Journal of Medical Virology 22:67-76 (1987)

An Enzyme Labelled Nuclear Antigen lmmunoassay for Detection of Cytomegalovirus IgM Antibodies in Human Serum: Specific and Non-Specif ic Reactions C.M. Nielsen, K. Hansen, H.M.K. Andersen, J. Gerstoft, and B.F. Vestergaard

Departments of Enterovirus (C. M. N., €3. E V.) and Rubella (K. H., J. G.), Statens Seruminstitut, Copenhagen, and Institute for Medical Microbiology, Rrhus University, Rrhus (H. M. K. A.), Denmark

A p-capture enzyme linked immunosorbent assay was developed for detection of lgM antibody t o cytomegalovirus (CMV). Virus-specific IgM was detected using horseradish peroxidase labelled nuclear CMV antigen (CMV-ELA). False-posi- tive reactions caused by Paul-Bunnell-Davidsohn (PBD) positive sera and anti- nuclear antibody (ANA) positive sera were identified in a combination assay employing enzyme labelled nuclear control antigen (CO-ELA) in parallel to the CMV-ELA. Four of five PBD positive and 30 of 31 ANA pc)sitive sera reactive with the CMV-ELA were identified as false positive reactions in the combined ELA-assay. The reactivity in PBD-positive sera could not be explained by anti- genic cross reactivity between CMV and Epstein-Barr virus. and the results further suggested that different cell specified components of the CMV-ELA were respcn sible for the reactivity of PBD-positive as compared to ANA-positive sera. One of 314 healthy blocxi donors. 12 of 12 patients with primary CMV infection. and 1 1 of IS patients with secondary CMV infection had detectable CMV IgM antibodies. Comparison of different CMV-ELAs revealed that pronounced differ- ences in specificity as well as sensitivity may exist.

Key words: cytomegalovirus. nuclear antigen, CMV IgM antibodies, enzyme immunoassaj

INTRODUCTION

A p-capture irnrnunoassays employing enzyme labelled antigens (ELAs) for the detection of CMV-specific IgM were first described by Schrnitz et al [ 1980) and van Loon et a1 [ 19811. ELAs have also been applied for IgM detection in other viral

Acccptcd lor publication Septembcr I . 1986

Address reprint requests to Carsten MBllcr Nielsen. Enterovirus Department. Statens Scruniinstitut. 80. Arnagcr Boulevard. DK 2300 Copcnhagcn S. Denmark.

(c) 1987 Alan R. Liss, Inc.

68 Nielsen et al

infections and for detection of other antibody classes [Schmitz, 1982; van Loon et al, 1985a.b; Bonfanti et al, 19851. ELA-assays differ from indirect assays and conven- tional p-capture assays in that they are not subject to false positive reactions with sera containing IgM rheumatoid factor [Schmitz et al, 1980; van Loon et al, 19811 and they possess the additional advantage of not being influenced by epitope blocking from specific IgG, which may cause false negative results in indirect assays [Cohen et al, 19761.

The ELA principle also differs from other test principles in regard to the control of reaction specificity. In the indirect assay [Cappel et al, 1978; Kangro, 19801 a control antigen may be included for identification of non-specific reactions, and the conventional p-capture immunoassay employs a virus specific indicator antibody selectively recognizing viral antigen bound to IgM on the solid phase [Yolken and Leister, 1981; Sutherland and Briggs, 19831. The ELA principle, in contrast, pre- cludes the possibility of including controls for reaction specificity in the assay proper. The specificity attained with this technique is thus entirely dependent on the specificity of the labelled antigen.

We describe an ELA-immunoassay employing enzyme labelled nuclear CMV antigen for the detection of specific IgM antibodies in serum. The utility of this assay for the serological diagnosis of postnatal CMV infections was evaluated by examina- tion of sera from cases of primary and secondary CMV infections as defined by complement fixation (CF) serology. The specificity of the assay was evaluated by examination of PBD-positive and ANA-positive sera.

A second immunoassay employing enzyme labelled control antigen was devel- oped for the analysis of possible reactions with cell encoded labelled constituents of the nuclear antigen. The potential of this assay to identify sera giving spurious reactions with the CMV-ELA was investigated.

MATERIALS AND METHODS Cell and Virus Propagation

Human fetal lung fibroblasts were propagated in Roux bottles in Eagle’s MEM supplemented with antibiotics and 10% fetal calf serum (FCS) for outgrowth and 2% FCS for maintenance. CMV, strain AD 169, from infected culture medium was inoculated onto confluent monolayers of fibroblasts using a multiplicity of infection of approximately 0.1. Culture media were changed on the day after inoculation. Cells were harvested 6-8 days postinoculation, when a nearly complete cytopathogenic effect (CPE) had developed. Uninfected fibroblasts were harvested by the same procedure.

Preparation of Enzyme Labelled Nuclear Antigens

Nuclei from CMV infected and uninfected cells were prepared by modification of previously reported procedures [Stagno et al, 1978; Mintz et al, 19801. Cells were detached by trypsin-versene treatment, washed once in PBS pH 7.4 and incubated at 0°C in double distilled water made 2% with FCS, using 1 ml per Roux bottle. After 15 minutes 0.5% v/v NP 40 was added and the suspension was homogenized in a tight fitting Dounce homogenizer until the nuclei appeared freed of cytoplasmic remnants (20-40 strokes). The nuclei were separated from cytoplasmic debris by pelleting through Ficoll 400 (Pharmacia) 22.5% w/v in PBS pH 7.4 at 1800 x g for

Enzyme Immunoassay for CMV IgM 69

15 minutes at 4°C. The pellets were washed once in PBS pH 7.4, resuspended in 0.01 M carbonate-bicarbonate buffer pH 9.5 using 0.25 ml per Roux bottle, and sonicated on an icebath. The antigen was finally clarified at 8OOO X g for 20 minutes at 4°C. The supernatant was stored at -70°C or used immediately for enzyme labelling.

Nuclear antigens were pre-adjusted to a concentration of 0.5-1 .O mg protein per ml [Bradford, 19761 and labelled with horseradish peroxidase (RZ > 3.0, Kem- En-Tec. Copenhagen, DK) using 300 pyrogallol units per mg protein, but otherwise according to the procedure of Wilson and Nakane [ 19781. ELAs were kept at 70°C in PBS pH 7.4 made 2.5% v/v with FCS, 0.05% w/v with sodium azide, and 50% v/v with glycerol. Optimal antigen prepared from one Roux bottle provided enzyme labelled antigen sufficient for 4000-6000 test wells.

Sera

Serum specimens included in the study were: 75 sera from blood donors known to be negative for CMV-specific CF-and neutralizing antibodies; 314 sera from unselected blood donors; 50 sera with heterophile antibodies as determined in the PBD-test; 97 sera with ANA as determined by indirect immunofluorescence; 105 sera from 33 patients showing sero-conversion or significant titre increase in the CF-test, and 78 sera from 35 patients presenting with elevated CF-titres (2 128).

lmmunoassay Procedures

Flat-bottomed polystyrene microtest plates (Immunoplate I , Nunc DK) were incubated overnight at room temperature (RT) with 100 p1 per well of rabbit Ig anti- human p-chain (code A 091, Dakopatts, DK) diluted 1:lOOO in PBS pH 7.4. The plates were washed 5 times in PBS, 0.05% v/v Tween 20 (PBS-Tw). Aliqouts of 100 pl serum diluted 1 : 100 in PBS-Tw, 1 % w/v BSA (PBS-Tw-BSA) were added followed by incubation for 1 hour at 37°C in a waterbath. The plates were washed as above and incubated for two hours at RT with 100 fil per well of enzyme labelled antigen diluted in PBS-Tw-BSA supplemented with 5 % FCS. The plates were washed again and allowed to react for 30 minutes in the dark at RT with orthophenylenediamine dihydrochloride and H202 as substrate. The reaction was terminated by addition of 1 M HzS04, 150 pl per well. The optical density at 492 nm (OD 492) was determined in a Multiscan photometer (Titertek) blanked with substrate.

Selected specimens were tested in duplicate against both enzyme labelled CMV antigen (CMV-ELA) and enzyme labelled control antigen (CO-ELA). CO-ELAs were selected and preadjusted to match CMV-ELAs in a reactivity with known spuriously reactive ANA sera and a pool of sera from blood donors serving as negative control.

Determination of Base Line Reactivity and Expression of Results

Sera from 75 blood donors repeatedly negative (2 2 years) for CMV-specific CF- and neutralizing antibodies were examined. The mean OD 492 was 0.057, range = 0.031-0.095 and SD = 0.011. A serum pool matching the average OD 492 of these 75 sera was used as negative control in the assay.

Results were expressed as S/N ratios calculated as the mean OD 492 from wells incubated with sample (S) divided by the mean OD 492 of four negative control (N) wells incubated with the negative control serum pool. S/N ratios were calculated separately for the CMV-ELA and CO-ELA reactions. The reaction specificity was

Sera were inactivated for 30 minutes at 56°C and kept at -20°C.

70 Nielsen et a1

assessed by calculation of a specificity index (SPI). The SPI was defined as the S/N ratio obtained with CMV-ELA divided by the S/N ratio with CO-ELA. Only samples giving a CMV-ELA S/N ratio 2 3.0 combined with a SPI 2 3.0 were considered positive for CMV-specific IgM antibodies.

One of 314 unselected blood donor sera (0.3%) gave a positive response according to these criteria.

RESULTS CMV-Specific lgM as a Marker of Postnatal CMV Infection

The sera from patients with serological evidence of CMV infection were all examined for CMV-specific IgM. Thirteen cases (30 sera) presenting with elevated CF titers but without significant titer increase and without IgM antibodies were excluded as insufficiently documented. The remaining 55 patients were classified as primary, secondary, or undefined types of infection according to the following crite- ria: seroconversion in the CF-test was considered indicative of primary infection. Pre-existing CF antibodies but absence of IgM followed by a significant CF titer increase and/or appearance of IgM was considered indicative of secondary infection (reactivatiodreinfection). Cases displaying both CF and IgM antibodies in the first available serum specimen were classified as undefined IgM positive infections (Ta- ble I ) .

The temporal IgM response in primary infections is shown in Figure 1 . Twelve of 13 cases with seroconversion developed IgM antibodies. The single case without IgM was considered a false serwonversion due to passively acquired CF-reactive antibodies induced by plasma exchange therapy. In three cases IgM appeared prior to CF seroconversion preceeding this by I to 3 weeks. Strong IgM responses appeared in most cases coinciding with seroconversion. The six patients from whom late convalescent sera (beyond 10 weeks) were available displayed marked differences in the decline of IgM, which in three cases remained detectable for more than 6 months.

Eleven of 15 cases with secondary CMV infection mounted detectable IgM antibodies coinciding with the CF titre increase, while four remained negative. The IgM responses were generally weak and of short duration (mean 3 weeks, range 2-10 weeks).

The IgM response of the first IgM-positive sample from cases of primary, secondary, and undefined IgM positive infection sera shown in Figure 2. The group of primary infections showed stronger responses (mean S/N = 22.6, range = 5.7- 37.5) than the IgM-positive secondary infections (mean S/N = 5.9, range = 3.0- 17.0), but the two groups had overlapping values. The group of undefined infections,

TABLE I. Classification of Patients According to CF- and IgM Results

Typc o f Number Number with intection of cases CMV-specific IgM -__ _- Primary 12 (13)" 12 Secondary 15 1 1 Undcfined IgM pos. 27 27

"For cxplanation see text.

Enzyme Immunoassay for CMV IgM 71

z

t ..

!

i I

I - 2

1 0 10 20 30 WEEKS

PRIMARY SECONDARY UNDEFINED INFECTIONS INFECTIONS INFECTIONS

N.12 N i l 1 N = 2 7

Fig. 1. CMV-specific IgM reactions in serial sera from 12 patients with primary CMV infection. Weeks = interval in weeks from the collection of the sample preceeding seroconversion in the CF test.

Fig. 2 . CMV-specific IgM reactions in the first IgM positive sample in different types of CMV infection.

TABLE 11. Reactivity of CMV-ELAs With PDB-Positive and ANA-Positive Sera

Working PBD-positive sera ANA-positive sera CMV-ELA dilution Examined Reactive (%) Examined Reactive (%)

a 1500 50 21 (54) I1 14 (18) b 1:1500 50 7 (14) I1 26 (34) C 1:1600 41 5 (11) 91 31 (32)

encompassing both primary and secondary infections, spanned the entire range of reactivities covered by the former groups (mean S/N = 25.1, range 3.1-40.3).

Reactivity of Enzyme Labelled IgM Antigens With PBD-Positive and ANA-Positive Sera

PBD-positive and ANA-positive sera were examined with three CMV-ELAs (a, b, and c) from different batches of nuclear antigen. The ELAs were employed in dilutions providing for similar responses with CMV-IgM positive reference sera and with the negative control. Results are given in Table II. A significant proportion of both PBD-positive and ANA-positive sera reacted with all three labelled antigens. However, the highest proportionate reactivities of the ANA-positive sera were re- corded with those ELAs with which the PBD-positive sera showed least reactivity and vice versa. The results obtained with CMV-ELAs a and b were compared quantitatively by means of scatter diagrams (Figs. 3 and 4), which illustrate the completely different response patterns obtained with the two ELAs, and that some sera gave strong responses.

72 Nielsen et a1

Fig. 3. Comparison of IgM reactions in PBD-positive sera examined with two different enzyme labelled CMV antigens.

Fig. 4. Comparison of IgM reactions in ANA-positive sera examined with two different enzyme labelled CMV antigens.

Specificity Assessment Using Enzyme Labelled Control Antigen PBD-positive and ANA-positive sera were examined in a combined ELA-assay

employing CMV-ELA (CMV-ELA c) and CO-ELA in dilutions providing for similar S/N ratios with sera known to give spurious reactions. Results are shown in Figure 5. Four of the five PBD-positive and 30 of the 31 ANA-positive sera reactive with the CMV-ELA were identified as false positively reacting in the combination assay.

A comparison of the CMV-ELA results with those obtained in the combined ELA-assay is shown in Figure 6 for 50 sera from serologically defined CMV infections. The two ELA methods correlated well, although the SPI-values of some sera were slightly lower than the corresponding S/N ratios. One specimen with a S/N ratio of 13.7 gave a SPI value of 0.8, indicating the reactivity with the CMV-ELA was non-specific. This sample was obtained from a patient assigned to the group of undefined IgM-positive infections and was the only false-positive reaction recorded in the clinical material.

Fluctuations in Specific Activity Between Different CMV Antigen Preparations

Figure 7 shows the results obtained by analysis of 65 sera from serologically defined CMV infections using two different batches of labelled CMV antigen. The two batches were prepared by identical procedures and had similar optimal working dilutions (CMV-ELAb 1:1500 and CMV-ELAc 1:1600). Fifty-two of the 65 sera gave higher S/N ratios with ELAc than with ELAb. Fourteen specimens displaying S/N ratios below cut-off level with CMV-ELAb were reactive with CMV-ELAc. One specimen of borderline reactivity with CMV-ELAb gave a reaction below cut-off with CMV-ELAc .

k

2 25 . z 0 U

20- 0

> z t 0 u- 2 15- v)

0 4

(0 F g 10-

Y - > I

F U

5-

3.0.-

Enzyme Immunoassay for CMV IgM 73

5 PBD-POSITIVE SERA ANA-POSITIVE SERA N=47 N=97

Fig. 5. Specificity assessment of IgM reactions in PBD-positive and ANA-positive sera by a combined immunoassay using enzyme labelled CMV antigen and enzyme laklled control antigen. SPI, specificity index.

t

Fig. 6. Comparison of S/N with specificity index (SPI) results of CMV IgM reactions in sera from case of CMV infection defined by CF serology.

Fig. 7. Comparison of the CMV IgM reactions obtained with two different enzyme labelled CMV antigens in sera from different types of CMV infection defined by CF serology.

74 Nielsen et al

DISCUSSION

The key findings of this study are two-fold: first, the demonstration of strong non-specific reactions and a solution for this problem in the form of the combined ELA assay employing CMV-ELA and CO-ELA in parallel; and second, the variation in specific activity between seemingly similar antigen preparations leading to differ- ences in specificity as well as sensitivity.

The finding that PBD-positive and ANA-positive sera frequently give spurious reactions with enzyme labelled nuclear CMV antigens is somewhat at variance with the high specificity reported by others [Schmitz et al, 1980; van Loon et al, 19811. This discrepancy could be explained by the smaller numbers of sera included in these studies and in the study of van Loon et al [ 19811 ANA-positive sera were not included. Furthermore Schmitz et al [ 19801 prepared their ELAs from antigens semipurified by density gradient centrifugation thereby probably improving the specificity of the labelled antigen.

Addition of unlabelled control antigen in excess of the specific probe has also been employed to improve the specificity of ELA-assays [Schmitz, 1982; van Loon et al, 1985b]. However, neither density gradient centrifugation nor addition of control antigen were capable of securing the specificity with the sera included in this study (data not shown). We therefore aimed at identifying non-specific reactions rather than attempting to eliminate them. The combined ELA-assay developed for this purpose proved to be an efficient safeguard against false-positive reactions.

Several authors have reported that PBD positive sera may react in immunoas- says for CMV-specific IgM detection [Schmitz et al, 1972, 1980; Schmitz 1982; Rasmussen et al, 1982; Sutherland and Briggs, 1983; Kangro et al, 19841. A cross- reaction at the level of the virus-specific IgM has been suggested as the underlying mechanism [Schmitz, 19821. Our data, however, do not support this supposition. First, the same group of PBD-positive sera displayed considerable differences in the proportion of specimens reactive with different CMV-ELAs, although the ELAs were used in dilutions providing similar responses with IgM positive reference sera. Second, only one of the five PBD-positive sera reactive with the CMV-ELA gave a CMV-specific IgM reaction in the combined ELA-assay while the remaining four were identified as spurious reactions. These observations indicate that IgM antibodies against cell encoded antigens cause the vast majority of the reactions in PBD-positive sera rather than EBV-specific IgM antibodies cross-reacting to CMV. The compara- tive analysis of different CMV-ELAs further suggests that the IgM reactivity of PBD- positive sera is directed against epitopes different from those responsible for the reactivity of ANA-positive sera.

The variation in specific activity between different antigen preparations proba- bly reflects differences in the level of CMV-specified antigen expression in the viral cultures, which, even under similar culture conditions cannot be judged optimally by observation of CPE. The ratio of enzyme labelled CMV specified protein to total enzyme labelled protein--a key determinant of both specificity and sensitivity-may consequently vary substantially between different antigen preparations.

Our results concerning the diagnostic value of CMV-specific IgM are in good agreement with those obtained by others [Cappel et al, 1978; van Loon et al, 1981; Kangro et al. 1982; Pass et al, 1983; Sutherland and Briggs, 19831. IgM was detectable in all cases of true primary and in 73% of secondary infections. The latter

Enzyme lmmunoassay for CMV IgM 75

group was exclusively made up by immunocompromised patients, which might account for the high proportion of IgM positive cases.

Distinction between primary and secondary CMV infection is important for both epidemiological and prognostic reasons. Our data confirm that the magnitude of the IgM response is of limited value in this respect. Van Loon et al [ 1985b] recently reported that detection of CMV-specific IgE might be a better indicator for this purpose, since IgE was detected almost exclusively in sera from cases of primary infection.

The results of the present study show that the achievement of reliable results with enzyme labelled nuclear CMV-antigens depends on critical selection of antigens of high specific activity combined with efficient measures for controlling the assay specificity. Fulfillment of these conditions makes the ELA-assay a sensitive and specific technique for detection of IgM antibodies against CMV.

ACKNOWLEDGMENTS

We thank E. Ehlen Knudsen, the Bloodbank, M. Hbier-Madsen, the Autoim- mune Lab., and J . Henrichsen, the Streptococcal Dept., SSI, for valuable help in providing serum specimens. The technical assistance of Ms. B. Frederiksen and Ms. A.C. Pelving and secretarial help of Mrs. B. Rothstein is highly appreciated.

REFERENCES

Bonfanti C. Meurman 0. Halonen P (1985): Direct enzyme imrnunoassay for detection o f specific IgG antibcxfy t o rubella virus by use o f enzyme labelled antigen. Journal o f Virological Methods I I : 161 - 170.

Bradford MM ( 1976): A rapid and sensitive method for quantitatton of microgram quantitites of protein utilizing the principle o f protein-dye binding. Analytical Biochemistry 72:248-254.

Cappcl R. de Cuypcr F. de Brakeleer J (1978): Rapid detection of IgG and IgM antibtxlies for cytomegalovirus by thc enzyme linked immunosorbcnt y (ELISA). Archives of Virology 58:253-258.

Cohcn IR. Norius LC, Julian AJ (1967): Competition between and effectiveness o f IgG and IgM antibcxiics in indirect fluorescent antibody and other tests. Journal of Immunology 98: 143- 149.

Kangro HO ( 1980): Evaluation of a radioimmunoassay for IgM-class antibodies against cytomegalovirus. British Journal o f Experimental Pathology 61 :512-520.

Kangro HO. Griffiths PD. Huber IJ, Heath ARB (1982): Specific IgM class antibody production following infection with cytornegalovirus. Journal of Medical Virology 10:203-2 12.

Kangro HO. Btx)th JC. Bakir TMF. Tryhorn Y. Sutherland S (1984): Detection of IgM antibodies against cytomcgaloviruh: Comparison o f two radioimmunoassays. enzyme-linked immunosorhent assay and immunofluorescent antibody test. Journal of Medical Virology 14:73-80.

Mint1 L. Mincr RC. Yeager AS ( 1980): Anticomplement immunofluorescence test that uses isolated fibroblast nuclei for detection o f antibodies to human cytomegalovirus. Journal of Clinical M icrohiology 12 : 562-566

Pass RF. Griffiths PD. August AM ( 1983): Antibody respmse to cytomegalovirus after renal transplan- tation: Comparison o f patients with primary and recurrent infections. The Journal of Infcctious Disease5 147:40-46.

Rasmussen L, Kelsall D, Nelson R, Carney W, Hirsch M. Winston D. Preiksaitis AJ, Merigan TC ( 1082): Virus-specific IgG and IgM antibodies in normal and immuntxompromised subjects infected with cytomegalovirus. Journal of Infectious Diseases 145: 19 I - 199.

Schmitz H. Volz C. Krainick-Riechert CM. Scherer M (1972): Acute Epstein-Barr virus infections in children. Medical Microbiology and Immunology 158~58-63.

Schniitz H. von Deimling W. Flchmig B (1980): Detection of IgM antibodies to cytomcgalovirus (CMV) using an enzyme-labelled antigen (ELA). Journal o f General Virology 50:59-68.

76 Nielsen et al

Schniitz H (1082): Detection of immunoglobulin M antibody t o Epstein-Barr virus by use o f an cnrynic labelled antigen. Journal o f Clinical Microbiology 16:361-366.

Stagno S, Reynold DW. Smith RJ (1978): Use of isolated nuclei in the indirect fluorescent-antibody test for cytomcgalovirus infection: Comparison with microneutralisation. anticomplement and con- ventional indirect fluorescent antibody assays. Journal o f Clinical Microbiology 7:486-489.

Sutherland S. Briggs J D (1983): The detection of antibodies to cytomegalovirus in the sera of renal transplant rccipients by an IgM antibody capture assay. Journal of Medical Virology 11: 147-150.

Van Loon AM. Hccsscn FWA. van der Logt JTM, van der Veen J (1981): Direct enzyme-linked imrnunosorhent assay that uses peroxidase-labelled antigen for determination of immunoglobulin M antikxly l o cytomegalovirus. Journal of Clinical Microbiology 13:416-422.

Van l < w n AM. van der Logt JTM. Heessen FWA. van der Veen J (1985a): Use of enzyme-labclled antigen for the detection o f immunoglobulin M and A antibodies to herpes simplex virus in wrum and cerebrospinal fluid. Journal of Medical Virology 15: 183- 195.

Van Loon AM. van der Logt JTM. Heessen FWA, van der Veen J (1985b): Quantitation o f immunoglob- ulin E antihody to cytomcgalovirus by antibody capture enzyme-linked immunosorbent assay. Journal of Clinical Microbiology 2 I :558-561.

Wilson MB. Nakane PK ( 1978): Recent developments in the peritdate method for conjugating horserad- ish peroxidasc (HRPO) to antibodies. In Knapp W. Holubar K, Wick G (eds): "lmmunofluores- cense and Related Staining Techniques." Amsterdam: ElsevieriNorth-Holland Biomedical Press,

Yolkcn RH. Leistcr FJ ( 1981): Enzyme irnrnunoassays for measurement of cytomegalovirus immuno- pp 215-224.

globulin M antibody. Journal of Clinical Microbiology 14:427432.