Solid-phase immunoradiometric assay for serum amyloid a protein using magnetisable cellulose particles

Journal of Immunological Methods, 54 (1982) 213-221 Elsevier Biomedical Press

213

Solid-Phase Immunoradiometric Assay for Serum Amyloid A Protein Using Magnetisable Cellulose Particles F.C. De Beer ~, R.F. Dyck 2 and M.B. Pepys Immunological Medicine Unit, Department of Medicine, Royal Postgraduate Medical School, Du Cane Road, London WI2 OILS, U.K.

Received 12 January 1982, accepted 20 March 1982)

An immunoradiometric assay for h u m a n serum amyloid A protein (SAA) was developed using magnetisable cellulose particles as the solid phase. Rabbit antiserum to SAA was raised by immunization with SAA isolated from acute-phase serum by gel filtration in formic acid. The antiserum was rendered monospecific for SAA by solid-phase immunoabsorption with normal h u m a n serum, which contains only traces of SAA, and some was coupled covalently to the cellulose particles, lmmunopurified anti-SAA antibodies were isolated from the monospecific anti-SAA serum by binding to, and elution from insolubilized acute-phase serum and were radiolabelled with n251. The assay was calibrated with an acute phase serum which contained 6000 times more SAA than normal sera with the lowest detectable level of SAA, and an arbitrary value of 6000 U / I was assigned to this standard. Sera were tested in the native, undenatured state and there was no increase in SAA immunoreactivity following alkali treatment or heating. The assay range was from 1-2000 U / I so that all SAA levels above 6 U/1 could be measured on a single ( l : 6) dilution of serum. The intra- and interassay coefficients of variation were 11.7 and 15.0% respectively. A m o n g 100 healthy normal subjects (50 male, 50 female) the median SAA level was 9 U/I, range < 1-100, with 93% below 20 U/1 and only 2% below the lower limit of sensitivity of the assay ( 1 u/i). Key words: serum am)'loid A protein - - ntagnetisahle celhdose particles - - intrnunoradiometric assay - sohd-phase IRMA

Introduction Reactive

systemic amyloidosis

inflammatory amyloid

are composed

about

8500.

about

12,500

is a r a r e b u t s e r i o u s c o m p l i c a t i o n

of some chronic

diseases. The fibrils which form the bulk of the deposits in this type of An can

of amyloid

immunochemically be

isolated

from

A protein

(AA), a peptide

cross-reactive

protein

plasma.

amino

The

of molecular

with acid

molecular sequence

weight weight

from

Present address: Department of Medicine. University of Stellenbosch, Republic of South Africa. 2 Present address: Department of Medicine, University of Saskatchewan, Saskatoon, Canada. 022-1759/82/0000-0000/$02.75 ~CJ 1982 Elsevier Biomedical Press

the

214

N-terminal of this serum amyloid A protein (SAA) is identical with the sequence of AA and AA may therefore be derived from SAA by proteolytic cleavage of the C-terminal portion of the precursor molecule (reviewed by Glenner, 1980a,b). In its native state in the plasma, SAA exists in a high molecular weight form and most of it is associated with high density lipoprotein, behaving as an apolipoprotein (Benditt and Eriksen, 1977). SAA is also an acute-phase reactant and its serum level increases rapidly and extensively following most forms of tissue injury, inflammation, infection or in malignant neoplasia (Rosenthal and Franklin, 1975; Gorevic et al., 1976; McAdam et al., 1978). Such increases in concentration may provide the material for deposition as AA fibrils. It is not clear why only a minority of individuals with chronically elevated SAA levels develop amyloidosis, but there is a series of distinct polymorphic forms of SAA and it has been suggested that some of these may be more amyloidogenic than others (Gorevic et al., 1976; Bausserman et al., 1980). Assays of SAA used hitherto have depended on cross-reactive antibodies raised by immunization with AA isolated from tissue amyloid deposits and have been standardised with samples of AA, of denatured SAA or of standard acute-phase serum (Rosenthal and Franklin, 1975; Sipe et al., 1976a; Benditt and Eriksen, 1977; Ignaczak et al., 1977; McAdam et al., 1978; Benson and Cohen, 1979). Unfortunately much of the antigenicity of AA is 'concealed' in native SAA (Franklin, 1976; Sipe et al., 1976b; Ignaczak et al., 1977) and in some but not all published assays, full immunoreactivity of SAA was detectable only after denaturation of serum samples with acid or alkali (McAdam et al., 1978; Hijmans and Sipe, 1979). We report here a new immunoradiometric assay (IRMA) for SAA using rabbit antiserum raised against SAA isolated from acute-phase serum and magnetisable cellulose particles to provide an easily separated solid phase. The method uses undenatured serum samples, is sensitive, precise and reproducible and has the extended range characteristic of IRMA based on this high capacity solid phase.

Materials and Methods

Serum amyloid A protein (SAA) SAA was isolated from acute-phase serum by making it 10% in formic acid and then performing sequential gel filtration on Sephadex G-100, Sephadex G-75 and Sephadex G-50 superfine (Pharmacia Hounslow, Middx), eluting with 10% formic acid after the method of Anders et al. (1975). Fractions containing SAA were located by double immunodiffusion using rabbit anti-AA serum kindly provided by Dr. Martha Skinner, Arthritis Section, Boston University, Boston, MA, U.S.A. Anti-SAA antiserum SAA was coupled to glutaraldehyde-activated polyacrylamide beads (Bio-Gel P300, 50-100 mesh from Bio-Rad Laboratories, Watford, Herts) as described by Ternynck and Avrameas (1976). Rabbit antiserum to SAA was raised by repeated immunization with these beads in Freund's complete adjuvant (Difco, West Molesey,

215 Surrey). This antiserum contained antibodies against other serum proteins in addition to SAA and these were removed by absorption with insolubilized normal human serum (NHS). The concentration of SAA in N H S is so low that there was only minimal loss of anti-SAA activity in this procedure. Five millilitres of normal human serum were coupled to 5 g of CNBr-Sepharose (Pharmacia) according to the manufacturer's instructions. After passing the crude rabbit anti-SAA serum over the S e p h a r o s e - N H S column it no longer gave any precipitation lines with N H S in gel immunodiffusion tests, but retained the capacity to stain known AA-positive amyloid in indirect immunofluorescence testing. Double immunodiffusion in gel testing for anti-SAA activity was performed as follows: samples were placed in 1% w / v agarose gel (Indubiose A37, IBF, Gennevilliers) in 0.023 M veronal buffer, p H 8.6 containing 4% w / v P E G 6000 (Sigma, Poole, Dorset) and incubated at 37°C for 24 h. Gels were then washed in 5% w / v NaC1/0.01% w / v sodium azide for 48h before pressing and staining with either Coomassie blue or Sudan black by standard techniques. Anti-SAA antibodies The I g G fraction of the NHS-absorbed rabbit anti-SAA serum was isolated using Sepharose-protein A (Pharmacia) according to the manufacturer's instructions. After being concentrated to 40 m g / m l this IgG preparation gave no precipitation lines with N H S in gel immunodiffusion tests over a wide range of antibody and antigen dilutions. When tested against acute-phase serum it gave a single immunoprecipitation line which identified with that produced by a commercial anti-AA serum (Atlantic Antibodies, American Hospital Supply, Didcot, Oxon.) and was stainable with Sudan black, indicating that the antigen was associated with lipid (Fig. 1). Purified specific anti-SAA antibodies were isolated by elution from a solid phase immunoabsorbent prepared as follows. Five millilitres of acute-phase human serum from a patient with myocardial infarction were coupled to 3 g of CNBr-Sepharose (Pharmacia) according to the manufacturer's instructions. The column was washed sequentially with 0.2 M glycine-HC1 buffer, p H 2.2, 1.0 M K 2 H P O 4 and phosphate-buffered isotonic saline, p H 7.2 (PBS). One hundred and sixty milligrams

a,

ff

Fig. 1. Immunodiffusionin gel analysis of IgG fraction of anti-SAA serum after absorption with NHS. a: 1, absorbed IgG fraction of anti-SAA serum; 2, commercial anti-AA serum (Atlantic Antibodies); 3, acute-phase human serum, b: 1, absorbed IgG fraction of anti-SAA serum; doubling dilutions from neat of acute phase human serum in surrounding wells. Stained with Sudan black.

216 of the IgG fraction of monospecific rabbit anti-SAA serum were passed over the Sepharose-acute-phase serum column which was then thoroughly washed with 0.01 M Tris-buffered 1.14 M saline containing 0.01 M E D T A until the A2~ 0 of the effluent was zero, before sequential elution at 4°C with 0.2 M glycine-HCl buffer, p H 2.2 and 1.0 M K z H P O 4. One and a half milligrams of IgG were recovered, and shown to be anti-SAA by its precipitation of SAA from acute-phase serum in double immunodiffusion testing. The precipitation line was stainable with Sudan black and identified with the precipitation line produced by the commercial anti-AA serum. The eluted anti-SAA antibody was finally dialyzed against distilled water and lyophilized.

Solid-phase anti-SAA Cellulose/iron oxide particles, a magnetisable solid phase prepared as described elsewhere (Robinson et al., 1973) and kindly provided by Dr. G.C. Forrest, Serono, London, were activated with CNBr (1 g / g of solid phase) at p H 11.5 (Axen et al., 1967). Eighty milligrams of the lgG fraction of the absorbed, specific rabbit anti-SAA serum were coupled to 1 g of activated solid phase at pH 8.6. Residual active sites were blocked with ethanolamine and, after alternate washes with pH 4.0 acetate buffer and p H 8.0 borate buffer, the particles were stored at 4°C in 20 ml of PBS containing 2 m l / l Tween-20 (Sigma) and 0.1 g / l NAN,.

Radiolabelling of anti-SAA lmmunopurified anti-SAA antibody (75 ~g) was labelled with 1251 (1 mCi) from Nal2SI (IMS30, Radiochemical Centre, Amersham, Bucks.) using chloramine T ( B D H Chemicals, Poole, Dorset) (Greenwood et al., 1963). There was 75% incorporation of radioiodine into protein, and after passing the reaction mixture over a column of Sephadex G-25 (Pharmacia, PD10) all the activity associated with protein was precipitable by trichloroacetic acid. The labelled anti-SAA was stored at 4°C in solution in PBS containing 10 g/1 bovine serum albumin (BSA) (Sigma) and 1 g / l sodium azide.

Standards for assay calibration Acute-phase serum from a patient with myocardial infarction was adopted as the standard and was stored in aliquots frozen at -20°C. The SAA level in this serum was found to be extremely high, approximately 6000-fold higher than the lowest levels detected in normal sera. It was therefore arbitrarily assigned a value of 6000 SAA units/l.

Denaturation of SAA SAA in the acute-phase standard serum was denatured by making it to a final concentration of 0.1 M N a O H , incubating at room temperature (21 23°C) for 4 h and neutralizing with 0.1 M HCI. This alkalinization has the same effect on immunogenicity as denaturation with 10% formic acid (Van Rijswijk, 1981). Acutephase standard was also denatured by incubation at 52°C for 3 h (Karlin et al., 1976).

217

Serum samples Sera from 100 normal adult volunteer HBs antigen-negative blood donors were obtained by courtesy of Dr. T.D. Davis, North London Blood Transfusion Service, Edgware, and stored at - 2 0 ° C before assay. Sera from patients who had recently suffered acute myocardial infarction were used to provide acute phase materials.

lmmunoradiometric assay The test was performed in 55 m m × 12 ram, 3.5 ml plastic tubes (Sarstedt, 55.484) and the diluent used for all reagents was PBS containing 2 ml/l Tween-20 and 10 g / l BSA (Sigma). Seventy five microlitres of solid-phase anti-SAA was dispensed into each tube followed by a source of SAA, either a known standard (25 /~l) or an unknown sample (25/~1 of a 1 : 6 dilution of serum). The standards were prepared by serial dilution of the acute-phase serum, to which an SAA level of 6000 U/1 had been assigned, to provide samples containing 2000, 1000, 100, 10 and 1 U/I. The tubes were incubated for 60 min at room temperature (21-23°C), and shaken sufficiently vigorously to keep the solid phase in suspension either by hand or in a horizontal mechanical shaker. The solid phase was then sedimented by locating each tube vertically over a powerful magnet and the supernatant fluid discarded by decantation. The particles in each tube were washed with a 2.0 ml volume of diluent, resedimented and the washings decanted. One hundred microlitres of ~251-1abelled anti-SAA (1500 dps) was then dispensed into each tube and all tubes were incubated again for 60 min at room temperature (21-23°C) whilst being shaken as before. At the end of this time the tubes were counted in a y-scintillation counter (NE 1600, Nuclear Enterprises, Edinburgh), the solid phase was resedimented on the magnet, the supernatant discarded and the pellet washed with diluent as before. Activity bound to the solid phase was counted and the percentage of counts bound was calculated for each tube. The non-specific uptake of label onto solid phase which had been incubated with diluent instead of serum was less than 5% and the value for such blank tubes which were included in each run was subtracted from the assay and calibration tubes.

Results

Quantity of solid-phase anti-SAA Experiments using varying dilutions of solid phase showed that 16.6 g / l of solid phase had sufficient capacity for antigen binding to detect SAA levels up to 2000 U/1. This allowed an adequate assay range for serum samples diluted 1 : 6 since SAA levels > 6000 U / I are very rarely seen. With lower dilutions of acute-phase sera, yielding SAA levels in the samples higher than 2000 U / l , there was 'diversion' of labelled antibody reacting with excess fluid-phase antigen, which is a recognized feature of this form of I R M A (De Beer and Pepys, 1982).

Effect of removal of unbound cold antigen before addition of 1-'51-labelled anti-SAA When the SAA-containing sample to be assayed was not separated from the solid

218

2O

z E

5

0

i

1;

i000

SAA concentration (u/I) Fig. 2. Standard I R M A curves obtained with ( 0 ) and without ( O ) washing of the solid-phase anti-SAA after incubation with SAA-containing material and before addition of ~zSI-labelled anti-SAA.

phase before addition of labelled anti-SAA, the calibration curve was altered so that diversion of label occurred at much lower SAA levels (Fig. 2). It was therefore necessary to adopt the procedure, described in Materials and Methods, of a 2-stage assay, a preliminary incubation of solid phase with sample followed by washing and then addition of label.

Effect of denaturation of SAA on the assay Standard curves were constructed with denatured standard sera. Sodium hydroxide denaturation brought about no significant change in immunoreactivity of SAA. Heat denaturation decreased immunoreactivity and depressed the standard curve.

TABLE I C O E F F I C I E N T S OF V A R I A T I O N OF T H E I R M A FOR SAA Mean ± S.D. SAA concentration ( U / l ) (coefficient of variation%). Sample

A B C

Assay number

Inter-assay variation b

1~

2

3

4

5

2755~338 153 ~+ 17 6+~ 1

2270 188 <6

2790 118 6

3 190 136 8

3496 142 <6

2900--+417(14.4%) 147±23(15.6%) NA t

a Five replicates of each sample ( A - C ) were tested. Mean intra-assay coefficient of v a r i a t i o n = 11.7% b Samples A - C were tested in 5 separate assays. Overall mean coefficient of variation = 15%. NA = not applicable.

219 25

20

C

5

0

f

lb

lOb

lOOb iooo

SAA concentration (u/I) Fig. 3. Typical standard curve obtained in the IRMA as routinely performed (see Results). Each point represents the mean ± S.D. of triplicate estimations of a single calibration sample.

Reproducibility and sensitivity of the assay Standard conditions were adopted as follows: mix 75/~1 of solid-phase anti-SAA at 16.6 g / l with 25 #1 of SAA standard serum or unknown, incubate for 1 h then wash, add 100/~1 of 125I-labelled anti-SAA containing 1500 dps of activity, incubate 1 h, count, wash and count. Three individual sera, previously shown to cover the clinical range, were diluted 1 : 6 and 5 replicates of each were run in a single assay (Table I, assay 1). The overall intra-assay coefficient of variation was 11.7%. The same 3 sera were run in 5

TABLE II SERUM C O N C E N T R A T I O N OF SAA ( U / I ) IN VOLUNTEER BLOOD DONORS OF D I F F E R E N T AGES Age (years)

18-27 28-37 38-47 48-61

Males

Females

Median

Interquartile range

Range

n

Median

Interquartile range

Range

n

7 12 9 12

5-15 5-18 7-14 8-16

<1-16 < 1-100 6-33 4-25

18 16 8 8

8 9 9 9

6-9 7-12 6-15 9-16

1-33 4-16 4-19 1-58

20 12 7 11

220 separate assays (Table I, assays 1-5). Fresh sets of standards were prepared for each assay and each standard dilution was tested in triplicate to establish the standard curve. The overall inter-assay coefficient of variation was 15%. Over a period of 6 weeks these standard curves were very similar to each other; a typical example is shown in Fig. 3. It should be noted that SAA levels as low as 1 U/1 were detectable.

Serum SAA concentration in normal health), adults Serum samples were assayed neat and in 1:6 dilution. The results obtained are shown in Table II. Ninety three per cent of these normal sera had levels below 20 U/I. There were 2 males amonst the whole group of 100 subjects in whom no SAA, i.e., a level below 1 U / l , was detectable even in neat serum. Discussion

A major problem in assays for SAA is that the protein is amphipathic and is normally associated with H D L particles. When it is dissociated by denaturation its antigenicity is altered and it is heterogeneous in terms of physical properties, particularly solubility under physiological conditions, and chemical association with fragments of other proteins particularly albumin and prealbumin (Marhaug and Husby, 1981). Previous assays for SAA have used antibodies raised against AA and have been calibrated either with standards of isolated, denatured AA (Rosenthal and Franklin, 1975: Sipe et al., 1976a; Benditt and Eriksen, 1977) or SAA (McAdam et al., 1978; Hijmans and Sipe, 1979) or with a standard acute-phase serum (Benson and Cohen, 1979). In some assays denaturation of serum samples has been essential for full expression by SAA of antigenicity cross-reactive with AA, and whether this is necessary presumably depends on the precise specificity of the anti-AA antibodies used. In its native state in the H D L particle SAA evidently expresses some epitopes which are not detected by anti-AA antisera, while the epitopes specific for denatured A A are concealed. In the present assay we elected to use antibodies raised against SAA and immunopurified them for labelling by elution from native SAA in H D L in whole serum. Our method does not therefore require denaturation of serum, a matter of some logistic importance in the handling of many samples, and it detects SAA in its physiological state. We have used an acute-phase serum with a high SAA level as an arbitrary calibration standard. Relating the values obtained to true weights or molarities of native SAA presents considerable theoretical and technical problems. SAA itself is polymorphic (Bausserman et al., 1980) whilst the H D L particle is complex and there is no information yet available on qualitative or quantitative aspects of the insertion and antigenic display of SAA within or upon it. The I R M A approach with a high capacity solid phase provides the extended assay range which is desirable in quantitating acute-phase proteins, particularly those which increase over such a broad incremental .range as SAA. We were able to measure raised SAA levels in all cases in a single serum dilution, although neat serum was required for detection of the low normal range. Nonetheless the assay was sufficiently sensitive to measure SAA concentration in 98% of normal individuals.

221 T h e p r o c e d u r e s used were technically simple and the overall assay time relatively short so that 50 serum samples could be processed to p r o v i d e results within 3 h. Th e m a g n e t i s a b l e particles greatly facilitate phase separation and by starting batches at a p p r o p r i a t e intervals a single w o r k e r c o u l d process several h u n d r e d samples per day. T h e r ep r o d u ci b i l i t y of results o b t a i n e d seems quite acceptable and c o m p a r e s reasonab l y well with other similar methods. U n f o r t u n a t e l y no c o m p a r i s o n in this respect with o t h er published assays for S A A is possible since none of these q u o t e their inter- or intra-assay coefficients of variation. We have applied o u r assay to several h u n d r e d sera from well characterized patients with varying degrees of activity of a variety of diseases including r h e u m a t o i d arthritis, j u v e n i l e c h r o n i c arthritis, systemic lupus erythematosus, C r o h n ' s disease and ulcerative colitis ( D e Beer et al., 1982). T h e results are of both a c a d e m i c and practical interest and suggest that regular m o n i t o r i n g of S A A levels using a sturdy, accessible and r e p r o d u c i b l e assay like this m a y have a role in m a n a g e m e n t of some c h r o n i c i n f l a m m a t o r y disorders.

Acknowledgements This work was s u p p o r t e d by M R C P r o g r a m m e G r a n t G 9 7 9 / 5 1 to MBP. We thank Dr. G o r d o n Forrest, Serono, L o n d o n for coupling a n t i - S A A to magnetisable cellulose particles and Miss J o a n R o b i n s for expert secretarial assistance.

References Anders, R.F., J.B. Natvig, T.E. Michaelsen and G. Husby~ 1975, Scand. J. lmmunol. 4, 397. Axen, R., J. Porath and S. Ernback, 1967, Nature (London) 214, 1302. Bausserman, L.L., P.N. Herbert and K,P.W.J. McAdam, 1980, J. Exp. Med. 152, 641. Benditt, E.P. and E. Eriksen, 1977, Proc. Natl. Acad. Sci. U.S.A. 74, 4025. Benson, M.D. and A.S. Cohen, 1979, Arthritis Rheum. 22, 36. De Beer, F.C. and M.B. Pepys, 1982, J. Immunol. Methods 50, 299. De Beer, F.C., R.K. Mallyor, E. Fagan, J.G. Lanham, G.R.V. Hughes and M.B. Pepys, 1982, Lancet ii, 231. Franklin, E.C., 1976, J. Exp. Med. 144, 1679. Glenner, G.G., 1980a, New. Engl. J. Med. 302, 1283. Glenner, G.G., 1980b, New Engl. J. Med. 302, 1333. Gorevic, P.D., C.J. Rosenthal and E.C. Franklin, 1976, Clin. Immunol. Immunopathol. 6, 83. Greenwood, F.C., W.M. Hunter and J.S. Glover, 1963, Biochem. J. 89, 114. Hijmans, W. and J.D. Sipe; 1979, Clin. Exp. Immunol. 35, 96. Ignaczak, T.F., J.D. Sipe, R.P. Linke and G.G. Glenner, 1977, J, Lab. Clin. Med. 89, 1092. Karlin, J.B., D.J. Juhn, T.I. Start, A.M. Scanu and A.H. Rubenstein, 1976, J. Lipid Res. 17, 30. McAdam, K.P.W.J., R.J. Elin, J.D. Sipe and S.M. Wolff, 1978, J. Clin. Invest. 61. 390. Marhaug, G. and G. Husby, 1981, Clin. Exp. Immunol. 45, 97. Robinson, P.J., P. Dunnill and M.D. Lilly, 1973, Biotechnol. Bioeng. 15, 603. Rosenthal, C. and C. Franklin, 1975, J. Clin. Invest. 55, 746 Sipe, J.D., T.F. lgnaczak, P,S. Pollock and G.G. Glenner, 1976a, J. lmmunol. 116, 1151. Sipe, J.D., K.P.W.J. McAdam, B.F. Torain and G.G. Glenner, 1976b, Brit. J. Exp. Pathol. 57, 582. Ternynck, T. and S. Avrameas, 1976, in: Immunoadsorbents in Protein Purification, Suppl. 3, ed. E. Ruoslahti. Scand. J. Immunol. (Universitetsforlaget, Oslo) p. 29. Van Rijswijk, M.H., 1981, Amyloidosis, Thesis. University of Groningen (Drukkerij Van Denderen B.V., the Netherlands).