Modification of the ELISA for the estimation of tetanus antitoxin in human sera

Journal of Biologrcal Standardization

(1987) 15, 143-157

Modification of the ELISA for the estimation of tetanus antitoxin in human sera*

0. Simonsen,? C. Scbout and 1. Heron?

The use of indirect ELISA for the quantitation of tetanus toxin neutralizing antibodies in human sera is limited by marked overestimations in low ritered sera. The reasons for the discrepancy between the results obtained by ELISA and by in vivo assay and modifications of the ELISA ro overcome the problem were investigated. Catching ELISA and indirect ELISA using trays coated with the contaminant proteins in toxoid preparations indicated that antibodies to contaminants were only partly responsible for the discrepancy and the introduction of these modifications did not solve the problem. In ELISA competition experiments with toxin neutralizing monoclonal antibodies, rhe human immunoglobulins irrelevant in toxin neutralization, but detectable in indirect ELISA, were found to be difficult to inhibit in their binding to the solid antigen phase. These might represent antitoxins bound bivalently to the solid phase but with affinities in monovalenr binding insufficient for toxin neutralization or other coupled antibodies due to conformational changes of the antigen. A competition ELISA with toxin in solution was therefore developed to assessselectively the antitoxin capable of binding the antigen in solution and by this approach the in vivo activities of even low titered sera were accurately predicted. This antigen competition ELISA may be easily introduced into routine tetanus serology and the principle may also be of value for the in IWW~ detection of functional antibodies to other antigens.

INTRODUCTION The determination of tetanus antitoxin in human serum samples may conveniently be carried out by enzyme linked immunosorbent assay (ELISA). In a previous paper’ serum samples from a number of subjects with different vaccination histories and of * Received for publication 7 August 1986. t Vaccine Department, State Seruminstitute, 0092-l

157/87/020143+

15 $03.00/O

Amager Boulevard 80, 2300 Copenhagen S, Denmark.

@ 1987 The International

Assocmion

of Biological

Staodardizarmn

143

0. SIMONSEN ET AL.

different ages were investigated using ELISA as well as a toxin neutraIitation method in mice. Sera containing high and intermediate levels of neutralizing tetanus antitoxin showed a satisfactory correlation between ELISA and neutralization assay. If, however, ELISA results were used to predict protective antibody levels in sera from subjects with incomplete vaccination histories or subjects whose primary vaccination had been completed many years earlier, a difficulty was encountered in the form of a very poor correlation between ELISA and in uivo titres. Neutralizing activity in samples with antitoxin concentrations below 0.16 IU ml- ’ could not be predicted safely from the ELISA results. In many samples the ELISA value markedly overestimated the neutralizing antitoxin content. The extent of this discrepancy was inversely proportional to the neutralizing antitoxin content of the samples. This phenomenon has also been observed by other investigators who compared ELISA and in Gvo results for low titered sera’ and a similar discrepancy was encountered when hemagglutination assay and in uivo assay results were compared.’ The present investigation was initiated to study closely such low titered antisera in ELISA and to carry out modifications of the in vitro system that would provide an explanation for the poor correlation and thus lead to improvements in the system. The strategy was to focus on (a) the basic parameters of the indirect ELISA, (b) the importance of the purity of the tetanus antigens employed in ELISA, and (c) refined assessments of antibody binding including Ig class specificity and competition methods. MATERIALS

AND

METHODS

Serum samples

The target group of sera for the studies were sera with antitoxin concentrations below 0.2 IU ml- *, below which level disagreement between ELISA and in vivo assay results is observed. i From a study of tetanus antitoxic immunity in the Danish population, 100 sera in this category were randomly selected and 15 sera containing 0.2-5 IU ml- i were considered for comparison. In vivo neutralization

assay

The neutralization method described by Ibsen” was used and the neutralizing activity in serum was expressed in antitoxin activity units per milliliter (UrnI-‘).’ Routine enzyme linked immunosorbent assay

Flat-bottomed microtiter trays with 96 wells (Nunc immunoplates) were used throughout. Tetanus toxoid with a purity about 2000 Lf mg- ’ of protein-N was used as the antigen and added at a concentration of O-075 Lfml- ’ in carbonate buffered saline (pH 9.5) during the antigen coating step. The buffer used for all washing procedures was phosphate buffered saline (PBS) containing 1% Triton X 100. Buffer for blocking as well as for the dilution of serum and conjugates was identical with the washing buffer apart from the addition of 2% bovine serum albumin. Human sera were prepared in serial twofold dilution steps. On each plate a local human standard serum with a known tetanus antitoxin concentration (4 IU ml- ‘) was titrated in paraIle1 with the test sera. Antibodies were allowed to bind for I h at room temperature. After three washing cycles, 100 ~1 of horse radish peroxidase conjugated to rabbit anti-human immunoglobulin (DAKOPATTS, Denmark, code no. P2 12) 144

MODIFICATION

OF THE

ELBA

OF TETANUS

ANTITOXIN

diluted 1:5000 was added and the mixture was held for 1 h at room temperature. After washing the substrate cromogen (100 pl), o-phenylenediamine (OPD), was added. The reaction was stopped after 30 min by addition of H$O* (1 N, 150 pl), and the optical densities (OD) were read at 490nm (NJ 2000 Immunoreader, TEKNUNC, Denmark). Antigens and antibodies Tetanus toxin obtained from the production unit of the department was used. The purity was about 2200 Lf mg- ’ of protein-N. The contaminant proteins present in the preparation were kindly provided by Dr C. Schou of the Vaccine Department and had been prepared by the passage of unpurified tetanus toxin preparations though several sepharose 4B-columns coupled to monocolonal antibodies against tetanus toxin. The contaminant preparations contained no tetanus toxin protein as assessed in crossed immunoelectrophoresis with polyclonal horse tetanus antitoxin. The protein content was adjusted to be identical with that of the tetanus toxoid used in the routine ELISA. A panel of monoclonal mouse antibodies (MoAb) against tetanus toxin/toxoid was available (manuscript in preparation). Six of these were found to neutralize tetanus toxin in uivo when mixed with the toxin and injected subcutaneously into mice. These six monoclonal antibody preparations were obtained in the form ofculture supernatants and mixed together in dilutions according to their relative ELISA titres to yield a monospecific polyclonal hybridoma antibody reagent. The mixture was found to bind in the ELISA at a dilution of 10p4. Some of the antibodies were purified using protein A affinity chromatography and the purified IgG was dialysed against PBS (pH 7.5) and used for coating ELISA plates in catching ELISA. Peroxidase conjugated rabbit anti-mouse Ig (DAKOPATTS Code no. P260) was used at a dilution of 1:500 for indirect ELISA to estimate the binding of monoclonal antibodies. In the antibody competition ELISA with human antibodies, the rabbit anti-mouse Ig conjugate was used in a form from which human cross reactive specificities had been removed by absorption with sepharose 4B coupled with irrelevant mouse monoclonal antibodies. Peroxidase conjugated rabbit anti-human IgG (DAKOPATTS Code no. P2 14) was used in the IgG specific indirect ELISA. The reagent was used at a dilution of 1:5000.

STATISTICAL

METHODS

The antitoxin concentration of a serum assayed by ELISA was calculated relative to the concentration in the standard serum, titrated in parallel, by standard statistical methods using the linear parts of the titration curves.’ For two parallel lines each derived from twofold dilution series the corresponding relative potency, R, between the sera was obtained from:

where x is the horizontal distance between the two lines. In competition experiments the titration curves for a serum with and without competitor were compared, and the competition expressed by: Gdl~cw~ttlout = lAX

the

0. SIMONSEN

ET AL.

were the antitoxin concentrations determined in ELISA with where Cwith and Cwichout and without competitor, respectively (Fig. 1). The concentration of antitoxin which was inhibited was calculated according to the following: C wlrh

-

c wthout

=

cwlthour

x

(1

-

2’,)

k-see%+ log2 dilution

Fig. 1. competition

Titration curves in competition ELISA without (*) and with (0) competitor. is calculated from the horizontal distance between the two lines.

Percentage

RESULTS The basic problem of the poor correlation between neutralizing activities determined in mice and antitoxin concentrations determined by ELBA can be seen in Fig. 2. It will be noted that the lack of correlation between the methods is more pronounced at the lower levels of neutralizing activity. Changes of basic parameters of the ELISA

For the methodological studies of the routine ELBA three groups each of four sera were selected at random. Group I with in viva titers of less than 0*00005 Urn I- ‘, group II with titers 0.00 l-O*005 U ml- i and group III with titers of00 l-0.5 U ml-‘. The effect of the toxoid concentration in the coating procedure, the incubation time with the human serum samples, and the concentration of the peroxidase conjugated rabbit antibody were studied. The human serum samples were,titrated in plates in parallel with the standard serum and the potencies of the individual sera were calculated relative to the standard under the various conditions. Table 1 gives a summary of these results. A figure of I.00 in the table means that the group of sera under the indicated 146

MODIFICATION

OF THE

ELBA

0~001 -1”: . I l

.

ANTITOXIN

:

. .

.

OF TETANUS

.

.

0~0001

c..

+-.%

%k--

I 0.1

Routine ELISA

I I

(IU ml-‘)

Fig. 2. Comparison of tetanus antitoxin concentration assessed in indirect ELISA and neurralizing activity assessed in vivo (115 sera). Line of identity is given.

conditions had estimated potencies which, on average, were identical with the potencies calculated in the routine ELBA. Results above 1.0 indicate relative potencies higher than those obtained by the routine methodology. Changes in the incubation period from 1 h to shorter times gave slight increases in the relative potencies whereas longer incubation periods had no effect. Changing the antigen concentration by coating with less or more antigen than in the routine ELBA had an effect particularly on sera in groups I and II. The problem of ELBA overestimation became more marked the lower the protective antibody concentration and the higher or lower the antigen coating. Increases in the concentration of the peroxidase conjugated secondary antibody gave a pronounced increase in the OD values obtained but when potencies were calculated under these conditions none of the groups of sera was found to benefit by changing this parameter. On the contrary a certain increase in relative potency was observed subsequent to this amplification of the system. The reliability of the ELISA was investigated in repetition experiments. When titrations were repeated six times on the same plate the variance of the log OD for one dilution was found to be 0.00 11, thus the 95% confidence limits of a re!ative potency, R, calculated by the parallel-line approach, were (R/l* 14, R X 1.14). If the relative potency had been calculated from one dilution the 95% confidence limits would have been (R/2*20, R X 2.20).

147

0. SIMONSEN

E7’ AI.

1. A comparison of results of ELBA under various conditions relative to the results obtained under routine conditions (indicated by bold-faced type) in three groups of Sef% with different neutralizing activities (Group I mean = 0~00005 UrnI-‘, Group II mean = 0.004 Ll ml-‘, Gro up III mean = 0.3 LJ ml- ‘). The inhuence of the various conditions on OD levels is illustrated by the OD of the standard serum diluted l/1000 (neutralizing activity = 0.004 U ml ‘) TABLE

Mean values of relative potencies determined by ELISA ELISA conditions -___-~~ ~~-~. ~-~ ~~~ Incubation time (min) 5 30 60 120 Coating (antigen concentration, 7.5 0.75 0.075 0.0075 Conjugate (dilution) l/500 l/100 l/5000 I/ 10000

Group I

Group II

Group III

OD for the standard serum

1.88 1.22 I.00 1.oo

1.10 I40 I.00 1.00

1.28 1.08 1.00 1.oo

0.24 0.38 0.47 0.47

6.67 2.63 1.00 4.27

2.57 1.72 I.00 2.91

I.56 I.42 I.00 0.9 I

0.58 0.58 0.47 0. I6

2.10 1.77 I.00 0.92

2.04 1.75 I.00 1.15

1.10 I .09 1.00 0.54

2.20 1.30 0.47 0. 17

~~ ~~

Lf ml- ‘)

The effect of tray variation was found to be only slight. When titrations were repeated on six different plates the 95% confidence limits for a relative potency were (R/1-20, R X 1.20). Catching ELBA Monoclonal antibodies against tetanus toxin were employed as catching reagents in microtiter plates which were subsequently incubated with tetanus toxin or tetanus toxoid. Thirty sera with toxin neutralizing activities between O-00004 and 0.008 Uml-’ were titrated using these ELISA plates and the potencies of the sera were calculated relative to the standard serum titrated in parallel (Table 2). The potencies assessedby the catching methods were on average, 57% of those determined in the routine ELBA and nearly identical with the toxin and toxoid antigens. The ELBA scores on plates coated with the proteins present as contaminants in tetanus toxin were determined using the same serum samples. The absorbance obtained at different serum dilutions on these plates was subtracted from the absorbance figure obtained by routine ELBA plates and these corrected absorbances were used for calculating potencies relative to the standard. Table 2 gives the figures from these corrected values and it appears that, on average, 45% of the potency calculated from the routine ELBA was obtained. This reduction in potency was of the same order of magnitude as that obtained by catching ELISA. 148

MODIFICATION

OF THE ELISA OF TETANUS

ANTITOXIN

TABLE 2. A comparison of results from 30 sera with low in viva activities (0.008 U ml-‘) when assayed by routine ELISA, in catching ELISA with toxin or toxoid as antigen and in ELISA with correction for binding to contaminant proteins

Routine ELBA Loglo antitoxin

concentration

Mean SD

Catching ELBA ~~ Toxin Toxoid

-2.027 0.280

-2-280 0.37 1

04094 100

0.0053 56

Mean antitoxin concentration (W/ml) Mean values relative to routine ELISA (%)

-2.262 0.369 0.0055 59

Corrected ELBA -2.180 O-46, 0~00~42 15

To assess the practical value of catching ELISA the 115 sera were investigated (Fig. 3). It appeared that the problem of the lack of correlation between in viva and in vitro potencies was only slightly diminished and sera which in vivo contained less than 0.0 1 U ml- * (on average 0*0002 U ml- ‘) and in the routine ELISA contained 0.0 19 IU ml- ’ on average, in the catching system scored an average of 0.0 10 IU ml- ’ (range 0*003-0.05 IU ml-‘). The introduction of the catching ELBA therefore did not solve the problem of the marked overestimation of sera with low neutralizing activities.

I 0.01 Catching

I I

I 0.1 ELISA

(IU ml-‘)

Fig. 3. A comparison of tetanus antitoxin concentration assessedin catching ELBA and neutralizing activity assessedin viva (115 sera). Line of identity is given. 149

0. SIMONSEN

ET A,!,.

IgG class spec& ELISA

Taking advantage of the knowledge that only IgG antibodies possesstoxin neutralizing activity in vtvos the panel of the 1 15 sera was tested in ELISA employing a specific anti-IgG conjugate. From the potencies obtained in routine ELISA compared to the potencies in the IgG specific ELISA (Fig. 4) it was obvious that this change in the detection system had no effect on the potency estimation of the sera tested.

L -E

. Ii

. .,!

. . . . .. :.,l. -$-r

. -.

‘2 *.*.:. .. ...+ . ..“. . .I’ . .: ;

I o-01

I 0.1 Routine ELISA

Fig. 4. A comparison ofretanus antitoxin ELISA (1 I5 sera). Line of identity is given.

Sloper of ELBA

titration

concentration

I I (IU ml“b

assessed in routine ELISA and in IgG specific

curves

The assumption that sera with low neutralizing activities might contain antibodies of different quality compared with sera with high levels was examined in several ways. As it has been suggested that the slopes of the dose-response curves in ELISA are related to antibody affinity’ the slopes of the ELBA curves obtained for the groups of sera concerned were examined (Table3). The mean slope of the groups decreased with decreasing in vivo activity but the variability between sera was marked making the correction by means of average slope of ELISA values to obtain measures of neutralizing activity unjustified. Competition ELISA

using monoclonal antibodies

Competition between serum antitoxin and monoclonal antibodies (MoAb) in binding to the solid phase antigens in ELISA was studied in two ways, i.e. (a) by the measurement of the displacement of the binding of MoAb in the presence of serum and (b) by the measurement of the displacement of serum antitoxin binding in the presence of MoAb. Displacement was studied by the reduction in OD obtained by anti-mouse and anti-human conjugates, respectively. In the presence of the human standard serum (antitoxin concentration 4-O IU ml-‘) almost complete inhibition of binding of the mixture of the neutralizing MoAb concerned was obtained whereas no inhibition was observed when the standard serum was diluted more than l/526 (antitoxin concentration 0.0077 IU ml-‘). At a dilution of l/l6 50% inhibition occurred (Fig. 5). 150

22 19 16 6

23

29

No. of sera

Median (U ml- ‘)

0*00005 0~00035 0.0055 0.050 0.18 2.0

Interval (LJ ml-‘)

o*ooo 1 0~0001-0~00099 0~001-0~0099 0~01-0*099 0.1-0.99 1

In vivo activity

Range (IU ml- ‘) __. 0.0065-0.033 0~008C--0.032 0*0075-o* 1’3 0.026-0.20 0.14-0.70 1.2-5.6

Median (IUmlI’l --__ 0.017 0,015 0.025 0.13 0.25 2.5 Median 0.65 0.70 0.79 1.00 0.95 1.02

Range 0.48-0.87 0.52-0.98 0~59-1*01 o-79-I.15 0*75-1.06 0.94-l-24

__-~

___Relative slope

Result of routine ELBA

Median 12 12 19 29 29 32

Range 12-50 12-50 12-47 13-39 13-56 19-42

MoAb

Percent inhibition

ELISA

12 12-29 12-56 29-69 44-7 1 65-94

Range

_I_

:,: 75

12 12 20

Median

Toxin in solution

in competition

3. The results of routine ELISA and competition ELBA with MoAb or toxin in solution relative to the in vim activities of 115 sera. The slope of ELBA titration curve for each serum is given relative to the slope of the standard strum

TABLE

0. SIMONSEN

ET AL.

III 2O

III I I I I I I I .-& 22 *3 24 25 26 27 29 29 *W $1 $2 Serum dhtion

Fig. 5. Competition ELISA. The binding of toxin neutralizing of human serum (antitoxin concentration 4.0 JU ml-‘).

monoclonal antibodies in the presence

In order to investigate the possibility of predicting the in uivu activity of sera from their ability to compete with neutralizing MoAb the panel of 115 sera was tested in this system and the dilution ofa serum yielding 50% inhibition ofMoAb binding was taken as a measure of the potency. In uivo activities greater than 0.1 U ml-’ were well predicted by this approach but sera with lower in viva activities were not capable of inhibiting MoAb binding by 50% even when undiluted. This system was thus found unsuited for the determination of the potencies ofsera of the problematic type with low in uivo activity. If less than 50% inhibition was chosen as the measure a problem was encountered in that sera without neutralizing activity exhibited inhibition in the same order of magnitude as sera with in vivo activities below 0.1 U ml-‘. In competition experiments performed to study the inhibition of serum antitoxin binding a mixture of the undiluted culture supernatant MoAbs were used to achieve the most effective competition. The MoAb mixture was found to inhibit only 36% of binding of a l/100 dilution of the standard serum. When the MoAb mixture was diluted less pronounced inhibition was observed and no inhibition was found when the MoAb mixture was diluted more than l/100 (Fig. 6). The panel of human sera was assessedin this competition system and the results were expressed as the percent inhibition exhibited by the undiluted MoAb. Table 3 gives the results and it appears that samples containing the smaller amounts of antibody on average became displaced to a lesser extent than sera containing higher levels although considerable variation was observed, particularly in the groups of low titered sera. Competition ELISA

llsing toxin in solution

In pilot experiments tetanus toxin and/or toxoid was mixed with several dilutions of human sera and incubated overnight. The mixtures were then transferred to ELISA plates and the binding of antibody to the solid phase was assessed. This system was compared with a simplified version in which the human serum dilutions were added 152

MODIFICATION

I I IO0

OF THE

ELBA

I

I

I

IO’

102

103

OF TETANUS

ANTITOXIN

I 104

Moab dilutm

Fig. 6. Competition ELISA. The binding of human serum antitoxin toxin neutralizing monoclonal antibodies.

in the presence of a mlxture of

directly to ELISA plates to which the soluble antigens had been added beforehand together with the dilution buffer. It was observed that these two versions ofthe antigen competition assay gave identical results and thus the latter method was chosen for convenience for the large scale investigation of samples. The presence of soluble tetanus toxin was responsible in a dose dependent way for the inhibition of the binding of antibodies to the solid phase (Fig. 7). Binding of the standard serum diluted from 1/ 100 by 1 Lf ml-’ of soluble tetanus toxin was almost totally inhibited, whereas this toxin concentration inhibited sera with low neutralizing activities only slightly. The concentration, C n, of antitoxin bound to toxin in solution was calculated from titration curves of twofold dilutions of serum titrated in parallel with and without toxin: Ce = C(l - R), where C is the total antitoxin concentration determined conventionally relative to the standard serum, and R is the relative potency of serum with and without toxin competition, respectively. As relative potencies above 0.88 (l/ l-14) were not found to be significantly different from identity, calculated values of Cs less than 0.12C are given as truncated values. The quantitative significance of Cn in the prediction of neutralizing activity of sera was investigated in pilot experiments with artificial mixtures of dilutions of the standard serum and dilutions of a serum devoid of neutralizing activity (Table 4). The concentration of neutralizing antitoxin in the mixtures, calculated from the quantitative composition, was found to correspond to the CB given by competition ELISA with 0.1 Lfml-‘. The reliability of the antigen competition ELISA for the general prediction of in viva activity of sera was investigated by the application of the system on the panel material and the results are shown in Fig. 8. A very satisfactory correlation between neutralizing 153

0. SIMONSEN

ET AL.

Standard

serum

,

SF30

,

,

,

] ,, ,

,

,

,

,

,

,

,

20 2’ 22 23 24 25 2s 27 20 2’ 2’ 23 24 25 26 2 Serum dllutlon

Fig. 7. Competition ELBA with toxin in solution. The binding of serum antitoxin co the solid antigen phase in the absence of toxin in solution (A) and with toxin added in different concentrations: B, 10m3 Lfml-‘; C, IO-’ Lfml-‘; D, 10-l Lfml-‘; E, 1 Lfml-‘; F, 10 Lfml-‘. Results of testing the standard serum l/100 (in viva activity 0.04 UrnI-‘) and a serum, S80, with low in viuo activity (0.0008 U ml-‘) are given.

I 0-I Competition

ELISA

I I (IlJml-‘)

Fig. 8. A comparison of the serum content of tetanus antitoxin able to bind toxin in solution (competition ELBA) and neutralizing activity (in viva assay) in 115 sera. Values are truncated at 0*004 ILJml-’ above which level specific values are given in competition ELISA. Regression line and lines corresponding to -t2 SD for the specific values (N = 53) are given. 154

MODIFICATION

TABLE

4.

OF THE

ELBA

OF TETANUS

ANTITOXIN

Neutralizing antitoxin activity estimated by competition

ELBA with toxin in solution in six artificial mixtures of the standard serum l/100 (in viva activity = ELBA activity = O-04 Uml-‘) and a serum devoid of neutralizing activity (ELISA activity = O-0275 IU ml-‘) relative to the neutralizing activities calculated from the quantitative compositions of the mixtures. Calculated and observed activities in conventional ELISA are given for comparison Neutralizing Proportion

activity

(U ml-‘)

ELBA

activity

(IU ml-‘)

of the

standard serum (%) 100 79 50 23 8 0

Calculated

Estimated

Calculated

Observed

0.04 0.03 16 0.02 O-0092 0.0032 o*ooo 1

0.04 0.028 0.0 19 0.0 104 0.0028 0.0028

0.04 0.0374 0.0338 0.0302 0.0285 0.0275

0.04 0.0354 0.0343 0.0305 O-0282 0.0275

activities predicted by the competition ELBA and values assessedin vivo was observed. For values in the range in which specific values were given by the competition ELISA (>0*004 IU ml- ‘) the correlation coefficient was 0.98 (N = 53) and the regression line was nearly identical to the identity line (O-20 < P < 0.30). The prediction of in r&o values less than O-004 U ml-’ was also found to be reliable (observed values below O-005 IU ml-‘, N = 62). The 95% confidence limits of competition ELBA values in the prediction of in vivo activity at the protective level (0.0 1 U ml-‘) were found to be O-0055 and 0.022 IUml-’ (Fig. 8).

DISCUSSION An in z&o method for tetanus antitoxin determination which would assay sera at all antibody levels and provide results closely correlated with those obtained in neutralization tests in animals would be of great general value. Indirect ELISA has generally been found well suited for the estimation of tetanus antitoxin levels in human serum2~738but investigators who examined sera with low antitoxin concentrations called attention to disagreement between the ELISA and the in vivo assay results with such sera.* When we subsequently investigated a large number of sera the reliability of ELISA for sera with antitoxin levels above 0.2 IU ml-’ was confirmed but for sera with lower antitoxin levels non-reliability with considerable overestimation was found. This problem, which has been focused on in this report, is for simplicity referred to in what follows as ‘the problem’. A simple way to circumvent the problem would be to make a fixed cut-off point so that estimates

of antibody

content

by ELISA below a fixed level were not considered

to

be reliable. This might be acceptable for clinical purposes to determine whether patients were sufficiently immunized as protective antitoxin levels are safely predicted by ELISA values above 0.16 IU ml- i. i Many sera with ELBA values below this level, however, contain protective amounts as assessedin viva and for population studies this approach is unsuitable because the incidence of subjects with a serum antitoxin content below the protective level will be the main interest. 155

0. SIMONSEN

ET AL

The problem encountered when comparing results obtained using methods which are so different as ELISA and neutralization tests presumably reflects the basic phenomenon that in the in vim assay only the binding of immunoglobulin is assessed whereas in the neutralization system, apart from the binding of the antibodies to the toxin, certain qualitative functional attributes of the antibodies are involved. An attempt to solve the problem, as described in a recent paper,’ was to use the OD values obtained with sera containing no tetanus toxin neutralizing antibodies as representative of the background and to subtract these OD vlaues from the values obtained using test sera. We found that a considerable variation occurred in the ODs given by different negative sera and were left with the problem that if the upper 95% confidence limit of the OD obtained by negative sera was chosen for subtraction, low titered sera might be scored as negative in ELBA and sensitivity in the lower range was totally lost! The effect of introducing changes in antigen purity in the system by the application of catching MoAb to focus toxin, gave a certain amount of improvement but quantitatively this was moderate. This refinement, as well as the results obtained by ELBA on contaminant protein coated plates, indicated that antibodies reacting with contaminant proteins are detected in the routine ELISA. The observation that the detection of only IgG antibodies in ELISA had no influence on the low titered sera is in accordance with the assumption that such sera are mainly derived from subjects many years after tetanus immunization at which time serum contains predominantly IgG antibodies. ‘(’ The idea that differences between low and high titered sera might be due, to some extent, not only to the quantity of antibodies but also to the quality, was considered and the possibility of functional avidity differences was supported by the fact that the slopes of the dose-response curves in ELISA decreased on average with decreasing antibody content of the samples. From studies of MoAb against the hapten DNP such a correlation between affinity and slopes has been suggested. ” Whether this is true also for polyclonal antibodies with specificity for tetanus toxin/toxoid has, however, not been proven, but the slope results might support the assumption of avidity differences and the reasons for exploring competition ELISA methods mainly was derived from this basic concept. It was anticipated that antibodies in sera with a low content of neutralizing antitoxin in ELISA would bind with a lower average avidity than antibodies in sera with higher neutralizing activity and it had previously been suggested that only high avidity antibodies are toxin neutralizing. iz It was therefore expected that antibody binding to the solid phase would be more susceptible to competition with toxin neutralizing monoclonal antibodies the lower the in viva activity of the sera. A tendency towards the opposite, however, was observed, which indicated that either the concept was wrong or the antibodies that are assessedin low titered sera are not toxin specific because they cannot be displaced by toxin specific MoAb. Hence the rationale for introducing the antigen competition ELISA was: (a) the functional effect ofavidity differences might be detected in such a system, as suggested by a recent study of monoclonal antibodies, ” (b) antibodies reactive with toxin in solution might be detected selectively which, in essence, is the mechanism in the in vivo neutralization assay in which toxin and serum are mixed in solution prior to injection, and (c) the system would possibly exclude antibodies which bind only to solid phase antigens either because of conformational changes of the antigen or due to the possibility of bivalent antibody binding, the latter 156

MODIFICATION

OF THE

ELISA

OF TETANUS

ANTITOXIN

having been suggested as essential for relative strong coupling in ELBA of low avidity antibodies. “,I* The results were such that antitoxin binding in the majority of the high and intermediate level sera was readily inhibited by soluble toxin whereas many low titer sera were inhibited considerably less and to varying degrees ranging from no inhibition to 30%. When the amount ofantitoxin in each serum calculated as bound to the antigen in solution was considered as a measure of the neutralizing antitoxin content,

a good

correlation

with

the in viz10 neutralization

results

was obtained,

indicating that this modification of the ELISA system conveniently solved the basic problem. Whether the antigen competition system solves the problem primarily because of antibody avidity the presence of non-toxin

it is a simple way of correcting

for

antibodies remains to be investigated. This new ELISA modification which, in essence, is a double serum titration

on

conventionally

differentiation

of whether

coated ELISA plates, one row with antigen

in solution

and one without,

may be introduced easily into routine serology for population studies and for the control ofvaccine potency in animals as well. The principles of the antigen competition ELISA may, moreover, be of value in the detection of functional antibodies to other antigens. REFERENCES 1. Simonsen 0, Bentzon MW, Heron I. ELISA for the routine determination of antitoxic immunity to tetanus. J Biol Stand 1986; 14: 231-240. 2. Gentili G, Pini C, Collotti C. The use ofan immunoenzymatic assay for the estimation of tetanus antitoxin in human sera: a comparison with seroneutralization and indirect haemagglutination. J Biol Stand 1985; 13: 53-59. 3. Simonsen 0, Kjeldsen K, Heron I. Immunity against tetanus and effect of revaccination 25-30 years after primary vaccination. Lancet 1984; II: 1240-1242. 4. Ipsen J. Systematische und zufallige fehler quellen bei messung kleiner antitoxin mengen. Z Immunitatsforsch 1942; 102: 347-368. 5. Ourth DD, McDonald AB. Neutralization of tetanus toxin by human and rabbit immunoglobulin classes and subunits. Immunology 1977; 33: 807-815. 6. Steward MW, Lew A. The importance of antibody affinity in the performance of immunoassays for antibody. J Immunol Methods 1985; 78: 173-190. 7. Melville-Smith ME, Seagroatt VA, Watkins JT. A comparison of enzyme-linked immunosorbent assay (ELISA) with the toxin neutralization test in mice as a method for the estimation of tetanus antitoxin in human sera. J Biol Stand 1983; 1 I: 137-144. 8. Sedwick AK, Ballow M, Sparks K, Tilton RC. Rapid quantitative microenzyme-linked immunosorbent assay for tetanus antibodies. J Clin Microbial 1983; 18: 104-109. 9. Ambrosch F, Wiedermann G, Miiller H. Eine neue Mikro-ELISA-Methode zur Bestimmung der Tetanus-Antikorper. Zbl Bakt Hyg 1984; A 258, 173-182. 10. Kodo H, Gale RP, Saxon A. Antibody synthesis by bone marrow cells in vitro following primary and booster toxoid immunization in humans. J Clin Invest 1984; 73: 1377 1384. 11. Lew A. The effect of epitope density and antibody affinity on the ELISA as analysed by monoclonal antibodies. J Immunol Methods 1984; 72: 171-176. 12. Jerne NK. A study ofavidity based on rabbit skin responses to diphtheria toxin-antitoxin mixtures. Acta Path Microbial Stand 195 1; suppl. LXXXVII. 13. Friquet B, Chaffotte AF, Djavadi-Ohaniance L, Goldberg ME. Measurements of the true affinity constant in solution ofantigen-antibody complexes by enzyme-linked immunosorbent assay. J Immunol Methods 1985; 77: 305-319. 14. Nimmo GR, Lew AM, Stanley CM, Steward MW. Influence of antibody affinity on the performance of different antibody assays. J Immunol Methods 1984; 72: 177-187 157