Fusion of Anthopleurin-B to AAV2 increases specificity of cardiac gene transfer

Abstract

AAV-mediated gene therapy has become a promising therapeutic strategy for chronic diseases. Its clinical utilization, however, is limited by the potential risk of off-target effects. In this work we attempt to overcome this challenge, hypothesizing that cardiac ion channel-specific ligands could be fused onto the AAV capsid, and narrow its tropism to cardiac myocytes. We successfully fused the cardiac sodium channel (Na_v1.5)-binding toxin Anthopleurin-B onto the AAV2 capsid without compromising virus integrity, and demonstrated increased specificity of cardiomyocyte attachment. Although virus attachment to Na_v1.5 did not supersede the natural heparan-mediated virus binding, heparan-binding ablated vectors carrying Anthopleurin-B eliminated hepatic and other extracardiac gene transfer, while preserving cardiac myocyte gene transfer. Virus binding to the cardiac sodium channel transiently decreased sodium current density, but did not cause any arrhythmias. Our findings expand the knowledge of attachment, infectivity, and intracellular processing of AAV vectors, and present an alternative strategy for vector retargeting.

INTRODUCTION

Heart disease accounts for 1 in every 3 deaths in the United States each year, at the rate of 1 death every 40 seconds.¹ Despite several decades of basic and clinical research efforts, conditions resulting from primary cardiomyocyte dysfunction, such as heart failure and arrhythmias, continue to lack long-term effective and safe therapeutic options.^2–5 Most current therapies focus on symptomatic control and decrease of disease progression or recurrence. A definitive cure remains elusive for most cardiac diseases.

Gene therapy has become a promising strategy to address this void in the therapeutic armamentarium. The adeno-associated virus (AAV) has arisen as one of the most promising vectors for this purpose,^6–8 due to its ability to infect non-dividing cells, low immunogenic and oncogenic profile, long-term transgene expression, and potential to correct genetic mutations through in vivo gene targeting.⁹ As it relates to myocardial applications, AAV has been suggested to have less infection efficiency than adenovirus,¹⁰ but a much longer duration transgene expression.¹¹ However, clinical utilization of AAV as a cardiac gene therapy vector is potentially limited by the risk of off-target effects, given the promiscuous nature of virus attachment across multiple tissue types and organ systems, and the lack of cardiac specificity in naturally occurring AAV serotypes.¹² Despite the many excellent contributions made to date towards overcoming this limitation,^13–17 viral tropism remains largely indiscriminate, and the need for a highly specific AAV vector for gene therapy still exists.

In this work we hypothesized that cardiac ion channel-specific ligands could be fused onto the AAV capsid, using ion channels to retarget the vector to cardiomyocytes. As a proof of concept, we report that the modified sea anemone toxin Anthopleurin-B(R12S,K49Q) (ApB), can be successfully packaged onto the AAV2 capsid, and that this modification increases specificity of cardiac myocyte attachment without compromising infection efficiency. This strategy could potentially be employed with other tissue-specific membrane proteins and any AAV serotype to broadly increase specificity of gene transfer.

MATERIALS AND METHODS

Plasmids

pXX6, pIM45, pTRdsRed1, pVP1,3 (pIM45_T138A), and pVP2-GFP (pIM45_{T138M, EagI}eGFP_MluI) where provided by Nicholas Muzyczka, PhD ¹⁸. Plasmid containing a mutant form of Anthopleurin-B (ApB_R12S,K49Q) was provided by Kenneth Blumenthal, PhD.¹⁹ ApB_R12S,K49Q was amplified by standard PCR using 5’EagI-ApB_R12S,K49Q and 3’MluI-ApB_R12S,K49Q primers. ApB_R12S,K49Q amplicons and pVP2-GFP were digested with EagI and MluI restriction enzymes (NEB, Ipswich, MA) to create sticky ends, and subsequently ligated with T4 DNA Ligase (NEB, Ipswich, MA). pVP2-ApB integrity was confirmed by restriction digestion and DNA sequencing (MWG Biotech Huntsville, AL). pIM45_R585,588A, pVP1,3_R585,588A, and pVP2-ApB_R585,588A were generated by site-directed mutagenesis performed with QuickChange II XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) using previously published primers:²⁰ FWD:GTATCTACCAACCTCCAGGCGGGCAACGCGCAAGCAGCTACCGCAGATGTC. REV:GACATCTGCGGTAGCTGCTTGCGCGTTGCCCGCCTGGAGGTTGGTAGATAC. All mutations were confirmed by DNA sequencing (MWG Biotech, Huntsville, AL).

Cells

AAV-293 (Agilent, Santa Clara, CA), H9C2 (ATCC, Manassas, VA), and HEK293 (ATCC, Manassas, VA) cells were grown with DMEM, 10% FBS and 1% P/S. A stable line of HEK293 cells expressing functional cardiac sodium channels (α-subunit of Na_v1.5) was provided by Ira Cohen, MD, PhD. These cells were grown with DMEM, 10% FBS, 1% P/S, and 1% G148. Media from all cells was changed every 2–3 days. Neonatal rat ventricular myocytes (NRVM) were freshly isolated from 2 day old Sprague-Dawley rat neonates (Charles River Laboratories, Wilmington, MA). Briefly, hearts were harvested and minced in ice-cold calcium-and magnesium-free 1x HBSS solution, and incubated rocking for 12 hours at 4°C in sterilized 2 mg/ml Trypsin solution (Sigma, St. Louis, MO). Subsequently, the tissue solution was incubated for 1 minute at 37 °C with 5% FBS, 1% P/S DMEM to inactivate the enzyme. After liquid phase removal, minced hearts were incubated in 0.5 mg/ml collagenase solution (Worthington Biochemical Corporation, Lakewood, NJ) and shaken at 125 RPM for 2 minutes at 37°C. Cell-containing supernatant was aspirated and incubated in ice; more collagenase solution was then added to remaining pieces of myocardium to repeat the process until all tissue was dissolved. Cell solution was filtered with a cell strainer (BD Biosciences, San Jose, CA), and spun at 2000 rpm for 5 minutes at 4°C. Supernatant was discarded and pellet was resuspended with feeding medium (5% FBS, 1% P/S DMEM), and subsequently plated in T175 flasks, incubating it for 2 hours at 37°C, 5% CO₂ to remove non-cardiomyocyte cells. Finally, cellular supernatant was removed, cells counted, and re-plated to 70% confluence in feeding medium containing 1:200 BrdU (Sigma, St. Louis, MO), and incubated at 37°C, 5% CO₂. After 48 hours, feeding media was exchanged by new one containing 1:2000 BrdU. All NRVM experiments were performed after 5 days of initial cellular plating, when cardiomyocytes were spontaneously contracting.

Virus production

AAV vectors were made using standard triple and quadruple calcium phosphate transfection protocols with a 1:1:1 or 1:1:1:1 stoichiometry in AAV-293 cells, as previously described.²¹ Helper plasmid pXX6 and genome plasmid pTRdsRed1 were common to all vectors. Capsid plasmids were as follows: pIM45 for AAV2; pIM45_R585,588A for HA-AAV2; pVP1,3 and pVP2-ApB for AAV2-ApB; and pVP1,3_R585,588A and pVP2-ApB_R585,588A for HA-AAV2-ApB. After cell lysis and Benzonase (Sigma, St. Louis, MO) treatment, vectors were purified by discontinuous iodixanol gradients ultracentrifugation (Sorvall Ultracentrifuge Ultra Pro 80), and subsequently desalted, concentrated, sterilized, and stored in aliquots at −80 °C. Virus titer was determined by qPCR ²². HA-AAV2 and HA-AAV2-ApB vectors for in vivo use were produced by Vector BioLabs (Philadelphia, PA).

Capsid protein immunoblotting

Virus capsid composition was analyzed by standard immunoblotting techniques. Briefly, 60 μg of protein from purified virus stocks were denatured by heating at 95 °C for 5 minutes, fractionated by SDS-PAGE (7% Tris-Acetate gel, XT Tricine buffer, Bio-Rad, Hercules, CA), transferred to a nitrocellulose membrane, blocked with 5% non-fat dry milk, and blotted for 8 hours at 4 °C with anti-AAV Ab clone B1 (mouse, 1:500, American Research Products, Inc., Belmont, MA). Membrane was washed with blocking buffer and incubated with anti-Mouse Ab (1:10000, SCBT, Dallas, TX) conjugated with horseradish peroxidase. Bands were detected with ECL Supersignal (Pierce, Thermo Scientific, Rockford, IL).

DNA and RNA isolation, purification, and processing

Total DNA from virus stocks, heparin column chromatography, and attachment, infectivity, and biodistribution experiments was isolated and purified with DNeasy Blood and Tissue kit (Qiagen, Valencia, CA). Total DNA and RNA from H9C2 infectivity experiments were isolated and purified with AllPrep (Qiagen, Valencia, CA). DNA and RNA concentration was determined with a nanodrop, and samples were subsequently normalized. cDNA was generated from isolated RNA using SuperScript III First-Strand Synthesis Supermix (Invitrogen, Carlsbad, CA), following manufacturer’s instructions, with a PTC-200, Peltier Thermal Cycler. RT-PCR (cDNA) product was directly added to the qPCR plate for quantification. Double stranded DNA (dsDNA) isolation was performed by ssDNA digestion of total DNA using S1 Nuclease (Promega, Madison, WI) at a working concentration of 1U/ng DNA, according to manufacturer’s instructions. Enzymatic solution was initially incubated at 37 °C for 30 minutes, and subsequently at 70 °C for 10 minutes to inactivate the enzyme, prior addition of 1 μl of Stop buffer (0.3 M Tris -pH8.0- + 0.05 M EDTA) per tube. Solution was diluted 10 times prior to qPCR quantification, in order to prevent inactivation of PCR reaction.

qPCR experiments

dsRed1 and 28S genes were quantified by standard qPCR methods using a 7500 Real Time PCR System, (Applied Biosystems, Life Technologies Corporation, Carlsbad, CA), Power SYBR Green PCR Master Mix (#4367659, AB), and 500 nM of primers as follows:

• dsRed1: FWD: 5’-GTGAAGCTGAAGGTGACCAAG-3’;

• REV: 5’-GCTTCTTGTAGTCGGGGATGT-3’.

• 28S: FWD: 5’-GTTGTTGCCATGGTAATCCTGCTCAGTACG-3’;

• REV: 5’-TCTGACTTAGAGGCGTTCAGTCATAATCCC-3’.

All amplifications were performed in triplicate, and both dsRed1 and 28S reactions from the same samples were run simultaneously in the same plate. CT values were translated into absolute GC/ml values using the standard curve method from known amplicon quantities. The amplicon level of detection was 2.59 GC/ul.

Transmission electron microscopy

5 μl of AAV2(R585,588A) and AAV2-ApB(R585,588A) purified stock solutions (1e12 GC/ml) were loaded onto a 400 mesh carbon-coated Formvar copper grids (Electron Microscopy Sciences, Hatfield, PA) and left for 1 min at room temperature. Excess fluid was drawn off, and the sample was washed three times with PBS. 5 μl of 2 % uranyl acetate was subsequently added for 10 seconds, and the grid was dried at room temperature for 10 min before viewing under a TECNAI G² 12 transmission electron microscope (FEI, Hillsboro, OR). Photos were taken with AMT camera (AMT Corp., Woburn, MA). Capsid size was determined by measuring and averaging the three intersecting diagonals of each full virus particle (n=150 sequential full particles) under 50% magnification of 3.5e5x photos (7e5x final magnification). The number for full and empty capsids was also recorded from 150 sequential particles from each virus type, at the same magnification.

Patch clamp analysis of sodium current

In order to assess binding of ApB to Nav1.5 when fused to a larger protein, we exposed 500 nM of GFP-ApB or AAV2-ApB to freshly isolated NRVM in Tyrode’s solution, and performed whole cell recordings of the cardiac sodium current (I_Na). Cells were exposed to control patching solution (CTL x 2, n=10 each time), AAV2 (n=9), AAV2(R585,588A) (n=10), AAV2-ApB (n=10), or AAV2-ApB(R585,588A) (n=10). Cells were placed in a 30 mm dish with 2 ml of patching solution, and virus was added to the solution for a final working concentration of 5e8 GC/ml. I_Na was recorded by ruptured-patch whole cell voltage clamp at room temperature. Microelectrodes were pulled from borosilicate capillary glass and lightly fire-polished to resistance 0.9–1.5 MΩ when filled with electrode solution of CsF 120 mM, MgCl2 2 mM, HEPES 10 mM, EGTA 11 mM, TRIS GTP 0.3 mM, phosphocreatine 14 mM, Mg ATP 4 mM, creatine phosphokinase 2 mM, and brought to a pH of 7.3. NRVM were placed in the solution containing NaCl 25 mM, N-methyl D-glucamine 120 mM, CsCl 5 mM, MgCl2 1 mM, CaCl2 2 mM, NiCl2 1 mM, glucose 10 mM, HEPES 10 mM (pH 7.3). I_Na was elicited from a holding potential of −75 mV with depolarizing voltage pulses to −20 mV or −10 mV for 16 ms. The peak current density (pA/pF) was calculated from the ratio of current amplitude to cell capacitance. Command and data acquisition were operated with an Axopatch 200B patch clamp amplifier controlled by a personal computer using a Digidata 1200 acquisition board driven by pCLAMP 9.0 software (Axon Instruments, Foster City, CA). In order to control for individual variability in electrophysiological parameters, all groups from each experiment were tested from the same myocyte isolation.

Heparin column affinity chromatography

In order to evaluate the heparan-binding ability of the modified AAV2 vectors, heparin column affinity chromatography was performed as follows: 1-ml HiTrap heparin HP columns (GE Healthcare, Pittsburgh, PA) were equilibrated by washing with 10 ml of TD buffer (dH2O, 5× PBS, 5 mM MgCl2-6H2O, 12.5 mM KCl). Firstly, 1 ml of 1 × 10⁹ GC/ml of purified virus (AAV2, AAV2-ApB, HA-AAV2, and HA-AAV2-ApB) was applied to the column; columns were then washed with 1 ml of TD buffer; and finally, 1 ml of eluting solution (TD/1 M NaCl buffer) was applied to the columns. The flow rate for all chromatography steps was 1 ml/min. Virus concentration of the flow-through from each step was quantified by qPCR after total DNA isolation.

Intracellular genome processing experiments

In order to evaluate the intracellular genome processing of AAV2-ApB in cardiac myocytes, we conceived the assessment three major determinants of infectivity:

Virus exposure time

1×10⁵ H9C2 cells were plated into T12.5 flasks; after 24hs, 1×10⁹ GC of virus was added to the flasks (1e9 GC/ml) and incubated at 37 ºC, 5% CO2 for 1 minute, 15 minutes, 1 hour, 6 hours, 24 hours, and 72 hours. After incubation time, virus-containing media was discarded; cells were washed 3 times with PBS, and subsequently covered with fresh virus-free media. Cells were harvested after 12 days of incubation.

Cell harvest time

1×10⁵ H9C2 cells were plated into T12.5; after 24hs, 1×10⁹ GC of virus was added to the flasks (1e9 GC/ml) and incubated at 37 ºC, 5% CO2 for 24 hours. After incubation time, virus-containing media was discarded; cells were washed 3 times with PBS, covered with fresh virus-free media, and subsequently incubated for 1, 3, 6, 9, and 12 days (from the time of infection). Cells were harvested after incubation.

Virus infection titer

1×10⁵ H9C2 cells were plated into T12.5 flasks; after 24hs, 1×10⁶, 1×10⁷, 1×10⁸, and 1×10⁹ GC of virus was added to the flasks (in 1 ml total volume) and incubated at 37 ºC, 5% CO2 for 24 hours. After incubation time, virus-containing media was discarded; cells were washed 3 times with PBS, and subsequently covered with fresh virus-free media. Cells were harvested after 12 days of incubation.

All cell flask infection experiments were done in triplicate. After final incubation, media was discarded; cells were washed three times with cold PBS, and harvested with RLT buffer by scraping. Cellular solution was homogenized with QIA shredder (Qiagen, Valencia, CA) and frozen at −80 °C, for later DNA and RNA purification and quantification by qPCR.

Attachment experiments

1×10⁶ H9C2, NRVM, HEK293, and HEK293-Na_v1.5(α) cells were plated into T12.5 flasks; after 24hs, media was discarded, cells were abundantly washed three times with chilled PBS, and incubated for 30 minutes in 1 ml of serum-free DMEM containing 80 μM of Dynasore (Sigma, St. Louis, MO) at 37 °C, 5% CO2, to arrest endocytosis. Immediately after, 1×10⁸ GC/ml of AAV vectors was added to the cells and incubated at 37 ºC, 5% CO2 for 60 minutes in Dynasore-containing media. Cells were subsequently incubated on ice (4 °C) for 5 minutes, washed three times with PBS to detach loosely bound virus, and harvested by scraping (n=3 per data point). Total DNA was isolated and quantified by qPCR.

Infectivity experiments

2×10⁶ H9C2 and NRVM cells were plated into T12.5 flasks; after 24hs, cells were infected with AAV vectors (1×10⁸ GC/ml) and incubated for 24 hours at 37 ºC, 5% CO2. After incubation time, virus-containing media was discarded; cells were washed 3 times with PBS, and subsequently covered with fresh virus-free media for 9 days. After incubation, media was discarded; cells were washed three times with PBS, and harvested by scraping. All cell flask infectivity experiments were done in triplicate. Total DNA was isolated and quantified by qPCR.

Biodistribution experiment

Eleven Sprague-Dawley rats were randomized and anesthetized (5% + 2.5% isoflurane) prior to delivery of 1×10¹² GC of AAV vectors via tail vein injection (HA-AAV2 n=5, HA-AAV2-ApB n=5, PBS n=1). A baseline ECG was taken from each animal after sedation and prior to injection. Cardiac telemetry was performed for 10 minutes immediately after virus injection. Animals underwent routine care, and were sacrifice after 6 weeks. Major organs (brain, testes, lung, heart, liver, spleen, kidneys, pancreas, intestine) were harvested and snap frozen in liquid nitrogen. Frozen tissue was pulverized with a mortar and pestle; total DNA was isolated and purified with DNeasy (Qiagen, Valencia, CA), quantified with a nanodrop and normalized. dsRed1 and 28S genes were absolutely quantified by qPCR. The animals for this study were maintained in accordance with the Policy on Humane Care and Use of Laboratory Animals from the Office of Laboratory Animal Welfare, National Institutes of Health. The experimental protocol was approved by the Institutional Animal Care and Use Committee.

Statistical Analysis

Subjects were randomized to treatment group, and investigators were blinded to subject study group during data assessment. Continuous variables with normal distribution were assessed using the student’s t-test. For data that were discontinuous, categorical, or not normally distributed, we assessed differences using nonparametric Fisher’s exact or Mann Whitney U tests. Multiple measures were corrected using the method of Bonferroni. Data are presented as mean ± SEM or median, [range].

RESULTS

Successful packaging of ApB onto the AAV2 capsid

The modified sea anemone toxin ApB is a high-affinity, high-specificity cardiac sodium channel (Na_v1.5) ligand.¹⁹ Prior study of ApB has shown no effect on peak sodium current (I_Na) and increase in the late sodium current.^23,24 To confirm continued sodium channel binding ability of ApB when bonded to larger particles, we initially fused ApB to green fluorescent protein (GFP) and assessed sodium channel binding ability of the ApB-GFP fusion product. Our results showed that ApB continued to bind the sodium channel, but the nature of the interaction changed. The ApB-GFP fusion product reduced peak I_Na by 77% and had no effect on the late current (Suppl. Fig.1).

We next inserted ApB into the open reading frame of the AAV2 cap gene after residue T138, using the plasmid pVP2-GFP (pIM45_{T138M, EagI}eGFP_MluI) that overexpresses VP2 (Suppl. Fig.2). We confirmed successful insertion by sequencing and then we produced virus by transfecting AAV-293 cells with the plasmids pXX6 (helper), pTRdsRed1 (genome), pVP1,3 (pIM45_T138A; providing VP1 and VP3 capsid proteins), and pVP2-ApB (pIM45_T138M, _EagIeApB_MluI; providing the modified VP2 capsid protein) with a 1:1:1:1 stoichiometry. We harvested and purified the resulting virus and confirmed expression of the modified capsid by Western blot analysis (Fig. 1a). The modified VP2 capsid protein extracted from AAV2-ApB virus was 5 kD heavier than the unmodified VP2 protein due to the 49 amino acid insert, confirming the presence of VP2-ApB in the AAV2-ApB virus.

Figure 1. Structural and functional characterization of AAV capsid modifications.

(a) Western blot of AAV2 and AAV-ApB capsids showing capsid proteins VP1, VP3 and VP3. The insertion of ApB at the N-terminus of VP2 is evident by the 5 KD increase in its molecular weight. (b) Percentages of AAV particles quantified by qPCR from the three separate recovered solutions after heparin column affinity chromatography (flow-through, wash, and eluate), confirming successful ablation of heparan-binding site. (c) AAV2 and AAV2-ApB particles loaded onto a 400 mesh carbon-coated Formvar copper grids visualized under a transmission electron microscope, at 3.5 × 10⁵X magnification (scale bar = 25 nm). (d) Measurement of the three intersecting diagonals of AAV full particles under 7 × 10⁵X effective magnification reveals that AAV-ApB is 3.1 nm (11%) smaller than AAV2 with no other evident morphological difference (n=150 per group).

We also created heparan-binding ablated AAV2-ApB by site-directed mutagenesis of the R585,588 sites to eliminate the heparan sulfate binding site, as previously described. ²⁰ Heparin column affinity chromatography was performed to assess the attachment properties of these vectors (Fig. 1b). Both wild type AAV2 and AAV2-ApB had strong heparin-binding ability. Both heparan-binding ablated AAV2 (HA-AAV2) and heparan-ablated AAV2-ApB (HA-AAV2-ApB) had almost complete loss of heparan binding.

Insertion of ApB induced contraction of the AAV2 capsid

In order to further evaluate the capsid changes induced by ApB insertion, we evaluated the vectors using transmission electron microscopy (Fig. 1c). Overall the shape of the AAV-ApB capsid was identical to that of wild type AAV2, but the AAV2-ApB particles were smaller on average than wild-type AAV2. We assessed virus size by measuring and averaging the three intersecting diagonals of full particles under 70,0000x effective magnification. Surprisingly, given the presence of the large peptide insert, AAV2-ApB particles averaged 3.1 nm (11% smaller than wild type AAV2 virus, p<0.0001) (Fig. 1d). The percent of empty capsids was 2% in AAV2 vs 7.3% in AAV-ApB (p=0.NS), suggesting that the packaging capacity of this new vector is not unaffected by this conformational change.

AAV2-ApB attached to the cardiac sodium channel

In order to assess the ability of AAV2-ApB to bind to the cardiac sodium channel as hypothesized, we exposed AAV2, AAV2-ApB, HA-AAV2 and HA-AAV2-ApB vectors to freshly isolated neonatal rat ventricular myocytes (NRVM), and measured I_Na density (Fig. 2). NRVM exposed to both AAV2-ApB and HA-AAV2-ApB vectors had a statistically significant 52% and 55% decrease in peak I_Na compared to control cells (p=0.0002 and p=0.022, respectively), whereas cells exposed to wild-type AAV2 and HA-AAV2 did not display any change in sodium current compared to unexposed cardiomyocytes. These results confirm the presence of ApB on the viruses, and they demonstrate intact sodium channel binding of ApB when expressed within the virus capsid.

Figure 2. Sodium current evaluation of cardiac myocytes exposed to AAV2 and AAV-ApB vectors.

Peak I_Na density of isolated neonatal rat cardiomyocytes exposed to (a) AAV2 and AAV2-ApB, or (b) HA-AAV2 and HA-AAV2-ApB vectors, compared to unexposed control myocytes (n = 9 for AAV2, n=10 for all other groups). Peak I_Na density decreases by 52% and 55%, respectively, after AAV2-ApB and HA-AAV2-ApB exposure.

ApB preserved the attachment efficiency to cardiac myocytes after ablation of the AAV2 heparan-binding ligand

We evaluated the attachment efficiency of AAV2, AAV2-ApB, HA-AAV2, and HA-AAV2-ApB vectors to immortalized (H9C2 line) and primary cultures (NRVM) of cardiac myocytes (Fig. 3a). To evaluate attachment, cells were pre-treated with dynasore, a cell-permeable inhibitor of dynamin, which arrests endocytosis by preventing the formation of clathrin-coated vesicles.^25,26 We found attachment efficiency to be superior in immortalized cardiomyocytes compared to isolated cardiomyocytes, but there was no statistically significant difference between AAV2 and AAV2-ApB vectors in either H9C2 cells or NRVMs.

Figure 3. Attachment of AAV and AAV-ApB vectors.

Attachment efficiency to (a) isolated and immortalized cardiac myocytes, and (b) non-cardiac cells with and without expression of the cardiac sodium channel (Nav1.5) was evaluated by DNA quantification of dsRed, normalized to 28S. Cells were first chilled with PBS, and pretreated with Dynasore to prevent internalization of particles after virus exposure. Cells were exposed to virus for 60 minutes, subsequently incubated at 4 °C for 5 minutes, and washed three times to detach loosely bound virus; cells were then harvested by scraping (n=3 per data point). Attachment of HA-AAV-ApB is superior to that of HA-AAV in both immortalized and isolated cardiomyocytes (p<0.01 and 0.03 respectively), as well as in non-cardiac cells stably expressing the cardiac-sodium channel (p=0.04); attachment to non-cardiac cells lacking Na_v1.5 is not statistically different between vectors.

We next assessed the AAV vectors that had heparin-binding ablated. As expected, HA-AAV2 attachment efficiency was markedly reduced in both types of cardiomyocytes due to the ablation of the primary viral binding site. On the contrary, HA-AAV2-ApB had preserved virus binding to cardiac myocytes. In both immortalized and primary cultures of cardiac myocytes, HA-AAV2-ApB attachment was not statistically different than wild type AAV2 or AAV2-ApB, and it was significantly better than HA-AAV2.

ApB increased attachment efficiency in non-cardiac cells expressing functional cardiac sodium channels

We also performed identical attachment experiments on non-cardiac cells (HEK293) using the heparin ablated vectors (Fig. 3b). We found that attachment to non-cardiac cells is not significantly different between vectors. However, HA-AAV2-ApB attachment efficiency in Na_v1.5-expressing HEK293 cells was significantly superior to that of HA-AAV2, albeit to a lesser degree than that in either immortalized or isolated cardiac myocytes.

DNA second strand synthesis and RNA transcription were not affected by capsid insertion of ApB onto AAV2

We compared intracellular processing of AAV2-ApB to wild type AAV2 in H9C2 cells as a function of virus dose (1e6 - 1e9 GC/ml), exposure time (1 minute - 72 hours), and post-exposure harvest time (1 – 12 days) (Fig. 4). We measured total transgene DNA as a marker of virus internalization, the transcriptionally active double-stranded transgene DNA (dsDNA) as a marker of successful virus internal processing (uncoating through second strand synthesis steps),²⁷ and transgene RNA as a marker of effective gene transcription. The low virus titers prevented visualization of dsRed1 protein expression under fluorescent microscopy (data not shown). Overall, we saw no differences between wild type AAV2 and AAV2-ApB for any of these variables, showing that ApB expression within the AAV2 capsid did not disrupt internalization or intracellular processing.

Figure 4. Evaluation of intracellular genome processing of AAV2 and AAV-ApB vectors in cardiac myocytes.

qPCR quantification (dsRed/28s) of total DNA, double-stranded DNA (dsDNA), and RNA from immortalized cardiomyocytes subjected to three different AAV infectivity variables: Virus titer (1×10⁶ – 1×10⁹ GC/ml), virus exposure time (1 minute - 72 hours), and post-exposure harvest time (1 – 12 days) (n=3 per each data point).(*p=0.048). Overall, AAV-ApB and AAV2 had identical intracellular processing.

We found that internalized transgenic DNA, dsDNA and RNA increased as a function of virus dose for both wild type AAV2 and AAV2-ApB. Transgene was undetectable with infection titer of 1e6 GC/ml. For the dose range of 1×10⁷ – 1×10⁹, we found a 10–15 fold increase for every log order increase in virus titer.

Regarding the exposure time variable, we found that infectivity does not significantly change with increasing exposure time up to 12 hours. This seems to suggest that the attachment process is completed very early after initial virus exposure.

Regarding time from initial exposure to harvest of infected cells, dsDNA comprised ~5–10% of total internalized DNA as early as 24 hours post-infection, and this ratio remained stable for at least 12 days. Both total and double stranded viral DNA decreased over time following infection, and RNA transgene progressively increased throughout the experiment timeline. Overall, there were no significant differences comparing AAV2 and AAV2-ApB, indicating that attachment, infectivity, and intracellular processing remained intact for AAV2-ApB.

Using the optimal conditions from the AAV2 and AAV2-ApB experiments, we subsequently evaluated infectivity of the heparan-ablated vectors in both immortalized and freshly isolated cardiomyocytes. We infected cells with 1×10⁸ GC/ml and incubated for 24 hours at 37 ºC, 5% CO2, and harvested cells after 9 days, for subsequent total DNA genome quantification using qPCR (Fig. 5). HA-AAV2-ApB was statistically superior to HA-AAV2 in both H9C2 cells and NRVMs, by 1.7-fold and 4-fold, respectively (p<0.01 each). These findings further confirm the positive effect of ApB on cardiomyocyte attachment and infectivity.

Figure 5. Infectivity of HA-AAV and HA-AAV-ApB in cardiac myocytes.

Isolated and immortalized cardiac myocytes were infected with 1×10⁸ GC/ml of HA-AAV2 and HA-AAV-ApB vectors and incubated for 24 hours, prior to cell washing and further incubation for 9 days in virus-free media. After harvesting, gene transfer was quantified by absolute qPCR DNA quantification of dsRed, normalized by 28s, revealing that HA-AAV2-ApB infectivity is superior to HA-AAV2 in both cell types (n=3 per data point).

HA-AAV2-ApB preserves cardiac binding and reduces off-target binding in vivo

We assessed the biodistribution of the heparan binding-ablated vectors in Sprague-Dawley rats after tail vein infusion. Since the in vitro results showed that heparan binding behavior dominated for AAV2 and AAV2-ApB, for these experiments we only evaluated animals receiving the heparan-binding ablated HA-AAV2 or HA-AAV2-ApB. In these in vivo experiments, off-target gene transfer was considerably reduced with HA-AAV2-ApB (Table). Of the 5 subjects receiving HA-AAV2-ApB, gene transfer to the liver was completely eliminated in all 5; transgene was also undetectable in several other organs after HA-AAV2-ApB exposure, including the brain (undetectable in 4 of 5), pancreas (undetectable in 4 of 5) gonad (undetectable in 2 of 5), and small intestine (undetectable in 1 of 5). In contrast, all animals in the HA-AAV2 group had detectible transgene in all of these organs, except 2 HA-AAV2 animals had no detectible transgene in the pancreas. Cardiac gene transfer was slightly reduced by HA-AAV2-ApB. Immunohistochemistry and fluorescent microscopy of histological slides did not confirmed dsRed1 protein expression of either virus (data not shown).

Table.

Transgene expression after tail vein infusion in rats

	HA-AAV2-ApB		HA-AAV2		p
	median	range	median	range
heart	1.1 × 10−6	3.3 × 10−7, 2.1 × 10−6	6.0 × 10−6	2.5 × 10−6, 1.2 × 10−5	0.03
liver	0	0, 0	4.4 × 10−7	1.5 × 10−7, 8.3 × 10−7	0.007
lung	5.0 × 10−7	0, 1.7 × 10−6	4.1 × 10−6	3.1 × 10−7, 1.2 × 10−5	0.15
brain	0	0, 2.8 × 10−6	1.5 × 10−6	3.6 × 10−7, 2.2 × 10−6	0.15
gonad	1.2 × 10−7	0, 2.6 × 10−7	4.8 × 10−7	3.1 × 10−7, 9.4 × 10−7	0.007
kidney	3.6 × 10−6	1.8 × 10−6, 1.6 × 10−5	2.1 × 10−5	4.2 × 10−6, 1.9 × 10−3	0.03
spleen	1.2 × 10−3	2.7 × 10−4, 6.1 × 10−3	6.1 × 10−4	2.2 × 10−4, 4.8 × 10−3	0.42
pancreas	0	0, 6.2 × 10−8	1.7 × 10−8	0, 2.9 × 10−7	0.34
intestine	1.4 × 10−6	0, 5.5 × 10−6	2.4 × 10−6	7.7 × 10−7, 8.8 × 10−5	0.70

Values are given as gene copies of transgene per gene copy of 28S

Given the potential, though unlikely, pro-arrhythmic risk of transient, very low dose I_Na blockade by AAV-ApB vectors, we subjected the animals to 10 minutes of telemetry monitoring during and immediately after tail vein injection. All animals remained in sinus rhythm, with no evidence of arrhythmias. We saw no ECG changes in the animals receiving HA-AAV2-ApB compared to baseline ECG in those animals and compared to animals receiving HA-AAV2 and a sham animal that received only PBS. Specifically, we saw no QRS widening, AV block, atrial or ventricular tachy- or bradyarrhythmias with ApB exposure.

DISCUSSION

AAV2 is a promising vector for gene therapy applications, but its utility is limited by the widespread distribution of its primary receptors.^28–32 Several other AAV serotypes have been identified, some with much better cardiac infectivity than AAV2; however, off-target (i.e. non-cardiac) gene transfer is also much higher with these serotypes.¹² Given that AAV2 is the most widely studied serotype, it has been the focus of multiple efforts to retarget the capsid to specific cell types, mostly by inserting linear targeting sequences into the capsid VP3 ORF, which is the most abundant but smallest capsid protein.^16,33–35 Warrington et al. discovered that the VP2 capsid protein is able to tolerate large peptides insertions at the site T138;¹⁸ this site has also has been shown to be at the surface of the particle following insertion of smaller epitopes.³⁶ Subsequently, some major accomplishments have been made taking advantage of this finding, such as GFP insertion for in vitro capsid labeling,³⁷ or luciferase insertion for in vivo and in vitro virus quantification.³⁸ One other work recently reported the use of this capsid site to retarget the vector to tumors in a strategy similar to ours.³⁹

We confirmed our hypothesis that a high-affinity, high-specificity cardiac ion channel ligand can be packaged onto the AAV2 capsid, and allow infection of cardiomyocytes through a de novo receptor. As a proof-of-concept, we chose the voltage-gated cardiac sodium channel Na_v1.5, which is expressed in high abundance and specificity on cardiac myocytes. We then selected as ligand a modified version of the sea anemone toxin, Anthopleurin-B(R12S,K49Q), that has the ability to discriminate between cardiac and neuronal sodium channels like Anthopleurin-A, while still retaining the high-affinity binding properties of Anthopleurin-B.¹⁹ We successfully inserted this ligand at the N-terminus of the VP2 protein of the AAV2 by standard molecular techniques, creating the AAV2-ApB virus. Since we wanted to minimize the wild-type virus binding mechanism, and study the AAV-ApB-Na_v1.5 interaction in isolation, we performed site-directed mutagenesis to eliminate heparan sulfate proteoglycan binding capacity; we confirmed this functional deletion by heparin column affinity chromatography. Both the ligand insertion and VP2 overexpression was confirmed by immunoblotting, with an antibody that recognizes the C-terminus of all capsid proteins. Upon examination of the AAV2 and AAV2-ApB vectors under high-power transmission electron microscopy, we found that the size of the AAV2-ApB vectors is significantly smaller than AAV2. However, the percentage of empty capsids of this new vector was not significantly different than wild type AAV2, suggesting this size difference did not interfere with packaging of our transgene insert. A more comprehensive study with varying transgene sizes could be performed to more fully understand the possibility of alterations in packaging from the smaller virus size. To our knowledge, this is the first report of this finding. In addition, we confirmed that attachment, internalization and gene expression functions remained intact with AAV2-ApB. Consistent with published literature, the dsDNA of AAV2 vectors account for a very small percentage of the total genome input, due to slow formation, instability and degradation throughout, which is known to lead to poor transduction efficiency.^40–42 Despite this shortcoming, the AAV2 capsid remains still the best studied to date, and the most suitable for our capsid modification and retargeting purposes. A systematic study of the AAV2-ApB vector’s structural and functional properties is warranted, but currently outside the scope of this work.

We confirmed the preserved binding of AAV2-fused ApB to cardiac sodium channels by patch clamp analysis of freshly isolated cardiac myocytes, using both wild-type and heparan binding-ablated AAV2 vectors. Although the toxin is known to prevent the inactivation of the channel and increase sodium flux,¹⁹ AAV2-bound ApB causes blockade of I_Na, likely due tosteric hindrance with attachment of the toxin to the channel and then occlusion of the channel by the virus particle. Another possible mechanism is reduction of the current by channel endocytosis triggered by virus binding as has been described for other surface proteins. Since the number of blocked channels is small relative to the overall number of surface sodium channels, AAV2-ApB does not alter cardiac electrophysiology in any detectible way during in vivo systemic delivery in rats, and the survival and growth of these animals was comparable. Like any other sodium channel blocker, dose is the relevant variable here, and further use of AAV2-ApB should keep in mind that functional effects could be theoretically possible if the administered dose was high enough.

AAV2-ApB did not increase attachment or infectivity to cardiac myocytes while the heparan-binding site was intact, but ApB fusion to AAV2 preserved the cardiac myocyte attachment and infective properties when heparan binding was ablated. The preservation of cardiac myocyte binding and uptake with heparan binding-ablated AAV2-ApB illustrates the principle of selective replacement of target receptor with additional capsid modifications to eliminate off-target binding. Furthermore, HA-AAV2-ApB attachment to non-cardiac cells is not significantly different than that of HA-AAV2, as one would expect. Interestingly, however, when these same HEK293 cells are transformed to express the functional α-subunit of the cardiac sodium channel (Nav1.5), HA-AAV2-ApB attachment efficiency is significantly superior to that of HA-AAV2, confirming our hypothesis of Na_v1.5-mediated virus attachment through ApB.

Finally, we evaluated the biodistribution of these vectors in adult rats via systemic delivery through tail vein injection. The most important finding of the in vivo experiments was the complete elimination of liver gene transfer with AAV-ApB. Other off-target gene transfer was also significantly reduced. Cardiac gene transfer was slightly reduced in vivo, unlike the in vitro findings. The cardiac effect may indicate reduced escape from the vasculature of the larger particle, or some other difference between the in vitro and in vivo environment. Further experimentation is needed to define and eliminate this effect.

With the HA-AAV2-ApB vector, the vast majority of the transferred gene was present in the spleen after 6 weeks of infection, similar to other AAV vectors. This pattern of virus biodistribution has been associated with a high titer of pre-existent AAV neutralizing antibodies that mediates vector sequestration by follicular dendritic cells.^43–45 These data suggest that the remaining off-target gene transfer occurs by heparan-independent virus interaction. Further capsid modifications may improve cardiac binding and eliminate the remaining off-target binding as well.

CONCLUSION

In summary, we present the first known effort to retarget AAV to cardiac myocytes using an ion channel as a de novo receptor. In this work, we confirm our hypothesis that a high-affinity cardiac ion channel ligand can be packaged onto the overexpressed VP2 capsid protein, and increase the attachment and infection of cardiomyocytes in vitro, using the naturally occurring, cardiac sodium channel complex. These findings expand the knowledge of attachment, infectivity, and intracellular processing of AAV vectors and present an alternative strategy for vector retargeting. Further work is needed in this area, specifically to expand this strategy to more efficient AAV serotypes and to other cardiac specific ion channels and a combination of such ligands, and to further investigate the immunogenic properties of these novel custom capsids.

Supplementary Material

1. Supplemental Figure 1: ApB fusion product blocks the sodium current.

Example of sodium current tracings from cardiac myocytes exposed to Tyrode’s solution alone (black line) or 500 nM ApB-GFP in Tyrode’s solution (pink line). Peak current is reduced by 77% with ApB-GFP exposure, likely representing steric hindrance. Late current is not affected by ApB-GFP exposure (zoomed box).

2. Supplemental Figure 2: Capsid plasmids for AAV2 and AAV2-ApB vectors.

Capsid plasmid pIM45 was used for AAV2 vector production. pVP1,3 plasmid and pVP2-ApB were used for AAV2-ApB production, in 1:1 stoichiometry. The T138A mutation in pVP1,3 impedes the translation of VP2 protein; the T138M mutation in pVP2-ApB overexpresses the VP2-ApB capsid protein production, impeding the VP1 and VP3 production as well.

Highlights.

• The AAV2 capsid can accommodate insertion of the sodium channel binding toxin ApB

• ApB-AAV2 blocks sodium current indicating that it accessible from the outside of the virus and that it retains channel binding capacity.

• Intracellular processing of ApB-AAV2 did not differ significantly from AAV2.

• The combination of ApB insertion and heparan sulfate binding deletion in AAV2 led to a virus with increased specificity of cardiac myocyte attachment in vivo in rats.