key: cord-0738962-t9jo76ja
authors: Hachim, A.; Kavian, N.; Cohen, C. A.; Chin, A. W.; Chu, D. K.; Mok, C. K. P.; Tsang, O. T.; Yeung, Y. C.; Perera, R. A.; Poon, L. L.; Peiris, M. J.; Valkenburg, S. A.
title: Beyond the Spike: identification of viral targets of the antibody response to SARS-CoV-2 in COVID-19 patients
date: 2020-05-02
journal: nan
DOI: 10.1101/2020.04.30.20085670
sha: 795d9fb74e8b97ee9c3b97a85dc88cf385036fff
doc_id: 738962
cord_uid: t9jo76ja

Background: The SARS-CoV-2 virus emerged in December 2019 and caused a pandemic associated with a spectrum of COVID-19 disease ranging from asymptomatic to lethal infection. Serology testing is important for diagnosis of infection, determining infection attack rates and immunity in the population. It also informs vaccine development. Although several serology tests are in use, improving their specificity and sensitivity for early diagnosis on the one hand and for detecting past infection for population-based studies, are priorities. Methods: We evaluated the anti-SARS-CoV-2 antibody profiles to 15 SARS-CoV-2 antigens by cloning and expressing 15 open reading frames (ORFs) in mammalian cells and screened antibody responses to them in COVID-19 patients using the Luciferase Immunoprecipitation System (LIPS). Results: The LIPS technique allowed us to detect antibody responses in COVID-19 patients to 11 of the 15 SARS-CoV-2 antigens tested, identifying novel immunogenic targets. This technique shows that antigens ORF3b and ORF8 allow detection of antibody early in infection in a specific manner and reveals the immuno-dominance of the N antigen in COVID-19 patients. Conclusion: Our report provides an unbiased characterization of antibody responses to a range of SARS-CoV-2 antigens. The combination of 3 SARS-CoV-2 antibody LIPS assays, i.e. N, ORF3b, and ORF8, is sufficient to identify all COVID-19 patients of our cohort even at early time-points of illness, whilst Spike alone fails to do so. Furthermore, our study highlights the importance of investigating new immunogens NSP1, ORF3b, ORF7a and ORF8 which may mediate immune functions other than neutralization which may be beneficial or harmful to the patient.

The acute pandemic respiratory disease COVID-19 is caused by a novel coronavirus 51 that belongs to the species Severe acute respiratory syndrome-related coronavirus 52 (SARS-CoV-2) (1, 2). There are at least 70 vaccine candidates against COVID-19 in 53 various stages of testing, development and human clinical trials (5). Urgency in 54 vaccine development and deployment to mitigate the pandemic has left some 55 fundamental immunological research questions outstanding, especially regarding the 56 range of virus immunogens that elicit antibody responses, their kinetics, specificity, 57 breadth, longevity and impact for long-term protection or immune-pathology. Nucleocapsid (N). The trimer S protein is cleaved into S1, containing the receptor 70 binding domain, and S2 subunits (8, 9) , and S2 is further cleaved into S2' to form the 71 viral fusion peptide (6). The S protein is critical for viral entry and is a neutralising 72 target, as it is with SARS-CoV and MERS-CoV and is a key target for diagnostic tests 73 and vaccine development (10). The S1 subunit of SARS-CoV-2 shares about 70% 74 . 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.

The copyright holder for this preprint this version posted May 2, 2020. . https://doi.org/10.1101/2020.04.30.20085670 doi: medRxiv preprint 7 by LIPS, whilst there was no difference observed in LIPS responses to subunits S1, 135 S2 and S2' in high or low MN COVID-19 patients (Figure 2c , p=0.0049). 136

We next investigated the presence of antibodies specific to previously 137 uncharacterized ORFS to the SARS-CoV-2. As we could not produce the full 138

ORF1ab due to its extended length (>21,000 bp, (1)), we cloned and expressed a 139 representative antigen, non-structural protein 1 (NSP1). Bioinformatic predictions 140 have revealed a putative role for NSP1 in suppressing the antiviral host response (8). 141

We used LIPS to detect antibodies specific to NSP1, ORF3a, ORF3b, ORF6, 142

ORF7a, ORF7b, ORF8 and ORF10 ( Figure 3 ). COVID-19 patients had significantly 143 higher NSP1-antibody levels compared to negative controls (mean of 5301+/-854.7 144 LU versus 3683+/-726.4 LU, p<0.0001, Figure 3a ). Furthermore, significantly higher 145 antibody levels in COVID-19 patients were detected towards ORF3a, ORF3b, 146

ORF7a, ORF7b and ORF8 (p=0.0302, p<0.0001, p<0.0001, p=0.0019, p<0.0001, 147 Globally, among the antibody responses tested in our LIPS assay, we detected 153 significant levels of antibodies specific to 11 antigens: N, M, S, S1, S2', NSP1, 154

ORF3a, ORF3b, ORF7a, ORF7b and ORF6 in the COVID-19 population (Table 1) point detection). We did not detect a significant production of antibodies to 4 of the 157 SARS-CoV-2 antigens tested: E, S2, ORF6 and ORF10 proteins. 158 . 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 May 2, 2020. .

Comparison of the global SARS-CoV-2 antibody responses from 15 COVID-19 159 patients reveals that anti-N antibodies dominate the humoral response detected by 160 LIPS (Figure 3i ), whilst other antigens make lower and similar contributions to the 161 magnitude anti-SARS-Cov-2 antibody responses (Figure 5a) . 162 163 Absence of sex-and age-effect in our cohort 164

Epidemiological data from many countries suggest that men develop more severe 165 COVID-19 related symptoms than women (21). We therefore evaluated the effect of 166 gender in our cohort of COVID-19 patients for antibody responses and found no 167

difference across all antigenic targets tested (p=0.6289, Supplementary Figure 2a) . 168

Similarly, the SARS-CoV-2 antibody response was comparable between patients 169 aged below and above 60 years old (p=0.9363, Supplementary Figure 2b) . 170 171 Combined antigen test panels as a potential diagnostic tool for COVID-19 172

We found significantly higher levels of 11 of the 15 antigens tested in the COVID-19 173 populations. Next, their ability to correctly identify COVID-19 patients was 174 determined. Therefore, a cut-off value of LIPS antibody signals, based on the mean 175 plus three standard deviations of the healthy negative control group (22-24), was 176 calculated for each of these 11 antigens, along with the sensitivity and specificity of 177 each test (Table 3 , and Figure 4 ). The cut-off value of all 11 tests showed a high 178 specificity of the SARS-CoV-2 LIPS assays with no samples from the negative 179 control group being above the cut-off. On the other hand, 8 antigens showed a low 180 sensitivity (below 75% for M, S, S1, S2', NSP1, ORF3a, ORF7a, ORF7b, Figure 4b) , 181

hence being insufficient to correctly identify all the COVID-19 patients with high rates 182 of false negatives (Table 3) , limiting their use for LIPS diagnostics. On the other 183 . 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 May 2, 2020. (Table 3) . Of note, ORF8 is the only of the 11 tests to 185 correctly identify all COVID-19 patients including at day 4 after onset of illness 186 ( Figure 4a) . 187

As many LIPS tests showed low sensitivity, we then used an approach based on the 188 combination of LIPS antibody LU signals for the 11 separate SARS-CoV-2 antigen 189 tests to efficiently detect SARS-CoV-2 exposure in this population (22-24). This 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 May 2, 2020. To investigate specificity of the antibody responses to the panel of antigens, we 220 subtracted the mean LU levels of healthy negative controls from the mean LU of 221 COVID-19 infected patients and compared this difference across each antigen 222 ( Figure 5b ). We found that this difference was significantly higher for the N-specific 223 antibody response compared to all the other relevant antibody responses (M, S, S1, 224 S2', NSP1, ORF3a, ORF3b, ORF7a, ORF7b, and ORF8) (p<0.0001, Figure 5b Whilst results for ORF3b were significant against the remaining 8 antigens (M, S, S1, 231 . 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 May 2, 2020. . S2', NSP1, ORF3a, 3b, and 7b), excluding ORF7a ( Figure 5c ). Therefore, ORF3b 232 and ORF8 are newly identified as specific and unique antigenic targets. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 2, 2020. . https://doi.org/10.1101/2020.04.30.20085670 doi: medRxiv preprint results, establishing consistency among assays for the full-S protein. N antibody 257 responses detected were also elevated in patients, but the correlation between LIPS 258 and ELISA N antibody assays was lower. 259

We next cloned all the available ORFs of the virus ORF1ab (as NSP1 only), ORF3a, 260

ORF3b, ORF6, ORF7a, ORF7b, ORF8 and ORF10, to acquire more extensive 261 information of the immunogenic targets of the virus. We found that 6 of these 8 ORFs 262

induced a humoral immune response in the patients (NSP1 (ORF1ab), ORF3a, 263

ORF3b, ORF7a, ORF7b and ORF8). Furthermore, by calculating a cut-off value 264 based on the mean of the negative population + 3 standard deviations, we could 265 ascertain that only 3 out of these 11 antigens were useful in diagnostic tests with high 266 performance: N, ORF3b, and ORF8. By using the combined results of several 267 antibody signals and calculating a cut-off value for these responses, we found that 268 the sum of the 11 relevant antibody tests was highly sensitive and specific. Moreover, 269 combining only the 3 of the most informative and sensitive antigens, N, ORF3b and 270 ORF8, we achieved a sensitivity and specificity of 100%, correctly identifying all the 271 COVID-19 patients versus negative controls. A larger data set from COVID-19 272 patients and negative controls is needed to further refine and confirm the 273 performance of our test (30). 274

The Spike protein is responsible for virus entry into host cells and is the main antigen 275 that elicits neutralizing antibodies (10). Whilst we detected significant differences in 276 the magnitude of responses by LIPS between patients and controls for S, S1 and 277 S2', these antigens did not show high sensitivity levels, especially for sera collected 278 early post disease onset. The Spike protein is a trimer on the surface of virions, and 279 the conformation of our Ruc-S antigen should be assessed by conformation 280 dependent monoclonal antibodies to confirm whether the conformation of S protein in 281 . 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.

The copyright holder for this preprint this version posted May 2, 2020. . https://doi.org/10.1101/2020.04.30.20085670 doi: medRxiv preprint 13 our assay is well retained for antibody binding and affinity (31). Further LIPS tests 282 using the S RBD antigen alone are underway to assess the proportion of RBD-283 specific antibodies as an important target of neutralizing antibodies (31). The S 284 protein also elicits non-neutralizing antibodies targeted to conserved epitopes (11), 285 and among our cohort an absence of in vitro neutralization has been observed for 286 some patients, especially at early time-points. 287

The combination of multiple antigens by LIPS beyond the Spike could be the basis 288

for supplementary serological tests useful to determine SARS-CoV-2 exposure to 289 overcome false negative results. We found that the single test detection of N, S or 290 replicative fitness, this issue may undermine the utility of ORF8 alone in serological 306 . 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.

The copyright holder for this preprint this version posted May 2, 2020. . 14 testing. In addition, NSP1 and ORF7a LIPS tests showed high significance for 307 COVID-19 patients. Though little is known about their function, bioinformatic 308 predictions reveal that both NSP1 and ORF7a could be involved in suppressing the 309 antiviral host response (6, 8). Therefore, the combined use of multiple antigens that 310 are sensitive and specific is needed for diagnostic serology. 311

Most commercially available or published serological tests use only the S antigen, 312 with a few using both the S and N antigens (11, 36, 37). Using extensive testing of 313 the virus antigens, we have shown that additional targets are important for early 314 detection of antibody responses and identification of COVID-19 patients. Potential 315 cross-reactivity of SARS-CoV-2 antibodies with other coronaviruses antibodies 316 cannot be excluded, but several recent data report that this cross-reactivity is minimal 317 (26). Therefore, the approach of combining several relevant antigens that are unique 318 to SARS-CoV-2, and immunogenic boosting specific antibodies would overcome 319 issues of cross-reactivity and increase the sensitivity of serological assays. 320

Screening of additional negative pre-pandemic samples with our LIPS approach is 321 needed to refine issues of cross-reactivity with other HCoV. The high orders of 322 magnitude and range of antibody quantity measured by our SARS-CoV-2 LIPS assay 323 is an advantage compared to ELISA Optical Density measurements. These 324 advantages may help with the rapid screening of recovered COVID-19 patients for 325 elevated antibodies to donate serum for the treatment of other patients by passive 326 transfer (38). 327

Our study does not report on antibody functions and data on Fc mediated functions 328 of SARS-CoV-2 antibodies are also still lacking (39). Therefore, Fc mediated 329 functions of antibodies directed against internal proteins from ORF3b and ORF8 will 330 be the focus of future studies. Furthermore, only IgG is bound to the protein A/G 331 . 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.

The copyright holder for this preprint this version posted May 2, 2020. greater sample size and asymptomatic cases are necessary to test the cut-off 343 sensitivity scores; and to confirm the diagnostic value of our LIPS assays and the 344 relevancy of ORF3b and ORF8 antibodies. Late-convalescent sera are needed to 345 assess antibody waning. We also need to investigate antibody profiles with disease 346 severity. 347

In conclusion, we found that COVID-19 patients not only produce antibodies to the 348 Spike protein, but also to other structural and non-structural proteins. The antibody responses associated with COVID-19. We still need to investigate whether 355 antigens other than the virus spike confer protection. Such information will help 356 . 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.

The copyright holder for this preprint this version posted May 2, 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.

The copyright holder for this preprint this version posted May 2, 2020. 

Our study enrolled a total of 26 patients with RT-PCR confirmed COVID-19 infection 362 (Table 1) Table 1 , and primers sequences in 375

Supplementary table 2). A RT-PCR was performed using extracted SARS-CoV-2 376 vRNA to amplify target genes corresponding to Structural and Non-structural proteins 377 of the virus (Table 1, (6)) using Platinum SuperScriptIII One Step RT-PCR system. 378

The bands were then extracted using Qiagen gel extraction kit (Qiagen, Germany) 379 and digested with BamHI and NotI or KpnI-HF and XhoI (New England Biolabs, 380 USA). Extracted products were ligated using T4 DNA ligase (New England Biolabs) 381 into the pREN2 plasmid (from Peter Burbelo NDICR, NIH). Plasmids were 382 transformed using DH10B competent cells and purified using PureYield Plasmid 383 midi-prep system (Promega). 384 . 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.

The copyright holder for this preprint this version posted May 2, 2020. .

Constructs with pREN2-Renilla luciferase plasmid containing the SARS-CoV-2 387 antigen of interest were transfected into Cos1 cells using Fugene 6 (Promega) as per Enzyme-linked immunosorbent assay 408 . 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.

The copyright holder for this preprint this version posted May 2, 2020. .

ELISA assays were performed with the available SARS-CoV-2 proteins Spike 409 (S1+S2) and Nucleoprotein (N) proteins. Briefly, recombinant S and N proteins 410 (Sinobiological) were coated on 96-well flat-bottom immunosorbent plates (Nunc 411

Immuno MaxiSorp, Roskilde, Denmark) at a concentration of 100 ng/ml, in 100/μl 412 coating buffer (PBS with 53% Na 2 CO 3 and 42% NaHCO 3 , pH 9.6) at 4°C overnight. 413

An additional plate coated with a non-specific protein (blocking buffer, PBS with 5% 414 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.

The copyright holder for this preprint this version posted May 2, 2020. .

Multiple amino acid alignments of structural proteins from HKU1 (AY597011.2), 434

HCoV-229E (AF304460.1), HCoV-OC43 (AY391777.1) and HCoV-NL63 435 (AY567487.2) were compared versus SARS-CoV-2 (MN908947) using CLUSTAL Antibody levels are presented as the geometric mean +/-standard deviation (stdev). 441

For the calculation of sensitivity and specificity, cut-off limits for each antigen were 442 derived from the mean value plus three standard deviations of the controls. Non-443

parametric Mann-Whitney U tests were used to compare the antibody levels 444 between COVID-19 and negative groups, using the GraphPad 8 Prism software. 445 . 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.

The copyright holder for this preprint this version posted May 2, 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.

The copyright holder for this preprint this version posted May 2, 2020. . https://doi.org/10.1101/2020.04.30.20085670 doi: medRxiv preprint . 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.

The copyright holder for this preprint this version posted May 2, 2020. . cryptic epitope in the receptor-binding domains of SARS-CoV-2 and SARS-CoV. 550

Science. 2020. 551 . 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.

The copyright holder for this preprint this version posted May 2, 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.

The copyright holder for this preprint this version posted May 2, 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.

The copyright holder for this preprint this version posted May 2, 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.

The copyright holder for this preprint this version posted May 2, 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.

The copyright holder for this preprint this version posted May 2, 2020. Luciferase titers (LU) * ns S1 S2 S2' ✱✱ . 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.

The copyright holder for this preprint this version posted May 2, 2020. . Luciferase titers (LU) S1 S2' a b . 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 May 2, 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 May 2, 2020. . https://doi.org/10.1101/2020.04.30.20085670 doi: medRxiv preprint

A new coronavirus 455 associated with human respiratory disease in China

Emergence of a novel human coronavirus threatening 457 human health

SARS-CoV-2 Vaccines: Status Report

Viral dynamics in mild 461 and severe cases of COVID-19. Lancet Infect Dis. 2020. 462 5. World Health Organization. Coronavirus disease (COVID-2019) situation 463 reports 2020

Protein Structure 466 and Sequence Reanalysis of 2019-nCoV Genome Refutes Snakes as Its 467 Intermediate Host and the Unique Similarity between Its Spike Protein Insertions and 468 HIV-1

Breadth of concomitant immune responses prior to patient recovery: a case 471 report of non-severe COVID-19

Genomic 473 characterization of the 2019 novel human-pathogenic coronavirus isolated from a 474 patient with atypical pneumonia after visiting Wuhan

Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Coronavirus Infections-More Than Just the 479 Common Cold

Cross-reactive antibody response between SARS-CoV-2 and SARS-CoV infections

Breadth of concomitant immune responses prior to patient recovery: a case 488 report of non-severe COVID-19

Neutralizing antibody 490 responses to SARS-CoV-2 in a COVID-19 recovered patient cohort and their 491 implications

Antibody Responses 1 Year after Symptom Onset, South Korea

MERS coronavirus in dromedary camel herd, Saudi Arabia

T cell-mediated immune response to 499 respiratory coronaviruses

A simplified immunoprecipitation method 501 for quantitatively measuring antibody responses in clinical sera samples by using 502