Journal of Immunological Methods, 151 (1992)67-86

67

© 1992ElsevierScience PublishersB.V, All rightsreserved0022-1759/92/$05.00

JIM06288

Multiple epitope interactions in the two-step sandwich immunoassay S. Amarasiri Fernando ~ and George S. Wilson Department of Chemistry, Unil~ersityof Kansas, Lawrence, KS 66045, USA

(Received5 August1991,revisedreceived6 January 1992,accepted 13January 1992) The 'hook" effect as related to the two-step sandwich immunoassay has been investigated experimentally and theoretically. The multiple epitope interactions between the analyte and the labeled antibody cause a 'hook' in the two-step sandwich immunoassay. Three different analytes and monoclonal antibodies were chosen to carefully demonstrate the effect of the analyte characteristics on this immunoassay. Two monoclonal antibodies against two different epitopes of biosynthetic human growth hormone (hOH) was the simplest model for this study. The sandwich immunoassay for hGH shows no 'hook' effect. The non-covalent dimeric form of hGH (D-hGH) possesses two repealing epitopes which is the simplest model for an analyte having a discrete number of repealing epitopes. The D-hGH assay demonstrated a 'hook' effect in the two-step sandwich immunoassay if the labeled antibody was allowed to interact with more than one epitope. In a third system multiple epitope interactions with the labeled antibody were observed using fcrritin. The effect of the analyte concentration and the liquid-phase antibody have been examined to elucidate the nature of these various interactions. The cause of the 'hook' effect in the two-step sandwich immunoassay is attributed to the desorption of the bound analyte due to a conformational change after the labeled antibody interacts with several epitopes of the adsorbed analyte. Key words: Two-stepsandwich immunoassay;Antibody;Analyte;Capture antibody:Labeled antibody;Epitope

Introduction The immunometric assay is increasingly recognized as a potentially impo~ant immunoassay technique (Miles and Hales, 1968). This assay offers several advantages over limited reagent Cornespondence to: G.S.Wilson,Department of Chemistry, Universityof Kansas. Lawrence, KS fi6045,USA. I Present address: Department of Pharmaceutical Chemistry, Universityof Kansas. Lawrence, KS 66045,USA. Abbreuiations: hGH,biosynthetichuman growthhormone; I~I-hGH, tzsl-labeledbiosynthetichuman growth hormone; D-hGH, human growthhormone dimer; GHC 101 and GHC 072, antiboman growth hormone monoclonal antibodies; QCI054 and FEF021,anli-ferrltlnmonoclonalant~'odies.

labeled assays (e.g., competitive binding assay, radioimmunoassay (RIA)) for analytes, including lower detection limits, higher specificity, wider working range and shorter incubation time. In a two-site immunometrie assay, the entity to be measured is 'sandwiched' between two antibodies which recognize either equivalent epitopes (symmetrical) or different epitopes (asymmetrical). One of the antibodies (capture antibody) is either covalently bound, or physically adsorbed on a solid-phase. In conventional, two-site immunometric assays, the antigen and the labeled antibody react sequentially with the capture antibody (Sevier et al., 1981). Excess reagent is removed before the addition of the next reagent.

fis The response is generated by the labeled antibody. and it is directly proportional to the aualyte concentration. One such 'sandwich type' immunometric assay (Woodhead et al., 1974) is the immunoradiometric assay (IRMA). Since IRMA was developed, the use of sandwich immunoassays has rapidly expanded, particularly in the area of clinical diagnosis for biologicaily active analytes (Gosling, 1990). One of the most serious analytical problems associated with the two-step sandwich immunoassay is the high dose "hook' effect - a paradoxical decrease in response at high analyte concentrations. Polyclonal antibodies have been commonly used as the solid-phase antibody in the two-step immunometrie assay (Miles et al., 1974). With the advent of monoclonal antibodies (K6hler and Milstein, 1975), an additional, simplified procedure was adopted involving two monoclonal antibodies directed against spatially distant antigenic sites (epitopes) on the same molecule (David et al., 1981). Analytes consisting of non-overlapping but repeating epitopes, however, permit a single monoclunal antibody to act concurrently and effectively as both capture and tracer antibody for the same assay (Chi et al., 1987). Rodbard and Feldman (1978) developed a theoretical model to optimize conditions for the two-step sandwich immunoassay which permits performance evaluation of the ideal assay system. Their model predicts the effects on the dose-response curve of random errors in the concentration of reagents, the rate constants for antigenantibody complex formation or the reaction time. It fails, however, to provide an explanation for the 'hook' effect, because the solid- and liquidphase antibody populations are assumed to be homogeneous. Other reports have also suggested improvements in models for IRMA (Rodbard and Lewald, 1970; Rodbard and Weiss, 1973). Studies were undertaken to evaluate the possible factors (Miles et al., 1974) which affect performance in sandwich immunoassays. According to the literature, analytes such as fel ritin (Casey et al., 1979) and human growth hormone (Miles, 1977) are initially captured by the polyclonal solid phase antibody but are released into solution during the second step of the assay. The immunological properties of ferritin and human growth

hormone differ widely yet behave similarly in these assays. Furthermore, their reports suggest that the 'hook' may be due to low affinity solidphase antibodie:, inadequate washing, insufficient amounts of !abeled antibody for the second step, or excessive incubation times. Rodbard et al. (1978) soon established a theoretical basis for the 'hook' effect using extended mathematical models based on two mechanisms. Their model suggests that the solid-phase antibody is heterogeneous and exhibits binding sites with differing affinity constants. Rapid dissociation of the analyte from the low affinity sites occurs during the second 'wash' period. This accentuates the 'hook' effect. Systems consisting of homogeneous solid-phase antibodies may still demonstrate a 'hook' effect if washing is incomplete after the first incubation with the antigen. The results obtained from theoretical studies on two-step immunometric assays suggested that a 'hooked' response could be obtained under conditions which do not involve low affinity antibodies (Ryall et al., 1982). This simulation of the 'hook' effect involved homogeneous antibodies of high binding affinity, but the concentration of labeled antibody added in the second assay step was insufficient. This model assumed that ferritin was lost from the surface and might have formed soluble complexes with labeled antibody during the second-step of the assay. The proposed model was specially designed for large molecules such as ferritin and some of the assumptions may not be applicable for analytes which do not possess mnltiple identical epitopes. According to their theoretical ealculations (Ryall et al., 1982) an additional population of low affinity antibodies on the solid-phase will demonstrate a 'hook' in the twostep immunoassay, as has been suggested by other reports. Ferritin has been used as the model antigen in experimental studies to investigate in detail the fundamental analytical problems in the two-step sandwich immunoassay (Perera and Worwood, 1982). Dual labeling provided a more detailed understanding of the probable reasons for the 'hook' in a sandwich assay involving a polyclonal solid-phase antibody. By monitoring the secondstep of the assay, it was reported that a significant amount of ferritin bound at the first-step

was lost dt;r;ag the second incubation. Another experiment, without the labeled antibody, demonstrated that 11)% of the bound ferritin was lost during the second-step of the assay. These data dearly show that the hook effect in a polyclonal system results ip part from heterogeneity of the capture antibody. The 'hook' was not eliminated in an assay developed using monoclonal capture antibodies. Moreover, the studies by Perera and Worwood (1982) indicated that nonspecific interactions should be eliminated to provide assays without 'hook" effects. However. no studies have investigated the influence of the characteristics of the analyte on 'hook' effects in two-step sandwich immunoassays. It is evident, therefore, that there are a number of possible causes of the 'hook' effect. As this phenomenon is clearly detrimental to reliable immunoassays, the conditions under which the effect can be observed must be delineated and its occurrence predicted. This study integrates the various theoretical approaches with experimental tests of analyte characteristics. To gain insight into the variables affecting the 'hook' effect, monoclonal antibodies have been used throughout for all experiments. The performance of the two-step sandwich immunoassay has been compared for different analyres to model the "hook' effect in terms of the charaeteris.~i~s of the analytes. Biosynthetic human growth hormone (hGH) has been selected as the simplest model to illustrate the binding behavior for an antigen having two different nonoverlapping epitopes. This protein is known to form an extremely stable non-covalent dimer (DhGH) which will serve as the simplest case of an analyte with two repeating epitopes. Ferritin was employed as a model for an antigen having mul'Apie epitopes. The influence of differing analytc and antibody concentrations, and non-specific interactions of the analyte with the solid-phase were all investigated to identify conditions under which the 'hook' effect may be observed. Kinetic studies were performed in order to explain the interactions that may release analyte from the solid-phase. All of the possible reasons previously reported for the "hook' effects in two-step sandwich immunoassay can be summarized as follows: low

affinity of the solid-phase antibodies, inadequate washing, insufficient amounts of labeled antibody. excessive incubation times and non-spccilic interactions. This study centers on the high dose 'hook' effect which can be exhibited under conditions of both excess and inadequate amounts of liquid-phase antibodies.

Experimental section Materials Biosynthetic human growth hormone (hGH) and purified dimeric hGH (D-hGH) were do. nated by Eli Lilly (Indianapolis, IN). The ~ttowing reagents were donated by Hybritech (San Diego, CA): monoclonal antibodies for hGH, GHC 072 and GHC 101, anti-ferritin antibod[~.~, FEF021 and QCI054 (F(ab')2). The 1251-FEF02[ and IzSI-GHC 101 (approximately i],,+ [.o..,1,,

q, [o,,.],,+ lad,,.],, =

The unbound analyte, H,) is washed away from the reaction medium as equilibrium is established. The concentration of the analyte washed away from the reaction medium is [H(t)]ll. Assuming R = [HQI(,)]0/[Hu)]0, a relationship can be developed which would then allow simulation of the binding curves for the first step of the two-site ]mmunometric assay (Ekins et al., 1968). Rz + AR+ B ~ 0 where A = K l ( h - q l ) - 1 and B = - q i K l . It is possible to write [Hu)]t~=h/(l + R), [Qt(,)]0 = R / K I and [HQIIj( ~= h R / ( 1 + R). Second assay step - reaction with the labeled antibody. The second step of the assay is de-

where H(, denotes the analyte lost from the immobilized antibody. As the analyte leaches from the solid-phase antibody, Qi, some capture antibodies will be unoccupied. The analyte is assumed to migrate from the solid-phase to react with the liquid-phase antibody primarily resulting in the formation of soluble complex HQ~:

[HO~,,]- K~[H.,][O~,,] Ka=nhk . It should be noted that all of the desorl-~d analyte might not form HQ~t ) complex. ~dso the unreacted labeled antibody is Q~tv If me analyte has more than two equivalent epitopes then the reaction scheme can be extended. Moreover several labeled antibodies could react with HQ* to form soluble complexes and Q~' HQ* is the simplest case. The reaction is shown below:

scribed in equation 6 as follows. [Q~HQ~,,] = K2[HQ~t,][Q~t,)]

HOti~)= O~t, # Q* HQII~, where HQ~(~) is the remaining bound analyte which was not involved in forming the complex, Q*HQtu r The excess or unreacted labeled antibody is denoted as Q~tv There is assumed to be no reaction between antibody Q~ and the solidphase. Under equilibrium conditions of the second step the following relationship can be developed.

(8)

The stoichiometric equilibrium constant for thc second step is K 2. The miclxascopic binding constant is k b. Again n b represents the number of repeating epitopes of the analyte. H, which can interact with Q* antibody. Some of the complexed analyte (HQI(~)) may dissociate from the solid-phase to form uncomplexed analyte, H(t I. The dissociation reaction can be symbolized as follows: I IQ I h), most of the antigen molecules are complexed in the first step of the assay protocol and hence the re ',lting response generating complex predominates in the second step. However the assay response is estimated to be a plateau at infinite dose since an insufficient concentration of labeled antibody is used. According to Fig. 3, the highest analyte concentration is 10 nM and the ~lid-phase antibody concentrations for curves F and G are 20 nM and 50 nM, respectively. Calculated results show that almost all analyte is bound to the solid phase (Qt) under the conditions used to generate curves F and G. As earlier predieted~ curves F and G could demonstrate a "hook" at high analyte concentration if the labeled antibody concentration is not increased selectively. According to model 2, in which the analytc has two repeating epitopes, high capacity solidphase antibodies should yield a plateau at high analyte concentrations. The effect o f labeled antibody concentration. The performance of two-step sandwich immunoassays can be controlled by selecting appropriate concentrations of labeled antibody. Theoretically, sufficiently high concentrations of la-

G,H,I 20

Is

D

2 4 6 ~t AnolyLe concentrotlon (n M)

10

Fig, 4. Simulateddose-response curvesshowingthe effect of the concentration of the labeled antibody(q~) on the 'hook' effect in the Iwo-step sandwich immunoassay. Theoretical parameters (model 2): KI = I nM-L, qt = 2.5 nM, q* (nM)= l0 (A). 5 (B), 2.5 (C). 1.0(D). 0.5 (F). 50 pM (GI, 10pM (H), 5 pM (l). Allalyteconcentralionrange(h) is 0.1-10 aM.

beled antibody are expected to give proportional amounts of signal generating complex even at higher analyte concentrations which would suppress the 'hook'. However the amounts of both solid- and liquid-phase antibodies should be carefully adjusted to generate a linear calibration curve which covers significantly higher analyte concentration in these assays. This study is limited to model 2 in which the effect of labeled antibody concentration was investigated for a 2.5 nM concentration of solid-phase antibody selected from Fig. 3 (curve C). Curve E in Fig. 4 was generated using parameters similar to curve C from Fig. 3. Curve E shows the 'hook' effect because the selected capture antibody concentration (qt) is greater than the labeled antibody concentration (q~). if the concentration of labeled antibody is decreased while keeping both solid-phase and analyte concentrations the same, the resulting calculated data show that the 'hook' shifts to slightly lower analyte concentrations. These data are shown in curves F-I, Fig. 4. The sharpness of the 'hook" increases as the amount of labeled antibody is decreased (curves F-I).

Moreover the positive slope (sensitivity) of these curves is increased. The labeled antibody is insufficient for the generation of curves F-I. Since some of the labeled antibody is also consumed in soluble complex formation, there will consequently be insufficient antibody for the formation of the signal generating sandwich complex, especially at higher analyte concentrations. Increased concentrations of labeled antibody should be sufficient for the formation of signal generating and soluble analyte complexes in the second step of the assay, although the assay response is diminished as the normalized responses are compared. Again note that curve D in Fig. 3 shows a 'hook' effect because q~ >q*. The concentrations qL and q~ are 2.5 nM and 1.0 nM respectively. Curves A - C in Fig. 4 do not show a 'hook' effect because the parameter ql is 2.5 nM while q~ varies from 2.5-10 nM (q~ ;~ q~). The normalized assay sensitivity is rapidly decreased as the concentration of the labeled antibody is raised, although the concentration of Q*PQ~(~) is amplified with respect to the labeled antibody concentration. According to this study one can avoid the 'hook' effect by maintaining q l > q ~ and the highest analyte concentration (h) for the calibration curve should be approximately 0.aqt (aM). A similar result has previously been reported by Ryall et al. (1982). The effect o f analyte concentration on assay response. The experimental results of hGH and D-hGH are presented together in order to compare and contrast the behavior of these two similar antigens in the immunoassay. The assay for hGH was designed using GHC 072 and GHC 101 as the solid- and the liquid-phase antibodies (GHC 0 7 2 / h G H / G H C 101 system). D-hGH was analogously assayed in two different modes. The labeled antibody was GHC 101 for both systems. These two systems are symbolized as GHC 072/ D - h G H / G H C 101 and GHC 101 / D - h G H / G H C 101. Assay for hGH - GHC 0 7 2 / h G H / G H C 10l sygtem Analytes such as hGH, which contain no repeating epitopes, are permitted to interact with two different antibodies so that the developed immunoassay for the analyte shows no 'hook'.

A

60

20

~0 ~ . ' 0 ' ' ' ..... i,.'~ . . . . . . . . Analyte c o n c e n t r a t i o n (nM)

;~'0

Fig. 5. T w ~ t e p sandwich immunoassayfor hGH and D-hGH

(wcr the low analyleconcentration range. Analytes:hGlt {t~), D-hGH (o); solid-phase antibody: GHC 072 1~,. t~): labeled antibody: GHC I1)1; 1251-GHCIfil a~ncentralinn: 1.22 nM (o), 0.81 (0]; incubationtime: first step (4 hi, second-step (3 hL Theorelical cuwes: cu~e A, D-hGll I•]: cu~e n, hG|l (n). Theoretical parameters: curve A (rondel I'L Kt-15 .aM I K,-0.48 nM i, q~-80 nM. q~ =0.81 aM; curve B (model4),K I 3n.aMl,K~-InM r, ql-2nM, q~-1.2 nM. Analyteconcentrafitm(p) range is 0.!IS-12nM.

Such an assay should generate a sandwich as defined in equations 1 and 2. The resulting doseresponse curve for hGH should be similar to curve A, Fig. 1 (model 4). Fig. 5 shows experimental data points and the standard linear curve for hGH in which the binding behavior can be described using equations 1 and 2. As the hGH concentration is increased, the response increases progressively as expected (Fig. 6). Moreover, Fig. 7 indicates the saturation of the response at high levels of hGH indicating the expected behavior of model 4. The combination of data in Figs. 5-7 comprise the hypothetical dose-response curve in Fig. 1 (curve A). The binding behavior of hGH is supported by mo0el 4 and the simulated curves for each figure are marked B. Model 4 can explain the binding behavior because the liquidphase interactions of individual antibodies (GHC 101 and GHC 072) with hGH permit only 1 : 1 or ] :2 complexes (Fernando et al., 1992). The experimental values for the affinities of GHC 072

78 (K~) a n d G H C 101 ( K 2) are 1.1 n M -I a n d 3.8 n M - ~ respectively ( S p o r t s m a n et al., 1989), T h e s e affinities have been d e t e r m i n e d while the antibody is in solution. T h e best fit theoretical values o f K~ a n d K~ are 36-fold a n d 4-fold lower t h a n the experimental values. T h e affinity of the solid-phase antibody used to fit the theoretical curve is considerably lower t h a n the e x p e r i m e n t a l value which may be a t t r i b u t e d to the effects of covalent i mmo b ilizatio n o f the antibody. It is n o t possible, however, to generalize the results f r o m a single m o d el analyte. Nevertheless if the twostep sandwich i m m u n o a s s a y is d e s i g n e d for analytes having different n o n - c o m p e t i n g epi t ope s using two different antibody protocols, no a m b i g u o u s results will be o b t a i n e d for the u n k n o w n sample. A r s a y J'or D . h G H 101 s y s t e m

-

GHC

101/D-hGH/GHC

D - h G H is the non-covalent d i m e r of h G H which has b e e n characterized for its chemical, lO0-

A

10oi 6o

go ~

F"

.:

A

°

C

t, L

~40i

] 2o i

0 . . . . . . . . 0.'g . . . . . . . i.'0 . . . . . . . 4.~ . . . . . . . ~.'6 . . . . . . . ~.'s

Anelyte cencentrotion (/~M)

Fig. 7. Two-step sandwich immunoassay for hGH and D-hGH over a high analyte concentration range. Analytes: hGH (~), D-hGH (¢>, z~); solid-phase antibodies: GHC 101 (o), GHC 072 (o, D); labeled antibody; GHC 101; 12~I-GHC 10l concentration: 1.22 nM (o}, 0.81 (,>, t=); incubation time: first step (4 h) second step (3 b). Theoretleal cuwes: curve A. D-hGH (O); curve B, hGH (D); curve C, D-hGH (4). Theoretical parameters: curve A (model 1), Ks=0.9 nM -0, Kz=P.48 aM- t q I = 80 nM, q ~ = 2.0 riM; curve B (model 4), KI - 30 t~M- I, K z - l n M - t, ql = 2 nM, q~ = l.2 nM; curve C (model

gG

3), Kt=60 /LM I, K2~30 ~M I q1=40 nM, q~=0.81 nM. Analyte concentration (p) range is 4.0 nM-2.2 p.M.

°

a

0

100 200 300 AnolyLe c o n c e n t r o t T o n

4flO (nM)

500

Fig. 6. Two-step sandwich immunoassay for hGH and D-hGH at moderate analyle concenlrations. Analyles: hGH ([]), DhGH (o. =); solid*phase antibodies: GHC 101 t) and GHC 072 (o. t~); labeled antibody: GHC 101: t2SI-GHC 101 concentralion: 1.22 nM (t~), 0.81 Co. a l; incubation time: first step (4 h) second step (3 h). Theoretical curves: curve A, D-hGH C); curve B. hGH Co); curve C, D-hGH ( ~ ). Theoretical paramelers: curve A (model !). KI=0.9 nM -t, Kz = 0.48 nM -I, q= = 811 nM, q~ = 1.'7 nM; curve B (model 4), K t = 30/zM - I, K~ = 1 nM - =. q i = 3 riM. q~ = 1.2 riM. Analyt¢ concentration Cp) range is 3.0-0.45 IzM.

physical a n d biological p r o p e r t i e s (Becker et al., 1987) a n d serves as a mod e l for two k n o w n rep e a t i n g epitopes. T h e antibody c h o s e n for the solid- or liquid-phase s h o u l d contribute some additional selectivity to the b i n d i n g r e s p o n s e in the sandwich immunoassay. T h e r e p e a t i n g e p i t o p e s of D - h G H p e r m i t a single antibody (either G H C 101 or G H C 072) to f o r m a sandwich comp!cx. T h e sandwich i m m u n o a s s a y d e s i g n e d for D - h G H used G H C 101 as t h e solid- a n d liquid-phase antibodies in different concentration ranges. A s D - h G H has only two e pi t o p e s to interact with G H C 101, the d o s e - r e s p o n s e curve s h o u l d be similar to h G H . T h e d a t a are s h o w n in Fig. 7 for the high D - h G H c o n c e n t r a t i o n range. Curve C r e p r e s e n t s the s i m u l a t e d b i n d i n g d a t a for Fig. 7. S i m u l a t e d curve C was g e n e r a t e d using m o d e l 3. T h e b i n d i n g curves are similar to the data obtained for h G H high analyte c o n c e n t r a t i o n (Fig. 7). D a t a for D - h G H correlate with the hypotheti-

cal curve in Fig. 1. Model 3 does not fit to the experimental data in the moderate analyte concentration range which may be attributed to the difference in affinity of the solid- and liquid-phase antibodies (data not shown). This experiment suggests that a single antibody (GHC 101) can interact with both non-competing epitopes of DhGH to generate a dose-response curve without a 'hook' effect. Therefore if a two-step sandwich immunoassay is designed for an analyte having only two repeating epitopes which interact with a single antibody, one can predict a lack of aberrant results for the unknown sample. Assay for D-hGH - GHC 0 7 2 / D - h G H / G H C 101 system Alternatively a similar two-step sandwich immunoassay can be developed by selecting two antibodies for different epitopes of D-hGH. Theoretically, results should be similar to the case involving a single antibody, as discussed previously. This sandwich immunoassay is analogous to the assay developed for hGH, except for the analyte change (Figs. 5-7). The binding reactions can be explained using model 1, in which two epitopes of D-hGH are accessible for the interaction with the liquid-phase antibody. Note that the analytes in previous assays are not permitted to have multiple interactions with the labeled anti body. The two-step sandwich immunoassay was again constructed using G H C 072 and G H C 101 as the solid- and liquid-phase antibodies, respectively. At very low concentrations of D-hGH (below 12 riM), the response increases with the rise in concentration of D-hGH as shown in Fig. 5. The response obtained at higher D-hgI--I concentrations declined, however, resulting in a "hook' fFig. 6) which further decreased and maintained a plateau at infinite D-hGH concentrations (Fig. 7). It can be seen that curve A in Fig. 6 also exhibited a sharp 'hook" in the theoretical doseresponse curve. The theoretical curves can be fitted to the experimental data in Figs. 6 and 7 if the binding parameters, K 1 and q~ are changed 60-fold and 2-fold respectively, with respect to the values used to generate curve A in Fig. 5. The combination of Figs. 5, 6 and 7 should result in the hypothetical 'hooked' curve from Fig. 1 (curve B). The theory and experiments in Figs. 6 and 7

suggest that the descending limb of the 'hook' is not directly explained by assuming that D-hGH comes away from the solid-phase and forms soluble liquid-phase complexes at high analyte concentrations (model 1). The above resnlts clearly indicate that the assay performance is mostly determined by the characteristics of the analytc. Comparing the data for hGH with D-hGH permits the following conclusions to be drawn. For hGH, with no repeating epitopes, the 'hook' effect can be avoided by using two different antibodies. For D-hGH, using a single antibody avoids the 'hook' effect, but two different antibodies promote formation of the 'hook' at high concentrations of the analyte. If the analyte is capable of forming multiple adducts with several liquidphase antibodies at high analyte concentrations, a plateau will not result in the dose-response curve (Fig. 6). Moreover, at high analyte concentrations, more analytes will react with the solid-phase antibody in the first step of the sandwich immunoassay so that the labeled antibody might react in a random fashion in the second step, resulting in multiple interactions with some analytes. Multiple interactions of the solid-phase analyte and the liquid-phase antibody, may result in conformational changes of the analyte due to steric effects. Thus, it could preferentially lead to weakening of the interactions of the analyte with the solid-phase antibody. Therefore, the bound antigen may easily be released from the solidphase, resulting in a decrease in assay response at high analyte concentration. This effect could not be detected at low analyte concentrations. In the case of D-hGH, only two antibody molecules per desorbing analyte are released that can be detected with respect to the liquid-phase antibody since this is the only ~abeled antibody in the immunoassay. As the desorption of the analyte is determined in terms of the labeled antibody, the labeling efficiency could also contribute to such effects. A~say for ferritin - chemtcally immobilized antibodies Two different systems were studied in this section. The filet assay system consisted of a single antibody as the solid- and liquid-phase antibody (FEF021/hs-ferritin/FEF021). Two

too!

antibodies, QCl(154 as the solid-phase antibody and FEF021 as the liquid-phase antibody were used for the second system (QCI054/hs-ferritin/ FEF021). The experimental results are presented together for both systems.

o ..am___

go

C

FEF021/h.~.ferr#in /FEF021 and QCI054/hsferritin / FEF021 systems According to the studies of the D-hGH system, ferritin should also exhibit a 'hooked' response due to multiple interactions with the liquid-phase antibody in the two-step sandwich immunoassay. This phenomenon can be observed for ferritin as an analyte because it consists of several identical binding sites. If the capture antibody is physically adsorbed, the signal generating complex (e.g., Q~HQ I) or solid-phase antibody might desnrb from the plastic support leading to a more complicated assay result, in order to rule out the possibility that the capture antibody might desorb during the reaction, beads with anti-ferritin antibody covalently attached were employed. To design a sandwich immunoassay in which the solidphase antibody is covalently attached to the plastic beads (c-FEF021), as described in model 2, an individual monoclonal antibody against ferritin was chosen for both the solid- and liquid-phase antibody. The second assay was designed by replacing c-FEF021 with c-QCI054 (covalently attached QC1054 to the plastic beads) which is comparable to model 1. The labeled antibody for both assays was FEF021. Both assays were performed over three different concentration ranges (low, moderate and high) and the resulting data are shown in Figs. 8-10. Figs. 8 and 9 illustrate three sets of data with corresponding theoretical curves. Curve A represents the theoretical curve of model 2 for the c-FEF021 system. Curves B and C show the simulated curves of model 1 for e-QCI054 system. Fig. l0 shows four sets of data and relevant computer generated curves. Theoretical curves A and B represent the c-FEF021 system while curves C and D represent the cQCI054 system, respectively. All the binding parameters are described in the relevant legends. The affinity constant of FEF021 is 56 nM -~ and QC!054 has an affinity in the same range. To understand the effect of the amount of the labeled antibody, these assays were performed at different concentrations of the liquid-phase anti-

A

6O

m ~

40

20

....... g66 ...... ~ i~'o'~..... ~ ~'66'"" i~'o'~'" ' 'i.~bo

hs--ferriLin c o n c e n t r a t l o n

(pM)

Fig. 8. Two-step sandwich immunoa~say for hs-ferritin (wer a /ow analyte concentration range, Solid-phase antibodies: FEFfl21 (o), QC1054 (

Multiple epitope interactions in the two-step sandwich immunoassay.

The 'hook' effect as related to the two-step sandwich immunoassay has been investigated experimentally and theoretically. The multiple epitope interac...
1MB Sizes 0 Downloads 0 Views