A competitive inhibition ELISA for the quantification of human interferon-γ

Journal oflmmunological Methods, 162 (1993) 247-255 © 1993 Elsevier Science Publishers B.V. All rights reserved 0022-1759/93/$06.00

247

JIM 06721

A competitive inhibition ELISA for the quantification of human interferon-y A n n e B. Wilson, S.M. McHugh, J. Deighton, Pamela W. Ewan and P.J. L a c h m a n n Molecular Immunopathology Unit, Medical Research Council Centre, Cambridge, UK (Received 17 August 1992, revised 15 February 1993, accepted 5 March 1993)

A competitive enzyme immunoassay has been developed for the measurement of human interferon-y (IFN-y) in cell culture supernatants. The assay is based on the dose-dependent inhibitory effect of liquid phase IFN-y on the binding of a specific monoclonal antibody to recombinant IFN-y (rIFN-y) immobilized on microtitre plate wells. The extent of monoclonal anti-IFN-3, inhibition was determined by the uptake of alkaline phosphatase-conjugated goat anti-mouse IgG and the subsequent development of enzyme substrate colour. Absorbance readings were taken and results for test samples were extrapolated from standard rIFN-y inhibition curves constructed as logit-log plots. Assay performance was assessed using three different monoclonal antibodies (clones 20G7, H-22 and GZ-4). Optimum sensitivity was achieved with the antibodies of higher affinity, 20G7 and H-22, which gave reliable quantification of IFN-y over a wide range of concentrations from 0.4 ng/ml (3.4 IU/ml), or less, to 250 ng/ml ~ 2000 IU/ml). The inhibition assay incorporates the advantages of specificity, reproducibility and convenience of performance which are the hallmarks of monoclonal antibody-based ELISAs. However, compared to the sandwich ELISAs previously described for human IFN-y, it is considerably more economical in its use of monoclonal anti-IFN-y, requiring < 50 ng of a single antibody per 96 well plate. It also uses relatively small volumes of test samples (50/zl/well) which is particularly advantageous where limited amounts of cell culture supernatant are available for cytokine assays. Key words: IFN-y, human; IFN-y assay; ELISA, competitive

Introduction Correspondence to: A.B. Wilson, MIP Unit, MRC Centre, Hills Road, Cambridge, CB2 2QH, UK. Tel: (0223) 402471; Fax: (0223) 213556. Abbreviations: ELISA(s), enzyme-linked immunosorbent assay(s); IFN-y, interferon-y; PBMC, peripheral blood mononuclear cells; PBST-FCS, phosphate-buffered saline, pH 7.4, containing 0.05% Tween 20 and 5% heat-inactivated foetal calf serum; PHA, phytohaemagglutinin; p-NPP, disodium para-nitrophenyl phosphate; rlFN-y, recombinant interferon-y; SKSD, streptokinase-streptodornase; Thl, Th2, types 1 and 2 T helper cells.

Recent intensive research on T lymphocyte subsets and their functions has shown that IFN-y, as a product of type 1 T cells, plays a major part in the regulation of immune responses (Mosmann and Coffman, 1989; Street and Mosmann, 1991; Bloom et al., 1992). Imbalance between lymphokines synthesized by type 1 (Thl) and type 2 (Th2) helper T cells is believed to be a critical

248 factor in the pathogenesis of allergic diseases (Kapsenberg et al., 1991; Parronchi et al., 1991) and to influence the immunoregulatory control of parasitic infections (Del Prete et al., 1991; King and Nutman, 1991). Investigations on these conditions and other disorders of the immune system are dependent on the availability of accurate, specific assays for the detection and quantification of IFN-y and other cytokines. In the past, IFN-y assays were based solely on its biological activities, especially those associated with viral inhibition which can he measured either as a reduction in viral RNA synthesis (Allen and Giron, 1970) or as IFN-y-induced resistance of target cells to the cytopathic effects of a challenge virus (Familletti et al., 1981). These biological assays have major disadvantages associated particularly with their lack of specificity, time consuming performance and poor reproducibility. With the advent of monoclonal antibodies more reliable methods have been developed for quantifying IFN-y, such as immunoradiometric assays (Chang et al., 1984), reverse passive haemagglutination (Butterworth et al., 1988; Wilson et al., 1988) and enzyme-linked immunosorbent assays (ELISAs) as described for human IFN-y by Van der Meide et al. (1985), Oda et al. (1986), Gallati et al. (1987), Jitsukawa et al. (1987), and Hirai et al. (1989). All of these ELISA procedures were based on the 'sandwich' principle, involving the capture of IFN-y from test samples by solid phase-bound monoclonal or polyclonal (Hirai et al., 1989) anti-IFN-y antibody, and the subsequent detection of captured IFN-y using enzyme-labelled specific antibody which was either a second monoclonal reagent (Van der Meide et al., 1985; Gallati et al., 1987; Jitsukawa et al., 1987) or a polyclonal anti-IFN--/ preparation (Oda et al., 1986). These sandwich ELISAs also had some disadvantages. In particular, they required relatively large quantities of monoclonal antibodies, either as coating or detecting reagents, ranging in amounts from 100 /xg (Oda et al., 1986) to 1 mg or more (Gallati et al., 1987; Hirai et al., 1989) per 96-well microtitre plate. This level of antibody requirement may prove prohibitively expensive for laboratories where 'inhouse' antibodies are not available. The use of polyclonal antibodies in sandwich assays may also

produce high background reactions due to the presence of 'natural' antibodies, reactive with mouse IgG or with blocking proteins, which will require appropriate adsorption. The assay we have developed for human IFN--/ is simple in performance, reacts with the active forms of both natural and recombinant IFN-T, and is economical in reagents, requiring only 2 /zg rIFN- 7 and less than 50 ng of a single monoclonal anti-IFN-7 antibody per 96-well plate. Essentially, the procedure is an inhibition ELISA in which soluble IFN-y in test samples competes with solid phase-attached rIFN-y for the binding of a mouse monoclonal IFN-y-specific antibody. The amount of bound antibody is determined by uptake of enzyme-labelled anti-mouse IgG, and levels of IFN-y in unknown samples are calculated from a standard inhibition curve. The development of this assay and its application to measuring IFN-y in lymphocyte culture supernatants are described in this report.

Materials and methods

Interferon preparations Human rIFN-3, (100/xg/ml) was obtained as a lyophilized powder from Biogen, Geneva, Switzerland. The rIFN-y was derived from Escherichia coli and, after reconstitution, was 99% pure in a balanced salt solution containing human serum albumin. When standardized against the British reference human natural IFN-y the specific activity of our rIFN-y was equivalent to 8.4 x 106 IU/mg. Working aliquots of rIFN-y were stored at 4°C over a period of months without measurable deterioration. British Standard natural human IFN-y (3000 IU/ml) was supplied by the National Institute for Biological Standards and Control, South Mimms, England (batch NIBSC 82/587), and was stored in aliquots at - 20°C. Monoclonal anti-IFN- y antibodies The assay was initially developed using a mouse IgG1 monoclonal anti-human IFN-y, clone 20G7, kindly provided by Celltech, Slough, England. This antibody was > 88% pure by SDS-PAGE, and at a concentration of 1 mg/ml in 40 mM

249 phosphate, pH 7.5. Two other antibodies reactive with natural and recombinant human IFN-y were also tested in the inhibition assay. One of these, clone H-22 mouse IgG2a from Genzyme (item no. 1598-00, lot no. 82140) was supplied by NBS Biologicals, Hatfield, England as a 1 mg/ml solution in phosphate-buffered saline (PBS). The second, clone GZ-4 mouse IgG1, was obtained from Boehringer Mannheim UK., Lewes, England (cat. no. 1296 825, lot no. 14724300) as a lyophilized preparation and was reconstituted to give antibody at 200 p~g/ml in PBS with 20 mg/ml raffinose. Monoclonal H-22 had a much higher level of IFN-T neutralizing activity, requiring only 1 ng antibody to block the anti-viral cytopathic effect of 1 IU human IFN-y compared to 200 ng of GZ-4 antibody. All working solutions of antibodies were stored at 4°C, with the addition of 0.01 M sodium azide.

Other reagents Alkaline phosphatase-conjugated goat antimouse IgG antibody (adsorbed with human serum proteins), disodium para-nitrophenyl phosphate (p-NPP) and Tween 20 detergent were all obtained from Sigma Chemical Co., Poole, England. Foetal calf serum (FCS) was purchased from Globepharm, Esher, England, and was heat-inactivated at 56°C for 30 min prior to use. Tissue culture medium for peripheral blood mononuclear cell (PBMC) cultures was RPMI 1640 (Flow Laboratories, Irvine, Scotland) supplemented with 100 U / m l penicillin, 100 ~ g / m l streptomycin, 2 mM glycine (all from Sigma), 20 mM N-2-hydroxyethylpiperazine-N'-2 ethanesulphonic acid (Hepes, Merck, Poole, England) and 10% heat-inactivated human AB + serum (supplied by Regional Blood Transfusion Service, Cambridge, England). Inhibition ELISA for human IFN-T Initial investigations on the various parameters of the assay and their standardization are described in the results section. The assay, as developed with 20G7 antibody, was carried out as follows. Immulon II 96-well microtitre plates (Dynatech M 129B) were coated with rIFN-y at 200 ng/ml in 0.05 M carbonate-bicarbonate buffer, pH 9.6, overnight at 4°C, using 100 /zl ali-

quots/well. The plates were washed three times with phosphate - buffered saline, pH 7.4, containing 0.05% Tween 20 (PBST) and blocked with 5% heat-inactivated FCS in PBST (PBST-FCS) for 1 h at room temperature (RT). After removal of blocking reagent, 50/zl aliquots of test samples or of titrated standard rIFN-y, diluted in PBSTFCS, were dispensed into duplicate/triplicate series of wells, respectively, followed by 50 /xl aliquots of monoclonal 20G7 anti-human IFN-7 antibody, diluted 1/150000 in PBST-FCS. The plates were shaken briefly on a vortex mixer, then incubated overnight at 4°C. After three washes with PBST, alkaline phosphatase-conjugated goat anti-mouse IgG, diluted 1/600 in PBST-FCS, was added (100 /~l/well) and left to react for 4 h at RT. The plates were then washed three times with PBST, and given a final wash with distilled water before adding 100 /zl aliquots of p-NPP substrate at 1 mg/ml in 0.05 M diethanolamine buffer (pH 9.8). Following incubation for 1.5 h at 37°C, absorbance was measured at OD405 in a Bio-Rad microplate reader (model 3550). Appropriate positive and negative controls were included on each plate, and the results for the test samples were calculated from the standard inhibition curve using Bio-Rad Microplate Manager software.

Results

Establishment of optimum assay conditions The assay was developed using 20G7 monoclonal anti-IFN-y. This antibody was first titrated in duplicate rows of microtitre wells coated with different concentrations of human rIFN-y ranging from 5 /xg/ml down to 40 ng/ml. After overnight incubation at 4°C the bound antibody was detected with enzyme conjugated anti-mouse IgG. From the observed antibody binding curves, it was apparent that a relatively low concentration of rIFN-y was sufficient to coat the solid phase, and a solution of 200 ng/ml rIFN-y was selected for further development of the assay. This concentration of coating rIFN-y gave monoclonal anti-IFN-7 binding curves as shown in Fig. 1, with 50% maximum binding point (50% max bp) at a dilution of 20G7 antibody equivalent to

250

1/75000 of 1 mg/ml stock solution (~ 13 ng antibody/ml). Competitive inhibition assays were then set up in which the binding of 20G7 monoclonal antiIFN-y to solid phase rIFN-y was inhibited in a dose dependent manner by the addition of an equal volume (50 /.d/well) of titrated rIFN-y to the reaction mixture. In these assays, comparison was made between inhibition curves formed with anti-IFN-y antibody at a concentration equivalent to its 50% max bp (diluted 1/75000) and with the higher dilutions of 1/100 000, 1/150 000 and 1/200 000. As shown in Fig. 2, the sensitivity of the assay was increased using anti-IFN-y antibody at concentrations below the 50% max bp. A dilution of 1/150000 (6.6 ng/ml) was chosen as combining optimum sensitivity with convenience of assay performance, such as the time required for the enzyme substrate reaction (1-1.5 h at 37°C).

Sensitivity and reproducibility of the assay The present assay permits measurement of IFN-y levels over a wide range of concentrations. Using 20G7 monoclonal antibody and a logit-log curve fit program (Bio-Rad Microplate Manager software) to calculate results, 95% confidence intervals were obtained between 0.4 ng/ml (3.4 IU/ml) and 250 n g / m l ( ~ 2000 IU/ml) of IFN-y (Fig. 2c). Lower concentrations of IFN-y down to 0.1 ng/ml were detectable and were integrated into the standard curve, though with logit-log curve fit the results at this level may be less reliable. The amount of IFN-y required for 50% inhibition showed remarkably little variation giving a mean value_+ SD of 12.86 + 1.57 ng/ml over 30 assays. Intra-assay coefficients of variation (%CV) were calculated for inhibition of 20G7 antibody by five different concentrations of rIFN-y ranging from 0.1 ng/ml to 30 ng/ml. These were set up in triplicate on each of six microtitre plates and gave excellent plate-to-plate correlation, with %CV varying from 2.0 to 6.6 (mean %CV _+ SD, 4.6 + 1.8). Interassay reproducibility was assessed using solutions of human natural IFN-y at 10, 30, 100 and 300 IU/ml. When measured in duplicate in four separate assays, these IFN-y preparations

gave %CV of 12.3, 4.6, 13.4 and 17.0, respectively (overall mean %CV _+ SD, 11.8 + 5.2).

Use of antibodies with different IFN-y-neutralizing activities Two additional antibodies, clones H-22 and GZ-4, were assessed for their performance in the inhibition assay. These monoclonals had very different IFN--/ neutralizing activity which was higher for antibody H-22 than for GZ-4 (see the materials and methods section). Prior to use in the IFN-y assays, both antibodies were titrated on microtitre plates coated with rIFN-y (200 n g / ml), as already described for monoclonal 20G7. They were then tested in the inhibition assay using dilutions below the 50% max bp. Monoclonal H-22 antibody performed well, giving standard curves similar in range and sensitivity to those obtained with 20G7 antibody. As shown in Fig. 3a, H-22 anti-IFN-y could be used effectively at dilutions down to 1/300000 of 1 mg/ml stock (3.3 ng/ml). At this antibody concentration, the rIFN-y-inhibition curves exhibited 95% confidence intervals between 0.4 ng/ml and 250 or 500 ng/ml, and 50% inhibition was obtained with 8.9 _+ 1.9 ng/ml (n = 9). Reproducibility of assays using the H-22 antibody was measured over the same range of IFN-y

E c

1_ ........................ c

,<

0 10 3

. 10 4

105

106

~.

iI,

107

Monoclonal anti-IFN y dilution Fig. 1. Titration of 2007 monoc]onal anti-][FN-y antibody on microtitre plate coated with human rlFN-y (200 ng/ml), incubated overnight at 4°C. Bound antibody was detected with goat anti-mouse IgG-alkaline phosphatase (1/600), reacted for 4 h at RT. Substrate (p-NPP, 1 mg/ml) colour developed 1 h at RT. Graph shows 50% maximum binding point of monoclonal anti-IFN-y at dilution 1/75 000.

251

concentrations as described for 20G7 antibody, and gave intra- and inter-assay %CVs of 5.2 + 3.1 (range 2.6-9.4, n = 3) and 6.1 + 2.0 (range 4.08.8, n = 4), respectively. The antibody with low neutralizing activity, monoclonal GZ-4, gave standard curves shifted somewhat to the right compared to those produced by 20G7 or H-22 antibodies (Fig. 3b). This shift was indicative of a lower antibody binding affinity and resulted in a reduction in assay sensitivity. The inhibition curves for GZ-4 antibody, when used at a concentration of 8 n g / m l , had 95% confidence intervals between 2.5 and 500 n g / m l rlFN-T, and the lowest detectable level of rlFN- 7 was around 1.5 n g / m l . Over a series of 8 assays, the amount of rlFN-T required for 50% inhibition of GZ-4 antibody binding was much higher than that for the other two monoclonals at

a)

Curve Fit ~s.~

Excellent results were obtained in 20G7 antibody inhibition assays using human natural IFN-7 (British Standard reagent). This IFN- y preparation gave inhibition curves which closely paralleled those of our rIFN- y (Fig. 4), making it possible to monitor the rIFN-y activity in terms of I U / m l . Preliminary experiments using tissue culture medium (RPMI 1640 with 10% heat-inactivated human serum and other supplements) which had been spiked with either natural IFN-7 or rIFN-7 showed good recovery over a wide range of concentrations (Table I). Low levels of

Curve Fit:

:

1.8665 0.1530)1(1+(Conc/16

1188)^ 1.0959)+0.1530

A B S

1.53-

kbs.=( 1.5923-0.1392)/(l+(G3nc/14.66501 ^

I ,59.

B A N 0,85 C

1.3o.

A

0,73.

E

0.44-

N C

o,so-

0,16-

0.15. 0 ,'03

0/69

15,85

99,Z2

0/03

1000,00

CONCENTRATION

o.'69

15:05

99.~2 1ooo.oo

CONCENTRATION

a)

e)

Curve Fit: A B

1.06331+0.1392

0 R ~.Ol. B

R 1.19

E

Assay of IFN-T in tissue culture supernatants

b) I ,87

A B S O

42.5 + 9.4 n g / m l . The respective %CVs from intra- and interassay data obtained with GZ-4 were 4.3 + 2.7 (range 2.0-8.7, n = 3) and 10.6 + 7.5 (range 2.3-16.7, n = 3).

,abs.-=t 1.23060.1343)/(l+(O)nc/ll.8333)^

I ,26-

Curve Fit

0.999111+0.1343

A B

kbs.=( 1.11115-0.1132)/(l+(Q)nc/

1,00-

S 1.030 R 0.81B A N o.59-

S O R B A

o,6s-

N

0.47-

C E o.37-

C E

0.30-

0.15-

~l 0,03

,

0,69

,

15,85

i

99.22

CONCENTRATION

i

1000.00

6.6840) ^

1.110141+0.1132

0,03-

0,120.03

0.69

15,85

99.22

1000.00

C O N C E N T R A T I ON

Fig. 2. Standard curves for rlFN-T inhibition ELISAs comparing assays using 20G7 monoclonal anti-IFN-T antibody at four different dilutions of a 1 mg/ml solution: (a) 1/75000; (b) 1/100000; (c) 1/150000; (d) 1/200000. Graphs represent Iogit-log plots of absorbance (OD405) against concentration of human rlFN-T (ng/ml). Absorbance measured after substrate incubation for 1.5 h at 37°C. Assays show increasing sensitivity with decreasing antibody concentration.

252

a)

TABLE I (.ltrve I ' ] t : A s . = ( 0.7305 0,1143)/(1+(Conc/

/

8.4137) ^

1.0090)+().1143

RECOVERY OF NATURAL OR RECOMBINANT IFN-7 FROM 'SPIKED' TISSUE CULTURE MEDIUM CONTAINING 10% HUMAN SERUM AND SUPPLEMENTS

0,60

(see materials and methods section)

0.48

Concentration of

~

0,36

IFN-7 added 1U/ml

Amount of IFN-T measured IU/ml

Percentage recovery

~

0,24

Natural IFN- 3' 1000 300 100 30 10

1245 338 127 33.5 9.8

124.5 112.7 127.0 111.7 98.0

649 221 81.7 24.5 9.8

77.3 88.4 97.3 98.0 116.7

0.12

0,03

0,'69

15.'85

99122

1000.00

CONCENTRATrON

b)

R e c o m b i n a n t IFN- 3' Olrve I 'lt: Cbs.=(0.6405-0.1226)/(1+(Conc/37.4261) ^

0.65 A B S O

o.s4

R

o.44-

B A N

%

1.1771)+(}.1226

840 (100) a 250 (30) 84 (10) 25 (3) 8.4 (1)

a Figures in parentheses = rlFN- 7 (ng/ml).

0.34-

C

E

o,24-

0,T3-

0:69

0103

99'.22

,5:85

CONCENTRATI

,ooo oo

ON

Fig. 3. Logit-log graphs of rlFN-7 inhibition ELISAs comparing the reactivity of two different monoclonal antibodies; (a) clone H-22, with high IFN-T neutralizing activity, used at a concentration of 3.3 ng/ml (1/300000 dilution of 1 mg/ml

stock), and (b) clone GZ-4, with low neutralizing activity, used at 8 ng/ml (1/25000 dilution of 200 /zg/ml stock). Absorbance read after 1.5 h incubation at 37°C.

10Natural IFN O8-

7

c o r n in a n t l F N

-Y

E c ~

06-

~

04-

x~ <

o2-

oo 1

, ~

. lO

.

. lOO

. lOOO

10000

ol

1

I

lO

lOO

lOOO

I

I

I

I

I

I

, 10000o loooo

U/ml

(natural)

ng/ml

(recomb,)

I

Fig. 4. Competitive ELISA using 20G7 antibody showing parallel inhibition curves obtained with natural IFN-T (British Standard, NIBSC 82/587) and rlFN-T. Specific activity of the rIFN-T was equivalent to 8.4 × 106 IU/mg.

IFN-7 may be present in normal human serum. It was important, therefore, to include control samples of ceil-free, complete tissue culture medium (i.e., medium containing supplements and serum) in each assay. Changes in IFN-7 levels in cell supernatants were then assessed against any background level of serum IFN-T in the culture medium. The inhibition assay has been used successfully to quantify IFN-T production by human atopic and normal lymphocytes activated in vitro by mitogens or specific antigens. A typical pattern of IFN-7 release by stimulated PBMC from a patient allergic to the house dust mite, Dermatophagoides pteronyssinus, is shown in Fig. 5a. Despite high level IFN-7 production following stimulation with the mitogen phytohaemagglutinin (PHA), atopic T cells were not induced to release IFN-T when cultured with specific allergen (D. pteronyssinus). Of particular interest is our finding that IFN-7 production by atopic PBMC stimulated with a non-allergenic recall antigen, streptokinase-streptodornase (SKSD), is also markedly reduced, or totally lacking, when compared to the SKSD response of normal, nonatopic PBMC (Fig. 5b). These observations have

253

a) 40-

30 O~ C v 20

E Z LL.

•

PHA

[]

Dpter SKSD

[]

Control

•

PHA

[] [] []

Opter SKSD Control

10

0

4

20

26

44

50

72

96

20 144 168

Time (hours)

b) 40-

30 03 C v

20"

E o~ Z LI_

0

4

20

26

44

50

72

96

20

44 168

Time (hours) Fig. 5. Levels of IFN- 7 production measured in cell culture supernatants of peripheral blood mononuclear cells, (a) from a patient allergic to the house dust mite (D. pteronyssinus), and (b) from a normal donor. Lymphocytes (105/ml) were stimulated either with mitogen (PHA, 10 /xg/ml), with specific allergen (D. pteronyssinus extract, 1000 B U / m l ) or with a recall antigen (SKSD, 10 /xg/ml) Aliquots of supernatant were collected at intervals over a period of 7 days, centrifuged and stored at - 20°C until assayed.

been confirmed in other atopic patients and are reported in a wider study (McHugh et al., 1993).

Discussion

Immunochemical techniques employing monoclonal antibodies have proved superior to biological assays for the quantification of IFN-y because

of their specificity, reliability and ease of performance. The competitive ELISA described here shares these features with the sandwich enzymelinked antibody assays previously developed for human IFN-y. The competitive assay, however, has the added advantage of being particularly economical in its use of antibody, requiring only a single monoclonal anti-IFN-3, antibody in ng/ml amounts. Our assay also has a wide range of detection and, with the higher affinity antibodies (20G7 or H-22), will reliably quantify IFN-3, at concentrations between 0.4 ng/ml (3.4 IU/ml) and 250 ng/ml ( ~ 2000 IU/ml), although amounts of 0.1 ng/ml or less can be detected. The lower levels of sensitivity compare favourably with sandwich ELISAs for IFN-3, which have been reported to detect minimal levels from 1 or 5 ng/ml (Jitsukawa et al., 1987; Hirai et al., 1989) down to 0.05-0.06 ng/ml (Van der Meide et al., 1985; Gallati et al., 1987). The latter assays are obviously highly sensitive but they both required test sample aliquots of at least 200/~I, compared to the 50/xl aliquots used in our inhibition ELISA. Sample volume may well be a limiting factor, particularly when cell culture supernatants are required for measurement of multiple cytokines. Competitive assays are influenced by the binding affinity of the antibody employed, and, for optimum sensitivity, require a high affinity reagent. The effect of antibody affinity on our IFN-~/assay is apparent from a comparison of the inhibition curves obtained with H-22 and GZ-4 monoclonals (Fig. 3). Antibody H-22 provided an assay equal in range and sensitivity to that obtained using the 20G7 monoclonal, while GZ-4 antibody produced inhibition curves characteristic of lower affinity reactions. This resulted in a shift in the lowest level of detectable IFN-3, from 0.1 ng/ml for 20G7 or H-22 antibodies to 1.5 ng/ml for GZ-4. Selection of a high affinity antibody is important, therefore, for quantification of IFN-~/ at these low concentrations. All three monoclonal antibodies used in the present report showed good reactivity with solid phase-bound rlFN-3,, although some investigators (Jitsukawa et al., 1989) have observed diminished binding of their monoclonal antibodies to immobilised rlFN-3,. This was thought to be due to conformational changes in the rlFN-3, molecule,

254 with a consequent loss of antigenic epitopes. The same authors found that monoclonal antibody reactivity was greatly enhanced when plates were coated with a mixture of rlFN-7 and bovine serum albumin, and it is possible that the human serum albumin content of our rlFN-7 preparation had a similar enhancing effect on its reactivity with monoclonal antibody. The inhibition assay has proved highly satisfactory for use in our studies on in vitro IFN-7 production by human activated mononuclear cells (Fig. 5). For these assays, up to 30 test samples were set up in duplicate on each microtitre plate, together with a triplicate titration of standard rlFN-7 and negative controls containing no IFN7. We have found it important to include further controls with aliquots of cell-free tissue culture medium, complete with 10% human serum which may contain low levels of IFN-7. In other respects, the presence of 10% heat-inactivated serum in the culture medium appeared to have no adverse effects on the assay for IFN-3,, as shown by the excellent recovery obtained from complete medium spiked with natural or rlFN-7 (Table I). Other authors (Van der Meide et al., 1991) have reported inhibitory effects of fresh serum on IFN-3, measurement in ELISA. This was traced to complement activation in assay wells and the subsequent binding of IFN- 7 molecules to activated C3. It is obviously important, therefore, that serum in cell culture media should be heat-inactivated before use, or that other steps be taken to prevent complement activation in the IFN-7 assay. Using the inhibition ELISA, we have examined levels of IFN-7 release by mitogen- or antigen-stimulated PBMC from patients allergic to the house dust mite, D. pteronyssinus (McHugh et al., 1993; and Fig. 5). These investigations confirmed preliminary observations (Ewan et al., 1990) of impaired IFN-3, production by the PBMC of atopic patients. Despite a normal I F N - 7 response to mitogen (PHA), lymphocytes from atopic donors showed defective IFN-3, release when stimulated with antigens, indicating a diminished T h l cell response in these patients. It is particularly interesting that this defect is not limited to reactivity with specific allergen (D. pteronyssinus), but is also found in SKSD antigen

responses which resulted in reduced IFN-7 production compared to that of normal PBMC. Considering the important role of IFN-7 as an antagonist of interleukin-4 in the regulation of IgE antibody synthesis (P~ne et al., 1988), there is urgent need for a better understanding of the causes for impaired IFN-7 production in these atopic patients.

Acknowledgements This work was supported by the Medical Research Council of Great Britain, and by a grant to S.M.McH. from the National Asthma Campaign UK.

References Allen, P. and Giron, D.G. (1970) Rapid sensitive assay for interferons based on the inhibition of MM virus nucleic acid synthesis. Appl. Microbiol. 20, 317. Bloom, B.R., Salgame, P. and Diamond, B. (1992) Revisiting and revising suppressor T cells. Immunol. Today 13, 131 [review]. Butterworth, A.E., Wilson, A., Richardson, B.A., Harris, J., Gachui, R.K., Senior, D. and Coombs, R.R.A. (1988) Measurement of gamma-interferon in culture supernatants of antigen-stimulated lymphocytes from patients with Schistosoma rnansoni infections by reverse passive haemagglutination assay. Clin. Exp. Immunol. 71,241. Chang, T.W., McKinney, S., Liu, V., Kung, P.C., Vilcek, J, and Le, J. (1984) Use of monoclonal antibodies as sensitive and specific probes for biologicallyactive human y-interferon. Proc. Natl. Acad. Sci. USA 81, 5219. Del Prete, G.F., De Carli, M., Mastromauro, C., Biagiotti, R., Macchia, D., Falagiani, P., Ricci, M. and Romagnani, S. (1991) Purified protein derivative of Mycobacterium tuberculosis and excretory-secretoryantigens of Toxocara canis expand in vitro human T cells with stable and opposite (type 1 helper or type 2 helper) profile of cytokine production. J. Clin. Invest. 88, 346. Ewan, P.W., McHugh, S.M. and Romero, C. (1990)Defective interferon-y production in house dust mite allergy. Clin. Exp. Allergy20 (suppl. l), 50. Familleti, P.C., Rubinstein, S. and Pestka, S. (1981) A convenient and rapid cytopathiceffect inhibition assay for interferon. Methods Enzymol. 78, 387. Gallati, H., Pracht, I., Schmidt, J., H~iring, P. and Garotta, G. (1987) A simple, rapid and large capacity ELISA for biologically active native and recombinant human IFNy. J. Biol. Reg. Homeostat. Agents 1, 109.

255 Hirai, M., Terano, Y. and Miura, Y. (1989) Characterization of monoclonal antibodies to human interferon-gamma and their application to enzyme-linked immunosorbent assay (ELISA). Jpn. J. Med. Sci. Biol. 42, 127. Jitsukawa, T., Nakajima, S., Sugawara, I., Mori, S. and De Ley, M. (1987) Characterization of murine monoclonal antibodies to human interferon-gamma (IFN-~/) and their application for sandwich enzyme-linked immunosorbent assay (ELISA). Microbiol. Immunol. 31, 809. Jitsukawa, T., Nakajima, S., Sugawara, I. and Watanabe, H. (1989) Increased coating efficiency of antigens and preservation of original antigenic structure after coating in ELISA. J. Immunol. Methods 116, 251. Kapsenberg, M.L., Wierenga, E.A., Bos, J.D. and Jansen, H.M. (1991) Functional subsets of allergen-reactive human CD4 ÷ T cells. Immunol. Today 12, 392. King, C.L. and Nutman, T.B. (1991) Regulation of the immune response in lymphatic filariasis and onchocerciasis. Immunol. Today 12, A54. McHugh, S.M., Wilson, A.B., Deighton, J., Lachmann, P.J. and Ewan, P.W. (1993) IL-2, IL-6 and IFN-~/profiles from peripheral blood mononuclear cells of house dust mite allergic patients: a role for IL-6 in allergic disease. Submitted. Mosmann, T.R. and Coffman, R.L. (1989) Heterogeneity of cytokine secretion patterns and functions of helper T cells. Adv. Immunol. 46, 111. Oda, S., Akinaga, S., Kumagai, A., Inoue, A., Nakamizo, N. and Yoshida, H. (1986) Generation of a new monoclonal antibody and its application for determination and purifi-

cation of biologically active human gamma-interferon. Hybridoma 5, 329. Parronchi, P., Macchia, D., Piccinni, M.-P., Biswas, P., Simonelli, C., Maggi, E., Ricci, M., Ansari, A.A. and Romagnani, S. (1991) Allergen- and bacterial antigen-specific T-cell clones established from atopic donors show a different profile of cytokine production. Proc. Natl. Acad. Sci. USA 88, 4538. P~ne, J., Rousset, F., Bri~re, F., Chr&ien, I., Bonnefoy, J.Y., Spits, H., Yokota, T., Arai, N., Arai, K., Banchereau, J. and De Vries, J.E. (1988) IgE production by normal human lymphocytes is induced by interleukin 4 and suppressed by interferons 3, and t~ and prostaglandin E 2. Proc. Natl. Acad. Sci, USA 85, 6880. Street, N.E. and Mosmann, T.R. (1991) Functional diversity of T lymphocytes due to secretion of different cytokine patterns. FASEB J. 5, 171. Van der Meide, P.H., Dubbeld, M. and Schellekens, H. (1985) Monoclonal antibodies to human immune interferon and their use in a sensitive solid-phase ELISA. J. Immunol. Methods 79, 293. Van der Meide, P.H., de Labie, M.C.D.C., Wubben, J.A.M. and Borman, A.H. (1991) Complement-mediated inactivation of interferon-~/in ELISA systems. J. Immunoassay 12, 65. Wilson, A.B., Harris, J.M. and Coombs, R.R.A. (1988) Interleukin-2-induced production of interferon-3, by resting human T cells and large granular lymphocytes: requirement for accessory cell factors, including interleukin-1. Cell. Immunol. 113, 130.