Influence of AAV vector tropism on long-term expression and Fc-γ receptor binding of an antibody targeting SARS-CoV-2

Abstract

Long-acting passive immunization strategies are needed to protect immunosuppressed vulnerable groups from infectious diseases. To further explore this concept for COVID-19, we constructed Adeno-associated viral (AAV) vectors encoding the human variable regions of the SARS-CoV-2 neutralizing antibody, TRES6, fused to murine constant regions. An optimized vector construct was packaged in hepatotropic (AAV8) or myotropic (AAVMYO) AAV capsids and injected intravenously into syngeneic TRIANNI-mice. The highest TRES6 serum concentrations (511 µg/ml) were detected 24 weeks after injection of the myotropic vector particles and mean TRES6 serum concentrations remained above 100 µg/ml for at least one year. Anti-drug antibodies or TRES6-specific T cells were not detectable. After injection of the AAV8 particles, vector mRNA was detected in the liver, while the AAVMYO particles led to high vector mRNA levels in the heart and skeletal muscle. The analysis of the Fc-glycosylation pattern of the TRES6 serum antibodies revealed critical differences between the capsids that coincided with different binding activities to murine Fc-γ-receptors. Concomitantly, the vector-based immune prophylaxis led to protection against SARS-CoV-2 infection in K18-hACE2 mice. High and long-lasting expression levels, absence of anti-drug antibodies and favourable Fc-γ-receptor binding activities warrant further exploration of myotropic AAV vector-based delivery of antibodies and other biologicals.

Fig. 1. Functional elements of AAV-TRES6 vector constructs.

Depicted are schemes of the AAV-TRES6hu SGSG/GSGS, the AAV-TRES6hu LCV, and the AAV-TRES6 vector constructs. ITR inverted terminal repeat, VH variable region of the immunoglobulin heavy chain of the TRES6 antibody, CH-huIgG1 constant region of human IgG1 heavy chain, CH-muIgG2c constant region of the murine IgG2c heavy chain, SGSG/GSGS serine/glycine linker, F2A F2A self-cleaving peptide of the foot-and-mouth disease virus, VL variable region of the immunoglobulin light chain of TRES6 antibody, C-hu-Igk constant region of the human kappa light chain, C-mu-Igk constant region of the murine kappa light chain, WPRE woodchuck hepatitis virus posttranscriptional regulatory element, hGH polyA signal human growth hormone polyadenylation signal.

Fig. 2. Characterization of AAV vector constructs.

a Western blot analysis of supernatants of HEK293T cells transfected with the indicated AAV vector constructs stained against human IgG1. Supernatants of mock-transfected cells and purified TRES6hu antibody at the indicated amounts were included as negative and positive controls, respectively. b Tenfold serial dilutions of supernatants of cells transfected with the indicated AAV vector constructs were analysed by a flow cytometric binding assay for the SARS-CoV-2 spike protein (one representative is shown out of three experiments). c Anti-human IgG1 Western blot analysis of supernatants of HEK293T cells transduced with the indicated dose (vg copies per cell) of the AAV vector construct packaged in AAV2 capsids, 0.5 µg of purified TRES6hu antibody served as control. d Human iPSC-derived cardiomyocytes were transduced with the AAV-TRES6 vector construct packaged in AAVMYO capsids at the indicated dose. Mock transduced cells and 0.5 µg of purified TRES6 antibody served as controls. All uncropped blots are shown in Fig. S1.

Fig. 3. Antibody concentrations in sera and BAL samples.

TRIANNI mice (n = 4, ♀ = 2, ♂ = 2) received either AAV8 (red) or AAVMYO (green) particles at the indicated doses of the viral genome (vg) copies. Two weeks after vector application, the mice were sacrificed and antibody concentrations were determined in sera (a) and bronchoalveolar lavage (BAL) (b) samples.

Fig. 4. Murine challenge study after AAV-vectored delivery of TRES6.

Female K18 hACE2 mice (n = 6 per cohort, ♀ = 6) received intravenously (i.v.) either 5 × 10¹¹ vg AAV8-TRES6 (red) or 5 × 10¹¹ vg AAVMYO-TRES6 (green) 14 days prior to intranasal SARS-CoV-2 challenge using 300 FFU of Wuhan strain. Control mice received i.v. either 3.33 mg/kg TRES6 (purple), TRES N297A (blue), or isotype control (gray) 5 days prior to the challenge. Cohort 1 (n = 6) was euthanized on day 4 and cohort 2 (n = 6) was euthanized according to humane endpoints or latest at day 10 after the challenge. The percentage of surviving animals according to humane endpoints are shown (a). Statistical evaluation of survival data were performed using the Mantel-Cox test in comparison to isotype control (*p ≤ 0.1; **p ≤ 0.01; ***p ≤ 0.001). Cohort 1 (n = 6) was euthanized at day 4. Mice were monitored daily for body weight (b) and cohort 2 (n = 6) was euthanized according to clinical score (c). Data shown were presented as means ± standard errors (humane endpoints ≥20 points, indicated as †). Viral RNA was extracted from lung homogenates and quantified by SARS-CoV-2-specific RT-qPCR (d). Data points shown represent viral copy numbers in each animal with the geometric mean of each group. Each point represents one mouse, whereby circles (●) indicate a survival of four or ten days post-infection and other symbols indicate mice that had to be euthanized according to humane endpoints at day 5 (▲), day 6 (⬛), or day 7 (♦). Statistical evaluation of the body weight, clinical score, and viral load data were performed by Mann–Whitney U-test (*p ≤ 0.05; ***p ≤ 0.001; ****p ≤ 0.0001; not indicated: non-significant). LOD limit of detection, dpi days post-infection.

Fig. 5. Long-term kinetic of TRES6 antibodies.

TRIANNI mice received intravenously either 5 × 10¹¹ vg copies of AAV-vector particles (n = 6, ♀ = 3, ♂ = 3) or as control 5 mg/kg body weight TRES6 antibody (n = 3, ♀ = 2, ♂ = 1). a Each data point represents the mean serum antibody concentration in each group at the indicated time point ± SEM. There was no detectable antibody concentration in the control mice measurable from week 24 onwards. b Differences in the total amount of serum antibodies were quantified for vector-transduced mice by calculating the area-under-the-curve (AUC) between weeks 4 and 52 of the long-term kinetic for each animal. c At weeks 4 and 52, mice were sacrificed and the TRES6 concentration was determined in the bronchoalveolar lavage fluid of AAV-transduced mice by ELISA. Each bar shows the mean antibody concentration of the group ± SEM. d Sera from vector-treated mice that were obtained at weeks 4 and 52 were tested for neutralization in a pseudotype neutralization assay and were compared to sera from human vaccinees. The reciprocal of the dilution at which 50% inhibition is reached (reciprocal ID₅₀) is shown per mouse. The bars indicate the group mean value ± SEM. b, d Statistical evaluation of the data were performed by Kruskal–Wallis test (one-way ANOVA) and Dunn’s pairwise multiple comparison procedures as post hoc test (*p ≤ 0.05; ****p ≤ 0.0001; not indicated: p > 0.05).

Fig. 6. Anti-drug antibody response, serum AAV vector DNA levels, and RNA expression levels.

a Anti-drug antibody levels in TRIANNI mice 52 weeks after injection of the indicated AAV vector particles. The dotted line represents the assay´s sample cut-off, calculated as the sum of the mean of the values from naïve mice and twice their standard deviation. b Double-stranded DNase-resistant AAV vector DNA copy numbers per μl DNA extracted from murine sera obtained in weeks 2, 4, and 8 of the long-term kinetic experiment with the indicated AAV vector (group size n = 6, ♀ = 3, ♂ = 3). Each bar represents the mean value of each group ± range. For statistical evaluation, a two-way-ANOVA was performed (n.s.: p > 0.05). c AAV vector RNA copy numbers per ng total RNA extracted from the indicated organs at four and 52 weeks after injection of the indicated AAV vector particles (n = 6, ♀ = 3, ♂ = 3). Each bar represents the mean value of each group ± SEM. The fold reduction between the two time points is indicated when the time points are above the lower limit of quantification (LOQ, dotted line).

Fig. 7. Fc glycosylation and impact on Fc-γ receptor interaction.

a The composition of glycans attached to the amino acid N297 of the antibody Fc part was measured by mass spectrometry for AAV8 (red)- and AAVMYO (green)-delivered TRES6. The samples were obtained in week 4 (W4, filled) and week 12 (W12, shaded) during the long-term kinetic experiment. The binding of in vivo produced TRES6 (AAV8: red; AAVMYO: green) or TRE6 N297A (blue) to the murine Fc-γ receptors RI (b), RIIb (c), RIII (d), and RIV (e) was analysed. Sera from week 8 and week 52 during the long-term kinetic experiment were used and diluted to a TRES6 concentration of 100 ng/ml. Each bar shows the mean value of the group ± SEM. Statistical evaluation of the data were performed by Kruskal–Wallis test (one-way ANOVA) and Dunn’s pairwise multiple comparison procedures as post hoc test (*p ≤ 0.05; ***p ≤ 0.001; ****p ≤ 0.0001; not indicated: non-significant).