Recombinant adeno-associated virus type 2 pseudotypes: comparing safety, specificity, and transduction efficiency in the primate striatum. Laboratory investigation

Abstract

Object: Although several clinical trials utilizing the adeno-associated virus (AAV) type 2 serotype 2 (2/2) are now underway, it is unclear whether this particular serotype offers any advantage over others in terms of safety or efficiency when delivered directly to the CNS.

Methods: Recombinant AAV2-green fluorescent protein (GFP) serotypes 2/1, 2/2, 2/5, and 2/8 were generated following standard triple transfection protocols (final yield 5.4 × 10(12) particles/ml). A total of 180 μl of each solution was stereotactically infused, covering the entire rostrocaudal extent of the caudoputamen in 4 rhesus monkeys (Macaca mulatta) (3.0 ± 0.5 kg). After 6 weeks' survival, the brain was formalin fixed, cut at 40 μm, and stained with standard immunohistochemistry for anti-GFP, anticaspase-2, and cell-specific markers (anti-microtubule-associated protein-2 for neurons and anti-glial fibrillary acidic protein for glia). Unbiased stereological counting methods were used to determine cell number and striatal volume.

Results: The entire striatum of each animal contained GFP-positive cells with significant labeling extending beyond the borders of the basal ganglia. No ischemic/necrotic, hemorrhagic, or neoplastic change was observed in any brain. Total infusate volumes were similar across the 4 serotypes. However, GFP-labeled cell density was markedly different. Adeno-associated virus 2/1, 2/2, and 2/5 each labeled < 8000 cells/mm(3), whereas serotype 8 labeled > 21,000 cells, a 3- to 4-fold higher transduction efficiency. On the other hand, serotype 8 also labeled neurons and glia with equal affinity compared with neuronal specificities > 89% for the other serotypes. Moderate caspase-2 colabeling was noted in neurons immediately around the AAV2/1 injection tracts, but was not seen above the background anywhere in the brain following injections with serotypes 2, 5, or 8.

Conclusions: Intrastriatal delivery of AAV2 yields the highest cell transduction efficiencies but lowest neuronal specificity for serotype 8 when compared with serotypes 1, 2, and 5. Only AAV2/1 revealed significant caspase-2 activation. Careful consideration of serotype-specific differences in AAV2 neurotropism, transduction efficiency, and potential toxicity may affect future human trials.

Fig. 1

Transduction of nonhuman primate striatum by recombinant AAV2/2. Sections were double-stained for neuronal-specific nuclear protein (NeuN-neurons) and EGFP (AAV-transgene). A: Brightfield low-power mosaic image of striatum. Original magnification × 100. B–E: Windows of fluorescence GFP staining. Original magnification × 10. F: Low-power fluorescence mosaic image with GFP and 4,6′-diamino-2-phenylindole-dihydrochloride (DAPI) demonstrating high transduction efficiency at the border between the caudate and internal capsule. Original magnification × 10. G: GFP, NeuN, and combined GFP and NeuN fluorescence images with DAPI showing high transduction efficiency in GFP window, high neuronal density in NeuN window, and complete overlap in GFP and NeuN combined window with the recombinant AAV2/2 construct. Original magnification × 4 (lower row), × 20 (upper row). Cd = caudate; GPe = globus pallidus externus; GPi = globus pallidus internus; Ic = internal capsule; Pu = putamen.

Fig. 2

Graphs showing number of GFP positively labeled cells (A), transduction efficiency (B), cell type specificity (C), and estimated transduction number of neurons and glia (D), using unbiased stereological counts. A: Number of positively labeled cells in striatum with single-sided injection, in which AAV2/8 > AAV2/5 > AAV2/2 > AAV2/1. B: Transduction efficiency (number of GFP-positive cells/mm³), in which AAV2/8 > AAV2/5 > AAV2/2 > AAV2/1. C: Cell type specificity (percentage of labeled cells) demonstrating almost complete neuronal specificity with AAV2/1, AAV2/2, and AAV2/5, but mixed cell type specificity withAAV2/8. D: Cell type specificity applied to stereological results from panel A, for estimated total number of transduced neurons and glia in striatum.

Fig. 3

Serotype-specific transduction differences between recombinant AAV2/1, 2/2, 2/5, and 2/8. Triple-labeled confocal micrographs with merged images (larger images) and single channels (smaller images) in GFP, MAP2 (neurons), and GFAP (glia). Adeno-associated virus 2/1, 2/2, and 2/5 show predominantly neuronal colabeling of MAP2 and GFP, whereas AAV2/8 demonstrates an almost equal combination of GFAP and MAP2 with GFP. Counts from confocal images were used to generate cell type specificity measurements. Bar = 100 μm.

Fig. 4

Caspase-3 activation near the injection tract of AAV2/1. Green fluorescent protein, caspase-3, and DAPI fluorescence images near (A–C) and away from (D–F) injection tract. Caspase-3 activation noted surrounding and around the injection tract (B) of AAV2/1, but not noted with other serotypes. Caspase-3 activation not observed beyond the needle tract (E) in any serotype.