AAV8, 9, Rh10, Rh43 vector gene transfer in the rat brain: effects of serotype, promoter and purification method

Abstract

We compared adeno-associated virus (AAV) serotypes for expression levels of green fluorescent protein (GFP) in the adult rat hippocampus by biophotonic imaging. Preparations of AAV serotypes 8, 9, Rh10, and Rh43 incorporating cytomegalovirus (CMV) promoter-driven GFP were purified by a CsCl method. Neither AAV Rh10 nor AAV Rh43 produced greater levels of GFP than AAV8, which was used as a reference. For AAV9, there was an increase relative to AAV8. The CsCl-purified AAV8 displayed an astroglial transduction pattern in contrast to the expected neuronal expression of other AAVs. After preparing the same CMV-GFP plasmid in AAV8 with an iodixanol purification method, the expected neuronal pattern resulted. The astroglial expression with the CsCl AAV8 was probably due to relatively high levels of protein impurities. We compared the CMV promoter with the CMV/chicken beta-actin (CBA) promoter in the context of AAV8, both prepared by iodixanol, and found the CBA promoter to produce stronger GFP expression. At two doses of vectors optimized for serotype, promoter and purification, we did not observe serotype differences among AAV8, AAV9, or AAV Rh10. The purification method can therefore impact the transduction pattern as well as the results when comparing serotype strengths.

Figure 1. Green fluorescent protein (GFP) biophotonic imaging

(a) Serotype comparison of AAV8, AAV9, AAV10, AAV43 GFP vectors at a dose of 1.3 × 10¹¹ vector genomes (vg) and expression interval of 4 weeks. This set of vectors used the cytomegalovirus (CMV) promoter and CsCl purification. Two brains per group are shown. There was greater spread of epifluorescence with AAV9 relative to AAV8 (P < 0.05) or AAV43 (P < 0.001) and with AAV10 realtive to AAV43 (P < 0.05) in analysis of variance (ANOVA)/Bonferroni′s multiple comparison test (N = 8/serotype). (b) Promoter comparison with iodixanol-purified vectors. AAV8 GFP vectors with either the CMV promoter or the CMV/chicken β-actin (CBA) hybrid promoter at a dose of 7.8 × 10⁹ vg and expression interval of 3 weeks. There was 6.4-fold greater spread of GFP expression with the CBA promoter (P < 0.005, N = 4/promoter, t-test). (c) Serotype comparison of AAV8, AAV9, AAV10 GFP vectors with the CBA promoter and iodixanol purification at two doses and an interval of 3 weeks. The low dose was 2.6 × 10⁹ vg and the high dose was 7.8 × 10⁹ vg. There was a dose effect (P < 0.0001) although no serotype effect in a two-way ANOVA (N = 4–7/dose/serotype). However, there was a trend for more consistent expression with AAV9 at the low dose; see Results. AAV, adeno-associated virus.

Figure 2. Hippocampal green fluorescent protein (GFP) levels on Western blots

AAV9, AAV10, or AAV43 were compared to AAV8 (N = 5–7/serotype as indicated). All injections used a dose of 1.3 × 10¹¹ vector genome and expression interval of 4 weeks; 35 µg protein loaded in each lane. (a) AAV9 versus AAV8: GFP and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) immunoblots; the GFP signal was normalized to GAPDH for comparisons. There was 92% greater GFP expression from AAV9 relative to AAV8 (P < 0.05, t-test). (b) AAV10 versus AAV8. (c) AAV43 versus AAV8. No difference in b or c. AAV, adeno-associated virus.

Figure 3. Green fluorescent protein (GFP) epifluorescence on hippocampal sections

View of hippocampus with ×10 lens for all vectors tested in the study. (a–d), cytomegalovirus/woodchuck hepatitis virus post-transcriptional regulatory element (CMV/WPRE) GFP adeno-associated virus (AAV) vectors purified by CsCl, vector dose in a–d was 1.3 × 10¹¹. The AAV8 in a led to an unusual diffuse GFP expression pattern that was not observed with the other vectors, which normally resulted in labeling neuronal positions in CA1 and dentate gyrus (DG). (e) CMV/WPRE GFP AAV8 purified by iodixanol, vector dose in f–h was 7.8 × 10⁹ vector genome. f–h, chicken β-actin (CBA)/WPRE GFP AAV vectors purified by iodixanol, same vector dose as e. All with 4 week expression intervals. Serotypes are indicated. There was more complete transduction of pyramidal neurons in CA1 (above the p) and of granule cells (g) in the DG with (f) CBA AAV8 and (g) AAV9 purified by iodixanol. a–h, bar = 134 µm.

Figure 4. Non-neuronal/astroglial expression with the CMV-GFP AAV8 purified by CsCl

(a,b) The largely non-neuronal transduction pattern with the AAV8 vector purified by CsCl was confirmed by neuronal (NeuN, red) counterstaining. Green fluorescent protein (GFP) positive cells were found in the white matter, corpus callosum a, and along the needle track b which did not co-localize with NeuN. (c) Higher magnification of GFP expression from the vector in the AAV8 vector in CA1 of hippocampus. (d) The same section as c showed robust counterstaining for astroglia [glial fibrillary acidic protein (GFAP), red]. (e) Merger of d and e shows GFP/GFAP co-localization (arrows). (f) In contrast, the batch of AAV9 also purified by CsCl, did not result in pronounced astroglial staining as in d. (g) Merger of GFP expression and GFAP in f showed a non-astrocytic pattern for the AAV9. (h) The AAV9-derived GFP did co-localize with the neuronal marker NeuN (red) in CA1 on a merged image from an adjacent section as f,g. All at a vector dose of 1.3 × 10¹¹ vector genome and a 4 week interval. a,b, bar = 8.4 µm; c–h, bar = 5.2 µm. AAV, adeno-associated virus; CMV, cytomegalovirus.

Figure 5. Needle track damage with AAV9 iodixanol vector

(a) Injection of the vehicle diluent resulted in minor astrocytic scarring; glial fibrillary acidic protein (GFAP) staining is in red. (b) Injection of AAV9 vector purified by iodixanol. There was greater density of astrocytic staining in the area of the injections of the AAV vectors than as in a. Images in a and b were captured with equal camera settings. (c) Green fluorescent protein (GFP) expression on same section as b shows peak physical damage on the needle track, but the highly efficient GFP expression did not overlap with GFAP. The majority of the CA1 pyramidal neuron layer and dentate gyrus granule neuron layer were transduced in the injected area. The vector dose, 7.8 × 10⁹ vector genome, and a 4 week expression interval. a–c, bar = 134 µm. AAV, adeno-associated virus.

Figure 6. Protein analyses of adeno-associated viruses (AAVs) directly loaded onto gels

(a) A fixed amount of AAV [(1.3 × 10¹¹ vector genome (vg)] was run on sodium dodecyl sulfate polyacrylamide gel electrophoresis, transferred to a membrane and stained for proteins with Ponceau S. There were more proteins present in the CsCl preps relative to the iodixanol preps, and among the CsCl preps, the AAV8 had the highest amount of proteins. (b) Green fluorescent protein (GFP) immunoblot of CsCl vectors. The vg load for this blot was: AAV8, 5.3 × 10¹¹; AAV9, 6.0 × 10¹¹; AAV10, 1.1 × 10¹²; AAV43, 1.0 × 10¹². The CsCl AAV8 prep appeared to have the most GFP despite the lowest vg loaded. Similar results were found with equivalent vg loading of 1.3 × 10¹¹, while GFP bands were never detected in iodixanol preps in any of the serotypes tested with up to 2.5 × 10¹¹ vg loaded. (c) Antibody B1 immunoblot for AAV capsid proteins. Both CsCl and iodixanol preps showed the expected sizes and stoichiometry for VP1, VP2, VP3 proteins. Equal loading of 1.3 × 10¹¹ vg.

Figure 7. Pronounced spread of hippocampal gene transfer with AAV9 and AAV10

The green fluorescent protein (GFP) expression with the CBA-GFP AAV9 (a–d) and AAV10 (e–h) purified by iodixanol was robust and well targeted to the hippocampus. The spread posteriorally with AAV9 and AAV10 was to a greater extent than previously observed with other serotypes. (a–c) AAV9, serial hippocampal sections from front to back. The section in c is 2 mm from the needle track and contains efficient transduction of the dentate gyrus granule layers. (b,f) Sections with 4′,6-diamidino-2-phenylindole nuclear counterstain demonstrates selective hippocampal targeting. (d,h) Sections from contralateral hippocampus showing anterograde GFP axons projecting from ipsilateral side. Throughout the study, the AAV9 more successfully transduced the pyramidal neuron layer than did the AAV10 (a–c compared to e–g). The vector dose, 7.8 × 10⁹ vector genome, and a 3 week expression interval. a–h, bar = 535 µm. AAV, adeno-associated virus; CBA, chicken β-actin.