DQ ! MICRO BIO

B R I E F R E P OR T

SARS-CoV-2 neutralization and serology testing of COVID- 19 convalescent plasma from donors with nonsevere disease

Thomas J. Gniadek1 | Joshua M. Thiede2 | William E. Matchett3 |

Abigail R. Gress2 | Kathryn A. Pape3 | Jessica K. Fiege3 | Marc K. Jenkins3 |

Vineet D. Menachery4 | Ryan A. Langlois3 | Tyler D. Bold2

1Department of Pathology and Laboratory Medicine, NorthShore University HealthSystem, Evanston, Illinois 2Division of Infectious Diseases and International Medicine, Department of Medicine, Center for Immunology, University of Minnesota Medical School, Minneapolis, Minnesota 3Department of Microbiology and Immunology, Center for Immunology, University of Minnesota Medical School, Minneapolis, Minnesota 4Department of Microbiology and Immunology, University of Texas Medical Branch, Galveston, Texas

Correspondence Tyler D. Bold, 2101 6th St SE, WMBB 2-118, Minneapolis, MN 55455. Email: [email protected]

Funding information This work was supported by UMN Department of Medicine funding (T.D.B.), NIH T32AI055433 (J.M.T.), T32HL774126 (W.E.M.), 1R01AI153602 (V.D.M.), and by private donations to the NorthShore Foundation for the NorthShore COVID-19 convalescent plasma program.

Abstract

Background: The transfer of passive immunity with convalescent plasma

is a promising strategy for treatment and prevention of COVID-19,

but donors with a history of nonsevere disease are serologically heteroge-

nous. The relationship between SARS-Cov-2 antigen–binding activity and neutralization activity in this population of donors has not been

defined.

Study Design and Methods: Convalescent plasma units from 47 individuals

with a history of nonsevere COVID-19 were assessed for antigen-binding activ-

ity of using three clinical diagnostic serology assays (Beckman, DiaSorin, and

Roche) with different SARS-CoV-2 targets. These results were compared with

functional neutralization activity using a fluorescent reporter strain of SARS-

CoV-2 in a microwell assay.

Results: Positive correlations of varying strength (Spearman r = 0.37-0.52)

between antigen binding and viral neutralization were identified. Donors age

48 to 75 years had the highest neutralization activity. Units in the highest ter-

tile of binding activity for each assay were enriched (75%-82%) for those with

the highest levels of neutralization.

Conclusion: The strength of the relationship between antigen-binding activity

and neutralization varies depending on the clinical assay used. Units in the

highest tertile of binding activity for each assay are predominantly comprised

of those with the greatest neutralization activity.

The transfer of passive immunity with convalescent plasma is a promising strategy for treatment and prevention of COVID-19.1 Challenges include the het- erogeneity of SARS-CoV-2 antibody responses2 and the variety of assay techniques to measure antigen- binding activities.3 Although individuals who

experience more severe COVID-19 disease tend to show greater SARS-CoV-2 antibody titers,4 the major- ity of individuals diagnosed with COVID-19 (and potential convalescent plasma donors) have nonse- vere disease.5 It is therefore especially important to assess the variability in antigen-binding activity of con- valescent plasma from donors with less severe disease and to determine how readily available serologic mea- surements correlate with functional neutralization activ- ity in this population.

Thomas J. Gniadek, Joshua M. Thiede, and William E. Matchett contributed equally to this study.

Received: 12 August 2020 Revised: 28 August 2020 Accepted: 31 August 2020

DOI: 10.1111/trf.16101

Transfusion. 2020;1–7. wileyonlinelibrary.com/journal/trf © 2020 AABB 1

Viral neutralization titers measure the ability of antibodies to prevent viral infection of a eukaryotic cell line in vitro. Both live and pseudotyped virus assays exist for SARS-CoV-26 as well as surrogate assays that measure blockage of the Spike–ACE2 pro- tein interaction.8 Pseudotyped and recombinant pro- tein assays require less restrictive biosafety procedures and facilities; however, results may differ from live viral assays since pseudotyped viruses typically express only a single viral entry protein. Neutralizing antibodies that impact other viral targets and processes may therefore not be detected by pseudotyped viral assays.9 Readouts of neutralization assays also vary, but standard viral plaque reduction readouts are labor-intensive. Fluorescence-based assays use an engineered viral particle with a fluorescent protein gene expressed upon viral infection of eukaryotic cells.10 One recently developed fluorescence microwell assay correlated well with plaque reduction but was faster and could be readily automated.7 In contrast, live SARS- CoV-2 neutralization assays remain available only in spe- cialized laboratories and are not widely used to screen con- valescent plasma due to complexity of implementation, high cost, low throughput, and biosafety concerns.

In addition, clinical assays that measure the bind- ing of antibodies against SARS-CoV-2 to specific viral antigens have been rapidly developed. These assays were primarily designed to diagnose past COVID-19 exposure and none of the currently available assays measure the ability of antibodies to neutralize the virus or prevent viral entry into cells. However, many of these clinical assays can be run on high-throughput, automated instruments. As a result, these assays can be performed at low cost, low safety risk to laboratory staff, and high throughput and in almost any clinical laboratory.

Initial reports have suggested some correlation between binding antibody activity and viral neutraliza- tion titers.11 Data from other coronaviruses suggests that plasma with detectable antibodies at a 1-in-160 or 1-in-320 dilution using a clinical binding assay should have high neutralizing titers as well.11 However, these assumptions have not been rigorously tested and the dilution of samples requires specific assay validation and an extra step in the testing process. Even without dilu- tion, many clinical serology assays provide a quantitative serologic score of antibody reactivity.

Despite these uncertainties, there is currently high demand for COVID-19 convalescent plasma for use in clinical trials and under the recent FDA emergency use authorization. Therefore, we sought to determine in a real-world cohort of donors with a history of nonsevere COVID-19, the relationship between antigen-binding

activity measured by several FDA-approved clinical diagnostic assays under emergency authorization, and neutralization activity against live SARS-CoV-2 recom- binantly engineered to express fluorescent mNeonGreen protein in infected Vero E6 cells.7,10

1 | MATERIALS AND METHODS

1.1 | Convalescent plasma donor recruitment

The NorthShore University HealthSystem COVID-19 convalescent plasma collection program was established in April 2020. This program and associated human sub- ject research, performed in in accordance with the ethi- cal standards of the Helsinki Declaration, were approved by the NorthShore University HealthSystem Institutional Review Board. All potential donors pro- vided written consent for the study and provided infor- mation about their COVID-19 disease history and demographics. Disease history was reported in free-text format and the absence of a reported symptom was assumed to indicate that the symptom was not present. For the donors included in this study, reported symp- toms included fatigue (49%), myalgia (47%), cough (47%), anosmia (43%), headache (40%), and other symp- toms 20% each (Table S1). None of the donors had a his- tory of hospitalization for COVID-19.

1.2 | Sample collection, storage, and transport

Samples were collected at the time of donation using serum separator tubes (BD, Franklin Lakes, NJ), cen- trifuged, aliquoted, and frozen at −80°C. For each sam- ple included in this study, an aliquot was shipped to the University of Minnesota for viral neutralization titer measurement on dry ice.

1.3 | Clinical serology testing

An aliquot was thawed and tested at NorthShore using the Elecsys anti-SARS-CoV-2 (Roche Diagnostics), Access SARS-CoV-2 IgG (Beckman Coulter), and LIAISON SARS-CoV-2 S1/S2 IgG (DiaSorin). Both the quantitative cutoff index and the qualitative results were recorded. Samples were divided into tertiles based on the Roche assay results, and then 47 were randomly selected to equally sample each tertile.

2 GNIADEK ET AL.

1.4 | SARS-CoV-2 S1 RBD immunoglobulin enzyme-linked immunosorbent assay

The anti–SARS-CoV-2 S1 RBD total immunoglobulin assay employed a standard indirect enzyme-linked immunosorbent assay (ELISA), described in detail else- where, using a secondary antibody recognizing all human immunoglobulin isotypes (goat anti-human IgG H+L-HRP, Invitrogen/ThermoFisher).12,13 Threefold serum dilutions were tested: 1 in 50 to more than 1 in 12 150. ELISA titer was reported as the dilution at which absorbance of each sample tested exceed the 50% maximal absorbance signal for a positive control sample on the same ELISA plate.

1.5 | Live SARS-CoV-2 virus neutralization assay

Vero E6 cells (2.5 × 104) were seeded in each well of a 96-well black/clear flat bottom TC-treated plate (Falcon) and incubated in DMEM overnight at 37°C with 5% CO2 before infection. Plasma samples were twofold serially diluted (from 1 in 20 to 1 in 5120) in DMEM and incubated with mNeonGreen SARS-CoV-2 at 37°C for 1 hour. Medium was removed from cells and the virus–plasma mixture was added to achieve a final multiplicity of infection of 0.1 plaque-forming units per cell. The cells were incubated at 37°C with the virus–plasma mixture for 24 to 26 hours. After incubation, cells were fixed in 4% paraformaldehyde at 4°C for 30 minutes. The paraformaldehyde–virus–plasma mixture was removed, cells were washed once with PBS, and 50 μL of PBS was added to each well. The fluorescence signal was determined by reading the plates on a hybrid multimode reader (Synergy H1, BioTek ), using excitation/emission wavelengths of 488/517 nm. Percent maximal infection was determined for each dilution as the ratio of the fluores- cent signal to the maximal signal for non–serum-treated controls in the same plate. A nonlinear regression method was used to determine the dilution that neutralized 50% of mNeonGreen fluorescence (NT50) by using Prism 8 (Gra- phPad). If a plasma titration failed to generate 50% inhibi- tion within the range of concentrations tested, a titer value of one-half (10) of the highest serum concentration tested was ascribed to it. Each sample was tested in duplicate.

1.6 | Statistical analysis

Assay results were compared using linear regression and Spearman correlation. Results from each assay were bro- ken into equal tertiles for comparison; tertiles were

compared using the Kruskal-Wallis test. Comparisons of NT50 between two groups used the Mann-Whitney U-test.

2 | RESULTS

2.1 | Fluorescent SARS-CoV-2 neutralization

We tested 47 units of convalescent plasma units from individuals with a history of nonsevere COVID-19 for viral neutralization activity. There was a broad range of neutralization against live SARS-CoV-2, with 6 (13%) units demonstrating high NT50 values of more than 500, and six (13%) with undetectable NT50 values of less than 20 (Figure S1). Neutralization assays results yielded a robust calculation of NT50 with R

2

values of more than 0.9 for curve fitting (Figure 1A) in the majority of samples with detectable neutralization activity.

2.2 | Binding and neutralization assay comparison

We compared the neutralization activity in these samples to binding activity as measured by an in-house ELISA for S1 RBD total immunoglobulin as well as three clinical diagnostic assays that use different viral antigenic targets: Beckman (IgG anti-S1 RBD), DiaSorin (IgG anti-S1/S2 protein), and Roche (total anti-Nucleocapsid immuno- globulin). The geometric mean, median, and quartiles (first, second, third, and fourth) for each assay were as follows: Beckman (2.75, 2.81, 1.2, 2.81, 8.75, 35.06), DiaSorin (31.41, 31.9, 16.85, 31.9, 59.6, 187), and Roche (10.84, 16.11, 4.38, 16.11, 41.70, 109.5). There was a uni- versally positive relationship between NT50 and all four binding assays tested, with Spearman correlation r rang- ing from 0.37 to 0.52 and R2 values of 0.17 to 0.40 reflecting a weak linear relationship (Figure 1B,C). The strongest positive correlation between assays was the Beckman and DiaSorin assays (Spearman r = 0.85), which measure different aspects of anti-Spike protein– binding activity. The Roche total antinucleocapsid assay had the lowest overall correlation with NT50, but stronger positive correlation with the Beckman and DiaSorin assays (Figure 1C).

We identified several individual donors with dis- cordant binding and neutralization activity, including some with high neutralization and low binding activi- ties in individual assays (Figure 1A,B, open hexagon

GNIADEK ET AL. 3

and open inverted triangle). We also identified individ- ual units characterized by low neutralization despite relatively high binding activity in individual assays (Figure 1A,B, black square and black inverted triangle).

2.3 | Tertile analysis to enrich for highly neutralizing convalescent plasma units

Because the optimal NT50 that corresponds to functional immune protection is not known, we assigned each

A

B

C

FIGURE 1 Comparison of nonsevere COVID-19 convalescent plasma SARS-CoV-2 neutralization activity and antigen-binding assays. A, Dilution curves with example of nonlinear curve fitting showing R2 and NT50 calculation for five highly neutralizing units (open

symbols) and 3 units with low or undetectable neutralization activity (black, filled symbols). B, Correlation plots comparing all assays tested,

with linear curve fitting R2 value shown. Several samples with either high or low neutralization activity, as represented by specific symbols

in Figure 1A, are also represented by the same symbols in Figure 1B. C, Spearmanʼs correlation r values for pairwise comparisons of each

assay tested

4 GNIADEK ET AL.

convalescent plasma unit to a neutralization tertile and sought to determine how different donor and binding characteristics enrich for units found in the top two tertiles of neutralization activity. These included units with NT50 values of either 1 in 78 to 180 or more than 1 in 180. There was a significant difference in NT50 depending on donor age. Donors in the oldest tertile (age 48-75 years) had the highest enrichment for the top two neutralization tertiles (P = .04). However, no

other donor demographic including sex, fever, symp- tom duration, or reported COVID-19 symptoms corre- lated with NT50 (Figure 2A and S2). We observed no association between neutralization or binding and the time from either symptom onset or symptom end to sample collection (Figure S2). Finally, there was no statistically significant correlation between the number of symptoms reported and NT50 (Pearson correlation = 0.10, P = .48).

44

50

6

33

47

20

25

0

75

T1 T2 T3

Serologic Tertile

Beckman

44

44

13

40

40

20

19

13

69

T1 T2 T3

T3

T2

T1

N eu

tr al

iz at

io n

T er

til e

Serologic Tertile

S1 RBD ELISA

50

50

0

33

40

27

19

6

75

T1 T2 T3

Serologic Tertile

DiaSorin

31

56

13

53

27

20

19

13

69

T1 T2 T3 0

Serologic Tertile

20

40

60

80

Roche 100

T1 T2 T3 8

16

32

64

128

256

512

1024

2048

Serologic Tertile

N T5

0

S1RBD IgG ELISA p = 0.0107

T1 T2 T3

Serologic Tertile

Beckman p = 0.0059

T1 T2 T3

Serologic Tertile

Diasorin p = 0.0028

T1 T2 T3

Serologic Tertile

Roche p = 0.0076

19 -3

7 38

-4 7

48 -7

5 8

16

32

64

128

256

512

1024

2048

Age p = 0.0436

N T5

0

Ma le

Fe m

ale

Sex p = 0.6122

Fe ve

r

No F

ev er

Fever p = 0.1892

1- 7

8- 14

15 -2

8

Symptom Duration (Days) p = 0.7697

A

B

C

FIGURE 2 Enrichment for highly neutralizing nonsevere COVID-19 convalescent plasma by donor characteristics and serologic tertile. A, Dot plots demonstrating NT50 values for all units tested as categorized by key donor characteristics. Dotted horizontal lines

represent transition points between neutralization tertiles. B, Comparison of NT50 values between tertiles of binding activity for each

serologic assay tested. C, Heatmaps indicating the percentage of each binding tertile that is made up of units from each neutralization tertile.

Significance was assessed by Mann-Whitney U test for comparisons of two donor categories. Kruskal-Wallis test was used to compare groups

of three categories

GNIADEK ET AL. 5

Most highly neutralizing antibodies were in the top tertile of binding activity for each assay. A total of 69% of the units in the top tertile of binding activity for the S1 RBD ELISA assay were in the highest tertile of neutrali- zation activity and 13% in the middle neutralization ter- tile (82% were highly neutralizing;, see Figure 2B,C). Clinical serologic assays were similar in this regard, 75% to 82% of the units in the highest binding tertile were highly neutralizing. Only the S1 RBD ELISA contained no units with undetectable neutralization activity in the highest binding tertile (Figure 2B), whereas each clinical assay contained at least one nonneutralizing sample in the highest binding tertile.

2.4 | Using multiple assays to predict neutralization titer

Results from the three emergency use authorization– approved assays were combined by summing the assay- specific tertile rank for each sample (range 3-9; with 9 indicating the highest reactivity tertiles across all three assays). For samples with a score of 7 or higher, 15.8% were in the lowest NT50 tertile, 10.5% were in the middle NT50 tertile, and 63.1% were in the highest NT50 tertile.

3 | DISCUSSION

These findings illustrate the difficulty in deriving infor- mation about functional antibody responses using anti– SARS-CoV-2–binding assays. Although a correlation exists, the relatively high discordance rate in donors with a history of nonsevere COVID-19 may adversely affect the interpretation of convalescent plasma clinical trial data. Furthermore, the discordance observed implies that SARS-CoV-2 patients develop a broad antibody repertoire against multiple proteins and epitopes, each of which may only partially contribute to the overall neutralization of the virus.

Several larger studies that included hospitalized patients with more severe disease have shown higher COVID-19 antibody–binding reactivity in individuals who had been hospitalized or received treatment for COVID-19. Interestingly, COVID-19 antibody levels, as measured by binding assays, are high in hospitalized patients during their hospitalization, and formation of these antibodies is not related to a decline in viral load.2

Future studies are needed to understand the significance of these observations. It is possible that high-titer– binding antibodies with low neutralization potential lead to antibody-dependent enhancement, suggesting that

selecting convalescent plasma units based on high bind- ing activity alone may cause harm.14 Similarly, future studies are needed to determine if developing high-titer– binding antibodies with low neutralization potential is a poor prognostic sign.

It is possible that combining binding titer results from multiple assays may produce a better prediction of neu- tralization titer than any single assay alone. In this study of limited sample size and three different emergency use authorization–approved assays, there did not appear to be an increased predictive power by combining results. However, additional studies are warranted, since any inconsistency in convalescent plasma potency could obscure a signal of potential benefit in the setting of a clinical trial.

In this study, increasing age correlated with NT50, consistent with other reports,15 perhaps due to relatively more severe disease processes and greater viral exposure in this age group, although the same association was not seen with symptom duration or fever. However, given the heterogeneity of COVID-19 disease symptoms, it is unclear how well symptom history reflects the degree of immunologic stimulation.

Although the neutralizing antibody dose needed for clinical benefit is unknown, units in our top two neutrali- zation tertiles had activity consistent with current FDA recommendations (NT50 > 160 or > 80). Other factors such as the number of units transfused, the viral load at the time of transfusion, the degree of irreversible end- organ damage at the time of transfusion, and the extra- cellular fluid volume of the recipient may be critical fac- tors in determining whether a unit with a certain concentration of neutralizing antibody shows clinical benefit.

ACKNOWLEDGMENTS The authors thank Christine Ronayne for administrative and logistical laboratory support and members of the UMN BSL3 program.

CONFLICT OF INTEREST The authors declare no potential conflict of interest.

ORCID William E. Matchett https://orcid.org/0000-0001-7515- 2226 Tyler D. Bold https://orcid.org/0000-0001-7360-6546

REFERENCES 1. Bloch EM, Shoham S, Casadevall A, et al. Deployment of con-

valescent plasma for the prevention and treatment of COVID- 19. J Clin Investig. 2020;130:2757–2765.

6 GNIADEK ET AL.

2. Wölfel R, Corman VM, Guggemos W, et al. Virological assess- ment of hospitalized patients with COVID-2019. Nature. 2020; 581:465–469.

3. Krammer F, Simon V. Serology assays to manage COVID-19. Science. 2020;368:1060–1061.

4. Long QX, Liu BZ, Deng HJ, et al. Antibody responses to SARS- CoV-2 in patients with COVID-19. Nat Med. 2020;26:845–848.

5. Tenforde MW, Billig Rose E, Lindsell CJ, et al. Characteristics of adult outpatients and inpatients with COVID-19 - 11 Aca- demic Medical Centers, United States, March-May 2020. MMWR Morb Mortal Wkly Rep. 2020;69(26):841.

6. Crawford KHD, Eguia R, Dingens AS, et al. Protocol and reagents for pseudotyping lentiviral particles with SARS-CoV-2 spike protein for neutralization assays. Viruses. 2020;12(5):513.

7. Muruato AE, Fontes-Garfias CR, Ren P, et al. A high- throughput neutralizing antibody assay for COVID-19 diagno- sis and vaccine evaluation. Nature Communications. 2020;11 (1). http://dx.doi.org/10.1038/s41467-020-17892-0

8. Tan CW, Chia WN, Qin X, et al. A SARS-CoV-2 surrogate virus neutralization test based on antibody-mediated blockage of ACE2–spike protein–protein interaction. Nat Biotechnol. 2020; 38:1073–1078.

9. Du P. Comparisons of VLP-based ELISA, neutralization assays with native HPV, and neutralization assays with PsV in detecting HPV antibody responses in HIV-infected women. J AIDS Clin Res. 2015;6(3):433.

10. Xie X, Muruato A, Lokugamage KG, et al. An infectious cDNA clone of SARS-CoV-2. Cell Host Microbe. 2020;27: 841–848.e3.

11. Salazar E, Kuchipudi SV, Christensen PA, et al. Convalescent plasma anti-SARS-CoV-2 spike protein ectodomain and receptor binding domain IgG correlate with virus

neutralization. J Clin Investig. 2020; http://dx.doi.org/10. 1172/jci141206.

12. Thomas SN, Altawallbeh G, Zaun C, et al. Initial determination of COVID-19 seroprevalence among outpatients and healthcare workers in Minnesota using a novel SARS-CoV-2 total antibody ELISA. SSRN Electron J. 2020; [Preprint].

13. Stadlbauer D, Amanat F, Chromikova V, et al. A detailed pro- tocol for a serological assay, antigen production, and test setup. Curr Protoc Microbiol. 2020;57:e100.

14. Coish JM, MacNeil AJ. Out of the frying pan and into the fire? Due diligence warranted for ADE in COVID-19. Microbes Infect. 2020; http://dx.doi.org/10.1016/j.micinf.2020.06.006.

15. Wang X, Guo X, Xin Q, et al. Neutralizing antibody responses to severe acute respiratory syndrome coronavirus 2 in coronavi- rus disease 2019 inpatients and convalescent patients. Clin Infect Dis. 2020; http://dx.doi.org/10.1093/cid/ciaa721.

SUPPORTING INFORMATION Additional supporting information may be found online in the Supporting Information section at the end of this article.

How to cite this article: Gniadek TJ, Thiede JM, Matchett WE, et al. SARS-CoV-2 neutralization and serology testing of COVID-19 convalescent plasma from donors with nonsevere disease. Transfusion. 2020;1–7. https://doi.org/10.1111/trf. 16101

GNIADEK ET AL. 7