Regulation of virus neutralization and the persistent fraction by TRIM21

Abstract

Despite a central role in immunity, antibody neutralization of virus infection is poorly understood. Here we show how the neutralization and persistence of adenovirus type 5, a prevalent nonenveloped human virus, are dependent upon the intracellular antibody receptor TRIM21. Cells with insufficient amounts of TRIM21 are readily infected, even at saturating concentrations of neutralizing antibody. Conversely, high TRIM21 expression levels decrease the persistent fraction of the infecting virus and allows neutralization by as few as 1.6 antibody molecules per virus. The direct interaction between TRIM21 and neutralizing antibody is essential, as single-point mutations within the TRIM21-binding site in the Fc region of a potently neutralizing antibody impair neutralization. However, infection at high multiplicity can saturate TRIM21 and overcome neutralization. These results provide insight into the mechanism and importance of a newly discovered, effector-driven process of antibody neutralization of nonenveloped viruses.

Fig 1

ADIN is essential for efficient neutralization of adenovirus. (A) Antihexon monoclonal antibody 9C12 was titrated against adenovirus on clonal fibroblast (MEF) lines derived from wild-type (WT) (n = 8) (circles) or TRIM21 knockout (K21) (n = 8) (squares) mice. Cells were either untreated (closed symbols) or treated with IFN-α (open symbols). Data represent mean remaining infectivities (I/I₀) from the addition of antibody versus a PBS-treated control infection ± standard errors of the means. (B) TRIM21 mRNA levels were quantified by qPCR in unstimulated (white bars) or IFN-α-stimulated (hatched bars) MEF cells. n.d., not detected. (C) Levels of the transgene that disrupts the TRIM21 locus in K21 cells were measured by qPCR. Shading is as described above for panel B. (D) The previously solved structure of the TRIM21-IgG Fc complex (14) showing interactions between TRIM21 (yellow) (residues labeled TR) and the IgG Fc HNH motif (blue) (residues labeled Fc). (E) Antihexon monoclonal antibody 9C12 was mutated at three positions or expressed as the wild type. Antibodies were titrated against adenovirus on HeLa cells. Open circles, WT 9C12; closed circles, H433A mutant; open squares, N434A mutant; closed squares, N434D mutant; triangles, H435A mutant; diamonds, K219R mutant.

Fig 2

ADIN mediates neutralization by few antibody molecules. (A) 9C12-adenovirus complexes were immunogold labeled and examined by electron microscopy. Gold particles are visible as black dots around the AdV capsid and show 0, 1, 2, and 3 regions labeled, respectively (from left to right). Scale bar, 50 nm. (B) Virions were scored for the number of gold particles per virus at increasing concentrations of 9C12. The distribution of gold particles per virus (black bars) was found to approximate a Poisson distribution (black lines), given the mean number of gold particles per virus (red dashed line). These data were used to calculate the average number of NAb molecules per virion. (C) PBS (circles) or 9C12 at 1.5 μg/ml (squares) was added to AdV for 1 h before 100-fold dilution into complete DMEM and incubation for various amounts of time before addition to cells. No increase in titer was observed for NAb-labeled virus over time, confirming that complexes do not dissociate upon dilution. (D and E) The infectivity of diluted NAb-virus complexes was assayed with MEF (D) and HeLa (E) cells treated with IFN-α, permitting a direct quantification of the average number of NAb molecules per virion at 1/e neutralization. Data represent the natural logarithm of remaining infectivity, ln(I/I₀). (F) Stoichiometry of 9C12-A488–adenovirus after pelleting of complexes. Virus and antibody concentrations were quantified by qPCR and fluorescence spectroscopy, respectively. Data are fitted to a Michaelis-Menten curve (R² = 0.94), giving a K_d of 28 nM and a maximum binding of 205 antibody molecules per virus (standard error of the mean, ±10). Data are represented as means ± standard errors of the means. A parallel infection assay with IFN-treated HeLa cells estimated the number of antibody molecules required for 1 natural logarithm of neutralization (λ) to be 5.2.

Fig 3

TRIM21 levels determine neutralization efficiency. (A) IFN-α was added to HeLa cells (circles) or HeLa cells transduced with TRIM21 shRNA (squares), and TRIM21 mRNA levels were quantified by RT-qPCR. (B) TRIM21 protein levels were quantified by immunoblotting and normalized to β-actin levels. (C) Relative TRIM21 mRNA and protein levels are closely correlated (P < 0.001 by Pearson's product-moment correlation; R² = 0.83). (D) Virus neutralization assays were performed on these cells, which display an initial gradient of neutralization that varies with TRIM21 levels. (E) The number of antibody molecules per virus at each concentration of 9C12 was calculated by using binding parameters obtained as described above (Fig. 2F). Linear regression was used in order to calculate the number of molecules of 9C12 bound at 1/e neutralization (λ). λ was found to decrease exponentially with the TRIM21 concentration, reaching a plateau at 4.2 antibody molecules per virus (dashed line). For panels D and E, circles indicate HeLa cells, and squares indicate HeLa cells expressing TRIM21-directed shRNA. Points for panels A and C to E are shaded white to black according to IFN levels.

Fig 4

Infection and persistence are influenced by the neutralization mechanism and viral MOI. (A) Neutralization data from Fig. 3D are shown with the x axis extended to show the persistent fraction range. The PF remains approximately constant within the range of 1.5 to 100 μg/ml 9C12, but its level is determined by the cellular TRIM21 concentration. (B) The average PF, calculated from all data at ≥5 μg/ml 9C12 (n = 7) (±standard errors of the means), decreases with TRIM21 levels before remaining constant at ∼6% remaining infectivity. Symbols and shading as for Fig. 3D (C) Remaining infectivity of adenovirus, assayed alongside a fluorescence pelleting assay, permitting quantification of the number of antibody molecules per virus in the persistent fraction. (D and E) Two polyclonal sera, pAb1 (D) and pAb2 (E), were titrated against AdV on HeLa cells (circles) or HeLa cells treated with siRNA directed toward TRIM21 (squares), which were either IFN stimulated (open symbols) or untreated (closed symbols). (F) Remaining infectivity after AdV was permitted to adsorb to the surface of HeLa cells before the addition of neutralizing monoclonal antibody 9C12. Symbols are as described above for panels D and E. (G) AdV-GFP, preincubated with 9C12, was added to HeLa (circles) or TRIM21 shRNA knockdown (squares) cells at a range of MOIs. A high MOI of AdV-GFP relieved the neutralization of 9C12-labeled AdV-RFP in HeLa cells but to a lesser extent in TRIM21 knockdown cells. Michaelis-Menten curve fitting (black lines) reveals that the maximal saturation values are significantly different between the two conditions (P = 0.0008 by F test).