Characterisation of the antibody-mediated selective pressure driving intra-host evolution of SARS-CoV-2 in prolonged infection

Abstract

Neutralising antibodies against the SARS-CoV-2 spike (S) protein are major determinants of protective immunity, though insufficient antibody responses may cause the emergence of escape mutants. We studied the humoral immune response causing intra-host evolution in a B-cell depleted, haemato-oncologic patient experiencing clinically severe, prolonged SARS-CoV-2 infection with a virus of lineage B.1.177.81. Following bamlanivimab treatment at an early stage of infection, the patient developed a bamlanivimab-resistant mutation, S:S494P. After five weeks of apparent genetic stability, the emergence of additional substitutions and deletions within the N-terminal domain (NTD) and the receptor binding domain (RBD) of S was observed. Notably, the composition and frequency of escape mutations changed in a short period with an unprecedented dynamic. The triple mutant S:Delta141-4 E484K S494P became dominant until virus elimination. Routine serology revealed no evidence of an antibody response in the patient. A detailed analysis of the variant-specific immune response by pseudotyped virus neutralisation test, surrogate virus neutralisation test, and immunoglobulin-capture enzyme immunoassay showed that the onset of an IgM-dominated antibody response coincided with the appearance of escape mutations. The formation of neutralising antibodies against S:Delta141-4 E484K S494P correlated with virus elimination. One year later, the patient experienced clinically mild re-infection with Omicron BA.1.18, which was treated with sotrovimab and resulted in an increase in Omicron-reactive antibodies. In conclusion, the onset of an IgM-dominated endogenous immune response in an immunocompromised patient coincided with the appearance of additional mutations in the NTD and RBD of S in a bamlanivimab-resistant virus. Although virus elimination was ultimately achieved, this humoral immune response escaped detection by routine diagnosis and created a situation temporarily favouring the rapid emergence of various antibody escape mutants with known epidemiological relevance.

Fig 1. Course of the first and second episode of SARS-CoV-2 infection.

(a) The figure displays the duration of hospitalisation following the initial diagnosis (day 1) (grey bar), the time point of the first vaccination (V), and the detection of SARS-CoV-2 RNA in nasopharyngeal swabs by qRT-PCR. Positive and negative PCR results are indicated by squares and crossed-out squares, respectively. The day of sampling is indicated on the x-axis. (b) The graph illustrates the time course of the administration and dosage of antiviral therapeutics received by the patient, with the timeline indicated on the x-axis. (c) The relative viral load in nasopharyngeal swabs during the first episode is indicated by Ct values of qRT-PCR; the timeline of sampling is delineated on the x-axis.

Fig 2. Intra-host evolution of SARS-CoV-2 during the first episode of infection.

The relative frequency of mutations in spike (% of reference sequence) is marked with a colour scale as indicated. The day of sampling after the initial diagnosis (day 1) and the days on which two consecutive samples were sequenced are indicated. GISAID accession numbers of the consensus sequences are provided in the S3 Table.

Fig 3. Detection of SARS-CoV-2 wild-type-reactive binding antibodies by routine serologic assays.

Peripheral blood samples obtained from the patient on the days indicated were tested by the quantitative Abbott Anti-RBD-CMIA (a) and the GenScript cPass sVNT (b). Antibody levels detected by Anti-RBD-CMIA and sVNT are given as arbitrary units (AU)/mL, cut-off ≥50 AU/mL, and signal reduction (% inhibition, cut-off >30%) as compared to the negative control. The GenScript cPass sVNT was performed with technical duplicates at a serum dilution of 1:20. The cut-off of the sVNT is indicated by the horizontal dotted line. (c) Qualitative results of the Mikrogen line blot testing for Anti-N IgG, as well as Anti-S1 and Anti-RBD IgM and IgG, are indicated as follows: open squares: non-reactive, grey squares: reactive below cut-off, black squares: reactive above cut-off. Administration of bamlanivimab (B), first vaccination (V) and therapy with sotrovimab (S) is indicated by vertical dotted lines.

Fig 4. Detection of variant-specific neutralising antibodies.

The reactivity of the patient’s sera with wt S, S:S494P, S:E484K S494P, S:Delta141-4 E484K S494P and S Omicron BA.1 was determined by pVNT. The mean inhibition of the GFP signal (% reduction as compared to the untreated control) was determined for four technical replicates at a serum dilution of 1:20, 1:80 and 1:320. Inhibition by ≥50% at a dilution of 1:20 was rated as a positive result. The half-maximal inhibitory activity of serum samples (IC50) was calculated using best-fit curves. The IC50 values are presented as reciprocal serum dilutions. IC50 values exceeding 1000 or below 1 are depicted as 1000 or 1, respectively. The interval between the first detection of additional mutations in the NTD and RBD in variant S:S494P on day 40 and virus elimination on day 56 is indicated by grey shading. The administration of bamlanivimab (B), the first vaccination (V) and sotrovimab (S) therapy are indicated by vertical dotted lines. The horizontal dotted line represents the cut-off value for the lowest serum dilution tested (1: 20). The inhibition of wt S and S variants at individual serum dilutions is depicted in the S1 Fig.

Fig 5. Detection of variant-specific inhibitory antibodies.

The reactivity of patient sera with wt RBD and RBD E340K (a), RBD S494P and RBD E484K S494P (b), and RBD Omicron BA.1 and RBD Omicron BA.1 E340K (c) was determined by sVNT using N-terminally secNLuc-tagged RBD antigens. Sera were tested at a dilution of 1:20. The mean values of two independent experiments are shown. The cut-off value of the sVNT for wt RBD was set to ≥25% inhibition (dotted line). Values below 0% were set to zero. The period between the first detection of additional mutations in the NTD and RBD in variant S:S494P on day 40 and virus elimination on day 56 is shaded in grey. The day of sampling after the initial diagnosis (day 1) is indicated on the x-axis. The administration of bamlanivimab (B), the first vaccination (V) and sotrovimab (S) therapy are indicated by vertical dotted lines.

Fig 6. Variant-specific serum reactivity in the Ig class capture RBD-EIA.

IgM reactivity (a), IgA reactivity (b), and IgG reactivity (c) of the patient’s sera. Antibody reactivity against wt RBD, RBD S494P, and RBD E484K S494P was determined from days 2 to 391, and reactivity against RBD Omicron BA.1 from days 131 to 391. Additionally, IgG reactivity with RBD E340K and RBD Omicron BA.1 E340K was tested from days 131 to 391 (c). Sera were analysed at a dilution of 1:100. Mean values of two independent experiments are shown. The cut-off was set at 300 rlu. Signal strength is given in relative light units (rlu). Values below 0 rlu were set to zero. The period between the first detection of additional mutations in the NTD and RBD in variant S:S494P on day 40 and virus elimination on day 56 is shaded in grey. The day of sampling after the initial diagnosis is indicated on the x-axis. The administration of bamlanivimab (B), the first vaccination (V) and sotrovimab (S) therapy are indicated by vertical dotted lines.

Fig 7. Variant-specific serum reactivity in the Ig class capture NTD-EIA.

(a) IgA reactivity, (b) IgG reactivity. Sera were tested at a dilution of 1:100. The means of two independent experiments are shown. Antibody reactivity is given in relative light units (rlu). Values below 0 rlu were set to zero. The cut-off was set at ≥ 100 rlu. The day of sampling after the initial diagnosis is indicated on the x-axis. The administration of bamlanivimab (B), the first vaccination (V) and sotrovimab (S) therapy are indicated by vertical dotted lines.