Enzyme-linked immunosorbent assay using recombinant envelope protein 2 antigen for diagnosis of Chikungunya virus

A suitable amount of 44 kDa rE2 of CHIKV was produced in E. coli cells, and its purity was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot (Fig. 1a and b). The optimal concentration of recombinant rE2-CHIKV for coating ELISA plate wells was determined based on a clear detection of anti-CHIKV antibodies using mouse hyperimmune serum, which was 10 times higher than the cut off. Higher concentrations of rE2 were not chosen because they did not increase the assay sensitivity. Thus, 1 μg/ml of rE2-CHIKV per well after 18 h of coating incubation were chosen as suitable for the assay. Importantly, cross-reactivity of other mouse hyperimmune sera was not observed for alphaviruses, such as AURV, EEEV, MAYV, MUCV, and WEEV in the rE2-CHIKV ELISA (Fig. 2).

Results obtained by rE2-CHIKV in IgG and IgM ELISAs analyzing 59 serum samples of CHIKV suspected cases were compared to those obtained using enzyme immunoassay with infected cultured cells (EIA-ICC), commercial immunochromatography assay and plaque reduction neutralization test of 50% of plaques (PRNT₅₀) (Table 1 and Additional file 1: Table S1). The IgG rE2-CHIKV ELISA showed a sensitivity of 89.66% and a specificity of 100% when compared to PRNT₅₀ results, considering 80 as a titer cutoff. Importantly, all 26 positive results in rE2-CHIKV ELISA were also positive by PRNT₅₀, and three false negative results were observed. The comparison of IgG CHIKV PRNT₅₀ results with those obtained with CHIKV EIA-ICC and commercial CHIKV immunochromatography assay showed 82.76 and 89.66% sensitivity and 96.67 and 100% specificity, respectively (Table 2). Additionally, IgG positive serum samples in the rE2-CHIKV ELISA showed 73% of high, 8% of medium and 19% of low avidity (Additional file 1: Table S1). The Bayesian analysis estimated a sensitivity of 92.48% and specificity of 79.04% for the IgM rE2-CHIKV ELISA. Sensitivities and specificities for IgM detection of the EIA-ICC were 74.01 and 55.59% and for the Commercial assay were 91.31 and 94.21%, respectively (Table 3). In addition, IgM positive samples showed 25% of high, 29% of medium and 46% of low avidity (Additional file 2: Table S2).

The rE2 of CHIKV was used as an antigen in an indirect ELISA for diagnosis of CHIKV infection. Concentrations of rE2, from 0.5 μg/ml to 8 μg/ml, diluted in 0.05 M Carbonate-Bicarbonate buffer pH 9.6 (Sigma-Aldrich, USA) were added to wells of microplates (Corning, USA) and as negative control, Escherichia coli cells extract was diluted in same dilution and added to the other half of the plate. Plates were incubated for 18 and 36 h in a wet chamber at 4 °C and washed between each assay step with 150 μl of PBS-T (Phosphate-buffered saline (PBS) with 0.05% (v/v) Tween®). After washing three times, plates were blocked for 2 h at 37 °C with 150 μl of 10% (w/v) non-fat dry milk in PBS-T (blocking solution) to reduce background. In sequence, plates were washed three times and for positive CHIKV IgG, 50 μl of different dilutions (1:100 to 1:800) of a mouse hyperimmune anti CHIKV serum diluted in blocking solution was added to the positive and negative wells. Plates were incubated 1 h at 37 °C and washed four times. After, it was added 50 μl of a 1:2000 horseradish peroxidase (HRP)-conjugated to a goat immunoglobulin anti-mouse IgG (‘Fab’ specific, Sigma-Aldrich, USA) diluted in blocking solution. Plates were incubated 1 h at 37 °C, washed five times and it was added 100 μl of 2,2′-Azinobis [3-ethylbenzothiazoline-6-sulfonic acid]-diammonium salt (ABTS) peroxidase substrate (KPL, USA) per well and incubated for ~ 15 min at 37 °C for test revealing. The plates were read in Titertek Multiscan MMC/340 Microplate Reader at an optical density (O.D.) of 405 nm. ELISA cutoff values were calculated as the mean O.D. of negative controls plus three standard deviations (SDs). All samples were tested in duplicate. Sample with an average O.D. above the cutoff value was considered as positive [40].

In order to evaluate cross-reactivity of antibodies to alphaviruses in the rE2-CHIKV ELISA, mouse hyperimmune sera to AURV strain BeAr10315, EEEV strain BEAN-1999, MAYV strain BeAr20290, MUCV strain BeAn-8 and WEEV strain SpAn14723 were tested at different dilutions (1:100 to 1:800). These hyperimmune sera had their reactivity previously confirmed by indirect immunofluorescence tests [41] against their reciprocal viruses infecting tissue culture cells, labelling for nuclei, cytoplasm and the virus (Additional file 3: Figure S1).

A total of 59 serum samples from patients clinically suspected of CHIKV infection were tested by rE2-CHIKV ELISA. Human assays used the HRP-conjugated goat anti-human IgG ‘Fab’ specific (Sigma-Aldrich, USA) diluted at 1:2000 or the HRP-conjugated goat anti-human IgM ‘Fc5μ’ (Merck Millipore, USA) diluted at 1:25000. Positive samples were diluted from 1:100 to 1:25,600.

Results obtained for the 59 human sera in IgG and IgM rE2-CHIKV ELISAs were compared to those obtained by an in-house EIA-ICC using C6/36 cells [42] and also to those obtained by a commercial immunochromatographic test for quick detection of anti-CHIKV IgG and IgM antibodies (Lumiquick, USA).

To evaluate the specificity of anti-CHIKV antibodies in human serum samples, a PRNT₅₀ was performed in Vero cells. Briefly, human sera previously inactivated were serially diluted (1/10 to 1/20,480) in Dulbecco’s Modified Eagle’s Medium (DMEM) (Vitrocell, Brazil). Subsequently, 1.75 × 10² PFU of CHIKV, strain BzH1, cordially provided by professor Benedito Antônio Lopes da Fonseca, were mixed with each serum and these mixtures were incubated for 1 h at 37 °C. Then, 200 μl of each mixture was inoculated into Vero cell monolayers in 6 well plates. Cells were incubated 1 h at 37 °C under gently rocking. In sequence, 2 ml of pre-warmed DMEM containing 1% agar (Sigma, USA) and 3% fetal bovine serum (FBS) (Vitrocell, Brazil) were added to each cell monolayer well, and plates were incubated for 2 days at 37 °C at 5% CO₂ atmosphere. Finally, the cells were fixed in the wells with 4% formaldehyde (Labsynth, Brazil) solution for 2 h and stained with 1% crystal violet (Merck, USA) during 5 min, for plaque visualization. Plaque reduction was calculated for each serum by comparing their respective number of plaques to the positive control, which was inoculated with 1.75 × 10² PFU of CHIKV. The cut off titer for positivity was set as 80.

Human IgG and IgM positive samples in the rE2-CHIKV ELISA were submitted to an avidity assay [43]. Briefly, after primary antibody incubation, 100 μl of a 6 M urea solution diluted in PBS or only PBS was added to the wells, and the plates were incubated for 10 min at 37 °C in a wet chamber. Then, plates were washed four times with PBS-T. The subsequent steps of the rE2-CHIKV ELISA were performed as described above. Relative avidity index (RAI) was calculated for each sample by dividing the liquid O.D. in urea treated wells by those in untreated wells (PBS), and proportions were showed in percentage. Samples with RAI > 60% were considered as having high avidity, 40 to 60% as medium avidity, and < 40% as low avidity.

The sensitivity and specificity of rE2-CHIKV ELISA, EIA-ICC and immunochromatographic test for IgM detection were estimated by the Bayesian method introduced by Joseph et al. [44]. The Bayesian method allows us to estimate the accuracy of the test in the absence of a gold standard, by incorporating into the analysis prior knowledge about the performance measures. Prior information about test sensitivities and specificities was obtained from the literature [13, 15, 36] and data available from LumiQuick Inc. (Santa Clara, USA), and it was represented by the use of beta (a,b) probability distributions, where the values of “a” and “b” determine the shape of the curve. These values were assessed using the function epi.betabuster of the epiR package in R software. This function calculates the values “a” and “b” based on a prior estimate and a 100(p)% uncertainty range, defined by a lower (LL) or upper limit (UL). Bayesian analyses were based on Markov chain Monte Carlo (MCMC) computations and performed with OpenBUGS (Imperial College and MRC, United Kingdom, available at www.​mrc-bsu.​cam.​ac.​uk/​software/​bugs). Posterior inferences were based on summaries of 1,000,000 iterations with a sampling lag of 10, after a burn-in of 10,000 iterations. The final results were presented as a median and a 95% probability interval (PI, percentiles 2.5 and 97.5) of each parameter estimate.