key: cord-353209-qkhfp66l authors: Steiner, Daniel J.; Cognetti, John S.; Luta, Ethan P.; Klose, Alanna M.; Bucukovski, Joseph; Bryan, Michael R.; Schmuke, Jon J.; Nguyen-Contant, Phuong; Sangster, Mark Y.; Topham, David J.; Miller, Benjamin L. title: Array-based analysis of SARS-CoV-2, other coronaviruses, and influenza antibodies in convalescent COVID-19 patients date: 2020-06-16 journal: bioRxiv DOI: 10.1101/2020.06.15.153064 sha: doc_id: 353209 cord_uid: qkhfp66l Detection of antibodies to upper respiratory pathogens is critical to surveillance, assessment of the immune status of individuals, vaccine development, and basic biology. The urgent need for antibody detection tools has proven particularly acute in the COVID-19 era. We report a multiplex label-free antigen microarray on the Arrayed Imaging Reflectometry (AIR) platform for detection of antibodies to SARS-CoV-2, SARS-CoV-1, MERS, three circulating coronavirus strains (HKU1, 229E, OC43) and three strains of influenza. We find that the array is readily able to distinguish uninfected from convalescent COVID-19 subjects, and provides quantitative information about total Ig, as well as IgG- and IgM-specific responses. What they do not provide, however, is a broader understanding of the human immune response to SARS-CoV-2 infection, or illuminate potential relationships between COVID-19 infection and previous infections (and immunity to) other respiratory viruses including circulating coronaviruses that cause the common cold. To address these goals, multiplex analytical techniques are required. A bead-based multiplex immunoassay for six coronaviruses infecting humans (pre-SARS-CoV-2) has been reported, 12 and more recently a 4-plex assay on the Quanterix platform focused on SARS-CoV-2 antigens has been described. 13 Despite these advances, there remains a significant need for analytical methods able to rapidly quantify antibodies not only to SARS-CoV-2, but also to other coronaviruses, and other pathogenic viruses. Most importantly, these must be able to discriminate among responses to different closely related viruses and different antigens from the same virus. To address this need, we have developed a prototype 15-plex array on the Arrayed Imaging Reflectometry (AIR) platform. AIR is a label-free multiplex sensor method in which the surface chemistry and deposition of capture molecules to form a microarray on a silicon chip are carefully controlled such that s-polarized HeNe laser light at a 70.6º incident angle to the chip undergoes total destructive interference within the surface film. 14 Binding to any probe spot on the array degrades the antireflective condition in proportion to the amount of material bound, yielding an increase in the reflected light as observed by a CCD camera. By comparing the intensity of the reflected light to an experimentally validated model, the thickness change for each spot, and therefore the quantity of each analyte in the sample, may be precisely and sensitively determined. 15 We have previously reported the utility of influenza antigen arrays fabricated on the AIR platform for assessment of anti-influenza antibodies in human, animal, and avian serum, 16, 17 both as a tool for viral surveillance and for assessment of the efficacy of a candidate vaccine. We have also demonstrated that AIR is scalable at least to 115-plex assays, used for discriminating different influenza virus serotypes. 18 , 19 We therefore anticipated that the platform would be useful as a way to quantify anti-SARS-CoV-2 antibodies, antibodies to other coronaviruses including circulating ("common cold") strains, and other respiratory pathogens including influenza. Here, we discuss the development and testing of a mixed coronavirus / influenza antigen panel on AIR, and its application to analyzing the coronavirus antibody profile of a cohort of convalescent COVID-19 patients and subjects of unknown disease status. Material sources: For AIR assays, SARS-CoV-2, SARS-CoV, MERS, and Influenza Type A and B antigens were obtained from Sino Biological, Inc., and are described in more detail below. Most antigens were supplied as lyophilized material and reconstituted at the recommended concentrations using 18-MΩ water, while the remaining antigens were supplied frozen on dry ice. PBS-ET was prepared as phosphate buffer (10 mM monobasic sodium phosphate, 10 mM dibasic sodium phosphate, 150 mM NaCl) with 0.02% w/v Tween-20 and 5 mM EDTA. Aminereactive substrates for fabrication of AIR arrays were provided by Adarza BioSystems, Inc. For ELISA assays, SARS-CoV-2 full-length spike and RBD were produced in-house using a mammalian expression system, 20,21 as was influenza A/H1N1/California 2009 hemagglutinin. HCoV-229E and HCoV-OC43 spike proteins (baculovirus-expressed) were obtained from Sino Biological. Tetanus toxoid (TTd) was obtained from Calbiochem. Antigen probe formulation: Prior to microarray fabrication, antigens were buffer-exchanged and concentrated using Amicon centrifugation filters (EMD Millipore) into phosphate buffer at pH 5.8 and pH 7.2 prior to use. During development, several printing concentrations and/or solution pH values of each antigen were tested, along with sugar additives (glycerol, trehalose) in order to optimize spot uniformity and morphology as well as initial probe thickness. 22 Preparation of arrays: Arrays were printed on amine-reactive silicon oxide substrates (Adarza BioSystems, Inc.) using a Scienion SX piezoelectric microarrayer (Scienion, A.G.) with spot volumes of approximately 300 pL. Six spots were printed for each antigen, the final layout of which is shown in Figure 1 . The number of spots arrayed was not critical to robust analytical performance or statistical analysis. Each spot consists of approximately 300 pixels when imaged by the CCD in an AIR chip reader (Adarza BioSystems, Inc.), with each pixel representing a discrete interrogation of a unique probe surface region. Therefore, averaging these pixel values together produces an inherently reliable measure of analyte-to-probe response. Dilutions of polyclonal anti-fluorescein (anti-FITC, Rockland Inc.), were printed as negative intra-array controls. After printing, chips were mounted onto adhesive strips at appropriate spacing for 96-well plates, and then placed into 50 mM sodium acetate buffer (pH 5) for 5 minutes. Next, a 1.5% BSA solution was added to each well resulting in a final BSA concentration of 0.5% to passivate the remaining amine-reactive surface functionality. After blocking for 20 minutes, the chips were transferred to new wells containing 20% fetal bovine serum (Gibco) in PBS-ET as a secondary block, and incubated for 40 min. This step was required to reduce nonspecific binding from human serum at the assay endpoint. The chips were then rinsed briefly (5 min) in new wells containing PBS-ET, then transferred to wells containing Microarray Stabilizer Solution (Surmodics IVD). After a 30-minute incubation, the chips were dried at 40 °C in an oven for 60 min. This last step renders the sensors shelf-stable, until use in assays performed later. IgM, or 20% FBS as a negative control. Each of these conditions was produced in duplicate. Secondary antibodies were diluted to 1 μg/mL for both goat α-hIgG (Jackson Immunoresearch) and rabbit α-hIgM (Rockland, Inc.) in Adarza diluent. After one hour of incubation with secondary antibodies at room temperature, chips were washed twice for 5 minutes in PBS-ET, then rinsed with water and dried with nitrogen as before. Data analysis: AIR images were analyzed using the Adarza ZIVA data analysis tool. Probe spots with major defects or debris were manually flagged and eliminated, and minor defects in spot quality were automatically identified and excluded from the median intensity measurement. The median intensity values were converted to median thickness values using a best-fit line to an experimentally derived reflectance model. 14 Then, the median thickness values were further processed in Microsoft Excel as described below, and are referred to simply as "thickness" hereafter. While anti-FITC spots were designed to serve as an intra-chip normalizer, these were not used as such due to the unexpected presence of anti-goat IgG antibodies in some single donor human serum samples. Therefore, the blank area served an intra-chip normalizer to mitigate any variation in the reactivity of the surface chemistry between AIR chips. The thickness of the blank area was subtracted from the thickness of each probe spot to produce "normalized thickness" values for each probe spot. All of the normalized thickness values across replicate chips (n=2) were averaged together (maximum of n=24 probe spots) for each antigen, and the standard deviation was calculated. The average thickness for each antigen in the fetal bovine serum (FBS) control was subtracted from the average thickness obtained for each antigen in each subject sample to produce the "normalized thickness change (Δ Thickness)." In the case of the polyclonal antibody titration, the control chip was incubated in a matrix of FBS and PNHS. ELISA assay: Serum IgG titers specific for SARS-CoV-2 proteins and selected non-coronavirus proteins were determined by ELISA as described previously. 23 Human serum standards were used to assign weight-based concentrations of antigen-specific IgG as previously described, with the limit of assay sensitivity set at 0.5 μg/mL for all antigens. 23, 24 Results: . This is as expected given the prevalence of these viruses in the general population. Addition of an anti-SARS-CoV-2 polyclonal antibody raised against the SARS-CoV-2 spike protein receptor binding domain (RBD) at 1 μg/mL produced a strong signal on all three RBDcontaining antigens (S1 + S2 ECD, S1, and RBD). Overall response to the polyclonal antibody was well-behaved, and titrated to zero as expected ( Figure 2C and D) . Quantitative data are presented in Ångstroms of build. At the highest concentrations, significant cross-reactive binding to the HCoV-229E spike protein was observed, as well as some binding to the HCoV-OC43 spike protein and MERS S1. Calculated limits of detection 25 for these data were 43.3 ng/mL (SARS-CoV-2 S1 + S2 ECD), 40.7 ng/mL (SARS-CoV-2 S1), and 25.1 ng/mL (SARS-CoV-2 RBD). However, these should be viewed as provisional, and subject to optimization. Response of a commercial anti-SARS-CoV-2 rabbit polyclonal antibody (pAb) on the array. (A) array exposed to array exposed to 20% FBS + 10% PNHS; (B) array exposed to 1 μg/mL anti-SARS-CoV-2 pAb in 20% FBS + 10% PNHS. Strong responses to SARS-CoV-2 S1+S2 ECD, S1, and RBD are observed, as well as smaller cross-reactive responses to HCoV-229E, HCoV-OC43, and MERS spike proteins; (C) quantitative data for the titration. Convalescent serum array responses were compared to an ELISA assay ( Figure 5 ). As ELISA values were all IgG-specific, and AIR data discussed thus far (obtained in a "label-free" mode) was a combination of IgG and IgM-specific responses, these results would not be expected to match precisely. Differences in the expression system used for antigen production (baculovirus for commercial antigens used in AIR; HEK 293T cells used for antigens used in the ELISA assays) could also lead to differences. However, overall trends for SARS-CoV-2 antigens correlate well, as shown in Figure 6 . To provide further detail with regard to the response, AIR assays were run using secondary anti-IgG and anti-IgM antibodies to determine class-specific responses for a subset of samples ( Figure 7 ). was the case with assays run using the laboratory AIR assay, analysis using the ZIVA system readily discriminated between negative and convalescent samples (Figure 8 ). Three putative convalescent COVID-19 samples gave responses on all SARS-CoV-2 antigens that were below the threshold for a positive response (two standard deviations above the average of the 16 negative samples). This is analogous to the AIR and ELISA results obtained for sample HD2146, as described above. The remaining 12 convalescent samples gave strong responses on at least one SARS-CoV-2 antigen, with many responding strongly to both RBD and S2 ( Figure 8 ). with at least one SARS-CoV-2 antigen response above threshold. Health and disease result from many factors, including the overall landscape of a person's immune system. As such, methods for profiling antigen-specific antibody titers to a range of diseases in addition to the disease of primary current interest are of utility when studying the disease. To that end, we have presented preliminary data on a 15-plex array on the AIR platform, developed in response to the need to study SARS-CoV-2 but incorporating antigens for other coronaviruses and influenza. Responses to SARS-CoV-2 antigens on the array effectively discriminated between serum samples from uninfected and COVID-19 convalescent subjects, with generally good correlation to ELISA data. Follow-up assays demonstrated that exposure of the arrays to anti-IgG and anti-IgM antibodies enabled discrimination of antibody isotype. An important aspect of this work is the ability to evaluate anti-SARS-CoV-2 immunity in the context of the individual's overall immune landscape. Because available chip real estate allows for substantial expansion of the multiplex capability of the array, in ongoing efforts we will add additional antigens for other strains of influenza (by analogy to our previous work 17 ), as well as other upper respiratory infections such as respiratory syncytial virus and metapneumovirus. Other coronavirus antigens including nucleocapsid (N) are also likely candidates for addition to the array, as they are known to produce an immune response (as seen in the ELISA results, for example). Thus, the flexibility of the AIR platform will prove useful not only in the current pandemic, but as other viruses inevitably emerge. Report from the American Society for Microbiology COVID-19 international summit Targets of T cell responses to SARS-CoV-2 coronavirus in humans with COVID-19 disease and unexposed individuals Anti-SARS-CoV-2 virus antibody levels in convalescent plasma of six donors who have recovered from COVID-19 Serology assays to manage COVID-19 Developing antibody tests for SARS-CoV-2" The Lancet SARS-CoV-2 seroconversion in humans: a detailed protocol for a serological assay, antigen production, and test setup Diagnostic value and dynamic variance of serum antibody in coronavirus disease Development and clinical application of a rapid IgM-IgG combined antibody test for SARS-COV-2 infection diagnosis Profiling early humoral response to diagnose novel coronavirus disease (COVID-19) Test performance evaluation of SARS-CoV-2 serological assays Development and evaluation of a multiplexed immunoassay for simultaneous detection of serum IgG antibodies to six human coronaviruses Ultra-sensitive high-resolution profiling of anti-SARS-CoV-2 antibodies for detecting early seroconversion in COVID-19 patients A theoretical and experimental analysis of Arrayed Imaging Reflectometry as a sensitive proteomics technique Validation of Arrayed Imaging Reflectometry Biosensor Response for Protein-Antibody Interactions: Cross-Correlation of Label-Free, Arrayed Sensing of Immune Response to Influenza Antigens A multiplex label-free approach to avian influenza surveillance and serology Crowd on a chip: Label-free human monoclonal antibody arrays for serotyping influenza Characterizing emerging canine H3 influenza viruses SARS-CoV-2 seroconversion in humans: a detailed protocol for a serological assay, antigen production, and test setup A serological assay to detect SARS_CoV-2 seroconversion in humans Investigation of non-nucleophilic additives for reduction of morphological anomalies in protein arrays Memory B Cell Expansion by Seasonal Influenza Virus Infection Reflects Early-Life Imprinting and Adaptation to the Infecting Virus Assignment of weight-based antibody units to a human antipneumococcal standard reference serum, lot 89-S Statistical method for determining and comparing limits of detection of bioassays We thank Alicia Papalia for assistance with human samples, and Florian Krammer for the generous donation of plasmids for SARS-CoV-2 antigen production (S1 + S2 ECD and RBD).