Several rAAV Vectors Efficiently Cross the Blood–brain Barrier and Transduce Neurons and Astrocytes in the Neonatal Mouse Central Nervous System

Abstract

Noninvasive systemic gene delivery to the central nervous system (CNS) has largely been impeded by the blood–brain barrier (BBB). Recent studies documented widespread CNS gene transfer after intravascular delivery of recombinant adeno-associated virus 9 (rAAV9). To investigate alternative and possibly more potent rAAV vectors for systemic gene delivery across the BBB, we systematically evaluated the CNS gene transfer properties of nine different rAAVEGFP vectors after intravascular infusion in neonatal mice. Several rAAVs efficiently transduce neurons, motor neurons, astrocytes, and Purkinje cells; among them, rAAVrh.10 is at least as efficient as rAAV9 in many of the regions examined. Importantly, intravenously delivered rAAVs did not cause abnormal microgliosis in the CNS. The rAAVs that achieve stable widespread gene transfer in the CNS are exceptionally useful platforms for the development of therapeutic approaches for neurological disorders affecting large regions of the CNS as well as convenient biological tools for neuroscience research.

Introduction

Many neurological disorders are caused by single gene mutations leading to either loss of function, or gain of a deleterious new property/function. Gene therapy is potentially beneficial in these disorders, either through the delivery of the normal gene, in the former case, or vehicles that can silence a miscreant gene and protein by RNA interference, in the latter case.^1,2 Targeted infusion of gene delivery vectors into discrete structures in the central nervous system (CNS) is highly effective for diseases with a localized lesion, or those where local expression of the transgene is sufficient to modify the overall disease phenotype.^3,4,5 For diseases that affect large areas of the CNS, however, local injection of gene delivery vectors is less than optimal, because it provides transgene expression only to limited regions in the CNS (near the injection site). Direct local injection also entails surgical risks and clinical costs. An ideal approach to address CNS disorders that affect large areas of the brain and spinal cord is to administer vectors through the vasculature.^6,7 This has been the holy grail of CNS-directed gene therapy for many decades. In addition to the obvious clinical applications, this technology may offer an alternative to traditional transgenesis for basic neuroscience research. This technology may be used to rapidly and economically generate somatic CNS transgenics or gene knockdown animal models without costly and time intensive genetic manipulations.

Presently, a critical impediment to widespread CNS gene transfer via the vasculature is the blood–brain barrier (BBB), which consists of many components, including endothelial tight junctions, astrocytic end-feet, pericytes, and cellular basement membranes; the actual barrier to CNS permeation is a consequence of these anatomic structures and functional features such as low pinocytic activity. These preclude the entry into the CNS of >98% of small molecule drugs and almost all macromolecule drugs, such as therapeutic proteins and gene delivery vectors.^8,9

Significant efforts have been made to develop and identify effective and safe vectors to deliver genes to the CNS through systemic administration.^{7,10,11,12,13} Among nonviral and viral vectors, recombinant adeno-associated viruses (rAAVs) have demonstrated a great potential in CNS gene transfer.^14,15,16 In human clinical trials, direct injection of AAV vectors gave rise to sustained transgene expression and therapeutic effect.^17,18,19 Emerging self-complementary AAV (scAAV) vectors hold extra advantages over their single strand counterparts due to higher gene delivery efficiency.²⁰ Recently, Foust et al. demonstrated that intravenous administration of scAAV9 resulted in extensive transduction of neuronal cells in neonates and astrocytes in adult mice.⁷ Importantly, when administered systemically to deliver the survival motor neuron gene (SMN1) to postnatal day 1 (P1) mice with spinal muscular atrophy, scAAV9 rescued a uniformly lethal cell death phenotype.⁶ This is an outstanding accomplishment in CNS gene therapy, although it remains possible that the host serological response to AAV9 may preclude widespread use of this vector.

Recently Gao et al. cloned a diverse family of more than 120 novel primate AAVs, including AAV9, with unique tissue/cell tropisms and varying efficiencies of in vivo gene transfer.^21,22 We report here a survey of nine scAAV vectors for their CNS gene transfer properties after systemic administration. This study was undertaken to identify alternative and possibly more effective vectors for the CNS gene transfer, with a focus on newly-isolated serotypes or natural variants for enhanced-permeation of the BBB and improved delivery of enhanced green fluorescent protein (EGFP) to the CNS following facial vein injection on P1. AAV9 was also included as a positive control. Except for rAAV2 and rAAV5, all other seven vectors crossed the BBB with varied transduction efficiency, among which rAAVrh.10, rAAVrh.39, rAAVrh.43, rAAV9, and rAAV7 rank in the top five, mediating robust EGFP expression in both neuronal and glial cells throughout the CNS. The performance of rAAVrh.10 is comparable to that of rAAV9, if not better. Our data document that the ability to cross the BBB in neonatal mice is not restricted to rAAV9; rAAVs continue to hold great potential as efficient vectors for systemic gene delivery to the CNS.

Results

Intravenous injection of rAAVs mediated widespread transduction in neonatal mouse CNS

Twenty one days after vector administration in P1 mice, we compared the CNS transduction profiles of the following recombinant AAV vectors encoding EGFP: rAAV1, rAAV2, rAAV5, rAAV6, rAAV6.2, rAAV7, rAAV9, rAAVrh.10, rAAVrh.39, and rhAAVrh.43. The vectors used in this study were comparable in purity and morphological integrity (Supplementary Figure S1). As assessed by the scoring system of Cearly et al.³ (see Materials and Methods), the positive control rAAV9 was indeed among the best performers; seven out of nine other rAAVs tested (rAAV1, rAAV6, rAAV6.2, rAAV7, rAAVrh.10, rAAVrh.39, and rAAVrh.43), but not rAAV2 and rAAV5, also gave rise to EGFP expression throughout the CNS (Table 1). However, the apparent number of EGFP-positive cells (Table 1) varied among CNS structures in a vector dependent manner. The region with the highest EGFP transduction was the hypothalamus followed by medulla, striatum, hippocampus, cortex, and cerebellum. In contrast, the transduction efficiency in olfactory bulb and thalamus was relatively low (Table 1). We also assessed average EGFP signal intensity/pixel in 12 different CNS regions to derive a more quantitative assessment of gene transfer efficiency of each rAAV, 12 CNS regions (Figure 1a). For the eight rAAV vectors that achieved CNS transduction after intravenous injection, the mean EGFP signal intensity/pixel was relatively low in cortex, habenular nucleus, cornu ammonis, dentate gyrus, thalamus, cerebellum, and olfactory bulb, moderate in choroid plexus and caudate-putamen, but high in hypothalamus, medulla and amygdala (Figure 1a). Next, the average EGFP signal intensity/pixel in the brain (average of 12 regions) for different rAAVs were compared in Figure 1b. AAVrh.10, AAVrh.39, and AAVrh.43 stood out for their overall gene transduction efficiency in brain, followed by AAV7, AAV9, and AAV1 (Figure 1b). Those eight effective serotypes also mediated EGFP expression throughout the spinal cord, to different degrees. The same quantitative analysis was performed for each rAAV in the cervical, thoracic and lumbar sections of the spinal cord (Figure 1a); the average EGFP signal intensity/pixel of the three sections for different rAAVs were also compared (Figure 1b). Overall, rAAV1, rAAV9, rAAVrh.10, rAAVrh.39, and rAAVrh.43 displayed stronger transduction in the spinal cord with the highest EGFP signal intensity/pixel observed in cervical, followed by thoracic and lumbar sections of the spinal cord (Figure 1a,b). For rAAV2 there were only a few EGFP-positive cells in hippocampus, cortex, and hypothalamus, while none was observed in most CNS regions in AAV5-injected mice except in the hypothalamus. The following is a more detailed description of our findings in different CNS structures. The selection of CNS structures for “zoom-in” analysis was based on their relevance to neurodegenerative diseases in humans (e.g., striatum in Huntington's disease, hippocampus, and cortex in Alzheimer's disease, cerebellum in spinal-cerebellar ataxias, spinal cord in amyotrophic lateral sclerosis, and spinal cord injury), and distinct transduction profiles by different rAAVs.

Table 1. Transduction characteristics of AAV serotypes following intravascular injections into neonatal mouse brain.

Figure 1.

Quantification of GFP intensity levels in the brain and spinal cord of 21 day-old mice infused at postnatal day 1 (P1) with various rAAVs. 4 × 10¹¹ genome copies of ten different rAAV vectors were injected into neonatal P1 pups via the superficial temporal vein. Mice were killed 21 days after injection, and 40 µm thick brain and spinal cord cryosections were stained with an anti-EGFP antibody for immunofluorescence. (a) EGFP signal intensity/pixel in different brain and spinal cord regions was measured for each rAAV vector. (b) Average EGFP signal intensity/pixel in brain and spinal cord corresponds to the average value across all respective structures analyzed individually in (a). CSP, cervical section of spinal cord; D, dorsal region; EGFP, enhanced green fluorescent protein; LSP, lumbar section of spinal cord; rAAV, recombinant adeno-associated virus; TSP, thoracic section of spinal cord; V, ventral region.

Striatum. Previous studies have shown that systemic injection of rAAV9 in neonatal mice yields robust striatal transduction.⁷ In this study, a large number of cells with neuronal morphology in this region were also transduced by rAAVrh.10 (Figure 2), which was confirmed by costaining with a neuronal marker as described below. Other vectors, including rAAVrh.39, rAAVrh.43 and rAAV7, also mediated moderate transduction in striatum (Figure 2). In contrast, rAAV6, rAAV6.2, and rAAV1 resulted in relatively lower EGFP expression in this structure (Figure 2).

Figure 2.

Strong and widespread EGFP expression in mouse brain after neonatal intravenous injection of rAAVs. 4 × 10¹¹ genome copies of rAAVs vectors were injected into neonatal postnatal day 1 (P1) pups, and distribution of EGFP expression in the brain was analyzed at 21 days postinjection. Forty micrometer thick cryosections were stained with an anti-EGFP antibody for immunofluorescence. The regions shown are: olfactory bulb, striatum, hippocampus, cortex, hypothalamus, cerebellum, and medulla. Representative sections are shown for each rAAV. Bars = 100 µm. EGFP, enhanced green fluorescent protein; rAAV, recombinant adeno-associated virus.

Hippocampus. Large numbers of EGFP-positive neurons were observed bilaterally in all regions of the hippocampus, namely dentate gyrus, hilus, CA1, CA2, and CA3 for the mice received intravenous rAAVrh.10, rAAV9, rAAV7, rAAVrh.39, and rAAVrh.43 (ranked by transduction efficiency in this structure, Table 1 and Figures 1,2). In addition to the neuronal transduction pattern, we also observed EGFP-positive cells with morphologic appearance of astrocytes (Figure 2). This was further confirmed by double staining with antibodies against EGFP and astrocytic marker as described below. For intravenously delivered rAAV1, rAAV6, and rAAV6.2 vectors there were only small numbers of EGFP-positive cells in the hippocampus (Figure 2).

Cortex. AAV7, AAV9, AAVrh.10, AAVrh.39, and AAVrh.43 vectors achieved moderate EGFP transduction in cortex (Table 1 and Figures 1,2). The morphology of transduced cells was consistent with both neurons and astrocytes as further confirmed by cellular marker staining and confocal microscopic analysis described below. Prominent EGFP-positive cells were typically observed in the ventrolateral regions of the cortex, including posterior agranular insular cortex, piriform cortex, lateral entorhinal cortex, posterolateral cortical amygdaloid nucleus, and posteromedial cortical amygdaloid nucleus (Figure 2). Strong EGFP signals spread from +1.5 to −3.3 mm in relation to the Bregma (0.0 mm; data not shown). The cortical transduction efficiency of rAAVrh.10, rAAV9, rAAVrh.39, and rAAVrh.43 was comparable (Table 1 and Figures 1,2). AAV1, AAV6, and AAV6.2 vectors also transduced cells in the cortex, albeit at considerably lower levels than the rAAV vectors mentioned above (Figure 2).

Hypothalamus. As indicated above, the most impressive EGFP signal was observed in the hypothalamus for all eight effective vectors. Intravenous administration of rAAVrh.10 resulted in EGFP expression in the entire hypothalamus, followed by rAAVrh.39, rAAV7, rAAV6.2, rAAVrh.43, rAAV9, rAAV1, and rAAV6 (Figures 1,2 and Table 1). Interestingly most EGFP-positive cells in this structure have an astrocytic morphology which was ascertained by immunostaining for an astrocytic cell type specific marker as described below. The extremely strong and widespread astrocytic EGFP signal tended to obscure direct examination of morphological details of other transduced cells. However, this was clarified by double immunofluorescent staining of tissue sections with antibodies for EGFP and neuronal cell markers as described below.

Cerebellum. EGFP-positive cells and fibers were easily detected in cerebellum for all rAAV vectors except for AAV2 and AAV5 (Table 1 and Figures 1,2). A large number of EGFP-expressing cells were found in the Purkinje and granule cell layers for rAAV7, rAAV9, rAAVrh.10, rAAVrh.39, and rAAVrh.43 (Figure 2). Interestingly, the transduction profile of rAAV1 vector was restricted to cells in the granule cell layer, while rAAV6 and rAAV6.2 were localized in cells in the Purkinje cell layer (Figure 2).

Medulla. As above, all rAAVs, except for rAAV2 and rAAV5, mediated moderate to robust EGFP expression in medulla with most green cells being present in the outer rim (Figure 2). Transduction efficiencies of these rAAV in this region are ranked in the following order: rAAVrh.39 = rAAVrh.43 > rAAV.rh10 > rAAV1 > rAAV9 > rAAV7 > rAAV6.2 > rAAV6 (Table 1 and Figure 1a). The morphology of most EGFP-transduced cells was consistent with astrocytes (please see below).

Spinal cord. rAAVrh.10, rAAV9, rAAVrh.39, and rAAVrh.43 gave rise to very robust EGFP expression in cervical gray and white matter, while rAAV1, rAAV6.2, and rAAV7 showed moderate EGFP expression (Table 1 and Figures 1,3). For rAAV1 the EGFP signal was observed only in white matter. The transduction ability of all effective rAAVs decreased from cervical to lumbar spinal cord, although EGFP-positive cells were still visible in the latter region. Large populations of EGFP-positive cells with astrocytic morphology were observed throughout the spinal cord (Figure 3). In addition, rAAVrh.10, rAAV9, rAAVrh.39, rAAVrh.43, and rAAV7 also transduced cells with motor neuron morphology in the ventral regions of spinal cord (Figure 3). Ascending dorsal column fibers showed clear EGFP signal. In addition, dorsal root ganglia displayed remarkable transduction with strong EGFP expression in dorsal root ganglia neurons (Figure 4 and Supplementary Figure S2). The identities of rAAV transduced cell types in the spinal cord were characterized by co-immunofluorescence staining with antibodies against EGFP and cell type specific markers as described below.

Figure 3.

EGFP expression in mouse spinal cord after neonatal intravenous injection of rAAVs. 4 × 10¹¹ genome copies of rAAVs were injected into neonatal postnatal day 1 (P1) pups, and distribution of EGFP expression in the spinal cord was analyzed at 21 days postinjection. Forty micrometer thick cryosections from cervical, thoracic and lumbar regions were stained with an anti-EGFP antibody for immunofluorescence. Representative sections are shown for each rAAV. Bars = 100 µm. EGFP, enhanced green fluorescent protein; rAAV, recombinant adeno-associated virus.

Figure 4.

EGFP expression in dorsal root ganglia after neonatal intravascular infusion of rAAVs. EGFP expression in dorsal root ganglia neurons was analyzed by double immunofluorescence staining for EGFP (green) and a neuronal-specific marker (NeuN, red). Bars = 75 µm. EGFP, enhanced green fluorescent protein; rAAV, recombinant adeno-associated virus.

Intravenous administration of AAV vectors leads to transduction of different cell types in the CNS

To confirm the identity of transduced cells in different regions of the CNS, we performed double immunofluorescent staining with antibodies for EGFP and NeuN (generic neuronal marker), glial fibrillary acid protein (GFAP; astrocyte marker), calbindin-D28K (Purkinje cell marker), choline acetyl transferase (ChAT; motor neuron marker), and tyrosine hydroxylase (TH; marker of dopaminergic neurons) (Figure 5 and Supplementary Figure S3). The immunostaining results showed that a large number of NeuN positive cells expressed EGFP throughout the mouse brain, which indicated widespread neuronal transduction. The regions with high density of transduced neurons included striatum, hippocampus, cortex, and hypothalamus. rAAVrh.10, rAAV9, rAAV7, and rAAVrh.39 vectors are very efficient in mediating neuronal transduction, followed by rAAV6.2, rAAV1, and rAAV6 (Figures 2,5 and data not shown). The most abundant transduced cells throughout the CNS were GFAP-positive astrocytes with small cell bodies and highly ramified processes (Figure 5). The calbindin-D28K immunostaining confirmed the identity of a large number of transduced cells in the cerebellum as Purkinje cells, with robust EGFP expression in both cell body and their tree-like processes (Figure 5). The rAAVs proficient in transducing Purkinje cells include: rAAVrh.10, rAAV9, rAAVrh.39, rAAV7, rAAV6.2, and rAAVrh.43. rAAV1 and rAAV6 only transduced a small portion of Purkinje cells with low EGFP intensity (Figure 2). Transduction of motor neurons was confirmed by the presence of large EGFP⁺/ChAT⁺ cells in the ventral spinal cord for several rAAV vectors (Figure 5 and data not shown). rAAVrh.10, rAAV9, rAAV7, rAAVrh.39 showed comparable high efficiency transduction of motor neurons (Figure 3). TH⁺ dopaminergic neurons in the substantia nigra were also transduced (Figure 5 and Supplementary Figure S3).

Figure 5.

Analysis of transduced cell phenotype in mouse CNS after neonatal intravascular delivery of rAAVs. Forty micrometer thick sections of brain and spinal cord from AAV-injected mice were stained for immunofluorescence using antibodies against EGFP and cell-type specific markers for neurons [neuronal-specific marker (NeuN)], astrocytes [glial fibrillary acid protein (GFAP)], cerebellar Purkinje cells (calbindin-D-28k), spinal cord motor neurons [choline acetyl transferase (ChAT)], and dopaminergic neurons in the sustantia nigra [Tyrosine hydroxylase (TH)]. All rAAVs were examined, but for each cell type, only one representative picture is shown. Scale bars sizes are indicated in each picture. AAV, adeno-associated virus; CNS, central nervous system.

Intravenous administration of AAV vectors mediated robust transduction in ventricles and brain blood vessels

Strong EGFP expression was observed in choroid plexus cells in lateral, 3rd and 4th ventricles of the animals infused with rAAVrh.39, rAAVrh.10, rAAVrh.43, rAAV7, and rAAV9 (ranked by transduction efficiency, Table 1 and Figure 6). Ependymal cells lining the ventricles were also transduced. An interesting observation regarding the distribution of EGFP-positive cells was the apparent gradient with the highest number of transduced cells in periventricular regions and progressively lower numbers with increasing distance to the ventricles. This phenomenon was more obvious in areas around the 3rd and 4th ventricles than the lateral ventricles (Figure 6). Extensive EGFP signal was also found associated with blood vessels throughout the brain and spinal cord. This was verified by dual immunofluorescent staining with antibodies directed to EGFP and a blood vessel endothelium specific marker, CD34²³ (Supplementary Figure S4a,b). Unlike the distinct rAAV transduction profiles in different regions of the brain parenchyma, the EGFP transduction of blood vessels throughout the CNS seemed to be quite uniform for any given vector. However, as observed in the brain parenchyma, capabilities of different rAAVs to transduce blood vessels were variable. While rAAV2 produced limited blood vessel transduction and rAAV5 resulted in almost no transduction, all other rAAVs mediated moderate (e.g., rAAV6) to highly efficient (e.g., rAAVrh.10) blood vessel transduction in the CNS (data not shown).

Figure 6.

Efficient transduction of brain ventricular structures after neonatal intravascular delivery of recombinant adeno-associated virus (rAAVs). The choroid plexus appears to be transduced at high efficiency for multiple rAAV. Bars = 100 µm.

Intravenous injection of AAV vectors did not cause microgliosis

Brain sections were also stained with antibody against Iba-1 to label microgial cells. The level of Iba-1-positive staining in the brain of rAAVrh.10-infused mice (same observation for the other rAAV vectors; data not shown) was comparable to that in phosphate-buffered saline (PBS)-injected or naive mice (Supplementary Figure S5). This result suggested that intravascularly delivered rAAVs do not cause sustained inflammation in the CNS of mice 3 weeks after the injection of P1 neonates. However, it remains to be determined if facial vein injection of rAAVs caused any acute microgliosis in the CNS at the neonatal stage since the present analysis was performed 3 weeks later.

Discussion

In this study, we have evaluated the CNS transduction profile of 10 different rAAV vectors delivered by intravascular infusion in neonatal mice. Eight out of ten rAAVs can cross the BBB and mediate gene transfer to the neonatal mouse CNS with varying degrees of efficiency (Figures 1,2,3 and Table 1). After systemic administration, rAAVrh.10, rAAVrh.39, rAAVrh.43, and rAAV9 are the most effective rAAVs with similar transduction capabilities and cellular tropism, as assessed by overall EGFP expression in the CNS. Specifically, a number of regions in the mouse CNS, including striatum, hippocampus, cortex, hypothalamus, cerebellum, medulla, and cervical spinal cord, all revealed substantial EGFP expression. In addition, rAAV6.2 and rAAV7 were also very effective, but the efficiencies of EGFP transduction in the targeted regions were consistently lower than those obtained with the four top vectors. Still further down the spectrum were rAAV1 and rAAV6, which did achieve some CNS transduction, but their efficiency was quite low. Finally, rAAV2 and rAAV5 revealed little or no CNS gene delivery after systemic administration (Table 1). It is worth noting that native EGFP expression was clearly detectable in brain and spinal cord sections for most of the rAAVs without immunostaining, which further demonstrates the robustness and extensiveness of CNS transduction in our study (Supplementary Figure S6).

Our findings hold considerable clinical significance for gene therapy of CNS-related disorders, especially for young patients. For a variety of neurological diseases, treatment during infancy will be necessary to prevent irreversible CNS injury. The capacity of rAAVs to transduce large numbers of neuronal cells in different regions may be relevant for treating neurological diseases such as spinal muscular atrophies,²⁴ neuronal ceroid lipofuscinoses,²⁵ and spinocerebellar degenerations.²⁶ In fact, the remarkable efficiency of some rAAV vectors in transducing Purkinje and granule layer cells opens interesting and promising possibilities to develop new therapies for spinocerebellar ataxias. Transduction of astrocytes by rAAVs expressing secreted neurotrophic factors may be also beneficial for a number of neurodegenerative diseases such as Canavan's disease²⁷ and amyotrophic lateral sclerosis.²⁸ Furthermore, the broad vascular transduction in the CNS may be relevant for treating brain ischemia and stroke.²⁹ The potential clinical application of intravascular rAAV-mediated gene delivery may also extend to the peripheral nervous system.³⁰ Efficient transduction of dorsal root ganglia may open new therapeutic possibilities for patients suffering from chronic pain.³¹ From any of these perspectives, transduction patterns generated from our study may serve as initial proof-of-concept data for diverse neurological disorders.

The utility of these newer AAVs is not confined to clinical application; systemic gene delivery to the CNS should also be useful as a convenient method to manipulate gene expression in the course of basic neuroscience research. Effective and stable transgene expression in the CNS by intravenous administration of rAAVs may be applied to establish somatic transgenic animal models, which is a potentially cheaper, faster, and simpler method than conventional transgenesis. Somatic CNS gene knockdown animal models may also be created using the method described here.³²

In addition to the global delivery properties of several rAAVs described above, some rAAVs indeed demonstrated unique transduction profiles in the CNS. For instance, rAAV1 displayed an interesting property in that it transduced primarily granule cells in the cerebellum, while rAAV6 and rAAV6.2 transduced mostly Purkinje cells, and yet others transduced both types of cells (Figure 2). This suggests that once across the BBB, the rAAVs display different tropisms, which can be attributed to the capsid given that the vector genome used in all vectors was the same.

The fact that different AAV serotypes can efficiently transduce brain capillary endothelial cells, neurons, and astrocytes strongly suggest that these vectors are able to extravasate from the circulation and reach the CNS parenchyma, possibly by crossing the BBB.³³ Although the exact molecular mechanism remains unknown, one view is that AAV crosses the endothelial barrier by a transcytosis pathway.³⁴ In our study, choroid plexuses and their surrounding parenchymal tissue were efficiently transduced. In addition, there was an apparent gradient of EGFP intensity from periventricular (higher) to deep parenchymal (lower) tissue. These observations may suggest that an alternative route for AAV for entering the neonatal mouse CNS could be through the choroid plexus,³⁵ followed by widespread distribution via CSF and/or interstitial fluid flow to transduce neuronal and glial cells.

Global and efficient gene transfer to CNS by systemic delivery of rAAVs indeed holds great promise for developing new treatments for devastating CNS disorders. However, the most effective vectors for CNS gene delivery also retain strong natural liver tropism. Inevitably, systemic administration of those rAAVs will generate robust liver transduction and widespread gene transfer into other tissues including skeletal muscle, heart, pancreas, and even antigen presenting cells. When a strong ubiquitous promoter such as a hybrid promoter carrying the chicken β-actin promoter fused to a portion of the CMV I/E enhancer is used to drive robust transgene expression in the CNS, off-target over-expression of transgenes from rAAVs in the peripheral tissues could potentially elicit untoward toxic and immunological responses. Thus, intravascular delivery of high-dose rAAVs may entail safety concerns. Presumably, neuronal- or glial-specific promoters, such as synapsin-1, and GFAP promoters may be used to restrict gene expression to a specific cell type but at lower levels.³⁶ An alternative method to achieve targeted CNS gene delivery is to utilize the power of RNA interference to detarget the peripheral tissues by post-transcriptional regulatory mechanisms. By adding microRNA binding sites into the 3′ end of the transgene cassettes, we have successfully spared liver, heart, and skeletal muscle from transgene expression after systemic administration of AAV vectors, whereas the CNS transduction was largely unaffected.³⁷

Materials and Methods

AAV production. All scAAV vectors were produced by trans-encapsidation of rAAV vector genome flanking by inverted terminal repeats from AAV2 with the capsids of different AAVs using the method transient transfection of 293 cells and CsCl gradient sedimentation as previously described.³⁸ Vector preparations were titered by quantitative PCR. Purity of vectors was assessed by 4–12% SDS-acrylamide gel electrophoresis and silver staining (Invitrogen, Carlsbad, CA). Morphological integrity of each vector used in the study was examined by transmission electron microscopy of negative stained recombinant AAV virions at Electronic Microscopy Core, University of Massachusetts Medical School, Worcester, MA, USA. The expression of EGFP in the scAAV vector genome is directed by ubiquitous hybrid CMV enhancer/chicken β-actin promoter.

Neonatal mouse injections. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Massachusetts Medical School. Wild-type C57BL/6 mice littermates were used. Mice breeding were conducted using programmatic timing method. Pregnant mice were monitored daily from embryonic day 17 to 21 to ensure the newborn pups could be dosed with vectors on P1. The mother (singly housed) of each litter to be injected was removed from the cage. Vectors were diluted to concentration of 4 × 10¹² genome copies/ml in PBS and 100 µl of solution was subsequently drawn into 31G insulin syringes (BD Ultra-Fine II U-100 Insulin Syringes). P1 pups of C57BL/6 mice were anesthetized using isoflurane. For intravenous injections, a dissection microscope was used to visualize the superficial temporal vein (located just anterior to the ear). The needle was inserted into the vein and the plunger was manually depressed. Correct injection was verified by noting blanching of the vein. Each pup received 4 × 10¹¹ genome copies of different scAAVCBEGFP vectors (rAAV1, rAAV2, rAAV5, rAAV6, rAAV6.2, rAAV7, rAAV9, rAAVrh.10, rAAVrh.39, rAAVrh.43; n = 6–8 mice per group). After the injection pups were carefully cleaned, rubbed with their original bedding, and then returned to their original cage. The mother was then reintroduced to the cage after brief nose numbing using ethanol pads.

Histological processing. The study animals were anesthetized 21 days postinjection, then transcardially perfused with 15 ml of cold PBS followed by 15 ml of fixation solution containing 4% paraformaldehyde (v/v) with 0.2% of glutaraldehyde (v/v) in PBS. Then the whole carcasses were postfixed in fixation solution for 5 days. Spinal cords and brains were extracted under a bright-field dissecting microscope, rinsed in PBS, and then cryoprotected in 30% sucrose (w/v) in PBS at 4 °C. Once the tissues sank to the bottom of the sucrose solution, they were embedded in Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA) and frozen in a dry ice/ethanol bath. The tissue blocks were stored at −80 °C until sectioning. Serial 40 µm floating sections of the entire brain were cut in a Cryostat (Thermo Microm HM 550, Thermo Scientific, Kalamazoo, MI). For the spinal cord, 3 mm length sections were taken from cervical, thoracic and lumbar regions, and then serial 40 µm transverse sections prepared as above.

Immunostaining and microscopy imaging analysis. Brain and spinal cord sections were stained as floating sections in 12-well plates. Sections were washed three times in PBS for 5 minutes each time, and then incubated in blocking solution containing 1% Triton-X100 (v/v) (Fisher, Pittsburgh, PA), 5% dry-milk (w/v), and 10% goat serum (v/v) (Invitrogen) for 2 hours at room temperature. Then the sections were incubated with primary antibodies diluted in blocking solution at 4 °C overnight. The following day tissue sections were washed twice in 0.05% Tween-20 (v/v) in PBS (PBST) and once with PBS, with each washing step lasting 10 minutes. Afterwards sections were incubated with appropriate secondary antibodies in blocking solution at room temperature for 2 hours. Sections were washed again as above before mounting on glass slides. Vectashield with 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA) was used to coverslip all slides, and then they were analyzed using a fluorescent inverted microscope (Nikon Eclipse Ti, Nikon Instrument, Melville, NY) or a Leica TSC-SP2 AOBS confocal microscope equipped with a ×63 oil lens and a DM-IRE2 inverted microscope. The primary antibodies used in this study were as follows: rabbit anti-GFP (Invitrogen), goat anti-ChAT and mouse anti-NeuN (both from Millipore, Billerica, MA), mouse anti-GFAP (Cell signaling, Danvers, MA), rat anti-CD34 (Abcam, Cambridge, MA), mouse anti-Calbindin D-28k (Sigma, St Louis, MO) and mouse antityrosine hydroxylase monoclonal antibody (Millipore). The secondary antibodies used in the study included: DyLight 488 AffiniPure Donkey antirabbit IgG (Jackson ImmunoResearch, West Grove, PA); DyLight 549 AffiniPure Donkey Anti-Goat IgG (Jackson ImmunoResearch); DyLight 549 Affinipure Goat antirat IgG (Jackson ImmunoResearch); DyLight 594 AffiniPure Goat antimouse IgG (Jackson ImmunoResearch); goat antirabbit IgG-Alexa fluro 488 (Invitrogen) and goat antimouse IgG-Alexa fluro 568 (Invitrogen).

Semi-quantitative and quantitative comparison of EGFP transduction by different vectors. To generate a quantifiable and comparable data format, we first used a semi-quantitative scoring system analogous to that developed by Cearley et al. to estimate transduction efficiency of different rAAV vectors in different regions of the mouse CNS.³ Briefly, regions with no EGFP-positive cells were marked as (−). Regions with very few EGFP-positive cells were scored (+), regions with some EGFP-positive cells were ranked as (++), regions with many EGFP-positive cells were marked as (+++). Finally, regions filled with EGFP-positive cells were marked as (++++).

Next, we selected 12 subanatomically and functionally important regions in the brain as well as cervical, thoracic, and lumbar sections of the spinal cord for quantitative analysis of images that were taken on a Nikon Eclipse Ti inverted microscope equipped with a Retiga 2000-RV CCD cooled camera. Nikon NIS elements AR software version 3.2 was used for intensity quantification. Prior to quantification, optimal light source intensity and exposure times were obtained by plotting an intensity/exposure time curve using fluorescence reference slides (Ted Pella, prod. 2273; Ted Pella, Redding, CA). It was found that the intensity and exposure times had linear correlation. In addition, overexposure and extreme underexposure distorts the linear correlation. We have therefore used the maximum intensity (ND1) and a 20 ms exposure for all sections to avoid overexposure. For quantification, fixed region of interest was used to quantify the brightest area of any given brain region. A mean intensity (total intensity/size of region of interest) was obtained for each region of all serotypes.

SUPPLEMENTARY MATERIAL Figure S1. Analysis of purity and morphological integrity of rAAV vectors. A. Silver stained SDS-PAGE analysis of CsCl gradient purified rAAVCBEGFP vectors used in this study. Approximately 1.5 x 10¹⁰ virus particles each of rAAVs 1, 2, 5, 6, 6.2, 7, 9, rh10, rh39 and rh43 were loaded in the corresponding lane. B. Transmission electron microscopy of negative stained recombinant AAV virions. rAAV virions were spread on a freshly prepared carbon coated- Formvar support film and stained with 1% uranyl acetate for transmission microscopy. The images of virus particles from representative vector lots were taken at 92,000X and presented. Figure S2. Transduction of neonatal mouse dorsal root ganglia by systemically delivered rAAVs 1, 6, 6.2 and rh43. Neonatal P1 pups received 4×10¹¹ GCs of rAAVs and EGFP expression in dorsal root ganglia analyzed at 21 days post-injection. Forty μm thick cryosections were stained with anti-EGFP antibody for immunofluorescence. Scale bars represent 75 μm. Figure S3. Transduction of dopaminergic neurons in the substantia nigra of neonatal mice by systemically delivered rAAVrh.10. Neonatal P1 pups received 4×10¹¹ GCs of rAAVs and EGFP transduction of dopaminergic neurons analyzed at 21 days post-injection. Forty μm thick cryosections were stained by immunofluorescence with anti-tyrosine hydroxylase (TH, Red) and EGFP (Green) antibodies. The scale bar in the merged image is 50 μm. Figure S4. Transduction of the brain capillary vessels by intravascularly delivered rAAVs. Neonatal P1 pups that received 4×10¹¹ GCs of rAAVs were sacrificed 21 days after injection. Forty μm thick cryosections of the brains were stained with: (a) anti-EGFP antibody (rAAV1, rAAV6, rAAV6.2, rAAV7, rAAV9, rAAVrh.10, rAAVrh.39 and rAAVrh.43); (b) anti-EGFP and anti-CD34 antibodies (rh.10 only) for immunofluorescence. Scale bars represent 100 μm. Figure S5. Evaluation of microgliosis in mice brain after systemic delivery of rAAVs to P1 neonates. Brain sections of animals treated with different rAAVs (and controls) were stained with anti-Iba-1 antibody for immunofluorescence. Shown are representative Iba-1 stained brain sections from naïve, PBS-injected and rAAVrh10-injected mice. Figure S6. Native EGFP expression in mice CNS after systemic delivery of rAAVs to P1 neonates. Neonatal P1 pups received 4×10¹¹ GCs of rAAVs and EGFP expression analyzed at 21 days post-injection. Forty μm thick cryosections were mounted on slides and EGFP expression observed without immunostaining. The exposure times for each image are indicated.

Supplementary Materials

Figure S1.

Analysis of purity and morphological integrity of rAAV vectors. A. Silver stained SDS-PAGE analysis of CsCl gradient purified rAAVCBEGFP vectors used in this study. Approximately 1.5 x 10¹⁰ virus particles each of rAAVs 1, 2, 5, 6, 6.2, 7, 9, rh10, rh39 and rh43 were loaded in the corresponding lane. B. Transmission electron microscopy of negative stained recombinant AAV virions. rAAV virions were spread on a freshly prepared carbon coated- Formvar support film and stained with 1% uranyl acetate for transmission microscopy. The images of virus particles from representative vector lots were taken at 92,000X and presented.

Figure S2.

Transduction of neonatal mouse dorsal root ganglia by systemically delivered rAAVs 1, 6, 6.2 and rh43. Neonatal P1 pups received 4×10¹¹ GCs of rAAVs and EGFP expression in dorsal root ganglia analyzed at 21 days post-injection. Forty μm thick cryosections were stained with anti-EGFP antibody for immunofluorescence. Scale bars represent 75 μm.

Figure S3.

Transduction of dopaminergic neurons in the substantia nigra of neonatal mice by systemically delivered rAAVrh.10. Neonatal P1 pups received 4×10¹¹ GCs of rAAVs and EGFP transduction of dopaminergic neurons analyzed at 21 days post-injection. Forty μm thick cryosections were stained by immunofluorescence with anti-tyrosine hydroxylase (TH, Red) and EGFP (Green) antibodies. The scale bar in the merged image is 50 μm.

Figure S4.

Transduction of the brain capillary vessels by intravascularly delivered rAAVs. Neonatal P1 pups that received 4×10¹¹ GCs of rAAVs were sacrificed 21 days after injection. Forty μm thick cryosections of the brains were stained with: (a) anti-EGFP antibody (rAAV1, rAAV6, rAAV6.2, rAAV7, rAAV9, rAAVrh.10, rAAVrh.39 and rAAVrh.43); (b) anti-EGFP and anti-CD34 antibodies (rh.10 only) for immunofluorescence. Scale bars represent 100 μm.

Figure S5.

Evaluation of microgliosis in mice brain after systemic delivery of rAAVs to P1 neonates. Brain sections of animals treated with different rAAVs (and controls) were stained with anti-Iba-1 antibody for immunofluorescence. Shown are representative Iba-1 stained brain sections from naïve, PBS-injected and rAAVrh10-injected mice.

Figure S6.

Native EGFP expression in mice CNS after systemic delivery of rAAVs to P1 neonates. Neonatal P1 pups received 4×10¹¹ GCs of rAAVs and EGFP expression analyzed at 21 days post-injection. Forty μm thick cryosections were mounted on slides and EGFP expression observed without immunostaining. The exposure times for each image are indicated.