We developed a multiplexed assay on a plasmonic-gold platform for measuring IgG and IgA antibodies and IgG avidity against both Zika virus (ZIKV) and dengue virus (DENV) infections. In contrast to IgM cross-reactivity, IgG and IgA antibodies against ZIKV nonstructural protein 1 (NS1) antigen were specific to ZIKV infection, and IgG avidity revealed recent ZIKV infection and past DENV-2 infection in patients in dengue-endemic regions. This assay could enable specific diagnosis of ZIKV infection over other flaviviral infections.
Zika virus (ZIKV) and dengue virus (DENV) are mosquito-borne flaviviruses that co-circulate throughout tropical and subtropical regions of the Western Hemisphere1. On the basis of clinical criteria alone, ZIKV is difficult to reliably distinguish from dengue fever and other infections that cause similar systemic febrile illness2,3. The overlap in clinical presentations and the potential for severe fetal outcomes, including microcephaly, congenital neurologic malformations and fetal demise4,5, and nonfetal manifestations, such as Guillain–Barré syndrome6, underscore the importance of accurate ZIKV diagnostics.
Although ZIKV reverse transcription (RT)–PCR tests provide specific ZIKV diagnoses7, given the high incidence of mild or asymptomatic ZIKV infections, patients might not present for medical attention at a point when viremia is detectable by RT–PCR. The US Centers for Disease Control and Prevention (CDC) recommends that ZIKV IgM-antibody testing be performed on specimens collected more than 4 d after illness onset8. However, current US Food and Drug Administration (FDA)-authorized ZIKV IgM assays under Emergency Use Authorization require confirmation by plaque-reduction neutralization testing (PRNT) to resolve false positives resulting from cross reactivity3. PRNT is a high-complexity method with a long turnaround time and limited availability9. The development of a broadly available serological test with a wide diagnostic window and the ability to differentiate ZIKV from DENV infection is essential.
Here we developed a multiplexed assay on a nanostructured plasmonic gold (pGOLD) platform9,10,11 for the detection of IgG, IgA and IgG avidity against ZIKV and DENV-2 antigens in patients' sera. The pGOLD platform is capable of amplifying near-infrared fluorescence by up to ∼100 times, allowing for sensitive analysis of multiple analytes over a 6–7-log dynamic range (Supplementary Fig. 1)9. Antigen arrays were developed on pGOLD for multicolor, simultaneous detection of the IgG, IgM and IgA antibody subtypes in type 1 diabetes and toxoplasmosis, and results matched reference laboratory tests10,11.
We constructed a microarray comprised of ZIKV NS1 and DENV-2 antigen on pGOLD slides (Supplementary Fig. 2). The fabricated biochip captured IgG and IgA antibodies against ZIKV or DENV antigens in human serum, and these were subsequently labeled with anti-human IgG-IRDye680 and anti-human IgA-IRDye800 (Supplementary Fig. 2). The amounts of IgG and IgA bound to each antigen were analyzed through the fluorescence intensities of IRDye680 and IRDye800. In a separate assay, IgG and IgM antibodies were detected with anti-human IgG-IRDye680 and with anti-human IgM-IRDye800, respectively. The assay was validated through comparison with a commercial DENV-infection test kit (Supplementary Fig. 3).
We also developed an IgG avidity test to differentiate recent from past infections. In the pGOLD avidity test, a urea-treatment step was introduced to remove weakly bound ZIKV or DENV IgG antibodies, which left behind only IgG antibodies with a strong affinity for the antigens on pGOLD. ZIKV IgG avidity was calculated by dividing ZIKV IgG levels, as measured using urea treatment, by IgG levels, as measured without urea treatment. Given that ZIKV IgG affinity is expected to increase gradually after infection, a low ZIKV IgG avidity suggested recent ZIKV infection, whereas a high ZIKV IgG avidity level suggested past ZIKV infection. The same analysis applied to the DENV IgG avidity test.
Serum samples from five groups of patients were obtained: (i) Group1_Z: 29 patients clinically diagnosed with ZIKV infection (sera collected at 2–93 d from illness onset) from dengue-endemic Colombia (Supplementary Table 1); (ii) Group2_D: 64 patients with DENV infection (Supplementary Table 2); (iii) Group3_H: 50 individuals with no history of ZIKV or DENV infection. (iv) Group4_Z: 49 patients from the DENV-endemic Dominican Republic with acute ZIKV infection positive by ZIKV RT–PCR (Supplementary Table 3); (v) Group5_DD: eight patients with acute secondary DENV infection.
Although ZIKV NS1 antigen could improve IgM test specificity over envelope antigen12,13, we still observed overlap of ZIKV NS1 IgM antibody levels in the sera of patients from Group1_Z and Group2_D (Fig. 1), owing to IgM cross-reactivity1. However, the levels of IgG (Fig. 1a) and IgA (Fig. 1c) antibodies against ZIKV NS1 antigen in the sera of patients with ZIKV infection were markedly higher than those in the sera of individuals with DENV infection and from control individuals. A ZIKV IgG cutoff of 0.14 was able to differentiate 26/29 (89.7%) individuals with ZIKV infection from those with DENV infection (Fig. 1a and Supplementary Fig. 4). The ZIKV IgG level was high even in sera from patients with ZIKV infection collected during the first 7 d of illness (Fig. 1d); it was highest in samples collected between 16 d and 50 d of illness, and it remained positive ∼100 d after symptom onset (Fig. 1d). The results suggested that in dengue-endemic regions, owing to weak IgG cross-reactivity for ZIKV and DENV, one should use a ZIKV IgG cutoff referenced to a population previously infected with DENV.
The ZIKV IgA levels detected in samples from Group1_Z patients were substantially higher than those detected in samples from Group2_D and Group3_H patients (Fig. 1c). With a cutoff of 0.25, the ZIKV IgA test offered a high negative percentage agreement (98.4%) and reasonable positive percentage agreement (69.0%), as compared to clinical diagnosis (Fig. 1c and Supplementary Fig. 4). Unlike ZIKV IgG, ZIKV IgA and IgM antibodies were not detectable 65 d after illness onset (Fig. 1d as compared to Fig. 1e,f). The transient nature of the IgA response (undetectable after 65 d of illness onset, Fig. 1f) means that a positive test for ZIKV IgA confirms recent ZIKV infection, thereby adding a tool for acute ZIKV infection diagnosis with high specificity and sensitivity.
In the sera of Group1_Z patients, we also detected high IgG antibody levels to DENV2 antigens (Fig. 1g), as well as to DENV serotype 1, 3 and 4 antigens (Supplementary Figs. 5 and 6). Given that DENV is endemic to Colombia, it was likely that these patients had been previously exposed to DENV14. Indeed, we detected high (>0.6) DENV IgG avidity levels for all of the Group1_Z patients (Fig. 1h), confirming past DENV infections. All patients with ZIKV infection in Group1_Z showed low (<0.5) ZIKV IgG avidity levels (Fig. 1i), which is in agreement with the fact that the Colombian patients had recently contracted ZIKV infection for the first time. The trend of increasing ZIKV IgG avidity with increasing time after illness onset (Fig. 1i) revealed the potential of the ZIKV IgG avidity test for differentiating recent and past ZIKV infections.
The early appearance of ZIKV IgG and IgA in Group1_Z patients prompted us to further investigate ZIKV IgG and IgA in patients with acute-phase infections, using a cohort of ZIKV-RNA-positive samples from Group4_Z. For comparison, we also tested eight DENV serum samples from patients with acute secondary DENV infection (Group5_DD), which were confirmed by positive DENV-RNA RT–PCR, a positive DENV IgG test (Fig. 2c) and high DENV IgG avidity (Fig. 2d).
Consistent with our findings from Group1_Z, we also detected positive ZIKV IgG and/or IgA levels in ∼47% of the samples from Group4_Z (Fig. 2). Because all patients in Group4_Z had acute ZIKV infection with specimens collected 2–6 d after symptom onset, the sensitivity of ZIKV IgG and IgA test for this cohort was lower than that for the Columbia cohort (Group1_Z) (Supplementary Figs. 4 and 7). All ZIKV-IgG-positive samples in Group4_Z showed low (<0.2) ZIKV IgG avidity (Fig. 2b), confirming acute ZIKV infection.
Importantly, samples from patients with secondary DENV infection in the acute phase (Group5_DD) showed positive DENV IgG and IgA levels (Fig. 2c,f), but low ZIKV IgG and IgA levels (Fig. 2a,e), which suggests that the positive ZIKV IgG and IgA levels observed in the two ZIKV cohorts from dengue-endemic regions were not due to pre-existing, cross-reactive anti-DENV antibodies, but were nascent, specific anti-ZIKV antibodies.
In summary, we developed a multiplexed assay to detect IgG, IgA and IgG avidity against ZIKV and DENV antigens. The high levels of ZIKV NS1 IgG and IgA antibodies detected within days of illness onset are specific for ZIKV infection, a finding that differs from commonly pursued IgM testing plagued by cross-reactivity. The early appearance of ZIKV IgG and IgA could be due to prior exposure of the patients to DENV and, consequently, secondary flavivirus infections, given that robust IgG and IgA responses can occur early on during a secondary-infection event15. Thus, highly specific serologic diagnosis of ZIKV infection can be achieved by using the pGOLD ZIKV IgG/IgA test in dengue-endemic regions where most adults have had previous DENV infections.
For early acute-phase ZIKV infections (<6 d after symptom onset) in dengue-endemic regions, the IgG/IgA assay detected 47% of ZIKV infections with high specificity. Although not ideal, the assay might still have diagnostic utility in the early acute phase, particularly in resource-limited settings that lack RT–PCR capability. The IgG/IgA test is powerful in the convalescent phase (sensitivity and specificity greater than 90% and 98%, respectively, Supplementary Table 4). A follow-up ZIKV IgG test weeks after illness onset could facilitate highly sensitive and highly specific ZIKV diagnosis. The addition of the IgG avidity test provided further data on the timing of infection, which would be valuable in dengue-endemic regions.
As compared to conventional singleplex methods, the assay presented here substantially reduces the number of tests required for diagnosis. The assay uses only 1 μl of human serum and provides results in ∼2 h, with satisfactory reproducibility (Supplementary Table 5). The assay is ready to be validated by research using large sample sizes, and has been submitted to the FDA in an application for Emergency Use Authorization for clinical diagnostics. The IgG/IgA avidity assay on pGOLD could facilitate laboratory screening for Zika fever during the pandemic spread of ZIKV infection, especially in regions with endemic DENV infection.
Serum samples from five groups of patients were obtained: (i) Group1_Z: 29 patients from DENV-endemic Colombia (purchased from Medical Research Networx) who were clinically diagnosed (people showing ZIKV symptoms who lived in or had traveled to regions with ZIKV transmission confirmed by RT–PCR16) with ZIKV infection 2–93 d after symptom onset during the recent outbreak in Colombia (samples collected from the end of 2015 to early 2016 period (Supplementary Table 1; data provided by Medical Research Networx)); (ii) Group2_D: 14 patients with a clinical diagnosis of DENV infection from Colombia, Ecuador and Honduras, collected before the introduction of ZIKV to the Americas (purchased from SeraCare Life Sciences), as well as 50 patients with a clinical diagnosis of DENV infection from Sri Lanka, which has had no known ZIKV cases (Supplementary Table 2 for DENV patient data); (iii) Group3_H: 50 control individuals with no history of ZIKV or DENV infection; (iv) Group4_Z: 49 patients from DENV-endemic Dominican Republic (purchased from Boca Biolistics) confirmed to have acute ZIKV infection (2–6 d after symptom onset) by laboratory ZIKV RT–PCR tests (Supplementary Table 3; data provided by Boca Biolistics); (v) Group5_DD: eight patients from Sri Lanka with secondary DENV infection (confirmed to have acute DENV infection by a laboratory DENV RT–PCR test and previous DENV infection with high IgG avidity). 50 serum samples from patients (from Sri Lanka) in Group2_D, 50 control individuals (from US) from group 3_H and eight serum samples from group 5_DD were collected, with Institutional Review Board (IRB) approval from Stanford University. Informed consent was obtained for the use of human samples.
All serum samples were de-identified. Investigators were not blinded to sample group allocation.
Multiplexed antigen microarray fabrication on pGOLD slide.
0.2 mg/ml ZIKV NS1 antigen (sequence strain: Uganda MR 766, produced in HEK293 human cells, Native Antigen Company, UK), 0.33 mg/ml dengue 2 antigen (purified dengue 2 virus particles, strain: 16681, cultured in vero cells, Microbix Biosystems, Canada) were prepared and delivered to pGOLD (Nirmidas Biotech, California) slides using GeSiM Nano-Plotter 2.1. Each microarray consists of three microarray spots of ZIKV NS1 antigen and three microarray spots of dengue 2 antigen. ∼3.2 nl of antigen solution was delivered to each spot. Microarray followed a 2 × 3 layout, spot diameter is ∼400 μm and the distance between spots is 1,000 μm. 16 identical microarrays were formed on each pGOLD slide. The fabricated biochips were vacuum sealed and stored at −20 °C before use. In a different microarray fabrication process, DENV1, 3 and 4 antigens (purified DENV 1, 3 and 4 virus particles, West Pacific 74, CH53489 and TVP-360 strains respectively) were also immobilized on pGOLD for antibody testing.
The fabricated biochip is integrated in a module in which 16 identical microarrays on each biochip are separated into 16 wells to process 16 samples. Each well was incubated with human sera (400 times dilution) for 40 min, followed by incubation of a mixture of anti-human IgG-IRDye680 conjugate and anti-human IgA-IRDye800 conjugate for 15 min (in a separate assay, anti-human IgG-IRDye680 and anti-human IgA-IRDye800 were applied to label-captured IgG and IgM antibody). Anti-human IgG/IgM/IgA secondary antibodies were purchased from Vector Laboratories with catalog #AI-3080, AI-3020, AI-3030, respectively. Each well was washed with washing buffer between each incubation procedure. 14 samples, together with two reference samples (one serum sample IgG, IgM and IgA positive for ZIKV NS1 antigen and DENV antigen, and one serum sample with negative IgG, IgM and IgA binding to ZIKV NS1 antigen and DENV antigen) were applied to each biochip.
The assay system is composed of the printed antigen array on a pGOLD slide, slide frame, buffers, plate washer and a dual-channel (700-nm/800-nm) scanner, which can be easily deployed in clinical and public-health laboratories.
ZIKV and DENV IgG avidity test.
The fabricated biochip is integrated in a module in which 16 identical microarrays on each biochip were separated into 16 wells to process 16 samples. Each well was incubated with human sera (400 times dilution) for 40 min, followed by incubation of 10 M urea/PBST solution for 10 min. Then, anti-human IgG-IRDye680 conjugate was applied to each well and incubated for 15 min. Each well was washed with washing buffer between incubation procedures. Two reference samples (one serum sample IgG, IgM and IgA positive for ZIKV NS1 antigen and DENV antigen, and one serum sample with negative IgG, IgM and IgA binding to ZIKV NS1 antigen and DENV antigen) were applied to each biochip, but were not treated with 10 M urea/PBST solution.
Qualitative RT–PCR testing.
ZIKV RT–PCR results were obtained from the vendor (Boca Biolistics). According to the vendor, ZIKV RT–PCR results were obtained with the LightMix Modular Zika Virus Real Time PCR Assay running on the COBAS Z480 or Light Cycler 2.0 system, by a clinical reference laboratory. DENV RT–PCR results were obtained with the CDC DENV-1–4 real-time RT–PCR kit. The assay was performed in multiplex on the Rotor-Gene Q instrument, as described in the package insert.
After the assay process, each biochip was scanned with a MidaScan-IR near-infrared scanner. MidaScan-IR is a dual channel (700 nm and 800 nm) near-infrared confocal microscope scanner for imaging tissues, cells and microarrays on standard glass or plasmonic slides. IRDye680 and IRDye800 fluorescence images were generated, and the median fluorescence signal for each channel on each microarray spot was quantified by MidaScan software. For each sample, each antigen and each channel, the average of the three median fluorescence signals for three spots is calculated and normalized by reference samples through a two-point calibration.
The ZIKV IgG cutoff is defined by mean ZIKV IgG levels of DENV-infected samples + 3 s.d. The ZIKV IgM cutoff is defined by mean ZIKV IgM levels of control samples + 3 s.d. The ZIKV IgA cutoff is defined by the mean ZIKV IgA levels of control samples + 3 s.d. The DENV IgG cutoff is defined by the mean DENV IgG levels of control samples + 3 s.d. The DENV IgM cutoff is defined by mean DENV IgM levels of control samples + 3 s.d. DENV IgA cutoff is defined by mean DENV IgA levels of control samples + 3 s.d. The cutoff, as determined by the mean + 3 s.d. method, resulted in the optimal combination of sensitivity and specificity of the ZIKV/DENV IgG/IgA assay on pGOLD (Supplementary Fig. 4).
For each sample that is positive for ZIKV IgG, ZIKV IgG avidity is calculated by dividing the normalized ZIKV IgG result of the sample tested with urea treatment (see ZIKV and DENV IgG avidity test section) by the normalized ZIKV IgG result of the sample without urea treatment (see 'Multiplexed-assay process'). DENV IgG avidity is calculated in the same way.
Measurements were performed three times for all values presented in this work, and the average is presented with error bars in relevant figures. The error bars (i.e., Fig. 1d–f) demonstrate the s.d. of three measurements for each value. Comparisons of ZIKV IgG/IgM/IgA levels measured for different patient cohorts (Fig. 1a–c) were assessed using an unpaired two- tailed Student's t-test. Results were considered to be significant for P < 0.01. In all box plots, measurement data (points) with median value (center line), average value (center small box), 1%, 25%, 75% and 99% value lines (box lines) are presented.
Any supplementary information and source-data files are available in the online version of the paper.
Waggoner, J.J. & Pinsky, B.A. J. Clin. Microbiol. 54, 860–867 (2016).
Duffy, M.R. et al. N. Engl. J. Med. 360, 2536–2543 (2009).
Zanluca, C. et al. Mem. Inst. Oswaldo Cruz 110, 569–572 (2015).
Rasmussen, S.A., Jamieson, D.J., Honein, M.A. & Petersen, L.R. N. Engl. J. Med. 374, 1981–1987 (2016).
Brasil, P. et al. N. Engl. J. Med. 375, 2321–2334 (2016).
Cao-Lormeau, V.M. et al. Emerg. Infect. Dis. 20, 1085–1086 (2014).
Waggoner, J.J. et al. Emerg. Infect. Dis. 22, 1295–1297 (2016).
US Centers for Disease Control and Prevention—Division of Vector-Borne Diseases. Memorandum (CDC, 2016). Available at https://www.cdc.gov/zika/pdfs/denvchikvzikv-testing-algorithm.pdf
Tabakman, S.M. et al. Nat. Commun. 2, 466 (2011).
Zhang, B., Kumar, R.B., Dai, H. & Feldman, B.J. Nat. Med. 20, 948–953 (2014).
Li, X. et al. J. Clin. Microbiol. http://dx.doi.org/10.1128/JCM.02132-16 (2016).
Steinhagen, K. et al. Euro Surveill. 21, 30426 (2016).
Wong, S.J. et al. EBioMedicine http://dx.doi.org/10.1016/j.ebiom.2017.01.008 (2017).
Villar, L.A., Rojas, D.P., Besada-Lombana, S. & Sarti, E. PLoS Negl. Trop. Dis. 9, e0003499 (2015).
Welliver, R.C. et al. J. Pediatr. 96, 808–813 (1980).
Rojas, D.P. et al. Euro. Surveill. 21, 30283 (2016).
H. Dai is a scientific adviser and share-holder of Nirmidas Biotech and served as a consultant for this work. The pGOLD platform was licensed to Nirmidas Biotech by Stanford University.
About this article
Cite this article
Zhang, B., Pinsky, B., Ananta, J. et al. Diagnosis of Zika virus infection on a nanotechnology platform. Nat Med 23, 548–550 (2017). https://doi.org/10.1038/nm.4302
Controlled copper in situ growth-amplified lateral flow sensors for sensitive, reliable, and field-deployable infectious disease diagnostics
Biosensors and Bioelectronics (2021)
Epilepsy & Behavior Reports (2020)
Trends in Microbiology (2020)
Reviews in Medical Virology (2020)