COVID-19 diagnosis and study of serum SARS-CoV-2 specific IgA, IgM and IgG by a quantitative and sensitive immunoassay

Background The current pandemic of the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has caused a great loss in lives and economy. Detecting viral RNAs on nasopharyngeal and throat swabs is the standard approach for SARS-CoV-2 diagnosis with variable success. Currently, there are only a few studies describing the serological diagnostic methods that involve the detection of SARS-CoV-2-specific IgM and IgG. Here, we aimed to develop a more quantitative and sensitive serological test for COVID-19 diagnosis, monitoring and clinical investigation, based on the detection of antigen-specific IgA as well as IgM and IgG in blood in response to SARS-CoV-2 infection. Methods In this investigation, we report the development of a set of validated diagnostic kits for detecting serum IgA, IgM, and IgG specific to SARS-CoV-2 nucleocapsid protein (NP) and receptor-binding domain (RBD) of the spike protein by chemi-luminescence immuno-analysis. The kits were tested with a cohort of 216 sera from 87 laboratory-confirmed COVID-19 patients, and 483 sera from SARS-CoV-2 negative or healthy individuals as negative controls. A standard receiver operating characteristic (ROC) analysis was conducted to evaluate the diagnostic accuracy. Using the kits, serum levels of IgA, IgM, and IgG were analyzed, in response to SARS-CoV-2 infection and COVID-19 pathogenesis. Findings The diagnostic kits based on the RBD antigen outperformed those based on the NP. RBD-specific IgA, IgM, and IgG detection kits showed sensitivities of 98.6%, 96.8%, and 96.8%, and specificities of 98.1%, 92.3%, and 99.8%, respectively. In addition, using purified RBD-specific immunoglobulins from a serum pool of COVID-19 patients as standards, the serum concentrations of RBD-specific IgA, IgM, and IgG proteins were determined. The concentrations varied widely among different patients. Median concentration of IgA and IgM reached peaks at 16-20 days after illness onset at 8.84 μg/mL and 7.25 μg/mL, respectively, while median concentration of IgG peaked during 21-25 days after illness onset at 16.47 μg/mL. Furthermore, the serum IgA level positively correlates with COVID-19 severity. Interpretation Our immunoassay of measuring SARS-CoV-2 specific antibodies IgA, IgM, and IgG in serum provides a better serological testing with improved sensitivity and specificity. Data of IgA, IgM, and IgG responses in blood of COVID-19 patients may provide novel insight for the monitoring and treatments of COVID-19. The kits are also suitable for epidemiological studies and vaccine validations.

The diagnostic kits based on the RBD antigen outperformed those based on the NP. RBD-specific IgA, 70 IgM, and IgG detection kits showed sensitivities of 98· 6%, 96· 8%, and 96· 8%, and specificities of 98· 1%, 71 92· 3%, and 99· 8%, respectively. In addition, using purified RBD-specific immunoglobulins from a   CC-BY-NC-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) The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint a sensitivity below 70%. 9-13 Therefore, there is an urgent need for more reliable and rapid diagnostic 126 methods to screen SARS-CoV-2 infected people including those who do not have overt symptoms. A 127 serological test of virus-induced antibody production has unique advantages in clinical diagnostics, 128 especially for identifying people who acquired immunity against pathogens without noticeable 129 symptoms. 14 When the virus invades host, the body produces large amounts of immunoglobulin (Ig) by 130 the immune system and released into blood, among them, IgG, IgM, and IgA isotypes. 15 It has been 131 widely believed that IgM is the first antibody to be transiently synthesized in response to the virus 132 invasion. 15 IgG is a major class of immunoglobulins found in the blood, comprising 75% of total serum 133 immunoglobulins and has long-term immunity and immunological memory. 15, 16 Therefore, a 134 combination of IgM and IgG has been used in various serological tests for detecting infection of SARS-

135
CoV-2 as previously used for SARS and other coronaviruses. 10,11,14,[17][18][19][20][21] In contrast, IgA, which is mainly 136 produced in mucosal tissues to hinder virus invasion and replication but also detected in blood (~15% of 137 total immunoglobulins in blood), 22 has not been widely used in serological tests for detecting coronavirus 138 infection. IgA's production kinetics and roles in anti-viral immunity of IgA are even less known.

139
Currently, only a few published studies reported diagnosis of COVID-19 by using ELISA or "flow 140 immunoassay" for detection of serum IgM and IgG with limited accuracy, 10, 11, 17-20 although SARS-CoV-141 2 specific IgA in serum was also detected in recent papers or a preprint. 12, 23, 24 The kinetics of antibody 142 responses in COVID-19 remains undefined, specifically for IgA production.

143
In this study, we designed and evaluated a set of sensitive and quantitative kits to measure serum IgA,   CC-BY-NC-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) The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint underline illnesses; the most common one was hypertension in eighteen patients (20· 7%). A total 216 168 serum samples were taken from the 87 COVID-19 patients.

169
Negative controls and potentially interfering non-COVID-19 patient serum samples were collected in 170 order to evaluate the reliability of the kits. This cohort contains 330 sera from obviously healthy people, 171 fifteen sera from once suspected cases (RT-qPCR negative but had typical manifestation of pneumonia) 172 and 138 sera from other patients with different underlying diseases. All sera were stored at -20°C.

185
Diagnostic kit preparation and testing 186 Briefly, the purified NP or RBD viral antigens were coated to magnetic particles to catch SARS-CoV-187 2 specific IgA, IgM, and IgG in patient sera. Then a second antibody that recognizes IgA, IgM, or IgG is

207
. CC-BY-NC-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)
The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint Based on the clinical RT-qPCR diagnosis results of SARS-CoV-2 infection, receiver operating 208 characteristic (ROC) analysis was conducted using MedCalc software to determine the optimal cut-off 209 value (criterion) and evaluate the diagnostic value of NP-or RBD-specific IgA, IgM, and IgG kits.

210
The specificity and sensitivity of the antibody kits were calculated according to the following formulas:

214
In order to analyze the correlation of serum antibody levels and age with disease severity, we first used 215 the Kruskal Wallis test 26 to test if there is any significant difference of IgA among the three groups (Mild,

216
Moderate, Severe). Then Dunn's test 27 was used to perform a pair-wise test between each group, and 217 Benjamini-Hochberg procedure 28 was used to adjust p-values. All the above analyses use R software 218 version 3.6.1 29 . A p value less than 0· 05 was judged statistically significant.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint and 153 non-COVID-19 patients also indicate that our RBD-based detection kits did not cross-interact 250 with antibodies raised against other human coronaviruses presenting in ~15% of common cold cases and 251 also causing pneumonia. Taken together, our detection systems are highly specific to SARS-CoV-2 252 RBD.

253
We attempted to analyze the kinetics of all the three isotypes of antibodies when multiple serum    were significantly older (median age of 62· 5) than those patients with moderate (median age of 46) and 287 mild symptoms (median age of 30). Remarkably, we found that IgA concentrations in severe cases were 288 significantly higher than mild or moderate cases (figure 3A). IgG levels in moderate and severe COVID-289 19 patients were also higher than mild cases (p < 0· 0001) ( figure 3C). The observation that serum IgG 290 levels were higher in severe and moderate than mild COVID-19 patients have been previously reported 291 . CC-BY-NC-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)
The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint 10, 21 . We also provided here a novel observation that serum IgA levels correlate with COVID-19 severity 292 (figure 3A), how the levels and roles of different types of antibodies as related to COVID-19 severity 293 remain to be determined.

295
Discussion 296 Compared to sampling of nasopharyngeal or throat swabs, blood extraction is more convenient and 297 reliable. Furthermore, serum antibody test is more convenient, fast and accurate, and with other 298 advantages over the detection of viral RNA. 9, 17 We report here an improved serological kit that can 299 sensitively and quantitatively detect serum levels of IgA as well as IgM and IgG. Together with recent 300 reports by others 9-12, 17-20, 23, 24 , the serological data that we obtained from 216 serum samples of 87

306
The nucleocapsid protein (NP) is the most abundant protein in coronaviruses, which was reported to 307 be highly immunogenic and often used as a diagnostic marker for coronaviruses such as SARS-CoV 31 .

308
The RBD of the spike protein on viral surface is the ligand binding to the major host receptor ACE2; 309 therefore RBD could be a main target for neutralization antibodies. 32,33 In this study, we explored the 310 possibility of using either NP or RBD as an immobilized antigen in for developing a clinical COVID-19 311 diagnostic kit. Our data (supplementary figure 2) showed RBD-based diagnostic kits were better 312 performed than that of NP in detecting all the three types of antibodies. A few previous studies reported 313 that RBD-based IgM and IgG detection is better than NP once a comparison was made 19, 23 , and the 314 measurement is agreeable with the titers measured by virus neutralization assays 23, 24 . We provided here 315 the evidence that RBD as an immobilized antigen is also better than NP in detection serum IgA from 316 COVID-19 patients. The exact mechanisms of difference between the use of two types of viral antigens 317 remain to be resolved. It could be that the NP as a highly basic protein interacts with acidic residues in 318 complementarity determining region in antibodies is less specific. It could also be due to the fact that the 319 NP antigen is expressed in bacteria as most investigators do, and the RBD protein we used is expressed 320 in a human cell line enabling critical glycosylation and high-affinity binding to antibodies raised in 321 COVID-19 patients. Nonetheless, we showed that our serological kits based on SARS-CoV-2 spike 322 protein RBD as an immobilized antigen provide a high sensitivity and specificity for detecting IgA, IgM, 323 and IgG in a quantitative manner.

325
Our serological kits have overall good performance 326 Our kits have much higher accuracy than RT-qPCR (sensitivity less than 70%) for detecting viral 327 RNA 9-13 , and published immune-assays such as " flow immunoassay" and ELISA in earlier studies 10-12, 328 17-21, 23, 24 . When we combined RBD-specific IgA and IgG kits together, the sensitivity, specificity and 329 overall agreement elevate to 99· 1%, 100%, and 99· 7%, respectively (table 1). In addition, this RBD-330 based detection kit may also help to screen and detect neutralization antibodies targeting SARS-CoV-2 331 RBD, because this peptide domain is exposed on viral surface and functions as a ligand binding to the 332 host cell surface receptor ACE2.

333
. CC-BY-NC-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)
The copyright holder for this preprint this version posted April 22, 2020.

335
Although one serum collected at the 4th day after illness onset was diagnosed as positive by our IgM 336 kit (not IgA or IgG kits in this study), the IgM kit overall showed a lower diagnostic specificity of 92· 3% 337 compared to that of IgG and IgA (figure 1). IgM is known to have relatively lower affinity toward 338 antigens compared with that of IgG or IgA. In addition, IgM often causes false positive signals as we 339 also observed (supplemental table 2), due to its pentameric structure 34 . To the contrary, IgA or IgG 340 antibody does not have this problem. Our RBD-specific IgG kit showed high specificity of 99· 8% (figure 341 1) but relatively low sensitivity of 96· 8%. This is expected, because that most (6/7) false negative cases 342 were samples collected at 4-10 days after illness onset when IgG production is likely very low.

343
Our RBD-based IgA kit showed high sensitivity and specificity of 98· 6% and 98· 1%, except two sera 344 collected at the 4th day after illness onset. All other sera (2 at the 6th day, 3 at the 7th day, 1 at the 8th

367
We observed the presence of high-level of RBD-specific IgA in COVID-19 patients' sera. It is widely 368 believed that mucosal plasma cells are a major production source of IgA, which is rapidly transported

374
. CC-BY-NC-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)
The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint IgA is traditionally recognized to play an anti-inflammatory role and prevent tissue damage at mucosal 375 sites. However, recent reports also demonstrated that serum IgA is involved in the formation of immune 376 complexes to amplify inflammatory responses. 38 Serum IgA induced proinflammatory cytokine 377 production by macrophages, monocytes and Kupffer cells in non-mucosal tissues including liver, skin 378 and peripheral blood. 39 In this study, we observed that IgA was present in COVID-19 patients' serum, 379 and its levels positively correlated with COVID-19 severity ( figure 3A). In our cohort, we also observed 380 that IgG levels were associated with worse clinical outcomes (figure 3C), as previously described 10, 21 .

381
The latter phenomena has been suggestive of possible antibody-dependent enhancement (ADE) of

393
These observations suggest that gut may be an important place for anti-viral response to coronaviruses, 394 and large amounts of secretory IgA could be detected in these mucosal tissues in addition to that in blood.

396
Weakness of this study

397
The current study at the present form has several limitations. We used 216 serum samples from 87 398 confirmed COVID-19 patients in this study, and serum samples were not available every day for each 399 patient. The earliest collected serum is at the 4th day after self-reported illness onset, and the last one

406
There were also few cases of COVID-19 patients whose symptoms remained mild and serum samples 407 were collected during hospitalization. Therefore, this study of the correlation between antibody levels 408 and disease severity needs further verification.

409
In summary, this study reports a novel sensitive and quantitative serological testing kit of detecting 410 IgA as well as IgM and IgG, for the diagnostics of COVID-19. Due to its high specificity and sensitivity, 411 this kit could sensitively and quantitatively measure levels of IgA in blood and other tissues. The

412
serological study also provides valuable information for monitoring and understanding of COVID-19.

414
Acknowledgements 415 . CC-BY-NC-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.

456
. CC-BY-NC-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)
The copyright holder for this preprint this version posted April 22, 2020.               . CC-BY-NC-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)
The copyright holder for this preprint this version posted April 22, 2020. 544 . CC-BY-NC-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) The copyright holder for this preprint this version posted April 22, 2020.

566
. CC-BY-NC-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) The copyright holder for this preprint this version posted April 22, 2020. . https://doi.org/10.1101/2020.04.17.20064907 doi: medRxiv preprint