Platform for isolation and characterization of SARS-CoV-2 variants enables rapid characterization of Omicron in Australia

Abstract

Genetically distinct variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have emerged since the start of the COVID-19 pandemic. Over this period, we developed a rapid platform (R-20) for viral isolation and characterization using primary remnant diagnostic swabs. This, combined with quarantine testing and genomics surveillance, enabled the rapid isolation and characterization of all major SARS-CoV-2 variants circulating in Australia in 2021. Our platform facilitated viral variant isolation, rapid resolution of variant fitness using nasopharyngeal swabs and ranking of evasion of neutralizing antibodies. In late 2021, variant of concern Omicron (B1.1.529) emerged. Using our platform, we detected and characterized SARS-CoV-2 VOC Omicron. We show that Omicron effectively evades neutralization antibodies and has a different entry route that is TMPRSS2-independent. Our low-cost platform is available to all and can detect all variants of SARS-CoV-2 studied so far, with the main limitation being that our platform still requires appropriate biocontainment.

Fig. 1. Sensitive isolation of all SARS-CoV-2 VOC and key VUI over the period of February to May 2021.

a, Rapid end-point titres versus average Ct values (Ct values are averages of the Ct values across the viral genes detected (that is, primarily E, RdRP and N)) recorded using the HAT-24 cell line for rapid and sensitive isolation of SARS-CoV-2. Variant B.1.517.1 (open blue circles) is used herein for comparison, with samples derived from community spread in Australia from December 2020 to the end of January 2021 (n = 50). b–d, Rapid appearance of viral CPE (see large spherical viral syncytia) following overnight (c) versus 48 h (d) culture with a swab positive for the VOC Beta (b is mock uninfected control). CPE appearance is representative of all VOC. Scale bars in b–d, 10 μm. Source data

Fig. 2. Titration and neutralization of SARS-CoV-2 viral stocks in VeroE6 and HAT-24 cell lines.

Virus stocks were serially diluted in 5-fold steps and added to cells in octuplicate. Cell nuclei were enumerated with high-content microscopy and cell numbers normalized to mock-infected controls where 100% represents cell numbers for mock-infected controls and 0% represents cell numbers for the highest viral concentration. a, Dose-dependent loss of nuclei in VeroE6-TMPRSS2, VeroE6 and HAT-24 cells at 20 h post infection (hpi). b,c, Titrations of a panel of SARS-CoV-2 isolates including the ancestral virus strain, VOC and VUI. Readout occurred at 72 hpi for VeroE6 (b) and 20 hpi for HAT-24 (c) cells. d, Representative fluorescence images of HAT-24 cells stained with Hoechst-33342, showing progressive loss of nuclei with increasing virus concentrations. e, Enumeration of stained nuclei with high-content imaging platform (IN Carta, Cytiva) using image thresholding and segmentation algorithms. Scale bars for d and e, 50 μm. f,g, Visual scoring of CPE at 72 hpi was done to calculate TCID₅₀ ml⁻¹ values in VeroE6 (f) and HAT-24 (g) cells (coloured bars). LD₅₀ values at 20 hpi are also displayed (small bars with pattern). Shown are the mean ± s.d. from at least n = 3 experiments. h–p, Neutralization assays were performed in high-throughput format with both VeroE6 and HAT-24 cells using live virus isolates from the VOC: Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1) and Delta (B.1.617.2), as well as the VUI: Epsilon (B.1.429), Zeta (P2), Eta (B.1.525), Kappa (B.1.617.1), Lambda (C.37) and C.36. ‘Wildtype’ virus from the same clade containing the dominant D614G mutation (Clade B - B.1.319) and ancestral Wuhan-like virus with the original D614 background (Clade A - A.2.2) were also included. h,i, Neutralization assay of SARS-CoV-2 isolates with monoclonal AB-3467 in HAT-24 (h) and VeroE6 cells (i). j, Spearman correlation of IC₅₀ values in h and i (r = 0.9371, P < 0.0001, two-tailed). k,l, Neutralization assay of SARS-CoV-2 isolates by WHO control sample G in HAT-24 (k) and VeroE6 cells (l). m, Spearman correlation of IC₅₀ values in k and l (r = 0.8671, P = 0.0005, two-tailed). n,o, Neutralization assay of SARS-CoV-2 isolates by the Plasma Alliance control sample in HAT-24 (n) and VeroE6 (o) cells. p, Spearman correlation of IC₅₀ values in n and o (r = 0.7972, P = 0.0029, two-tailed). Shown are the mean ± s.d. of technical replicates done in quadruplicate. Each panel is representative of a minimum of three independent experiments. In h, i, k, l, n and o, dose-response curves and interpolated IC₅₀ values were determined using the sigmoidal 4PL model of regression analysis in GraphPad Prism software version 9.1.2. Source data

Fig. 3. Contemporary live SARS-CoV-2 variants and evasion of the vaccine and convalescent response.

a–f, Neutralization assays were performed in high-throughput format with both VeroE6 and HAT-24 cells using live virus isolates from the VOC: Alpha, Beta, Gamma and Delta, as well as the VUI: Epsilon, Zeta, Eta, Kappa, Lambda and C.36. ‘Wildtype’ virus from the same clade containing the dominant D614G mutation (Clade B - B.1.319) and ancestral Wuhan-like virus with the original D614 background (Clade A - A.2.2) were used as controls. a–d, ID₅₀ neutralization titres presented for 12 live variants for peak BNT162b2-vaccinated samples (a,c) (n = 24) and convalescent samples (b,d) for HAT-24 (a,b) and VeroE6 cells (c,d) (n = 23). e,f, Correlation of ID₅₀ for vaccine versus convalescent samples respectively for HAT-24 and VeroE6 cells. r = 0.8322 (P = 0.0013, two-tailed) for vaccine samples (e) and r = 0.6853 (P = 0.0170, two-tailed) for convalescent samples (f). g–j, Fold reduction in ID₅₀ values for data presented in a–d when compared to the earliest Clade A variant A.2.2. Shown are the mean ± s.d. for n = 24 donors in g and i, and n = 15 donors in h and j. Mean fold change for each variant is indicated at the top of the panel. Significance testing was done using Friedman’s test with Dunn’s multiple comparison. In g, P = 0.011 for Delta, P = 0.0001 for Epsilon and P < 0.0001 for Eta, Kappa and Lambda. In h, P < 0.0001 for Beta, P = 0.0015 for Zeta and P = 0.0105 for Lambda. In i, P = 0.0014 for Alpha, P = 0.0001 for Beta and P < 0.0001 for Kappa. In j, P = 0.0002 for Beta, P < 0.0001 for Kappa and P = 0.0230 for Lambda. Interpolated ID₅₀ values were determined using the sigmoidal 4PL model of regression analysis in GraphPad Prism software version 9.1.2. Source data

Fig. 4. Resolution of variant fitness in vivo by combining rapid and sensitive end-point viral titres with diagnostic PCR values.

a, Nasopharyngeal swabs were obtained from December 2020 to June 2021. In brief, samples were taken, placed into viral transport media and frozen at −80 °C in 100 µl aliquots within 24 h. Samples were filter-sterilized using 0.22 μm centrifugal filters and then co-cultured with the HAT-24 line. b,c, 20 h of culture: uninfected cells (b) versus infected cells (c) with a swab Ct <20. Note: CPE is scored through the rapid appearance of large spherical syncytia. d,e, 96 h of culture: uninfected confluent well (d) and low-level infection with CPE revealed as a combination of spherical syncytia and the formation of plaques (e). f,g, 96 h of culture: low-level infection with CPE revealed by extensive fusion across the viral cell sheet and the formation of plaques (f); extensive infection where CPE has resolved early as spherical syncytia and has become granular in appearance over time (g). h, Rapid end-point titres versus average Ct values as described in Fig. 1. Here the titres are scored within 20 h versus 96 h and correlated to the average Ct of all genes detected in diagnostic PCR values. All samples were collected as outlined in a and are from the Delta outbreak in Sydney in June 2021 (n = 82; r = −0.8585, P < 0.0001, two-tailed). i, Comparison of the early clade B.1.517.1 (n = 15) versus the VOC Delta (n = 80). Swabs for B.1.517.1 were collected as outlined in a but from December 2020 to January 2021. Spearman coefficient of correlation between end-point titres and Ct values, r = −0.8582 (P < 0.0001, two-tailed) and r = −0.8543 (P < 0.0001, two-tailed) for Delta and B.1.517.1, respectively. j, Correlation of Delta infectious titres versus age. r = 0.1048 (P = 0.3709, two-tailed). k, Titres derived from primary ALI cultures, 3 d and 7 d post infection using the R-20 platform. Shown are the mean ± s.d. from n = 3 experiments. Shaded areas in h and i represent 95% confidence intervals for each linear regression. Source data

Fig. 5. Rapid isolation and characterization of the SARS-CoV-2 Omicron variant.

a, Primary Omicron sample (Ct 22) was filter-sterilized using 0.22 μm centrifugal filters and then co-cultured with the HAT-24 line. Left: 48 h of culture. Right: 72 h of culture. As with other SARS-CoV-2 variants, extensive syncytia accumulated within the HAT-24 line. Scale bars, 200 μm. b, End-point titres of Omicron plotted against the diagnostic PCR Ct value as described in Fig. 4. Spearman coefficient of correlation between end-point titres and Ct values for Omicron, r = −0.7288 (P < 0.001, two-tailed). Delta and the early variant B.1.517.1 linear regressions are overlaid for comparison. Data are from the 96 h timepoint. c, Titration of the TMPRSS2 inhibitor Nafamostat using the HAT-24 cell line. Note: the lack of inhibition in Omicron versus Delta at all concentrations tested reveals limited usage of TMPRSS2 by Omicron. Shown are the mean ± s.d. of technical replicates done in quadruplicate. Each panel is representative of a minimum of three independent experiments. d, Potential alternative pathways of viral entry in Omicron versus other SARS-CoV-2 variants. All SARS-CoV-2 variants can enter via endocytosis (left pathway) or at the plasma membrane by TMPRSS2 cleavage. The change in tropism of Omicron either reflects viral fusion primarily at the endosome using Cathepsin L or fusion at the membrane using an alternate protease to TMPRSS2 (designated herein as ?). Source data

Fig. 6. Humoral neutralization of live SARS-CoV-2 variants in convalescent and vaccinated donors and concentrated human IgG plasma samples.

a–c, Neutralization assays were performed in a high-throughput format in HAT-24 cells using live virus isolates from the VOCs Delta (B.1.617.2), Gamma (P.1), Beta (B.1.351), Omicron (B.1.1.529) and the Clade A Wuhan-like virus, with the original D614 background (A.2.2) as a control. ID₅₀ neutralization titres are presented for five variants for convalescent donors from Clade A and B (B.1.319) (first wave) (a) (n = 10), Clade 20F (D614G + S477N) (second wave) (b) (n = 10) and third (VOC Delta) wave (c) (n = 10). d,e, ID₅₀ neutralization titres presented for 5 live variants for vaccinated donors from Clade A and B (D614G) (first wave) (d) (n = 10) and Clade 20F (D614G + S477N) (e) (second wave) (n = 9). F, Sera from healthy donors one month post third dose of BNT162b2 vaccine (n = 28). G, ID₅₀ neutralization titres across five live variants with concentrated polyclonal IgG from either convalescent and vaccinated donors or convalescent donors only. Data in a–g indicate the mean ID₅₀ of technical replicates for individual samples (circles) with the geometric mean and 95% confidence interval shown for each variant. Horizontal dotted lines in a–g represent the lowest serum dilution used and thus the limits of detection. Titres below the limit of detection are indicated with red diamonds. Fold-change reductions in ID₅₀ neutralization titres compare VOC to the ancestral variant, as well as Beta to Omicron. Significance testing was done using Kruskal Wallis test with Dunn’s multiple comparison test. In b, P = 0.0035 for ancestral versus Beta; in c, P = 0.0024, 0.0365 and <0.0001 for ancestral versus Beta, Gamma and Omicron, respectively; in d, P = 0.0028 and <0.0001 for ancestral versus Beta and Omicron, respectively; in e, P = 0.0022 and <0.0001 for ancestral versus Beta and Omicron, respectively; in f, P < 0.0001 for ancestral versus Beta and Omicron, respectively. h, Summary of neutralization reductions across Omicron and other VOC. Black dots represent the mean of all participants (except for Omicron where only groups of individuals with detectable neutralization against Omicron were included). Error bars are 95% CIs of the geometric mean. Unbounded error bars are present for groups where no detectable neutralization against Omicron was observed. i, Summary of fold reduction of neutralization potency across all variants tested using human polyclonal IgG derived from more than 20,000 US plasma donors (Poly_IgG-1033). This represents the collective activity of neutralizing antibodies derived from both convalescent and vaccinated plasma donors. j. Fold reduction of neutralization potency across all variants tested using human polyclonal IgG derived from COVID-19 convalescent donors’ plasma. Fold reduction values are representative of fold reduction from n = 3 independent experiments. Source data