Tropism of engineered and evolved recombinant AAV serotypes in the rd1 mouse and ex vivo primate retina

Abstract

There is much debate on the adeno-associated virus (AAV) serotype that best targets specific retinal cell types and the route of surgical delivery-intravitreal or subretinal. This study compared three of the most efficacious AAV vectors known to date in a mouse model of retinal degeneration (rd1 mouse) and macaque and human retinal explants. Green fluorescent protein (GFP) driven by a ubiquitous promoter was packaged into three AAV capsids: AAV2/8(Y733F), AAV2/2(quad Y-F) and AAV2/2(7m8). Overall, AAV2/2(7m8) transduced the largest area of retina and resulted in the highest level of GFP expression, followed by AAV2/2(quad Y-F) and AAV2/8(Y733F). AAV2/2(7m8) and AAV2/2(quad Y-F) both resulted in similar patterns of transduction whether they were injected intravitreally or subretinally. AAV2/8(Y733F) transduced a significantly smaller area of retina when injected intravitreally compared with subretinally. Retinal ganglion cells, horizontal cells and retinal pigment epithelium expressed relatively high levels of GFP in the mouse retina, whereas amacrine cells expressed low levels of GFP and bipolar cells were infrequently transduced. Cone cells were the most frequently transduced cell type in macaque retina explants, whereas Müller cells were the predominant transduced cell type in human retinal explants. Of the AAV serotypes tested, AAV2/2(7m8) was the most effective at transducing a range of cell types in degenerate mouse retina and macaque and human retinal explants.

Figure 1

Scanning laser ophthalmoscopy of degenerate mouse retinas injected with three GFP-expressing AAV serotypes. (a–c) Mice were injected at 13–14 weeks of age and assessed 3 weeks post injection by confocal scanning laser ophthalmoscopy (cSLO). Images centred on the optic disc were acquired from each of the eyes injected with each of the following serotypes: (a) AAV2/8(Y733F), (b) AAV2/2(quad Y-F) and (c) AAV2/2(7m8). The same autofluorescence settings were used. (d) SLO images were quantified by setting a threshold pixel value and determining the area of each image above this threshold and (e) the intensity of the signal (‘Pixel value’) in the area above the threshold (mean±s.e.m.; n=4, AAV2/8(Y733F) and AAV2/2(quad Y-F); n=3, AAV2/2(7m8)). *P<0.05, **P<0.01. Scale bar, 1 mm.

Figure 2

Statistical analysis of immunohistochemistry data comparing delivery routes and AAV serotypes in degenerate rd1 mouse retinas. (a) Ordinary two-way analysis of variance (ANOVA) tests examining the effect of delivery route (intravitreal or subretinal) and AAV serotype on the area above threshold for each of the three layers of the retina were examined. Only the effect of delivery route and AAV serotype on the area above threshold in the inner plexiform layer showed a statistically significant interaction between delivery route and area above threshold, F(1, 11)=5.2, P=0.043 (n=3, all groups except intravitreal AAV2/2(7 m8) (n=2)). Simple effects analysis showed that the area above threshold in the inner plexiform layer from a subretinal injection with AAV2/8(Y733F) was significantly lower than the area associated with an intravitreal injection with AAV2/2(7m8). (b) Ordinary two-way ANOVA tests examining the effect of delivery route and AAV serotype on the pixel value in each of the three layers of the retina showed no statistically significant interactions. *P<0.05.

Figure 3

Confocal fluorescence micrographs of degenerate mouse retinas injected with three serotypes of AAV-GFP and double labelled for GFP and retinal cell markers. Degenerate rd1 mouse retinas were injected either intravitreally or subretinally with an AAV vector expressing GFP driven by a ubiquitous promoter, with one of three AAV serotypes: AAV2/8(Y733F), AAV2/2(quad Y-F) or AAV2/2(7m8). Vertical sections were double immunolabelled for GFP and retinal cell markers: (a) calbindin; (b) protein kinase-Cα (PKCα); (c) glutamate decarboxylase 67 (GAD67); (d) glycine transporter 1 (GlyT1) or (e) brain-specific homeobox/POU domain protein 3a (Brn3a). Cell bodies that were immunopositive for both GFP and the cell marker are encircled. Arrowheads indicate GFP-expressing retinal pigment epithelium (RPE). The RPE cannot be identified in the majority of the panels because of it detaching during sample preparation. See Table 1 for a summary of the colocalisation results. GCL, ganglion cell layer; INL, inner nuclear layer. Scale bar, 20 μm.

Figure 4

Macaque fovea explants transduced with three serotypes of AAV-GFP. The 3 mm diameter discs of macaque retina centred on the fovea were cultured for 8 days in AAV-GFP solution before being fixed and immunolabelled with anti-GFP antibody. Three different AAV serotypes were used to transduce different explants: AAV2/8(Y733F), AAV2/2(quad Y-F) and AAV2/2(7m8). Images were acquired using the same settings. The outer nuclear layer is nearest the camera in all images. Scale bar, 500 μm.

Figure 5

Fluorescence micrographs of foveal macaque retina explants transduced with three serotypes of AAV-GFP and double labelled for GFP and rhodopsin. Macaque explants from the fovea were transduced with AAV-GFP with one of three serotypes—AAV2/8(Y733F), AAV2/2(quad Y-F) or AAV2/2(7m8)—or no AAV. Explants were fixed, sectioned and double immunolabelled with antibodies against GFP and rhodopsin (1D4) to show the distribution of GFP in the retinal layers. All sections are oriented with photoreceptors up. Note that these images are from tissue that is unique from those samples shown in Figure 4. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bar, 200 μm.

Figure 6

Confocal fluorescence micrographs of macaque foveal explants transduced with AAV2/8(Y733F)-GFP, co-labelled for retinal cell markers. (a) The majority of GFP-positive cells in macaque foveal explants were calbindin positive, indicating they were cones. (b) However, a small number of GFP-positive photoreceptors were calbindin negative (arrowheads) that, combined with the slender inner segment morphology of these cells and the more inner location of the cell bodies, suggests these are rods. Calbindin- and GFP-positive cells were also identified in the ganglion cell layer (*). (c) At the very periphery of a flat-mounted explant, a PKCα-positive bipolar cell was co-labelled (arrowhead). GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bar, 20 μm.

Figure 7

Fluorescence micrographs of human retina explants transduced with three serotypes of AAV-GFP. Retina from patients requiring retinectomy was cultured for 1 week in the presence of AAV-GFP of three different serotypes: (a) AAV2/8(Y733F), (b) AAV2/2(quad Y-F) or (c) AAV2/2(7m8). Explants were sectioned and immunolabelled for GFP and cell markers, such as protein kinase-Cα (PKCα, used in (a) and (b) only). GFP was largely limited to the inner nuclear layer (INL) of AAV2/8(Y733F) transduced retina, but extended throughout all retinal layers in retina transduced with AAV2/2(quad Y-F) and AAV2/2 (7m8). GCL, ganglion cell layer; ONL, outer nuclear layer. Scale bar, 20 μm.

Figure 8

Fluorescence micrographs of human retina explants transduced with either AAV2/8(Y733F)-, AAV2/2(quad Y-F)- or AAV2/2(7m8)-GFP, or no AAV and co-labelled for GFP and (a and e) protein kinase-Cα (PKCα), (b) rhodopsin, (c) glial fibrillary acidic protein (GFAP) or (d) calbindin. (a, e) No cells were clearly double labelled for GFP and PKCα. (b) Explants transduced with both AAV2/2(quad Y-F)-GFP and AAV2/2(7m8)-GFP were identified as co-labelled for GFP and rhodopsin, suggesting transduced rods (arrowheads). (c) AAV2/2(quad Y-F)-GFP and AAV2/2(7m8)-GFP transduced retinas showed GFP colocalisation with GFAP (arrowheads). (d) Sections of AAV2/2(7m8)-GFP transduced retina were noted to have cells that were both GFP and calbindin positive (arrowhead). (e) No GFP was identified in retina not exposed to either AAV. All sections are oriented with the photoreceptors up. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer. Scale bar, 20 μm.