key: cord-0427205-6326lwuc
authors: Maurel Ribes, A.; Bessiere, P.; Guery, J.-C.; Joly Featherstone, E.; Bruel, T.; Robinot, R.; Schwartz, O.; Abravanel, F.; Izopet, J.; Joly, E.
title: A simple, sensitive and quantitative FACS-based test for SARS-CoV-2 serology in humans and animals
date: 2021-09-13
journal: nan
DOI: 10.1101/2021.09.06.21262027
sha: f87d638d9e86bc83361050ec35a1f120a64542dd
doc_id: 427205
cord_uid: 6326lwuc

Serological tests are important for understanding the physiopathology and following the evolution of the Covid-19 pandemic. Assays based on flow cytometry (FACS) of tissue culture cells expressing the spike (S) protein of SARS-CoV-2 have repeatedly proven to perform slightly better than the plate-based assays ELISA and CLIA (chemiluminescent immuno-assay), and markedly better than lateral flow immuno-assays (LFIA). Here, we describe an optimized and very simple FACS assay based on staining a mix of two Jurkat cell lines, expressing either high levels of the S protein (Jurkat-S) or the mCherry fluorescent protein (Jurkat-R, which serve as an internal negative control). We show that this Jurkat-S&R-flow test has a much broader dynamic range than a commercial ELISA test and performs at least as well in terms of sensitivity and specificity. Also, it is more sensitive and quantitative than the hemagglutination-based test HAT, which we described recently. The Jurkat-R&S-flow test requires only a few microliters of blood; thus, it can be used to quantify various Ig isotypes in capillary blood collected from a finger prick. It can be used also to evaluate serological responses in mice, cats and dogs. FACS tests offer a very attractive solution for laboratories with access to tissue culture and flow cytometry who want to monitor serological responses in humans or in animals, and how these relate to susceptibility to infection, or re-infection, by the virus, and to protection against Covid-19.

Introduction 42 Over the past year, our world has been thrown into disarray by a pandemic caused by a new 43 coronavirus, SARS-CoV-2. With a mortality rate around 1%, this new virus is not as pathogenic as 44 previous coronaviruses such as SARS-CoV (9.6%) and MERS (35%), but it transmits faster from 45 human-to-human (Fani et al. 2020 ), probably because of a large proportion (ca. 50%) of 46 asymptomatic carriers (Wu et al. 2021 ; Long et al. 2020 ). Consequently, in less than two years 47 since its discovery in Wuhan, China, SARS-CoV-2 has spread all over the world and caused more 48 than 200 millions confirmed cases and over 4 millions confirmed deaths ( 49 https://www.who.int/emergencies/diseases/novel-coronavirus-2019 ). 50

Biotechnology has proven a great asset in combating the pandemic, with the rapid 51 development of diagnostic tests and, more recently, of vaccines. Diagnostic tests detect directly 52 either the viral nucleic acid or viral proteins in nasopharyngeal swabs. Serological tests, by 53 contrast, detect antibodies developed in response to infection by the virus or vaccination, or a 54 combination of the two. Since the presence of antibodies in serum usually correlates with 55 elimination of the virus and the patient's recovery, serological tests have not been very helpful 56 in the clinic. They have, however, proven an essential tool to follow the spread of the pandemic 57 by evaluating seroprevalence in populations, and they are now set to become essential to 58 evaluate the immunity of individuals as well as populations (Koopmans and Haagmans 2020) . . CC-BY 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 13, 2021. ; https://doi.org/10.1101/2021.09.06.21262027 doi: medRxiv preprint While these recent publications show that the presence of antibodies does correlate with 68 protection against the Covid-19 disease, and in particular against the more serious forms of the 69 disease, one of the more burning questions that remains to be answered is how long this 70 protection will last? Another crucial question concerns whether there will be differences in the 71 duration of this protection depending on which vaccine was used, and whether an individual 72 had been infected by the SARS-CoV-2 virus before being vaccinated. 73 74

Obtaining answers to this type of questions should be greatly facilitated by access to simple, 75 cheap and quantitative serological tests, which would work both in humans and in animal 76 models. To date, however, although a multitude of commercial serological tests have been 77 developed to detect the presence of antibodies in the serum of patients (Farnsworth and 78 Anderson 2020), those are mostly ill-suited for use in research laboratories, not only because of 79 their price, but also because they are not or only poorly quantitative. 80 81

The most commonly used methods for Covid-19 serodiagnostic are either ELISA (Enzyme-82 Linked ImmunoSorbent Assays) or CLIA (ChemiLuminescent ImmunoAssays). Whilst those 83 methods show very good sensitivity and specificity, they also have several significant 84 drawbacks: 85

i) The commercial versions are based on using volumes of serum or plasma which exceed the 86 amounts which can be readily obtained by finger prick, and therefore require venipuncture, 87 and hence trained personnel to collect the samples, and elaborate logistics to handle those 88 samples. 89

ii) They are relatively expensive (ca. 500 € per plate of 90 tests for commercial ELISA or CLIA ) 90 and not easily modular (i.e. a whole plate will often have to be used even if only a few tests 91 are to be performed). Whilst in-house ELISAs are a possible alternative, they are difficult to 92 standardize and require high amounts of recombinant antigen (Amanat et al. 2020 To be able to perform  117  HAT on whole blood with the sensitivity and simplicity which we set out the reach, and to  118 attempt to make it quantitative, various modifications and improvements had to be tested, and 119

in order to do this, we felt that we needed a simple quantitative test that would allow us to 120 evaluate the amount of antibodies present in the whole-blood samples we were using more 121 simply and cheaply than by using ELISA or CLIA. 122 123

In this regard, the S-flow test (Grzelak et al. 2020) , which uses flow cytometry (FACS) 124

performed on human cells expressing the S protein, appeared as a very promising approach 125 since it is very simple to run and its performances compared favorably with three other 126 serological tests (two ELISA directed towards the S or N proteins, and a luciferase immuno-127

precipitation system (LIPS) combining both N and S detection). 128

The S-flow method initially described used HEK cells (Grzelak et al. 2020) , which are adherent 129 cells. We felt that it would be better to use cells growing in suspension, not only because it 130 makes it a lot easier to grow large numbers of cells, but also because those can be used directly, 131 without having to be detached from the plastic, which we feared could possibly alter the cells' 132 characteristics, and introduce a possible source of variability between assays. 133

In this regard, Horndler and colleagues have recently described an assay inspired by the S-134 flow assay, but based on Jurkat cells expressing both the full-length native S-protein of SARS-135

CoV-2 and a truncated form of the human EGFR protein, which is used as an internal control for 136 the expression level of the S protein ). 137

One of the caveats of using human cells to express the S protein, however, is that those cells 138

will also express other antigens, and MHC molecules in particular, which can be the targets of 139 allo-reactive antibodies present in certain individuals, and not others. 

Jurkat-S&R-flow: basic principles 162 163

Jurkat cells expressing high levels of the SARS-CoV-2 spike protein, which we subsequently 164 refer to as Jurkat-S, or J-S, were obtained by means of transduction with a lentiviral vector, 165 followed by three rounds of cell sorting. A second population of Jurkat cells, in which all cells 166 express the mCherry fluorescent protein, and which we subsequently refer to as Jurkat-R, or J-R, 167

was also obtained by lentiviral transduction (see M&M: Materials and Methods section). 168 169

For the Jurkat-S&R-flow test, we simply prepare a mix of equivalent numbers of the two cells 170

lines, J-S and J-R, and use either sera or plasma at a final dilution of 1/100 to label 2.10 5 cells of 171 this mix (see M&M for details). After this primary step of labelling, the cells are washed before 172

being incubated with a fluorescent secondary antibody, and a final wash is performed before 173 analysis by flow cytometry. 174 175

The advantage of using such an approach is that it guarantees that the test cells (Jurkat-S) 176 and the control cells (Jurkat-R) are labelled in exactly the same conditions. Comparing the levels 177

of staining between test and control cells can then be carried out without any risk of a 178 difference between the two being due to a difference in the course of the labelling procedure 179 (e.g. less of the primary or secondary antibodies). Accessorily, another significant advantage of 180

such an approach is that it reduces the number of FACS samples to be processed by a factor of 181 two. 182 183

The mCherry signal allows simple separation by gating during analysis of the control cells 184 from the test cells ( Figure 1 ). In samples labelled just with the secondary antibody (neg. cont.) 185 or with pre-pandemic plasma which does not contain antibodies against the spike protein (neg 186 plasma), similar green signals are found on J-S et J-R populations. When there are antibodies 187 against the spike protein present in the plasma or serum used to label the cells, this will result in 188 a marked difference in the green signals detected on the J-S cells compared to the J-R cells. The 189 difference in the fluorescence intensity between the two populations will provide a quantitative 190 evaluation of the amounts of antibodies in the serum or plasma used to label the cells (first 191 column). The numbers shown in red correspond to the relative specific staining (RSS = signal J-S 192 -signal J-R / signal neg. cont.). 193 . CC-BY 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. Pos. 378

Pos. 855

Neg. Allo ++ S neg 30

Allo +++ S pos 508

Allo + S neg 10

. CC-BY 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 13, 2021. ; https://doi.org/10.1101/2021.09.06.21262027 doi: medRxiv preprint If isotype-specific secondary reagents are used, an evaluation of the respective amounts of 223 various classes of antibodies can also be obtained (second column). It should, however, be 224 noted that, because different secondary reagents do not necessarily recognize the various Ig 225 isoforms with the same efficiency, this provides only a very rough analysis of the relative 226 amounts of Ig-G, -A and -M. Within a set of samples, however, this can provide very simple 227 means to compare samples with one another (Table S1 ). 228 229

As alluded to in the introduction, one of the possible caveats of using a human cell line to 230 express the S protein is that some blood samples will contain allo-reactive antibodies directed 231 against that cell line, possibly as a consequence of a pregnancy, or past history of receiving a 232 blood transfusion or organ transplant (Hickey et al. 2016 ; Karahan et al. 2020 ). The third column 233

of Figure 1 shows examples of such samples containing marked levels of alloreactive antibodies, 234

i.e. samples for which the J-R cells show significant levels of staining compared to the same cells 235 labelled with just the secondary antibody. Based on our results collected on more than 350 236 clinical samples, we evaluate that ca. 30 % of samples will contain allo-reactive Abs that will 237 result in levels of staining of Jurkat cells that are more than five-fold that of the signal obtained 238

for the negative control (and 3-6 % more than ten-fold). Of note, we did not notice an increased 239 frequency of allo-reactivity in samples from women compared to men, which suggests that allo-240 reactivity after pregnancy is not a major cause in the origin of those allo-reactions. 241 242

A difficult question with all serological tests is that of where to set the threshold beyond 243 which the specific signals detected can confidently be considered as positive, which will be 244 directly linked to the balance between sensitivity and specificity of the assay. Based on the 245 analyses of various cohorts of positive and negative samples (some of which will be presented 246 further down in this manuscript), for the Jurkat-S&R-flow, we have settled for a threshold of RSS 247 =20, i.e. twenty-fold the value of the negative control stained just with secondary-antibodies. 248

The reason for using the geometric mean of fluorescence intensity (GMFI) of the negative 249 control as an internal reference is that, in flow cytometry, the numerical values of fluorescence 250 intensities will be totally dependent upon the cytometer settings, and the voltages applied to 251 the PMT in particular. But we find that this can be somewhat compensated by such an 252 approach. For example, in the conditions used in our experiments, the geometric mean of 253 fluorescence intensity (GMFI) of the negative control had a value of ca. 2 when analyzed on a 254

FACScalibur flow cytometer. With a threshold set at 20, samples were thus considered as 255 positive if the difference in the GMFI of the J-S and J-R populations was above 40 (numbers 256

shown in red in Figure 1 are J-S -J-R / 2). When the very same samples were analyzed on a 257

Fortessa flow cytometer ( Figure S1 ), the value of the GMFI for the negative control was 24-fold 258

higher, but the RSS values obtained closely resembled those obtained with the same samples on 259

the FACScalibur (red numbers in Figure S1 and all Jurkat cells. Such alloreactive signals can be higher in J-R for certain serum or plasma 268 . CC-BY 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 Inspection of the dot plot for this very alloreactive sample provides the explanation for the 278 slightly odd shapes of the FL1/FL3 gates we used to discriminate Jurkat-S from Jurkat-R cells. 279

This was necessary because, on the FACScalibur, the fluorescent signal of mCherry is best 280 captured by the FL3 channel, but the version of the Cellquest program used for acquisition on 281 this cytometer does not allow for FL3/FL1 compensation. Consequently, samples with extremely 282

high FL1 signals showed some 'bleeding' into the FL3 channel. We found that this could not be 283 satisfactorily treated by post-acquisition compensation with the Flowjo analysis software either, 284

and thus resorted to drawing such gates to separate J-S from J-R populations. 285 286 We had elected to use a green/red combination for test and control cells for two reasons: 1) 287 secondary antibodies labeled with green fluorescent dyes such as fluorescein or Alexa 488 are 288 the most commonly available, and also usually the cheapest. 2) All flow-cytometers, even the 289 most basic ones, are equipped with a 488 nm laser which allows to perform green/red analyses. 290

Another important consideration is that one of the goals of this study was to set up a test which 291 could be used by as many research teams as possible, including those based in institutes from 292 less affluent countries, which are less likely to have access to recent, state-of-the-art multi-laser 293 flow cytometers. 294 295

When the same samples as shown on Figure 1 were analyzed using a Fortessa flow 296 cytometer, the picture was quite different ( Figure S1 ). This more recent flow cytometer is 297 endowed with several lasers, including a 561 nm Yellow-Green laser, which is much better 298 suited for the excitation of the mCherry fluorescent protein, thus yielding much higher signals 299 which, since they are acquired on a different laser line from the green signals, require absolutely 300 no compensation. 301 302

In many flow cytometry facilities, users are required to fix any samples that have been in 303 contact with materials of human origin. Whilst this was not the case for us, we still tried 304

analyzing the same samples after those were fixed with 1% formaldehyde and kept for 4 days at 305

4°C before re-analysis. As can be seen on the right part of Figure S1 , formaldehyde-fixation 306 resulted in a 7-fold drop in the intensity of the mCherry signals, which made it impossible to 307 separate J-R from J-S population on the FACScalibur (first line). On the other hand, analysis on 308 the Fortessa was still comfortably possible. Of note, whilst formaldehyde did not noticeably 309

alter the fluorescence signals of the alexa-488 dye, it did result in a twofold increase of the 310 green auto-fluorescence of the negative controls, hence resulting in a twofold reduction of the 311 RSS compared to those obtained on unfixed cells. 312 313 314 . CC-BY 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 result of the analysis of these two cohorts with the three serological tests are presented 334

in Figure 2 , with colors used to represent the HAT results. The results of the first cohort ( Figure 2A) The results of the second cohort, which comprised a few Covid patients, but also a large 370

proportion of blood samples randomly picked among those from patients hospitalized for 371

conditions unrelated to Covid-19, yielded a much less clear picture than the first one. As can be 372 seen on Figure 2B As a proof of concept that this was doable and useful, we used capillary blood which one of 421 the authors collected by pricking his finger at various intervals to document the time course of 422 his serological response after vaccination, and the results are shown in panel A of Figure 3 . One 423 somewhat surprising finding was with the low levels of IgM recorded, which may in part be 424 explained by the fact that, as a rule, anti-IgM secondary antibodies tend not to work as well as 425 those against the other isotypes. One should note, however, that, with the same anti-IgM 426 secondary reagent, signals of the same order of magnitude were found for IgG and IgM with the 427 20/130 reference serum (Figure 1) , as well as in several other samples (Table S1 ). The 428

observation that the signals obtained with an anti-IgG reagent are similar, and even often a bit 429

higher to those obtained with the pan-reactive anti-IgGAM is something which we tend to find 430 in most samples (Table S1) , and which had already been underlined by Grzelak First, we used sera of mice which had been immunized once or twice with inactivated SARS-444

CoV-2 virus injected intra-peritoneally. As can be seen on panel B of Figure 3 , whilst the sera of 445 control mice did not react significantly with Jurkat cells, those of immunized mice showed 446 strong specific reactions against the spike protein, which were even higher in those having 447 received two injections. Of note, the reactions against the Jurkat-R cells also went up in 448 correlation with the injections, albeit to a much lesser degree than against the S protein. This is 449

presumably due to the fact that the preparations of inactivated virus used for the 450 immunizations probably contained some bovine proteins from the serum used in the tissue 451 culture medium that would have the capacity to bind to the surface of the Jurkat cells, such as, 452

for example, beta-2-microglobulin binding to MHC class I molecules. The Jurkat-S&R-flow 453 method therefore seems to work very satisfactorily in mice, and since it requires only a few µL 454 of blood, it is well suited to follow serological responses over time. 455 456 We then turned our attention to cats and dogs, which have both been shown to be 457 susceptible to infections by the SARS-CoV-2 virus ( S&R-flow test, using the appropriate secondary antibodies. As can be seen on panel C of Figure  464 3, we found a very good correlation between the levels of specific staining of Jurkat-S cells and 465 the neutralization titers of the same sera. Whilst those are very preliminary observations which 466 will need to be strengthened by many more samples, and in particular with samples collected 467 before the Covid-19 pandemic as negative controls, those results show that the Jurkat-S&R-flow 468 test can clearly be used to evaluate the levels of antibodies against the S protein of the SARS-469

CoV-2 virus in cats and dogs, and could presumably also be adapted to other animals, be they 470 pets, farm animals or laboratory models used to study the viral infection. supported by the results shown in Figure S2 , i.e. that, for most samples, we found that 498 incubation at room temperature resulted in a sizeable increase in the amounts of antibodies 499

binding to the spike-expressing cells, suggesting that, at the surface of live cells, the capacity of 500 the spike protein to fluctuate between various conformations will expose different epitopes, 501

and allow the binding of more antibodies. 502 503

One striking aspect of the results shown on Figure 2 is in the difference of performance of 504 the tests between the two cohorts of samples tested. On the one hand, nearly perfect scores 505

were obtained for all three tests on a cohort of sera comprised either of control samples 506 collected before 2019, or of positive sera from PCR-positive Covid-19 patients. On the other 507 hand, the situation was much less clear-cut for the cohort comprising blood samples picked 508 more or less randomly and blindly among those available as left-overs from the hematology 509 department and was, therefore, more akin to a 'real' population. For this second cohort, an 510 additional confounding factor may have been that, since all the samples were from hospitalized 511 patients, some sera may have been poly-reactive due to inflammatory pathologies unrelated to 512

Covid-19. 513 514

All in all, this difference between the two cohorts is reminiscent of the common observation 515 that the performances of clinical tests are often much lower on real populations than those 516 obtained by the manufacturers on very carefully controlled and standardized populations. Our 517 results indeed bring support to the view that the performance of any given serological test will 518 be entirely dependent on the set of samples used to measure it: whilst it is relatively easy to 519

reach an almost perfect score on a cohort comprised only of highly positive and completely 520 negative samples such as the one used for Figure 2A , the situation becomes much less clear 521 when using a set of samples more closely resembling the general population, in which the 522 positive or negative nature of many samples will remain uncertain, and from which it would 523 thus seem futile to try to make precise calculations of sensitivity and specificity. This being said, 524

performing several tests in parallel on such cohorts is very useful to compare the performance 525 of those tests with one another. Whilst HAT is not as sensitive as an anti-RBD ELISA or the Jurkat-S&R-flow test, this relatively 547 low sensitivity may not really be a problem for using HAT to help decide when to revaccinate 548 people since low levels of antibodies are unlikely to be fully protective. By performing titrations, 549

the HAT assay can also provide a quantitative assessment of the levels of circulating antibodies, 550

which have been shown to correlate strongly with sero-neutralisation titers (Lamikanra et al. 551 2021). In the current version of HAT, however, such a quantification can only be performed in a 552 laboratory environment because it requires separation of the plasma or serum from the red 553 blood cells. Making use of the Jurkat-S&R-flow test as a reference, we are currently in the 554 process of completing work on a modified version of HAT that will be compatible with being 555 performed pretty much anywhere, with no specialized equipment, and will provide a 556 quantitative evaluation of the levels of antibodies in a single step (Joly et al. man in prep.). 557 Topro-3 and  570 for performing the Mycoplasma tests; Emmanuelle Naser and Penelope Viana from the IPBS 571 flow cytometry facility for their assistance, and the very helpful staff of the Toulouse EFS. 572 573

Funding: The first part of this project was funded by a private donation. The second part was 574 funded by the ANR grant HAT-field to EJ. 575

Reagents 578  579  Polyclonal anti-human and anti-mouse Igs secondary antibodies, all conjugated to Alexa-488,  580 were from Jackson laboratories, and purchased from Ozyme EcoRI sites. 604

The lentiviral vector for the expression of mCherry was obtained by replacing the GFP 605 sequence of GFP by that of mCherry in the pCDH-EF1α-MCS*-T2A-GFP plasmid 606

(https://systembio.com/shop/pcdh-ef1α-mcs-t2a-gfp-cdna-single-promoter-cloning-and-607 expression-lentivector/) 608 Lentiviral infectious supernatants were obtained after transient transfection of HEK cells with 609 pLV-or pCDH-derived vectors, together with the packaging R8-2, and VSV-G plasmids. 610

Supernatant was harvested 24h and 48h post transfection, passed through a 45 μm filter and 611 stored in aliquots at -80°C. For the transduction of Jurkat cells, those were distributed in a 6 well 612 plate at 1.5. 10 6 cell per well, in a volume of 500 µL of tissue cuture medium (RPMI, 10 % FCS, 613

1% PS, 2% Hepes), and 20 µL of the infectious supernatants were added, as well as 5 µL of 614

Lentiblast premium (OZBiosciences). The plate was then spun at 1000g for 60 min at 32°C, 615 before adding 2.5 ml of tissue culture medium and returning the plate to the 37°C incubator. 616

Selection with puromycin was then performed at a concentration of 10µg/mL. 617

After a few days, the population of Jurkat cells thus obtained was stained for flow cytometry 618 analysis using a highly reactive serum from a covid-19 patient, and it was found that most cells 619 expressed the S protein, but at low to intermediate levels.

To obtain a population that would 620 express higher levels, we submitted this population to three successive rounds of sterile cell 621 sorting using a FACSAria Fusion cell sorter (Beckton Dickinson), selecting each time the 5 % of 622 cells with the brightest staining. Cells were placed back in culture and reamplified after each 623 round of selection. At the time of the second round of sorting, the cell sorter was also used for 624 single cell cloning, but all of the dozen clones obtained by this means showed lower expression 625 that the sorted population. The expression of the S protein in the population of Jurkat cells 626 obtained after these three rounds of sorting was found to remain expressed in all cells, and at 627 similar levels, for more than 50 successive passages, over many weeks of continuous cell 628 culture. 629 630

The Jurkat-R cells were obtained by successive transduction with the pCDH-GFP lentiviral 631

vector described above, then with the empty pLV lentiviral vector, followed by selection with 632

Puromycin at 10µg/mL. FACS analysis revealed that the mCherry fluorescent protein was 633 expressed in 100 % of the cells of the population thus obtained, which therefore did not need to 634 undergo any cell sorting. counted, and sufficient numbers harvested to have a bit more than 10 5 cells of each per sample 644 to be tested. Cells from both cell lines were then spun, and resuspended in their own tissue 645 culture medium at a concentration of 2.2 10 6 cells/ mL before pooling equal volumes of the two. 646

Plasmas or sera to be tested were diluted 1/10, either in PBS or in PFN (PBS / 2% FCS / 200 647 mg/L sodium azide ). 10 µL of these 1/10 dilutions were then placed in U-bottom 96 well plates, 648 before adding 90 ul per well of the Jurkat-S&R mix. 649

The plates were then incubated for 30 minutes at room temperature before placing them on 650 ice for a further 30 minutes. As can be seen on the supplementary Figure 2 , we have found that 651 this initial incubation at room temperature results in a marked increase of the staining signals 652 for most antibodies. 653

All subsequent steps were carried out in the cold, with plates and washing buffers kept on 654

ice. After the primary staining, samples were then washed in PFN, with resuspending the cells 655 by tapping the plate after each centrifugation, and before adding the next wash. the tubes, and those were stored at 4°C for 4 days before they were once again analyzed on 676 both machines. 677

Post-acquisition analysis of all the samples was performed using the Flowjo software (version 678 10.7.1) 679 680

Cost of the Jurkat-S&R-flow test 681

The cost per sample of the Jurkat-S&R-flow test lies in large part with the price of the 682 secondary antibodies used. Typically, one vial of secondary antibody costs about 150 € for 1ml, 683

and using 30-50 ul at a 1/200 dilution will provide for at least 5000 samples. The cost of a 684 polyclonal antibody is thus of the order or 3-5 cts per sample. 685

Each sample requires 2.10 5 jurkat cells, which is roughly the amount obtained with 0.5 ml of 686 standard tissue culture medium, which costs around 50 €/L, i.e. 2.5 cts/sample. All in all, if one 687 adds the cost of TC, buffers for the washes and disposable plastics, we estimate that the cost 688 per sample will be of the order of 10 cts. 689

The cost of access to a flow cytometer will be extremely variable between laboratories and 690

institutes. If the cost of access is 20 € per hour, and one runs 200 samples per hour, this will add 691 another 10 cts to the cost per sample. 692 693

HAT tests 694

HAT tests were performed using the IH4-RBD reagent diluted at 1 ug/ml in PBN rather than in 695 straight PBS as originally described (Townsend et al. 2021 In an attempt to increase the sensitivity of the method by detecting partial 715 hemagglutinations, the plates were then returned to a horizontal position for a further two 716 hours, and the procedure of tilting the plate and taking pictures was repeated. 717

After transferring the pictures to computer files, the hemagglutination tests were scored by 718

three independent assessors, of which two were blinded. 719

The scoring system was as follows: A score of 1 was given only to those samples which 720

showed complete hemagglutination after one hour. A score of 0.5 was given to samples which 721

showed either partial hemagglutination after one hour, or partial or complete hemagglutination 722 after the second incubation of 2 hours. 723

This revealed that, whilst there was 100% agreement between the three assessors for the 724 scoring of complete hemagglutination, the scoring of partial hemagglutination proved to be 725 much more subjective and variable, and only those samples which had been scored as partials 726 by all three assessors were finally considered as bona fide partial hemagglutinations. Figure S1 : Using a flow cytometer with a 561 nm Yellow-Green laser improves the mCherry 947 signals 948

The same samples shown on Figure 1 were also analyzed on a Fortessa flow cytometer. The 949 561 nm yellow-green laser was used for the excitation of the mCherry fluorescent protein, 950 which led to much brighter signals, and thus to much easier separation of the Jurkat-R (mCherry 951 pos) and Jurkat-S (mCherry neg) populations than when the samples were acquired on a 952

FACScalibur. 953

As in figure 1 , the numbers in the upper right corners of the histogram overlays plots 954

correspond to the GMFI of the two histograms (Red: Jurkat-R; green: Jurkat-S), and the big red 955

numbers to the left of the plots to the RSS. Of note, although the GMFI values were much higher 956 on the Fortessa than those recorded on the FACScalibur, the RSS values of the various samples 957

were all very similar between the two machines. 958 959

After the samples had been run on both flow cytometers, those were fixed in 1% 960 formaldehyde by adding an equal volume of PBS + 2% formaldehyde to the remainder of each 961

sample. The tubes were then stored à 4°C for 4 days before re-analyzing them on both 962

cytometers. As can be seen on the right hand side of the figure, the red fluorescence of the 963 mCherry was still detectable on the Fortessa after this procedure of fixation, despite the 964 intensity of the signals having dropped by about 7 fold. On the FACScalibur, however, this drop 965 of the mCherry fluorescent signals precluded the separation of the Jurkat-S from the Jurkat-R 966

cells. If samples are to be fixed, performing the Jurkat-R&S-flow test will thus require access to a 967 flow cytometer equipped with a 561 nm Yellow-Green laser. As an alternative, we are currently 968 exploring the possibility of transforming the Jurkat-S&R-flow test into a Jurkat-S&G-flow test, 969

where the negative control cells would be expressing GFP, which is excited at 488 nm, and using 970 secondary antibodies conjugated to red fluorochromes such as Alexa647, excited by a 633 nm 971 red laser, for the detection of the primary antibodies. 972 973

Of note, in the dot plots of the Positive and Strong Positive fixed samples, faint clouds of FL1 974 negative cells can be seen in the mCherry neg window. Those presumably correspond to Jurkat-975 R cells which were dead before the fixation step, and from which the mCherry had thus leaked 976 out. When analyzing populations of live cells, such cells are gated out with the To-pro 3 live 977

stain, but this no longer works for population of fixed cells. This small problem could probably 978 be circumvented using a fixable dead cell staining dye such as the LIVE/DEAD Fixable Far Red 979 stain (ThermoFischer L34973). 980 981 982 983 984 Figure S2 : Incubation of the samples at room temperature for the first part of the primary 985 staining step markedly improves the staining signals for most antibodies. 986

For a large fraction of monoclonal and polyclonal antibodies reacting with the Spike protein 987 expressed at the surface of cells, we have repeatedly observed that a step of incubation at room 988 temperature can result in a marked improvement of the staining signals compared to 989 performing the staining continuously on ice. 990

For this experiment, we selected a panel of 5 monoclonal antibodies reacting with different 991 sites of the spike RBD domain, a panel of 4 plasmas reacting with various intensities, as well the 992 20/130 reference serum. 993

For all the samples, we performed eight steps of 3-fold dilutions, and placed 10 ul of the 994 diluted samples in two parallel U bottom 96 well plates. The first plate was kept at room 995 temperature (RT) whilst the other one was placed on ice. The Jurkat-S&R mix was then added to 996 the wells of the plate at room temperature. The tube was then placed on ice for a few minutes, 997

before distributing the mix in the wells of the second plate. After 30 minutes, the plate which 998 had been sitting at room temperature was also placed on ice, and incubated for another 30 999 minutes. The rest of the staining procedure was then exactly the same for the two plates, all 1000 performed in the cold as described in M&M. 1001 1002 

At the time of collection, the blood 759 samples were diluted with 200 µL of PBS + 5mM EDTA (Since whole blood is roughly 50% RBCs 760 and 50% plasma, this 1/5 dilution of the blood actually corresponds to a 1/10 dilution of the 761 plasma). The plasma was then separated from the RBCs after centrifugation, placed in another 762 tube with sodium azide, and stored at 4°C until the day of the assay. 763 764 Experiments on mouse sera 765 Virus preparation and inactivation 766 SARS

Viral stocks were propagated in 300 cm 2 flasks (Dutscher) in which 10 3 tissue 769 culture infectious dose 50 (TCID50) were inoculated in 100 ml of medium for 3 days at

Culture supernatants containing the viral stocks were harvested and inactivated with 771 BPL (Fischer) overnight at 4°C. Virus was then concentrated by ultracentrifugation at 25 000 rpm 772 (for 2 hours at 4°C) on a 20% sucrose cushion in Ultra-Clear centrifuge tubes

Pellets were resuspended in PBS, protein concentration was quantified by BCA 774

BPL-inactivated SARS-CoV2 stocks at protein concentrations ranging 775 from 1 to 5 µg/µL were stored at -80°C until use

Mice were euthanized and serum was collected at 779 day 14 post-primary immunization and at day 7 post-secondary challenge. 780 781 Experiments on Cat and Dog sera 782 Serum samples were collected from cats and dogs belonging to owners who developed COVID-783 19-like symptoms and subsequently tested positive for SARS-CoV-2 infection by RT-qPCR. , at least 784 one month after the owners' recoveries. Samples and data collections were conducted according 785 to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee Sciences 786 et Santé Animale n•115

Sero-neutralisation assay

After 60 minutes at 37°C, the 796 virus-serum dilutions were removed before adding DMEM complemented with 2% of heat-797 inactivated fetal bovine serum and 1% of penicillin-streptomycin. Cells were then incubated for 798 72 h at 37 °C with 5% of CO2. Individual wells were then screened by eye under the microscope 799 for cytopathic effects. Monoclonal anti-SARS-CoV CR3022 antibody (BEI Resources, NIAID, NIH) 800 was used as a positive control

Clinical 820 performance of a rapid test compared to a microplate test to detect total anti SARS-CoV-2 821 antibodies directed to the spike protein

SARS-CoV-2 reinfection in a cohort of 823 43,000 antibody-positive individuals followed for up to 35 weeks

Antibody testing for COVID-19: A report 825 from the National COVID Scientific Advisory Panel

A serological assay to detect SARS-CoV-827 2 seroconversion in humans

SARS-CoV-2 neutralizing antibody 829 structures inform therapeutic strategies

Household Cases Suggest That Cats 831 Belonging to Owners with COVID-19 Have a Limited Role in Virus Transmission

Susceptibility of ferrets, cats, dogs, and different domestic animals to SARS-834 coronavirus-2

Serological evidence of human infection with 836 SARS-CoV-2: a systematic review and meta-analysis. The Lancet

The antigenic anatomy of SARS-CoV-2 839 receptor binding domain

Evaluating 10 Commercially Available 841 SARS-CoV-2 Rapid Serological Tests by Use of the STARD (Standards for Reporting of 842 Diagnostic Accuracy Studies) Method

Current State 844 of Knowledge about Role of Pets in Zoonotic Transmission of SARS-CoV-2

Sensitive detection of SARS-846

CoV-2 seroconversion by flow cytometry reveals the presence of nucleoprotein-reactive 847 antibodies in unexposed individuals

Comparison of the COVID-2019 (SARS-CoV-2) 849 pathogenesis with SARS-CoV and MERS-CoV infections

SARS-CoV-2 Serology: Much Hype, Little Data

Correlates of protection against symptomatic and 853 asymptomatic SARS-CoV-2 infection

COVID-19-neutralizing antibodies 855 predict disease severity and survival

A flow cytometry-based assay for serological 857 detection of anti-spike antibodies in COVID-19 patients

A comparison of four serological assays for 859 detecting anti-SARS-CoV-2 antibodies in human serum samples from different populations

A simple, sensitive, and low-cost FACS assay for 862 detecting antibodies against the native SARS-CoV-2 spike protein

Association of SARS-CoV-2 Seropositive 865

Antibody Test With Risk of Future Infection

SARS-CoV-2 variants, spike mutations 867 and immune escape

Alloantibody generation and effector 869 function following sensitization to human leukocyte antigen

Flow cytometry multiplexed method for the 872 detection of neutralizing human antibodies to the native SARS-CoV-2 spike protein

Breadth and function of antibody response 875 to acute SARS-CoV-2 infection in humans

Comparison of kinetics of immune 877 responses to SARS-CoV-2 proteins in individuals with varying severity of infection and 878 following a single dose of the AZD1222

Immune responses to a single 880 dose of the AZD1222/Covishield vaccine in health care workers

Antibodies to SARS-CoV-2 protect 883 against re-infection during outbreaks in care homes

Comparison of two assays to detect 886

IgG antibodies to the receptor binding domain of SARS-CoV-2 as a surrogate marker for 887 assessing neutralizing antibodies in COVID-19 patients

Pre-existing Alloreactive T and B Cells and

Their Possible Relevance for Pre-transplant Risk Estimation in Kidney Transplant Recipients. 891 Frontiers in medicine 7

Common Human Coronaviruses and SARS-CoV-2 Using Coronavirus Antigen Microarray

Neutralizing antibody levels are highly 896 predictive of immune protection from symptomatic SARS-CoV-2 infection

Assessing the extent of SARS-CoV-2 circulation 899 through serological studies

Comparability of six different 901 immunoassays measuring SARS-CoV-2 antibodies with neutralising antibody levels in 902 convalescent plasma: from utility to prediction

Rapid response flow cytometric assay for the 904 detection of antibody responses to SARS-CoV-2

SARS-CoV-2 Seropositivity and Subsequent 907 Infection Risk in Healthy Young Adults: A Prospective Cohort Study

Clinical and immunological assessment of 910 asymptomatic SARS-CoV-2 infections

Antibody Status and Incidence of 912 SARS-CoV-2 Infection in Health Care Workers

Human monoclonal antibody 915 combination against SARS coronavirus: synergy and coverage of escape mutants

A comparative review of immunoassays for COVID-918 19 detection

SARS-CoV-2 lateral flow assays for possible use 920 in national covid-19 seroprevalence surveys (React 2): diagnostic accuracy study

Preexisting and de novo humoral immunity to 923 SARS-CoV-2 in humans

Mapping Neutralizing and Immunodominant 925 Sites on the SARS-CoV-2 Spike Receptor-Binding Domain by Structure-Guided High-926 Resolution Serology

Flow cytometry detection of 928 sustained humoral immune response (IgG + IgA) against native spike glycoprotein in 929 asymptomatic/mild SARS-CoV-2 infection

Convergent antibody responses to SARS-931

CoV-2 in convalescent individuals

Infection of dogs with SARS-CoV-2

A haemagglutination test for rapid detection of 935 antibodies to SARS-CoV-2

The unique features of SARS-CoV-2 937 transmission: Comparison with SARS-CoV, MERS-CoV and 2009 H1N1 pandemic influenza 938 virus

Structural basis for the neutralization of 940 SARS-CoV-2 by an antibody from a convalescent patient

Because polyclonal secondary antibodies will not all react with their targeted primary antibodies 1008with the same efficiency, however, the actuals signals for the various subclasses of antibodies 1009should not be taken as a true reflection of the amounts of each subclass of antibody. 1010If a more quantitative evaluation was needed, monoclonal secondary antibodies could be 1011 used instead of polyclonal reagents, making sure that all those isotypes-specific secondary 1012 monoclonal antibodies were being used well above the saturating concentration, and 1013 conjugated to similar amounts of the same fluorochrome. This would, however, increase the 1014 cost of the procedure considerably.