Novel multiplex assay platforms to detect influenza A hemagglutinin subtype‐specific antibody responses for high‐throughput and in‐field applications

Background Detections of influenza A subtype‐specific antibody responses are often complicated by the presence of cross‐reactive antibodies. We developed two novel multiplex platforms for antibody detection. The multiplexed magnetic fluorescence microsphere immunoassay (MAGPIX) is a high‐throughput laboratory‐based assay. Chembio Dual Path Platform (DPP) is a portable and rapid test that could be used in the field. Methods Twelve recombinant globular head domain hemagglutinin (GH HA1) antigens from A(H1N1)pdm09 (pH1N1), A(H2N2), A(H3N2), A(H5N1), A(H7N9), A(H9N2), A(H13N9), B/Victoria lineage, B/Yamagata lineage viruses, and protein A control were used. Human sera from U.S. residents either vaccinated (with H5N1 or pH1N1) or infected with pH1N1 influenza viruses and sera from live bird market workers in Bangladesh (BDPW) were evaluated. GH HA1 antigens and serum adsorption using full ectodomain recombinant hemagglutinins from A(pH1N1) and A(H3N2) were introduced into the platforms to reduce cross‐reactivity. Results Serum adsorption reduced cross‐reactivity to novel subtype HAs. Compared to traditional hemagglutination inhibition or microneutralization assays, when serum adsorption and the highest fold rise in signals were used to determine positivity, the correct subtype‐specific responses were identified in 86%‐100% of U.S. residents exposed to influenza antigens through vaccination or infection (N=49). For detection of H5N1‐specific antibodies in sera collected from BDPW, H5 sensitivity was 100% (six of six) for MAGPIX, 83% (five of six) for DPP, H5 specificity was 100% (15/15), and cross‐reactivity against other subtype was 0% (zero of six) for both platforms. Conclusion MAGPIX and DPP platforms can be utilized for high‐throughput and in‐field detection of novel influenza virus infections.


| INTRODUCTION
Hemagglutination inhibition (HI) and virus microneutralization (MN) assays are gold standards for detection of antibody responses to influenza; both tests primarily detect antibodies to hemagglutinin (HA) of influenza viruses. 1,2 However, implementation of these assays requires propagation of live viruses and appropriate biological safety levels, which often limits assay portability and throughput. Utilizing noninfectious recombinant proteins may broaden the applications of serologic assays, making them more suitable for field applications that are often at remote locations lacking heightened laboratory containment facilities.
Following novel influenza virus emergence events among humans such as the 2009 H1N1 pandemic, or H5N1 and H7N9 outbreaks, serologic studies need to be conducted quickly to assess the levels of exposure and immunity in populations. Rapid public health responses required in such scenarios highlight the need for sensitive and specific serologic assays to detect influenza subtypespecific antibody responses that are quick, high throughput, and ideally portable. [2][3][4] Given that the traditional ELISA often lacks subtype specificity 5 and multiplexing capabilities 6 to detect antibody responses to influenza, alternative assays continue to be evaluated to improve the performance of influenza serologic assays. A multiplex protein microarray using globular head domain HA1 (GH HA1) was developed to investigate antibody profiling, and high sensitivity and specificity were achieved by bivariate model of microarray using HAs from pH1N1 and A/South Carolina/1918 (H1N1) viruses. 4,7,8 Recently, a mPLEX-Flu assay using full-length HA was also used to evaluate HA profiles and demonstrated good correlation between the median fluorescence intensity (MFI) and anti-HA antibodies in ELISA. 9 Detections of antibody responses to influenza in humans are often complicated by the complex antibody profiles generated from past exposure through infection or vaccination with multiple influenza viruses. Novel influenza viruses may share common epitopes with seasonal viruses that an individual may have been exposed to previously, causing both within-subtype (homosubtypic) or cross-subtype (heterosubtypic) cross-reactivity. This cross-reactive immunity may provide broader immune protection to influenza virus infection, but at the same time also poses challenges when serology is used to assess exposure with novel influenza viruses.
Here, we developed two novel multiplex assay platforms: the magnetic multiplexed fluorescence microsphere immunoassay (MAGPIX) and the Chembio Dual Path Platform (DPP) utilizing multiple subtype rHAs to address high-throughput needs in the laboratory setting and the portability required for in-field applications, respectively. We incorporated the use of GH HA1 as detection antigens and antibody adsorption using full ectodomain HAs to reduce the effects of antibody cross-reactivity to improve assay performance. Panels of well-characterized human serum specimens collected from persons exposed to known subtype influenza viruses or vaccines were used to assess assay sensitivity, specificity, and cross-reactivity. In addition, sera from workers in Bangladesh live poultry markets that were potentially exposed to highly pathogenic H5N1 viruses were also evaluated to recapitulate the scenario in which these assays may be applied in the field in the event of novel influenza virus outbreaks.

| Serum adsorption assay
One milliliter of 1% blue latex beads (300 nm in diameter, Thermo

| Chembio Dual Path Platform
The  Figure S1) to allow sera antibodies to bind to immobilized antigens on the membrane. After a 10-minute incubation, five drops of running buffer were added to the buffer port ( Figure S1) to allow colloidal gold-conjugated PA to bind to GH HA1-antibody complex for test line or to PA for control line, respectively. Results were read using a Chembio Rapid Influenza Immunity Test Reader (QIAGEN, Hilden, Germany) after 15 minutes of incubation at room temperature.

| Data analysis
Fold rises in MFI and DPP values (S2 value/S1 value) were calculated to measure antibody binding. Due to wide dynamic ranges of the readout of the MAGPIX and DPP assays, S1 samples with values lower than baseline level were normalized to a set value as discussed previously. 13 For serum samples from US residents, any MFI lower than 1000 in MAGPIX and DPP value lower than 100 in DPP in S1 samples were arbitrarily adjusted to 1000 and 100, respectively. Because the baseline antibody levels in S1 sera collected in Bangladesh were lower than those in S1 sera collected in the US (data not shown), low MFI and DPP values in Bangladesh S1 sera were arbitrary adjusted to 400 and 50, respectively. When fold rises against multiple subtype HAs (≥2) were observed, the subtype HA with the highest fold rise in MFI or DPP value was considered positive. Fisher exact test and paired sample t test were performed using GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA), and P values of <0.05 were considered statistically significant.
T A B L E 1 Sera from influenza-vaccinated or influenza-infected persons used in the study for evaluation of sensitivity and cross-reactivity

| Homosubtypic and heterosubtypic antibody responses following vaccination and natural infection are detected by MAGPIX and DPP
In untreated sera from persons vaccinated with a H5N1 (clade were statistically not significant except MFI rise to H7 and DPP value rises to H2 and H7 in pH1N1-exposed persons (n=36, P<.05) ( Figure 1C,D).

| Cross-reactive antibodies can be removed by serum adsorption without loss of subtypespecific antibodies
To analyze subtype-specific antibody responses due to antigen exposure, cross-reactive antibodies were removed by serum adsorption with latex beads conjugated with ectodomain rHAs from pH1N1 and H3N2. Following adsorption of S2 sera from H5N1-vaccinated persons (n=13), antibodies against pH1 and H3 in MAGPIX and pH1/H3 in DPP were significantly reduced (P<.05, Figure 2A,B); cross-reactive antibodies to the novel subtype H7 GH HA1 were also removed (P<.05, Figure 2A,B). H5 GH HA1-specific antibodies in H5N1 vaccinees remained post-adsorption (P>.05, Figure 2A,B). This was also seen in paired serum samples from H5N1 vaccinees after serum adsorption ( Figure 2C,D).

| Improved sensitivity and reduced crossreactivity by serum adsorption for MAGPIX and DPP
S2/S1 fold rise threshold in MFI and DPP values was analyzed using paired human serum samples collected from either influenza infection or vaccination (Table 1). Some paired serum specimens showed a greater than twofold rise against multiple GH HA1 antigens due to cross-reactivity or unknown exposure to avian influenza virus(es) (Table S1 and data not shown). Therefore, the highest fold rise value was used as a threshold to interpret results ( Table 2). After serum adsorption with pH1/H3 rHA-conjugated latex beads, the sensitivity for pH1/H3 in DPP increased from 67% to 86% (P=.052) ( Table 2) and cross-reactivity to H5 decreased from 11% to 3% (P=.357) ( Table 2). The highest S2/S1 fold rise threshold post-serum adsorption correlated well with the correct HA subtype-specific antibody response when fold rises to multiple rHAs were observed (

| DISCUSSION
Here, we developed and evaluated two novel multiplex platforms, MAGPIX and DPP, for the detection of subtype-specific antibodies to influenza. We have improved assay sensitivity and specificity and reduced cross-reactivity by two strategies. Firstly, GH HA1 antigens that contain more subtype-and strain-specific epitopes were utilized as detection antigens to eliminate cross-reactive epitopes in the HA2 domain. 14   that had been infected and vaccinated with multiple subtypes of viruses or had been exposed to novel influenza in Bangladesh poultry markets 10,15 to analyze both platforms. When sera were pre-adsorbed with pH1/H3 rHA-conjugated latex beads and the highest fold rise to antigen was considered as positive, >83% sensitivity (six of six or five of six), 100% specificity (15/15), and 0% cross-reactivity (zero of six) were achieved for both platforms to detect exposures to H5N1 viruses (Table 3). These data indicated that these assays may have sufficient sensitivity and specificity to detect even mild or asymptomatic influenza virus infections. 10 DPP is a portable lateral flow platform which is easy to use. Results from DPP can be obtained in a non-laboratory setting in less than 30 minutes, making it an appealing choice for infield applications. MAGPIX is cost-effective and high throughput, requires limited sample volume and staff time, and reduces aliquot errors. 6,9,16,17 Both platforms are multiplex assays with several influenza subtype antigens, ideal for serum specimen collected from locations where multiple subtypes of influenza viruses cocirculate. Our results suggested that the optimized DPP and MAGPIX platforms described here could be used for in-field and high-throughput applications to support both routine influenza surveillance and outbreak responses of HPAI influenza.
As a result of exposure(s) to seasonal influenza viruses and/or vaccine(s), antibodies to novel HA in unexposed populations increases with age and may vary by geographic location. 3,[18][19][20][21] People who were born before 1967 showed significantly higher reactivity against H2 GH HA1 than those born after 1970 (P<.05) and could not be removed by serum adsorption. They also showed higher antibody baselines (S1) against other group 1 HAs, such as, pH1, H5 VN, and H9 due to potential exposure to H2N2 virus and more seasonal H1N1 viruses.
Therefore, age-matched controls are necessary to determine baseline level of pre-existing antibodies in different birth year cohorts. 13,22 A complex relationship between antibodies against seasonal and novel avian influenza viruses was observed in humans 23  not by HI in primary H1N1 or H3N2 influenza virus-infected children. 5 While the HI assay measures the presence of antibodies that inhibit HA binding to host cells, both MAGPIX and DPP assays were designed to detect total binding antibodies to GH HA1. In this study, the use of GH HA1 antigens improved the specificity of MAGPIX and DPP platforms for the detection of HA subtype-specific antibody responses, although cross-reactive antibodies against other subtype HA(s) were still observed without adsorption ( Figure 1C,D).
We demonstrated that serum adsorption in combination with the use of GH HA1 reduced cross-reactivity to other subtypes of GH HA1 without loss of subtype-specific antibodies against H5 and H2 GH HA1 antigens (Figures 2 and 3). Serum adsorption improved sensitivity from 67% to 86% and reduced cross-reactivity from 11% to 3% (Table 2). Upon exposure to one influenza antigen, antibody rises to multiple rHA subtypes were observed in both DPP and MAGPIX assays (Table S1), but the antigen with the highest fold rises can be used to identify correct subtype exposures in the majority of cases after serum adsorption ( Table 2).
Our study has several limitations: ( In summary, portable lateral flow (DPP) and high-throughput platform (MAGPIX) using GH HA1 proteins, combined with serum adsorption of ectodomain pH1/H3 rHAs, showed advantages in detecting HA subtype-specific antibody responses. To our knowledge, this is the first study in which serum adsorption with rHA-conjugated beads has been used to remove cross-reactive antibodies in evaluating binding antibody assays. These platforms can be utilized in influenza serologic surveillance and outbreak responses to novel HPAI influenza virus infections, especially when the virus isolation and RT-RCR results are not available or cannot be accurately interpreted without further serologic confirmation.