Development and Evaluation of a Broad Bead-Based Multiplex Immunoassay To Measure IgG Seroreactivity against Human Polyomaviruses

Human polyomavirus serology assays.

IgG seroreactivities against VP1 were measured using a customized Luminex xMAP assay, as previously described, albeit expanded to include all currently known HPyVs (12, 16, 20). In short, synthetic DNA sequences of VP1 (Table 1) (gBlocks; IDT, San Jose, CA, USA), either wild type (JCPyV, KIPyV, WUPyV, HPyV12, NJPyV, LIPyV) or codon optimized (MWPyV, HPyV10, STLPyV), were cloned into pGEX-5x-3 vectors (GE Healthcare Life Sciences, Chicago, IL, USA) and expressed as GST-VP1.tag fusion proteins in Escherichia coli BL21 Rosetta bacteria. Expression of each newly expressed GST-VP1 fusion protein was analyzed by glutathione-Sepharose 4B purification and SDS-PAGE (10%) separation, followed by Coomassie staining.

The GST-VP1.tag fusion protein was subsequently coupled to glutathione-casein-linked polystyrene beads (Bio-Rad Laboratories, Hercules, CA, USA.) Each bead is color coded by fluorescent dyes, which allows distinction between the different analytes in a single well. The coupling of the complete GST-VP1.tag fusion protein to the bead was verified on a Bio-Plex apparatus using mouse anti-tag antibodies (1:100; a kind gift from M. Pawlita), followed by anti-mouse immunoglobulin-phycoerythrin for detection (1:250; Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA) (which were incubated for 30 min each in the dark at room temperature).

In the HPyV multiplex immunoassay, serum samples (1:100) were incubated for 1 h in blocking buffer (1 mg/ml casein, 0.5% polyvinyl alcohol, 0.8% polyvinylpyrrolidone, 2.5% Super ChemiBlock [Chemicon International, Billerica, MA, USA], 2 mg/ml GST bacterial lysate in phosphate-buffered saline) to suppress potential nonspecific binding to the beads or to GST (20, 21). In the meantime, the GST-VP1 fusion proteins were coupled to glutathione-casein-linked polystyrene beads and the serum samples were subsequently incubated with the mixture of GST-VP1 beads (for 1 h in the dark at room temperature). For detection of a VP1-directed human IgG response, biotinylated goat anti-human IgG (H+L) was used (1:1,000; Jackson ImmunoResearch Laboratories Inc., West Grove, PA, USA), followed by streptavidin–R-phycoerythrin (SAPE; 1:1,000; Invitrogen, Waltham, MA, USA) (which were incubated for 30 min each in the dark at room temperature). As a positive control, a serially diluted mixture of four serum samples with known seroreactivity against various polyomaviruses was included in each test run (12). The seroreactivity was measured in a Bio-Plex 100 analyzer (Bio-Rad Laboratories, Hercules, CA, USA). Specific seroreactivity was defined by subtracting the median fluorescence intensity (MFI) values for both a blank sample and beads coupled to an irrelevant GST fusion protein (simian virus 40 small T antigen).

For a comparison between the in-house GST-VP1-based immunoassay and the VP1-VLP enzyme-linked immunosorbent assay (ELISA), 396 serum samples were analyzed in both assays for BKPyV IgG detection, as described previously (22). Our assay uses the VP1 protein from BKPyV genotype Ib1, while the VLP ELISA uses the VP1 protein from BKPyV genotype Ib2 (with 98.6% VP1 amino acid similarity existing between genotypes Ib2 and Ib1) (22).

HPyV12 and NJPyV VP1 seroreactivity confirmation.

To demonstrate the antigenicity of the HPyV12 and NJPyV GST-VP1 antigens used, two synthetic peptides (HPyV12 VP1 [VPKSVTDVTAKIQC] and NJPyV VP1 [SIHPNDIAKLPEED]) were generated (GenScript, Nanjing, China) and used to immunize rabbits. These peptides were chosen on the basis of their expected antigenicity in VP1 (15, 19) and the low amino acid similarity with other HPyV VP1 proteins. The polyclonal rabbit antisera raised against these peptides were used in a 1:100 dilution for the recognition of GST-HPyV12- and NJPyV VP1-coupled beads (which were incubated for 30 min in the dark at room temperature). Detection was performed with anti-rabbit immunoglobulin-biotin (1:1,000; Dako, Santa Clara, CA, USA) and SAPE (which were incubated for 30 min each in the dark at room temperature).

Competition analysis.

To gain further insight into cross-reactivity, VP1 antigen competition experiments were performed, as described previously (12). Serum samples with known seroreactivity were serially diluted from 1:100 to 1:409,600 and incubated with regular blocking buffer containing either GST or the GST-VP1 fusion proteins (∼2 mg/ml). For this purpose, only serum samples with measured seroreactivity above 5,000 MFI at a 1:100 serum dilution were selected.

Study population.

For evaluation of the HPyV multiplex serology assay, anonymized serum samples from a cohort of 87 healthy blood donors (HBD) (23) and a cohort of 65 immunocompromised kidney transplant recipients (KTR) (24) were tested. The participants gave written informed consent, and the study adhered to the Declaration of Helsinki principles.

Statistical analysis.

Squared Pearson correlation coefficients (r²) were calculated to determine intertest reliability. The correlation between the assessed HPyVs was further examined by calculating Spearman rank correlation coefficients (ρ). Statistical analysis was performed in IBM SPSS Statistics (version 23) software. When necessary, the significance level (α = 0.05) was adjusted according to the Bonferroni method for multiple comparisons.

Expression and coupling of HPyV VP1 to polystyrene beads.

To extend the in-house multiplex immunoassay to all currently known HPyVs, the VP1 genes of JCPyV, KIPyV, WUPyV, MWPyV, HPyV10, STLPyV, HPyV12, NJPyV, and LIPyV were individually cloned and expressed as GST-VP1 fusion proteins. Expression of glutathione-purified GST-VP1 fusion proteins was checked by Coomassie-stained SDS-PAGE and found to be comparable for all HPyVs (Fig. 1A). GST-VP1-containing crude bacterial extracts were purified and coupled to the glutathione-casein cross-linked beads. A tag sequence was included at the C terminus of each GST-VP1 fusion protein to check for efficient antigen binding and saturation of the beads. This was shown in a dilution series of GST-VP1-containing crude bacterial extracts (Fig. 1B). For convenience, it was decided to use a dilution of ∼1 mg/ml of each GST-VP1 crude extract in the HPyV VP1 multiplex immunoassay.

Antigenicity of GST-VP1 in the HPyV multiplex immunoassay.

Serum samples from HBD and immunocompromised KTR were tested to analyze the performance of the HPyV multiplex immunoassay. A broad range of seroreactivities that spanned the entire dynamic range of the assay (0 to 25,000 MFI units) was observed. Overall, comparable results were obtained for both sample sets (Fig. 2A and B). The measured seroreactivities against HPyV9, HPyV12, NJPyV, and LIPyV were generally lower than those against most other HPyVs, with the exception of some outliers.

To ensure the antigenicity of the HPyV12 and NJPyV VP1 preparations, polyclonal rabbit antisera were raised against specific HPyV12- and NJPyV-derived immunogenic peptides. These rabbit antisera recognized the relevant HPyV VP1 antigen (see Fig. S1 in the supplemental material), demonstrating the ability of our assay to detect HPyV12 and NJPyV antibody reactivity.

Reproducibility of the HPyV multiplex immunoassay.

The reproducibility of the HPyV multiplex assay was determined by calculating the squared Pearson's correlation coefficients between repeated measurements while using beads independently coupled to VP1 fusion proteins. These analyses were highly reproducible, with r² values ranging from 0.84 to 0.98 (Fig. S2A to J). Furthermore, we compared the use of different fluorescent beads for the same GST-VP1 fusion protein, which was tested for three HPyVs (BKPyV, KIPyV, and HPyV10) and revealed reproducible results, with r² values ranging from 0.77 to 0.95 (Fig. S3A to C). A historical comparison between seroresponses obtained in 2013 for six of the current HPyV targets with those for the HBD population revealed highly reproducible results (r² range, 0.71 to 0.97) (Fig. S4A to F) (12).

Specificity of the HPyV multiplex immunoassay.

Due to the VP1 amino acid sequence similarity between different HPyV species, varying between 21.6% and 78.5% (Fig. 3, red values at top right), one might expect epitope sharing and therefore a certain degree of cross-reactivity among (related) HPyVs. To evaluate this, a correlation matrix of the HPyV seroresponses was generated (Fig. S5), and Spearman rank correlation coefficients were calculated for each HPyV combination for the HBD population (Fig. 3, blue values at bottom left). The KTR population showed comparable data (not shown). The lack of measured seroreactivity against HPyV9, HPyV12, NJPyV, and LIPyV did not allow a meaningful correlation analysis, and these viruses were therefore excluded from this analysis. Overall, we observed little correlation between the seroreactivities determined against the individual HPyVs. A moderate correlation between HPyV6 and HPyV7 was observed in both the HBD population (Spearman's ρ = 0.49, P < 0.001) and the KTR population (Spearman's ρ = 0.44, P < 0.001). Despite 78.5% VP1 amino acid sequence similarity between BKPyV and JCPyV, no correlation between these types was measured (Spearman's ρ = −0.14, P = 0.19). Between species (HPyV10) members MWPyV and HPyV10, a high correlation was observed (Spearman's ρ = 0.92, P < 0.001), which can be explained by their high VP1 amino acid sequence similarity (98%) (Fig. 4).

To gain more insight into cross-reactivity, antigen competition experiments were performed. In these experiments reactive serum samples were titrated and preincubated with soluble GST-VP1 of various HPyVs before being exposed to antigenic beads coated with the relevant HPyV VP1. Figure 5 shows some examples of the results of these analyses. A complete overview of the selected serum samples tested in this way can be found in Fig. S6. Overall, little competition between VP1 antigens from HPyVs belonging to different species was observed. Preincubation with JCPyV VP1 did not show a reduction in BKPyV seroreactivity in three out of four experiments (Fig. 5A and S6A4 and A5); however, in one competition experiment, a substantial reduction was seen (Fig. S6A3). Vice versa, preincubation with BKPyV VP1 reduced JCPyV seroreactivity in two out of four competition experiments (Fig. S6B1 and B3). Between closely related species HPyV6 and HPyV7, partial antigen competition indicative of limited cross-reactivity was observed (Fig. 5F and G and S6F1, F2, and G2). As expected, HPyV10 species members MWPyV and HPyV10 showed high levels of cross-reactivity in this analysis (Fig. 5J and K). Interestingly, in three out of six HPyV10 competition experiments, preincubation with MWPyV VP1 did not block HPyV10 seroreactivity (Fig. S6K1, K4, and K5). A summary of the results of the competition experiments is shown in Table 2.

Comparison between the GST-VP1-based HPyV multiplex immunoassay and a VP1 VLP-based ELISA.

To learn more about the antigenicity of the GST-VP1 fusion proteins that we used, we compared the seroresponses measured for BKPyV in our method to those obtained with a VLP-based ELISA. Although differences in especially the presentation of conformational epitopes were anticipated, the BKPyV seroreactivities measured by both methods were quite similar (Fig. 6) (22). A high Spearman correlation coefficient (ρ = 0.823, P < 0.001) was observed between the optical density (OD) values obtained with the VP1 VLP ELISA and the MFI values obtained with the GST-VP1 immunoassay.