Journul of’ Virological Mrthod.s, 40 (1992) 77-84

(‘ 1992 Elsevier Science Publishers B.V. i All rights reserved / 0166-0934/92:$05.00

17

VIRMET 01398

A nylon membrane enzyme immunoassay for rapid diagnosis of influenza A infection

Gilles Duverlie”, Laurent Houbart”, Bertrand Visse”, Jean-Jacques Chomelb, Jean-Claude ManuguerraC, Claude HannounC and Jeanne Orfila”

” Virologk. CHU-Hcipiiul Sud. Amiem f Frunc~e) I ” Virologir, Fuulth dc MPdecinc, L,wn ( Fronw) and

’ Ecologk Virulr, Institut Pastcur. Paris (Franw)

(Accepted 19 May 1992)

Summary

A new membrane-enzyme immunofiltration assay (MIFA) was developed for rapid diagnosis of influenza A infection. The pretreated specimens were dispensed into a 1.2 pm BiodyneTM B nylon membrane-bottomed microplate and vacuum filtration was applied. Blocking solution, peroxidase-conjugated anti-influenza A nucleoprotein monoclonal antibody, washing buffer and substrate were added in that order. The assay was completed within 30 min. Out of 103 nasopharyngeal swabs collected in transport medium, 31 isolates of influenza A virus were obtained and 22 specimens were detected directly by the MIFA technique. The 9 isolation-positive MIFA-negative specimens required 6 days or more for viral detection in cell culture, and probably contained a very low quantity of virus. The 72 cell culture negative specimens were also negative by MIFA. Comparison with a classical immunocapture assay (ICA) gave a better sensitivity for MIFA, as only 15/103 specimens were positive by ICA. MIFA is a rapid test with 71% sensitivity and 100% specificity. It was also very useful to test the cell culture supernatants, as a sensitivity of 100% was obtained with MIFA when the immunofluorescence technique was positive. The same technique could be readily carried out on the same plate for other respiratory viruses since capture antibody is not used.

Influenza A; Membrane assay; Antigen detection

Correspondcwcr to: G. Duverlie. Virologie, CHU-HGpital Sud. X0054 Amiens Ccdex I, France.

78

Introduction

In 1984, an influenza surveillance system, the ‘GROG’ (Regional Influenza Surveillance Group) was set up in ‘Be de France’ and since 1988 has extended to many other regions (Hannoun et al., 1984). The main aim is to identify the viruses in circulation as quickly as possible and to detect antigenic variants, in order to adapt the composition of vaccines as efficiently as possible. The isolation of epidemic strains is essential for these studies. Rapid diagnosis of influenza is desirable to facilitate both vaccination and specific chemotherapy, particularly for inadequately protected high-risk subjects. A rapid diagnostic test, if reliable and inexpensive, would also be useful for large epidemiological surveys. ‘Immunofluorescence techniques have been widely used for nasophar- yngeal aspirates (Liu, 1956; Grandien et al., 1985; McQuillin et al., 1985; Pothier et al., 1986). However, these are not suitable for the examination of nasopharyngeal swabs if only few cells are available, or when the samples are frozen or not quickly processed in the laboratory. Moreover, these techniques are fastidious and time-consuming when many specimens and many respiratory viruses are tested. Immunocapture ELISA is a potentially sensitive, specific and rapid test to detect viral antigens when delayed examination is required of samples kept at various temperatures or frozen (Berg et al., 1980; Bishai’ et al., 1981; Sarkkinen et al., 1981; Pothier et al., 1988; Chomel et al., 1989; Hornsleth et al.? 1990). Automation is possible for immunoenzymatic techniques, or at least semi-automation for reading and calculation of the results, Generally, a minimum of 3 h is necessary to complete the test, and precoated plates are required. Recently, membrane enzyme immunoassays were reported to be useful for virai detection (Bode et al., 1984; Hornsleth et al., 1990; Fulton et al., 1988; Waner et al., 1991). These have a high capture capacity and allow the development of very rapid tests, particularly with immunofiltration enzyme immunoassays. A rapid membrane enzyme immunoassay to detect influenza A nucleoprotein antigen within 30 min is described (MIFA). The viral antigens were non-specifically trapped on a nylon membrane and detected using peroxidase-conjugated mouse monoclonal antibodies. Objective photometric readings were obtained.

Materials and methods

During an outbreak of influenza A infection in the winter of 1989/1990, physicians collected 103 nasopharyngeals swabs in vials containing 2 ml of tryptose-phosphate supplemented with 0.5% gelatin and antibiotics as the transport medium. Patients were between 1 and 64 years old (26 & 16). Samples were mailed to the laboratory at ambient temperature and received after an average of 1.4 f 0.6 days. When received, cell culture isolations were

carried out and the remaining samples were frozen (-70°C). Samples were thawed only once before testing by the immunocapture and membrane immunoassays. The cell culture supernatants of the 3 1 isolates were also tested. One was diluted in the transport medium for use as a positive control and for testing the reproducibility of the immunoassays. The transport medium was used as a negative control. Eight nasopharyngeal aspirates positive for respiratory syncytial virus, as tested by immunofluorescence technique, were also tested by MIFA for influenza A virus.

~solatjon of viruses

When received, each specimen was inoculated onto monolayers of Madin- Darby canine kidney cells (MDCK) maintained in a serum-free medium containing 2 pg of trypsin per ml (Frank et al., 1979). Inoculated cultures were incubated at 34°C and viewed daily up to a maximum of 2 weeks to detect morphological changes. Every second day, a haemagglutination test was carried out and the type of influenza virus isolate was identified by indirect immunofluorescence using monoclona1 A and B antibodies (IA-52 and IB-82, Biosys, France) applied to cells scraped from culture tubes (Pothier et al., 1986). These monoclonal antibodies have a good specificity since they are directed against nucleoprotein antigens, which are type-specific and highly conserved (Pothier et al., 1989).

Itnmunocapture assay

This classical immunocapture assay (ICA) was described previously (Pothier et al., 1989). Briefly, 100 ~1 of rabbit anti-influenza A or anti-influenza B polycfonal antibodies diluted in carbonate buffer 0.05 M (pH 9.6) were dispensed into microplates containing wells of Immulon I Removawell (Dynatech). After an overnight incubation at 4°C the following reagents (100 ~1) were added in successive steps, each step preceded by a wash of the wells with phosphate-buffered saline pH 7.4 (PBS) containing 0.05% Tween 20, and each reagent was incubated for 1 h at 37°C: undiluted sample; anti-A (IA- 52) or anti-B (IB-82) nucleoprotein monoclonal antibodies diluted in PBS- Tween 20 containing 1% bovine serum albumin; biotinylated anti-mouse IgG immunoglobulin and peroxidase-conjugated avidin D (BA-2000 and A-2004, Vector Laboratories, CA). Following a final wash, the ortho-phenylenediamine substrate was added and the reactions stopped with sulphuric acid after 15 min at room temperature. Readings of absorbances were carried out at 490 nm, and the ratios were calculated by dividing the absorbance of each sample by the absorbance of the negative control. Samples with absorbance equal to or greater than 2 times the negative control absorbance value were considered to be positive.

80

Membrane immunqfiltration asa?> f MIFA /

The specimens (160 ,ul) were pretreated by adding the same volume of a reagent containing 50 mM Tris, 150 mM NaC1, 1 mM EDTA, 1% N- acetylcysteine, 2.5% Empigen BBTM detergent. The mixture was vortexed for 30 seconds and processed 3 min later. The mixtures and similarly treated controls (positive and ne ative) were dispensed into a 96-well membrane- bottomed plate. Biodyne &I B Silent MonitorTM plates (1.2 pm) and the appropriate vacuum filter holder (TV-l for filtration and transfer) were purchased from Pall Biosupport, France. Vacuum filtration was performed at an operating pressure of 120 mmHg. For each successive step, one volume of reagent was added and sucked through before stopping the vacuum and incubating with an equal volume of the same reagent. Accordingly, 300 ,~l of PBS containing 1% casein and 0.2% Tween 20 was added after the filtration of the treated specimens, sucked through, and a further equal volume added and then incubated for 8 min to saturate the membrane. The PBS-Casein-Tween 20 was sucked through and then a volume of 100 ,uI of peroxidase-conjugated monoclonal antibody (IA-52) was added and sucked through. Vacuum was stopped and a further equal volume of the conjugated monoclonal antibody was dispensed and incubated for 8 min. Six washings with PBS-Tween 20 and a first volume of 100 ,uI of ortho-phenylenediamine substrate were then sucked through. One further equal volume of substrate was incubated for 8 min and the reaction stopped by adding 100 ,~l of H2S04 1N. The contents of the plates were then transferred by filtration to a flat bottomed plate. One volume of 100 ~1 of NaCl 150 mM was finally added to wash through all the coloured substrate solution. Plates were read at 490 nm. The ratios were calculated as already described and the samples with absorbance equal to or greater than 2 times the negative control absorbance value were considered to be positive.

Results

Thirty-one (311103) specimens were identified as influenza A positive by cell culture technique (30.1%), 22 (21.3%) by MIFA and 15 (14.6%) by ICA. The average time for viral isolation was 4 + 2.8 days. Five specimens were found to be positive for influenza B virus by cell culture, identified by immunofluor- escence and ICA. These 5 specimens were negative by influenza A immunoassays. The 8 nasopharyngeal aspirates, positive for respiratory syncytial virus by immunofluorescence, were negative by MIFA for influenza A. All the 72 culture negative specimens were also negative by both immunoassays, the specificity of which was therefore of 100%. Sensitivity of the immunoassays as compared to the cell culture isolation was 48% (15/3 1) for ICA and 7 1% (22/3 1) for MIFA (Table 1). One specimen was ICA positive and MIFA negative and 8 specimens were ICA negative and MIFA positive. Statistical analysis of paired data by McNemar’s test is significant (P ~0.05).

TABLE I

Results of the immunoassays with the samples positive by cell culture isolation”

ICA MIFA

Positive 15h 22’ Negative 16 9 Sensitivity 48% 71%

“31 culture positive samples (31/103). The 72 culture negative samples were also negative with both the immunoassays (specificity: 100%). bl (lCA+ MIFA-). ‘8 (MlFA + , ICA-).

but only 9 discordant pairs were observed (Armitage, 1971). Borderline ratio values of 1.630, 1.787 and 1.775 were observed for 3 ICA negative and MIFA positive specimens. The specimen ICA positive and MIFA negative had a MIFA ratio of 1.598. The mean of the ratios was 2.840 f 0.457 for ICA and 4.486 + 0.868 for MIFA. The mean of the differences between each pair of ratio (MIFA minus ICA) was 1.647 f 0.749, and therefore significantly different from zero (P ~0.03). Correlation coefficient between ratio of ICA and MIFA was 0.506 (P = 0.04). Results of immunoassays are related to the isolation time (t-test; P = 0.01). Positive ICA and MIFA assays had a mean time for isolation of 2.8 It 2.6 days and 3.3 i 2.5 days, respectively. Negative ICA and IFA assays had a mean time for isolation of 5.3 + 2.5 days and 6 + 2.6 days, respectively. Cut-off of the enzyme immunoassays was reevaluated with culture negative samples. The mean of the ratios was 1.102 + 0.288 and 1.144 f 0.254 for ICA and MIFA, respectively. Therefore, the cut-off values were about the means of the negative samples plus 3 standard deviations (p + 3 S.D.). The reproducibility of MIFA was evaluated with a low positive control. The coefficient of variation of the ratios was 7.8% (R = 30).

Discussion

MIFA is more sensitive than a classical immunocapture assay when compared to a cell culture technique for influenza A viruses. The same monoclonal antibodies were used, but the mode of capture of the viral antigens with each of the two methods was different. A specific polyclonal capture antibody was used for ICA while in MIFA the membrane only non-specifically trapped the viruses. The nylon membrane offers a total surface area about 100 fold higher than the coated wells, and this large surface in the MIFA should be sufficient to ensure a greater capture of the viral antigens. In a preliminary step we also examined membrane coating with capture antibodies, but better results were not found. The membrane is also positively charged by a high density of quaternary ammonium groups making it strongly cationic to bind the proteins. Correlation between the ratios of the two techniques and final results were not very good (Y = 0.506) and the sensitivity was 71% for MIFA and 49% for ICA

82

using nasopharyngeal swabs when compared to a reference cell culture isolation. A comparable sensitivity would be expected, whatever the subtype of epidemic strains, as shown by immunofluorescence studies with IA-52 monoclonal antibody (Pothier et al., 1986). The enzyme immunoassays showed high specificities (100%) probably due in part to their moderate sensitivity and no-cross reaction was observed with the 5 influenza B cell eulture positive specimens or the 8 respiratory syncytial virus positive samples. This is not surprising since the monoclonal antibodies have already been shown to give good specificity (Pothier et al., 1986 and 1989). The sensitivities of both MIFA and ICA would probably have been better if nasopharyngeal aspirates were tested, because larger quantities of viral antigens in the specimens would have been obtained. Unfortunately, such samples are not available from our sentinel doctors network, because they are time-consuming and not readily accepted by patients. Hornsleth et al. (1990) obtained a sensitivity and a specificity of 88% and loo%, respectively, with an immunocapture assay similar to that described in this study (ICA), using only nasopharyngeal secretions containing epithelial cells. Similarly, with an immunocapture assay, Chomel et al. (1989) reported a sensitivity of 78%, by testing nasal swabs and aspirates collected mostly from children. They also detected positive immunocapture samples, confirmed by blocking assay, which were not found to be positive by cell culture. As in the present study, all the immunocapture assays had a high specificity. Using essentially nasopharyngeal washes processed within 4 h, the sensitivity obtained by Waner et al. (1991) was 100% with the DirectigenTM test (an enzyme immunoassay membrane test), but the specificity was only 88%. For MIFA, the cut-off could not be lowered to obtain a sensitivity near 100% without reducing dramatically the specificity. Negative ICA and MIFA results, with positive cell culture isolation results, were probably related to a low quantity of virus and viral antigens in the specimens, as suggested by the isolation time delay; certainly for MIFA, all but one were isolated after 6 days. As specimen collection quality was of paramount importance in the different studies, the parallel comparison reported here between MIFA and ICA emphasizes the importance of membrane enzyme immunoassays, particularly for rapid diagnosis. While a time-resolved fluoroimmunoassay has been described as having a very high sensitivity, it requires expensive fluorometric equipment and conjugates are not readily available (Walls et al., 1986; Nikkari et al., 1989). For MIFA, the efficient pretreatment of the specimens ensured good filtration with a membrane of 1.2 pm pore size. Preliminary assays with homogenized aspirations were also satisfactory. Classical non-ionic and zwitterionic detergents were tested, but finally we found Empigen BB in conjunction with N-acetylcysteine to be very efficient for the technique used and any difficulties with filtration were essentially avoided (Miller et al., 1986; Jennings et al., 1990). Reproducibility of MIFA was found to be acceptable for an immunoassay (C.V. = 7.8%). A grey zone of 1.8 to 2.2 around the cut-off of the ratio could be considered. Quantitative results are not essential for the

83

diagnosis, but the objective reading improves the quality of the test. Moreover, MIFA has many advantages as compared to an immunocapture assay. Capture antibodies are not required and solid phase preparation is not necessary before the test. Other respiratory viruses could be tested easily in the other wells, without the requirement of a combination of different solid phases or large series of specimens. The complete run is shorter (30 min vs. 4 h 15 min), and washing apparatus was not required. The 96”well membrane-bottomed plates are ready to use, well adapted for rapid diagnosis, and the equipment is not expensive. The transfer system is efficient and simple. Cross-contamination was not encountered. Supernatants of cell culture can also be tested rapidly by this technique. In this study a labelled specific monoclonal antibody was used to reduce the time required, but a two-step detection would be possible using a biotin-streptavidin system. This system would be longer (45 min vs. 30 min), but a better sensitivity could probably be expected and several respiratory viruses or other antigens could be tested at the same time without requiring a battery of labelled monoclonal antibodies. Further parallel studies are also necessary to assess the sensitivity of MIFA compared to the cell culture technique, especially with regard to the nature of the clinical samples. In this study, MIFA provided a rapid diagnosis in 71% of the clinical samples, but cell culture isolation seems to be always necessary for a definitive result (and also for the study of the epidemic strains). In this way, MIFA could also be a very simple technique for follow-up of the cell culture supernatants.

Acknowledgements

We wish to thank all the sentinel doctors and the GROG members of Picardie.

References

Armitage, P. (1971) Statistical Methods in Medical Research. Blackwell, Oxford. Berg, R.A., Rennard, S.I., Murphy, B.R., Yolken, R.H., Dolin, R. and Straus, SE. (1980) New

enzyme immunoassays for measurement of influenza A/Victoria/3/75 virus in nasal washes. Lancet i, 851-853.

Bishai’, F.R., Galli, R. and F&on, R.E. (1981) Enzyme-linked immunosorbent assay for detection and identification of influenza viruses A and B in clinical specimens. Lancet ii, 756-757.

Bode, L., Be&in, L. and Kohler H. (1984) Nitrocellulose enzyme-linked immunosorbent assay (NC- ELISA): a sensitive technique for the rapid visual detection of both viral antigens and antibodies. J. Viral. Methods. 8, 11 l-121.

Chomel, J.J., Thouvenot, D., Onno, M., Kaiser, C., Gourreau, J.M. and Aymard, M. (1989) Rapid diagnosis of influenza infection of NP antigen using an immunocapture ELISA test. J. Virol. Methods. 25, 81-92.

Frank, A.L., Couch, R.B., Griffis, A. and Baxter, B.D. (1979) Comparison of different tissue culture for isolation and quantitation of influenza and parainfluenza viruses. J. Clin. Microbial. 10, 32- 36.

84

F&ton, R.E.. Wong, J.P.. Siddiki, Y.M. and Tso, M.S. (1988) Sensitive tluorogenic enzyme immunoassay on nitrocellulose membrane for quantitation of virus. J. Virol. Methods. 22, 149 164.

Grandien, M., Petterson. C.A., Gardner. P.S., Linde, A. and Stanton, ‘4. (19X5) Rapid viral diagnosis of acute respiratory infections: comparison of enzyme-linked immunosorbent assay and the immunofluorescence technique for detection of viral antigens in nasopharyngeal secretions. J. Clin. Microbial. 22. 757 -760.

Hannoun. C., Dab, W. and Cohen, J.M. (1989) A new influenza surveillance system in France: The Be de France ‘GROG’.]. Principles and Methodology. Eur. J. Epidemioi. 5, 285-293.

Hornsleth, A. and Jankowski, M. (1990) Sensitive immunoassay for the rapid diagnosis of influenza A virus infections in clinical specimens. Res. Viral. 141, 373 ~384.

Jennings, M. and Erturk, M. (1990) Comparative studies of HSV-I antigens solubilised from infected cells by using non-ionic or zwitterionic detergents. J. Med. Viral. 3 1. 98- 108.

Liu, C. (1956) Rapid diagnosis of influenza infection from smears by means of fluorescein Iabelled antibody. Proc. Sot. Exp. Biol. Med. 92. 883.887.

McQuillin, J., Madeley, CR. and Kendal, A.P. (198.5) Monoclonal antibodies for the rapid diagnosis of influenza A and B virus infection by immunofluorescence. Lancet ii, 911 914.

Miller, H.R., Phipps. P.H. and Rossier. E. (1986) Reduction of non-specific fluorescence m respiratory specimens by pretreatment with N-acetylcysteine. J. Clin. Microbial. 24. 470-471.

Nikkari, S., Halonen, P., Kharitonenkov, I.. Kivivirta, M., Khristova, M., Waris, M. and Kendal. A. (1989) One-incubation time-resolved lluoroimmunoassay based on monoclonal antibodies in detection of influenza A and B viruses directly in clinical specimens. J. Virol. Methods. 23. 29 39.

Pothier, P.; Denoyel, G.A.. Ghim, S., Prudhomme de Saint Maur, G. and Freymuth, F. (1986) USC of monoclonal antibodies for rapid detection of influenza A virus in nasophary~~eal secretions. Eur. .I. Clin. Microbial. 5, 3366339.

Pothier, P., Bow. J.B., Lhuiliier, M. and Hannoun, C. (1988) Detection rapide des antigenes grippaux dans les produits pathologiques a I’aide d’anticorps monoclonaux. Rev. Franc. Lab. 169, 7- 10.

Sarkkinen. H.K.. Halonen, P.E. and Salmi, A.A. (1981) Detection of influenza A virus by radioimmunoassay and enzyme-immunoassay from nasopharyngeal specimens. J. Med. Viral. 7. 212-220.

Walls, H.H., Johansson, K.H.. Harmon. M.W., Halonen, P.E. and Kendal, A.P. (1986) Time resolved fluoroimmunoassay with monoclonal antibodies for rapid diagnosis of influenza infection. J. Clin. Microbial. 24, 907-912.

Waner, J.L., Sarah. J.T., Shalaby, H.. Murphy, P. and Wall, L.V. (1991) Comparison of directigen Flu-A with viral isolation and direct immunofluorescence for the rapid detection and identi~~ation of influenza A virus. J. Clin. Microbial. 29. 479 -482.