Molecular analysis of AAV5-hFVIII-SQ vector-genome-processing kinetics in transduced mouse and nonhuman primate livers

Abstract

Valoctocogene roxaparvovec (AAV5-hFVIII-SQ) is an adeno-associated virus serotype 5 (AAV5)-based gene therapy vector containing a B-domain-deleted human coagulation factor VIII (hFVIII) gene controlled by a liver-selective promoter. AAV5-hFVIII-SQ is currently under clinical investigation as a treatment for severe hemophilia A. The full-length AAV5-hFVIII-SQ is >4.9 kb, which is over the optimal packaging limit of AAV5. Following administration, the vector must undergo a number of genome-processing, assembly, and repair steps to form full-length circularized episomes that mediate long-term FVIII expression in target tissues. To understand the processing kinetics of the oversized AAV5-hFVIII-SQ vector genome into circular episomes, we characterized the various molecular forms of the AAV5-hFVIII-SQ genome at multiple time points up to 6 months postdose in the liver of murine and non-human primate models. Full-length circular episomes were detected in liver tissue beginning 1 week postdosing. Over 6 months, quantities of circular episomes (in a predominantly head-to-tail configuration) increased, while DNA species lacking inverted terminal repeats were preferentially degraded. Levels of duplex, circular, full-length genomes significantly correlated with levels of hFVIII-SQ RNA transcripts in mice and non-human primates dosed with AAV5-hFVIII-SQ. Altogether, we show that formation of full-length circular episomes in the liver following AAV5-hFVIII-SQ transduction was associated with long-term FVIII expression.

Graphical abstract

AAV vectors are a well-established transgene delivery method, but the steps that transform oversized vector genomes into stable structures capable of transgene expression remain uncertain. Here, we demonstrate the formation of durable, circular, full-length episomes in hepatocytes of mice and non-human primates dosed with valoctocogene roxaparvovec (AAV5-hFVIII-SQ).

Introduction

Hemophilia A is an X-linked bleeding disorder caused by a deficiency in factor VIII (FVIII) coagulation protein activity.^1,2 People with hemophilia A are susceptible to spontaneous and traumatic bleeding into soft tissues and joints that can result in painful, disabling arthropathy and impaired quality of life, including possible intracranial hemorrhage and early death.^2,3 Hemophilia A is currently managed with chronic administration of exogenous FVIII, either prophylactically or in response to bleeding events.¹

Gene therapy with valoctocogene roxaparvovec (AAV5-hFVIII-SQ) is being developed for long-term management of severe hemophilia A.4, 5, 6 AAV5-hFVIII-SQ is a recombinant, replication-incompetent adeno-associated virus serotype 5 (AAV5) gene therapy vector containing a genome comprising a single-stranded, codon-optimized, B-domain-deleted human FVIII gene (hFVIII-SQ) controlled by a hepatocyte-selective promotor, a synthetic polyadenylation (poly(A)) sequence, and AAV2-derived double-stranded inverted terminal repeats (ITRs) on the 5′ and 3′ ends.4, 5, 6 The total vector genome is more than 4.9 kb in length,⁵ exceeding the optimal packaging capacity of AAV vectors.7, 8, 9 As a result, incomplete genomes may be present in the vector preparation. In a murine model of hemophilia A treated with AAV5-hFVIII-SQ at doses from 2 × 10¹³ to 6 × 10¹³ vg/kg, normal to supraphysiological levels of circulating hFVIII-SQ protein were produced, and improvements in bleeding time and reductions in blood loss were achieved.⁴ In addition, comparable doses in primates produced similar levels of circulating hFVIII-SQ protein.⁴ In a phase 1/2 clinical trial, a single infusion of AAV5-hFVIII-SQ dosed at 6 × 10¹³ and 4 × 10¹³ vg/kg resulted in a clinically relevant reduction in bleeding through 5 and 4 years of follow-up postinfusion, respectively, for adult men with severe hemophilia A.^5,6,10 The most common adverse events were transient, asymptomatic alanine aminotransferase increases that resolved without clinical sequelae.^5,6,10 In a phase 3 trial, an infusion of 6 × 10¹³ vg/kg AAV5-hFVIII-SQ provided significantly increased FVIII activity and reduced bleeding at 52 weeks for adult men with severe hemophilia A.¹¹ While these findings represent a substantial leap forward in the clinic, there are gaps in our biological understanding of the mechanisms behind long-term expressions of FVIII.

Multiple complex processes are involved in achieving successful and durable transgene expression using an AAV vector (Figure 1). Following dosing and target cell receptor-mediated uptake of the viral capsid, the single-stranded, linear vector DNA is transported into the nucleus.12, 13, 14 Vector-genome-processing can then occur, transforming the single-stranded vector DNA to double-stranded linear genomes via second-strand DNA synthesis.^9,13, 14, 15, 16, 17, 18, 19, 20 For vector genomes exceeding the packaging limit of the AAV capsid, the majority of vector genomes packaged are fragmented or clipped.21, 22, 23 To generate full-length vector genomes in cells, overlapping vector genomes of opposite strand polarities are first annealed, followed by second-strand synthesis and/or homologous recombination.²⁴ This process is facilitated by Rad51C, a single-stranded DNA-binding protein that promotes strand annealing in response to double-strand breaks.⁹ The resultant double-stranded linear genomes are converted into stable monomeric and concatemeric circular episomes when the ITRs on the 5′ and 3′ ends of the linear vector genome structure undergo homologous recombination or non-homologous end-joining.25, 26, 27, 28, 29 Presumably, only full-length vector genomes spanning from the promoter to the poly(A) sequence are capable of giving rise to stable transgene expression. Preclinical studies in lung, liver, and muscle transduced by AAV gene therapy in mice and nonhuman primates, and liver in canines, demonstrate that circularized monomeric and concatemeric episomes are the major DNA species associated with long-term, persistent expressions of transgene products in the target cells.^25,28,30,31

Figure 1.

Model of AAV genome-processing and forms associated with long-term expression

DNA-PKcs, DNA-dependent protein kinase catalytic subunit; DS, double-stranded; DSB, double-strand break; H, head (5′ end); ITR, inverted terminal repeat; SS, single strand; T, tail (3′ end); vg, vector genome.

Preclinical results, as well as the sustained circulating FVIII activity observed in participants who received gene transfer in the phase 1/2 clinical trial, suggest that the oversized AAV5-hFVIII-SQ vector genome is successfully assembled and repaired into full-length, circular episomes in hepatocytes following AAV treatment. In this study, we explore the kinetics and structures of AAV5-hFVIII-SQ vector-genome-processing at regular intervals up to 6 months following dosing in mice and non-human primates. These findings provide insights into the complex molecular mechanisms behind successful in vivo transductions of liver using an AAV gene therapy platform.

Results

AAV5-hFVIII-SQ vector-genome-processing in mouse livers

To assess vector-genome-processing kinetics soon after vector administration, 8-week-old Rag2^−/− FVIII^−/− double knockout mice were used as an animal model of hemophilia A. Mice were dosed with 3.5 × 10¹³ vg/kg AAV5-hFVIII-SQ (the size distribution of AAV5-hFVIII-SQ is shown in Figure S1), and livers were analyzed 24 h and 1, 3, 5, and 8 weeks thereafter. Quantitative real-time polymerase chain reactions (quantitative real-time PCR) using probes and primers specific to the hFVIII-SQ transgene were used to quantify hFVIII-SQ DNA abundance (SQ amplicon; Figure S2). Mice dosed with AAV5-hFVIII-SQ showed an initial sharp decline in overall hFVIII-SQ DNA in the first 3 weeks postadministration; by week 5 postdose, 97.3% of hFVIII-SQ vector genomes had been degraded compared with by 24 h postdose (Figure 2A). Comparison of levels of cytoplasmic and nuclear vector genomes using DNA in situ hybridization showed that an elimination of genomes within the cytoplasm was responsible for the majority of the decline in vector DNA observed in the first 3 weeks; vector genome signals in the nuclei also declined by 82% between 1 and 3 weeks (Figure S3). Consistent with droplet digital PCR (ddPCR) results, by 8 weeks, the vector DNA signals decreased 99% and 91% in the cytoplasm and in nuclei, respectively, compared with after 24 h. In the liver, levels of hFVIII-SQ RNA assessed by quantitative real-time PCR increased through 8 weeks postdose (Figure 2B) in contrast to the observed pattern of decreased DNA. Consistent with increasing liver hFVIII-SQ RNA levels, circulating plasma hFVIII-SQ protein, quantified via enzyme-linked immune-absorbent assay (ELISA), also increased over time (Figure 2C).

Figure 2.

Assessment of AAV5-hFVIII-SQ vector genome levels and transgene expression in livers from mice dosed with 3.5 × 10¹³ vg/kg AAV5-hFVIII-SQ using quantitative real-time PCR

(A) Overall hFVIII-SQ vector DNA levels as measured by SQ amplicon in mouse liver over 8 weeks, (B) Liver hFVIII-SQ RNA levels over 8 weeks. (C) Plasma hFVIII protein levels over 8 weeks. Data are mean ±1 standard deviation. At all timepoints, n = 10. (A) and (B) use a log scale y axis. hFVIII-SQ, human Factor VIII; hr, hour; wk, week.

Because 97% of the overall genomes were degraded, we hypothesized that incomplete genomes lacking ITRs were being preferentially degraded over the first 8 weeks. To assess this, we used a variant of the ddPCR method, referred to as drop-phase ddPCR, to determine the contiguity of vector genomes between the 5′ end, the 3′ end, and the middle (Figure S4). This method can detect and quantify the co-occurrence of separate target sequences individually labeled with different fluorescent probes that are present in a single droplet and therefore linked on a contiguous DNA strand. Various primers and probes were designed to generate short amplicons spanning approximately 1,000 base pairs internal to the 5′ ITR (eg, R1–R6 amplicons) or 3′ ITR (eg, R7–R11 amplicons) of AAV5-hFVIII-SQ. To confirm the validity of this method, we determined that full-length linear genomes and circular monomers and concatemers of known DNA forms could be detected (Figure S5). Then, each of these primers/probes was combined with a primer pair/probe that amplified the middle of the genome (SQ amplicon) in each duplex reaction (Figures 3A and S2). The relative proportions of vector genomes containing sequences spanning approximately 1 kb internal to either 5′ or 3′ ITR to vector genomes containing the SQ sequences increased significantly from 24 h to 3 weeks and increased only slightly from 3 weeks to 8 weeks post AAV5-hFVIII-SQ administration (Figures 3B–3F). Both vector genomes lacking either ITR but containing the SQ region (single positive in the drop-phase ddPCR reaction) and vector genomes containing the SQ region connected to an ITR (double positives) were rapidly lost during the first 3 weeks postdose, with the rate of loss decreasing substantially after week 3 (Figures 3G and 3H). Incomplete vector genomes (as defined by single positives), particularly those lacking either the 5′ or 3′ ITR, were preferentially degraded at a faster rate over 8 weeks than were vector genomes that contained either ITR (double positives).

Figure 3.

Relative proportions of AAV5-hFVIII-SQ vector genomes as detected with drop-phase ddPCR in mouse liver following dosing with 3.5 × 10¹³ vg/kg AAV5-hFVIII-SQ

(A) Location of ddPCR amplicons used. (B–H) Proportions of vector genomes containing each amplicon relative to SQ amplicon at (B) 24 h, (C) 1 week, (D) 3 weeks, (E) 5 weeks, or (F) 8 weeks postdosing, and proportion of genomes over time (G) with both SQ and 5′ ITR amplicons or without the 5′ ITR and (H) with both SQ and 3′ ITR amplicons or without the 3′ ITR. (I) Portion of individual amplicon sites A, B, C, and D that are on DNA fragments unconnected to other sites (unconnected/total normalized for each site independently) over time. Data are mean ±1 standard deviation. At all timepoints, n = 10. bp, base pairs; hFVIII-SQ, human factor VIII; ITR, inverted terminal repeat; hr, hour; pA, polyadenylation sequence; wk, week.

To further investigate the mechanism and kinetics of degradation during this period, we examined vector structural connectivity with a set of 4 drop-phase amplicons spaced along the vector (amplicon locations are shown in Figure S2). We used a set of paired amplicon reactions to identify the fraction of each amplicon that was unconnected to any adjacent amplicons, representing potential intermediates on the degradation pathway, where sites are contained on small fragments disconnected from the rest of the vector genome. Partial vector fragments, containing one end or the other, are largely either degraded or assembled into more stable full-length forms after being released from the capsid. All 4 sites were predominantly connected in the first sample, from 24 h after dosing, when much of the DNA is likely still in the original vector form. If released DNA were rapidly degraded or assembled, we would only observe connectivity similar to the 24 h sample or increased connectivity. If released DNA were largely degraded by exonuclease activity from the exposed single- or double-stranded ends of the partial vector, we would observe a decrease in the relative amount of connected middle amplicons and an increase in disconnected ITR-proximal sites, while degradation from the ITR ends would create an increase in fragments that contain only the middle sites. Instead, all 4 amplicons had similar kinetics (Figure 3I), with each site reaching maximal fragmentation with about 60% unconnected to any other site, at 1-week postdose, when the majority of vector genomes were disappearing. As with the ITR proximal sites, the vector DNA remaining at 8 weeks was more highly connected. This suggests that vector genome degradation includes an active endonuclease fragmentation mechanism where released DNA is broken into smaller fragments that have sufficient persistence to be the majority form at the time point of maximal DNA degradation.

Circular full-length vector-genome-processing kinetics in mouse liver

We next examined the kinetics of the formation of circular full-length or ITR-fused AAV5-hFVIII-SQ vector genomes over 24 weeks in mice dosed with 2 × 10¹³, 3.5 × 10¹³, and 6 × 10¹³ vg/kg in 2 separate studies. DNA was extracted from mouse livers, and samples were digested with Plasmid Safe ATP-Dependent DNase (PS-DNase; Lucigen, Middleton, WI, USA) to hydrolyze linear DNA and isolate circular DNA. Full-length genomes were quantified using drop-phase ddPCR with primers/probes on the proximal end of the 5′ and 3′ ITRs (the D-segment) to generate amplicons R1 and R11 (Figure S2), thus spanning the entire vector genome unit (5′ D-segment, promoter, transgene, and poly(A) signal, 3′ D-segment). ITR-fused genome forms were identified using primers, and probe against a DNA sequence formed only when the 5′ and 3′ ITRs are joined by homologous recombination (Figure S6).

DNA samples were also digested with the restriction enzyme KpnI to enumerate each full-length head-to-tail vector genome units within concatemeric forms following PS-DNase treatment (Figures 4A and S6). No circular full-length vector genome DNA copies or ITR fusions were detected at 24 h postdosing; however, they were present at 1-week postdose and increased slowly through 8 weeks (Figure 4B), as did ITR fusion genomes (Figure 4C). Levels of both circular full-length episomes (Figure 4D) and ITR fusions (Figure 4E) remained stable through 24 weeks, and higher levels of circular full-length genomes resulted from the higher dose. Despite the reduction in overall vector genome DNA, repaired circular full-length and ITR-fused vector genome DNA species were formed and increased over time. Levels of circular full-length genomes were similar to levels of ITR-fused genomes, suggesting that most ITR-fused genomes are full-length genomes.

Figure 4.

Circular episome formation in mice treated with AAV5-hFVIII-SQ

(A) Schematic diagram showing enzymatic treatment of samples to reveal quantities of head-to-tail full-length genome units in circular episomes using drop-phase ddPCR. (B and C) Over 8 weeks postdosing with 3.5 × 10¹³ vg/kg AAV5-hFVIII-SQ, levels of (B) full-length circular hFVIII-SQ vector DNA and (C) circular ITR-fused vector genomes in the liver. (D and E) Over 24 weeks, (D) full-length circular hFVIII-SQ vector DNA levels in livers from mice dosed with either 6 × 10¹³ or 2 × 10¹³ vg/kg and (E) circular ITR-fused vector genomes in livers of mice dosed with 6 × 10¹³ vg/kg. (F) Schematic representation of circular episome structures using specific probes as detected by Southern blotting. (G) Southern blot showing circular episome vector concatemeric structure (left panel) and orientation (right panel). Full-length vector genomes were identified with linked R1-R11 amplicons and are defined as sequences including the D-segment of the 5′ ITR, the promoter, the h-FVIII-SQ transgene, the polyadenylation signal, and the D-segment of the 3′ ITR. The same primers and probes were used in both southern blots shown in (G). In (G), J1–7 represent individual mice. hFVIII-SQ, human factor VIII; H, head (5′ end); HH, head-to-head orientation; HT, head-to-tail orientation; TT, tail-to-tail orientation; M, monkey; PS, Plasmid Safe ATP-Dependent DNase; T, tail (3′ end); vgDNA, vector genome DNA; wk, week; Veh, vehicle.

Southern blotting was performed to visualize the structure and orientation of genome forms using probes near the 5′ (head) and 3′ (tail) ends with monomeric/multimeric AAV5-hFVIII-SQ DNA used as a control (Figure S7). Consistent with ddPCR analysis, Southern blotting demonstrated that the circular episomes were present persistently at 4 through 24 weeks postdosing; bands migrated at sizes corresponding to full-length monomeric and concatemeric forms, mainly in the head-to-tail configuration (Figures 4G and S8).

Levels of hFVIII-SQ RNA transcripts were also assessed in mice. The quantity of RNA transcripts was significantly correlated with levels of overall hFVIII-SQ vector genomes as measured by the SQ amplicon (r = 0.46; p < 0.001; Figure 5A) but were more strongly correlated with full-length hFVIII-SQ vector genomes (r = 0.64; p < 0.0001; Figure 5B) and ITR-fused genomes (r = 0.60; p < 0.0001; Figure 5C).

Figure 5.

Correlation of transgene RNA and DNA after AAV5-hFVIII-SQ treatment

(A–C) In the mouse liver, correlation of (A) hFVIII-SQ RNA with overall hFVIII-SQ vector genomes (including incomplete genomes) as measured by SQ amplicon, (B) hFVIII-SQ RNA with full-length hFVIII-SQ vector genomes, and (C) hFVIII-SQ RNA with ITR-fused genomes. Full-length vector genomes were identified with linked R1-R11 amplicons and are defined as sequences including the D-segment of the 5′ ITR, the promoter, the hFVIII-SQ transgene, the polyadenylation signal, and the D-segment of the 3′ ITR. hFVIII-SQ, human factor VIII; ITR, inverted terminal repeat; vgDNA, vector genome DNA.

Detection of circular episomes in non-human primate liver

In a separate study, male cynomolgus monkeys were intravenously infused with 6 × 10¹³ vg/kg AAV5-hFVIII-SQ, and livers were analyzed 13 or 26 weeks after vector administration. Southern blotting detected circular episomes in multimeric forms in the livers of these monkeys at 13 and 26 weeks, while no signal was detected in animals dosed with vehicle (Figures 6A and S9). As in mice, full-length circular episomes preferentially formed in the head-to-tail configuration. Levels of full-length hFVIII-SQ vector genomes as detected by drop-phase ddPCR were significantly correlated with FVIII RNA transcript levels in non-human primate livers (r = 0.74, p < 0.05; Figure 6B), similar to results in mice.

Figure 6.

Circular episome formation in non-human primates treated with AAV5-hFVIII-SQ

Assessment of liver circular episomes in non-human primates dosed with AAV5-hFVIII-SQ. (A) Southern blots showing circular episome vector concatemeric structure and orientation. (B) Correlation of liver hFVIII-SQ RNA with circular full-length vector genome DNA as measured by drop-phase ddPCR. Full-length vector genomes were identified with linked R1-R11 amplicons and are defined as sequences including the D-segment of the 5′ ITR, the promoter, the hFVIII-SQ transgene, the polyadenylation signal, and the D-segment of the 3′ ITR. The same primers and probes were used in both Southern blots shown in (A). hFVIII-SQ, human factor VIII; H, head (5′ end); HH, head-to-head orientation; HT, head-to-tail orientation; M, monkey; PS, Plasmid Safe ATP-Dependent DNase; T, tail (3′ end); TT, tail-to-tail orientation; Veh, vehicle; vgDNA, vector genome DNA; wk, week.

Discussion

In recent years, AAV-based gene transfer has rapidly advanced in the clinic, with multiple FDA-approved products on the market and even more in phase 3 trials.^12,32 At the molecular level, however, much remains unknown. In this report, we studied the kinetics and structures of AAV5-hFVIII-SQ vector genomes in the livers of mice and non-human primates for up to 6 months following dosing.

Consistent with previous reports, treatment of mice with AAV5-hFVIII-SQ was followed initially by a sharp decline in overall vector genome DNA: more than 97% of the hFVIII-SQ genome was degraded by 8 weeks postdose.^4,33 The elimination of genomes in the cytoplasm and within the nuclei contributes to this decline. Incomplete genomes, particularly those lacking either 5′ or 3′ ITRs, were preferentially degraded at a faster rate. However, full-length circular episomes were observed in the liver 1-week postdose. While the total number of vector genomes present declined, the proportion of full-length duplex ITR fusions and circular episomes increased over time. Circular episomes were also present in non-human primates and persisted in the liver for up to 6 months, the longest time point assessed. These results are consistent with the proposed genome-processing model following AAV-mediated gene transduction, where most single-stranded vector genomes are destroyed, and a small portion are assembled into stable circular episomes that resist degradation and mediate stable protein expression (Figure 1).^27,28,30

Levels of FVIII RNA present in the mouse and non-human primate liver samples correlated with the levels of the full-length hFVIII-SQ genome capable of giving rise to transcription. Protein presence was first detected 3 weeks after dosing; the latency is expected due to the slow uncoating of the AAV5 capsid and the time required to process genomes into forms capable of giving rise to stable transcription and then the translation and secretion of sufficient protein quantities above the limit of detection. Head-to-tail circular genomes that form after AAV administration appear to be more stable than non-ITR, linear, and double-stranded genomes.^13,14,25,27 Similar to previous studies using other AAV gene therapies,^{13,15,25,27,34,35} head-to-tail episomes are the primary forms of circular genomes in mice and non-human primates treated with AAV5-hFVIII-SQ; this configuration also appears to support stability of episomes and more efficient transcription of the transgene.^25,36, 37, 38 Together, these data suggest that following AAV5-hFVIII-SQ transduction, the formation of circular episomes in the liver, particularly in the head-to-tail orientation, leads to efficient transcription and long-term FVIII expression in the liver.

In a canine model of severe hemophilia A, treatment with AAV-canine-FVIII-SQ resulted in persistent FVIII activity and correction of bleeding phenotypes for up to 10 years postdose.³¹ Analysis of terminal liver samples revealed more than 95% of remaining vector genomes were present in episomal forms, suggesting that episomes, not vector genomes integrated into the nuclear genome, may be the major DNA form responsible for long-term transgene expression;³¹ however, expression arising from integrated genomes cannot be ruled out. Nevertheless, levels of full-length episomes present in canine livers were positively associated with levels of plasma FVIII activity.³¹ Together with our results, these data demonstrate that circularized episomes can resist degradation and support long-term expression of transgenes, including oversized genomes such as AAV5-hFVIII-SQ.^{13,25,27,35,39}

In these preclinical studies, limitations include the relatively small sample sizes and the short maximum follow-up time of 6 months. Evidence of longer-term durability is pending additional follow-up. PS-DNase also degrades 25%–35% of circular plasmid DNA (data not shown), as the DNA extraction process likely linearizes and shears large concatemeric circular episomes, rendering them sensitive to PS-DNase.³⁴ Therefore, the levels of circular vector DNA quantified following PS-DNase treatment represent the lowest possible value, but the proportion of linear versus circular genomes cannot be calculated accurately as it is unknown how many circular episomes are sheared during the extraction process. Additionally, expression arising from the integration of a small portion of full-length genomes into the host nuclear genome cannot be ruled out and may contribute to the results presented here.

Following treatment with AAV5-hFVIII-SQ, we observed dose-dependent increases in the quantities of circular full-length and ITR-fused vector genomes in the livers of mice and non-human primates that were associated with long-term expression of hFVIII-SQ. These results, obtained using a novel molecular technique, support the hypothesis that circular episomes formed in the weeks following AAV-vector-mediated gene transfer lead to long-term expression of gene product. The results reported here with the oversized AAV5-hFVIII-SQ vector align with previous findings with standard-sized vectors.13, 14, 15^,25, 26, 27, 28^,30,35,40, 41, 42 Future studies will assess the transduction pattern and kinetics in human participants with severe hemophilia A following a single infusion of AAV5-hFVIII-SQ through serial liver biopsies if such samples can be obtained, allowing for the association of FVIII activity with vector genome forms over time.

Materials and methods

Mouse study design

Rag2^−/−FVIII^−/− double knockout (DKO) mice were used in all analyses. Only male mice were used, as hemophilia A occurs predominantly in men. FVIII^−/− mice (B6; 129S-F8^tm1Kaz; Jackson Laboratory, Bar Harbor, ME, USA), which have less than 1% of normal mouse FVIII activity and model the severe hemophilia A phenotype,⁴³ and Rag2^−/− mice (B6.129S6-Rag2^tm1Fwa; Taconic Biosciences, Rensselaer, NY, USA), which are incapable of generating mature T or B lymphocytes and thus are unlikely to produce antibodies against foreign proteins,⁴⁴ were crossed to generate a DKO line as previously described.⁴ Mice were administered the AAV5-hFVIII-SQ vector in prepared vehicle (0.001% Pluronic F-68 in Dulbecco’s phosphate-buffered saline) or vehicle alone via a single intravenous bolus tail injection of 4 μL/g body weight. Following deep anesthetization with inhaled isoflurane and exsanguination via cardiac puncture, liver samples were harvested. Blood was collected with sodium citrate anticoagulant (0.38% final concentration). All in vivo mouse experiments were performed in accordance with institutional guidelines under protocols approved by the Institutional Animal Care and Use Committee of the Buck Institute (Novato, CA, USA).

To evaluate vector-genome-processing kinetics soon after dosing, 8-week-old mice (n = 10 per group) were treated with vehicle or 3.5 × 10¹³ vg/kg of AAV5-hFVIII-SQ and euthanized at 24 h or 1, 3, 5, or 8 weeks after vector administration. To analyze vector processing over a longer period of time, 8-week-old mice (n = 7 per group) were treated with vehicle or 2 × 10¹³ or 6 × 10¹³ vg/kg of AAV5-hFVIII-SQ and euthanized at 4, 12, or 24 weeks after vector administration.

Non-human primate study design

Male cynomolgus monkeys were received from Charles River Labs (Wilmington, MA, USA) and housed at their facility in Reno, NV, USA. All study procedures were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of the Charles River Labs facility. The animals were approximately 3 years old and weighed approximately 2.5 kg on average at the time of dosing. All monkeys were prescreened for total antibodies and neutralizing antibodies against AAV5; only those that screened negative for both were used in the study.⁴⁵ Eight male cynomolgus monkeys were intravenously dosed with 6 × 10¹³ vg/kg AAV5-hFVIII-SQ and euthanized 13 or 26 weeks after vector administration. At necropsy, liver samples were taken and snap-frozen for molecular analyses.

Molecular methods

DNA enzyme treatment procedures

DNA was extracted from frozen intact liver tissues using a MagMAX DNA Multi-Sample Ultra kit with the KingFisher Flex System (ThermoFisher Scientific, Waltham, MA, USA) following manufacturer’s instructions. Extracted DNA was diluted to 20 ng/μL. Samples were digested with PS-DNase (Lucigen, Middleton, WI, USA) to hydrolyze all linear forms of DNA and isolate circular DNA. The number of circular genomes is underestimated using PS-DNase because the DNA extraction process would likely shear/linearize some large concatemeric circular episomes, rendering them sensitive to PS-DNase (it is unclear the extent to which circular episomes are sheared during the extraction process), and PS-DNase has been shown to degrade 25%–35% of circular plasmid DNA (data not shown). Prior to droplet generation, 200 ng of total DNA was incubated for 16 h at 37°C with 50 units/μg of PS-DNase in 33 mM Tris-acetate (pH 7.5), 66 mM potassium acetate, 10 mM magnesium acetate, 0.5 mM dithiothreitol, and 1 mM adenosine triphosphate. PS-DNase was then inactivated with a 20 min incubation at 80°C, and samples were diluted to 2 ng/μL in 10 mM Tris-Cl (pH 8.5) and 0.05% pluronic F-68. For the ddPCR reactions, 5 μL of dilute sample was used.

ddPCR was also performed with another set of samples treated with both PS-DNase and KpnI restriction enzymes to quantify individual vector genome units within circular concatemers. DNA samples were treated with PS-DNase, heat inactivated, and diluted, as previously described.⁶ Prior to the droplet generation step, 4 units of KpnI-HF enzyme (New England Biolabs, Ipswich, MA, USA) were added to the ddPCR reaction mix and incubated at 37°C for 30 min.

ddPCR procedures

Quantities of vector genome forms in samples were measured with ddPCR, which captures individual DNA molecules in thousands of water-oil emulsion droplets prior to PCR amplification with fluorescent tags. Individual droplets are then counted as negative or positive using fluorescence, and Poisson statistics are applied to the fraction of positive droplets to estimate the copy number of target DNA molecules per sample. In this analysis, a variant of ddPCR called drop-phase ddPCR was performed to detect and quantify the levels of paired target sequences together on a single DNA molecule to measure the contiguity of the DNA molecule using 2 different fluorescent tags (FAM and HEX). The number of double-positive droplets is then calculated, and the total copy number of molecules with both target sequences is estimated using the software QuantaSoft (Bio-Rad, Hercules, CA, USA), which includes an algorithm that accounts for the probability that some double-positive droplets may occur due to chance. ddPCR of the endogenous gene TFRC (transferrin receptor protein 1) was also performed to provide a normalization reference for calculating vector copy numbers per diploid genome from the concentration of droplets containing HEX or FAM fluorophores.

The primer and probe sets used for ddPCR and drop-phase ddPCR are presented in Table S1. Fusions of the 5′ and 3′ ITRs (head-to-tail) were detected using forward and reverse primers located on the 5′ and 3′ ends of the linear genome that only produced an amplicon when 5′ and 3′ ITRs were fused. Cross-ITR amplification was performed in a 3-step ddPCR reaction with 35 s annealing at 59°C, 65 s further extension at 72°C, and 30 s denaturation at 94°C. Amplification was detected at high efficiency from circular constructs containing head-to-tail ITR fusions but not from linear constructs, head-to-head constructs, or starting vector. Full-length vector genomes capable of giving rise to stable hFVIII-SQ transcription were detected with drop-phase ddPCR reactions identifying the co-occurrence of amplicons of R1 and R11, which overlap with the D-segments of the ITR on the 5′ and 3′ ends of the genome, respectively (Figure S2).

Following treatment with KpnI and/or PS-DNase, 10 ng of DNA were used in each ddPCR reaction. The reaction mixture contained 1x ddPCR Supermix for Probes without deoxyuridine triphosphate (Bio-Rad), 250 nM each of forward and reverse primers, 900 nM of probes, and 5 μL of sample for a final reaction volume of 25 μL. A Bio-Rad Auto Droplet Generator was used to generate droplets from reaction mix and QX200 Droplet Generation Oil for Probes (Bio-Rad), which were then transferred into a 96-well plate. PCR was performed in a C1000 Touch Thermal Cycler (Bio-Rad) as follows: 10 min at 95°C, 40 cycles of 30 s at 95°C and 1 min at 58°C, 10 min at 98°C, and hold at 4°C. Samples were read using a QX200 droplet reader (Bio-Rad), and the total concentration of target sequences and linked copies of target sequences were processed with QuantaSoft software (Bio-Rad).

Drop-phase ddPCR for sites A, B, C, and D was performed similarly, without enzymatic pre-treatment. The fraction of unconnected amplicons for each site was calculated by subtracting the percentage of linkage for that site to each adjacent site, compensating for the linkage of those 2 sites to each other. For example,

$$Fraction\left(unlinkedB\right)=1-(linkage\left(A\&B\right)+linkage\left(B\&C\right)-linkage(A\&C)).$$ (Equation 1)

Drop-phase ddPCR results were validated using linear and circular dsDNA fragments containing the full AAV5-hFVIII-SQ vector genome with both ITRs. Figure S5A shows R1 and R11 duplex ddPCR results using a linear DNA fragment containing the full AAV5-hFVIII-SQ vector genome with both ITRs as a template, with or without KpnI digestion (which cuts within the middle of the genome). Primers and probes amplifying R1 and R11 regions correspond to the 5′ ITR D loop and partial promoter and 3′ ITR D loop and partial FVIII transgene, respectively (Figure S2). Without KpnI digestion, R1 and R11 single positive copy numbers are fairly similar to the R1/R11 double-positive copy numbers. A slightly lower number of R1/R11 double-positive copies was observed, possibly due to damage to the ends of the linear DNA fragments. With KpnI digestion to separate the R1 and R11 regions into 2 separate fragments, R1-R11 double-positive copies are absent, consistent with the fact that they are no longer connected. In contrast, R1 and R11 single positive copies are similar to reactions without KpnI digestion.

One caveat of using ddPCR is that one droplet captures a single molecule and counts it as one positive unit, regardless of whether it is a monomer or mulitimer containing multiple units of vector genomes. This means the number of vg units could be undercounted if concatemers exist in the samples. To that end, we used the restriction enzyme KpnI to make cuts within the circular genome and release individual vector genome DNA units from concatemers. Therefore, an increase in the quantities of vector genomes upon KpnI digestion suggests circular concatemers are present in the sample. In Figure S5B, circular monomeric DNA was generated using the same linear DNA fragment used previously and was subjected to drop-phase ddPCR using R1 and R11 primers and probes sets. Similar copy numbers were observed with or without KpnI digestion in each comparison for the monomeric DNA. In contrast, if multimeric circular controls as templates were used (Figure S5C), copy numbers increased upon KpnI digestion.

In situ hybridization to detect hFVIII-SQ DNA

Formalin-fixed paraffin-embedded mouse liver sections (5 μm) were collected on Superfrost Plus slides using RNase-free conditions. An RNAscope in situ hybridization protocol was performed using a Ventana Discovery Ultra Autostainer (Tucson, AZ, USA), RNAscope (Newark, CA, USA), Universal 2.5 Reagent Kit, and custom-generated hFVIII-SQ probes. Slides were counterstained with 4′, 6-diamidino-2-phenylindole, and mounted with Fluoromount G. (catalogue number [cat. no.] 17984-25, Electron Microscopy Sciences). Slides were imaged on an Axio Scan.Z1 (Zeiss) slide scanner using a Plan-Apochromat 20x/0.8 objective equipped with a Hamamatsu Orca Flash camera. The intensity of hFVIII-SQ DNA in situ hybridization signal within hepatocyte nuclei and/or hepatocyte cytoplasm was quantified using Visiopharm image analysis software (Hoersholm, Denmark).

Southern blot procedures

Southern blotting was used to identify the configuration of circular episomes. For each sample, 12 μg of DNA was digested with 50 units each of EcoRI and HindIII restriction enzymes (New England Biolabs), which do not cut within vector genomes for 2 h, followed by PS-DNase treatment at 37°C for 16 h. The reaction was halted by heat inactivation at 80°C for 20 min; an additional set of samples was also incubated at 37°C for 2 h with 10 units of KpnI. For all samples, DNA was isolated with phenol-chloroform extraction, precipitated in ethanol, and resuspended in 30 μL of nuclease-free water. Purified DNA samples were mixed with 6X gel loading dye (New England Biolabs), loaded into a 0.7% agarose gel containing 0.5X SYBR Safe DNA Gel Stain dye (ThermoFisher Scientific), and electrophoresis was performed at 30 to 35 V for 16 to 18 h at room temperature. DNA was depurinated in 0.5 M NaOH and 1.5 M NaCl denaturing buffer for 30 min at room temperature with gentle agitation, neutralized in 1.5 M NaCl and 0.5 M Tris-HCl (pH 7.0) neutralizing buffer for 30 min, and then soaked in 20X SSC transfer buffer (3M NaCl and sodium citrate [pH 7.0]) for 30 min.

DNA was then transferred to a positively charged nylon membrane using a Whatman Nytran SuPerCharge TurboBlotter system (Sigma-Aldrich, St. Louis, MO, USA) and immobilized with a UV Crosslinker (VWR, Radnor, PA, USA). Biotin-labeled probes binding to both ends of the vector genome were generated using the Biotin PCR labeling kit (PromoCell GmbH, Heidelberg, Germany); the sequences are presented in Table S2. Southern blot hybridization was performed overnight at 65°C using 30 ng of each biotinylated probe and 100 μg of sheared salmon sperm DNA (ThermoFisher Scientific). A ThermoFisher Chemiluminescent Nucleic Acid Detection Module kit was used to detect hybridization and the membrane was imaged with a ChemiDoc Imager (Bio-Rad).

Monomeric/multimeric AAV5-hFVIII-SQ vector genomes ladder generation

Multimeric forms of AAV5-hFVIII-SQ vector genomes were produced for use as a ladder using an enzymatic synthesis method.⁴⁶ Briefly, pJ200-hFVIIISQ plasmid was digested with AscI (50 U/μg DNA) to release an ITR-ITR fragment containing the entire vector genome for 3 h at 37°C followed by heat inactivation for 20 min at 80°C. The cleaved DNA was electrophoresed on a 0.7% agarose gel at 85 V for 2 h. The 5 kb fragment was excised from the gel and purified using a DNA gel extraction kit (Qiagen) according to the manufacturer’s instructions. Concentration of the DNA fragment was determined by NanoDrop 8000 (Thermo Fisher Scientific). Overall, 500 ng of DNA fragment was incubated with T4 ligase (800 U/μg DNA) for 2 h at 16°C and heat inactivated for 15 min at 65°C. To enrich circular DNA forms, samples were incubated with PS-DNase (5 U/μg DNA) at 37°C for 2 h and heat inactivated for 20 min at 80°C. Multimeric forms of AAV5-hFVIII-SQ vector genomes were purified using phenol-chloroform extraction⁴⁷ and ethanol precipitation with glycogen solution (VWR) as an inert carrier. In brief, 1/10 volume of 3M sodium acetate, pH 5.2, and 1/20 volume of glycogen solution (20 mg/mL in stock) were added to the nucleic acid solution. Either 1 volume of isopropanol or 2.5 volume of ethanol was added to the solution and gently mixed. The mixed solution was incubated at −20°C for 1 h and centrifuged at 10,000 RPM for 10–15 min at 4°C. The supernatant was discarded and gently rinsed with cold 70% ethanol followed by centrifugation at 10,000 RPM for 2–5 min at 4°C. Once supernatant was discarded, the pellet was air-dried at room temperature. The dried pellet was dissolved in nuclease-free water or TE buffer. Purified DNA was used as a multimeric control in Southern blot analyses and ddPCR assays.

RNA quantification analysis

Total RNA and DNA was extracted from liver samples using an AllPrep DNA/RNA micro kit (Qiagen, Hilden, Germany). Concentration of extracted RNA was measured using a NanoDrop 8000 spectrophotometer (Thermo Fisher Scientific) and then diluted to 200 ng/μL. For each sample, 2 μg RNA was reverse-transcribed to generate first-strand cDNA using SuperScript VILO Master Mix (Life Technologies, Carlsbad, CA, USA). cDNA for use as a matrix for the standard curve was generated by reverse-transcribing RNA extracted from untreated mouse liver tissue purified using a QIAquick PCR Purification kit (Qiagen). The amount of FVIII-SQ RNA was normalized to the quantity of input liver RNA, determined by absorbance at 260 nm using a spectrophotometer.

Quantitative real-time PCR reactions of 20 μL were set up with 2X iQ Multiplex Powermix (Bio-Rad), 300 nM of forward and reverse primers, 150 nM of fluorescent probes targeting hFVIII-SQ cDNA, 100 ng of cDNA, and DNase/RNase-free water (Invitrogen, Carlsbad, CA, USA). Primer and probe sequences are listed in Table S3. Samples were loaded into clear 384-well PCR plates and run on a Roche Light Cycler 480 II (Roche, Basel, Switzerland) at 95°C for 2 min then for 50 cycles at 95°C for 3 s and 60°C for 20 s. Fluorescent signals at 483–533 nm were used to detect the hFVIII-SQ amplicon, and cross-point cycle values were calculated for each reaction by the Light Cycler 480 Software v. 1.5.1 using Abs quant/second derivative max analysis (Roche).

hFVIII-SQ protein quantification

Levels of hFVIII-SQ protein in sodium-citrate-anticoagulated plasma samples were measured using a sandwich ELISA with a human-specific anti-FVIII capture (GMA-9023, Green Mountain Antibodies, Burlington, VT, USA) and a detection (F8C-eIA, Affinity Biologicals, Ancaster, ON, Canada) antibody pair to specifically measure human FVIII. Nunc MaxiSorp high-binding black polypropylene plates (Thermo Fisher Scientific) were coated with 4 μg/mL of anti-FVIII (domain A2) antibodies. Samples were diluted 1:10 in Diluent Buffer (6% BSA in 1X TBS-T) and incubated for 1 h at room temperature. Sheep anti-FVIII antibodies conjugated to horseradish peroxidase were added and incubated for 1 h at room temperature. After a final wash, QuantaBlu substrate solution (Thermo Fisher Scientific) was used for detection. Relative fluorescent units were detected on a Molecular Devices FlexStation 3 instrument (Molecular Devices, San Jose, CA, USA), and concentrations were extrapolated from a standard curve prepared by spiking clinical grade recombinant B-domain-deleted hFVIII (Xyntha) in human FVIII-deficient plasma. Raw data was acquired using SoftMax Pro 5.4.1 (Molecular Devices).

Statistical analysis

Pearson’s correlation between levels of duplex linear and circular R1-R11 linked genomes per diploid genome or ITR-fused genomes and levels of FVIII-SQ transcript per μg of total RNA was performed using Prism software (GraphPad Software, San Diego, CA, USA).

Distribution of materials and data

Materials and protocols will be distributed to qualified scientific researchers for non-commercial academic purposes. The AAV5-hFVIII-SQ vector and the AAV5-hFVIII-SQ vector sequence are part of an ongoing development program, and they will not be shared.