Antibodies to lactalbumin interfere with its radioimmunoassay in human plasma

149

Clinica Chimica Acta, 87 (1978) 149-157 @ Elsevier/North-Holland Biomedical Press

CCA 9489

ANTIBODIES TO LACTALBUMIN INTERFERE RADIOIMMUNOASSAY IN HUMAN PLASMA

U. STEVENS

a, D.J.R.

LAURENCE

WITH ITS

a.* and M.G. ORMEROD

b

a Unit of Human Cancer Biology, London Branch, Ludwig Institute for Cancer Research, Royal Marsden Hospital and b Institute of Cancer Research, Royal Cancer Hospital, Haddow Laboratories, Clifton Avenue, Sutton, Surrey, SM2 5PX (U.K.) (Received

February

lOth,

1978)

Summary Two radioimmunoassays for human lactalbumin have been established using a rabbit antiserum. One assay uses a second antibody to separate bound from free label; the other uses polyethylene glycol to precipitate gamma globulin non-specifically. We have confirmed that about half the normal human population have a substance in their blood which inhibits the binding of lactalbumin to the rabbit antibody. Comparison of the two assays has demonstrated that this material is not lactalbumin but a naturally occurring antibody. We have shown that it is in the IgG fraction of human plasma and is probably a crossreacting antibody to bovine lactalbumin. None out of fifteen males and fourteen out of fifty eight non-pregnant, nonlactating females had low levels of lactalbumin in their blood (0.6-2.0 ng/ml). Our assay could not detect a statistically significant difference between normal women and those with either benign breast disease or metastatic mammary carcinoma.

Introduction Lactalbumin, the B protein of lactose synthetase, is formed specifically by mammary epithelial cells, and is a major protein of human milk whey. Sensitive radioimmunoassays (RIA) for this protein have been established with the object of investigating its production in breast cancer [l-3]. All the assays used a rabbit antiserum as the primary reagent and separated free from bound lactalbumin with a second antiserum to rabbit gamma globulins. It was reported that some apparently healthy males had detectable levels of lactalbumin (>l ng/ml) in their serum [ 1,3]. Using a similar RIA, we have confirmed that * To whom correspondence should be addressed.

150

the blood of some normal males contains a substance which inhibits the binding of lactalbumin to the primary rabbit antibody. Further investigation has shown that this material is not lactalbumin but a substance that binds lactalbumin and is probably a cross-reacting antibody to bovine lactalbumin. We have studied the properties of this antibody and have shown that it is contained in the IgG fraction of human plasma. In a double antibody RIA the rabbit antibody is precipitated specifically but the presence of a naturally occurring human antibody reduced the amount of labelled lactalbumin available for binding and hence the amount of label precipitated. If bound label is separated from free label by non-specific precipitation of gamma globulins, then the human antibody will be precipitated along with the rabbit and the total amount of label bound will be increased. We have established such an alternative RIA using polyethylene glycol [4] as the nonspecific precipitant. This paper reports a comparison of the effect of plasmas on the two RIAs which shows that about half the normal human population have antibodies to lactalbumin. We have studied the properties of this antibody and have shown that it is contained in the IgG fraction of human plasma. Materials and Methods Human lactalbumin was purified from human milk by the method of Phillips and Jenness [ 51. Bovine lactalbumin was purchased from the Sigma Chemical co. The anti-human lactalbumin serum was raised in rabbits by Sera-Lab Limited. 5.0 mg antigen in complete Freund’s adjuvant was injected intradermally at multiple sites. Six weeks later a booster dose of 1.0 mg lactalbumin was injected. The blood used for these experiments was taken eleven weeks after the booster injection. The antiserum was used in the RIA at a dilution of 1 in 25 000. Lactalbumins were labelled with lz5 I using Sepharose-coupled lactoperoxidase and hydrogen peroxide [ 61. Antibody, labelled human lactalbumin and dilutions either of known standards or unknown plasmas were incubated overnight at 4°C. For the double antibody RIA, the bound fraction was separated by the addition of donkey anti-rabbit gamma globulin serum (Wellcome, Reagents, Beckenham) with a further incubation at 4°C for 24 h, followed by centrifugation and aspiration of the supernatant. In the alternative RIA, a solution of ice-cold polyethylene glycol was added to give final concentration of 10% (w/v) and the resulting precipitate collected by centrifugation. Typical standard curves for the two assays are shown in Figs. 1 and 2. The limit of detection of lactalbumin was 100 pg per tube or 500 pg/ml plasma in either assay procedure. Bovine lactalbumin gave no inhibition at the dose range illustrated in Figs. 1 and 2. When comparing inhibition curves of purified lactalbumin with samples of plasma we plotted log(b,/b - 1) against log(lactalbumin concentration) [7] where b is the fraction of labelled bound and b,, is the fraction bound with no lactalbumin added. This type of plot is often close to linearity [8] and it depicts effectively non-parallel curves of inhibition [9]. The curves are shown in Figs. 1 and 2.

151

10 (b,,/b) I

1.0 _

0.1 c

0.01 1

0.01 i 0.1

1.0 nglml

10 lactalbumin

100

0.1

I 1.0 ngjml

10 lactalbumin

100

Fig. 1. Double-antibody RIA for human lactalbumin. Comparison of inhibition curves for lactalbumin purified from milk and plasma lactalbumin. 0, purified lactalbumin; A, plasma C.L. (360 mg/ml lactalbumin) (pregnant woman 30 weeks); 0, plasma M.R. (240 mg/ml lactalbumin) (pregnant woman 31 weeks). Fig. 2. As for Fig. 1 using the polyethylene glycol RIA. 0, purified lactalbumin; A, plasma C.B. (200 mg/ ml lactalbumin) (pregnant woman 29 weeks): 0, plasma A.D. (240 mg/ml lactalbumin) (pregnant woman 34 weeks).

Human antibodies to bovine lactalbumin were purified from 50 ml human plasma by the addition of ammonium sulphate to 45% saturation at 0°C. The resulting precipitate was collected by centrifugation, washed in a 45% saturated solution of ammonium sulphate, and dissolved in distilled water. After dialysis against 0.02 M phosphate buffer, pH 7.0, the proteins were applied to a 24 X 300 mm column of DEAE-cellulose (DE-52, Whatman) pre-equilibrated in the same buffer. The antibodies did not bind to the column and were collected in the initial break-through fraction. Bound proteins were eluted with 0.15 M NaCl, 0.02 M phosphate, pH 7.0 followed by buffer containing 0.3 M NaCl. Fractions were then applied to a column of Sephadex G-200 (24 X 850 mm) and protein eluted with 0.1 M Tris/HCl, 1.0 M NaCl pH 8. Fractions were tested for IgG, IgA and IgM by the Mancini method with plates purchased from Hoechst Pharmaceuticals. Samples of plasma were collected using EDTA as an anti-coagulant. The normal group were personnel within the laboratory. Preoperative samples were obtained from 24 female patients with histologically proved benign breast disease. All patients with cancer of the breast were women attending the Royal Marsden Hospital, Sutton, who had overt metastatic disease. Plasma samples were obtained from women between 30 and 35 weeks of pregnancy attending the antenatal clinic at Epsom District Hospital. The nursing mothers were wives of laboratory personnel. Results We have screened about 100 samples of plasma, of which only those from nursing mothers or pregnant women were found to contain high (>50 ng/ml)

152

levels of lactalbumin. Inhibition curves were constructed for four of the plasmas from women who were over 30 weeks pregnant; the curves were parallel to the curve given by a purified lactalbumin standard in both the double antibody and the PEG assay (Figs. 1 and 2, respectively). There is no reason to doubt that these plasmas genuinely contain lactalbumin. For the other samples of plasma we recorded the following parameters: A, the amount of label precipitated by polyethylene glycol in the absence of rabbit antibody; B, the amount precipitated by polyethylene glycol in the presence of rabbit antibody; C, the label precipitated in the double antibody RIA. We found that the 15 normal males could be divided into two groups (Fig. 3). In one group, the plasma had little inhibiting effect in either RIA; expressed as b/b, where b is fraction bound and b,, is the fraction bound in absence of sample or standard, the values for A, B and C were typically 0.2, 1.0 and 1.0, respectively. In the second group of plasmas, A and B were elevated (B often greater than 1.0) and C was depressed (between 0.3 and 1.0). We interpret this result as the demonstration of a binding protein for lactalbumin in the plasma. Using the double antibody assay alone it might have been concluded that these plasmas contained lactalbumin. The plasmas from the normal females were divided into five groups (four of which are shown in Fig. 4). Five of the plasmas were negative (A = 0.2, B and C Y 1.0) like the first group of males while seven contained binding protein (A > 0.2, B > 1.0, C < 1.0) like the second group of males. Three of the plasmas from non-pregnant non-lactating females contained lactalbumin (estimated to be 0.7, 0.8 and 1.3 ng/ml); in those plasmas A = 0.2, B and C < 1.0. The fourth group in Fig. 4 are six out of 22 plasmas from pregnant women

1.0

%

i

01

1

A

B

C

A

B

C

Fig. 3. RIA data using plasmas for normal males. A: label precipitated by PEG in absence of rabbit antiserum. B: label precipitated by PEG in presence of rabbit antiserum. C: label precipitated by double antibody. The label precipitated is shown as a fraction of the label precipitated in the presence of rabbit antiserum and in the absence of a plasma sample. The plasmas have been divided into two grouts on the basis of the values of A, B. and C obtained.

153

%,

1

A Fig.

4.

RI.4

women

I

C

A

using

plasmas

6 data

in the

figure

(on

C

6 from

the right

A

6

apparently

hand

side)

healthy

were

all over

A

c females. 30

weeks

8 Details

C as in Fig.

3. The

fourth

group

of

pregnant.

who were between one and six weeks from term. In this group A was elevated (>0.2) while both B and C were depressed (50 ng/ml). The data for healthy people together with those for women with benign and malignant metastatic diseases of the breast are summarised in Table I. The differences between the three groups of women were not statistically significant @ > 0.25 by Chi-square test with Yates’ correction). Since it is known that humans frequently have antibodies to the proteins of bovine milk [lo], we labelled bovine lactalbumin and observed its reaction with human plasmas. Labelled lactalbumin (either human or bovine) was incubated with samples of human plasma at 4°C overnight. The gamma globulins were then precipitated by 10% PEG solution and the radioactivity in the precipitate estimated. As is shown in Fig. 5 those plasmas which bound human

TABLE THE

I PERCENTAGE

BODY

TO

Sex

OF

PEOPLE

LACTALBUMIN Disease

WITH

IN THEIR status

DETECTABLE

No.

Percentage La&albumin

Neither 47

53 47

33

8

25

67

19

16

37

41

5L

14

35

52

Benign

breast

24

Female

Breast

cancer

Female

All

healthy,

Antibody

0

Female

= apparently

WITH

with

20

Normal

* Normal

THOSE

15

Normal

FfXlXlle

combined

AND

15

Male

*

LACTALBUMIN

BLOOD

non-pregnant,

non-lactating.

ANTI-

154

% bound

l

human lactalb.

L

IO

10 bovine

lactalbumin

Fig. 5. The percentage of labelled lactalbumin human lactalbumin. The bound was separated ground’ level of binding was 12%.

% bound

bound by different human plasmas comparing bovine to from the free label by precipitation with PEG. The ‘back-

lactalbumin at above the control level of 12% also bound the bovine lactalbumin above this level and vice versa. For further study we prepared plasmas from some out-dated titrated human blood from the blood bank. The selection of one plasma which contained antibodies to lactalbumin enabled us to work with 200 ml titrated plasma. A radioimmunoassay was then established using undiluted human plasma as the source of antibody and either labelled human or labelled bovine lactalbumin. Inhibition curves using unlabelled protein from the two species (Fig. 6) showed that bovine lactalbumin could displace both bovine and human label from the antibody but human lactalbumin only displaced human label, it competed ineffectively for binding sites when bovine lactalbumin label was used. Egg white lyso-

% inhib.

10

ng/ml

100 lactalbumin

1000

Fig. 6. Inhibition curves for PEG RIA using human plasma as a source of antibody and either human or bovine labelled lactalbumin. Inhibitors were either unlabelled human or bovine lactalbumin. 0, labelled human. human inhibitor; n, labelled human, bovine inhibitor; 0, labelled bovine, human inhibitor; 0, labelled bovine. bovine inhibitor.

155

3.10.1 I.02 M Q A 280 nr

0.02M

@

O.lSM

NaCl

pti 7.0

0.02M

@

0.30M

N&l

pli

7.0

”

B

A

C

DE

F

24 1 _

1.1 0.

150

100

50 fraction

no.

Fig. ‘7. Proteins eluted from DEAE cellulose column. Proteins precipitated by 45% saturated ammonium sulphate were applied to a DE-52 column in 0.02 M phosphate, pH 7.0. Proteins in fraction A did not bind to the column. Fractions B and C were eluted with 0.02 M phosphate, 0.15 M NaCl. pH 7.0. A contained IgG and the antibody to bovine lactalbumin. B contained IgG and IgA. C contained I$G. IgA and a substance which cross-reacted with lactalbumin in the RIA.

zyme which has sequences of amino acids in common .with lactalbumin [ll] did not inhibit the binding of bovine lactalbumin to the human antibody. The antibody had the same chromatographic properties as IgG. It precipitated in an ammonium sulphate solution at 45% saturation. On a DEAE column it eluted with the electrophoretically slow fraction of IgG (fraction A

I

n

L 40

4

50 fraction

60 no.

Fig. 8. Fraction C from DE-52 column applied to Sephadex G-200 column. Arrows show the positions of peaks from blue dextran. IgG and albumin, respectively. Fractions in areas I. II and III were bulked and tested in the RIA. I and III contained the unidentified material which cross-reacted with la&albumin in the RIA.

156

in Fig. 7) and on Sephadex G-200 co-eluted with IgG as identified by immunodiffusion. As a by-product we also discovered a substance which was an inhibitor in the PEG-RIA with either rabbit antibody and human label or with human antibody and bovine label. It was precipitated by 45% saturated ammonium sulphate solution but separated from the antibody on the DEAE column (fraction C in Fig. 7). On a Sephadex G-200 column the activity split into two; about half eluted in the void volume and about half eluted between IgG and albumin (that is, it had a molecular weight of about 100 000) (see Fig. 8). This material was not unique to the individual plasma chosen for study; it was also found in other plasma samples. Discussion The observation that some healthy males apparently had lactalbumin, a typical milk protein, in their blood was unexpected [ 1,3]. The contradiction has been resolved by our observation that, in a RIA using a non-specific precipitant, the fraction bound was increased rather than decreased by these plasmas. The simple explanation of this result is that the plasmas contain a protein which reacts with human lactalbumin and which is precipitated by PEG. Indeed we have shown that there is an antibody which reacts with both human and bovine lactalbumin in these plasmas. The difference between inhibition of binding in a double antibody RIA due to antigen and that due to antibody might be detected by assaying serial dilutions of the unknown sample and demonstrating non-parallelism with the standard curve. In practice the level of inhibition is often too low to permit an accurate comparison to be made. Also an assay is designed to give maximum sensitivity and these conditions are not necessarily the optimum for demonstrating non-parallelism of inhibition curves. As has been shown in this paper, the difference between antigen and a naturally occurring antibody to it can be readily demonstrated by comparison with a RIA using a non-specific precipitant. The antibody co-purified with IgG. Lactalbumin is the B protein of lactose synthetase and will bind to the A protein, galactosyl transferase, this reaction shows little species specificity [ 121. Although the A protein might interfere in a radioimmunoassay, this does not account for our observations since its physical properties are different from IgG [13] and it would have separated from the gamma globulin on purification. The data in Fig. 7 suggest that the human antibodies react with more than one antigenic determinant on bovine lactalbumin and that not all determinants are shared with the human protein. This explains why the human lactalbumin did not displace labelled bovine lactalbumin from the human antibody despite the fact that labelled human lactalbumin reacted with this antibody. During our study of the properties of the antibody we uncovered a material which inhibited the binding of labelled lactalbumin to antibody in the RIA but whose physical properties were quite different from lactalbumin. To date, we have not been able to identify this substance.

157

Once the spurious results due to naturally occurring antibodies had been eliminated, a comparison of healthy women with those suffering from either benign or malignant diseases of the breast confirmed the previous conclusion that lactalbumin is not a useful marker of mammary disease at the levels of sensitivity we are using [ 1,141. We examined plasmas from a group of pregnant women because we wanted to observe the behaviour in the RIA of plasmas which contained both antibodies and medium levels of human lactalbumin. It was interesting to note that production of lactalbumin in humans begins before parturition. We did not make a comprehensive study of women through the different stages of pregnancy but from the samples available to us we could detect lactalbumin by at least the 20th we,ek of pregnancy. In contrast, in the mouse, lactose synthesis is controlled by the production of lactalbumin which rises rapidly at parturition [15]. Acknowledgements We thank Mr. S. Pate1 for technical assistance, Mrs. M. Woollard for collecting the plasma samples, Dr. D. Frizel, Epsom District Hospital for supplying the samples of plasma from pregnant women and Professor A.M. Neville of the Ludwig Institute for Cancer Research for his support and advice. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Kleinberg, D.L. (1975) Science 190. 276-278 Rose, H.N. and McGrath, C.M. (1975) Science 190.673475 Woods, K.L. and Heath, D.A. (1977) Clin. Chim. Acta 18.129-133 Stevens, U., Laurence, D.J.R. and Ormerod. M.G. (1978) in Liquid Scintillation Counting, (Crook. M.A. and Johnson, P., eds.), Heydon and Sons, London, in Press Phillips. N.I. and Jenness, R. (1971) Biochim. Biophys. Acta 229. 407-410 David. G.S. and Reisfeld, R.A. (1974) Biochemistry 13,1014-1021 Ekins, R.P. and Samols. E. (1963) Lancet ii. 202 Rodbard, D. and LewaId, J.E. (1970) Acta Endocrinol. (Copenhagen) 64. SUPP~. 147, 79-98 Ormerod, M.G. (1977) in 5th Meeting of Int. Research Group for Carcinoembwonic Proteins.

Vol. 5

CoPen-

hagen, in press Rothberg, R.M. and Farr, R.S. (1965) Pediatrics 35, 571-588 Brew, K., CastelIino, F.J., Vanama, T.C. and Hi& R.L. (1970) J. Biol. Chem. 245. 4570-4582 Tanashi. N., Brodbeck. IJ. and Ebner, K.E. (1968) Biochim. Biophys. Acta 154, 247-249 Ley, J.M. and Jenness, R. (1970) Arch. Biochem. Biophw. 138.164-169 Coombes. R.C.. Gazet, J.S., Sloane, J.P.. Powles, T.J.. Ford, H.T., Laurence, D.J.R. and Nevfl% A.M. (1977) Lancet i, 132-134 Turkington, R.W.. Brew. K., Vanaman. T.C. and Hill, R.L. (1968) J. Biol. Chem. 243, 3382-3387