- Email: [email protected]
Immunoassays for inhibin and its subunits Further applications of the synthetic peptide approach
Journal oflmmunological Methods, 165 (1993) 167-176
167
© 1993 Elsevier Science Publishers B.V. All rights reserved 0022-1759/93/$[16.00
JIM 06818
Immunoassays for inhibin and its subunits Further applications of the synthetic peptide approach Nigel G r o o m e a n d M a r t i n O ' B r i e n School of Biological and Molecular Sciences, Oxford Brookes Unit:ersity, Gipsy Lane, Headington, Oxford OX3 0BP, UK
(Received 30 March 1993, revised received 18 May 1993, accepted 28 May 1993)
We describe the preparation of a new rat monoclonal antibody (CRC1) to the N-terminal sequencc of the 43 kDa subunit of human ovarian inhibin, and its use together with other anti-peptide monoclonal antibodies, in two-site immunoassays for the detection of inhibin-related material in biological fluids. The Fab fraction of a mouse monoclonal antibody (R1) to the N-terminal portion of the 20 kDa a subunit, coupled to alkaline phosphatase, was used for detection, and either CRC1 or a monocional antibody (E4) to t h e / 3 - A subunit were used as capture antibodies. The E 4 / R I combination, expected to measure dimeric bioactive inhibin, could detect less than 2 p g / m l of recombinant inhibin in diluent, gave good recovery of activity spiked into human blood, and could measure significant levels of immunorcactivity in sera from women undergoing ovulation induction, and in some normal women. Scra from post-menopausal women contained undetectable levels. Apparent inhibin levels in human follicular fluid were increased six-fold by p r e t r e a t m e n t with 8 M urea, suggesting masking of epitopes in this fluid. Activin cross-reactivity in the assay was 0.05%. The R 1 / C R C 1 assay, expected to measure only large molecular weight forms of inhibin or its a subunit, could detect immunoreactivity in human FF diluted 50,000-fold, and in all sera tested, although the levels in the hyperovulated women were higher. By contrast to the E 4 / R 1 assay much of the immunoreactivity was labile during the clotting process, or subsequent assay, and reliable measurements on blood with this assay will require special samplc collection procedures. These results demonstrate the value of anti-peptide monoclonal antibodies in the study of inhibin, and the results obtained with CRC1 show that antibodies useful for immunoassays can somctimes be obtained without the purified target molecule being available for immunization or screening. Key words: lnhibin; Activin; Synthetic peptide; Monoclonal antibody; ELISA
Correspondence to: N.P. Groome, School of Biological
Sciences, Brookes University. Headington, Oxford OX3 OBP, UK. Tel.: +44-865-819255; Fax: +44-865-819928). AbbreL'iations: BSA, bovine serum albumin; ELISA, enzyme-linked immunoadsorbent assay; FF, follicular fluid; IVF, in vitro fertilization; MgCle, magnesium chloride; NaCI, sodium chloride; PBS, phosphate-buffered saline.
Introduction Inhibin subunits are known to be present in bovine and human FF in a wide range of molecular forms (Knight et al., 1989; Miyamoto et al., 1986). Some of these have inhibin bioactivity, others have activin activity, and some have no
168
known biological activity. Presently it is not possible to quantify many of these forms by specific immunoassays, and the development of such assays is a formidable challenge in view of the low concentration of these forms present in human serum and FF, and the limited availability of recombinant forms. It seems unlikely that a full understanding of the role of inhibin heterogeneity will be achieved without such assays. Inhibin assays specific for bioactive inhibin would be useful for numerous clinical studies to replace the existing radioimmunoassays, which are unable to distinguish dimeric inhibin from forms of the c~ subunit (Schneyer et al., 1990), and bioassays, which are tedious and prone to interference from other hormones. An essential step in immunoassay development is the preparation of antibodies of high specificity and affinity. Inhibin and activin have proved to be poor immunogens, possibly because they are closely conserved during evolution. Some investigators, including ourselves, have found synthetic peptide immunogens particularly useful (Groome et al., 1990; Groome and Lawrence, 1991). The advantages of the synthetic peptide approach are: (1) specificity is largely predetermined by the choice of peptide, thus avoiding the need for extensive epitope mapping; (2) coupling of peptides to carrier proteins before immunization can provide additional sites for helper T cell recognition, thus improving antibody production to the peptide; (3) it does not require large amounts of purified natural or recombinant material. Although some workers have viewed polyclonal antibodies raised to peptides as an alternative to monoclonal antibodies, we have concentrated on the latter in the hope of producing defined reagents for long term use. In the present paper we describe the immunization and screening procedures which have now enabled us to make a new rat monoclonal antibody reacting with the N-terminal sequence of the 43 kDa form of the human inhibin a subunit, a molecule which is not presently available in purified form. We show how this antibody, with others to the Nterminal portion of the 20,000 ot subunit (Groome et al., 1990) and the 82-114 sequence of the /3-A subunit (Groome and Lawrence, 1991), can be
used to configure very sensitive two-sitc enzyme immunoassays with alkaline phosphatase as the label. Preliminary data are reported which indicate that these assays should prove applicable to measurements on human blood samples. These studies reinforce our earlier claim that high affinity monoclonal antibodies suitable for use in ultrasensitive two-site immunoassays can readily be made using synthetic peptide immunogens (Groome, 1991). In the present report we show that purified forms of the intended analyte are not always an essential requirement for the development of such assays. Furthermore, in our original assay (Groome, 1991) activin showed a 5.2% cross-reaction and post-menopausal human serum (not expected to contain dimeric inhibin) gave a strong signal. In attempts to overcome these specificity problems we reversed the roles of the two antibodies used for capture and as label, with the great improvement in assay specificity reported here.
Materials and methods
Monoclonal antibodies The preparation of clone E4, making a mouse IgG2b antibody reacting with the inhibin /3-A subunit (Groome and Lawrence, 1991), and clone R1 making a mouse IgG2a antibody reacting with the inhibin a subunit (Groome et al., 1990), have been described elsewhere, as has their use in two-site immunoassays for recombinant human inhibin and activin (Groome, 1991). Clone CRC1, making a rat monoclonal antibody reactive with the N-terminal portion of the 43 kDa form of the human inhibin a subunit, was made using as an immunogen a synthetic peptide of sequence H A L G G F T H R G S E P E C . This particular sequence was chosen because N- and C-terminal peptides are often able to elicit antibodies which recognizc the parent protein. The peptide was coupled to tuberculin through the thiol group of cysteine by heterobifunctional chemistry as previously described (Groomc et al., 1990; Groome and Lawrence, 1991). Immunization of animals (Lewis rats) and fusion of spleen cells to the S P 2 / 0 myeloma line were carried out as in our previous work (Groome and Lawrence, 1991).
169
Identification of antibodies reacting with inhibin forms in human FF
pected to bind a variety of molecular forms of inhibin, including a subunits unassociated with a /3 subunit. To screen the individual supernatants from a fusion of rat spleen cells with the S P 2 / 0 myeloma cells 5 0 / z l of each supernatant were transferred into wells prepared as above and incubated for 2 h at room temperature with agitation. The microplates were washed and then incubated for 1 h with 50/~1 of a 5 / z g / m i solution of a biotinylated mouse monoclonal antibody specific for rat kappa light chains (Serotcc MCA 187B). After another wash, 50 /~1 of a 1 in 1000 dilution of streptavidin/peroxidase (Serotec STAR 5A) were added for 1 h at room temperature. After a final wash, 50 /~1 of TMB peroxidase substrate (Dynatech) were added for 30 min. From this screening process clone CRCI was identified as producing the most promising antibody. Subsequent isotyping with a Serotec kit revealed this to be a rat IgG2b antibody. The fact that this hybridoma was of rat × mouse origin meant that ascites production had to be performed in nude mice and purification done using protein G chromatography. Biotinylation was carried out as described previously
A major problem was anticipated when it became necessary to screen the fusion supernatants, because no purified preparation of 43 kDa inhibin was available. An indirect capture procedure was used for this purpose, which is why we used rats as the spleen cell source. Firstly, Nunc Immunoplates (type 1 with certificate) were coated overnight with 100 p.I/well of 0.1 M sodium bicarbonate pH 8.0 containing 1 0 / z g / m l of purified IgG from monoclonal R1. The following day the wells were blocked by incubation with 150 /~! of a 1% ( w / v ) solution of BSA (Sigma A3294) for 1 h at room temperature. A 1 in 10 dilution of pooled human FF (from an 1VF clinic) was made in a diluent consisting of 0.05 M sodium phosphate buffer containing 1.2% ( w / v ) NaCI, 0.5% (w/v) Tween 80 and 1% ( w / v ) BSA. 100/~1 of this solution were added to each well on the R1 coated microplates, and the plates incubated in the cold room for 4 h. The R1 antibody was known to be useful for immunoaffinity extraction of inhibin and its subunits from FF (Groome et al., 1990), and during this incubation it was ex-
Substrata Stre0tavii?
Colour
(6)
Substrata
Per°xidale
(5)
Streplavidin peroxidase
/ oo.
antibodies
(3) Rat in fusion supernatanls
g (1] R~
mouse anli ral kappa monoclonal
(4] B i o t i n y l a t e d
(MCA 187B) (2]
Inhibin forms captured from
human f o l l i c u l a r fluid
MAb
S C R E E N I N G T E C H N I Q U E S USED TO P R E P A R E MONOCLONAL ANTIBODY CRCl Fig. 1. Shows the principle of the ELISA screening assay used for preparation of the CRC1 monoclonal antibody.
170
(Groome et al., 1990). The screening procedure used to produce hybridoma CRC1 is shown in Fig. 1.
Two-site immunoassay for dimeric inhibin (using monoclonals E4 and RI) Two alternative procedures, using different commercially obtainable microplates, were employed to obtain a stable coat of the E4 antibody on the surface of microplates, whilst retaining the binding capacity of the antibody. These appear to give identical results in the inhibin immunoassays reported here. In the first of these, biotinylated E4 IgG was added to plates carrying covalently attached streptavidin (Bioproducts, Pias Grono Road, Wrexham, UK). In the second, plates carrying hydrazide groups were used to anchor the E4 IgG through the carbohydrate on its Fc region (Avidplate HZ, Bioprobe, Tustin, CA, USA). Streptavidin-coated microplates were first washed three times in distilled water and then coated with 50 ~1 of biotinylated E4 IgG at 2.5 p~g/ml overnight at room temperature. The diluent used for this stage was 1% ( w / v ) BSA in PBS. The following day the plate was washed with PBS, and excess sites on the polystyrene blocked by the addition of 150 p.I of 1% (w/v) BSA, 0.1% (w/v) sodium azide in PBS. Plates could be stored in this form for several months. To couple E4 lgG to Avidplates required prior oxidation of the antibody following the manufacturer's instructions. E4 IgG at a concentration of 1 m g / m l in 50 mM sodium acetate buffer pH 5.0, was mixed with 0.1 vol. of 100 mM sodium periodate and placed in the dark for 30 min. The reaction was stopped by the addition of 0.01 vol. of ethylene glycol, and the solution desaited into 50 mM sodium acetate buffer pH 5.0. The oxidised IgG (50 ~ l / w e l l ) was directly added to the wells of Avidplates at a concentration of 5 ~.g/mi, after dilution in 50 mM acetate buffer pH 5.0. The plates were left overnight at room temperature, and then blocked with 150 t~l of I% (w/v) BSA, 0.1 M Tris-HCl pH 7.5, 0.1% (w/v) sodium azide. Plates could be stored in this form for several months. Immediately before use the plate (streptavidin or hydrazide) was washed thoroughly with 0.05% ( w / v ) Tween 20 in 0.1 M Tris-HCI buffer pH 7.5
and blocked by the addition of 150 /zl of 1% (w/v) BSA in PBS. The plate was dried on paper towelling and to each well was added 100 /~l of recombinant human 32 kDa inhibin (Genentech) solution made as standards in an assay diluent containing 5% (w/v) Triton X-100, 10% (w/v) BSA, 5% (v/v) mouse serum in 0.1 M Tris-HCl buffer pH 7.5 containing 0.15 M sodium chloride. Samples containing unknown amount of inhibin were diluted in the same diluent. The plate was incubated overnight on a shaker at 4°C. The following day the plate was washed thoroughly and to each well was added 50 ~tl of a 1 ~ g / m l solution of the Fab fragment of clone R1 which had previously been coupled to alkaline phosphatase by heterobifunctional chemistry. The Fab fragment was prepared from purified IgG by standard pepsin digestion, the coupling to the enzyme as described previously (Ishikawa et al., 1983), and the final preparation was size fractionated to obtain forms with different molar ratios of enzyme and Fab. The fraction used in our studies has about 2 Fab molecules per alkaline phosphatase. After incubation for 1 h at room temperature on a shaker the plate was washed extensively over 15 min in wash solution and the wells emptied. Ultrasensitive detection of alkaline phosphatasc for the experiments reported here initially used a commercially available kit (AMPAK from Dako Diagnostics). 50 ~l of the substrate solution were added to each well and the plate was tightly covered and incubated for 2 h at 37°C. In order to extend the substratc reaction to this time it was necessary to make the kit substrate 1 mM in MgCl 2 (Dako personal communication). At the end of this time the plate was allowed to stand at room temperature for 15 min prior to adding the kit amplifier. The red colour development took place rapidly, and as soon as the zero analyte wells began to appear slightly red the reaction was stopped by the addition of 50 ~,1 of 0.4 M HCI. The absorbance of each well at 490 nm was read on a microplate reader (Biotek) with the reference wavelength at 620 rim, and data processing was done with associated software (Kineticalc). We have subsequently run these assays with in-house amplification reagents following published procedures (Stanley et al., 1985).
171
Exactly the same procedure was used as in the assay for dimeric inhibin except that the coated antibody was biotinylated CRC1. Since no relevant recombinant or purified standard for forms of inhibin containing large a subunits was available the standard used in this assay was a pool of human FF assigned an arbitrary potency of 50,000 U / m l . This FF was obtained from an IVF clinic.
dilution series started with a 1 / 2 dilution in the Triton assay buffer and further dilutions down to 1/32 were made in the same buffer. The recovery of inhibin in each sample was expressed as a percentage. For the purposes of calculating recoveries it was assumed that 50% of the blood volume was plasma. Thus a final concentration of inhibin in a plasma or serum of 2 n g / m l would correspond to 100% recovery of that added.
Localization of the epitopes for monoclonal RI and E4
Recovery of large forms of the inhibin et subunit from spiked human blood
Since monoclonal R1 was made by immunization with the 20 kDa a subunit 1-32 peptide sequence (Groome et al., 1990) and E4 with the /3-A 82-114 sequence (Groome and Lawrence, 1991) experiments were done to iocalise the reactive epitopes more closely. Overlapping 15mer synthetic peptides from the a and /3-A subunits were used in competition ELISA experiments to block the binding of each antibody to the corresponding long peptide immunogens adsorbed on to ELISA plates.
This was carried out as for dimeric inhibin except that the spiking was performed by adding 100 ~zl of pooled human FF (assumed to contain 50,000 U / m l ) to each tube, and this was finally measured in the R 1 / C R C 1 assay using as a standard the same FF preparation to make the standards. For this experiment the CRCI was attached to streptavidin plates. Full recovery of the added inhibin would thus correspond to 1000 U/ml.
Two-site assays for large forms of the a subunit (using monoclonals R1 and CRC1)
Recovery of dimeric inhibin from spiked human blood 10 ng of 32 kDa recombinant inhibin, dissolved in 0.I ml of 1% BSA, were added to different tubes for blood collection. The tubes included standard heparinised, citrated, and E D T A tubes for obtaining plasma, and a glass tube with no anticoagulant for obtaining serum. Immediately the inhibin was in the tubes one of the authors (N.P.G., male aged 45) was bled into the tubes up to the 10 ml mark. The samples were immediately mixed and returned to our laboratory. Within 30 min, preparations of heparinised, citrated and E D T A plasma were obtained. Each was divided into three aliquots for storage before assay: one aliquot frozen, one kept at 4°C, and the other kept at room temperature. The sample without anticoagulant was allowed to stand at 4°C overnight. The following day, preparations of serum and all the plasmas were assayed for inhibin in the E 4 / R 1 assay described earlier (hydrazide plates), with a standard concentrations made using the same preparation of inhibin which had earlier been spiked into the blood. Each
Effect of urea pretreatment of human FF on apparent inhibm levels A 1/10 dilution of human FF was made in either 8 M urea, 0.1 M Tris-HCl pH 7.5 buffer or the assay diluent referred to earlier. Both samples were incubated at room temperature for 10 min before further dilutions were made in assay diluent. The samples were then assayed in the E 4 / R I two-site assay using E4 coupled to an Avidplate.
Results
Epitopes recognized by the antibodies used here The R1 antibody reacted with only one of the overlapping 15mer peptides from the a subunit, corresponding to sequence MSWPWSPSALRLLQR. The epitope of CRC1 was obviously in the first 15 amino acids of the 43 kDa a subunit used for immunization. The inability of this antibody to react with bovine inhibin (Fig. 2) suggests that some of the three amino acids in this region which differ between human and bovine inhibin (positions 3, 7 and 8) are important for CRC1
172 . . . .
..
.~,.,.; ,~
-
--
~ -:
-
TABLE I D A T A F O R R E C O M B I N A N T 32 kDa INHIBIN STAND A R D S IN T H E E 4 / R 1 T W O - S I T E ASSAY
t:i
Concentration of 32 kDa inhibin
~,~,~-:~ '.:
0 :
:
.
.
.
-.
,:., :..... . . . .
Fig. 2. Shows the application of the R 1 / C R C I two-site assay to serial dilutions of h u m a n FF (top two rows), bovine FF (middle two rows), and post-menopausal serum (bottom two rows). T h e dilutions start at 1 in 2 and double across the plate.
recognition. The E4 antibody reacted with only one of the overlapping peptides from the /3-A subunit corresponding to sequence KLRPMSML Y Y D D G Q N . In competition ELISA experiments the corresponding peptide from the /3-B subunit of inhibin was almost as reactive (five-fold less). We believe that our two-site assay using this antibody would measure inhibin-B with similar effectiveness to the inhibin-A tested here, although no inhibin-B is available to confirm this.
Two-site assay for dimeric inhibin with the E4 / R1 antibody combination Fig. 3 shows the results of an assay in which our two-site assay with antibodies E4 and R1 was used to detect recombinant human 32 kDa in-
Picograms/ml
32kDa
Mean absorbance
0.049, 0.046 0.050, 0.048 0.081.0.092 0.125.0.128 0.182, 0.193 0.295, 0.323 0.448, 0.487 0.688, I).706 0.985, 1.040 1.18, 1.16
0.048
(pg/ml)
,
.
Absorbance
3.9 7.8 15.6 31.25 62.5 125 250 500
0.087 0.127 0.188 0.309 0.468 0.697 l.l)l 1.17
hibin. The 3.9 p g / m l standard could be easily distinguished from zero by visual inspection of the freshly developed plate. Raw data from such an assay are presented in Table I, from which it can be seen that the 4 p g / m l standard gave twice the absorbance of the zero analyte wells. A striking difference between this assay and our earlier assay with the same antibodies was that the cross-reaction of activin was reduced from 5.2% to just 0.05% (data not shown).
Two-site assay for large forms of the c~ subunit with the R 1 / C R C I antibody combination Fig. 2 shows the results of the first assay we ran with CRC1 and RI as the antibody pair. The specificity of the reaction was shown by the very strong reaction with inhibin forms in human FF, the absence of any reaction with bovine FF, and the weak reaction with post-menopausal human
human
inhibin
500 2[~0 125 62.,5 31.2!.;, 1,5.6 7.~{ 3.9
Fig. 3. Shows the results obtained when the E 4 / R 1 two-site assay was applied to recombinant 32 kDa inhibin (top row) and h u m a n FF (bottom row). FF dilutions start at 1 in 10 and reduce in ten-fold steps.
173 TABLE 11 DATA FOR HUMAN FF DILUTIONS IN THE RI/CRC1 TWO-SITE ASSAY Undiluted FF was assumed to contain 50,000 U/ml. Concentration of human FF (arbitrary units)
Absorbance
Mean absorbance
0 1.9 3.9 7.8 15.6 31.25 62.5 125 250 500
0.006, 0.005 0.024, (I.035 0.052, (I.054 0.105, 0.113 0.192, 0.235 0.347, 0.364 0.628, 0.652 1.05, 1.07 1.41, 1.29 1.53, 1.52
0.006 0.030 0.053 0.109 0.214 0.356 0.640 1.072 1.352 1.52
serum. Subsequent alterations to the procedure were made to improve sensitivity, resulting in the procedure described in the materials and meth-
ods section. Table II shows data obtained with our new assay format using serially diluted human FF (assumed to contain 50,000 U / m l ) as standards. 1 U / m i was readily detectable.
Recot;ery of dimeric 32 kDa inhibin spiked into human blood Table III shows some of the data obtained from the experiment described in which 32 kDa inhibin was added to human blood. There was no evidence for interference by serum, since the assays on each sample diluted 1/2, 1/4, 1/8, 1/16 and 1/32 gave almost identical estimates for sample recovery, which was complete, within the probable limits of the experimental accuracy (only one blood tube of each type was used). As can be seen from the table the immunoreactivity measured by this assay appears to be stable even in serum left in contact with the clot overnight.
TABLE IV
TABLE III
RECOVERY OF LARGE FORMS OF INHIBIN a SUBUNIT A F r E R SPIKING OF HUMAN BLOOD WITH A 1/100 DILUTION OF HUMAN FF
RECOVERY OF 32 kDa INHIBIN 1N SERUM AND PLASMA AFTER SPIKING INTO HUMAN BLOOD
Assay using R1/CRC two-site assay standard is human FF assumed to contain 50,000 U/ml.
Assay dilution Citrated plasma separated from cells and allowed to stand overnight at room temperature
1/2
Recovery %
Assay dilution
1/2
78
1/4 1/8
58
1/32
80 77 78 83 77
Heparinised plasma frozen soon after collection
1/2 1/4 1/8 1/16 1/32
112 122 109 103 108
Citrated plasma separated from cells and allowed to stand overnight at room temperature
1/2 1/4 1/8 1/16 1/32
17 13 14 17 19
Citrated plasma frozen soon after collection
1/2 1/4
Serum from clotted blood allowed to stand overnight at room temperature
1/32
74 81 79 90 90
t/2 1/4 1/8 1/16 1/32
22 31 39 45 53
1/2 1/4 1/8 1/16 1/32
1113 119 118 117 112
FF added directly to pooled post-menopausal serum which had been stored for a long period at - 8ff'C
1/2
61 74 72 81 89
1/4
1/8 1/16
l/8 1/16 Serum from clotted blood allowed to stand overnight at 4°C before assay
Citrated plasma frozen soon after collection
Recovery %
1/16 1/32
47 53 63
1/4 I/8 1/16 1/32
174 TABLE V TWO-SITE ASSAY OF HUMAN FOLLICULAR FLUID USING THE E 4 / R I ANTIBODY COMBINATION Data show the effect of pretreatment of the sample with 8 M urea in increasing the apparent amount of inhibin present. Final dilution
1/10 1/20 1/40 1/ 80 1/160 1/320
Apparent inhibin concentrations relative to 32 kDa recombinant standard p g / m l A Urea treated
B Untreated
442 682 386 150 70 33.9
156 85.4 48.7 26.3 13.1 -
Recot.,ery of large forms of a subunit immunoreactit.'ity spiked into blood Table IV shows some of the data obtained from the experiment described in which human FF was spiked into human blood and the resulting serum or plasma assayed in the R 1 / C R C 1 two-site assay. By contrast to the results for recombinant 32 kDa inhibin above, the recoveries were lower and more variable, and it appeared that serum and plasma gave different profiles. The best recovery was with heparinised plasma stored frozen until the time of assay, the worst was citrated plasma stored overnight at room temperature, with clotted serum giving intermediate recoveries. In the serum sample it appeared that recovery improved uniformly as the sample was diluted from 1 / 2 to 1/32. This could be accounted for if proteolysis of the sample continued during the immunoassay itself. Application of the immunoassays to human serum samples Preliminary experiments were carried out to assess the possibility that each of these assays might have sufficient sensitivity and specificity to measure inhibin forms in human serum. Three serum samples were assayed from women stimulated to achieve hyperovulation, and four samples were assayed from post-menopausal women. In the E 4 / R 1 assay all four samples from post-
menopausal women had undetectable levels of immunoreactivity (assumed to represent dimeric inhibin). The absorbance values recorded were indistinguishable from the zero analyte buffer blanks. This would correspond to < 4 p g / m l in relation to the recombinant 32 kDa inhibin standard. By contrast, the three samples from hyperovulated women had apparent levels of 84 p g / m l , 150 p g / m l and 178 p g / m l . In the R 1 / C R C 1 assay the post-menopausal samples contained apparent levels of 4 U / m l , 8 U / m l , 5 U / m l and 8 U / m l whilc the samples from hyperovulatcd women had levels of 41, 72 and 49 U / m l . In view of the poor recovery likely in the R 1 / C R C I assay with these serum samples it is certain that these figures significantly underestimate the true levels. Future studies will determine whether the addition of proteasc inhibitors to plasma improves recovery.
Some obsert.,ations on the use of the E 4 / R I assay to detect dimeric inhibin in FF Table V shows application of the E 4 / R 1 twosite assay to human FF with and without pretreatment of the FF with 8 M urea. The data showed that the apparent amount of dimeric inhibin in FF increased approximately six-fold after urea treatment.
Discussion When synthetic peptides are used for immunization to produce monoclonal antibodies reacting with native proteins it is vital to use the native molecule for screening the fusion supernatants. In the present work we wished to make a monoclonal antibody to the N-terminal portion of the 43 kDa human a inhibin, a form which was not available to us either in recombinant or purified natural form. Instead, we used our existing mouse monoclonal antibody to the 1-32 sequence of the 20 kDa a subunit to capture inhibin molecular forms from crude human FF in order to produce a suitable solid phase for E L I S A screening. Detection of rat antibodies to the N-terminal region of the 43 kDa inhibin a subunit was then achieved using a biotinylated mouse monoclonai antibody
175 specific for rat kappa light chains, followed by streptavidin/peroxidase and peroxidase substrate. The procedure worked well, and the only disadvantage was that the resulting r a t / m o u s e hybridoma cannot be grown to produce ascites in mice. In view of the trend towards growth of hybridomas in bioreactors this was not a significant problem. The above screening method is ideally suited to identifying antibodies which will be good components of two-site immunoassays, particularly if the capture antibody used in the screening (such as RI) is intended to be used in combination with the new antibody. It avoids the possible problem of alterations to the conformation of proteins which can take place when they are directly coated to plastic. Although we used the method because we had no purified 43 kDa inhibin to use for screening, there is no reason why it could not be used with purified native or recombinant material. Presently we do not know exactly which of the many forms of inhibin react in our R 1 / C R C 1 two-site assay. However, the sensitivity of detection can be approximated by the ability of the assay to detect inhibin forms in human FF at a dilution of 1 in 50,000. Knight et ai. (1991) showed that human FF contained around 2 ~.g/ml of total inhibin, expressed in relation to a 32 kDa inhibin standard. On this basis our assay would be detecting 40 p g / m l of inhibin. This calculation can only be approximate in the absence of knowledge about which molecular forms are present in human FF and which are measurable by our assay. Certainly the Rl/CRC1 two-site assay is very sensitive. One future application of this assay may be to monitor purification of large forms of the human inhibin a subunit, which can then be used as a standard. Of interest may be our finding that the R 1 / CRC1 assay appears to be capable of detecting large forms of the inhibin a subunit in serum from both hyperovulated and post-menopausal women although in greater amounts in the former. Future studies will need to improve the stability of the large forms measurable in this assay during serum collection and the assay itself, as well as other determinants affecting recovery. It seems likely that protease inhibitors may be
usefully added to tubes before sample collection for this assay. In the present work we also applied the same Fab alkaline phosphatase conjugate to the detection of dimeric inhibin, using our E4 antibody (to the fl-A subunit) for capture. A striking improvement was noted in the specificity of this assay compared to that observed in our previous work (Groome, 1991), when the E4 was used for detection. In the earlier work activin had shown a 5.2% cross-reaction in the inhibin assay. Furthermore, when we applied the original assay to serum from post-menopausal women a large (presumably non-specific) signal was observed (unpublished observations). These factors appeared to render the R 1 / E 4 assay unsuitable for use on serum. By contrast in the new assay, activin had only a 0.05% cross-reaction, and post-menopausal sera gave no signal under conditions where hyperovulated women gave a clear signal. These observations demonstrate that if the specificity of a two-site assay is not as great as desired it is well worth reversing the roles of label and capture antibody to seek improvement. The high recoveries shown here for the dimeric form of inhibin spiked into human blood, suggest that plasma and serum are equally suitable for this assay. We cannot yet be certain that some of the inhibin in blood is not masked in a way not readily simulated by spiking experiments, and that the true inhibin levels are being underestimated. Neither can we exclude, from the rather detailed recovery experiments done on one male serum, the possibility that other sera would give lower recoveries. The six-fold greater levels of detectable inhibin in human FF after urea treatment suggest that in this fluid masking of epitopes by proteins such as follistatin is extensive, and may lead to underestimates if assays are applied directly. It is not clear at present if the urea pretreatment releases all of the potentially reactive inhibin. The development of two-site immunoassays for different molecular molecular forms of inhibin is a very difficult challenge. The present study reinforces the value of the synthetic peptide approach in producing the necessary antibodies, and it is notable that the two assays described here have sensitivities in the same range as very sensitive assays for other molecules which use
176
conventionally prepared antibodies. The main features of the present inhibin assays, which we recommend to others, are as follows. (1) The stable attachment of solid phase antibodies to microplates by procedures is likely to retain a high density of functional antibody binding sites, as opposed to simple adsorption, which may release antibody when incubated with serum, and may also lead to impaired affinity or loss of antibody activity. (2) The assay diluent, containing very high concentrations of Triton X-100 and BSA, appears to simulate serum in the dimeric assay so well that serum samples can be diluted in this solution, thus removing the need for analyte-free serum to dilute standards. (3) The use of a long capture stage (24 h), combined with a high density of functional antibody (point 1), may help to achieve high recoveries from serum by allowing time for the inhibin to dissociate from binding proteins. (4) The use of an antibody Fab fragment for enzyme labelling is effective in minimising nonspecific binding of the label, and contributes directly to the high sensitivity observed. It is also likely to avoid problems with heterophilic antibodies in human sera, particularly when combined with an assay diluent containing 5% mouse serum (Jones et al., 1992). (5) The use of the amplified alkaline phosphatase assay contributes to the high sensitivity and convenience of the present assays. The possibility that the sensitivity can be further increased is being explored. Even in their present form the assays described appear to have the potential to be useful for a number of timely studies on inhibin.
Acknowledgements The authors wish to thank the Cancer Research Campaign for financial support, Derek Whiteley for art work, Barbara Southall for pho-
tography and Dr Ian Morris for the samples of serum from hypcrovulated women. Genentech generously provided recombinant inhibin and activin.
References Groome, N. (1991) Ultrasensitive two-site immunoassays for inhibin and activin using monoclonal antibodies raised to synthetic peptides. J. Immunol. Methods 145, 65. Groome, N. and Lawrence, M. (1991) Preparation of monoclonal antibodies reactive with the B-A subunit of human ovarian inhibin. Hybridoma 10. 309. Grt~)me. N., Hancock, J., Betteridge, A. and Craven, R. (1990) Monoclonal and polyclonal antibodies reactive with the 1-32 amino-terminal peptide of the a subunit of human 32 k inhibin. Hybridoma 9, 31. lshikawa, E., Imagawa, M., Hashida, S., Shinji, Y., Hamaguchi, Y. and Veno, T. (1983) Enzyme labelling of antibodies and their fragments for enzyme immunoassay and immunocytochemistry. J. Immunoassay 4, 209. Jones, S.L., Cox, J.C., Shepherd, J.M., Rothel, J., Wood, P.R. and Radford, A.J. (1992) Removal of false positive reactions from plasma in an enzyme immunoassay for bovine interferon-y. J. Immunol. Methods 155, 233. Knight, P.G., Beard, A.J., Wrathall, J.H.M. and Castillo, R.J. (1989) Evidence that the bovine ovary secretes large amounts of monomeric inhibin a subunit and its isolation from bovine FF.J. Mol. Endocrinol. 2, 189. Knight, P.G., Groome, N. and Beard, A.J. (1991) Development of a two-site immunoradiometric assay for dimeric inhibin using antibodies raised against synthetic peptides. J. Endocrinol. 129, R9. Knight, P.G., Muttikrishna, S., Groome, N. and Webley, G.E. (1992) Most of the radioimmunoassayable inhibin secreted by the corpus luteum of the common marmoset monkey is of a non-dimeric biologically inactive form. Biol. Reprod. 47, 554. Miyamoto, K., l-tasegawa, Y., Fukuda, M. and lgarashi, M. (1986) Demonstration of high molecular weight forms of inhibin in bovine FF (bFF) by using monoclonal antibodies to bFF 32 k inhibin. Biochem. Biophys. Res. Commun. 136, 1103. Schneyer, A.L., Mason, A.J., Burton. L.E., Ziegner, J.R. and Crowley, W.F. (1990) Immunoreactive inhibin a subunit in human serum: implications for radioimmunoassay. J. Clin. Endocrinol. Metab. 70, 1208. Stanley, C.J. Johannsson, A. and Self, C.H. (1985) Amplification can enhance the speed and sensitivity of immunoassays. J. Immunol. Methods 83, 89.
167
© 1993 Elsevier Science Publishers B.V. All rights reserved 0022-1759/93/$[16.00
JIM 06818
Immunoassays for inhibin and its subunits Further applications of the synthetic peptide approach Nigel G r o o m e a n d M a r t i n O ' B r i e n School of Biological and Molecular Sciences, Oxford Brookes Unit:ersity, Gipsy Lane, Headington, Oxford OX3 0BP, UK
(Received 30 March 1993, revised received 18 May 1993, accepted 28 May 1993)
We describe the preparation of a new rat monoclonal antibody (CRC1) to the N-terminal sequencc of the 43 kDa subunit of human ovarian inhibin, and its use together with other anti-peptide monoclonal antibodies, in two-site immunoassays for the detection of inhibin-related material in biological fluids. The Fab fraction of a mouse monoclonal antibody (R1) to the N-terminal portion of the 20 kDa a subunit, coupled to alkaline phosphatase, was used for detection, and either CRC1 or a monocional antibody (E4) to t h e / 3 - A subunit were used as capture antibodies. The E 4 / R I combination, expected to measure dimeric bioactive inhibin, could detect less than 2 p g / m l of recombinant inhibin in diluent, gave good recovery of activity spiked into human blood, and could measure significant levels of immunorcactivity in sera from women undergoing ovulation induction, and in some normal women. Scra from post-menopausal women contained undetectable levels. Apparent inhibin levels in human follicular fluid were increased six-fold by p r e t r e a t m e n t with 8 M urea, suggesting masking of epitopes in this fluid. Activin cross-reactivity in the assay was 0.05%. The R 1 / C R C 1 assay, expected to measure only large molecular weight forms of inhibin or its a subunit, could detect immunoreactivity in human FF diluted 50,000-fold, and in all sera tested, although the levels in the hyperovulated women were higher. By contrast to the E 4 / R 1 assay much of the immunoreactivity was labile during the clotting process, or subsequent assay, and reliable measurements on blood with this assay will require special samplc collection procedures. These results demonstrate the value of anti-peptide monoclonal antibodies in the study of inhibin, and the results obtained with CRC1 show that antibodies useful for immunoassays can somctimes be obtained without the purified target molecule being available for immunization or screening. Key words: lnhibin; Activin; Synthetic peptide; Monoclonal antibody; ELISA
Correspondence to: N.P. Groome, School of Biological
Sciences, Brookes University. Headington, Oxford OX3 OBP, UK. Tel.: +44-865-819255; Fax: +44-865-819928). AbbreL'iations: BSA, bovine serum albumin; ELISA, enzyme-linked immunoadsorbent assay; FF, follicular fluid; IVF, in vitro fertilization; MgCle, magnesium chloride; NaCI, sodium chloride; PBS, phosphate-buffered saline.
Introduction Inhibin subunits are known to be present in bovine and human FF in a wide range of molecular forms (Knight et al., 1989; Miyamoto et al., 1986). Some of these have inhibin bioactivity, others have activin activity, and some have no
168
known biological activity. Presently it is not possible to quantify many of these forms by specific immunoassays, and the development of such assays is a formidable challenge in view of the low concentration of these forms present in human serum and FF, and the limited availability of recombinant forms. It seems unlikely that a full understanding of the role of inhibin heterogeneity will be achieved without such assays. Inhibin assays specific for bioactive inhibin would be useful for numerous clinical studies to replace the existing radioimmunoassays, which are unable to distinguish dimeric inhibin from forms of the c~ subunit (Schneyer et al., 1990), and bioassays, which are tedious and prone to interference from other hormones. An essential step in immunoassay development is the preparation of antibodies of high specificity and affinity. Inhibin and activin have proved to be poor immunogens, possibly because they are closely conserved during evolution. Some investigators, including ourselves, have found synthetic peptide immunogens particularly useful (Groome et al., 1990; Groome and Lawrence, 1991). The advantages of the synthetic peptide approach are: (1) specificity is largely predetermined by the choice of peptide, thus avoiding the need for extensive epitope mapping; (2) coupling of peptides to carrier proteins before immunization can provide additional sites for helper T cell recognition, thus improving antibody production to the peptide; (3) it does not require large amounts of purified natural or recombinant material. Although some workers have viewed polyclonal antibodies raised to peptides as an alternative to monoclonal antibodies, we have concentrated on the latter in the hope of producing defined reagents for long term use. In the present paper we describe the immunization and screening procedures which have now enabled us to make a new rat monoclonal antibody reacting with the N-terminal sequence of the 43 kDa form of the human inhibin a subunit, a molecule which is not presently available in purified form. We show how this antibody, with others to the Nterminal portion of the 20,000 ot subunit (Groome et al., 1990) and the 82-114 sequence of the /3-A subunit (Groome and Lawrence, 1991), can be
used to configure very sensitive two-sitc enzyme immunoassays with alkaline phosphatase as the label. Preliminary data are reported which indicate that these assays should prove applicable to measurements on human blood samples. These studies reinforce our earlier claim that high affinity monoclonal antibodies suitable for use in ultrasensitive two-site immunoassays can readily be made using synthetic peptide immunogens (Groome, 1991). In the present report we show that purified forms of the intended analyte are not always an essential requirement for the development of such assays. Furthermore, in our original assay (Groome, 1991) activin showed a 5.2% cross-reaction and post-menopausal human serum (not expected to contain dimeric inhibin) gave a strong signal. In attempts to overcome these specificity problems we reversed the roles of the two antibodies used for capture and as label, with the great improvement in assay specificity reported here.
Materials and methods
Monoclonal antibodies The preparation of clone E4, making a mouse IgG2b antibody reacting with the inhibin /3-A subunit (Groome and Lawrence, 1991), and clone R1 making a mouse IgG2a antibody reacting with the inhibin a subunit (Groome et al., 1990), have been described elsewhere, as has their use in two-site immunoassays for recombinant human inhibin and activin (Groome, 1991). Clone CRC1, making a rat monoclonal antibody reactive with the N-terminal portion of the 43 kDa form of the human inhibin a subunit, was made using as an immunogen a synthetic peptide of sequence H A L G G F T H R G S E P E C . This particular sequence was chosen because N- and C-terminal peptides are often able to elicit antibodies which recognizc the parent protein. The peptide was coupled to tuberculin through the thiol group of cysteine by heterobifunctional chemistry as previously described (Groomc et al., 1990; Groome and Lawrence, 1991). Immunization of animals (Lewis rats) and fusion of spleen cells to the S P 2 / 0 myeloma line were carried out as in our previous work (Groome and Lawrence, 1991).
169
Identification of antibodies reacting with inhibin forms in human FF
pected to bind a variety of molecular forms of inhibin, including a subunits unassociated with a /3 subunit. To screen the individual supernatants from a fusion of rat spleen cells with the S P 2 / 0 myeloma cells 5 0 / z l of each supernatant were transferred into wells prepared as above and incubated for 2 h at room temperature with agitation. The microplates were washed and then incubated for 1 h with 50/~1 of a 5 / z g / m i solution of a biotinylated mouse monoclonal antibody specific for rat kappa light chains (Serotcc MCA 187B). After another wash, 50 /~1 of a 1 in 1000 dilution of streptavidin/peroxidase (Serotec STAR 5A) were added for 1 h at room temperature. After a final wash, 50 /~1 of TMB peroxidase substrate (Dynatech) were added for 30 min. From this screening process clone CRCI was identified as producing the most promising antibody. Subsequent isotyping with a Serotec kit revealed this to be a rat IgG2b antibody. The fact that this hybridoma was of rat × mouse origin meant that ascites production had to be performed in nude mice and purification done using protein G chromatography. Biotinylation was carried out as described previously
A major problem was anticipated when it became necessary to screen the fusion supernatants, because no purified preparation of 43 kDa inhibin was available. An indirect capture procedure was used for this purpose, which is why we used rats as the spleen cell source. Firstly, Nunc Immunoplates (type 1 with certificate) were coated overnight with 100 p.I/well of 0.1 M sodium bicarbonate pH 8.0 containing 1 0 / z g / m l of purified IgG from monoclonal R1. The following day the wells were blocked by incubation with 150 /~! of a 1% ( w / v ) solution of BSA (Sigma A3294) for 1 h at room temperature. A 1 in 10 dilution of pooled human FF (from an 1VF clinic) was made in a diluent consisting of 0.05 M sodium phosphate buffer containing 1.2% ( w / v ) NaCI, 0.5% (w/v) Tween 80 and 1% ( w / v ) BSA. 100/~1 of this solution were added to each well on the R1 coated microplates, and the plates incubated in the cold room for 4 h. The R1 antibody was known to be useful for immunoaffinity extraction of inhibin and its subunits from FF (Groome et al., 1990), and during this incubation it was ex-
Substrata Stre0tavii?
Colour
(6)
Substrata
Per°xidale
(5)
Streplavidin peroxidase
/ oo.
antibodies
(3) Rat in fusion supernatanls
g (1] R~
mouse anli ral kappa monoclonal
(4] B i o t i n y l a t e d
(MCA 187B) (2]
Inhibin forms captured from
human f o l l i c u l a r fluid
MAb
S C R E E N I N G T E C H N I Q U E S USED TO P R E P A R E MONOCLONAL ANTIBODY CRCl Fig. 1. Shows the principle of the ELISA screening assay used for preparation of the CRC1 monoclonal antibody.
170
(Groome et al., 1990). The screening procedure used to produce hybridoma CRC1 is shown in Fig. 1.
Two-site immunoassay for dimeric inhibin (using monoclonals E4 and RI) Two alternative procedures, using different commercially obtainable microplates, were employed to obtain a stable coat of the E4 antibody on the surface of microplates, whilst retaining the binding capacity of the antibody. These appear to give identical results in the inhibin immunoassays reported here. In the first of these, biotinylated E4 IgG was added to plates carrying covalently attached streptavidin (Bioproducts, Pias Grono Road, Wrexham, UK). In the second, plates carrying hydrazide groups were used to anchor the E4 IgG through the carbohydrate on its Fc region (Avidplate HZ, Bioprobe, Tustin, CA, USA). Streptavidin-coated microplates were first washed three times in distilled water and then coated with 50 ~1 of biotinylated E4 IgG at 2.5 p~g/ml overnight at room temperature. The diluent used for this stage was 1% ( w / v ) BSA in PBS. The following day the plate was washed with PBS, and excess sites on the polystyrene blocked by the addition of 150 p.I of 1% (w/v) BSA, 0.1% (w/v) sodium azide in PBS. Plates could be stored in this form for several months. To couple E4 lgG to Avidplates required prior oxidation of the antibody following the manufacturer's instructions. E4 IgG at a concentration of 1 m g / m l in 50 mM sodium acetate buffer pH 5.0, was mixed with 0.1 vol. of 100 mM sodium periodate and placed in the dark for 30 min. The reaction was stopped by the addition of 0.01 vol. of ethylene glycol, and the solution desaited into 50 mM sodium acetate buffer pH 5.0. The oxidised IgG (50 ~ l / w e l l ) was directly added to the wells of Avidplates at a concentration of 5 ~.g/mi, after dilution in 50 mM acetate buffer pH 5.0. The plates were left overnight at room temperature, and then blocked with 150 t~l of I% (w/v) BSA, 0.1 M Tris-HCl pH 7.5, 0.1% (w/v) sodium azide. Plates could be stored in this form for several months. Immediately before use the plate (streptavidin or hydrazide) was washed thoroughly with 0.05% ( w / v ) Tween 20 in 0.1 M Tris-HCI buffer pH 7.5
and blocked by the addition of 150 /zl of 1% (w/v) BSA in PBS. The plate was dried on paper towelling and to each well was added 100 /~l of recombinant human 32 kDa inhibin (Genentech) solution made as standards in an assay diluent containing 5% (w/v) Triton X-100, 10% (w/v) BSA, 5% (v/v) mouse serum in 0.1 M Tris-HCl buffer pH 7.5 containing 0.15 M sodium chloride. Samples containing unknown amount of inhibin were diluted in the same diluent. The plate was incubated overnight on a shaker at 4°C. The following day the plate was washed thoroughly and to each well was added 50 ~tl of a 1 ~ g / m l solution of the Fab fragment of clone R1 which had previously been coupled to alkaline phosphatase by heterobifunctional chemistry. The Fab fragment was prepared from purified IgG by standard pepsin digestion, the coupling to the enzyme as described previously (Ishikawa et al., 1983), and the final preparation was size fractionated to obtain forms with different molar ratios of enzyme and Fab. The fraction used in our studies has about 2 Fab molecules per alkaline phosphatase. After incubation for 1 h at room temperature on a shaker the plate was washed extensively over 15 min in wash solution and the wells emptied. Ultrasensitive detection of alkaline phosphatasc for the experiments reported here initially used a commercially available kit (AMPAK from Dako Diagnostics). 50 ~l of the substrate solution were added to each well and the plate was tightly covered and incubated for 2 h at 37°C. In order to extend the substratc reaction to this time it was necessary to make the kit substrate 1 mM in MgCl 2 (Dako personal communication). At the end of this time the plate was allowed to stand at room temperature for 15 min prior to adding the kit amplifier. The red colour development took place rapidly, and as soon as the zero analyte wells began to appear slightly red the reaction was stopped by the addition of 50 ~,1 of 0.4 M HCI. The absorbance of each well at 490 nm was read on a microplate reader (Biotek) with the reference wavelength at 620 rim, and data processing was done with associated software (Kineticalc). We have subsequently run these assays with in-house amplification reagents following published procedures (Stanley et al., 1985).
171
Exactly the same procedure was used as in the assay for dimeric inhibin except that the coated antibody was biotinylated CRC1. Since no relevant recombinant or purified standard for forms of inhibin containing large a subunits was available the standard used in this assay was a pool of human FF assigned an arbitrary potency of 50,000 U / m l . This FF was obtained from an IVF clinic.
dilution series started with a 1 / 2 dilution in the Triton assay buffer and further dilutions down to 1/32 were made in the same buffer. The recovery of inhibin in each sample was expressed as a percentage. For the purposes of calculating recoveries it was assumed that 50% of the blood volume was plasma. Thus a final concentration of inhibin in a plasma or serum of 2 n g / m l would correspond to 100% recovery of that added.
Localization of the epitopes for monoclonal RI and E4
Recovery of large forms of the inhibin et subunit from spiked human blood
Since monoclonal R1 was made by immunization with the 20 kDa a subunit 1-32 peptide sequence (Groome et al., 1990) and E4 with the /3-A 82-114 sequence (Groome and Lawrence, 1991) experiments were done to iocalise the reactive epitopes more closely. Overlapping 15mer synthetic peptides from the a and /3-A subunits were used in competition ELISA experiments to block the binding of each antibody to the corresponding long peptide immunogens adsorbed on to ELISA plates.
This was carried out as for dimeric inhibin except that the spiking was performed by adding 100 ~zl of pooled human FF (assumed to contain 50,000 U / m l ) to each tube, and this was finally measured in the R 1 / C R C 1 assay using as a standard the same FF preparation to make the standards. For this experiment the CRCI was attached to streptavidin plates. Full recovery of the added inhibin would thus correspond to 1000 U/ml.
Two-site assays for large forms of the a subunit (using monoclonals R1 and CRC1)
Recovery of dimeric inhibin from spiked human blood 10 ng of 32 kDa recombinant inhibin, dissolved in 0.I ml of 1% BSA, were added to different tubes for blood collection. The tubes included standard heparinised, citrated, and E D T A tubes for obtaining plasma, and a glass tube with no anticoagulant for obtaining serum. Immediately the inhibin was in the tubes one of the authors (N.P.G., male aged 45) was bled into the tubes up to the 10 ml mark. The samples were immediately mixed and returned to our laboratory. Within 30 min, preparations of heparinised, citrated and E D T A plasma were obtained. Each was divided into three aliquots for storage before assay: one aliquot frozen, one kept at 4°C, and the other kept at room temperature. The sample without anticoagulant was allowed to stand at 4°C overnight. The following day, preparations of serum and all the plasmas were assayed for inhibin in the E 4 / R 1 assay described earlier (hydrazide plates), with a standard concentrations made using the same preparation of inhibin which had earlier been spiked into the blood. Each
Effect of urea pretreatment of human FF on apparent inhibm levels A 1/10 dilution of human FF was made in either 8 M urea, 0.1 M Tris-HCl pH 7.5 buffer or the assay diluent referred to earlier. Both samples were incubated at room temperature for 10 min before further dilutions were made in assay diluent. The samples were then assayed in the E 4 / R I two-site assay using E4 coupled to an Avidplate.
Results
Epitopes recognized by the antibodies used here The R1 antibody reacted with only one of the overlapping 15mer peptides from the a subunit, corresponding to sequence MSWPWSPSALRLLQR. The epitope of CRC1 was obviously in the first 15 amino acids of the 43 kDa a subunit used for immunization. The inability of this antibody to react with bovine inhibin (Fig. 2) suggests that some of the three amino acids in this region which differ between human and bovine inhibin (positions 3, 7 and 8) are important for CRC1
172 . . . .
..
.~,.,.; ,~
-
--
~ -:
-
TABLE I D A T A F O R R E C O M B I N A N T 32 kDa INHIBIN STAND A R D S IN T H E E 4 / R 1 T W O - S I T E ASSAY
t:i
Concentration of 32 kDa inhibin
~,~,~-:~ '.:
0 :
:
.
.
.
-.
,:., :..... . . . .
Fig. 2. Shows the application of the R 1 / C R C I two-site assay to serial dilutions of h u m a n FF (top two rows), bovine FF (middle two rows), and post-menopausal serum (bottom two rows). T h e dilutions start at 1 in 2 and double across the plate.
recognition. The E4 antibody reacted with only one of the overlapping peptides from the /3-A subunit corresponding to sequence KLRPMSML Y Y D D G Q N . In competition ELISA experiments the corresponding peptide from the /3-B subunit of inhibin was almost as reactive (five-fold less). We believe that our two-site assay using this antibody would measure inhibin-B with similar effectiveness to the inhibin-A tested here, although no inhibin-B is available to confirm this.
Two-site assay for dimeric inhibin with the E4 / R1 antibody combination Fig. 3 shows the results of an assay in which our two-site assay with antibodies E4 and R1 was used to detect recombinant human 32 kDa in-
Picograms/ml
32kDa
Mean absorbance
0.049, 0.046 0.050, 0.048 0.081.0.092 0.125.0.128 0.182, 0.193 0.295, 0.323 0.448, 0.487 0.688, I).706 0.985, 1.040 1.18, 1.16
0.048
(pg/ml)
,
.
Absorbance
3.9 7.8 15.6 31.25 62.5 125 250 500
0.087 0.127 0.188 0.309 0.468 0.697 l.l)l 1.17
hibin. The 3.9 p g / m l standard could be easily distinguished from zero by visual inspection of the freshly developed plate. Raw data from such an assay are presented in Table I, from which it can be seen that the 4 p g / m l standard gave twice the absorbance of the zero analyte wells. A striking difference between this assay and our earlier assay with the same antibodies was that the cross-reaction of activin was reduced from 5.2% to just 0.05% (data not shown).
Two-site assay for large forms of the c~ subunit with the R 1 / C R C I antibody combination Fig. 2 shows the results of the first assay we ran with CRC1 and RI as the antibody pair. The specificity of the reaction was shown by the very strong reaction with inhibin forms in human FF, the absence of any reaction with bovine FF, and the weak reaction with post-menopausal human
human
inhibin
500 2[~0 125 62.,5 31.2!.;, 1,5.6 7.~{ 3.9
Fig. 3. Shows the results obtained when the E 4 / R 1 two-site assay was applied to recombinant 32 kDa inhibin (top row) and h u m a n FF (bottom row). FF dilutions start at 1 in 10 and reduce in ten-fold steps.
173 TABLE 11 DATA FOR HUMAN FF DILUTIONS IN THE RI/CRC1 TWO-SITE ASSAY Undiluted FF was assumed to contain 50,000 U/ml. Concentration of human FF (arbitrary units)
Absorbance
Mean absorbance
0 1.9 3.9 7.8 15.6 31.25 62.5 125 250 500
0.006, 0.005 0.024, (I.035 0.052, (I.054 0.105, 0.113 0.192, 0.235 0.347, 0.364 0.628, 0.652 1.05, 1.07 1.41, 1.29 1.53, 1.52
0.006 0.030 0.053 0.109 0.214 0.356 0.640 1.072 1.352 1.52
serum. Subsequent alterations to the procedure were made to improve sensitivity, resulting in the procedure described in the materials and meth-
ods section. Table II shows data obtained with our new assay format using serially diluted human FF (assumed to contain 50,000 U / m l ) as standards. 1 U / m i was readily detectable.
Recot;ery of dimeric 32 kDa inhibin spiked into human blood Table III shows some of the data obtained from the experiment described in which 32 kDa inhibin was added to human blood. There was no evidence for interference by serum, since the assays on each sample diluted 1/2, 1/4, 1/8, 1/16 and 1/32 gave almost identical estimates for sample recovery, which was complete, within the probable limits of the experimental accuracy (only one blood tube of each type was used). As can be seen from the table the immunoreactivity measured by this assay appears to be stable even in serum left in contact with the clot overnight.
TABLE IV
TABLE III
RECOVERY OF LARGE FORMS OF INHIBIN a SUBUNIT A F r E R SPIKING OF HUMAN BLOOD WITH A 1/100 DILUTION OF HUMAN FF
RECOVERY OF 32 kDa INHIBIN 1N SERUM AND PLASMA AFTER SPIKING INTO HUMAN BLOOD
Assay using R1/CRC two-site assay standard is human FF assumed to contain 50,000 U/ml.
Assay dilution Citrated plasma separated from cells and allowed to stand overnight at room temperature
1/2
Recovery %
Assay dilution
1/2
78
1/4 1/8
58
1/32
80 77 78 83 77
Heparinised plasma frozen soon after collection
1/2 1/4 1/8 1/16 1/32
112 122 109 103 108
Citrated plasma separated from cells and allowed to stand overnight at room temperature
1/2 1/4 1/8 1/16 1/32
17 13 14 17 19
Citrated plasma frozen soon after collection
1/2 1/4
Serum from clotted blood allowed to stand overnight at room temperature
1/32
74 81 79 90 90
t/2 1/4 1/8 1/16 1/32
22 31 39 45 53
1/2 1/4 1/8 1/16 1/32
1113 119 118 117 112
FF added directly to pooled post-menopausal serum which had been stored for a long period at - 8ff'C
1/2
61 74 72 81 89
1/4
1/8 1/16
l/8 1/16 Serum from clotted blood allowed to stand overnight at 4°C before assay
Citrated plasma frozen soon after collection
Recovery %
1/16 1/32
47 53 63
1/4 I/8 1/16 1/32
174 TABLE V TWO-SITE ASSAY OF HUMAN FOLLICULAR FLUID USING THE E 4 / R I ANTIBODY COMBINATION Data show the effect of pretreatment of the sample with 8 M urea in increasing the apparent amount of inhibin present. Final dilution
1/10 1/20 1/40 1/ 80 1/160 1/320
Apparent inhibin concentrations relative to 32 kDa recombinant standard p g / m l A Urea treated
B Untreated
442 682 386 150 70 33.9
156 85.4 48.7 26.3 13.1 -
Recot.,ery of large forms of a subunit immunoreactit.'ity spiked into blood Table IV shows some of the data obtained from the experiment described in which human FF was spiked into human blood and the resulting serum or plasma assayed in the R 1 / C R C 1 two-site assay. By contrast to the results for recombinant 32 kDa inhibin above, the recoveries were lower and more variable, and it appeared that serum and plasma gave different profiles. The best recovery was with heparinised plasma stored frozen until the time of assay, the worst was citrated plasma stored overnight at room temperature, with clotted serum giving intermediate recoveries. In the serum sample it appeared that recovery improved uniformly as the sample was diluted from 1 / 2 to 1/32. This could be accounted for if proteolysis of the sample continued during the immunoassay itself. Application of the immunoassays to human serum samples Preliminary experiments were carried out to assess the possibility that each of these assays might have sufficient sensitivity and specificity to measure inhibin forms in human serum. Three serum samples were assayed from women stimulated to achieve hyperovulation, and four samples were assayed from post-menopausal women. In the E 4 / R 1 assay all four samples from post-
menopausal women had undetectable levels of immunoreactivity (assumed to represent dimeric inhibin). The absorbance values recorded were indistinguishable from the zero analyte buffer blanks. This would correspond to < 4 p g / m l in relation to the recombinant 32 kDa inhibin standard. By contrast, the three samples from hyperovulated women had apparent levels of 84 p g / m l , 150 p g / m l and 178 p g / m l . In the R 1 / C R C 1 assay the post-menopausal samples contained apparent levels of 4 U / m l , 8 U / m l , 5 U / m l and 8 U / m l whilc the samples from hyperovulatcd women had levels of 41, 72 and 49 U / m l . In view of the poor recovery likely in the R 1 / C R C I assay with these serum samples it is certain that these figures significantly underestimate the true levels. Future studies will determine whether the addition of proteasc inhibitors to plasma improves recovery.
Some obsert.,ations on the use of the E 4 / R I assay to detect dimeric inhibin in FF Table V shows application of the E 4 / R 1 twosite assay to human FF with and without pretreatment of the FF with 8 M urea. The data showed that the apparent amount of dimeric inhibin in FF increased approximately six-fold after urea treatment.
Discussion When synthetic peptides are used for immunization to produce monoclonal antibodies reacting with native proteins it is vital to use the native molecule for screening the fusion supernatants. In the present work we wished to make a monoclonal antibody to the N-terminal portion of the 43 kDa human a inhibin, a form which was not available to us either in recombinant or purified natural form. Instead, we used our existing mouse monoclonal antibody to the 1-32 sequence of the 20 kDa a subunit to capture inhibin molecular forms from crude human FF in order to produce a suitable solid phase for E L I S A screening. Detection of rat antibodies to the N-terminal region of the 43 kDa inhibin a subunit was then achieved using a biotinylated mouse monoclonai antibody
175 specific for rat kappa light chains, followed by streptavidin/peroxidase and peroxidase substrate. The procedure worked well, and the only disadvantage was that the resulting r a t / m o u s e hybridoma cannot be grown to produce ascites in mice. In view of the trend towards growth of hybridomas in bioreactors this was not a significant problem. The above screening method is ideally suited to identifying antibodies which will be good components of two-site immunoassays, particularly if the capture antibody used in the screening (such as RI) is intended to be used in combination with the new antibody. It avoids the possible problem of alterations to the conformation of proteins which can take place when they are directly coated to plastic. Although we used the method because we had no purified 43 kDa inhibin to use for screening, there is no reason why it could not be used with purified native or recombinant material. Presently we do not know exactly which of the many forms of inhibin react in our R 1 / C R C 1 two-site assay. However, the sensitivity of detection can be approximated by the ability of the assay to detect inhibin forms in human FF at a dilution of 1 in 50,000. Knight et ai. (1991) showed that human FF contained around 2 ~.g/ml of total inhibin, expressed in relation to a 32 kDa inhibin standard. On this basis our assay would be detecting 40 p g / m l of inhibin. This calculation can only be approximate in the absence of knowledge about which molecular forms are present in human FF and which are measurable by our assay. Certainly the Rl/CRC1 two-site assay is very sensitive. One future application of this assay may be to monitor purification of large forms of the human inhibin a subunit, which can then be used as a standard. Of interest may be our finding that the R 1 / CRC1 assay appears to be capable of detecting large forms of the inhibin a subunit in serum from both hyperovulated and post-menopausal women although in greater amounts in the former. Future studies will need to improve the stability of the large forms measurable in this assay during serum collection and the assay itself, as well as other determinants affecting recovery. It seems likely that protease inhibitors may be
usefully added to tubes before sample collection for this assay. In the present work we also applied the same Fab alkaline phosphatase conjugate to the detection of dimeric inhibin, using our E4 antibody (to the fl-A subunit) for capture. A striking improvement was noted in the specificity of this assay compared to that observed in our previous work (Groome, 1991), when the E4 was used for detection. In the earlier work activin had shown a 5.2% cross-reaction in the inhibin assay. Furthermore, when we applied the original assay to serum from post-menopausal women a large (presumably non-specific) signal was observed (unpublished observations). These factors appeared to render the R 1 / E 4 assay unsuitable for use on serum. By contrast in the new assay, activin had only a 0.05% cross-reaction, and post-menopausal sera gave no signal under conditions where hyperovulated women gave a clear signal. These observations demonstrate that if the specificity of a two-site assay is not as great as desired it is well worth reversing the roles of label and capture antibody to seek improvement. The high recoveries shown here for the dimeric form of inhibin spiked into human blood, suggest that plasma and serum are equally suitable for this assay. We cannot yet be certain that some of the inhibin in blood is not masked in a way not readily simulated by spiking experiments, and that the true inhibin levels are being underestimated. Neither can we exclude, from the rather detailed recovery experiments done on one male serum, the possibility that other sera would give lower recoveries. The six-fold greater levels of detectable inhibin in human FF after urea treatment suggest that in this fluid masking of epitopes by proteins such as follistatin is extensive, and may lead to underestimates if assays are applied directly. It is not clear at present if the urea pretreatment releases all of the potentially reactive inhibin. The development of two-site immunoassays for different molecular molecular forms of inhibin is a very difficult challenge. The present study reinforces the value of the synthetic peptide approach in producing the necessary antibodies, and it is notable that the two assays described here have sensitivities in the same range as very sensitive assays for other molecules which use
176
conventionally prepared antibodies. The main features of the present inhibin assays, which we recommend to others, are as follows. (1) The stable attachment of solid phase antibodies to microplates by procedures is likely to retain a high density of functional antibody binding sites, as opposed to simple adsorption, which may release antibody when incubated with serum, and may also lead to impaired affinity or loss of antibody activity. (2) The assay diluent, containing very high concentrations of Triton X-100 and BSA, appears to simulate serum in the dimeric assay so well that serum samples can be diluted in this solution, thus removing the need for analyte-free serum to dilute standards. (3) The use of a long capture stage (24 h), combined with a high density of functional antibody (point 1), may help to achieve high recoveries from serum by allowing time for the inhibin to dissociate from binding proteins. (4) The use of an antibody Fab fragment for enzyme labelling is effective in minimising nonspecific binding of the label, and contributes directly to the high sensitivity observed. It is also likely to avoid problems with heterophilic antibodies in human sera, particularly when combined with an assay diluent containing 5% mouse serum (Jones et al., 1992). (5) The use of the amplified alkaline phosphatase assay contributes to the high sensitivity and convenience of the present assays. The possibility that the sensitivity can be further increased is being explored. Even in their present form the assays described appear to have the potential to be useful for a number of timely studies on inhibin.
Acknowledgements The authors wish to thank the Cancer Research Campaign for financial support, Derek Whiteley for art work, Barbara Southall for pho-
tography and Dr Ian Morris for the samples of serum from hypcrovulated women. Genentech generously provided recombinant inhibin and activin.
References Groome, N. (1991) Ultrasensitive two-site immunoassays for inhibin and activin using monoclonal antibodies raised to synthetic peptides. J. Immunol. Methods 145, 65. Groome, N. and Lawrence, M. (1991) Preparation of monoclonal antibodies reactive with the B-A subunit of human ovarian inhibin. Hybridoma 10. 309. Grt~)me. N., Hancock, J., Betteridge, A. and Craven, R. (1990) Monoclonal and polyclonal antibodies reactive with the 1-32 amino-terminal peptide of the a subunit of human 32 k inhibin. Hybridoma 9, 31. lshikawa, E., Imagawa, M., Hashida, S., Shinji, Y., Hamaguchi, Y. and Veno, T. (1983) Enzyme labelling of antibodies and their fragments for enzyme immunoassay and immunocytochemistry. J. Immunoassay 4, 209. Jones, S.L., Cox, J.C., Shepherd, J.M., Rothel, J., Wood, P.R. and Radford, A.J. (1992) Removal of false positive reactions from plasma in an enzyme immunoassay for bovine interferon-y. J. Immunol. Methods 155, 233. Knight, P.G., Beard, A.J., Wrathall, J.H.M. and Castillo, R.J. (1989) Evidence that the bovine ovary secretes large amounts of monomeric inhibin a subunit and its isolation from bovine FF.J. Mol. Endocrinol. 2, 189. Knight, P.G., Groome, N. and Beard, A.J. (1991) Development of a two-site immunoradiometric assay for dimeric inhibin using antibodies raised against synthetic peptides. J. Endocrinol. 129, R9. Knight, P.G., Muttikrishna, S., Groome, N. and Webley, G.E. (1992) Most of the radioimmunoassayable inhibin secreted by the corpus luteum of the common marmoset monkey is of a non-dimeric biologically inactive form. Biol. Reprod. 47, 554. Miyamoto, K., l-tasegawa, Y., Fukuda, M. and lgarashi, M. (1986) Demonstration of high molecular weight forms of inhibin in bovine FF (bFF) by using monoclonal antibodies to bFF 32 k inhibin. Biochem. Biophys. Res. Commun. 136, 1103. Schneyer, A.L., Mason, A.J., Burton. L.E., Ziegner, J.R. and Crowley, W.F. (1990) Immunoreactive inhibin a subunit in human serum: implications for radioimmunoassay. J. Clin. Endocrinol. Metab. 70, 1208. Stanley, C.J. Johannsson, A. and Self, C.H. (1985) Amplification can enhance the speed and sensitivity of immunoassays. J. Immunol. Methods 83, 89.