A rapid method for the purification and radioimmunoassay of human α-fetoprotein

317

Clinica Chimica Acta, 64 (1975) 317-323 0 Elsevier Scientific Publishing Company,

Amsterdam

- Printed

in The Netherlands

CGA 7435

A RAPID METHOD FOR THE PURIFICATION AND RADIOIMMUNOASSAY OF HUMAN a-FETOPROTEIN

P.I. FORRESTERa, R.L. HANCOCKa, and F.L. LORSCHEIDERC**

D.M. HAYb, P.C.W. LAIC

bObstetrics and Gynaecology, Divisions of aMedical Biochemistry, Physiology, Faculty of Medicine, University of Calgary, Calgary,

(Received

and CMedical Alberta, T2N lN4

(Canada)

May 21, 1975)

Summary Human a-fetoprotein (AFP) was isolated from cord serum on an immunoadsorbent column obtained by covalently linking rabbit anti AFP to cyanogen bromide activated Sepharose. Bound AFP was eluted with 8 M urea with better than 50% recovery. The purified AFP was iodinated prior to its use in a double antibody radioimmunoassay. The purification and radioimmunoassay employ commercially available materials. A standard inhibition curve was obtained which allowed determination of AFP levels between 50 and 1000 ng/ml in human serum. The assay was verified by measuring AFP levels in normal female serum, pregnancy serum, cord serum, hepatoma ascitic fluid and a standardized AFP solution.

Introduction It is well established that the occurrence of a-fetoprotein (AFP) in blood serum is associated both with ontogenesis and with primary liver cancer in several species including man [ 11. Several radioimmunoassays (RIA) have been described which permit the detection of human AFP in serum of patients with hepatoma and noncancerous hepatic diseases [ 2,3], in maternal pregnancy serum [4,5] and in maternal serum of certain pregnancies complicated by fetal compromise [6--81. A common methodological requirement of these assays has been the isolation and purification of AFP by a series of gel filtrations under several pH conditions both for the purposes of antiserum production and for radiolabelling. In the present paper we have employed affinity chromato* Request for reprints Faculty of Medicine,

should be addressed to Dr F.L. Lorscheider, Division of Medical Physiology, 2920 24 Avenue N.W., Calgary. Alberta, Canada, T2N lN4.

graphy to produce an immunoadsorbent which permits the isolation of large quantities of AFP of sufficient purity for radiolabelling. This procedure and the resultant RIA are based entirely on commercially available materials. We are presently using this RIA for antenatal screening to diagnose fetal compromise in a wide variety of obstetrical complications. Materials and methods (i) Preparation

of immunoadsorbent

for isolation of human AFP

Two grams of freeze dried cyanogen bromide activated Sepharose 4B (Sigma Chemical Co., St. Louis, MO.) was allowed to swell in 50 ml of 1 mM HCl. The activated Sepharose was then washed with another 200 ml of 1 mM HCl, filtered through a sintered glass Buchner funnel and suspended in 20 ml of a buffer containing 0.1 M NaHCO, and 0.5 M NaCl, pH 8.3. The gamma globulin fraction from 2 ml of the rabbit antihuman AFP antiserum (Behring Diagnostics, Montreal, P.Q.) was obtained by the addition of saturated ammonium sulphate to yield a 40% saturated solution. After 15 min at O”C, the precipitate was collected by centrifugation at 10 000 X g for 10 minutes, and was subsequently dissolved in 2 ml of the NaHCO,/NaCl buffer described above. The gamma globulin solution was added to the activated Sepharose suspension. The slurry was kept at 4°C for 24 h and was then washed with 10 volumes of phosphosaline buffer (0.05 M sodium phosphate/O.5 M NaCl, pH 7.6). The immunoadsorbent was then placed in a small column (4.0 cm X 0.5 cm) and washed with 20 ml of the phosphosahne buffer containing 8 M urea. This step was carried out to remove any non-covalently attached gamma globulin. The column was equilibrated with 10 column volumes of the phosphosaline buffer to remove urea prior to use. (ii) Isolation

of human AFP from cord serum

Two ml of cord serum were passed through the immunoadsorbent column. Non-adsorbing proteins were eluted with 10 column volumes of phosphosaline buffer. After all the non-adsorbing proteins had been eluted (A280nm equal to zero), AFP was eluted with the phosphosaline buffer containing 8 M urea and 0.2 ml fractions were collected. The presence of AFP in these fractions was monitored by the Mancini radial immunodiffusion assay [9] using M-Partigen plates (Behring Diagnostics). After reconditioning the column with 10 column volumes of the phosphosaline buffer, it could be reused (at least 10 times) to isolate more AFP from a fresh sample of cord serum. (iii) Iodination

of AFP

a-Fetoprotein was iodinated by the method of Greenwood Four to six pug of AFP in 0.15 ml of 0.05 M sodium phosphate reacted for 1 min with 2 mCi of ’ 2‘1 (New England Nuclear Dorval, P.Q.; specific activity 17 Ci/mg iodine) and 200 pg After termination of the reaction by the addition of 240 pg of bisulphite in 0.1 ml of the phosphate buffer, 2 mg of unlabelled 0.2 ml of the phosphate buffer was added. The labelled AFP

et al. [lo]. (pH 7.6) was Corporation, chloramine-T. sodium metacarrier KI in was separated

319

01

0

I

I

100

300

200

400

I

500

ng AFP

Fig. 1. Standard curve for RIA of human AFP. Each Point represents the mean percent radioactivity found in the precipitate, corrected for non-specific binding. The range of values for three replicate assays is indicated by the bars.

from the unreacted ’ 251 by chromatograp h y on a G-50 Sephadex column (0.8 cm X 20 cm). The most active fraction from the first peak was further purified prior to use on a Sephadex G-100 superfine column by the method of Beamer et al. [ll]. (iv) Radioimmunoassay for AFP The ’ 2 ’ I-labelled AFP was diluted with a phosphate/bovine serum albumin (BSA) buffer (0.05 M sodium phosphate and 0.5% BSA, Nutritional Biochemicals, Cleveland, Ohio, pH 7.6), so that 0.1 ml aliquots contained approximately 10 000-12 000 cpm 12SI-labelled AFP. Each assay tube (10 mm X 75 mm) contained the following in order of addition: 0.2 ml of phosphate/ BSA buffer, 0.1 ml lz5 I-labelled AFP in phosphate/BSA buffer, 0.1 ml unlabelled standard AFP (Behring Diagnostics cy, -fetoprotein standard, concentration 280 pg/ml) diluted in pooled normal male serum or 0.1 ml unknown serum, 0.1 ml of a 1 : 800 dilution of rabbit antihuman AFP antiserum in phosphate/BSA buffer. After incubation at 4°C for 24 h, 0.1 ml of a 1 : 5 dilution of goat antirabbit gamma globulin (Calbiochem., San Diego, Calif.) in 1 : 20 normal rabbit serum (0.05 M phosphate buffer, pH 7.6) was added and incubation was continued for a further 24 h. The assay tubes were centrifuged at 5000 X g for 30 min and the supernatants removed by aspiration. The radioactivity of the precipitated ’ 25 I-labelled AFP was determined, the percent radioactivity bound to the primary antiserum was calculated and the amount of AFP was determined from the standard curve (Fig. 1) and expressed as ng/ml. Results (i) Isolation ,of AFP The elution profile for the isolation of AFP on the immunoadsorbent column is shown in Fig. 2. 275 pg AFP could be isolated from two ml of cord serum. This represents approximately 53% recovery of purified AFP.

t 4

5

0

IO

20 FRACTION

Fig.

2.

amount

25

30

35

NUMBER

Elution profile of AFP on an immunoadsorbe~t of AFP in 5 &l of each fraction was determined

column. Fractions of 250 1.r1were collected. The by RIA. The arrow indicates the point of elution

with 8 M urea.

One minute was found to be the optimal interval for the radiolabelling with chloramine-T. The most active fraction from the first peak off the G-50 column was further purified on a Sephadex G-100 column. The elution profile from this latter column showed a single symmetrical peak, the top 75% of which was pooled and used in the RIA. The resultant yield of iodinated AFP was found to be sufficient for 5000-10 000 assays. Disc gel electrophoresis of labelled and unlabelled purified AFP is shown in Fig. 3. It can be seen that the unlabelled AFP is approximately 80% pure. It can also be seen that approximately 66% of the radioactivity is bound to AFP. Only the major protein band had immunolo~c~ activity.

dislonce ni~,raled

!mmJ

Fig. 3. Polyacrylamide gel electrophoresis of labelled and unlabelled human AFP. Electrophoresis was carried out at pH 9.3 in 7.5% aerylamide. After electrophoresis the radioactivity was determined in 2 mm slices. The slices were eluted with 0.05 M phosphate buffer (PH 7.5) and the AFP precipitated with rabbit antihuman AFP as described in tbe Methods section for the RIA. The shaded areas above the peaks represent unlabelled AFP run under similar conditions and stained with Buffalo Blue Black stain.

321

(iii) Radioimmunoassay for AFP The following conditions were found to be optimal for the RIA: 1 : 800 dilution of rabbit antihuman AFP; 24 h incubation in the primary antiserum; the use of a 1 : 5 goat antirabbit gamma globulin (diluted in 1 : 20 normal rabbit serum in 0.05 M phosphate buffer, pH 7.6); 24 h incubation in the secondary antiserum. The standard curve was obtained using a standard AFP solution diluted in pooled normal male serum. Normal male serum was used as a diluent to ensure a constant and minimal nonspecific binding of 2.3%. (iv) Sensitivity of the assay The limit of sensitivity of the assay (defined as twice the standard deviation of the determination of zero) was 5.5 ng on the standard curve which is equivalent to 55 ng/ml serum. The limit of sensitivity could be lowered by using a higher dilution of primary antiserum, e.g. l/5000, if it is found necessary to work in the range O-50 ng/ml of serum. However, when a dilution of l/800 is used, the working range is equivalent to 50-1000 ng/ml which corresponds to the range found in normal pregnancy. (v) Precision of the assay The average coefficient of variation for 5 different serum samples, assayed in replicate (n = 10) on the same standard curve is 6.0% (range is 4.6-7.8). The average coefficient of variation for 5 different serum samples, assayed in replicate (n = 10) on three different standard curves is 7.2%. Another method of deriving an index of precision is by the measurement of differences between a series of duplicate determinations [ 121. By this latter method the average difference for 100 duplicate determinations in the working range of the assay (50-1000 ng/ml serum) is 15.6 ng/ml with a coefficient of variation of 3.2%. (vi) Specificity of the assay Cross reactivity between AFP and human albumin was undetectable at several concentrations ranging between 0.25 and 1.0 mg albumin per assay

TABLE

I

COMPARISON AND

OF

AFP

REFERENCE

LEVELS

AS

MEASURED

BY

Fluid

IN

Concentration

Normal

female

serum

Maternal

pregnancy

Newborn

cord

Ascitic Purified

fluid

pital,

000

obtained fluid Quebec,

Purified

patient**

m/ml)* at 37

‘provided

weeks by

Sapporo.

AFP

267

? 108

262

000

245

000

1700

* *

and stated

lyophilized

of Medicine,

(n = 50)

(pooled)

bepatoma

(1 000

Ascitic

SEVERAL

PHYSIOLOGICAL

FLUIDS

of AFP

(w/ml)

<20

(pooled) serum*

serum

from

AFP

* Serum ** ***

RIA

STANDARDS

000

gestation. Dr

Phil

to contain reference

Japan.

Gold, 230

McGill 000

standard

University

Medical

Clinic,

Montreal

General

Hos-

m/ml. provided

by Dr S. Nishi.

Hokkaido

University.

School

322

tube (which represents 2.5 to 10 mg albumin per ml serum). Only solutions known to contain AFP produced significant competitive inhibition on the standard curve. A comparison of AFP levels in several physiological fluids and reference standards using the RI,4 described herein, is made in Table I. Samples having AFP levels greater than 2000 ng/ml were diluted prior to assay. Discussion The method of isolating AFP described in this paper uses commercially available antiserum and has the advantage of providing high yields (50%) of purified AFP suitable for radiolabelling and antisera production. The methods for isolating and radiolabelling AFP described above yield a product which gives an initial percent binding on the standard curve of 35%. This initial percent binding can be increased to approximately 70-75% by a further purification step such as polyacrylamide gel electrophoresis (see Fig. 3). Isolation of AFP using the method of Page on con A Sepharose [13] yielded preparations of AFP which were only bound 10% by the antiserum as determined by RIA. Attempts to isolate AFP by the immunoprecipitation technique of Nishi [ 141 also failed due to the fact that in our system the antigenantibody complex could not be wholly dissociated at pH 1.8. Attempts were then made using an immunoadsorbent column as described by Nishi and Hirai [15]. This column bound AFP which could not be eluted at either pH 2.8 or 1.8. However, bound AFP could be eluted using 8 M urea as described by Pihko et al. [16]. It has been shown [17] that the type of antiserum employed is the major determinant as to whether the antigen-antibody complex can be dissociated at low pH. Thus the complex formed with a high titre antiserum produced over a relatively long time period is incapable of dissociation at low pH, whereas the complex formed by antiserum produced over a shorter time period can be more readily dissociated at low pH. Other recent papers described RIA techniques for quantitation of AFP which circumvent the necessity of having highly purified AFP [ 181. However, these methods require more elaborate techniques and are therefore not as simple as the more traditional RIA protocol reported in this paper. Normal male serum was used to dilute the standard AFP solutions. If instead these solutions were diluted in buffer the non-specific binding varied with the amount of AFP added. When the standards were diluted in normal male serum the non-specific binding was constant at about 2.3%. The repeatability of the assay is approximately 6-7% and changes in AFP concentrations as low as 16 ng/ml can readily be detected within the working range of 50-1000 ng/ml. To test the efficacy of this assay, AFP levels were determined in several physiological fluids as well as with a standardized ascitic fluid and with authentic and purified AFP. Table I shows that our assay is in reasonably close agreement with results reported elsewhere for pregnancy serum [4,5] and cord serum [4]. Our determinations of AFP in the reference ascitic fluid agree to within 6% of the value stated by Dr P. Gold (personal communication). The apparent discrepancy between our assay values and the lyophilized pure AFP

323

provided by Dr S. Nishi can be explained if his recent samples are more highly purified than those originally used by Behring Diagnostics in the calibration of their own reference standards. There is clearly a need for all investigators to utilize a single reference standard for the purposes of comparing their results. Acknowledgements This investigation was supported in part by the Medical Research Council of Canada, Grant No. MA 5292 and the World Health Organization Grant No. R/OO~87. The authors wish to thank the Foothills Hospital Research and Development Committee for the support of laboratory personnel costs associated with a portion of this study. The authors are indebted to Dr P. Gold for providing a standardized sample of human hepatoma ascitic fluid and to Dr S. Nishi for supplying a purified reference sample of human AFP. One of the authors (P.I.F.) is a recipient of a post-doctoral fellowship from the Lalor Foundation. References 1 Abelev, G.I. (1971) Adv. Cancer Res. 14.295-358 2 Ishiguro, T., Mataukura, S. and Muranaka, H. (1971) Jap. J. CIin. Pathol. 19,157-158 3 Silver, H.K.B., Gold, P. Fe&r, S., Freedman, S. and Shuster. J. 11973) Proe. NatI. Acad. Sci. U.S.A. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

70.526-530 Se~~ala, M. and Ruoslahti, E. (1972) Am. J. Obstet. Gynecol. 112, 208-212 Ishiguro, T. and Nishimura, T. (1973) Am. J. Obstet. Gynecol. 116,27-33 SeppaIa, M. and Ruoslahti, E. f1373) Am. J. Obstet. GynecoI. 11548-52 WaId, NJ., Brock, D.J.H. and Bonnsr, J. (1974) Lancet i, 765-767 Brock, D.J.H., Bolton, A.E. and Scrimgeour, J.B. (1974) Lancet i, 767-769 Mancini, G., Carbonara, A.O. and Heremans, J.F. (1965) lmmunochem~~ 2, 235-254 Greenwood, F.C., Hunter, W.M. and Glover. J.S. (1963) Biochem. .I. 89, 114-123 Beamer, W.G., Murr, SM. and Geschwind, 1.1. (1972) Endocrinology 90, 823-827 Snedecor, G. (1952) Biometrics 8, 85-86 Pa&, M. (1973) Can. J. Biochem. 51,1213-1215 Nishi, S. (1970) Cancer Res. 30, 2507-2513 Nishi, S. and Hirai. H. (1972) Biochim. Biophys. Acta 278, 293-298 Pihko, H., Lindgren, J. and Ruoslahti, E. (1973) Immunochemistry 10, 381-385 Ruoslahti, E., Pihko. H. and Seppaia, M. (1974) Transplant. Rev. 20,38+60 Johansson, S.G.O.. Sherman. M.S., Hellsing, K. and KjessIer, B. (1974) Lancet ii, 839-341