Rational Design and Engineering of a Modified
`Adeno-Associated Virus (AAV1)-Based Vector
`System for Enhanced Retrograde Gene Delivery
`BACKGROUND: After injection into muscle and peripheral nerves, a variety of viral
`vectors undergo retrograde transport to lower motor neurons. However, because of its
`attractive safety profile and durable gene expression, adeno-associated virus (AAV)
`remains the only vector to have been applied to the human nervous system for the
`treatment of neurodegenerative disease. Nonetheless, only a very small fraction of
`intramuscularly injected AAV vector arrives at the spinal cord.
`OBJECTIVE: To engineer a novel AAV vector by inserting a neuronal targeting peptide
`(Tet1), with binding properties similar to those of tetanus toxin, into the AAV1 capsid.
`METHODS: Integral to this approach was the use of structure-based design to increase
`the effectiveness of functional capsid engineering. This approach allowed the optimi-
`zation of scaffolding regions for effective display of the foreign epitope
`while minimizing disruption of the native capsid structure. We also validated an
`approach by which low-titer tropism-modified AAV vectors can be rescued by particle
`mosaicism with unmodified capsid proteins.
`RESULTS: Importantly, our rationally engineered AAV1-based vectors exhibited mark-
`edly enhanced transduction of cultured motor neurons, diminished transduction of
`nontarget cells, and markedly superior retrograde delivery compared with unmodified
`AAV1 vector.
`CONCLUSION: This approach promises a significant advancement in the rational
`engineering of AAV vectors for diseases of the nervous system and other organs.
`KEY WORDS: AAV-mediated gene delivery, Motor neuron disease, Rational design, Retrograde axonal
`transport, Vector engineering, Vector targeting
`Neurosurgery 76:216–225, 2015 DOI: 10.1227/NEU.0000000000000589 www.neurosurgery-online.com
`A
`deno-associated virus (AAV) has emerged
`as a promising vector for gene delivery
`because of its broad-tissue tropism, safety,
`ability to transduce both quiescent and dividing
`cells, and ability to mediate long-term gene expres-
`sion. Recent isolation of novel AAV serotypes and
`advances in vector engineering have further
`enhanced the utility of these vectors. Gene therapy
`holds a variety of advantages for the treatment of
`a variety of neurological diseases. In several of these
`diseases, an optimal therapeutic response is facil-
`itated by vectors that are capable of efficient
`retrograde axonal transport. In Parkinson disease,
`preservation and repair of the nigrostriatal pathway
`may require the expression of trophic genes in both
`the striatum (postsynaptic) and substantia nigra
`(presynaptic).
`1 This type of targeting is facilitated
`by the ability of viral vectors to undergo retrograde
`axonal transport from the postsynaptic site into
`the presynaptic neurons. Similarly, retrograde
`transport can facilitate the delivery of genes to
`target neurons that are widely disbursed as in the
`case of treating corticospinal neurons to enhance
`spinal cord regeneration. Finally, retrograde vector
`Adam S. Davis, PhD‡ *
`Thais Federici, PhD§*
`William C. Ray, PhD¶k
`Nicholas M. Boulis, MD§
`Deirdre O’Connor, PhD§
`K. Reed Clark, PhD‡ k#
`Jeffrey S. Bartlett, PhD**
`‡ Gene Therapy Center and ¶Battelle
`Center for Mathematical Medicine, The
`Research Institute at Nationwide Children’s
`Hospital, Nationwide Children’s Hospital,
`Columbus, Ohio; §Department of Neuro-
`surgery, Emory University, Atlanta, Georgia;
`kDepartment of Pediatrics, College of
`Medicine and Public Health and #Depart-
`ment of Molecular Virology, Immunology,
`and Medical Genetics, College of Medicine
`and Public Health, The Ohio State Univer-
`sity Columbus, Ohio; **Calimmune, Inc,
`Tucson, Arizona
`*These authors have contributed equally
`to this article.
`Correspondence:
`Adam S. Davis, PhD,
`Center for Gene Therapy,
`The Research Institute at Nationwide
`Children’s Hospital,
`700 Children’s Drive, W210,
`Columbus, OH 43205.
`E-mail: Adam.
`Davis@NationwideChildrens.org
`Received, March 6, 2014.
`Accepted, September 26, 2014.
`Published Online, December 29, 2014.
`Copyright © 2014 by the
`Congress of Neurological Surgeons.
`WHAT IS THIS BOX?
`A QR Code is a matrix
`barcode readable by QR
`scanners, mobile phones
`with cameras, and
`smartphones. The QR
`Code above links to
`Supplemental Digital
`Content from this
`article.
`ABBREVIATIONS: AAV, adeno-associated virus;
`ALS, amyotrophic lateral sclerosis; DMEM,
`Dulbecco-modified Eagle medium; DRG, dorsal
`root ganglia; DRP ,DNase-resistant particle; MOI,
`multiplicity of infection; PBS, phosphate-buffered
`saline; PCR, polymerase chain reaction; rAAV, re-
`combinant adeno-associated virus
`Supplemental digital content is available for this article.
`Direct URL citations appear in the printed text and are
`provided in the HTML and PDF versions of this article on
`the journal’s Web site (www.neurosurgery-online.com).
`RESEARCH—LABORATORY
`RESEARCH—LABORATORY
`216 |V O L U M E 7 6|N U M B E R 2|F E B R U A R Y 2 0 1 5 www.neurosurgery-online.com
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Sarepta Exhibit 1021, page 1
`
`
`
`
`
`
`
`delivery provides a minimally invasive means to access neurons in
`remote or sensitive structures, as is the case for dorsal root ganglia
`(DRG) in the treatment of neuropathy and pain or spinal cord
`motor neurons in the treatment of motor neuron diseases. Our
`laboratory has focused on both idiopathic and genetic motor
`neuron diseases. However, AAV-mediated retrograde gene delivery
`to motor neurons remains limited by inefficient axon binding and
`poor retrograde axonal transport.
`2,3
`Previously, we isolated Tet1, a 12-mer linear peptide that binds
`selectively to differentiated pheochromocytoma (PC12) cells,
`primary motor neurons, and DRG in vitro.4 We showed that
`Tet1 binds to the tetanus toxin GT1b receptor and described the
`in vivo neuronal binding properties and spinal cord uptake of Tet1
`after peripheral delivery,
`5 suggesting that insertion of this epitope
`into the AAV capsid might allow the generation of modified
`vectors with enhanced axon terminal binding and uptake.
`Although several studies have demonstrated genetic incorporation
`of peptide epitopes into the AAV2 capsid and subsequent
`modification of vector tropism,
`6-9 our group has defined sites
`amenable to peptide insertion in other serotype AAV capsids10 and
`has demonstrated that these peptide-modified vectors can also
`transduce target cells via these engineered interactions10-13 Because
`AAV1 vectors have proven superior to AAV2 vectors for neuronal
`gene delivery and retrograde axonal transport, we sought to modify
`the AAV1 capsid by Tet1 peptide insertion.
`2 Computational
`modeling was used to analyze capsid structure and conformation
`and to visualize and determine the optimal structural context for
`peptide display. This approach was instrumental in our ability to
`produce Tet1-modified AAV1 vectors and for these vectors to
`function properly and direct gene transduction to motor neurons.
`Through this work, we have combined the goals and approaches
`of computational molecular design and capsid structure analysis to
`provide tools for the rational mutagenesis and functional modi-
`fication of AAV vector particles. This approach used the analysis of
`3-dimensional capsid structure to guide the selection of appropri-
`ate amino acid sequences to create a desired property or function.
`The convergence of high-speed computing, a tremendous increase
`in capsid structural information, and a growing understanding of
`the forces that control protein structure and maintain essential
`viral functions has resulted in dramatic advances in our ability to
`engineer protein function and structure and to create novel,
`rationally designed virus-basedg e n et r a n s f e rv e c t o r s .S u c ha n
`enhancement in our ability to engineer vectors for specific
`functions may prove critical to the practical application of gene
`transfer for a variety of therapeutic paradigms in addition to the
`studies described here for sensory and motor neuron gene
`delivery.
`METHODS
`Construction of Modified AAV Helper Plasmids
`Modified AAV1 helper constructs encoding capsid proteins with Tet1
`motif insertions were generated by polymerase chain reaction (PCR)-
`based site-directed mutagenesis as previously described.10 Briefly, DNA
`primers were designed to encode the Tet1 motif and scaffolding
`sequences (Table) and used to direct PCR-based mutagenesis of the
`AAV helper plasmid pXR1.14,15 This plasmid contains the entire AAV
`genome, encoding AAV2 Rep proteins and AAV1 Cap proteins, less the
`2 viral inverted terminal repeats. PCR products were digested withDpnI
`endonuclease to eliminate the parental plasmid template and were
`propagated in DH-5a bacteria (Invitrogen Life Technologies, Grand
`Island, New York). The nonhomologous linker sequences included in
`the PCR primers to encode the Tet1 scaffolding sequences were also
`designed to contain restriction sites. Therefore, mini-prep plasmid DNA
`could be extracted from ampicillin-resistant colonies and screened by
`restriction endonuclease digestion for confirmation of epitope insertion.
`Furthermore, all constructs were also subsequently sequenced to confirm
`epitope insertion and lack of second-site mutations.
`Vectors
`AAV1eGFP or AAV1RFP (dsRed2) vectors were produced by triple
`transfection as previously described. 10 To produce AAV vectors
`comprising Tet1-modified capsid proteins, HEK 293 cells were trans-
`fected with modified AAV helper plasmids, constructed as described
`above, in place of unmodified pXR1. In instances when mosaic particles
`were generated, unmodified pXR1 was included at a 20% molar
`equivalent of Tet1-modified AAV helper plasmid as described
`previously.13 Transfections were carried out at 37/C176C with the use of
`the calcium phosphate transfection system (Invitrogen Life Technolo-
`gies) according to the manufacturer’s specifications. Forty-eight hours
`after transfection, cells were harvested by centrifugation at 500g for
`10 minutes and resuspended in phosphate-buffered saline (PBS), and
`vector was released in 3 freeze-thaw cycles. The crude lysate was clarified
`by centrifugation at 500g 10 minutes, and viscosity was reduced by the
`addition of Benzonase (250 U/mL) and incubation at 37 /C176C for
`30 minutes. Lysate was then fractionated on an iodixanol step gradient
`16
`and further purified by high-performance liquid chromatography as
`described previously.
`17 Final vector preparations were stored at220/C176C
`in PBS containing 20% glycerol. DNase-resistant particle (DRP) values
`were determined by real-time PCR assay.
`10
`Cell Lines
`Low-passage-number (passage number 20-40) HEK 293 cells18 and
`HeLa C12 cells19 were grown in Dulbecco-modified Eagle medium
`(DMEM) supplemented with 10% heat-inactivated fetal bovine serum,
`penicillin (100 U/mL), and streptomycin (100 U/mL) at 37/C176C and 5%
`CO2. PC12 pheochromocytoma cells were grown in DMEM supple-
`mented with 10% horse serum, 5% fetal bovine serum, and penicillin
`(100 U/mL). For differentiation, cells were exposed for 2 to 3 days to
`100 ng/mL of nerve growth factor (nerve growth factor 2.5S, Invitrogen
`TABLE. Summary of Tet1 Insertion Mutantsa
`Vector Designation
`Upstream
`Linker
`Tet1 Peptide
`Epitope
`Downstream
`Linker
`AAV1.D590_P591insTet1a AS HLNILSTLWKYR GLS
`AAV1.D590insTet1b ASDA HLNILSTLWKYR GLS
`AAV1.D590_P591insTet1c ASDA HLNILSTLWKYR ADGLS
`aAAV, adeno-associated virus; AS, alanine-serine; GLS, glycine-leucine-serine.
`AAV VECTOR FOR RETROGRADE DELIVERY
`NEUROSURGERY VOLUME 76 | NUMBER 2 | FEBRUARY 2015 |217
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Sarepta Exhibit 1021, page 2
`
`
`
`
`
`
`
`Life Technologies) in DMEM with 2% horse serum and 1% fetal
`bovine serum. C2C12 cells20 were cultured in Corning 6-well plates
`(Fisher, 07-200-80) with 1 · DMEM (Gibco, 11965-092) supple-
`mented with 10% Hyclone Cosmic Calf Serum (GE, SH30087.04IR)
`and 10 mg/mL ciprofloxacin.
`Primary Cell Cultures
`Spinal cords were obtained under sterile conditions from 15-day-old
`Sprague-Dawley rat embryos following an established protocol.21 DRG
`and perineural membranes were removed, and cords were cut into 2-mm
`sections, which were then trypsinized. Cells were collected, centrifuged,
`pelleted, and then resuspended in complete growth medium made in
`supplemented Neurobasal Medium (Invitrogen Life Technologies).
`Cells were plated on glass coverslips in multiwell culture plates precoated
`with poly-
`L-lysine (Sigma-Aldrich, St. Louis, Missouri).
`Campenot Chambers
`Campenot Teflon chamber dividers (Tyler Research, Edmonton,
`Alberta, Canada) were carefully attached to Collagen/Matrigel-coated
`35-mm culture dishes using silicone vacuum grease (Dow Corning,
`Midland, Michigan). Scratches were made in the Collagen/Matrigel with
`a pin rake. One drop of medium containing methylcellulose (1%) was
`placed onto the plate before the divider was set on the culture dish, which
`facilitated axon growth underneath the silicon grease barriers.
`22 DRG
`explants were plated into one of the compartments of the chambers in
`a small volume of media and allowed to adhere for 2 hours. The
`compartment was then filled with growth medium. The adjacent
`compartments were filled with media supplemented with 100 ng/mL
`nerve growth factor to encourage neurite growth into these compartments.
`Modeling
`Modeling of modified AAV capsid structures was carried out first
`with SWISS-Model (http://swissmodel.expasy.org/)23,24 and later with
`I-TASSER (http://zhanglab.ccmb.med.umich.edu/I-TASSER/).25-27 The
`capsid protein was initially modeled as a monomer and extended to
`a trimer after automated modeling. SWISS- Model experienced difficulty
`in finding adequate loop structures for the Tet1c mutant in the
`automated mode and generally produced results that did not appear
`physiologically relevant even when using the alignment mode with
`a forced alignment to the relevant region of PDB 1LP3. I-TASSER
`successfully predicted several potential structures for each Tet1 structure
`with no constraints on the modeling. The SWISS-Model results and the
`I-TASSER results agreed on the probable structure of the inserted Tet1
`sequence itself, including the location of the turn between the
`antiparallel b sheets. I-TASSER, however, positioned the Tet1 mutant
`insertions in physiologically realistic locations folded against the surface
`of the monomer, whereas SWISS-Model preferred to leave the insert as
`a free strand projecting into space. Although I-TASSER produced several
`potential results for each mutant, the majority of the structural
`differences between I-TASSER models were in the predicted structure
`of the N-terminal region of the protein, quite distant from the insert site.
`The position of the Tet1 mutants was generally as described in Figure 1,
`for which we used the highest-confidence I-TASSER-predicted result.
`The trimeric forms for the structures were constructed by structurally
`aligning the mutants with PDB 1LP3 and using the capsid-creation
`rotation symmetry matrixes from that file. The trimer was constructed of
`capsid subunits 0, 19, and 45. Only the Tet1a mutant required manual
`intervention, inversion of theb-turn curl direction, to avoid physiolog-
`ically impossible topological conflicts. To increase our confidence that
`the folded positioning of the insert sequence was correct, we performed
`brief (40 000 time step, 2 femtoseconds per step) molecular minimi-
`zations of the trimer interface regions using NAMD (Theoretical and
`Computational Biophysics group at the Beckman Institute, University of
`Illinois at Urbana-Champaign).
`28 Although insufficient to predict
`a lowest-energy conformation, our experience has been that these
`parameters are sufficient to detect dramatic spatial clashes caused by
`compositing independently modeled subunits. Each predicted mutant
`structure experienced some side-chain optimization during the minimi-
`zation run, but none displayed symptoms of impossibly close van der
`Walls interactions or other signs of physical implausibility. We therefore
`believe that the modeled locations of the Tet1 mutants are physiolog-
`ically reasonable possibilities for their placement.
`Immunocytochemistry
`Staining was performed with the neuronal cell marker Map-2. The
`protocol consisted of fixing the cell cultures for 40 minutes with 4%
`paraformaldehyde in 0.1 mol/L PBS (pH 7.4) and blocking them with
`0.1% Triton X-100 to 3% bovine serum albumin in PBS for 1 hour,
`before an overnight incubation with the primary antibodies. The next day,
`cells were washed 3 times with 1% bovine serum albumin/PBS and
`incubated for 1 hour with fluorochrome-conjugated secondary antibodies
`at room temperature. After washing, the coverslips were mounted on slides
`using Vectashield Mounting Medium with DAPI (Vector Laboratories,
`Burlingame, California).
`Comparative Gene Transduction Assays
`Titer-matched vector solutions (unmodified and Tet1-modified
`recombinant AAV1 [rAAV1]) were added to cultures and rocked gently
`to ensure equal distribution of the suspension over the cells. Typically, 3
`wells per condition were used. Cells were fixed and processed for analyses
`3 days after treatment. Images of transduced cells were acquired with
`a Nikon E400 microscope using a DS-Qi1 high-sensitivity cooled
`charge-coupled device camera and analyzed by use of the NIS-Elements
`imaging software (Nikon Instruments, Inc, Melville, New York) from 10
`randomly selected fields per slide. The percentage of green fluorescent
`protein (GFP)- or red fluorescent protein-positive cells was determined
`by dividing the number of cell marker-positive, DAPI-positive cells that
`were also GFP- or red fluorescent protein-positive by the total number of
`cell marker-positive, DAPI-positive cells. Graphs of the relationship
`between dose and percentage of transduced cells were then generated.
`Data were expressed as mean6 SEM.
`For quantification of gene transduction in Campenot chambers, the
`NIS-Elements imaging software was used to compare the fluorescence
`pixel intensity within the DRG cell bodies. By drawing a region of interest
`around the explants (white lines), we limited the area that was analyzed.
`Two channels of fluorescence were captured. The first channel was an
`internal control to compensate for variability of thickness, size, or density
`of cells within the explants. It consisted of DAPI staining (seeFigure A,
`Supplemental Digital Content 1, http://links.lww.com/NEU/A693).
`The second channel recorded levels of fluorescence resulting from gene
`expression (see Figure B, Supplemental Digital Content 1, http://
`links.lww.com/NEU/A693), which was then expressed as a ratio of the
`internal control. The software calculated the area of pixel intensity within
`each region of interest (Table), and a graph was generated.
`C2C12 cells at confluence were transduced with either control
`scrAAV1.eCBA.eGFP vector or scrAAV1tet1.eCBA.eGFP vector at
`DAVIS ET AL
`218 |V O L U M E 7 6|N U M B E R 2|F E B R U A R Y 2 0 1 5 www.neurosurgery-online.com
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Sarepta Exhibit 1021, page 3
`
`
`
`
`
`
`
`multiplicity of infection (MOI) of 10 000 and 33 000 in serum-free
`Gibco 1· DMEM. The cells were centrifuged at 1000 rpm for
`30 minutes at room temperature and incubated for 4 hours at 37/C176C
`and 5% CO2.A t4h o u r sa f t e rt r a n s d u c t i o n ,t h ew e l l sw e r es p i k e dt o
`af i n a lc o n c e n t r a t i o no f2 %s e r u ma n di n c u b a t e da t3 7/C176Ca n d5 %
`CO2 for approximately 48 hours. The cells were rinsed with Gibco 1·
`Hanks balanced salt solution and harvested with Hyclone 0.05%
`trypsin containing 25 mmol/L EDTA. The trypsin was neutralized
`with Gibco 1· DMEM containing 10% Hyclone Cosmic Calf Serum
`and 10mg/mL ciprofloxacin. The cells were centrifuged at 1000 rpm for
`20 minutes and resuspended in Teknova TMN200. Cells positive for GFP
`and mean florescence were determined by flow cytometry. Data were
`collected on a Becton Dickenson LSRII cytometer using BD FACSDiva
`software. A total of 30 000 events were collected for each sample, and
`viable cells were gated via a forward-scatter vs side-scatter plot. GFP-
`positive cells were subsequently gated from the viable cell population.
`Voltages were set based on autofluorescence of unlabeled cells.
`Statistical Analysis
`Cell counts were performed in 10 d ifferent fields per slide. All
`experiments were performed in triplicate. The effects of vector
`treatment and MOI were compar ed by use of 2-way analysis of
`variance; however, Tukey tests alsowere performed, and the data are
`presented as 2-way analysis of variance with the P values from the
`Tukey tests (P , .05).
`RESULTS
`Generation of AAV Capsids Containing Tet1
`Peptide Insertions
`We have previously shown that AAV1 can tolerate the insertion
`of exogenous peptides after VP1 amino acid 590.29 Importantly,
`we have also shown that the inserted sequences can be displayed
`on the surface of assembled AAV particles and can promote novel
`capsid-protein interactions.13,29 To use these findings for the
`purpose of generating Tet1-modified AAV particles, we intro-
`duced oligonucleotides encoding the Tet1 motif into the Cap
`ORF of the AAV1 helper plasmid pXR1
`14 by PCR-based site-
`directed mutagenesis, creating pXR1-Cap1.D590_P591insTet1a.
`Short peptide linkers, previously optimized for the display of
`heterologous ligands,
`11,13 were also included in an attempt to
`maintain local capsid flexibility and to promote efficient display of
`the Tet1 epitope on the surface of the assembled AAV vector
`particles (Table). In the first instance, these were made up of an
`alanine-serine upstream linker and a glycine-leucine-serine
`downstream linker flanking the Tet1 epitope (Tet1a designa-
`tion, Table).
`Tet1a-Modified AAV1 Capsid Proteins Inefficiently
`Package Vector Genomes and Fail to Mediate
`Efficient Gene Transfer
`To assess the impact of the Tet1a modification on DNA
`packaging and infectivity, we determined particle and infectious
`titers of AAV1 and Tet1a-modified AAV1, AAV1.D590_P591ins-
`Tet1a, vectors (Table). A real-time PCR assay was used to assess
`DRP titer,
`30 and transduction of HeLa C12 cells was performed
`to assess particle infectivity. Compared with unmodified AAV1
`capsids, the production of AAV1.D590_P591insTet1a capsids
`was 66.5-fold less efficient (5.956 0.41 · 10
`10 DRP per 1 mL
`for AAV1.D590_P591insTet1a vs 3.966 2.2 · 1012 DRP per 1
`mL for AAV1). Similarly, transduction of HeLa C12 cells was
`significantly impaired by roughly 3 orders of magnitude by
`the Tet1 modification (see Figure, Supplemental Digital
`Content 2, http://links.lww.com/NEU/A694).
`FIGURE 1. A molecular modeling comparison of linker sequence effect on Tet1 peptide display and perturbation of native
`structure. The adeno-associated virus (AAV) capsid 3-fold axis of symmetry is depicted as a solvent exclusion surface, with each
`monomer a unique shade of gray, whereas the Tet1 peptide and linker sequences have added van der Walls spheres and are shaded
`red and pink, respectively.A, I-TASSER modeling of Tet1 with C-terminal linker alanine-serine (AS) and n-terminal linker
`glycine-leucine-serine (GLS; Tet1a) suggests that the peptide folds into the interface between and partially beneath the interlocking
`arms of each monomer that creates the 3-fold axis of symmetry of capsid.B, the C-terminal linker ASDA and N-terminal linker
`GLS (Tet1b) slightly change the side-chain polarity and presentation geometry of the peptide so that it is more accessible to the
`capsid exterior; however, it is still partially buried under the capsid surface.C, our modeling predicts that the C-terminal linker
`ASDA and N-terminal linker ADGLS (Tet1c) provide the best opportunity for maximum surface exposure of the Tet1 peptide on
`the exterior of the capsid with the least disruption of the native capsid structure.
`AAV VECTOR FOR RETROGRADE DELIVERY
`NEUROSURGERY VOLUME 76 | NUMBER 2 | FEBRUARY 2015 |219
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Sarepta Exhibit 1021, page 4
`
`
`
`
`
`
`
`Computational Modeling of Tet1-Modified AAV1
`Capsid Proteins
`We hypothesized that conformational stresses imposed on the
`AAV1 capsid by the Tet1 peptide insertion might be responsible
`for the observed decreases in titers. We therefore pursued 2
`complementary approaches to rescue titer of peptide-modified
`AAV1 particles. First, we modeled the structure of the AAV1
`capsid at the 3-fold axis of symmetry to see if sequences flanking
`the Tet1 insert could be altered either to minimize distortion of the
`native capsid structure or to promote more efficient display of the
`targeting epitope. Modeling preformed with both I-TASSER and
`SWISS-Model and validated with molecular dynamics simulations
`in NAMD indicated that the Tet1 peptide favorably forms a pair of
`antiparallel b strands. Postulating that changing the length of the
`linker sequences and the linker side chain polarity might
`predispose this shortb sheet motif toward different positioning
`on the capsid surface, we modeled the structures of 2 additional
`Tet1 peptide modifications with different linker/scaffolding
`sequences (Tet1b and Tet1c; Table). When flanked by alanine
`and serine at the C-terminus and glycine, leucine, and serine at
`the N-terminus, as in the original Tet1a construct, the Tet1
`epitope folded “downward” and packed between and under the
`VP3 monomer arms that create the 3-fold axis of symmetry. This
`not only might effectively shield the epitope from any interaction
`with its targeted receptor but also appeared to distort the native
`capsid structure at the 3-fold axis of symmetry (Figure 1A).
`Modification of the C-terminal linker to include aspartic acid and
`alanine allowed slightly better presentation of Tet1, allowing it to
`stack alongside the monomer arm into which it was inserted;
`however, it was still partially buried (Figure 1B). However, the
`Tet1c modification, which includes additional alanine and
`aspartic acid residues in the N-terminal linker sequence, allowed
`the epitope to lie on top of the arm into which it was inserted,
`resulting in unobstructed display on the capsid surface (Figure
`1C). In fact, modeling of the Tet1c modification predicted the
`presentation of a“looped” version of the Tet1 epitope, with the
`C-terminal and N-terminal residues of the inserted sequences in
`close proximity to each other. Insertions such as this,
`which minimize the induced gap in and therefore
`might minimize distortion of the native structure, although
`facilitating more effective presentation of the targeting epitope,
`could potentially increase both particle titer and the ability of the
`modified particles to interact with their targeted receptors.
`Efficient Production of Mosaic Tet1c-Modified Vectors
`Capable of Mediating Specific Gene Transfer to
`Differentiated Pheochromocytoma (PC12) Cells
`Based on our computational modeling, we generated 2 additional
`packaging constructs, pXR1- Cap1.D590_P591insTet1b and
`pXR1-Cap1.D590_P591insTet1c, for building the modified
`Tet1b and Tet1c sequences into the AAV1 capsid. As suggested
`above and described previously by our group,
`13 these constructs
`were used with AAV helper plasmid pXR1, encoding unmodified
`AAV1 capsid proteins, to generate mosaic AAV particles contain-
`ing different ratios of Tet1-modified and unmodified capsid
`proteins. Particle titers were determined by real-time PCR, and
`gene transduction was assessed on HEK 293 and differentiated
`pheochromocytoma (PC12) cells. Mosaic particles generated from
`a mixture of 80% Tet1c-modified pXR1 plasmid (pXR1-Cap1.
`D590_P591insTet1c) and 20% unmodified pXR1 plasmid into
`HEK 293 packaging cells were produced 4-fold more efficiently
`than particles made up entirely of Tet1c-modified capsid proteins
`(P , .05), and significant increases in particle titer were realized
`with the use of optimized linker/scaffolding sequences flanking
`Tet1 (P , .05; Figure 2). These findings validate the ability of
`particle mosaicism to rescue titer of some tropism-modified AAV
`vector particles, as well as our computational modeling approach to
`rationally engineer modified AAV capsids. However, the real
`benefit of this strategy was evidenced by the ability of Tet1c-
`modified mosaic AAV1 vectors (Tet1c-rAAV) to mediate signif-
`icantly enhanced transduction of PC12 cells and significantly lower
`transduction of HEK 293 cells (Figure 3). Transduction of
`HEK 293 cells at 100 DRPs per cell decreased.55-fold from
`32.32 6 6.98% (mean6 SEM) to only 0.586 1.29% (P , .05),
`whereas at 1000 DRPs per cell, transduction decreased from
`FIGURE 2. Particle titers of Tet1-modified recombinant adeno-associated virus
`1 (rAAV1) vectors. DNase-resistant particle (DRP) titers were determined by
`real-time polymerase chain reaction assay.30 Particles comprising entirely Tet1-
`modified AAV1 capsids proteins are compared with mosaic particles assembled
`from HEK 293 cells transfected with an 80:20 mixture of plasmid DNA en-
`coding Tet1-modified AAV1 capsid proteins and unmodified AAV1 capsid
`proteins. All data are shown as the mean of triplicate determinations. Bars
`indicate the standard error of the mean. A 2-way analysis of variance test revealed
`that a significant effect depended on both capsid mosaicism and linker/scaffolding
`sequences flanking the Tet1 epitope (*P , .05).
`DAVIS ET AL
`220 |V O L U M E 7 6|N U M B E R 2|F E B R U A R Y 2 0 1 5 www.neurosurgery-online.com
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
`Sarepta Exhibit 1021, page 5
`
`
`
`
`
`
`
`57.42 6 6.84% to only 9.396 4.03% (P , .05; Figure 3A).
`Transduction of PC12 cells at 100 DRPs per cell increased from
`6.5 6 2.32% to 18.136 5.51% (P , .05) and at 1000 DRPs per
`cell from 14.216 3.02% to 25.726 10.92% (P , .05; Figure
`3B). By simultaneously increasing the transduction of PC12 cells
`while decreasing transduction of nonneuronal HEK 293 cells,
`a .155-fold (P , .05) increase in specificity is realized at the lower
`MOI, suggesting the utility of t hese engineered particles.
`Because axonal uptake and retrograde delivery depend equally
`on axon terminal affinity (neuronal binding) and available titer
`(lack of binding to surrounding muscle), increased neuronal
`specificity is likely more important than affinity alone.
`Although the technical ease of using immortalized cell lines
`led us to initially screen Tet1-modified vectors on PC12 and
`HEK 293 cells, having demonstrated increased specificity, we
`moved on to assess gene transduction in primary spinal cord
`cultures.
`Tet1c-Modified Vectors Mediate Enhanced Gene
`Transfer to Primary Spinal Neurons
`Embryonic day 15 spinal cord cultures (neuron/glia ratio, 8:2)
`were used to assess gene transduction mediated by Tet1c-modified
`AAV particles. To evaluate the specificity of Tet1c-rAAV-
`mediated gene transfer, cells were stained with antibodies against
`neuron-specific markers (MAP-2). Figure 4A illustrates an
`apparent increase in neuronal GFP expression in the Tet1c-
`rAAV treated cultures. At an MOI of 1000 Tet1c-rAAV1eGFP
`DRPs per cell, transgene expression was observed in 61.096
`13.23% of spinal cord neurons, whereas only 17.486 7.54% of
`spinal cord neurons in these cultures were transduced with
`unmodified rAAV1eGFP (P , .05; Figure 4B).
`Tet1c-Modified Vectors Mediate Enhanced Retrograde
`Axonal Transport to Primary DRG Neurons
`It has previously been demonstrated that wild-type AAV
`undergoes uptake and retrograde transport with limited
`efficiency.
`31 We hypothesized that modifying AAV using
`peptides with high affinity for motor neuron receptors might
`enhance axon terminal uptake. We evaluated retrograde gene
`delivery using the Tet1c-modified AAV1 vector in vitro. Axon
`terminals of DRG explants plated in Campenot chambers were
`exposed to Tet1c- and unmodified rAAV1 vectors. The
`percentage of transduced cell bodies in the DRG explants was
`counted to estimate the tendency of individual vectors to bind
`axon terminals, to undergo uptake, and to be transported in
`a retrograde fashion to the cell bodies. We were able to
`demonstrate that the uptake of Tet1c-rAAV1