Neutralization of SARS-CoV-2 variants by convalescent and vaccinated serum.

We tested human sera from large, demographically balanced cohorts of BNT162b2 vaccine recipients (n=51) and COVID-19 patients (n=44) for neutralizing antibodies against SARS-CoV-2 variants B.1.1.7 and B.1.351. Although the effect is more pronounced in the vaccine cohort, both B.1.1.7 and B.1.351 show significantly reduced levels of neutralization by vaccinated and convalescent sera. Age is negatively correlated with neutralization in vaccinee, and levels of variant-specific RBD antibodies are proportional to neutralizing activities.

Since its emergence in Wuhan, China in late 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread worldwide, causing widespread illness and mortality from coronavirus 2019 disease (COVID-19).[1] Continued SARS-CoV-2 transmission has led to the emergence of variants of concern (VOCs) that show evidence of increased transmissibility or resistance to prior immunity.[2] By early 2021, three major VOCs were widely recognized: B.1.1.7, first identified in southeast England in November 2020[3]; B.1.351, identified in November in South Africa; and P.1, identified in December Brazil[3,4]. These VOCs were associated with increases in infections and hospitalizations in their countries of origin, and all have increased in frequency in other regions, suggesting a competitive fitness advantage over existing lineages.[5]

Though a relatively small number of nonsynonymous mutations and deletions distinguish VOCs from earlier lineages (Supplementary Table 1), many of these encode residues in the spike protein, which interacts with the SARS-CoV-2 cellular receptor, angiotensin-converting enzyme 2 (ACE2), via its receptor-binding domain (RBD)[6,7]. RBD mutations could potentially increase transmissibility by enhancing binding to ACE2, or promote immune escape by altering epitopes that are the primary target of potently neutralizing antibodies.[7] In fact, the most prominent mutation that appeared early in the pandemic and rose to near-fixation in new strains was a substitution at spike residue position 614 (D614G) which positions the RBD in a more accessible configuration and confers greater infectivity but also greater susceptibility to neutralizing antibodies.[8,9]

In addition to sharing D614G and a N501Y substitution which is associated with greater ACE2 affinity,[10] VOCs have acquired other spike mutations, some of which are associated with resistance to antibody neutralization. These include E484K and K417N/T, both of which arose independently in the B.1.351 and P.1 lineages.[11-13] Epidemiological reports suggest that natural immunity to earlier SARS-CoV-2 lineages may confer limited protection from reinfection by B.1.351 or P.1[3,14], and prior analyses using relatively small numbers of vaccinee sera against pseudotyped or chimeric viruses showed reduced neutralization of B.1.351 and P.1[13,15]. The purpose of this study was to use neutralization assays with clinical virus isolates to rigorously examine the potency of antibodies elicited by the BNT162b2 vaccine or natural infection against the broader antigenic RBD variability within the B.1.1.7 and B.1.351 variants.

The three COVID-19 vaccines authorized for emergency use by the U.S. Food and Drug Administration (BNT162b2 [Pfizer–BioNTech], mRNA-1273 [Moderna], and Ad26.COV2.S [Janssen]) elicit immunity using a spike protein antigen derived from early isolates such as USA_WA1/2020 (WA1)[16]. RBD-binding antibody levels in adults who had received two doses of the BNT162b2 mRNA vaccine were determined by ELISA using recombinant RBD from WA1 (RBD-WA1) and RBDs with substitutions possessed by B.1.1.7 (N501Y) and B.1.351 (N501Y, E484K, K417N) (Table S1). Compared to that of RBD-WA1, vaccinated patient sera had a geometric mean 50% effective concentration (EC50) which was 1.3-fold lower (P=0.0411) for RBD-B.1.1.7 and 1.4-fold lower (P=0.0047) for RBD-B.1.351 (Figure 1A). BNT162b2-elicited antibodies also displayed potent neutralizing activity against WA1 in a 50% focus reduction neutralization tests (FRNT50) (geometric mean titer (GMT) 1:393 +/− 2.5) but decreased neutralization of B.1.1.7 (GMT 1:149 +/− 2.4) and B.1.351 (GMT 1:45 +/− 2.3), representing 2.6-fold (P<0.0001) and 8.8-fold (P<0.0001) reductions, respectively (Figures 1B and S1). The positive correlation between serum EC50 and NT50 was consistent for each matched variant-RBD pair, indicating that variant-specific RBD-targeted antibody concentration is proportional to live virus neutralization capacity against each lineage (Figure 1C).

Figure 1.

Serum antibody levels of BNT162b2 vaccine recipients and potency of sera to neutralize SARS-CoV-2 variants.

A) Serum antibody levels (EC50) that recognize the spike RBD of the wild type USA-WA1/2020 (WA1), B.1.1.7, and B.1.351 variants are shown. The RBD-B.1.1.7 carries the N501Y mutation corresponding to the B.1.1.7 variant and the RBD-B.1.135 has K417N, E484K, and N501Y mutations that are present in the B.1.351 variant. B) Comparison of neutralization titers (FRNT50) between WA1, B.1.17 and B.1.351 for BNT162b2 vaccinee sera. C) Correlation of variant matched RBD-specific antibody levels and neutralization titers (FRNT50) of the AW1 virus and the two variants. D, E) Correlations between neutralization titers of the B.1.1.7 (D) and B.1.351 (E) variants with the WA1 virus. The dotted diagonal lines indicate identical neutralization, and the solid diagonal black lines indicate 10-fold differences in neutralization. F-H) Correlation between participant age and neutralization titer against WA1 (F), B.1.1.7 (G), and B.1.351 (H). I) Effect of age range on the neutralization potency among the BNT162b2 vaccine recipients.

There was individual variation in the relative neutralization of the different variants. Neutralizing titers for WA1 and B.1.1.7 were highly correlated at the individual level (Figure 1D). In contrast, WA1 and B.1.351 FRNT50 titers correlated weakly at the individual level, with some individual’s serum potently neutralizing WA1 while having FRNT50 for B.1.351 below the assay limit of detection (1:20) (Figure 1E).

Older adults make up the most vulnerable population to COVID-19 and therefore have been prioritized for vaccination[19]. We found similar age-dependent decline in FRNT50 titers against each lineage in our study (Figures 1F-H). These differences were highly significant for all three variants between subgroups of younger (20-50 y.o. n=25) and older (>50 y.o. n=25) adults in our cohort (Figure 1I). There was no correlation between gender and neutralization titers after vaccination.

In contrast to the spike-specific antibody repertoire raised by BNT162b2 vaccination, the antibody response to SARS-CoV-2 infection is more antigenically diverse[7]. Overall, RBD binding activities against all lineages were significantly lower in convalescent sera compared to vaccinee sera across all sample timepoints (1-301 days post-PCR positive) (Figures 1A and 2A and 2B). Moreover, there was no observable difference in convalescent serum EC50 between RBD-WA1, RBD-B.1.1.7, and RBD-B.1.351 (Figure 2B). In convalescent sera, there was also no clear correlation between variant-specific RBD binding and neutralization (Figure 2C). To better capture the reduced antibody levels, we modified our ELISA protocol to reduce the limit of detection to 1:200 (compared with 1:1600 for vaccinee ELISAs). Differences in FRNT50 titer against WA1 and the VOCs were similarly reduced overall compared to vaccinee sera (WA1, GMTs 1:52.1 +/− 4.3; B.1.1.7, 1:36.8 +/− 3.0; B.1.351, 28.8 +/− 2.3) but showed substantially less variability with a 1.8-fold drop (CI) for B.1.351 and a 1.4-fold drop (CI) for B.1.1.7 relative to WA1 (Figures 2C and S2). Many convalescent sera fell below the FRNT limit of detection (Figures 2D-E): for WA1, 43% of convalescent cohort sera failed to neutralize ≥50% of input virus at the lowest dilution, and this proportion was even greater for the VOCs (B.1.1.7, 54%; B.1.351, 64%).

Figure 2.

Neutralization of SARS-CoV-2 variants by convalescent serum.

A) Quantification of serum antibody levels (EC50) that recognize RBD protein corresponding to the wild type (WA1), B.1.1.7, and B.1.351 variants. B) Comparison of neutralization titers between WA1, B.1.17 and B.1.351 for convalescent sera. C) Relationship between convalescent antibody levels and neutralization (FRNT50) of the different virus strains. D, E) Correlations between convalescent serum neutralization titer of the B.1.1.7 (D) and B.1.351 (E) variants with the WA1 virus. The dotted diagonal lines indicate identical neutralization, and the solid diagonal black lines indicate 10-fold differences in neutralization. F) Correlation of convalescent neutralization titers with time after first positive PCR test in COVID-19 patients. There is no significant correlation between days post positive PCR test and FRNT50 for the viral strains tested.

An important question surrounding the increasing evidence of reinfection in convalescent individuals by VOC is whether the severity of disease during the initial exposure is related to protection, or lack thereof, from reinfection[3]. Neutralizing titer in the convalescent cohort showed no significant correlation with the time between first confirmatory positive PCR result and sample collection, indicating relatively stable FRNT50 values over timescales up to 301 days (Figure 2F). Additionally, no correlation was found between neutralizing titer for any lineage and patient age, sex, or hospitalization for COVID-19 (Figures S3 and S4).

In this study we provide evidence of reduced antibody-mediated immunity to newly emerging SARS-CoV-2 variants B.1.1.7 and B.1.351 after immunization with the Pfizer-BioNTech COVID-19 vaccine or following natural infection. Our study involves a relatively large cohort, provides data well-balanced for gender and age distribution, controls for time since vaccination, and directly compares early-type and two newly emerging SARS-CoV-2 variants of global concern. Critically, we use authentic clinical isolates that display the native antigenic landscape of the virus, an approach that provides the best possible examination of antibody activity against these viruses.

While it is likely that the resistance of some VOCs to neutralization is driven by accumulated mutations in the RBD and the rest of the spike protein, and there is evidence that high levels of RBD-binding antibodies is a meaningful correlate of protection from isogenic lineages[7,17], other features of host immunity may contribute to protection. Specifically, the neutralization titers seen in our convalescent subjects, while lower overall, have a smaller gap in neutralizing activity between WA1 and VOCs than in BNT162b2 vaccinees. This difference between convalescents and vaccinees suggests that SARS-CoV-2 infection may elicit more broadly cross-reactive and potentially cross-neutralizing antibodies, even with reduced affinity for mutant RBDs. This notion has a strong foundation in coronavirus research, as there is substantial cross-reactivity of anti-SARS-CoV spike antibodies with SARS-CoV-2 spike[18]. Indeed, risk of reinfection by VOCs may be driven by generally low serological responses in most COVID-19 patients, rather than the presence of RBD mutations that allow immune escape. Other arms of the adaptive immune response that we did not explore here, such as T cell immunity, could also contribute to cross-lineage immunity[19].

A particularly significant finding was the negative correlation between age and neutralizing antibody titer against VOCs in vaccinees, given that age is the predominant risk factor for severe COVID-19[20] and patients of advanced age stand to benefit the most from vaccination. Longitudinal studies of this and other cohorts could examine the durability of vaccine-induced immune responses, and should be designed to resolve the nature of antibody responses induced by vaccination or natural infection that may correlate with broad cross-neutralization. This will be particularly important for developing vaccines that will be effective in vulnerable populations, including those of advanced age, against future SARS CoV-2 variants.

Table 1.

Demographic characteristics of study participants

Convalescent serum donors
Characteristic	Hospitalized		Total
	Yes (n=17)	No (n=37)	(N=54)
Median age – yr (range)	56 (22-88)	54 (1-80)	56 (1-88)
Sex – no. (%)
Female	11 (35.4)	20 (64.5)	31 (57.4)
Male	6 (26.1)	17 (73.9)	23 (42.6)
Symptomatic – no. (%)
No	3 (17.6)	1 (2.70)	4 (7.40)
Yes	14 (82.4)	36 (97.3)	50 (92.6)
Admitted to ICU – no. (%)
Yes	5 (29.4)	NA	5 (9.25)
No	12 (70.6)	NA	12 (22.2)
Median time between first positive COVID-19 PCR test and sample collection – days (range)	21 (1-217)	197 (22-302)	188.5 (1-302)
BNT162b2-vaccinated donors
Characteristic		Total (N=51)
Median age – yr (range)		50 (21-82)
Sex – no. (%)
Female		28 (54.9)
Male		23 (45.1)
Median time between vaccine doses – days (range)		21 (20-22)
Median time between second dose and sample collection – days (range)		14 (14-15)