown here, the addition of a significant excess of the tracer, even in

ENZYME

IMMUNOASSAY

VIA

AFFINITY

221

CHROMATOGRAPHY

1.6.

T 0

100

200

FIG. 5. Calibration curve for standards 10 fmol; (- - -) 50 to 1000 amol.

300

in the 50 amol

400

500

to 10 fmol

range.

several additions, does not override the lack of binding of the labeled antibody. Furthermore, higher nonspecific binding was also noted in these experimental conditions. As a result, the low signal-to-noise ratio significantly lowers the sensitivity of those methods. In this work, we aimed to take advantage of the efficiency of the chromatographic process in the second step of a noncompetitive immunoassay as well as in the first one. If the revelation step of classical immunoassays is performed by reacting the immobilized antigen with an adequately labeled antibody, the revelation step of RACIA is carried out by reacting the antigen with the

I 12

3

4

5

6

7

8

910

tiulnber

or

assay

FIG. 6. Repetitive determination of 0.1 amol of growth hormone. Revelation time: 180 min with NADP only and 15 min after addition of enzymes and INT. (0) Blank value; (0) assay value

600

Revelation

700

800

time:

(-1

900

1000

amole

25 s; (- - -) 200 s. Concentration

range:

(-1

I to

tracer under an immobilized form. Because further reaction between the antigen and the capture solid phase is imperative, the binding of the tracer on the solid phase has to be reversible. The sequence of the RACIA steps then has to be inverted as compared with classical immunoassays. Indeed, in RACIA, the antigen is first reacted with the tracer reversibly bound to a solid phase through its conjugated enzyme. The very large conjugate excess leads to a very high yield in the antigen binding, By competition with a substrate in solution, the antigen-conjugate complex and the conjugate excess then pass to the second part of the column containing the immobilized capture antibody, which will then bind the antigen-conjugate complex. At this stage, as in the former step, the immobilized antibody is in large excess, providing a highly efficient binding. Although not an absolute requirement, the reversibility of the capture antibody binding has substantial advantages. The turnover of the column can be increased since the revelation step is performed outside the immunoreactor, which also results in an increased flexibility of the procedure. Indeed, the capture solid phase can be adjusted to any antigen since different capture antibodies can be bound onto the same ligand. RACIA has been tested by using human growth hormone as a model antigen. Although assaying very low levels of human growth hormone could be of physiological relevance, the main aim of this work was to demonstrate that the sensitivity and the speed of immunoassays could be improved to a large extent by applying a new experimental procedure to noncompetitive immunoassays. The feasibility study of RACIA has been lim-

222

LEJEUNE

ited here to its fundamental aspect since, at this current stage of development, RACIA requires a high conjugate consumption. Studies are now in progress to develop compatible with very low microimmunoreactors amounts of conjugates. These will allow us to validate the process using serum samples and to study such important factors as specificity. We have demonstrated that an original chromatographic process using an enzymatic tracer as capture reagents of the specific antibodies and a specific displacement of the antigen-conjugate complex enhances, to a very significant extent, the performances of immunoassays in term of sensitivity and reduction of the incubation time.

CONCLUSIONS We have described a new system for high-speed determination of very low quantities of proteins. Owing to an original affinity chromatography process, incubation times for the development of antigen-antibody reactions have been drastically reduced. Moreover, high or very high affinity antibodies are not necessary since the sensitivity of the RACIA is not a function of antibody affinity. It seems the sensitivity is essentially determined by the tracer-specific activity and by nonspecific binding. This nonspecific binding is partly prevented by the high flow rate permitted by the system. The design of the assay procedure can be easily adapted for automation.

ET

AL.

ACKNOWLEDGMENTS This work We gratefully uscript.

was supported acknowledge

by BIOCODE (B-4000 Liege, Dr. M. MacNamara for revising

Belgium). this man-

REFERENCES 1. Collins, W. P. (1986) Alternative ley, New York. 2. Chessum, B. S., and Denmark, 3. Cantarero, B&hem.

L. A., Butler,

J. R. (1978)

pp. 77-86,

Lance&

8054,

Wi161.

J. E., and Osborne,

J. W. (1980)

Anal.

J. D., and Wilson,

G. S. (1983)

Anal.

105, 375.

4. Sportsman, J. R., Liddil, Chem. 55,771. 5. De Alwis, 6. Boitieux, Biomed.

Immunoassays,

W. U., and Wilson, J. L., Biron, Anal. 5,821.

M.

G. S. (1985) P., and Thomas,

Anal.

Chem.

D. (1987)

57,2754. J. Pharm.

7. Frankenne, F., Closset, J., Gomez, F., Scippo, M. L., Smal, Hennen, G. (1988) J. Clin. Endocrinol. Metab. 66, 1171.

J., and

8. Gomez, F., Pirens, G., Schaus, C., Closset, J., and Hennen, G. (1984) J. Immunoassays 5, 145. 9. Jue, R., Lambert, J. M., Pierce, L. R., and Traut, R. R. (1978) Biochemistry 1'7, 25, 5399. 10. Yoshitake, S., Imagawa, M., Ishikawa, E., Niitsu, Y., Urushizaki, I., Nisiura, M., Kanazawa, R., Kurosaki, H., Tachibana, S., Nakazawa, N., and Ogawa, H. (1982) J. Biochem. 92,1413. 11. Dean, P. D. G., Johnson, W. S., and Middle, F. A. (1985) Affinity Chromatography-A Practical Approach, pp. 34, 35, IRL Press, Oxford. 12. Landt,

M., Boltz,

S. C., and Butler,

L. G. (1978)

Biochemistry

1'7,

5,915. 13. Smal, them.

J., Closset, J. 225,283.

J., Hennen,

G., and Demeyts,

14. Stanley, C. J., Ellis, D. H., Bates, D. L., (1987) J. Pharm. Biomed. Anal. 5,811.

P. (1985)

Bio-

and Johannsson,

A.