Immunobiology of a rationally-designed AAV2 capsid following intravitreal delivery in mice

Abstract

Adeno-associated virus serotype 2 (AAV2) is a viral vector that can be used to deliver therapeutic genes to diseased cells in the retina. One strategy for altering AAV2 vectors involves the mutation of phosphodegron residues, which are thought to be phosphorylated/ubiquitinated in the cytosol, facilitating degradation of the vector and the inhibition of transduction. As such, mutation of phosphodegron residues have been correlated with increased transduction of target cells, however, an assessment of the immunobiology of wild-type and phosphodegron mutant AAV2 vectors following intravitreal (IVT) delivery to immunocompetent animals is lacking in the current literature. In this study, we show that IVT of a triple phosphodegron mutant AAV2 capsid is associated with higher levels of humoral immune activation, infiltration of CD4 and CD8 T-cells into the retina, generation of splenic germinal centre reactions, activation of conventional dendritic cell subsets, and elevated retinal gliosis compared to wild-type AAV2 capsids. However, we did not detect significant changes in electroretinography arising after vector administration. We also demonstrate that the triple AAV2 mutant capsid is less susceptible to neutralisation by soluble heparan sulphate and anti-AAV2 neutralising antibodies, highlighting a possible utility for the vector in terms of circumventing pre-existing humoral immunity. In summary, the present study highlights novel aspects of rationally-designed vector immunobiology, which may be relevant to their application in preclinical and clinical settings.

Fig. 1. A combination of three phosphodegron mutations leads to a synergistic increase in transduction of the murine retina.

Vectors were injected intravitreally and tissue was taken for analysis after three weeks. All data is presented as a column graphs showing the mean value for each group ±SEM. All statistical analyses are vs. the AAV2 group. a Representative images of retinal wholemounts depicting increased levels of GFP in the phosphodegron mutant groups compared to AAV2 WT control three weeks after bilateral intravitreal injection of 2E8 GC/eye. Quantification of GFP expression levels in wholemounted retina samples was performed in Volocity. b Percentage of GFP + RBPMS+ cells, and (c) mean GFP immunofluorescence/RBPMS + ROI/FOV was calculated. *p < 0.05, **p < 0.01, ***p < 0.001, parametric ANOVAs and Dunnett’s posthoc tests, n = 6–8. d Representative 40x epifluorescent microscope images of retinal sections transduced with AAV2, or phosphodegron mutant capsid AAV2. RGC retinal ganglion cell layer, INL inner nuclear layer, ONL outer nuclear layer. Quantification was performed in ImageJ. Here, (e) mean GFP fluorescence levels/FOV was calculated. *p < 0.05, **p < 0.01, ****p < 0.0001, Kruskal-Wallis and Dunn’s posthoc tests, n = 6–8. f Representative tilescan images showing wholemounted retinas from each group.

Fig. 2. Intravitreal injection of a triple phosphodegron mutant AAV2 induces humoral and cellular adaptive immune responses.

Vectors were injected and blood and tissue samples were taken three weeks later for analysis. All data is presented as a bar graph showing the mean value for each group ±SEM. All statistical analyses are vs. the AAV2 group. a Injection of phosphodegron mutant AAV2 increases murine sera neutralising antibody (NAb) levels. NAb titres were assessed using a HEK-293T/1E9 VP/mL AAV2.CMV.Luciferase system. Luminescence was measured across a range of dilutions of sera, and remaining infectivity was defined as the luminescent signal in each well divided by ‘Control’ values. b IC50 values for each group were calculated using non-linear regression (variable slope, four parameters) in GraphPad **p < 0.01, Brown-Forsythe ANOVA and Dunnett’s post hoc tests, n = 6. c Immunoglobulin ELISA assays were performed to assess the antibody subtypes responsible for the neutralising effect. **p < 0.01, ****p < 0.0001, two-way AVOVA and Dunnett’s posthoc tests, n = 6. d Representative 40x magnification epifluorescent microscope images showing that intravitreal injection of triple phosphodegron mutant AAV2 is associated with the presence of CD4+ T-cells in the retina. e Quantification of dataset was undertaken via manual counting of the number of CD4+ ROIs/FOV. *p < 0.05, **p < 0.01, Kruskal-Wallis and Dunn’s posthoc tests, n = 6–10. f Representative 40x magnification epifluorescence microscope images showing that intravitreal injection of triple phosphodegron mutant AAV2 is associated with the presence of CD8+ T-cells in the retina. g Quantification of dataset was undertaken via manual counting of CD8+ ROIs/FOV. *p < 0.05, **p < 0.01, Kruskal-Wallis and Dunn’s posthoc, n = 4–8.

Fig. 3. Intravitreal injection of AAV2 (TM) induces changes in splenic lymphocyte populations.

Vectors were injected via IVT and spleens were harvested for analysis after three weeks. Changes were observed in the levels of follicular helper T-cells and germinal centre B-cells, and in classical and myeloid dendritic cell MHC c. II expression. All data is displayed as column graphs, and show the mean for each group ±SEM. All statistical analyses are vs. the AAV2 group. a CD4 and CD8 T-cell gating strategy used in subsequent analyses. b Effector memory CD4 + T-cell (CD3 + CD4 + CD8- CD62L_lo CD44_hi) levels expressed as a percentage of total CD4 + T-cell levels, and representative flow plots. c Effector memory CD8 + T-cell (CD3 + CD4- CD8+ CD62L_lo CD44_hi) levels expressed as a percentage of total CD8 + T-cell levels, and representative flow plots. d Follicular helper CD4 + T-cell (CD3 + CD4+ CXCR5_hi PD-1_hi) levels expressed as a percentage of all CD4+ cells. **p < 0.01, Brown-Forsythe ANOVA and Dunnett’s T3 posthoc tests (n = 4), and representative flow plots. e B-cell gating strategy used for subsequent analysis. f Germinal centre B-cell (CD19+ IgM + /_lo IgD_lo CD95_hi GL7+ ) levels as a percentage of all splenic lymphocytes. ***p < 0.001, Kruskal-Wallis test and Dunn’s posthoc tests (n = 4), and representative flow plots. g Levels of MHC c. II expressed on germinal centre B-cells. **p < 0.01, one-way ANOVA and Dunnett’s posthoc tests (n = 4). h Percentage of class-switched B-cells (IgM_lo IgD_lo) in the germinal centres, as a percentage of all germinal centre B-cells. **p < 0.01, one-way ANOVA and Dunnett’s posthoc tests (n = 4), and representative flow plots. i Gating strategy used to delineate DC subsets for subsequent analysis j MHC c. II expression on conventional DCs subset 1 (cDC1) (CD11c^hi CD8a + XCR1 + ) dendritic cells. *p < 0.05, one-way ANOVA and Dunnett’s posthoc tests (n = 4). k MHC c. II expression on conventional DCs subset 2 (cDC1) (CD11c^hi CD11b+) dendritic cells. *p < 0.05, Student’s t test (n = 4). l MHC c. II expression on plasmacytoid (CD11c+ SiglecH+) dendritic cells.

Fig. 4. Intravitreal injection of a phosphodegron mutant AAV2 and high titre AAV2 leads to gliosis in the murine retina.

Vectors were injected via IVT and retinas were extracted for analysis after three weeks. All data is presented as column graphs, with the mean value for each group shown ±SEM. All statistical analyses are vs. the AAV2 group. RGC, retinal ganglion cell layer; INL inner nuclear layer, ONL outer nuclear layer. a Representative 40x objective epifluorescent microscopy images of retinal cryosections showing IBA1 immunoreactivity in groups receiving phosphodegron mutant AAV2 injections. b Corresponding quantification performed in ImageJ shows the level of IBA1 immunoreactivity per field of view (FOV). *p < 0.05, **p < 0.01, Kruskal-Wallis ANOVA and Dunn’s posthoc test, n = 4–5. c Representative 40x objective epifluorescent microscopy images of retinal cryosections showing GFAP immunoreactivity in groups receiving phosphodegron mutant AAV2 injections. Corresponding quantification was performed using Simple Neurite Tracer, an ImageJ plugin (https://imagej.net/SNT). Each GFAP+ fibril was identified in each FOV and the length of each fibril (d) and its fluorescence intensity (e) and was measured. *p < 0.05, **p < 0.01, one-way ANOVA and Dunnett’s posthoc tests, n = 4. f Representative 63x images acquired on an Leiss Airyscan Confocal microscope showing increased complexity of astrocytic dendritic arbours in AAV2 (TM). Data was analysed in ImageJ using by assessing the number of junctions (g) and quadruple points (h) of skeletonized images. *p < 0.05, **p < 0.01, one-way ANOVA and posthoc Dunnett’s tests, n = 5–8.

Fig. 5. Injection of phosphodegron mutant vectors is not associated with changes in electrophysiological function in the murine retina.

Vectors were injected via IVT and ERG was performed after three weeks. 250 ng of lipopolysaccharide (LPS) 24 h prior to ERG. a Average pSTRs, B-waves and A-waves from each treatment group (–4.37, –1.90 & 1.29 log cd.s/m² light intensities, respectively). b Column charts showing peak voltages across a range of light intensities for each treatment group. Data is presented as the mean value ±SEM. c Representative 40x confocal images of Tuj1+ cells (an RGC marker) in flatmounted retina tissue.

Fig. 6. Mutation of AAV2 capsids at selected phosphodegron regions attenuates neutralisation by heparan sulphate (HS), and anti-AAV2 neutralising antibody-containing sera.

All data is presented as bar graphs, with the mean value for each group shown ±SEM. All statistical analyses are vs. the AAV2 group. a VP3 capsid monomer showing the positioning of the three phosphodegron mutants, the five residues thought to mediate the binding affinity of AAV2 capsids to HSPG (blue) and the 14 residues thought to mediate binding of AAV2 to the AAV receptor (orange) (AAVR; KIAA0319). This model was generated in PyMol using 6ih9, a 2.8 Å resolution cryoelectron microscopy-derived structure of the AAV2 VP3 monomer. b AAV2 full 60mer capsid structure with highlighted phosphodegron mutations and heparin binding domains. Red = Y444F, Green = K556E, Yellow = S662V, Blue = R484, R487, K532, R585, and R588 heparin binding domains (HBDs), Orange = R471, D528, Q589, T592, S262, Q263, G265, A266, S267, N268, H271, N382 and Q385 AAVR binding domains (AAVR BDs). c RIVEM plot showing the positioning of the three phosphodegron mutants and their proximity to HBDs and AAVR BDs (using 6IH9 AAV2 coordinates, see (b) for colouring). d Schematic representation of mutations represented in PyMol. Mutations were introduced using PyMol’s Mutagenesis Wizard, and the optimal rotamer confirmation was selected in accordance with the software’s prediction. e Mutation of phosphodegron residues attenuates neutralisation by HS. AAV2 and mutant capsids were incubated with HS for 1 h, prior to addition to HEK293T cell media. Number of GFP+ cells was normalised to –HS controls for each vector group, allowing calculation of remaining infectivity (I/I0). *p < 0.05, ****p < 0.0001, one-way ANOVA and a Dunnett’s posthoc tests, n = 4. f Mutation of phosphodegron residues partially rescues neutralisation by AAV2 NAbs. AAV2 and mutant capsids were incubated with sera extracted from animals that had previously been injected intravitreally with AAV2, and the samples were tested to confirm the presence of NAbs. Calculation of remaining infectivity (I/I0) was performed as in (d). *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, one-way ANOVA and a Dunnett’s posthoc tests, n = 4.