key: cord-0748882-m7hsn64z
authors: McGuire, Bailey E.; Mela, Julia E.; Thompson, Vanessa C.; Cucksey, Logan R; Stevens, Claire E.; McWhinnie, Ralph L.; Winkler, Dirk F. H.; Pelech, Steven; Nano, Francis E.
title: Escherichia coli recombinant expression of SARS-CoV-2 protein fragments
date: 2021-06-23
journal: bioRxiv
DOI: 10.1101/2021.06.22.449540
sha: 1c7c7a166cd2bbd986dcbaa546ef478f5aa8f3cf
doc_id: 748882
cord_uid: m7hsn64z

We have developed a method for the inexpensive, high-level expression of antigenic protein fragments of SARS-CoV-2 proteins in Escherichia coli. Our approach used the thermophilic family 9 carbohydrate-binding module (CBM9) as an N-terminal carrier protein and affinity tag. The CBM9 module was joined to SARS-CoV-2 protein fragments via a flexible proline-threonine linker, which proved to be resistant to E. coli proteases. Two CBM9-spike protein fragment fusion proteins and one CBM9-nucleocapsid fragment fusion protein largely resisted protease degradation, while most of the CBM9 fusion proteins were degraded at some site in the SARS-CoV-2 protein fragment. All fusion proteins were expressed in E. coli at about 0.1 g/L, and could be purified with a single affinity binding step using inexpensive cellulose powder. Three purified CBM9-SARS-CoV-2 fusion proteins were tested and found to bind antibody directed to the appropriate SARS-CoV-2 antigenic region. The largest intact CBM9 fusion protein incorporates spike protein amino acids 540-588, which is a conserved region immediately C-terminal to the receptor binding domain that is widely recognized by human convalescent sera and contains a putative protective epitope.

We have developed a method for the inexpensive, high-level expression of antigenic protein fragments 25 of SARS-CoV-2 proteins in Escherichia coli. Our approach used the thermophilic family 9 carbohydrate-26 binding module (CBM9) as an N-terminal carrier protein and affinity tag. The CBM9 module was joined to 27 SARS-CoV-2 protein fragments via a flexible proline-threonine linker, which proved to be resistant to E. coli 28 proteases. Two CBM9-spike protein fragment fusion proteins and one CBM9-nucleocapsid fragment fusion 29 protein largely resisted protease degradation, while most of the CBM9 fusion proteins were degraded at some 30 site in the SARS-CoV-2 protein fragment. All fusion proteins were expressed in E. coli at about 0.1 g/L, and 31 could be purified with a single affinity binding step using inexpensive cellulose powder. Three purified CBM9-32 SARS-CoV-2 fusion proteins were tested and found to bind antibody directed to the appropriate SARS-CoV-2 33 antigenic region. The largest intact CBM9 fusion protein incorporates spike protein amino acids 540-588, which 34 is a conserved region immediately C-terminal to the receptor binding domain that is widely recognized by 35 human convalescent sera and contains a putative protective epitope. 36 37 INTRODUCTION 38 39 One of the public health surveillance tools needed to respond to the coronavirus disease 2019 pandemic is the ability to detect seroconversion to antigens of the SARS-CoV-2 virus, the causative agent 41 of COVID-19. The ability to detect antibodies that are specific to SARS-CoV-2 allows an assessment of the 42 level of probable immunity to COVID-19 in a population. At the individual level, the ability to detect anti-43 SARS-CoV-2 antibodies can help one assess their personal level of vulnerability to COVID-19 due to immunity 44 generated by either vaccination or from an indeterminate or asymptomatic infection. 45

The human antibody response to SARS-CoV-2 infection can include the development of antibodies 47 reactive with any of the 29 proteins encoded in the viral genome (1), including 16 non-structural proteins 48 (NSP's) encoded by the ORF1a/b gene. However most studies have focused on studying the antibody response 49 to the abundant spike and nucleocapsid proteins (2-11). Among the SARS-CoV-2 proteins, the spike protein 50 varies the most between coronaviruses (12); using it allows for the greatest specificity in an antibody assay as 51 well as the ability to differentiate reaction with SARS-CoV-2 from other common coronaviruses. As well some 52 antibodies reactive with the spike protein, especially those that react within or near its receptor binding domain 53 (RBD), are neutralizing (9, 11, (13) (14) (15) . Thus, detection of anti-spike protein antibodies may indicate a level of 54 immunity to 56 Some studies of anti-SARS-CoV-2 antibody response examine antibody reactivity with linear epitopes 57 using synthetic peptides that correspond to the primary structure of viral proteins (2-10). While this type of 58 assay misses many antibody responses against conformational and topographically assembled epitopes, it is the 59 most practical, both technically and economically. Although synthetic peptides are far less expensive than full-60 length recombinant SARS-CoV-2 spike protein, their production cost can still present barriers in resource-poor 61 health systems or when large quantities are needed. 62 6 amine, 1% yeast extract, 25 mM Na2HPO4, 25 mM KH2PO4, 50 mM NH4Cl, 5 mM Na2SO4, 2 mM MgSO4, 100 113 µM FeCl3) (26); the media was supplemented with carbenicillin (250 µg/mL) and chloramphenicol (10 µg/mL). 114

These cultures were grown at 24 °C for 16 h. The A600 was determined and the culture was added to 100 mL 115 fresh broth with antibiotics in a 1 L flask to give an A600 of 2. IPTG was added to a final concentration of 2 mM, 116 and the culture was incubated with shaking (250 RPM) for 3 h at 30 °C. The cultures were cooled on ice for 15 117 min, and then subjected to centrifugation for 10 min at 16,000 x g. The cell pellet from each 50 mL of culture 118 was suspended in 5 mL of solution B (500 mM NaCl, 10 mM MgCl2, 0.5% CHAPS, 50 mM potassium 119 phosphate, pH 7.0, 100 μg/mL lysozyme). 3 g of glass beads (≤106 µm, Sigma), and 5 µL Benzonase® nuclease 120 (Sigma) were added and the cell suspension was vortexed vigorously for two 1 min intervals, and then subjected 121 to centrifugation for 10 min at 16,0000 x g. The supernatant was added to 0.5 g of cellulose powder (Sigma, cat 122 # 435236) that had previously been equilibrated with solution A (500 mM NaCl, 50 mM potassium phosphate, 123 pH 7.0) in a 15 mL conical centrifuge tube. The tube was rocked for 1 h at room temperature, and the cellulose 124 powder was pelleted by centrifugation for 2 min at 4,000 x g. The cellulose resin was washed three times by 125 adding 12 mL of solution A, rocking for 15 min, and separation of the cellulose from the washing solution. This 126 was followed by three washes using solution C (150 mM NaCl, 50 mM potassium phosphate, pH 7.0). 127

Following the last wash with solution C, the cellulose was suspended in elution buffer (1 M glucose, 15 mM 128 NaCl, 10 mM Tris, pH 7.6), and rocked for 15 min. Centrifugation was applied to pellet the cellulose, and the 129 supernatant was removed and added to a Vivaspin® 6, 10 kDa MW cut off (Sartorius) protein concentrator to 130 change the buffer to 150 mM NaCl, 10 mM Tris, pH 7.6. 131

For SDS-PAGE analysis of the CBM9 fusion protein-expressing cultures, 1 mL of cultures were 132 centrifuged and the cell pellet suspended in 150 µL of Bug Buster (EMD Millipore), supplemented with 133 lysozyme and Benzonase® nuclease as described above. Cell extracts were analyzed by SDS-PAGE using 134 MOPS-SDS running buffer and TruPAGE™ gels (Sigma). The blood samples were collected from persons that had been confirmed as positive using a PCR genetic 161 test (designated as "COVID"), those that showed symptoms similar to COVID-19 but were not tested 162 (designated as "sick"), those that were healthy and asymptomatic (designated as "Control") and those that were 163 healthy and donated prior to April 2019 ("pre-COVID"). For recovery of the serum, the collected blood was 164 allowed to clot at room temperature. After about 2 h, possible clots sticking on the walls of the tubes were 165 released by scraping with pipette tips around the inside walls of the collection tubes before the samples were 166 centrifuged at 2000 x g for at least 15 min. When larger blood volumes (up to 10 mL) were processed, the top 167 layer containing the serum was transferred into a syringe and sterile filtered into storage tubes. Sera from small 168 blood samples (less than 500 µL) were transferred directly into the storage tubes without filtration. The 169 recovered serum was stored at -80 ºC. 170

All of the preparations of recombinant proteins were robotically spotted as 0.24 µL of a ~ 8 µM final 171 concentration (except GST-NSP2 SARS CoV2 [DU 66414] in Spot B1 was printed at a 6.5 µM concentration) 172

on to nitrocellulose membranes. The blots were washed three times with TBS (aqueous solution of 20 mM Tris-173 base and 250 mM NaCl; pH 7.5). These dot blot arrays were blocked with 2.5% BSA in T-TBS for 30 min. 174

After washing the membranes twice with T-TBS (TBS with 0.05% Tween 20), the arrays were incubated with 175 either affinity-purified rabbit polyclonal antibodies against SARS-CoV-2 protein sequences at 2 µg/mL or 176 serum from recovered COVID-19 patients and healthy controls at 1:200 dilution in T-TBS. The incubation was 177 carried out at 4 °C overnight. To detect the bound antibodies, the arrays were washed with T-TBS three times 178 followed by the incubation with goat anti-human IgG+IgA+IgM pAb or donkey-anti-rabbit IgG (HRP 179 conjugates, 1:20,000 dilution; Jackson ImmunoResearch, West Grove, Pennsylvania, USA). After 30 min 180 incubation, the arrays were washed six times with T-TBS, once with 0.125 M NaCl and rinsed twice with water. 181

The bound secondary antibody was visualized by enhanced chemiluminescence (ECL) on a Bio-Rad FluorS-182

Max scanner. The ECL scan was performed at a scanning time of 300 s. Eight images of the scanned array were 183 generated during that scanning time. of the viral proteins that elicit an antibody response are known and these specific fragments are used in the 198 serological assay. The goal of our research was to test if SARS-CoV-2 protein fragments known to elicit a 199 human antibody response could be produced inexpensively using a universally available microbial expression 200 system. The SARS-CoV-2 spike protein has been recombinantly produced in a number of hosts, including 201 engineered HEK-293 cells, insect cells, and insect larvae (17, 27, 28) . However, these expression systems 202 require growth media and bioreactors that are orders of magnitude more expensive than the material and 203 equipment used for microbial expression systems. Therefore, we felt that there was value in developing an 204 inexpensive microbial expression system that can produce SARS-CoV-2 protein fragments in substantial 205 amounts. 206

Our strategy was to use a standard E. coli T7 RNA polymerase approach to drive high levels of mRNA 207 transcription, and use a proven carrier protein module, a thermophilic family 9 carbohydrate-binding module 208 (CBM9), to carry the SARS-CoV-2 fragment at the C-terminus of the fusion protein ( Fig. 1 A-C) . Kavoosi et al. 209 (21) showed that the CBM9 module expresses at high levels, even when a protein was fused to the C-terminus. 210

In a separate work (22), these same researchers showed that linking CBM9 to a protein with a proline-threonine 211 rich linker ([PT]4P-IEGR) resulted in a fusion protein that was resistant to protease attack by endogenous E. coli 212

proteases (22). Thus, we adopted the use of (PT)4P as a linker between the CBM9 and SARS-CoV-2 spike 213 protein fragments, and an example of the gene organization is shown in Supplemental Fig. S1 . 214

The bulk of the studies on the antibody response to SARS-CoV-2 have been conducted using 215 overlapping synthetic peptides corresponding to SARS-CoV-2 proteins, primarily the spike protein, but 216 sometimes several proteins or the entire proteome. At the time that we initiated our studies, the work of Zhang 217 et al. (10) was one of the more comprehensive analyses of the human antibody response in COVID-19. These 218 researchers identified nine regions in the spike protein, designated ID-A through ID-I (Fig. 1D) , which were 219 recognized by antibodies from COVID-19 convalescent sera, and we chose these regions as well as one 220

nucleocapsid protein epitope to clone and express. An SDS-PAGE analysis of the expressed CBM9-spike 221 protein fusions (Supplemental Fig. S2 ) indicated which fusion proteins were most resistant to degradation by E. 222

coli and we chose a sub-group of the clones to study further (abandoning ID-H, ID-H2, ID-H3 and ID-I). Table  223 1 lists the amino acid sequences of the encoded spike (and one nucleocapsid ["N"]) protein regions in the clones 224

we constructed and chose to study further, and the specific epitope regions of the spike protein identified by 225

Zhang and co-workers (10) are underlined. 226 227

Purification and mass spectroscopy analysis of CBM9 fusion proteins.

From the recombinant CBM9 fusion clones that we chose, we expressed and isolated the recombinant 229 protein using powdered cellulose in a batch purification. The resulting purified proteins were subjected to mass 230 spectroscopy analysis to determine the molecular weight of the dominant purified product. The results (Table 1)  231 indicated, as expected, that all products had the N-terminal methionine removed. Most cloned products were 232 processed, presumably by endogenous E. coli proteases, so that some portion of the C-terminal end was 233 removed, effectively removing a few to several amino acids of the spike protein fragment. However, clones 234 expressing CBM9-(PT)4P, CBM9-ID-F, CBM9-H1 and CBM9-N produced, as the dominant purified product, 235 proteins that were 1 Dalton smaller than the predicted monoisotopic product. Since the CBM9-(PT)4P dominant 236 product was 1 Dalton smaller than predicted, we interpreted this to mean that an unidentified chemical 237 modification occurs, many of which are documented (29), on the CBM9 module to remove one atomic mass 238 unit. The clone expressing CBM9-ID-A was processed so as to remove only two amino acids from the B-cell 239 epitope identified by Zhang et al. (10) . 240

For further work we chose E. coli clones that expressed intact fusion proteins of CBM9 and fragments of 241 SARS-CoV-2 proteins as the dominant recombinant products. The expression of clones CBM9-(PT)4P, CBM9-242 ID-H1 and CBM9-N are shown in Fig. 2A and the purified products in Fig. 2B ; for comparison, the carrier 243 protein module CBM9 is also shown. By comparing the staining intensity of the protein band in the cell extracts 244

to the band of purified CBM9-N, we estimated that the clones expressed recombinant product at levels of about 245 100 mg/L upon IPTG-induction, and this is consistent with the 200 mg/L estimates of Kavoosi et al. (21) . These 246 experiments were performed using standard research growth flasks at an A600 of less than 10, and presumably 247 the levels of recombinant protein that are produced could be significantly increased using an optimized fed-248 batch bioreactor protocol. 249

The recombinant fusion proteins were isolated by binding to cellulose powder, using batch purification 250 (Fig. 2B ). Following this, the CBM9 and the CBM9-ID-F samples were heated to 70 °C for 10 minutes and the 251 CBM9-(PT)4P, CBM9-ID-H1 and CBM9-N samples were filtered sterilized, all before storing at 4 °C for at 252 least two weeks (Fig. 2B ). As well, samples were stored at -20 °C in 50% glycerol ( Fig. 2A last lane on right, 253 CBM9-N). All storage conditions preserved the integrity of the sample. However, heating to 70 °C seemed to 254 generate small amounts of multimers of the protein, consistent with previously reported observations for 255 hyperthermophilic enzymes (30). 256 We constructed a number of clones of the ID-H region (see Fig. S2C ), and it was fortuitous that the 257 CBM9-ID-H1 clone highly expressed a product that was largely resistant to E. coli proteases. The ID-H region 258

(residues 522-646) partially overlaps with the RBD (residues 319-541) of the SARS-CoV-2 spike protein. In the 259 3D-structure oriented with the RBDs at the top, the ID-H1 region (residues 540-588) slightly overlaps with and 260 lies below the RBD (Fig. 3A) . It is possible that the CBM9-H1 recombinant product is resistant to proteases, 261 while shorter than other CBM9 fusions that are susceptible, because the ID-H1 clone encodes a potential self-262 folding protein domain (Fig. 3B ). The region encompassed by CBM9-ID-H1 includes amino acid sequences 263 identified by several groups as B-cell epitopes, as defined by synthetic peptides that are recognized by 264 convalescent sera from COVID-19 patients (Fig. 3C ). 265

Interestingly, the region encompassed by the H1 clone does not encompass any of the amino acid 266 changes found among any of the variants of concern (12), but it does contain a putative protective epitope. Poh 267 et al. (9) found that titer of antibody in sera from COVID-19 convalescent patients that reacted with a peptide 268 corresponding to amino acids 562-579 of the spike protein correlated with the amount of in vitro pseudovirus 269 neutralization. Further, when the neutralizing sera was depleted of reactivity against the peptide representing 270 amino acids 562-579 the neutralization activity fell sharply. Such evidence indicates that a strategy to elicit 271 antibodies against this region may be an effective way of protecting against variants with amino acid changes in 272 the RBD of the SARS-CoV-2 spike protein.

CBM9-SARS-CoV-2 fusion proteins react with rabbit anti-spike protein sera and human sera.

As stated above, we used proline-threonine flexible linker regions to join the CBM9 module to SARS-276

CoV-2 spike protein and nucleocapsid protein regions. In this research the purpose of the linker was to allow 277 the SARS-CoV-2 protein fragment to be accessible to antibody binding. To determine if the linker 278

accomplished this, we reacted purified CBM9-(PT)4P, CBM9-ID-F, CBM9-ID-H1 and CBM9-N with purified 279 rabbit antibodies that had been raised to different portions of SARS-CoV-2 proteins (Fig. 4) and human sera 280 ( Fig. 5 and Fig. 6 ). In a semi-quantitative dot blot assay, we found that rabbit antibodies raised against the 281 appropriate fragments of the SARS-CoV-2 spike protein reacted strongly with CBM9-H1 (Fig. 4F, 4G ), but 282 only weakly, or not at all, with antibodies directed against other regions of the spike protein. Reaction of the 283 appropriate antibodies with CBM9-ID-F was moderately strong (Fig. 4C ), and likewise poor or not detectable 284 with the other antibodies. We had access to a small sampling of sera from COVID-19 confirmed (n=7), 285 13 COVID-19 suspected (n=13), and healthy individuals (n=20) (Fig. 5 and Fig. 6) . While this small sample set 286 size and the dot blot assay cannot provide an epidemiological story, the results did show that human sera clearly 287 reacted with the CBM9 fusions carrying ID-F and the nucleocapsid epitope. Many sera samples from both sick 288 and healthy individuals reacted strongly with ID-F, indicating that this region may be similar in other 289 coronaviruses or may be similar to another commonly encountered antigen. Surprisingly, the sera from several 290 individuals, both healthy and ill, reacted apparently more strongly with CBM9-ID-F (spike amino acids 450-291 469) than with the MBP-RBD fusion protein (spike amino acids 319-541), even though the latter encompasses 292 the ID-F region. These results may reflect a property of the antigen, such as accessibility of the SARS-CoV-2 293 portion to the antibody; or it may be that the ID-F region is an especially immunogenic region of the RBD. 294

Overall, with this small sample of sera there was little difference between the patterns of reactivity of the sera 295 from the sick and healthy groups. This high detection of anti-spike and anti-nucleocapsid immunoreactivity in 296 serum samples from healthy individuals is consistent with previous studies using two different serological tests 297 developed by Mesoscale Devices and Kinexus (31) . 298

In this work, we have demonstrated that fragments of the SARS-CoV-2 virus, fused to the CBM9 299 module through a flexible linker, could be produced at high levels -about 0.1 g/L -using universally available 300 equipment with inexpensive materials. The costs of cultivating E. coli are 10-to 100-fold less expensive than 301 the costs of growing non-microbial eukaryotic cells, which are the usual hosts for expressing SARS-CoV-2 302

antigens. Further, the cost of using cellulose powder for affinity purification of CBM9 fusion proteins is about 303 100-to 1,000-fold less than using the conventional immobilized nickel resin or a combination of traditional 304

protein purification columns, such as ion-exchange with size exclusion resins. Lastly, while we described the 305 production and isolation of CBM9 fusion proteins, there may be applications that require the separation of the 306 CBM9 module from the SARS-CoV-2 fragment. Often the components of a fusion protein are separated using a 307 highly specific protease cleavage site. Indeed, using this approach Kavoosi et al. (21) found that a CBM9-GFP 308 fusion protein remained largely intact even in the absence of protease inhibitors, unless cleaved with factor Xa 309 whose cleavage site was incorporated into the linker. While specific proteases may be found that function with a 310 14 specific CBM9 fusion, with or without addition of the cleavage site into the linker, it is also possible to use 311 chemical treatments that cleave with an acceptable level of specificity. For example, in the case of recombinant 312 product CBM9-ID-H1, 45 amino acids of the 49 amino acid SARS-CoV-2 fragment could be released by 313 treatment with hydroxylamine which attacks asparaginyl-glycine bonds (32) as CBM9 lacks an arginine-glycine 314 sequence. Thus, the use of CBM9-SARS-CoV-2 protein fragment fusions allows for the economical production 315 of antigens to be used for a variety of purposes, including in COVID-19 serological assays. 316 317 ACKNOWLEDGEMENTS 318 319

We thank Jeff Wang for aid in the drawing of blood samples from the study participants. Our thanks go also to 320 Akshra Atrey for the probing of the dot blot arrays. 321

Funding. This work was supported by grants from Natural Sciences and Engineering Research Council of 322

Canada (ALLRP 553524 -20 and RGPIN-2018-03747). 323

Conflict of interest statement. SP is the majority shareholder of Kinexus Bioinformatics Corporation. All other 324 authors declare no conflicts. 325 326 Table 1 . Amino acid sequence of B-cell epitopes cloned as fusions to CBM9, 462 and location of putative protease cleavage sites. 463 464

Epitope (amino acid numbers)

Amino acid sequence and calculated MW of processed protein*, and observed MW of CBM9 fusion protein

The proteins of severe acute respiratory syndrome coronavirus-2 (SARS CoV-2 or 329 n-COV19), the cause of COVID-19

ReScan, a multiplex diagnostic pipeline, 333 pans human sera for SARS-CoV-2 antigens

Linear B-cell epitopes in the spike and 338 nucleocapsid proteins as markers of SARS-CoV-2 exposure and disease severity

Peptide 342 microarray-based analysis of antibody responses to SARS-CoV-2 identifies unique epitopes with 343 potential for diagnostic test development

346 Systematic profiling of SARS-CoV-2-specific IgG epitopes at amino acid resolution

Immunoreactive peptide maps of SARS-CoV-2

Viral epitope profiling of COVID-19 patients reveals cross-reactivity and correlates of 356 severity

The landscape of antibody binding in SARS-358

CoV-2 infection

Two linear epitopes on the SARS-CoV-2 362 spike protein that elicit neutralising antibodies in COVID-19 patients

Mining of epitopes on spike protein of SARS-CoV-2 from 366 COVID-19 patients

CoV-2 spike protein elicit neutralizing antibodies in COVID-19 patients

373 SARS-CoV-2 variants, spike mutations and immune escape

A vaccine targeting the RBD 382 of the S protein of SARS-CoV-2 induces protective immunity

CoV-2 challenge via multiple mechanisms

Castelain 393 S, Helle F. 2020. Anti-spike, anti-nucleocapsid and neutralizing antibodies in SARS-CoV-2 inpatients 394 and asymptomatic individuals

SARS-CoV-2 seroconversion in humans: A detailed protocol for a serological assay, antigen production, 398 and test setup

Efficient production of recombinant SARS-401

CoV-2 spike protein using the baculovirus-silkworm system

A plasmid DNA-launched SARS-CoV-2 reverse genetics system 413 and coronavirus toolkit for COVID-19 research

Crystal structures of the family 9 416 carbohydrate-binding module from Thermotoga maritima xylanase 10a in native and ligand-bound forms

Inexpensive one-step 422 purification of polypeptides expressed in Escherichia coli as fusions with the family 9 carbohydrate-423 binding module of xylanase 10A from T. maritima

Strategy for selecting and characterizing 426 linker peptides for CBM9-tagged fusion proteins expressed in Escherichia coli

The pRSET family of T7 promoter expression vectors for Escherichia coli

JCat: A novel tool to 431 adapt codon usage of a target gene to its potential expression host

A one pot, one step, precision cloning method with high 434 throughput capability

Characterization of spike glycoprotein of SARS-CoV-2 on virus 439 entry and its immune cross-reactivity with SARS-CoV

Structure 442 of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor

Protein termini and their modifications revealed by positional 445 proteomics

Hyperthermophilic enzymes: sources, uses, and molecular mechanisms for 447 thermostability

A majority of 451 uninfected adults show preexisting antibody reactivity against SARS-CoV-2

Cleavage at Asn-Gly bonds with hydroxylamine

Structure, Function, and 456 Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Schrödinger L. 2020. The PyMOL Molecular Graphics System

CBM9-(TP)4P Not applicable

 Calculated: 22945.49 Observed: 22944.51*The calculated molecular weight was based on sequences assuming that the amino terminal methionine was 465removed. The C-terminal amino acid was chosen to best match the molecular weight of the dominant form 466 21 found in the MS experiments. Underlined sequence corresponds to synthetic peptide found to bind antibody as 467 determined by Zhang, et al. (10) . The highlighted region with the slash indicates the putative protease cleavage  468 site, by endogenous E. coli proteases. in an inverse PCR reaction using pRSET5A::CBM9-id-a to exchange the epitopes fused to CBM9. D. 477Representation of linear ID-A through ID-I regions with the amino acid numbers of the SARS-CoV-2 spike 478 protein recognized by antibody from COVID-19 convalescent sera, as described by Zhang et al. (10) ; RBD is 479 the receptor binding domain. In Panels A-C the bent arrow indicates T7 promoter region; half circle indicates 480 T7 ribosome binding site; "T" symbol represents transcriptional terminator. correspond to healthy individuals whose serum samples were collected in 2020. Panels N-U, "pre-COVID" 504 samples were from healthy individuals whose serum samples were retrieved prior to April 2019. Individuals are 505 27 also identified by sex (M for male and F for female) followed by age in years. The MBP-spike RBD protein 506 (spot position A1) includes amino acids 319-541 of the spike protein. 507 508 28 509 Figure 6 . Dot blot of proteins with sera from COVID-19 and sick individuals. Panels B-H, "COVID" samples 510 correspond to individuals that PCR-tested positive for SARS-CoV-2 RNA and whose serum samples were 511 collected in 2020. Panels I-U, "Sick" samples were from individuals who had COVID-19 symptoms and whose 512 serum samples were collected in 2020. Individuals are also identified by sex (M for male and F for female) 513 followed by age in years. The MBP-spike RBD protein (spot position A1) includes amino acids 319-541 of the 514 spike protein. 515