Whole-body biodistribution of AAV9P31 upon i.v., i.c.v. and intraparenchymal administration in C57BL/6J mice
The AAV9P31-CMV-GFP vector was administered to different groups of 7 weeks-old C57BL/6J mice using the following routes of administration: i.v. (4x1011 vg, equivalent to 2x1013 vg/kg); i.c.v. (1.2x1011 vg, unilateral), and intraparenchymal (6x1010 vg, intrastriatal, unilateral). One month later, mice were sacrificed and vector genomes were quantified in tissue samples by qPCR. We confirmed the efficient access of the vector to the brain and liver upon i.v. administration, as previously reported(21). In addition, we found that the lung is also a preferential site for accumulation of AAV9P31 genomes (Fig. 1A). Detailed analysis of different brain regions revealed a relatively homogeneous infection of all structures. In contrast, direct injection into the striatum led to high viral genome content in this location, together with the parieto-temporal cortex in the tract of injection. The access to other brain regions distant from the injection site such as brainstem and cerebellum was significantly lower (Fig. 1C). As expected, the spread of the vector to peripheral organs was low. The i.c.v. administration showed an intermediate phenotype, with clear accumulation of genomes in the brain, especially hippocampus, but relatively lower content in cerebellum and brainstem (Fig. 1B). However, no statistically significant differences were observed among brain structures because the inter-individual variability was relatively high. Immunohistochemical analysis of brain sections confirmed the pattern of transgene expression (GFP) in the brain of mice injected systemically with AAV9P31-CMV-GFP (Fig. 2). The efficacy of hepatocyte transduction following this route of administration was moderate but clearly higher than the i.c.v. route (supplementary figures B and C). Only the intraparenchymal route showed no systemic exposure to the vector (supplementary Fig. 1D).
In order to determine the tropism of the vector for different cell populations in the CNS, we performed multiple immunofluorescent stains in brain sections from C57BL/6J mice treated with AAV9P31-CMV-GFP. Co-staining was performed with GFP for AAV transduction and specific markers for neurons (NeuN), astroglia (GFAP), and microglia (Iba1). Although the conducted immunoperoxidase stains upon i.v. delivery of AAV9P31-CMV-GFP showed a widespread transduction throughout the entire brain, immunofluorescent stains revealed different patterns of cellular transduction between different brain regions. In this regard, both the cerebral cortex and the striatum showed a preferential and random transduction of astrocytes, as well as projection neurons (Fig. 3). Cortical astrocytes were transduced very strongly, with GFP expression extending over the thinnest astrocytic processes (Fig. 3, panels A and B). By contrast, neurons were found to be the only cell type expressing GFP within any examined subcortical locations such as the hippocampus, amygdaloid complex, thalamus, hypothalamus, mesencephalon, sensory and motor-related nuclei of brainstem and cerebellum (Fig. 4). Furthermore, GFP expression was never found in microglial cells regardless the brain territory under scrutiny (supplementary Figs. 2 and 3).
Spatial and temporal analysis of transgene expression upon i.v. administration of AAV9P31
In order to validate the results obtained with the AAV9P31-GFP vector and to gain insight into the kinetics of transgene expression, we employed the NanoLuc-expressing vector AAV9P31-CMV-NL. In the first experiment, 7 weeks-old C57BL/6J mice received the vector by i.v. injection (2x1013 vg/kg). One month later, mice were treated with an intraperitoneal injection of the substrate (furimazine), and light emission was detected by a luminometer in live animals (Fig. 5A). A strong signal was detected in the brain area and the abdomen. However, spatial definition of BLI is low, and light can be quenched by overlapping tissues such as bone or hair. In order to define which organs and brain regions were responsible for light emission, mice were sacrificed and luciferase activity was measured in tissue extracts (Fig. 5B). The results confirm widespread and relatively homogeneous transgene expression across all CNS structures, including cerebellum, brainstem and spinal cord. Interestingly, peripheral organs expressing high levels of NanoLuc were not only liver and lung, but also peritoneum, heart and skeletal muscle. Only residual expression was observed in the colon, gonads and uterus. In contrast, the prostate showed moderate levels, similar to the eyes. In general, these results are consistent with the detection of viral genomes in the same organs (Fig. 5C), although the ratio between NanoLuc expression and viral genomes is variable among tissues (supplementary Fig. 4). In order to study the kinetics of transgene expression, two weeks-old C57BL/6J mice received the AAV9P31 vector by i.v. injection, and light emission was monitored for up to 5 months (Figs. 5D and 5E). We found strong signals as early as 72 h after vector administration both in the head and abdominal regions. After a moderate decay during the first 3 weeks, transgene expression remained stable until the end of the experiment in both regions.
Biodistribution and transgene expression of AAV9P31 in Balb/c mice upon i.v. administration of AAV9P31
Next, we studied if the ability of AAVP31 to cross the BBB was maintained in other commonly used mouse strains such as Balb/c mice. To this end, the AAV9P31-CMV-NL vector was injected i.v. in this strain (2x1013 vg/kg), and the animals were sacrificed one month later to quantify vector genomes and luciferase activity in different brain structures and peripheral organs. The pattern of transgene expression (NanoLuc) in Balb/c mice was similar to C57BL/6J (Fig. 6A), in agreement with the conservation of CA-IV across different mouse strains(25), supplementary Fig. 5. The distribution of genomes was also common, with some exceptions such as higher vector genomes in the spleen and lower in the lung of Balb/c mice compared with C57BL/6J (Fig. 6B). Efficient and homogeneous transduction of brain structures was confirmed. Interestingly, Balb/c mice presented lower NanoLuc expression levels relative to the vector genome content in all tissues studied, in comparison with C57BL/6J (supplementary Fig. 4).
Biodistribution of AAV9P31 in rats and NHPs
After characterization of AAV9P31 in the most widely used mouse strains, we investigated to what extent this capsid can bypass the BBB in other species. Initial experiments were carried out by i.v. injection of the AAV9P31-CMV-GFP vector into adult (7 weeks-old) Sprague Dawley rats (2x1013 vg/kg). Immunohistochemical analysis of GFP in rat brains performed one month after vector injection revealed no transgene expression in most of the brain territories examined, despite moderate transduction in the liver (supplementary Fig. 1E). However, neurons transduced with GFP were constantly found in several different nuclei in the brainstem (Fig. 7), therefore suggesting an unexpected and enriched distribution of the targeted receptor specific for the brainstem. For the analysis of AAV9P31 capsid in NHPs, we compared the three routes of administration tested in C57BL/6J mice. In order to minimize the use of animals and to best compare different administration routes, the same macaques (n = 2) were injected with vectors expressing GFP, tdT and NanoLuc following i.v., i.c.v. and intraparenchymal injections, respectively. The i.v. dose was maintained as 2x1013 vg/kg. For the i.c.v. route, animals received a total of 4x1013 vg in bilateral injections (2x1013 vg in 1 ml/ventricle). The intraparenchymal injection consisted of 2x1011 vg in a volume of 5 µl injected unilaterally in the left putamen nucleus. Animals were sacrificed one month after injection, and the expression of the respective transgenes was analyzed by immunoperoxidase stains. In the case of NanoLuc, the antibody was directed against the N-terminal HA tag fused to the protein. No meaningful GFP expression was observed in the brain of NHPs injected i.v. with the AAV9P31-CMV-GFP vector, whereas positive hepatocytes were readily detected in the liver (supplementary Fig. 1F). GFP+ cells were only found through the injection tract from the insertion of the needle used for intraputaminal deliveries and thus leading to a mechanical disruption of the BBB in this location (Fig. 8A), therefore suggesting that the AAV9P31 capsid was unable to bypass the intact BBB in NHPs upon systemic delivery. That said, HA+ cells were detected in the injected putamen nucleus (Fig. 8B), indicating the ability of the vector to transduce neurons in the macaque brain following intraparenchymal deliveries. Moreover, the AAV9P31 capsid variant maintained the well-known retrograde spreading properties of the native AAV9 capsid and therefore external division of the globus pallidus (GPe) neurons projecting to the putamen (arkypallidal neurons) also became transduced (Fig. 8B’’). Similarly, neurons located in the left substantia nigra pars compacta innervating the left putamen nucleus through the nigrostriatal pathway were also retrogradely transduced upon intraparenchymal delivery of the vector (Figs. 8C’-C’’). Finally, the i.c.v. administration of AAV9P31-CMV-tdT led to a widespread neuronal transduction in the cerebral cortex, subcortical structures and the cerebellum (Fig. 8D-F’’’), likely mimicking the ability of the native AAV9 capsid to bypass the ependymal brain barrier instead of reflecting a receptor-mediated transport. At the cortical level, pyramidal neurons from upper and lower cortical layers became transduced. Neurons positive for tdT are organized in a columnar random pattern (Figs. 8D and 8D’). Moreover, a patchy distribution of neurons was noticed in the putamen nucleus, more evident in ventrolateral putaminal areas (Figs. 8D’’ and 8D’’’). A similar patchy distribution of transduced neurons was found in the thalamus (Figs. 8E and 8E’). A moderate neuronal transduction was found in the hippocampal formation (Fig. 8E’’), with a higher number of tdT+ neurons being found in the CA3 field (inset in E’’). Neurons expressing tdT were also found in deep brain structures such as the substantia nigra pars compacta (Fig. 8E’’’). Regarding the brainstem, the most abundant neuronal transduction was found in peri-ventricular locations such as the abducens nucleus (Fig. 8F’). At the level of the cerebellar cortex, climbing fibers were the only structures expressing tdT (Fig. 8F’’), whereas neurons located in deep cerebellar nuclei were strongly transduced (Fig. 8F’’’).