Journal of Immunological Melhods, 155 (1992)215-223

215

© 1992ElsevierSciencePublishersB.V. All rights yesCrved(1022-1759/92/$05.111)

JIM06475

A peroxidase-linked enzyme immunoassay for tumour necrosis factor a utilising alternative colorimetric or chemilumimetric substrates W.R. Lamb, R.S.H. Pumphrey and P.E.C. Brenchley North West Immunology Sen'ice, St Mary's Hospital, Whitworth Park, Manchester. M13 OJH. UK

{Received20 January lt)92,revisedreceived22 April 1992,accepted21 May 1992)

We present a double antibody immunoasaay for turnout necrosis factor a (TNFu) with a peroxidas¢ dependent endpoint which can be detected by absorbance or chemiluminescence depending on the choice of substrata. The chemilumimetric and colorimetric assays have a detection threshold in human serum of 3.9 pg/ml and 7.8 pg/ml respectively and are able to recognise both rTNFa and natural TNFet. Concentrations of TNF~, interleukin-lot (IL-10t), IL-/3, IL=2, IL-3, IL-6 or interferon-~, (IFN-~.) up to 5 ng/ml failed to show any cross-reactivity. The monocIonal antibody done 5-2, used in the assays, did not ncutralise r T N F , in the L929 bioassay. The assay was able to detect rTNF~ in the presence of excess concentrations of both TNFa receptors (p55 and p75). Removal of interference by rheumatoid factor was achieved by the absorbance of the polyclonal antiserum with mouse serum and the inclusion of 10 -2 M dithiothreitol in the buffer containing the TNFa polyclonal antiserum. The assay will be useful for the quantitation of endogenous human TNFa in serum, other body fluids and culture supernatants, and can also be used to monitor levels of rTNFa in clinical trials. Key words: Peroxidnse-linkedenzymeimmuncmssay;Tumor ne~rosisI'aetora; ELISA

Introduction Tumour necrosis factor a (TNFa) consists of a non-glycosylated polypeptide of approximately 17 kDa existing in multimers of 2, 3 or 5 U (Beutter et al., 1985a,b). Each monomer is a 157 amino Careespondence to." W.R. Lamb, Regional Immunology Service, St. Mary's Hospital. Hathersage Road, Manchester MI3 0JH, UK. Tel.; 061 276 6323; Fax: 061 276 6468, Abbreciations: TNF. tamour necrosis factor. IL, inlertenkin; ELISA,en~me-linkcdimmunoab~rbcntassay;BSA, bovineserum albumin;DTT, dithiothreitol;ABTS,2,2'-uzinobis-(3-ethylbenzthlazoliue-6-sulphonicacid); IF, interfem~; PBS. phosphale-bufferedsaline; CFA. complete Freunds adjuvant.

acid polypeptide processed from a 233 amino acid precursor (Wang et al., 1985). It shares approximately 30% homology with human TNF/~ (Lymphotoxin) (Penniea et al., 1984). There is no evidence for a larger family of TNF related proteins or any homology with any other cytokine. TNFa was originally known as cachectin (Carswell et al., 1975) and is a monocyte/macrophage derived protein with cytotoxic or cytostatic properties (Tavcrne et al., 1984). It is a cytokine with pleiotrophic biological activities such as inhibition of lipoprotcin iipase (Beutler et al., 1985a), activation of polymorphonuclear leukocytes (Shalaby et al., 1985; Tsujimoto et at., 1986), stimulation of bone resorption and inhibition of

216 bone tormation (Bertolini et al., 1986) and stimulation of fibroblast growth (Sugarman et al., 1985; Vilcek ct al., 1986). TNFa, either alone or in synergy with other factors, is thought to be involved in weight loss (Beutler and Cerami, 1986), pyrexia (Dinarello et al., 1986) and to be a major mediator of endotoxic shock (Tracey et al., 1986). More recently it has been implicated in renal (Maury and Teppo, 1987) and cardiac allograft rejection (Chollet-Martin et aL, 1990). Although cytolytic bioassays have been used for the detection of TNFa these are not specific, since TNF~ may also cause cytolysis in bioassays, probably through binding to the same cellular receptor (Aggarwal et al., 1985). More recently radioimmunoassays (Teppo and Maury, 1987) and enzyme-linked immunosorbent assays (Bringham and Aggarwal, 1987) have been developed for the specific detection of TNFa, but with a lower detection threshold of approximately 81) pg/ml, these lack the sensitivity of the bioassay. Here we de.scribe the development of a TNFa assay which can be used as an ELISA or enhanced chernihminescence assay with a lower detection threshold of under 8 pg/ml or 4 pg/ml respectively.

Materials and methods

Cytokines Recombinant (E. coli-derived) human TNFa was a kind gift from Asahi Chemicals, Japan, The rIL-la, rIL-l~, rlL-2, rIL-3, rlL-4, rlL-6, IFN-y, rTNF-, and rTNFI3 were purchased from British Biotechnology (UK). The I F N ~ was purchased from Wellcome Reagents and the natural human TNFa was purchased from Genzyme (UK).

detection of hyhridoma cells with HAT medium, the cultures were screened by an ELISA. Plates coated with rTNFa (1 9.g/ml) were incubated with hybridoma supematants and the murine antibodies captured were detected with a peroxidase conjugated rabbit anti-mouse antiserum. Positive cultures were subcloned by limiting dilution and monoclonal antibody produced by inoculation of hybridoma ceils into pristane primed BALB/c mice and purified from the resultant ascitic fluid by ammonium sulphate precipitation. The isotypes of the monoclonal antibodies were determined using a commercially available kit (Serotec, UK), Polyclonal antiserum was obtained by immunising NZW rabbits s.c. at several sites with rTNFa at a total dose of 50 ~g in CFA. Booster injections of 50 ~g rTNFa were given in IFA at four weekly intervals and the sera evaluated for TNFa activity in a sandwich ELISA.

Specificity of the monoclonal antibody: clone 5-2 Monoclonal antibody clone 5-2 was tested by Dr. S. Hopkins (Manchester, UK) for its ability to block the biological activity of TNFa in the L929 bioassay. The ability of the assay to detect T N F a / T N F a receptor complexes was tested. Briefly, five sera, with initial concentrations of TNFa of 36, 430, 740, 1020 and 1080 pg/ml, known to contain high levels of free TNFa receptor (free TNFa p55 and p75 receptor concentrations were measured by Dr. Gallati, Basel, Switzerland) were spiked with 250, 500 and 1000 pg/ml of rTNFa. These were incubated at 37°C for 1 h and the recovery of the spiked TNFtx measured using the chemilumimetric assay.

Sample collection and storage Production of monoclonal and polyclonal antibodies Female BALB/c mice were immunised intraperitoneally with 10 p,g rTNFce emulsified with CFA; booster injections of 10 /zg were given 4 and 8 weeks later. A final intravenous immunisation of 25 #g rTNF in isotonic saline was given 3 days before fusion. Spleen cells were fused with P3/X63Ag8.653 mouse myeloma cells using 45% polyethylene glycol 1500 and 10% dimethyl sulfox.;de (Galfr~ and Milstein, 1981). Following the

Whole blood was collected in blood tubes (Sarstedt Z/10) and allowed to clot. Serum was removed within I h and stored in multiple aliquots at -80°C until assayed, To test for in rive production or degradation during the period after collection but before storage the following experiment was conducted. Blood was collected from ten healthy volunteers, five of which were spiked with rTNFa ranging from 60 to 250 pg/ml. The samples were left at room temperature and aliquots taken at various time points.

217

Assay technique A single horseradish pcroxidase based assay was developed in which the endpoint could be measured with ',he use of either a colorimetdc or chemiluminescence substrate. The general assay is given below together with any substrat¢ dependent variations. The monoclonal anti-human TNF~ (clone 5-27 was adsorbed to the wells of immunoassay plates in 0.05 M carbonate buffer pH 9.6 (1 #g/rot, 100 ~l/well) for a minimum of 12 h at 4°C. The plates selected were lmmulon 2 for the ELISA and MicroFLUOR ¢W' for the luminescent assay (Dynatech Laboratories, USA). The plates were washed five times in wash buffer (0.1 M PBS, pH 7.2, 0.05% Tween 20) beiore and after being blocked with assay buffer (0.1% bovine serum albumin in PBS 0.1 M, pH 7.2; 150 p.I/well) for I h at room temperature. The wells were filled with either samples; standards or controls (100 t~l/well in duplicate). The samples were either neat serum, plasma, urine or tissue culture medium. The standard curve and controls were a pooled sample of the appropriate fluid from a series of TNFa negative controls spiked with rTNFa; background wells contained the appropriate fluid without rTNFa. The standard curve ranged from 3.9 pg/ml to 1 ng/ml rTNFc¢ in eight serial two-fold steps. The plates for the ELISA were incubated for a minimum of 12 h at 4°C, the luminescent assay plates were incubated for 4 h at room temperature, after which the plates were washed five times with wash buffer. The cross-reactivity of the polyclonal rabbit anti-human TNFo~ antisera to mouse immunoglobulins was blocked prior to use by a 30 rain incubation with mouse serum (10 #1/10 ml, 1/2500 antisera in assay buffer containing 10 -2 M dithiothseitol (DT]') at room temperature before its addition to the plate3) (100 td/well). After a 2 h incubation at room temperature the plates were washed five times with wash buffer. The peroxidase-conjugated donkey anti-rabbit IgG (heavy and light chainspecific) (Jackson immunorcsearch Labs, Stratech Scientific, OK) was diluted 1/2500 in assay buffer (the cross-reactivity to mouse immunoglobulins removed as described above) and added (100 ~l/w¢ll) to the plates. After a 1 h incubation at room temperature the plates were washed

five times in wash buffer. The ELISA nlates were developed for 15 rain at room tem~rature by the addition of 1 ram ABTS (2,2'-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid)/0.1 mM H20: in 0A M citrate phosphate buffer pH 5.0 and read on a I~natech Multiscan plate reader at a wavelength of 420 nm. The luminescent assay was developed using the Amerlite buffer and tablet packs (Amersham International, UK) and read on the Amedite enhanced luminescence microtiter plate reader (Amersham International, UK) as directed by the manufacturers.

Standardisation and quality control The standard curve was rTNFa spiked into the appropriate fluid and calibrated against the interim reference preparation obtained from the National Institute for Biological Standards and Control, OK (code no. 86/659), Two internal controls consisting of rTNFa at 25 and 500 pg/ml were assayed on each plate to monitor any possible drift of the standard curve.

ResuHs

Monoclonal antibodies A panel of monodonal antibodies reactive with TNFa was obtained. The monoclonai ~nti~_-qdy clone number 5-2 (IgG1 isotlq~) was ehoseh ;nr use in the assay after preliminary work had shown it to be most suitable. This antibody was shown not to neutralise the biological activity of TNFa in the L929 bioassa¥. Fig. i shows that in frye sera with high levels of free TNFa receptor (p55 and p75) the assay was able to detect spiked TNFa =l: 10%.

Development and characteristics of the TNFa assay The concentrations of the reagents and time of the incubation periods were optimiscd in preliminary experiments. The resultant assay protocol is given in the materials and methods section and typical standard curves for both the absorbance and cbemilumimetrie assays are shown in Fig. 2. Fig. 3 shows the effect of using different species sera as the diluent in the assay. None of these

218 were found to be suitable and so the diluent for the standard curve, controls and background of the assay was always matched with the sample type being analysed. These consisted of at least ten TNFa negative samples of serum, plasma or urine pooled together and stored at -80°C until used. The usual working range for the Amerlit¢ assay was 3,9-10~ pg/m] with the lowest detectable amount (mean of background + 3 SD) being 3,1 pg/ml TNFzt (mean of three assays). The background values of the Amerllte assay, defined here as the negative control of a percentagc of the top standard, was originally as high as 10%. This was discovered to be due to the crossreactivity of the rabbit anti-TNF