key: cord-0969043-gcsi1u5k
authors: Yuen, R.; Steiner, D.; Pihl, R.; Chavez, E.; Olson, A.; Baird, L.; Korkmaz, F.; Urick, P.; Sagar, M.; Berrigan, J.; Gummuluru, R.; Corley, R.; Quillen, K.; Belkina, A.; Mostoslavsky, G.; Rifkin, I.; Kataria, Y.; Cappione, A.; Lin, N.; Bhadelia, N.; Snyder-Cappione, J.
title: An ELISA protocol with resolution at high sample concentration reveals reactive antibodies to SARS-CoV-2 in unexposed individuals
date: 2020-09-18
journal: nan
DOI: 10.1101/2020.09.15.20192765
sha: 90d96a23985947d37e1f1ee53ab56c185701f752
doc_id: 969043
cord_uid: gcsi1u5k

The COVID-19 pandemic has significantly impacted work, economy, and way of life. The SARS-CoV-2 virus displays unique features including widely varying symptoms and outcomes between infected individuals. Sensitive measurement of SARS-CoV-2 specific antibodies would provide new insight into virus transmission dynamics, pre-existing cross-reactive immunity, and the nuances of SARS-CoV-2 pathogenesis. To date, existing SARS-CoV-2 serology tests have limited utility due to insufficient detection of antibody levels lower than what is typically present after several days of symptoms. To measure lower quantities of SARS-CoV-2 IgM, IgG, and IgA with higher resolution than existing assays, we developed a new ELISA protocol with a distinct plate washing procedure and timed plate development via use of a standard curve. This BU ELISA method exhibits very low signal from plasma or serum samples added to uncoated wells at as low as a 1:5 dilution. Use of this method revealed circulating SARS-CoV-2 receptor binding domain (RBD) and nucleocapsid protein (NP) reactive antibodies from blood samples drawn prior to May 2019. Of our pre-pandemic cohort, no SARS-CoV-2 RBD-reactive IgG antibodies were detected in subjects over 70 years of age, and SARS-CoV-2 NP-reactive antibodies were present at similar levels to infected subjects in some individuals and very low in others. Also, samples drawn in May 2020 from two individuals with no symptoms or no known virus exposure contained SARS-CoV-2 RBD-reactive antibodies at intermediate amounts compared with other subject groups (higher than pre-pandemic and lower than confirmed SARS-CoV-2 infected). The one asymptomatic SARS-CoV-2 convalescent subject in our study possessed comparable amounts of SARS-CoV-2 NP-specific IgM and IgG but drastically lower IgA than the symptomatic counterparts. Also, our assay detected positive signal from samples that gave negative results in a commercially available Lateral Flow Device (LFD) and the EUA approved Abbott IgG chemiluminescent microparticle immunoassay for SARS-CoV-2 antibody detection. We propose that this improved ELISA protocol, which is straightforward to perform, low cost, and uses readily available commercial reagents, is a useful tool to elucidate new information about SARS-CoV-2 infection and has promising implications for improved detection of all analytes measurable by this platform.

From the first reported case of COVID-19 caused by the virus SARS-CoV-2 in December 2019 1,2 there have been more than 28 million reported cases and 900,000 deaths worldwide as of September 11, 2020 . Common symptoms of SARS-CoV-2 infection include fever, cough, myalgia, and fatigue and these symptoms vary widely in magnitude, nature, and duration between individuals for reasons that are not clear to date 3,4 , with some individuals with confirmed infections remaining asymptomatic 5 . Epidemiological evidence indicates silent viral spread via asymptomatic individuals within communities and the extent of this form of transmission is currently unclear 6 . SARS-CoV-2 has homology to other alpha and beta 'common cold' coronaviruses in circulation, and cross-reactive T cell immunity to SARS-CoV-2 spike and nucleocapsid proteins are present in a substantial percentage of unexposed individuals [7] [8] [9] [10] . It is postulated that this cross-reactive immunity may influence the nature and severity of COVID-19 symptoms upon infection and impact disease course 11 .

Sensitive and accurate detection of virus-specific immune factors, such as antibodies, is imperative in order to measure rates of SARS-CoV-2 infections within communities with greater accuracy, to more fully define crossreactive immunity in unexposed individuals, and to gain new understanding about the nature of effective versus potentially deleterious immune responses upon SARS-CoV-2 exposure. Antibody measurements are of particular importance, as pathogen-specific immunoglobulins are a known first line of defense upon exposure and can prevent new infections. Antibody titers are used to assess both likelihood of protection from reinfection and general vaccine efficacy 12 . A variety of SARS-CoV-2 serological assays have been developed by multiple manufacturers and academic institutes and many are CE-marked and granted emergency use authorization (EUA) from the US Food and Drug Administration (FDA). Varieties include point-of-care rapid lateral flow assays, chemiluminescence immunoassays (CLIA), and enzyme-linked immunosorbent assays (ELISA) [13] [14] [15] [16] . These tests detect antibodies that primarily target the nucleocapsid protein (NP) or the spike (S) protein of SARS-CoV-2, and specifically the Receptor Binding Domain (RBD) of spike which is an immunodominant surface protein targeted by neutralizing antibodies and a main target antigen for vaccine development [16] [17] [18] . Some of these tests possess high sensitivity and specificity for detection of SARS-CoV-2 4 antibodies 14 days after diagnosis and/or symptom onset [19] [20] [21] [22] . However, others report negative results from individuals who are asymptomatic, mildly symptomatic, or symptomatic for less than 14 days, even when SARS-CoV-2 infection is confirmed 15, 23, 24 ; whether such individuals possess antibodies below the limit of the detection of the particular test used or lack these antibodies altogether is unresolved.

To enable detection of low levels of SARS-CoV-2-reactive antibodies, we modified the standard ELISA procedure, particularly the plate washing method, to improve sensitivity. Our protocol (the 'BU ELISA') allows clear SARS-CoV-2-reactive antibody signal resolution at sample dilutions as low as 1:5. Using this protocol we were able to detect low levels of SARS-CoV-2-specific IgM, IgG, and IgA with low background:signal ratio from plasma or serum from three groups of individuals: (1) subjects that varied by age, HIV infection, and systemic lupus erythematosus (SLE) disease status with all samples collected before November 8 th , 2019 ('prepandemic'); (2) eight individuals with no symptoms or known exposure to SARS-CoV-2 with samples collected in the spring of 2020 ('pandemic'), and (3) five subjects with confirmed SARS-CoV-2 infection ('SARS-CoV-2 infected'), including one asymptomatic individual and one who was hospitalized and received two convalescent plasma infusions. In addition, the BU ELISA measured signal that was not detected with Antagen's IgM IgG Lateral Flow Device (LFD) test and/or the Abbott IgG chemiluminescent microparticle immunoassay (CMIA).

Our results demonstrate the exceptional sensitivity of the BU ELISA protocol for the detection of low levels of SARS-CoV-2 reactive antibodies.

A modified ELISA protocol demonstrates low noise from high concentration human serum and plasma samples. The Enzyme-Linked Immunosorbent Assay (ELISA) is a commonly used method for the measurement of analytes in a suspension sample. While low cost and easy to adapt in most lab settings, a limitation of this platform is high background from some biological samples at low sample dilutions.

Specifically, optical densities (ODs) from sample dilutions lower than 1:100 is often sizeable and can mask the analyte of interest. This issue is particularly germane to serologic testing for SARS-CoV-2, as antibodies that 5 are cross-reactive in unexposed individuals, newly generated in asymptomatic and/or recent infections, induced from an encounter with low viral dose, or waned post convalescence may be missed because the antibody levels are below the limit of detection of current assays. To address this issue, we have developed an ELISA protocol with unique steps to reduce non-specific signal at low sample dilutions. One change is the plate washing procedure, which is performed manually by an operator using a multichannel pipettor and includes agitation and soaking steps with repeated complete removal of residual fluid as described (Methods and Supplemental Figure   1 ). ELISAs were performed that compared uncoated well OD values of five human plasma samples with plates washed with our method or an automated plate washer and the total levels of non-specifically bound IgG was determined. The manual washing procedure resulted in a notably lower average and range of ODs at 1:5, 1:10, and 1:25 dilutions as compared with the automated washer (Fig. 1A ). This BU ELISA protocol was run on samples from a total of 40 pre-pandemic and 13 healthy or SARS-CoV-2-infected pandemic subjects (Table 1) , with paired antigen coated and uncoated wells for seven sample dilutions (plate map shown in Supplemental Figure 2 ) for all but one subject (CONV1). The average ODs for uncoated, 1:5 diluted sample loaded wells from all 53 subjects were 0.16, 0.09, and 0.068 for IgM, IgG, and IgA respectively ( Figure 1B ). Given these low background OD values and the results from the wash method comparison, it's possible that details of our protocol other than the washing method may contribute to these low background ODs, such as the type of plates, the blocking buffer/sample diluent used, and the number and placement of washing steps (Methods and Supplemental Figure 1 ). This no coat 'noise' is remarkably consistent between multiple runs of a given sample (Supplemental Figure 3 ) and appears to be due to components within the sample, such as IgG and inflammatory factors 25 and not due to assay variability. Importantly, when ODs from uncoated wells with the same dilution of sample are not measured and properly subtracted, incorrect interpretation of results as positive can occur 26 ; therefore, the no coat values were subtracted from coated OD results for all results in this manuscript with the exception of one subject (CONV1). Next, to determine if age or disease status impacts background signal in our assay, we compared the results from uncoated wells for the 1:5 and 1:10 dilutions between younger (<34 years old), older (>64 years old), HIV+, SLE, pandemic, or SARS-CoV-2 infected groups and found no significant differences in signal with the exception of IgA between SLE and two other groups ( Figure 1C ).

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . https://doi.org/10.1101/2020.09. 15.20192765 doi: medRxiv preprint Modification of ELISA development duration based on standard curve signal detection enables accurate comparison of antibody levels between experimental runs by minimizing impact of OD drift. During assay development we noted differences in OD values in different experimental runs even with strict adherence to all procedures and length of steps. Therefore, for all sample runs, we included a standard curve using recombinant monoclonal IgM, IgG, and IgA antibodies that recognize SARS-CoV-2 RBD for each of the respective isotype assays and stopped the development reaction when there was a visible difference between the seventh dilution (1.37ng/ml) of the standard and the 'zero' (sample diluent only) well. Addition of these standards and timing of development in this manner helped to ensure accurate calculation of the relative antibody levels (Arbitrary Units, as described below) between samples run on different days, plates, and/or by different operators. The OD values of the standard curves following this development procedure for the IgM, IgG, and IgA assays for 17 representative runs are shown ( Figure 2 ). The development time of these runs to complete visualization of the standard curve development ranged from ~8-20 minutes, demonstrating the need to adjust substrate incubation time per experimental run to maximize signal detection.

SARS-CoV-2 RBD cross-reactive antibodies are present in some unexposed individuals. SARS-CoV-2 RBD IgM, IgG, and IgA ELISA assays were performed on 40 pre-pandemic samples (Table 1) using the BU ELISA protocol. For these experiments, in addition to the recombinant antibody (standard) controls mentioned above, a dilution series from a COVID-19+ subject (CONV1) was included on each plate as a biological positive control. CONV1 was hospitalized due to COVID-19 complications and received two convalescent plasma infusions four weeks prior to the collection of sample used in these assays. After first subtracting the average ODs of the blank wells with the same coat from the sample wells, and secondly subtracting the OD of the dilution-paired sample in uncoated wells, several subjects exhibited a clear positive signal to the SARS-CoV-2 RBD coat, with a linear loss of OD with sample dilution, such as subjects Y3, Y6, H3, S3, S8, and S9 ( Figure 3A ). To determine positive signal, we did not use a set cut-off value but instead developed a metric based on the relationship between no coat and coated OD values from individual subjects. This metric, called . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 7 Metric 1, is comprised of the following: (1) the OD from the RBD-coated well must be ≥2.5x the uncoated well from the paired sample dilution for at least two dilutions in the series, and (2) the OD from the RBD coated well must be ≥0.1 for at least one dilution. Following this guideline, roughly half (18/40) of the unexposed prepandemic individuals possessed reactive antibodies of at least one isotype to SARS-CoV-2 RBD, albeit often at very low levels compared with the CONV1 sample ( Figure 3A ).

Unlike younger counterparts, no IgM or IgG SARS-CoV-2 RBD reactive antibodies were detected in unexposed individuals >70 years old. We noted the low frequencies of subjects with SARS-CoV-2 RBD reactive antibodies in the older pre-pandemic subject group ( Figure 3A) ; therefore, to further investigate the potential impact of age on the presence of these antibodies, we re-classified the 40 pre-pandemic subjects into one of three age groups (less than 50, between 50-70, and greater than 70 years old) and calculated the total number of subjects that possess SARS-CoV-2 RBD reactive antibodies according to Metric 1 to each or any of the three isotypes in these groups. We found similar percentages of subjects with SARS-CoV-2 reactive antibodies in the <50 and 50-70 age groups; however, among those over 70 years old, only one of the nine individuals possessed detectable levels of SARS-CoV-2 RBD reactive antibodies, and only of the IgA isotype ( Figure 3B ).

Samples collected in May 2020 from two subjects with no symptoms or known exposure to SARS-CoV-2 contained SARS-CoV-2 RBD-specific IgG at levels higher than unexposed individuals. Samples from eight subjects collected in May 2020 with no symptoms or known SARS-CoV-2 exposure were tested for reactive antibodies to SARS-CoV-2 RBD using the BU ELISA protocol. Surprisingly, two of the eight subjects had SARS-CoV-2 RBD-specific IgG OD values over 3.5 at 1:5 dilution and ≥2.5 times higher than the corresponding no coat well for at least five dilutions (Figure 4) , a magnitude of antibody concentration not found in any of the 40 pre-pandemic samples ( Figure 3A ). These two subjects, C1 and C5, also showed significant levels of IgA SARS-CoV-2 RBD-reactive antibodies with ODs as high as only two of 40 pre-. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . 8 pandemic subjects (Y6 and H3, Fig. 3A ). Also, positive signal for SARS-CoV-2 RBD IgM reactive antibodies was found in subject C1; C1 was positive for all three isotypes tested ( Figure 4 ).

subjects using the BU ELISA protocol. Plasma or serum samples from five subjects with confirmed SARS-

CoV-2 infection were tested with the BU ELISA protocol. Four of these subjects were infected at a biotechnology conference in Boston, Massachusetts in March of 2020. Three of the four were symptomatic with length of symptoms ranging from 10 days (subject B1) to 61 days (subject B4), and one was asymptomatic (B2). Samples from these four subjects were collected during May-June 2020 more than 30 (range 36-59) days since the resolution of symptoms. The fifth subject, CONV1, received two transfers of convalescent plasma three days apart four weeks prior to sample collection and remained symptomatic at the time of sample collection. Clear positive signal was detected from all five subjects for all three isotypes ( Figure 4 ). CONV1

possessed the highest levels of IgM and IgA, and the highest SARS-CoV-2 RBD reactive IgG levels in this group was found from subject B03 (Figure 4) .

Arbitrary Unit (AU) calculations reveals reactive IgM and/or IgA in three pre-pandemic subjects at levels greater than or similar to SARS-CoV-2 convalescent subjects and a gradation of IgG among positive samples in the pre-pandemic, pandemic, and SARS-CoV-2 infected groups. To determine the relative amounts of SARS-CoV-2 RBD reactive antibodies between samples tested during different experimental runs, with different operators, and/or on different plates, we calculated the arbitrary units (AU) within each sample extrapolated from the standard curve from each individual run as described (Methods). This analysis yielded some surprising results; for example, the pre-pandemic subjects H3 and S3 possess SARS-CoV-2 RBD reactive IgM antibody levels higher than subjects B1, B2, and B4 from the SARS-CoV-2 infected group ( Figure 5A) ; also, subjects Y6 and H3 possess IgA levels similar to B1, B2, B3, and B4 ( Figure 5B ). IgG results yielded a different pattern of antibody levels between the groups; pre-pandemic subjects were found to . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. .

have the lowest quantities, followed by intermediate amounts in the pandemic group (driven by C1 and C5 results), and the highest IgG levels were present in the five SARS-CoV-2 infected subjects ( Figure 5C ). We performed further analysis, where we included the AU values from only subjects determined to have positive signal by both Metric 1 and Metric 2 (AU values greater than 30, 28, and 27 for IgM, IgG, and IgA as described in Methods and depicted as dotted lines on the graphs in Figure 5A , B, and C). From subjects determined to have positive signal by both metrics, a statistically significant difference was found between both the prepandemic and pandemic groups as compared with the SARS-CoV-2 infected for SARS-CoV-2 RBD-reactive IgG ( Figure 5D ).

CoV-2 infected samples tested, respectively, with no signal measured from pre-pandemic and pandemic subjects. Samples from our pre-pandemic, pandemic, and SARS-CoV-2 infected groups were tested with a lateral flow rapid test (Antagen Pharmaceuticals) for the qualitative determination of IgM and IgG antibodies to SARS-CoV-2. The test detected SARS-CoV-2 RBD reactive IgG exclusively in the confirmed SARS-CoV-2 infected group, with all four tested subjects showing clear IgG positive bands ( Figure 5E ). SARS-CoV-2 reactive IgM was detected from subjects B3 and B4, but not B1 or B2, which were found to have very low but positive SARS-CoV-2 RBD reactive IgM levels in the BU ELISA protocol (Figures 4, 5A ). Pre-pandemic subjects H3 and S3, both found to have higher levels of IgM than subject B4 by the BU ELISA had negative results using this test ( Figure 5E ). Also, none of the pandemic samples had positive results for IgG in this assay, including subjects C1 and C5 ( Figure The vast majority of pre-pandemic samples contained circulating IgG and IgA antibodies reactive with SARS-CoV-2 nucleocapsid protein (NP) with bimodal levels (high or low) found among the nonautoimmune subjects. Using the BU ELISA protocol, we measured IgM, IgG, and IgA levels reactive with . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . https://doi.org/10.1101/2020.09.15.20192765 doi: medRxiv preprint SARS-CoV-2 nucleocapsid protein (NP) from 20 subjects from our pre-pandemic cohort, all eight pandemic subjects, and B1, B2, B3, and B4 from the SARS-CoV-2 infected group. Seven dilutions were run for all samples with and without NP coated wells as with the SARS-CoV-2 RBD assays; sample dilution curves were generated and positive results were determined using Metric 1. To our surprise, all but one of the 20 prepandemic subjects possessed IgG antibodies reactive with SARS-CoV-2 NP ( Figure 6A ). The one negative subject, O4, is 95-100 years old. Only 5/20 pre-pandemic subjects scored positive for IgM, while 4/5 pandemic and 3/4 of the SARS-CoV-2 infected were positive for this isotype (Fig. 6A ). The majority of subjects in all groups possessed SARS-CoV-2 NP-reactive IgA ( Figure 6A ). As with the SARS-CoV-2 RBD results, relative antibody levels between subjects were quantified using Arbitrary Unit calculations from recombinant antibody standard curves. Two pre-pandemic subjects, O1 and S3, possessed equivalent or higher levels of SARS-CoV-2 NP reactive IgM as subject B3, the SARS-CoV-2 infected subject with the highest level of NP-reactive IgM of that group ( Figure 6B ). The SARS-CoV-2 NP IgG and IgA from the pre-pandemic group possessed levels in a large range between different subjects, with no observable association with age ( Fig. 6B ). Among the non-SLE pre-pandemic subjects tested in this assay, there appears to be bi-modal presence of antibody levels, with seven of the 15 subjects with levels all below 154 AU and the other eight with AU of at least 738 for IgG ( Figure 6B ).

Also, the one SARS-CoV-2 infected subject with no symptoms (B2) possessed a level of SARS-CoV-2 NP reactive IgA dramatically lower than any of the other three subjects included in this assay from this group ( Figure 6B ).

Abbott chemiluminescent microparticle immunoassay (CMIA) IgG antibody test used for the detection of SARS-CoV-2 infection was performed on all eight pandemic subjects and subjects B1, B2, B3, and B4. All eight pandemic samples scored negative by this test, and all four from the SARS-CoV-2 infected group scored positive (Red symbols, Figure 6B , and Supplemental Figure 4 ).

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . https://doi.org/10.1101/2020.09.15.20192765 doi: medRxiv preprint

A variety of SARS-CoV-2 serological assays have been produced in record time as antibody testing has the potential to play a pivotal role in combating the COVID-19 pandemic. Specifically, accurate and sensitive measurement of virus-specific antibodies: (1) could complement diagnostic testing, (2) provide information about the true prevalence of infection, (3) provide insight into anti-viral immunity, (4) help assess vaccine efficacy, and (5) will serve as a screening tool of convalescent plasma for clinical use. However, a lack of required sensitivity and specificity of many of the SARS-CoV-2 antibody tests available to date have led some to conclude that they have limited clinical utility in combating COVID-19 27 . Here, we present a modified ELISA protocol with exceptional sensitivity with high concentration samples that enables the detection of low levels of antigen-specific antibodies in human specimens.

The BU ELISA is straightforward, comprised of reagents that are readily available from commercial vendors and can easily be adapted for other applications and analytes. However, a major limitation of this assay is that it is currently considerably lower in throughput compared to other serological platforms. This protocol requires an operator for the manual wash steps, limiting the number of plates that can be run compared to automated methods. There is potential for throughput increase if automated washers/ELISA systems can be adapted to more closely mimic this protocol.

Other important features of our approach include the inclusion of paired sample dilutions with uncoated wells to enable detection of true antigen-reactive signal and adjustment of the length of substrate incubation based on standard curve development for OD standardization to enable direct comparison between samples on different plates. Quantification of relative antibody levels via Arbitrary Units (AU) or a similar method will be imperative for determining which convalescent samples have antibody levels sufficient for effective plasma transfer as well as other applications. However, while we believe this is a preferred approach for determining of relative output values within all samples, it is critical to note that the unique dynamics of the panoply of . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. .

antibodies of varying affinities and isotypes within a given specimen causes inherent confounding factors to serologic readouts. Specifically, all isotypes that bind a given antigen may be present in a specimen and there will be competition and steric hindrance that skew the accuracy of the final signal for a given isotype. For example, a specimen with a high level of SARS-CoV-2 RBD reactive IgM antibodies could have a lower detected signal for IgG and IgA due to IgM's pentameric conformation blocking many binding sites. Also, higher affinity antibody clones (IgG and IgA vs IgM, for example) may outcompete for binding sites of the coated antigen and thereby be detected more readily than others. We can account for this issue to some extent via measurement of all three major isotypes in all samples; however, for this reason we do not attempt to quantify antibodies as the number of molecules per sample in this report.

Direct comparison of our ELISA protocol with two commercially available serological assays for SARS-CoV-2, Antagen's DISCOVID LFD test and Abbott's CMIA IgG assay, yielded thought-provoking results. Both the LFD and CMIA tests performed well in identification of the confirmed SARS-CoV-2 infected subjects via detection of SARS-CoV-2 IgG to RBD (Antagen test) and NP (Abbott test). The BU ELISA protocol detected signal from many pre-pandemic and pandemic samples which all scored negative in the Antagen and Abbott tests ( Figure 6 and Supplemental Figure 4 ). We postulate that these tests contain manufacturer-curated portions of the RBD and NP proteins that have epitopes recognized by antibodies selectively found in SARS-CoV-2 infected individuals. In addition, low frequencies of B cells were found in unexposed subjects that bind SARS-CoV-2 spike at a lower affinity and with lower levels of clonal expansion and somatic hypermutation than a SARS-CoV convalescent subject 28 . These results support our findings and taken together, underscore the value of comprehensive testing of antibody reactivity to SARS-CoV-2 proteins in different subject groups to gain new insight into how pre-existing and newly primed immune responses interplay to impact infection outcomes.

Our results from pre-pandemic samples show that roughly half of the subjects tested possessed antibodies that bind to SARS-CoV-2 RBD, albeit at very low levels on average, as determined by our Metric 1 cutoff and via AU calculations (Metric 2) (Figures 2, 3) . These results are in contrast with the conclusions of other reports, . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. .

which state that RBD-reactive antibodies are not detected in unexposed individuals 13, 16 . However, in these studies the assays were run at higher sample dilutions and therefore low signal may have been missed or misinterpreted as noise. The clinical significance of these antibodies is unknown and while the levels are low in the blood, it is possible if they are present in higher concentrations in other sites, such as the mucosa. Also, none of the nine pre-pandemic subjects over the age of 70 possessed antibodies of the IgM or IgG isotypes that recognized SARS-CoV-2 RBD, with one of nine positive for reactive IgA ( Figure 3C ). As individuals over 70 are more likely to present with serious COVID-19 complications 29-32 future research investigating connections between age and the levels of these RBD-reactive antibodies in a larger pre-pandemic group is warranted. Also, as antibodies to RBD as associated with virus neutralization both in vitro and in animal models, 14, [33] [34] [35] performing detailed functional analyses of plasma samples from pre-pandemic samples with RBD-reactive antibodies is an important next step. Preliminary experiments from our group indicated that neutralization activity was not present in subjects C1, C5, H3 and S3 (data not shown) but future experiments are needed to more thoroughly address this question.

To our surprise, high quantities of reactive antibodies with SARS-CoV-2 NP was present in some of the prepandemic samples tested, with some subjects possessing levels similar to what was measured from confirmed SARS-CoV-2 infected subjects ( Figure 6 ). These results complement recent reports of high percentages of unexposed individuals with circulating T cells reactive with SARS-CoV-2 NP peptides 7-10 . We found it of particular interest that among the non-autoimmune pre-pandemic subjects screened for anti-NP reactive IgG antibodies, there is bimodal expression of signal, with roughly 7/15 possessing very low/no antibodies (highest AU 154) and 8/15 with AU values all over 738 ( Figure 6 ). These results could reflect infection history in recent months/years with common cold coronaviruses in the circulation in the United States with homology to SARS-CoV-2; next steps include measurement of both SARS-CoV-2 and seasonal common cold alpha (NL63 and 229E) and beta (OC43 and HKU1) coronavirus antibodies in individual pre-pandemic subjects for comparative analysis. Importantly, as with the recently reported cross-reactive T cell immunity, the role of the SARS-CoV-2 NP reactive antibodies reported here in unexposed individuals is unknown. This reactive immunity could be . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . https://doi.org/10.1101/2020.09.15.20192765 doi: medRxiv preprint helpful to the host, preventing symptoms upon infection and/or lessening the severity of COVID-19, as such a phenomenon been reported with influenza [36] [37] [38] [39] ; alternatively, this immunity could lead to worse infection outcomes. This latter possibility is of particular interest in light of a recent report of the higher in vitro assessed functional activity of anti-NP antibodies among hospitalized individuals with COVID-19 who later died as compared with surviving counterparts 40 . Gaining insight into the role, if any, of this cross-reactive immunity to SARS-CoV-2 in unexposed individuals could reveal why diverse disease outcomes exist upon viral exposure.

Our 'pandemic' subject group was comprised of eight subjects who self-reported no symptoms or viral exposure and donated blood at Boston Medical Center in May 2020; and surprisingly, two of the eight subjects (C1 and C5) possessed SARS-CoV-2 reactive IgG at concentrations higher than any of the 40 pre-pandemic samples yet lower than all five confirmed SARS-CoV-2 infected subjects, including one asymptomatic (Figures   3,4) . C1 and C5 also possessed anti-RBD antibodies of other isotypes as well (Figures 3,4) . As it is becoming evident that dose of SARS-CoV-2 exposure is a putative critical factor driving productive infection and COVID-19 onset and progression [41] [42] [43] [44] , we speculate that this intermediate level of RBD-reactive IgG could be due to exposure to a low innoculum of virus. As individuals with mild or asymptomatic infection develop less robust antibody responses and may have a more rapid decline of antibody levels as compared with symptomatic counterparts 24,45 these results could reflect another, even lower dose of virus exposure and response that is not measurable by other assays and could potentially reveal new information about viral transmission and immunity. Also, the gradation in SARS-CoV-2 RBD IgG antibody level between pre-pandemic, pandemic, and our confirmed SARS-CoV-2 infected subject groups indicate the possibility of using antibody levels as a tool for diagnostics.

Another notable result in this report is the SARS-CoV-2-NP specific isotype signature from the asymptomatic SARS-CoV-2 convalescent subject (B2) (Figure 6 ). This subject's specimen possessed levels of NP-specific IgM and IgG comparable to the other infected subjects, yet uniquely displayed extremely low levels of NPreactive IgA (Figure 6 ). IgA is a major isotype produced in the mucosa, and it is known that gastrointestinal . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The BU ELISA Protocol: Antibodies reactive to SARS-CoV-2 RBD or NP were assayed from sera or plasma as described in accompanying SOP (Supplemental Figure 1) . Briefly, wells of 96-well plates (Pierce 96-Well . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 18, 2020. . https://doi.org/10.1101/2020.09.15.20192765 doi: medRxiv preprint 16 Polystyrene Plates; cat#15041, Thermo Fisher Scientific) were coated with 50µl/well of a 2µg/ml solution of each respective protein in sterile PBS (Gibco) or with PBS only for 1 hour at room temperature. Coating solution was removed manually by a swift flick of the plates into a biohazard waste container. Next, 200µl per well of sterile PBS was added with a multichannel pipettor and liquid was removed via swift flick and the plate was banged on absorbent paper towels to remove residual liquid; this washing procedure was performed three times. Next, 200µl of casein blocking buffer (Thermo Fisher Scientific, cat#37528) was added to wells at room temperature for 1 hour. Next, plates were washed three times as previously described. Subject samples and . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 18, 2020. . https://doi.org/10.1101/2020.09.15.20192765 doi: medRxiv preprint Antigens: SARS-CoV-2 RBD was a gift from the Schmidt lab at the Ragon Institute and was expressed and purified as previously described 52 . SARS-CoV-2 NP was purchased from Sino Biological, (Cat# 40588-V08B). LFD tests: Antagen's DISCOVID IgM IgG LFD test was used to detect SARS-CoV-2 RBD specific IgM and IgG antibodies following manufacturer instructions. Briefly, 20µl of plasma or serum was added to the . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted September 18, 2020. .

indicated sample port, immediately followed by provided diluent, and incubated at room temperature before reading at 45 minutes. The results were scored as positive or negative for IgM and IgG by two independent readers blinded to donor sample status.

The SARS-CoV-2 IgG assay is a chemiluminescent microparticle immunoassay (CMIA) used for the detection of SARS-CoV-2 nucleocapsid protein-specific IgG in human samples. The assays were performed according to manufacturer's protocol.

Automated Washer: Plates were washed with Molecular Devices SkanWasher 400 microplate washer with three rounds of aspiration and wash with a final aspiration step for each run. This protocol was run twice after the coating, blocking, and sample incubation steps and three times after the addition of the secondary detection antibody step in the experiment shown in Figure 1A . Plates were rotated 180° between each run. Residual wash buffer was left in the plates (plates were not blotted post-wash) to mimic a fully automated system. 

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

. CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. .

The authors declare that the research was conducted in the absence of any personal, professional, or financial relationships that could potentially be construed as a conflict of interest.

Boston University filed for patent protection of this method on August 12, 2020.

We thank Dr. Aaron Schmidt/Dr. Jared Feldman and Dr. Galit Alter from the Ragon Institute of MGH, MIT, and Harvard for the gifts of the recombinant SARS-CoV-2 RBD antigen and monoclonal SARS-CoV-2 RBD IgG antibody, respectively. We also thank Dr. Andrew Lodge at ThermoFisher Scientific for his advice regarding initial reagent selection and protocol steps.

This work was supported by the Boston University's National Emerging Infectious Diseases Laboratories (NEIDL) Director's Fund. and IgA were detected in individual assays. One value from the 3,125 dilution of IgG from one sample was . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . 20 removed as it was presumed to be due to a pipetting error. (C) No coat OD values from the groups in this study, which includes pre-pandemic subjects sub-grouped via age, HIV, and SLE status as well as pandemic and confirmed SARS-CoV-2 infected subjects. 1-way ANOVA tests found no significant differences of 1:5 dilution OD values between the groups for all isotypes except IgA SLE versus pandemic (p=.0255) and IgA SLE versus SARS-CoV-2 infected (p=0.044). . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted September 18, 2020. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72314 Cases From the Chinese Center for Disease Control and Prevention

A Systematic Review of Asymptomatic Infections with COVID-19

Targets of T Cell Responses to SARS-CoV-2 Coronavirus in Humans with COVID-19

SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, and uninfected controls

SARS-CoV-2-reactive T cells in healthy donors and patients with COVID-19

Selective and cross-reactive SARS-CoV-2 T cell epitopes in unexposed humans

Pre-existing immunity to SARS-CoV-2: the knowns and unknowns

Vaccines: correlates of vaccine-induced immunity

A serological assay to detect SARS-CoV-2 seroconversion in humans

Convergent Antibody Responses to SARS-CoV-2 Infection in Convalescent Individuals

Severe Acute Respiratory Syndrome Coronavirus 2-Specific Antibody 25

Hospitalization Rates and Characteristics of Patients Hospitalized with Laboratory-Confirmed Coronavirus Disease 2019 -COVID-NET, 14 States

Coronavirus Disease 2019 Case Surveillance -United States

Case-Fatality Rate and Characteristics of Patients Dying in Relation to COVID-19 in Italy

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Primary exposure to SARS-CoV-2 protects against reinfection in rhesus macaques

Isolation of potent SARS-CoV-2 neutralizing antibodies and protection from disease in a small animal model

A SARS-CoV-2 surrogate virus neutralization test based on antibody-mediated blockage of ACE2-spike protein-protein interaction

Significant cross reactive antibodies to influenza virus in adults and children during a period of marked antigenic drift

A systematic review and meta-analysis of crossreactivity of antibodies induced by H7 influenza vaccine

Influenza Infection in Humans Induces Broadly Cross-Reactive and Protective Neuraminidase-Reactive Antibodies

Cross-reactive antibody responses to the 2009 pandemic H1N1 influenza virus

Distinct Early Serological Signatures Track with SARS-CoV-2 Survival

Viral dynamics in mild and severe cases of COVID-19

Risk of COVID-19 in health-care workers in Denmark: an observational cohort study

SARS-CoV-2 viral load predicts COVID-19 mortality

Exposure to SARS-CoV-2 in a high transmission setting increases the risk of severe COVID-19 compared with exposure to a low transmission setting?

Antibody responses to SARS-CoV-2 in patients of novel coronavirus disease 2019

The inflammatory function of human IgA

Dis)Trust your gut: the gut microbiome in age-related inflammation, health, and disease

Microbial translocation is a cause of systemic immune activation in chronic HIV infection

Evidence for Gastrointestinal Infection of SARS-CoV-2