Main

The continuous emergence of new variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused successive global waves of infection. Since its first report in November 2021 in South Africa, Omicron has become the dominating variant due to its high transmissibility and immune evasion1,2, with many Omicron sublineages emerging over time. The initial Omicron BA.1 was displaced by BA.2, which has further evolved to sublineages BA.2.12.1, BA.2.75, BA.2.75.2, BA.4 and BA.5, among which BA.5 is currently dominant in many countries. BA.4 and BA.5 have an identical spike sequence (defined as BA.4/5 hereafter), and their offspring, BA.4.6, BF.7 and BQ.1.1, are expanding in prevalence. As of 19 November 2022, the BA.2-derived sublineage BA.2.75.2 accounted for 0.8% of the total SARS-CoV-2 infection in the United States, whereas the BA.4/5-derived sublineages BA.4.6, BF.7, BQ.1 and BQ.1.1 accounted for 4.4%, 7.8%, 25.5% and 24.2% of total cases, respectively. In addition, another BA.5-derived sublineage, XBB, first identified in India in August 2022, is rapidly spreading in Europe and has been detected in the United States. XBB was predominant in Singapore, accounting for 54% of SARS-CoV-2 infections during the week of 3–9 October 2022 (Ministry of Health, Singapore; https://www.moh.gov.sg/).

SARS-CoV-2 spike mutations often contribute to immune evasion and/or transmission efficiency3,4,5,6,7,8,9. Previous studies showed that three or four doses of parental mRNA vaccine did not elicit robust neutralization against BA.4/5, supporting the development of bivalent vaccines that target both the ancestral spike and the BA.4/5 spike protein10,11,12. Because the newly emerged Omicron sublineages have accumulated additional spike mutations (Fig. 1a), it is important to examine the vaccine-elicited neutralization against these new sublineages. The goal of this study was to compare the neutralizing activities against six newly emerged Omicron sublineages (BA.5, BF.7, BA.4.6, BA.2.75.2, BQ.1.1 and XBB.1) using human sera collected from individuals who received four doses of parental mRNA vaccine or a BA.5 bivalent booster after 2–4 doses of parental mRNA vaccine.

Fig. 1: Neutralization against Omicron sublineages.
figure 1

a, Construction of Omicron sublineage-spike mNG SARS-CoV-2. Amino acid mutations, deletions (Δ) and insertions (Ins) are indicated in reference to the USA-WA1/2020-spike. L, leader sequence; ORF, open reading frame; NTD, N-terminal domain of S1; RBD, receptor binding domain; S, spike glycoprotein; S1, N-terminal furin cleavage fragment of S; S2, C-terminal furin cleavage fragment of S; E, envelope protein; M, membrane protein; N, nucleocapsid; UTR, untranslated region. b, FFRNT50 values of human sera after dose 4 parental mRNA vaccine. The P values (two-sided) for group comparison of GMTs are the following. USA-WA1/2020 versus others: <0.0001; BA.4/5-spike versus BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: 0.029, 0.001, <0.0001, <0.0001 and <0.0001; BF.7-spike versus BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: 0.103, <0.0001, <0.0001 and <0.0001; BA.4.6-spike versus BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: 0.0001, <0.0001 and <0.0001; BA.2.75.2-spike versus BQ.1.1-spike and XBB.1-spike: 0.24 and <0.0001; BQ.1.1-spike versus XBB.1-spike: 0.0028. The FFRNT50 values against BA.4/5-spike were reported previously11. c, FFRNT50 of 29 sera collected after BA.5 bivalent booster from individuals without infection history. The P values (two-sided) for group comparison of GMTs are the following. USA-WA1/2020 versus others: <0.0001; BA.4/5-spike versus BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: 0.844, <0.0001, <0.0001, <0.0001 and <0.0001; BF.7-spike versus BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: all <0.0001; BA.4.6-spike versus BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: all <0.0001; BA.2.75.2-spike versus BQ.1.1-spike and XBB.1-spike: 0.69 and <0.0001; BQ.1.1-spike versus XBB.1-spike: <0.0001. d, FFRNT50 of 23 sera collected after BA.5 bivalent booster from individuals with infection history. The P values (two-sided) for group comparison of GMTs are the following. USA-WA1/2020 versus others: <0.0001; BA.4/5-spike versus BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: 0.0049, <0.0001, <0.0001, <0.0001 and <0.0001; BF.7-spike versus BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: all <0.0001; BA.4.6-spike versus BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike: 0.0005, <0.0001 and <0.0001; BA.2.75.2-spike versus BQ.1.1-spike and XBB.1-spike: 0.114 and <0.0001; BQ.1.1-spike versus XBB.1-spike: <0.0001. For bd, bar heights and the numbers above indicate GMTs. Error bars indicate 95% confidence interval. The fold of GMT reduction against each Omicron sublineage, compared with the GMT against USA-WA1/2020, is shown in italic font. The dotted line indicates the limit of detection of FFRNT50. Statistical analyses were performed using the Wilcoxon matched-pairs signed-rank test for group comparison of GMTs.

Source data

To facilitate neutralization measurement, we engineered the complete spike gene from Omicron sublineage BA.4/5, BF.7, BA.4.6, BA.2.75.2, BQ.1.1 or XBB.1 into the backbone of mNeonGreen (mNG) reporter USA-WA1/2020 SARS-CoV-2 (Fig. 1a). Compared with wild-type USA-WA1/2020 (a strain isolated in January 2020), insertion of mNG gene at open reading frame 7 of the viral genome attenuated the virus in vivo13. So, the engineered live-attenuated mNG viruses can be used safely in a Biosafety Level 3 (BSL-3) facility with the correct procedures for neutralization and antiviral testing14. Passage 1 of recombinant BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike mNG USA-WA1/2020 viruses were sequenced to ensure no undesired mutations. Only the passage 1 virus stocks were used to determine the 50% fluorescent focus-reduction neutralization titers (FFRNT50) of vaccinated human sera, to ensure no additional spike mutations in the tested recombinant viruses.

Three human serum panels with distinct vaccination and/or SARS-CoV-2 infection history were analyzed. The first panel consisted of 25 sera obtained from individuals 23–94 days (median 47 days) after dose 4 of parental monovalent mRNA-1273 or BNT162b2 vaccine (post-dose-4 sera); these sera were collected from 16 March to 30 June 2022 (Extended Data Table 1). The second panel consisted of 29 sera collected from individuals 14–32 days (median 22 days) after BA.5 bivalent booster (BA.5 bivalent booster sera); these specimens were collected from 30 September to 22 October 2022 (Extended Data Table 2). All sera from the first and second panels tested negative against viral nucleocapsid protein (Extended Data Fig. 1), suggesting no previous or recent SARS-CoV-2 infection. The third panel consisted of 23 sera collected from individuals who were previously infected by SARS-CoV-2 (nucleocapsid antibody positive; Extended Data Fig. 1) and received a BA.5 bivalent booster 14–32 days (median 21 days) ago (BA.5 bivalent booster infection sera); the viral infection time and genotype could not be determined because most infections were asymptomatic; these samples were collected from 4 to 22 October 2022 (Extended Data Table 3). All participants from the second and third panels had also received two, three or four doses of parental monovalent mRNA vaccine before receiving the BA.5 bivalent booster. Extended Data Tables 13 summarize the serum information and neutralization for each serum panel.

Post-dose-4 sera neutralized USA-WA1/2020-spike, BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike mNG SARS-CoV-2 with geometric mean titers (GMTs) of 1,533, 95, 69, 62, 26, 22 and 15, respectively (Fig. 1b and Extended Data Table 1). The neutralizing GMTs against BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike viruses were 16.1-fold, 22.2-fold, 24.7-fold, 59-fold, 69.7-fold and 102-fold lower than the GMT against the USA-WA1/2020-spike virus, respectively (Fig. 1b). Compared with the GMT against the current dominant BA.4/5, the neutralizing GMTs against BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike viruses were reduced by 1.4-fold, 1.5-fold, 3.7-fold, 4.3-fold and 6.3-fold, respectively. The GMTs against BA.2.75.2 (26) and BQ.1.1 (22) were barely above 20, the detection limit of FFRNT, whereas the GMT against XBB.1 (15) was below the FFRNT detection limit. These results indicate that (1) four doses of parental mRNA vaccine do not elicit robust neutralization against the newly emerged Omicron sublineages when measured at 23–94 days (median 47 days) after dose 4, and (2) the rank of neutralization evasion is in the order of BA.4/5 < BF.7 ≤ BA.4.6 < BA.2.75.2 ≤ BQ.1.1 < XBB.1.

BA.5 bivalent booster sera, collected at 14–32 days (median 22 days) after boost, neutralized USA-WA1/2020-spike, BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike SARS-CoV-2s with GMTs of 3,620, 298, 305, 183, 98, 73 and 35, respectively (Fig. 1c and Extended Data Table 2). The neutralizing GMTs against BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike viruses were 12.1-fold, 11.9-fold, 19.8-fold, 36.9-fold, 49.6-fold and 103-fold lower than the GMT against the USA-WA1/2020, respectively (Fig. 1c). The data indicate that, although BA.5 bivalent booster elicits high neutralizing titers against BA.4/5 measured at 14–32 days after boost, the neutralization against BA.2.75.2 (98), BQ.1.1 (73) and XBB.1 (35) remains low after BA.5 bivalent booster.

BA.5 bivalent booster infection sera, collected at 14–32 days (median 21 days) after boost, neutralized USA-WA1/2020-spike, BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike SARS-CoV-2s with GMTs of 5,776, 1,558, 1,223, 744, 367, 267 and 103, respectively (Fig. 1d and Extended Data Table 3). The neutralizing GMTs against BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike viruses were 3.7-fold, 4.7-fold, 7.8-fold, 15.7-fold, 21.6-fold and 56.1-fold lower than the GMT against the USA-WA1/2020-spike SARS-CoV-2, respectively (Fig. 1d). Compared with BA.5 bivalent booster sera without infection history, BA.5 bivalent booster infection sera increased the neutralizing GMTs against USA-WA1/2020-spike, BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike viruses by 1.6-fold, 5.2-fold, 4.0-fold, 4.1-fold, 3.7-fold, 3.7-fold and 2.9-fold, respectively (compare Fig. 1c,d). The results suggest that (1) previous infection substantially increases the magnitude and breadth of neutralization for BA.5 bivalent booster, and (2) among the tested Omicron sublineages, XBB.1 exhibits the highest level of immune evasion.

Collectively, our neutralization results support two conclusions. First, the newly emerged Omicron sublineages continue to increase their immune evasion of vaccine-elicited and/or infection-elicited neutralization. Among tested Omicron sublineages, BA.2.75.2, BQ.1.1 and XBB.1 exhibit the greatest evasion against vaccine-elicited neutralization, suggesting the potential of these new sublineages to dethrone BA.5 as the dominant lineage in circulation. Second, individuals with SARS-CoV-2 infection history develop higher and broader neutralization against the current circulating Omicron sublineages after the BA.5 bivalent booster.

This study has several limitations. First, we have not examined the antiviral roles of non-neutralizing antibodies and cell-mediated immunity. These two immune components, together with neutralizing antibodies, protect patients from severe disease and death15,16. Unlike neutralizing antibodies, many T cell epitopes after vaccination or natural infection are preserved in Omicron spikes17. However, robust antibody neutralization is critical to prevent viral infection18. Second, we have not defined the spike mutations that contribute to the observed immune evasion of the newly emerged Omicron sublineages. Spike mutation F486V was previously shown to drive the immune evasion of BA.4/5 (ref.10). The new Omicron sublineages BA.2.75.2, BA.4.6, BF.7, BQ.1.1 and XBB.1 share the spike R346T mutation that was reported to confer higher neutralization evasion19. Third, the current results do not allow a direct comparison of neutralization between parental mRNA vaccine and BA.5 bivalent booster because of the differences in individuals’ demographics (for example, age), numbers of vaccine doses and serum collection time. Fourth, we do not know (1) how neutralizing titers related to protection against infection, severe disease or death; (2) when and which variants infected individuals from the BA.5 bivalent booster infection cohort; and (3) the insight related to the differences in vaccine dose for Moderna’s bivalent (Original and Omicron BA.4/BA.5) versus Pfizer/BioNTech’s BA.4/BA.5 adapted bivalent booster, and (4) the baseline of the neutralization titers before boost were not determined due to sample unavailability.

Our laboratory investigation, along with the recent real-world effectiveness of BA.5 bivalent booster20, supports a vaccine update strategy that future boosters should match new circulating SARS-CoV-2 variants. Given the advantage of mRNA vaccine platform that can rapidly adapt to new antigen sequences, the key challenge is to determine the future booster sequence before new variants become prevalent in circulation.

Methods

Ethics statement

All virus work was performed in a BSL-3 laboratory with redundant fans in the biosafety cabinets at The University of Texas Medical Branch (UTMB) at Galveston. All personnel wore powered air-purifying respirators (Breathe Easy, 3M) with Tyvek suits, aprons, booties and double gloves. The research protocol regarding the use of human serum specimens was reviewed and approved by the UTMB institutional review board (IRB) (no. 20-0070). No informed consent was required because these de-identified sera were leftover specimens from routine standard of care and diagnostics before being discarded. No diagnosis or treatment was involved either. The use of human serum specimens in this study was reviewed and approved by the UTMB IRB.

Cells

Vero E6 (American Type Culture Collection (ATCC), CRL-1586) cells were purchased from ATCC, and Vero E6 cells expressing TMPRSS2 (JCRB1819) purchased from SEKISUI XenoTech were maintained in a high-glucose DMEM containing 10% FBS (HyClone Laboratories) and 1% penicillin–streptomycin at 37 °C with 5% CO2. Culture media and antibiotics were purchased from Thermo Fisher Scientific. Both cell lines were tested negative for Mycoplasma.

Human serum

Three panels of human sera collected at UTMB were used in the study. Samples were collected based on availability. Varied ages with both genders are included. The population contains varied races or ethnicity, including White, Hispanic, Black and Asian. Individuals have received at least two doses of the Coronavirus Disease 2019 (COVID-19) vaccine with or without evidence of SARS-CoV-2 infection. The first panel consisted of 25 sera collected from individuals 23–94 days (median 47 days) after receiving dose 4 of parental vaccine mRNA-1273 or BNT162b2. This panel had been tested negative for SARS-CoV-2 nucleocapsid protein expression using Bio-Plex Pro Human IgG SARS-CoV-2 N/RBD/S1/S2 4-Plex Panel (Bio-Rad). The second panel consisted of 29 sera collected from individuals 14–32 days (median 22 days) after BA.5 bivalent booster from Pfizer (BA.4/BA.5 adapted bivalent booster) or Moderna (bivalent booster). All sera from this panel were tested negative for antibodies against SARS-CoV-2 nucleocapsid protein. The third panel consisted of 23 sera from individuals who were previously infected by SARS-CoV-2 (as determined by SARS-CoV-2 nucleocapsid ELISA), who were vaccinated with 2–4 doses of parental mRNA vaccine and who received a BA.5 bivalent booster 14–32 days (median 21 days) before serum collection. The genotypes of the infecting SARS-CoV-2 variants could not be determined for the third serum panel. Patient information was completely de-identified from all specimens. No informed consent was required because these de-identified sera were leftover specimens from standard of care and diagnostics before being discarded. The use of human sera for this study was reviewed and approved by the UTMB IRB (no. 20-0070). The de-identified human sera were heat-inactivated at 56 °C for 30 minutes before the neutralization test. The serum information is presented in Extended Data Tables 13.

Generation of recombinant Omicron sublineages mNG SARS CoV-2

Recombinant Omicron sublineage BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike mNG SARS-CoV-2s was constructed by engineering the complete spike gene from the indicated variants into an infectious cDNA clone of mNG USA-WA1/2020, as reported previously21,22. Spike sequences were based on BA.4/5 (BA.4: GISAID EPI_ISL_11 542270; BA.5: GISAID EPI_ISL_11542604; BA.4 and BA.5 have the identical spike sequence), BA.4.6 (GISAID EPI_ISL_15380489), BA.2.75.2 (GISAID EPI_ISL_14458978), BF.7 (GISAID EPI_ISL_14425795), BQ.1.1 (GISAID EPI_ISL_15542649) and XBB.1 (GISAID EPI_ISL_15232105). The full-length infectious cDNA clone of SARS-CoV-2 was assembled by in vitro ligation followed by in vitro transcription to synthesize the viral genomic RNA. The full-length RNA transcripts were electroporated in Vero E6-TMPRSS2 cells to recover the viruses. Viruses were rescued 2–3 days after electroporation and served as P0 stock. P0 stock was further passaged once on Vero E6 cells to produce P1 stock. The reason for using Vero E6 cells (rather than using Vero E6-TMPRSS2) to prepare the P1 virus is that the infectivity of the P1 virus can be affected by the cell types; because our established FFRNT assay uses Vero E6 cells, we chose to prepare the P1 viruses using the same Vero E6 cells. The spike gene was sequenced from all P1 stock viruses to ensure no undesired mutation. The infectious titer of the P1 virus was quantified by fluorescent focus assay on Vero E6 cells. The P1 virus was used for the neutralization test. The protocols for the mutagenesis of mNG SARS-CoV-2 and virus production were reported previously11. All virus preparation and neutralization assays were carried out at the BSL-3 facility at UTMB at Galveston.

FFRNT

Neutralization titers of human sera were measured by FFRNT using the USA-WA1/2020-spike, BA.4/5-spike, BF.7-spike, BA.4.6-spike, BA.2.75.2-spike, BQ.1.1-spike and XBB.1-spike mNG SARS-CoV-2s at BSL-3. The details of the FFRNT protocol were reported previously11. In brief, 2.5 × 104 Vero E6 cells per well were seeded in 96-well plates (Greiner Bio-One). The cells were incubated overnight. On the next day, each serum was two-fold serially diluted in the culture medium with the first dilution of 1:20 (final dilution range of 1:20 to 1:20,480). The diluted serum was incubated with 100–150 FFUs of mNG SARS-CoV-2 at 37 °C for 1 hour, after which the serum virus mixtures were loaded onto the pre-seeded Vero E6 cell monolayer in 96-well plates. After 1-hour infection, the inoculum was removed, and 100 μl of overlay medium (supplemented with 0.8% methylcellulose) was added to each well. After incubating the plates at 37 °C for 16 hours, raw images of mNG foci were acquired using Cytation 7 (BioTek) armed with a ×2.5 FL Zeiss objective with a wide field of view and processed using the Gen 5 software settings (GFP (469,525) threshold 4,000 and object selection size 50–1,000 μm). The foci in each well were counted using the Gen5 software and normalized to the non-serum-treated controls to calculate the relative infectivities. The FFRNT50 value was defined as the minimal serum dilution that suppressed >50% of fluorescent foci. The neutralization titer of each serum was determined in duplicate assays, and the geometric mean was taken. Extended Data Tables 1–3 summarize the FFRNT50 results. Data were initially plotted in GraphPad Prism 9 software and assembled in Adobe Illustrator. FFRNT50 of <20 was treated as 10 for plot purposes and statistical analysis. The above FFRNT50 protocol has been reliably used to support the clinical development of COVID-19 vaccines23. Thus, we applied the same FFRNT protocol to the current study.

Statistics and reproducibility

No statistical method was used to predetermine the sample size. The samples were collected based on availability. No data were excluded from the analyses. The experiments were not randomized. Patient information was blinded in the study. The investigators were blinded to sample identity during data collection and/or analysis. The experiments were performed in duplication. All attempts at replication were successful.

Continuous variables were summarized as the geometric mean with 95% confidence intervals or median. Sera with undetectable (<20) antibody titers were assigned an antibody titer of 10, for purposes of GMT calculations or statistical comparisons. Comparison between neutralization titers was performed using a Wilcoxon matched-pairs signed-rank test using GraphPad Prism 9.0. Absolute P values were provided. P < 0.05 was considered statistically significant. Images were assembled using Adobe Illustrator.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.