Breakpoints of AAV-ITR in the wild-type mouse livers transduced with a very high dose of rAAV8. (A) The distribution of 120 breakpoints within the 165-base AAV-DDITR identified in 60 ITR-ITR recombination junctions in the wild-type mouse livers transduced with 7.2 × 10¹² vg of AAV8-ISce I.AO3. Fifteen junctions that exhibited 165-bp DDITR are not included. The histogram collating the ITR breakpoints showed a clustering with a peak corresponding to nucleotide number 70. (B) Location of microhomology (MH) found at ITR-ITR junctions. Among 60 ITR-ITR junctions sequenced, 36 showed microhomology. The histogram gives the number of times each nucleotide was found within 36 microhomology regions at ITR-ITR junctions. Nucleotides 62 to 70 frequently are involved in ITR-ITR recombinations with microhomology. (C) Schematic representation of the hot spot of ITR breakage (between nucleotides 70 and 71, shown with an arrowhead) and the region frequently contributing to microhomology at ITR recombination (nucleotides 62 to 70).

rAAV dose-response study of the no-end double-stranded linear rAAV genomes in DNA-PKcs-deficient mouse livers transduced with AAV8-ISce I.AO3. Wild-type (B6) and DNA-PKcs-deficient [PK(−)] C57BL/6J mice were injected with AAV8-ISce I.AO3 via the tail vein at various doses as indicated above the lanes. rAAV genome forms in the livers were analyzed 6 weeks postinjection by Southern blot analysis with PstI (non-rAAV genome cutter) digestion and were probed with a β-lactamase probe. For the five right-most lanes, the same blot was intensified. Open arrowheads indicate no-end monomer (1) and dimer (2) rAAV genomes. Positions of concatemers (Conc) and supercoiled circular monomers (CM) are indicated with closed arrowheads. Multiple lanes under the same experimental conditions represent samples from different animals.

Possible mechanisms for the formation of no-end double-stranded linear rAAV genomes with covalently closed DDITR hairpin caps in the absence of DNA-PKcs or Artemis. Three possible mechanisms (models A, B, and C) are presented. Model A is the simplest model, in which the leading strand hits the 5′ end of the AAV-ITR hairpin DNA and is ligated to it. Models B and C are different in how double-stranded rAAV genomes are formed (second-strand synthesis versus annealing), but they share the same hairpin terminal structure. An rAAV terminus with two 145-base ITRs is reminiscent of a stalled replication fork; therefore, it might be a target for the endonuclease activity of the Mus81-Eme1 complex, which cleaves the replication fork as illustrated. Since no active DNA replication is ongoing at the cleaved fork, the DNA gap presumably is sealed, forming DDITR hairpin caps. Holliday junction resolvase-like activity cleaves a 22-base hairpin region of the ITR (see Fig. 6D), making four types of no-end rAAV genomes in the absence of DNA-PKcs or Artemis. Whether this cleavage occurs before, at, or after forming no-end rAAV genomes is not known.

Fate of the no-end double-stranded linear rAAV genomes in the DNA-PKcs-deficient mouse livers after two-thirds partial hepatectomy. DNA-PKcs-deficient mice injected with 1.8 × 10¹² vg of AAV8-ISce I.AO3 via the tail vein underwent two-thirds partial hepatectomy (PHx) 6 weeks postinjection. The liver DNA samples at PHx (PrePHx) and 8 weeks after PHx (PostPHx) were analyzed for rAAV genome copy numbers and molecular forms by Southern blot analysis in the same manner as that described for Fig. 1. Ten micrograms of total DNA was digested with an appropriate enzyme, separated on a 0.8% neutral agarose gel, and probed with a β-lactamase probe. In the right panel, a separate analysis was performed to demonstrate that there was no no-end double-stranded linear monomer rAAV genomes in PostPHx samples. The bands at the positions indicated with open arrowheads represent no-end double-stranded linear genomes. Concatemers (Conc) and supercoiled circular monomer genomes (CM) are shown with closed arrowheads. M, 0.5 and 0.05 μg of liver DNA from a DNA-PKcs-deficient mouse injected with 7.2 × 10¹² vg of AAV8-ISce I.AO3, serving as a no-end rAAV genome position marker. 0.0 to 300, rAAV vector genome copy number standards (double-stranded rAAV vector genome copy numbers per diploid genomic equivalents). Each lane represents individual mouse samples.

Analyses of in-vitro-synthesized dumbbell-shaped no-end double-stranded linear monomer rAAV genomes. (A) Schematic representation of the method for synthesizing dumbbell-shaped nicked and no-end double-stranded linear rAAV genomes in vitro. The starting material is a 4,915-bp double-stranded PvuII-PvuII DNA fragment comprising AAV-EF1α-hF.IX genomes with a 130-bp AAV2-ITR at each genome terminus (for the vector map, see panel E). (B) Southern blot analysis of the in-vitro-synthesized dumbbell-shaped no-end rAAV genome and its intermediates. The top three rows of pluses and minuses indicate how each substrate (i.e., the no-end genome and its intermediates) was made, and the bottom row of pluses and minuses shows Exo III or Plasmid Safe (PS) treatment of the synthesized substrates. Ten nanograms of DNA of each product was separated on a 0.8% agarose gel and subjected to Southern blot analysis with an hF.IX cDNA probe. Some samples contained 400 ng of pNKan, a supercoiled circular plasmid containing the kanamycin resistance gene, and were treated in the same manner in order to verify that there was no irrelevant loss of sample DNA during the experimental procedures. Pol, T4 DNA Pol; M1, a 4,915-bp marker (M); Lin, a control double-stranded linear DNA, which is the same as M1 on this blot. 1st Exo indicates the Exo III treatment of the PvuII-PvuII AAV-EF1α-hF.IX fragment (see panel A). (C and D) Further characterization of the in-vitro-synthesized no-end genome and intermediates by neutral and alkaline agarose gel Southern blot analyses. Experimental conditions are indicated above each panel. (D) Pol-treated substrates and Pol- and Lig-treated substrates were treated with or without Exo III or Plasmid Safe and subsequently were treated with DraIII or BamHI. M2, 9,607-bp dimer marker. (E) Map of the AAV-EF1α-hF.IX vector genome.

Southern blot analyses of rAAV genomes in wild-type and Artemis-deficient 129 mouse tissues transduced with AAV8-ISce I.AO3. (A) rAAV genome forms in the livers of wild-type (129) and Artemis-deficient [Art(−)] mice injected with 3.0 × 10¹¹ vg (lo), 1.8 × 10¹² vg (m), or 7.2 × 10¹² vg (hi) of AAV8-ISce I.AO3 via the tail vein. Total DNA was prepared from tissue samples 10 days postinjection. A Plasmid Safe digestion step was incorporated into the experimental procedure to remove single-stranded rAAV genomes and double-stranded genomes that may emerge in test tubes due to plus- and minus-strand annealing (see the text). Otherwise, the experimental procedure for Southern blot analysis was essentially the same as that for Fig. 1 and 4. For the left panel, sample DNA was digested with PstI first, and then it was treated with or without Plasmid Safe (PS). For the middle panel, sample DNA was further digested with PstI, BamHI, or AlwNI. For the right panel, sample DNA was first digested with EcoRI, and then it was treated with or without Exo III. For the middle and right panels, only the DNA samples prepared from the mice injected with 7.2 × 10¹² vg of rAAV were used. The absence of double-stranded linear monomer genomes and the presence of a considerable amount of circular monomers and concatemers in the PstI- and PS-treated Art(−) lo sample have been confirmed in a separate Southern blot analysis (data not shown). The bands with arrows represent those derived from no-end linear monomer genomes. (B) The demonstration of no-end linear rAAV genomes (indicated with arrows) in nonhepatic tissues of Artemis-deficient mice, but not of wild-type mice, by Southern blot analysis. The experimental procedure was the same as that for Fig. 4B. Ten micrograms of total tissue DNA per lane was used, except for the liver, for which 0.05 μg per lane was used. Conc and CM in each panel indicate the positions of concatemers and supercoiled circular monomers, respectively. Lin, a 2,943-bp double-stranded linear DNA fragment.

Southern blot analyses of rAAV genomes in nonhepatic tissues of wild-type and DNA-PKcs-deficient mice injected with AAV8-ISce I.AO3. (A) rAAV genome forms in the muscle (M), kidney (K), and heart (H1 and H2) in wild-type (B6) and DNA-PKcs-deficient [PK(−)] mice injected with 7.2 × 10¹² vg of AAV8-ISce I.AO3 via the tail vein. Total DNA was extracted from each tissue 6 weeks postinjection (for M, K, and H1) or 6 months postinjection (for H2). Sample H2 was prepared due to the insufficiency of sample H1. Ten micrograms of total DNA was digested with the restriction enzyme indicated above the lanes, was separated on a 0.8% neutral agarose gel, and was probed with a β-lactamase probe. The bands indicated with arrows represent those derived from no-end linear monomer rAAV genomes detected only in the absence of DNA-PKcs. Positions of concatemers (Conc) and supercoiled circular monomers (CM) are indicated with arrowheads. (B) Further characterization of the unusual monomer-length rAAV genomes found in the DNA-PKcs-deficient mouse tissues by Exo III digestion. Ten micrograms of tissue DNA first was digested with EcoRI followed by Exo III treatment, and then it was subjected to neutral agarose gel Southern blot analysis. As controls, an appropriate amount of liver DNA (Lv) from wild-type or DNA-PKcs-deficient mice injected with 7.2 × 10¹² vg of AAV8-ISce I.AO3 also was processed in the same manner for each blot (0.05, 0.05, and 0.025 μg of liver DNA for the heart, muscle, and kidney blots, respectively). Arrows indicate no-end double-stranded linear monomer rAAV genome. Positions of concatemers (Conc) and supercoiled circular monomers (CM) are indicated with arrowheads. M1, 3,053- and 2,177-bp markers; M2, 6,167-bp marker; Lin, a control double-stranded linear DNA. The results demonstrate that the unusual linear monomer rAAV genomes were Exo III resistant. (C) An rAAV dose-response study of the no-end double-stranded linear rAAV genomes in DNA-PKcs-deficient mouse heart, muscle, and kidney transduced with AAV8-ISce I.AO3. Ten micrograms of tissue DNA was digested with PstI, was separated on a 0.8% neutral agarose gel, and was subjected to Southern blot analysis. Doses of rAAV are indicated above the lanes. No-end double-stranded linear monomer genome bands are indicated with arrows. Total double-stranded rAAV genome copy numbers determined by a separate Southern blot analysis are indicated below the lanes (means ± |mean-each value|; n = 2). Each arrowhead indicates the position of large concatemer (1), double-stranded linear dimer (2), double-stranded linear monomer (3), supercoiled double-stranded circular monomer (4), and single-stranded rAAV genome (5). M, total liver DNA (0.05 and 0.005 μg for DNA in the left and right lanes, respectively) extracted from the DNA-PKcs-deficient mouse liver transduced with 7.2 × 10¹² vg of AAV8-ISce I.AO3. Multiple lanes under the same experimental condition represent samples from different animals.

Bisulfite PCR analyses of no-end rAAV genome terminal structures. (A) Location of bisulfate PCR primers. Hairpin ends of no-end rAAV genome (upper panel, 165-base DDITR) and single-stranded rAAV genome (lower panel, 145-base ITR) are shown. The letters A to D are ITR subregions. Numbers (1 to 165) indicate nucleotide positions in the ITRs. (B) Southern blot analysis of purified no-end double-stranded linear AAV8-ISce I.AO3 genomes recovered from DNA-PKcs-deficient [DNA-PKcs(−)] or Artemis-deficient [Artemis(−)] mouse livers. A portion of each purified DNA was separated along with 0.625 to 5 μg of unpurified liver DNA on a 0.8% neutral agarose gel and was subjected to Southern blot analysis with a β-lactamase probe. Purified no-end double-stranded linear monomer rAAV genomes (band position 4) are free from large concatemers (1), linear trimers (2), linear dimers (3), supercoiled circular monomers (5), and single-stranded genomes (6). (C) Bisulfite PCR products of AAV-ITR sequences. Chemically modified ITR sequences were PCR amplified from various templates with a primer set of DDITR F1 and DDITR R1 (lanes 1 to 10) or a set of ITR F2 and ITR R2 (lanes 11 to 13). The resulting PCR products were separated on a 2.5% NuSieve GTG agarose gel and stained with ethidium bromide. In lanes 1 to 4, a second band with various intensities (indicated with arrows) was observed in addition to the major band of the expected size. The bisulfite PCR templates are as follows: lanes 1 and 2, purified no-end linear AAV-ISce I.AO3 monomer genomes formed in DNA-PKcs-deficient (lane 1) and Artemis-deficient (lane 2) mice; lanes 3 and 4, purified no-end linear AAV-EF1α-nlslacZ monomer genomes formed in DNA-PKcs-deficient (lane 3) and Artemis-deficient (lane 4) mice; lanes 5 and 13, pDDITRAAV-EF1α-hFAH.AO plasmid carrying a 165-bp DDITR; lane 6, a control gel piece taken and processed as a no-end genome purification (see Materials and Methods); lane 7, no-DNA control; lanes 8 and 9, chemically modified total liver DNA from a C57BL/6 (lane 8) or 129 (lane 9) mouse transduced with 7.2 × 10¹² vg of AAV8-ISce I.AO3; lane 10, in-vitro-synthesized no-end AAV-EF1α-hF.IX genome; lane 11, single-stranded AAV8-ISce I.AO3 viral genome DNA; lane 12, single-stranded AAV8-EF1α-nlslacZ viral genome DNA. M, HaeIII-digested ΦX174 DNA. (D and E) Two possible mechanisms for the formation of the 143-base AAV-DDITR: a model involving a Holliday junction resolvase-like activity (D) and a premature ITR fold-back model (E). In the first model (D), one arm of the T-shaped hairpin structure of AAV-ITR is cleaved at symmetrically related positions, as shown with open or closed arrowheads, by Holliday junction resolvase-like activity. This results in the removal of a 22-base DNA hairpin containing the B or C subregion of the ITR, forming two types of simple hairpin structures (types 1 and 2). In the second model (E), the leading strand folds back when AAV-ITR is half copied, skipping one arm of the T-shaped hairpin structure. The ITR shown in this figure is a flip-oriented ITR.

Southern blot analyses of rAAV genomes in wild-type and DNA-PKcs-deficient mouse livers transduced with a high dose of AAV8-ISce I.AO3. (A) AAV-ISce I.AO3 genome map. The locations of restriction enzyme recognition sites and a β-lactamase probe used for Southern blot analyses and sizes of each digested DNA fragment (in bases or base pairs) are indicated. Pr, Tn3 prokaryotic promoter; pUC ori, pUC plasmid origin of replication; MLV-LTR, a portion of the Moloney murine leukemia virus long terminal repeat. (B) Analysis of rAAV genome forms by neutral agarose gel Southern blot analysis with a β-lactamase probe. Wild-type C57BL/6J mice (B6) or DNA-PKcs-deficient mice [PK(−)] were injected with 7.2 × 10¹² vg of AAV8-ISce I.AO3 via the tail vein (n = 2 mice each), and the liver tissues were harvested 6 weeks postinjection. rAAV genome copy numbers (BamHI/AlwNI digestion) and rAAV genome forms (PstI digestion) are shown with copy number standards (0 to 3,000 ds-vg/dge). Unusual bands that were not present in samples from wild-type mice were observed in PstI (noncutter) DNA digests in DNA-PKcs-deficient mouse livers (shown with arrows). The major unusual bands migrated slightly faster than a 3,049-bp marker, which is 20 bp shorter than full-length double-stranded linear AAV-ISce I.AO3 genomes (3,069 bp). Positions of large concatemeric rAAV genomes (Conc) and supercoiled circular monomer rAAV genomes (CM) are indicated with arrowheads. (C) Schematic explanation of how to discriminate between four different possible configurations of double-stranded linear monomer rAAV genomes by neutral and alkaline gel Southern blot analyses. Closed and open arrowheads indicate BamHI and AlwNI sites, respectively. Under neutral conditions, each linear monomer genome configuration digested with PstI, BamHI, or AlwNI would migrate as double-stranded DNA of the size indicated above each arrow, while under alkaline conditions it would migrate as single-stranded DNA of the size indicated below each arrow. The sizes of DNA fragments are calculated with the assumption that each rAAV genome terminus has a 145-base full-length ITR and each ITR hairpin cap at closed ends comprises two full-length ITRs (for more details, see Table 1). Only the DNA fragments detectable by the β-lactamase probe used in the study are shown. (D) Further characterization of the unusual monomer-length rAAV genomes that accumulated in DNA-PKcs-deficient mouse livers by neutral and alkaline gel Southern blot analyses. Numbers 1 to 16 are given to the left or upper left of each representative band on the blots, and the bands representing or derived from the unusual double-stranded linear monomer genomes, which were seen only in DNA-PKcs-deficient mice, are circled. Ten micrograms of total liver DNA (n = 2) was digested with the enzymes indicated above the lanes, separated on a 0.8% neutral agarose gel (upper panel) or a 1.0% alkaline agarose gel (lower panel), and subjected to Southern blot analysis. Samples were loaded in the same order for both blots. The result from PstI (non-rAAV genome cutter) digestion in the upper panel demonstrates that double-stranded linear monomer genomes (number 1 in the upper panel) are found only in DNA-PKcs-deficient mice and not in wild-type mice. The results from AlwNI digestion and BamHI digestion further confirmed that the bands (number 1 in the upper panel) are linear monomers, not circular genomes. The result from PstI digestion shown in the lower panel demonstrates that double-stranded linear monomer genomes in the DNA-PKcs-deficient mouse livers exhibited a dimer size (number 1 in the lower panel). AlwNI digestion and BamHI digestion excluded the possibility that the dimer-sized bands (number 1 in the lower panel) have a hairpin structure with an open end, because a majority of the bands migrated to head-to-head (number 8 in the lower panel) and tail-to-tail (number 11 in the lower panel) positions with AlwNI digestion and BamHI digestion, respectively. Based on these observations, we concluded that the unusual linear monomer genomes exhibit a no-end configuration (see panel C). For the expected size of each band, see Table 1. Very faint bands shown with arrows are the bands derived from linear genomes with open ends. Each band represents the following: 1, the unusual double-stranded (ds) linear monomer; 2, ds linear dimer; 3, ds linear trimer; 4 and 7, large concatemer; 5, supercoiled ds circular monomer; 6, relaxed ds circular monomer; 8, head-end; 9 and 12, head-tail; 10, head-head; 11, tail-end; 13, tail-tail; 14, 6,167-bp or 6,167-base dimer size marker; 15, 3,053-bp or 3,053-base monomer size marker; 16, 2,177-bp or 2,177-base size marker. (E) Southern blot analysis of Exo III-treated or Plasmid Safe-treated rAAV genomes from the liver. Total liver DNA (10 μg) was first digested with EcoRI or PstI and then with Exo III (Exo) or Plasmid Safe (PS), respectively. The digested DNA was analyzed by 0.8% neutral agarose gel Southern blot analysis along with naïve liver genomic DNA containing control linear (Lin) or supercoiled circular (Circ) double-stranded DNA. Lin, 2,943-bp linear DNA; Circ, a supercoiled circular plasmid containing the 2,943-bp fragment. After treatment with Exo III or Plasmid Safe, the supercoiled plasmid controls were treated with EcoRI to release the 2,943-bp fragment. The open arrowhead indicates the position of the unusual double-stranded linear monomer and dimer rAAV genomes. Positions of concatemers (Conc) and supercoiled circular monomers (CM) are indicated with closed arrowheads. For panels B and D, multiple lanes under the same experimental condition represent samples from different animals.