Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 109 (20), E1238-47

Gene Delivery to Mitochondria by Targeting Modified Adenoassociated Virus Suppresses Leber's Hereditary Optic Neuropathy in a Mouse Model

Affiliations

Gene Delivery to Mitochondria by Targeting Modified Adenoassociated Virus Suppresses Leber's Hereditary Optic Neuropathy in a Mouse Model

Hong Yu et al. Proc Natl Acad Sci U S A.

Abstract

To introduce DNA into mitochondria efficiently, we fused adenoassociated virus capsid VP2 with a mitochondrial targeting sequence to carry the mitochondrial gene encoding the human NADH ubiquinone oxidoreductase subunit 4 (ND4). Expression of WT ND4 in cells with the G11778A mutation in ND4 led to restoration of defective ATP synthesis. Furthermore, with injection into the rodent eye, human ND4 DNA levels in mitochondria reached 80% of its mouse homolog. The construct expressed in most inner retinal neurons, and it also suppressed visual loss and optic atrophy induced by a mutant ND4 homolog. The adenoassociated virus cassette accommodates genes of up to ∼5 kb in length, thus providing a platform for introduction of almost any mitochondrial gene and perhaps even allowing insertion of DNA encompassing large deletions of mtDNA, some associated with aging, into the organelle of adults.

Conflict of interest statement

Conflict of interest statement: W.W.H. and the University of Florida have a financial interest in the use of AAV therapies, and own equity in a company (AGTC Inc.) that might, in the future, commercialize some aspects of this work.

Figures

Fig. 1.
Fig. 1.
Nucleic acid analysis of infected LHON cell cultures. (A) Illustration shows that amino acids with their respective codons for human ND4 in the mitochondrial genetic code would result in a stop codon at amino acid 16 in the nuclear genetic code but not in the mitochondria, where this same TGA codon encodes for the amino acid tryptophan. For ND4 expression in the cytoplasm, the TGA codon must be changed to TGG, which encodes for tryptophan in the nucleus (allotopic expression). (B) Following 5 d of selection in restrictive media and 2 wk of growth in normal media, homoplasmic G11778A cybrids infected with WT ND4 packaged with COX8GFP VP2 exhibited better growth in standard glucose media compared with those infected with AAV VP2 lacking the MTS. (C) With a sense PCR primer flanking the H-strand promoter and antisense primer flanking the 3′end of ND4, the amplificant of endogenous mouse mtDNA would be greater than 10 kb. (D) In contrast, with our construct, amplification of the H-strand promoter adjacent to human ND4 would be ∼1.4 kb. (E) Agarose gel electrophoresis of mtDNA isolated from COX8GFP VP2 (lane 2) or VP2 transfected G11778A cells (lane 3) shows that in addition to endogenous mtDNA (16 kb), a smaller band likely representing our mitochondria-targeted HSP-ND4FLAG was only seen with COX8GFP VP2 transfection. Lane 1 is the molecular weight standards. (F) With an extension time of 2 min, PCR of mtDNA isolated from homoplasmic G11778A cells infected with ND4 delivered by COX8 VP2 revealed a 1.4-kb band (arrow) several weeks after growth in restrictive media (lanes 2–5). No PCR product was obtained from the VP2-delivered ND4 lacking the MTS (lane 6). PCR amplification of ND4 plasmid DNA is shown in lane 7. Lane 1 is the molecular weight standards. (G) Illustration of WT ND4 showing the two SfaN1 restriction sites that should cut twice, with the largest digestion product being 915 bp. With the second SfaN1 site lost in mutant G11778A ND4, a larger 1.2-kb fragment would be generated by digestion with SfaN1. (H) SfaN1 digestion of the above PCR-amplified HSP-ND4 DNA revealed a 915-bp band indicating the presence of two SfaN1 sites in the PCR product (lane 2). Lane 1 is the molecular weight marker. (I) RT-PCR of RNA isolated from mitochondrial pellets performed with the forward primer nested in ND4 and the reverse primer nested in the FLAG epitope revealed the expected 500-bp band only in LHON cybrids that received the COX8 VP2 AAV (lane 2) but not in RNA isolated from cybrids that received the VP2 construct lacking the MTS (lane 3). Expression of the endogenous ATP8 gene is seen in both samples (150-bp bands in lanes 2 and 3). Lane 1 is the molecular weight standards. (J) Bar plot of quantitative RT-PCR results comparing transcription of the human ND4 (hND4) introduced by the MTS AAV with the control shows an almost fourfold elevation. The ND4 mRNA level was indicated as a ratio to that of endogenous mitochondrial human ND1 gene (hND1).
Fig. 2.
Fig. 2.
Microscopy of infected cell cultures. (A) Fluorescence microscopy of 293T cells infected with VP2 AAV containing GFP without the MTS reveals nuclear and cytoplasmic expression of GFP. (B) Cells infected with the COX8GFP VP2 MTS AAV show punctate and perinuclear expression of GFP. A light micrograph (C) shows that the A20 antibody recognized fully assembled AAV virions in a COX8GFP fluorescent cell (D) colabeled by silver-enhanced A20 immunogold (E). A neuronal cell line, with large mitochondria containing G11778A mutated mtDNA, shows perinuclear COX8GFP (F) and perinuclear ND4FLAG (G). (H) Colocalization of COX8GFP and ND4FLAG is seen in two of these cells. (I) Transmission electron micrograph of cells infected with COX8GFP VP2 AAV containing the WT mitochondrial ND4 DNA reveals that silver-enhanced GFP immunogold was present in many mitochondria (arrows). GFP immunogold was also evident in the nucleus (N) of the cell to which the virus typically translocates. (J) In a COX8GFP VP2 AAV mitochondrial ND4-infected cell, FLAG-tagged immunogold (arrows) was clearly evident within mitochondria (arrows). (K) Control stained with only the secondary antibody, conjugated to immunogold, shows no background. (L) In another MTS AAV-infected cell, perinuclear mitochondria contained the ND4FLAG immunogold particles. N, nucleus.
Fig. 3.
Fig. 3.
Microscopy and immunoblotting of MTS AAV-infected retina and optic nerve. A retinal flat mount stained with the anti–III β-tubulin antibody (A) identified RGCs in which evidence of expression of the ND4FLAG chimera was apparent (B). The ND4FLAG chimera had a perinuclear distribution surrounding the nuclei of RGCs counterstained with DAPI. (C) ND4FLAG chimera colocalized with MitoTracker Green in RGCs identified by III β-tubulin as shown in A. (D) Longitudinal retinal sections were counterstained with MitoTracker Green. ND4FLAG was evident in cells of the RGC layer (E), and it colocalized with MitoTracker Green (F). RGCs labeled by the anti–III β-tubulin antibody (G) also expressed the ND4FLAG chimera (H) that colocalized with MitoTracker Green (I). In a flat-mount preparation, RGCs labeled by the Brn3a antibody had a perinuclear distribution of ND4FLAG (J) that colocalized with mitochondria immunolabeled by an antibody against porin (K). (L) ND4FLAG chimeric protein colocalized with porin. This finding was confirmed by software showing the peaks of FLAG immunofluorescence (red) colocalized with those of porin (green). (M) Bar plot shows that the number of ND4FLAG-expressing cells increased from almost half of Thy1.2-positive RGCs at 1 wk (1WK) to almost 90% at 1 mo (1M) to 2 mo (2M) after injection. (N) Immunoblotting of ND4FLAG protein in the retina could be detected up to 6 mo after intraocular injection (lane 2). Lane 1 is the molecular weight standards. Transmission EM of lightly fixed LR White resin-embedded optic nerves revealed silver-enhanced ND4FLAG immunogold (arrows) within optic nerve mitochondria (O) that were identified by mitochondrial SOD2 immunogold (arrows) (P).
Fig. 4.
Fig. 4.
Nucleic acid analysis of the infected rodent visual system. (A) Nine days after intraocular injections, PCR assay of mtDNA extracts probed for the human ND4FLAG gave the anticipated 1.4-kb band for both the retinas and optic nerves of COX8-targeted ND4FLAG. The untargeted VP2 was also positive for ND4FLAG in retinal mitochondrial (RM) extracts but was absent in optic nerve mitochondria (OM). (B) Quantitative PCR assay showed that in the retinal mitochondrial fraction (COX8-RM), human ND4 was ∼80% of its mouse homolog at 9 d postinjection. (C) Earlier, at day 3 postinjection, human ND4 levels in mice mitochondria were much lower. (D) Alignment of the DNA sequences obtained from COX8 (COX8hybrid)- or VP2 (VP2hybrid)-injected eyes with human or mouse ND4 and the sequencing chromatographs (COX8hybrid) (E) confirmed that the COX8 MTS VP2 delivered human ND4 in the infected murine optic nerve and retina. The DNA sequences after the 3′ end of the human ND4 (COX8hybrid) are homologous to the mouse tRNA histidine. The mouse tRNA histidine is just downstream of the murine ND4 in the mouse mitochondrial genome. cox8-OM, optic nerve mtDNA isolated from eyes injected with scAAV-mND4FLAG containing the VP2 cox8MTS; cox8-ON, nuclear DNA isolated from the optic nerves eyes injected with scAAV-mND4FLAG containing the VP2 cox8MTS; cox8-RM, retinal mtDNA isolated from eyes injected with scAAV-mND4FLAG containing the VP2 cox8MTS; cox8-RN, nuclear DNA isolated from the retinas of eyes injected with scAAV-mND4FLAG containing the VP2 cox8MTS; GFP-OM, optic nerve mtDNA isolated from AAV-GFP injected eyes; GFP-ON, nuclear DNA isolated from the optic nerves of AAV-GFP–injected eyes; GFP-RM, retinal mtDNA isolated from AAV-GFP injected eyes; GFP-RN, nuclear DNA isolated from the retinas of AAV-GFP–injected eyes; hND4, human ND4; mND1, mouse ND1; mND4, mouse ND4; vp2-OM, optic nerve mtDNA isolated from eyes injected with scAAV-mND4FLAG lacking the VP2 MTS; vp2-ON, nuclear DNA isolated from the optic nerves of eyes injected with scAAV-mND4FLAG lacking the VP2 MTS; vp2-RM, retinal mtDNA isolated from eyes injected with scAAV-mND4FLAG lacking the VP2 MTS; vp2-RN, nuclear DNA isolated from the retinas of eyes injected with scAAV-mND4FLAG lacking the VP2 MTS.
Fig. 5.
Fig. 5.
MTS AAV-mediated rescue of visual loss and optic atrophy of rodents. A scatterplot of amplitudes of individual animals (line = median value) (A) and averaged waveforms (B) of PERGs done 7 wk after intraocular injections of scAAV2/COX8 VP2 containing human ND4FLAG revealed no loss of amplitude relative to uninjected eyes of normal mice. One month after intraocular injections, a scatterplot of PERG amplitudes (C) and averaged waveforms (D) revealed that loss of visual function induced by an allotopic R340H mutant ND4 was prevented by ocular injection of the COX8-targeted human ND4 AAV. At this time point, PERG amplitude for the control left eyes injected with AAV-GFP and AAV containing mutant R340H ND4 decreased by 29% compared with eyes injected with COX8-delivered WT ND4 and also injected with R340H mutant ND4 AAV (P = 0.0151). Differences between the contralateral control eyes and eyes injected with VP2-delivered WT ND4 were not significant. In another series of experiments, both eyes of the same animal received injections of WT human ND4 delivered either by mitochondria-targeted COX8GFP VP2, VP2, or AAV-GFP. All eyes of these animals were also injected with R340H mutant ND4. Three months postinjection, a scatterplot (E) and averaged waveforms (F) show that the protective effect of COX8-delivered human ND4 in preventing loss of vision relative to the GFP control animals was highly significant (P = 0.0064). Relative to these controls, differences between mice that received the VP2-delivered human ND4 were not significant (P = 0.339). Twelve months after intravitreal injections, a scatterplot (G) and averaged waveforms (H) show that differences in PERG amplitude between eyes injected with AAV-GFP (and AAV containing mutant R340H ND4) relative to COX8-delivered WT ND4-injected eyes (that were also injected with R340H mutant ND4 AAV) were still highly significant (P = 0.0073). Differences between eyes injected with COX8-delivered WT ND4 AAV (that were also injected with R340H mutant ND4 AAV) relative to eyes injected with standard VP2 AAV containing the WT ND4 (that were also injected with R340H mutant ND4 AAV) were not significant. (I) Bar plot of postmortem optic nerve diameter differences shows that with both eyes injected with the mutant ND4, differences in optic nerve diameters between those treated with COX8-delivered WT ND4 relative to mock treatment with AAV-GFP were highly significant (**P = 0.0053). As an additional control, we injected AAV-GFP into the eyes of normal mice that received no other intraocular injections and were killed 1 y later. Comparisons of this normal group (injected only with AAV-GFP) with those that received the mutant R340H ND4 and mock treatment with AAV-GFP were also significant (*P = 0.047). Comparisons of the normal group with the COX8-treated eyes that received the mutant ND4 showed the least differences. (J and K) Gross specimens of two animals dissected 13 mo after intravitreal injections revealed significant thinning of the entire mock-treated left optic nerve from the globe to the optic chiasm in eyes injected with AAV-GFP and AAV containing mutant R340H ND4, whereas the opposite right eyes treated by injection of COX8-delivered WT ND4 (and also injected with R340H mutant ND4 AAV) were much thicker. Histology of cross-sections of the retrobulbar segment of optic nerve taken at the sites indicated by the arrows confirmed the atrophy of the untreated left optic nerves (L and N), in contrast to the preservation of optic nerve structure with treatment by COX8-delivered WT ND4 (M and O). (Scale bar, 100 μm.) C, optic chiasm; L, left eye; R, right eye.
Fig. 6.
Fig. 6.
One year after intravitreal injections, a transmission electron micrograph shows that relative to the complement of axons seen with COX8 MTS scAAV-ND4 treatment (A), axonal density is markedly reduced with mock treatment with AAV-GFP (B) or with treatment with VP2 scAAV-ND4 lacking the MTS (C). The empty spaces between remaining axons where axons were lost accounted for the cystic spaces seen by light microscopy. (D) Micrograph of a normal optic nerve injected with only AAV-GFP a year earlier is shown for comparison. (E) Bar plot shows that differences in optic nerve axon counts between COX8 MTS scAAV-ND4–treated eyes relative to treatment with VP2 scAAV-ND4 or mock treatment with AAV-GFP were highly significant (*P = 0.0178; **P = 0.0032). a, axon; e, empty spaces where axons were lost.
Fig. P1.
Fig. P1.
Potential mechanisms of mitochondrial-targeting sequence (MTS)-targeted AAV entry into mitochondria, where the MTS AAV uncoats. The human ND4 subunit gene may combine by homologous recombination into the mouse mitochondrial genome, where it replaces the mouse ND4, or the heavy strand promoter-human ND4 may stay episomal and permit translation of the ND4 FLAG chimera (mechanism 1). Alternatively, the MTS-targeted AAV may bind to the outer mitochondrial membrane and the human ND4 may enter into the mitochondria with dissolution of the viral capsid (mechanism 2).

Comment in

Similar articles

See all similar articles

Cited by 46 PubMed Central articles

See all "Cited by" articles

Publication types

LinkOut - more resources

Feedback