Structural and Functional Studies of Adeno-Associated Virus

Material Information

Structural and Functional Studies of Adeno-Associated Virus
Copyright Date:


Subjects / Keywords:
Amino acids ( jstor )
Capsid ( jstor )
Capsid proteins ( jstor )
Crystal structure ( jstor )
Crystals ( jstor )
Dependovirus ( jstor )
DNA ( jstor )
Genomes ( jstor )
Heparin ( jstor )
Receptors ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright the author. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:


This item is only available as the following downloads:

Full Text




Copyright 2005 by Hazel Christina Levy


iii TABLE OF CONTENTS page LIST OF FI ABSTRACT ........................................................................................................viii 1 BACKGROUND AND INTRODUCTION........................................................1 Adeno-Associated Virus Biolog y and Serotype Diversity...............................1 AAV Genome.................................................................................................2 AAV Life C ycle...............................................................................................4 Helper Virus Function.....................................................................................6 AAV Capsid St ructure....................................................................................8 Viral Rece ptors...............................................................................................9 AAV Serotype Antigeni c Evolut ion...............................................................14 Phylogentic Analysis of AAV DNA and Capsid Protein................................15 AAV Capsid Assembly and Genome Pa ckaging ..........................................17 The Location of C apsid Asse mbly................................................................17 Previous AAV Genome Packaging Studies ..................................................18 Genome-Mediated AAV Capsid Assembly Hypothes is................................20 Nucleic Acid Structure in Icosahedral Viruses..............................................21 Determination of Macromol ecular Stru ctures...............................................24 Structure Determination by X-ray Crysta llogra phy.................................25 Structure Determination by Cr yo-Electron Mi croscopy..........................28 Virus:Ligand Structures of Icosahedral Viruses...........................................33 2 MATERIAL A ND METHOD S........................................................................38 AAV Capsid Protein Ex pression Ve ctors......................................................38 Capsid Protein Expressi on...........................................................................40 Purification of Adeno-Associat ed Virus-Like Particle s..................................41 Western Blot Analysis for the De tection of Caps id Protein...........................43 Negative Stain Elec tron Micro scopy.............................................................43 SDS-PAGE and Silver Stain Anal ysis..........................................................44 Extraction of Nucleic Acid from Purified Virus-Like Particles........................44 Complexing of AAV2~Heparin Oligosacc haride...........................................45 Cryo-Electron Microsc opy............................................................................45 Three-Dimensional Rec onstruction of AAV..................................................46 Difference Map Analys is..............................................................................47 Pseudo-Atomic Model Construc tion.............................................................48


iv Generation of a Surface Map Bas ed on the AAV2 Atom ic Model................48 VLP Crystalliz ation.......................................................................................49 AAV1 Crystallizat ion Scr eens................................................................49 AAV2~Heparin Co-cryst allization Screens.............................................49 AAV5 Crystallizat ion Scr eens................................................................50 AAV5 X-ray Data Coll ection and R eduction...........................................50 3 CRYO-ELECTRON MICROSCOPY, THREE-DIMENSIONAL RECONSTRUCTION AND CRYSTALLIZA TION OF AAV SEROTYPE 1....51 Introducti on..................................................................................................51 AAV1 as a Gene T herapy Ve ctor..........................................................51 Capsid Surface Loops Associated with Tropism and Antigenicity.........52 Result s.........................................................................................................57 Construction of Expression Vector for AAV1 Capsid Protein.................57 Crystallization of AAV1 VL Ps.................................................................59 Cryo-EM Structure of AAV1 VLPs.........................................................59 Antibodies to Detect In tact AAV1 C apsids.............................................63 Discussio n....................................................................................................63 Correlations between Structur e Differences and Tropism.....................63 Correlations between Structure Di fferences and An tigenicity................67 4 COMPLEX OF AAV2 AND HEPA RIN OLIGOSAC CHARIDE......................69 Introducti on..................................................................................................69 HSPG is the Only Known Host-Cell Receptor Required for AAV2 Cell Entry...................................................................................................69 Mapped Heparin Bind ing Regi on...........................................................72 Mapped Epitopes for Neutralizin g Antibodies A2 0 and D37..................72 Structure of t he AAV2 C apsid................................................................74 Genome-Dependent Icosahedral Caps id Assembly Hypothesis...........75 Result s.........................................................................................................76 Density Maps.........................................................................................76 Modeling Heparin into Density at the Heparin Bi nding Region..............79 Modeling Fivefold HI Loop c hanges.......................................................79 Ordered Nucleic Acid Insi de the Capsid Shell.......................................82 Discussio n....................................................................................................85 Acidic and Basic Residues of the Heparin Bi nding Regi on....................85 Expressed AAV2 Capsids Nuclea se protect N on-Viral DNA.................92 Internal DNA Density and an Assembly and Packaging Hypothesis.....95 Genome Uncoati ng Hypothes is.............................................................99 Genome Pack aging.............................................................................100 Antigenicity and Hepar in Bindin g.........................................................100


v 5 PRODUCTION, PURIFICAT ION, CRYSTALLIZATION AND PRELIMINARY X-RAY STRUCTURAL STUDIES OF ADENOASSOCIATED VIRUS SEROTYPE 5.........................................................102 Introducti on................................................................................................102 Evolutionary Divergent Serotype AAV5...............................................102 AAV5 as a Gene Ther apy Vect or........................................................103 AAV5 Host-Cell re ceptors ....................................................................103 Structural Studi es of AAV.................................................................... 104 Results and Di scussion..............................................................................105 Crystallization Conditio ns for AAV5 Virus-Li ke Particles......................105 X-Ray Data Collection and Processing Re sults...................................107 6 SUMMARY AND FUTURE DIRECTIONS..................................................111 AAV Capsid Stru ctures..............................................................................111 AAV1 Cryo-EM Map............................................................................111 AAV5 X-ray Crystal Structur e..............................................................112 AAV Capsid Surface Evolutio n...................................................................112 The Heparin Densit y and Mutagenes is......................................................113 AAV2~Heparin Conformati onal Differ ences.........................................114 Changes at the capsid interior and an assembly hypothesis...............115 Future Directions to St udy AAV2 Asse mbly...............................................116 Future GH Loop Mutagenesis ....................................................................116 Charged Residue Mu tagenesis ..................................................................117 LIST OF REFE RENCES..................................................................................118 BIOGRAPHICAL SKETCH ...............................................................................131


vi LIST OF FIGURES Figure page 1-1 The genetic m ap of A AV..............................................................................3 1-2 The latent and lytic pathwa ys of the AAV li fe cycle......................................5 1-3 Models of AAV2 capsid................................................................................7 1-4 Alignment of parvoviru s capsid prot eins. ...................................................10 1-5 Neutralizing antibody epitopes for AAV2....................................................16 1-6 Ribbon representation of nucleoti des modeled into the X-ray crystal structure of C PV and MVMi ........................................................................22 1-7 Atomic model of MVMi amino acids within 5Ã… of the ordered nucleotides.23 1-8 Negative stain electron micr ograph of AAV capsid proteins.......................30 1-9 Atomic model of fibroblast growth factor (FGF) bound to an oligosaccharide of heparin sulf ate..............................................................36 1-10 Atomic model of Foot and Mouth Disease Virus bound to an oligosaccharide of heparin sulf ate..............................................................37 3-1 Sequence alignment of the AAV1 and AAV2 capsid protein amino acid sequence. ..................................................................................................54 3-2 Construction of adenovirus expre ssing AAV1 capsid protein.....................58 3-3 The pseudo-atomic model of AA V1...........................................................60 3-4 Cryo-EM reconstructi ons for AAV c apsids. ................................................62 3-5 Detail of threefold view of cr yo-EM reconstructi ons of AAVs.....................64 4-1 Basic residues surrounding AAV2 heparin binding region mapped by mutagenesis ...............................................................................................73 4-2 Three-dimensional reconstruc tions of adenovirus expressed AAV2 particles and the AAV2~ heparin comp lex..................................................77


vii 4-3 Equatorial sections of the reconstruc tions.................................................78 4-4 Cross section of the differenc e map superimposed onto the atomic model of the AAV2 capsid..........................................................................80 4-5 Atomic model describi ng AAV2~heparin co mplex.....................................81 4-7 Difference density associated with the interior face of the parvovirus capsid sh ell................................................................................................84 4-8 Nucleic acid contained in the viruslike protein capsid shells visualized on an agarose gel with ethi dium brom ide....................................................... 86 4-9 Atomic model of charged AAV2 resi dues in the heparin binding cleft. The atomic model is from the crystal structur e of AAV2 ............................88 4-10 Atomic model of an AAV2 VP3 pentam er..................................................91 4-11 Capsid protein residues with the shorte st radial distance from the internal DNA cage. ..................................................................................................94 4-12 DNA-facilitated pentam eric VP asse mbly...................................................98 5-1 Isolation and characterization of pur ified AAV5 empty viral capsids........106 5-2 X-ray diffraction image for a cr ystal of AAV5 empty (no DNA) viral capsids. ....................................................................................................108


viii Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy STRUCTURAL AND FUNCTIONAL STUDIES OF ADENO-ASSOCIATED VIRUS By Hazel Christina Levy December 2005 Chair: Mavis Agbandje-McKenna Cochair: Nicholas Muzyczka Major Department: Biochemistry and Molecular Biology Adeno-associated virus (AAV) holds signifi cant promise for the correction of human diseases. AAV can be used to transfer genes efficiently into primary cells in vivo , and in most cases, expression of the transgene appears to be long lived. There are several AAV serotypes, differing in their capsid protein amino acids sequences, ability to transduce different cell types and their affinity for binding cellular receptors. The cellular tropi sm of the different AAV serotypes is determined by the ability of the virus capsid to bind host cell receptors with different terminal carbohydrates; they m ediate cellular uptake of the virus via endocytosis. Receptor binding is thought to utilize regions on the capsid protein surface of the viruses that differ in amino acid sequence between the serotypes. This work has focused on AAV1, AAV2 and AAV5, which show broad, although


ix somewhat different, tissue tropism. We obtained stru ctural information about both AAV virus capsids alone, and capsids in complexes with the carbohydrate component of their receptors by cryo-e lectron microscopy (cryo-EM) and X-ray crystallography, in order to better understand the specific virus:host interactions that govern tropism. Cryo-EM structures of AAV1, AAV2, AAV2~heparin complex, and atomic models for AAV1 and the AAV2 heparin complex were determined and are analyzed in this work. Crystallizat ion conditions for AAV1, AAV2~heparin complex and AAV5 were also determined for the first time in this study. These data have been interpreted with respect to receptor binding, phenotypes, capsid protein assembly, capsid:DNA interactions, packaging and AAV capsid evolution, and allow an improved understanding of the st ructure and function of AAV capsid proteins for the development of AAV as a viral gene therapy vector.


1 CHAPTER 1 BACKGROUND AND INTRODUCTION Adeno-Associated Virus Biol ogy and Serotype Diversity Viral gene delivery and protein expressi on is the result of a cascade of events involving chemical interaction bet ween the virus and the host cell. In order to gain insight into the mechanisms that dictate the virus life cycle, it is important to have an understand ing of the structure an d function of the virus capsid shell and genome. Adeno-associated virus (AAV) is the prototypical helper dependent parvovirus (Muzyczka and Be rns, 2001) requiring coinfection with a helper virus for productive in fection. AAV is a nonpathogenic human parvovirus that was originally di scovered as an adenovirus contaminant. Currently, a number of AAV serotypes , including AAV1-11, are being studied for their use in gene therapy; thes e serotypes differ in their antigenic properties, host-cell tropism, and receptor bi nding characteristics (Atchison et al., 1965; Hoggan and Blacklow,1966; Melnick et al., 1965; Mayor and Melnick, 1966; Bantel-Shaal and Hauson, 1984; B antel-Schaal and zur Hausen, 1984; Georg-Fries et al., 1984; Xiao et al.,1999; Rutledge et el., 1998; Xiao et al., 1998; Gao et al., 2002, 2004; Mori et al., 2004). Most of the gene therapy applications to date have been with AAV serotype 2 (AAV2), which is currently the subject of clinical trials to correct diseases includ ing hemophilia and cystic fibrosis (Flotte et al., 2002; Kay et al., 2000).


2 It is now clear that for a variety of transduction purposes, serotypes other than AAV2 may be more useful. AAV1, for example, can transduce rodent skeletal muscle as much as a thousand fold more efficiently than AAV2 (Chao et al., 2000). AAV5 has been shown to have a better spread and to be more efficient in transducing both neuronal tissues and lung tissues compared to AAV2 (Davidson et al., 2000). Mo re recently, as many as 40 additional non-human primate and over 50 human serotypes hav e been identified and some of these appear to have unique tropisms (Gao et al., 2002, 2004). Additionally, several groups, have developed AAV2 mutants that c an accept ligand insertions into the capsid gene, which change the tropism of the virus and target transduction more efficiently to cell types that were previous ly intractable (Girod et al., 1999; Shi et al., 2001; Wu et al., 2000). AAV Genome Like the related autonomous parvoviru ses, AAV capsids are icosahedral, approximately 26 nm in diameter, and have a molecular weight of approximately 4 X 106 Daltons. They package a 4.7 kb si ngle stranded DNA genome (fig. 1-1A) containing two major open reading frames (ORF), rep and cap , three promoters p5, p19 and p40, and inverted terminal repeats. The AAV genome can be transcribed into six mRNA transcripts which are 3Â’-capped and 5Â’-polyadenylated (fig. 1-1B). The rep open reading frame codes for four proteins, Rep40, Rep52, Rep68, and Rep78. These proteins are involved in viral genome replication and packaging. Rep52, Rep68 and Rep78 have ATP-dependent helicase activity in the 3Â’-5Â’ direction (McLaughlin et al., 1988; Qiu and Brown, 1999; Samulski et al.,


3 Figure 1-1. The genetic map of AAV. A) The genetic map includes the rep and cap open reading frames as large red arrows, the inverted terminal repeats as large black arrows and the locations of the three promotors represented as thin, red bent arrows. B) The six mRNA transcripts are 3Â’-capped and 5Â’-polyadenylated and code for four replication proteins and three capsid proteins. The capsid proteins are translated from two alternately spliced mRNAs. One of these co ntains the entire capsid orf and makes VP1. The other synthesizes VP2 from an alternate upstream start codon, AC G (indicated by the dotted box), and VP3 from a conventional downstr eam ATG. C) Density map of the AAV2 capsid based on the atomic model surface density (Xie et al., 2002). Sixty copies of a combination of VP1, VP2 and VP3 capsid proteins assemble into an icosahedr on with 2, 3, and 5-fold rotational axes of symmetry.


4 1991), and Rep 68 and 78 have an N-terminal DNA binding domain allowing the proteins to bind the 5’ -terminal repeat s of the genome. The cap gene codes for three similar proteins VP1, VP2 and VP3 . The capsid proteins are in an approximate ratio of 1:1:8 (VP1:VP2:VP3) in the 60-subunit viral particle. The the 61 kDa VP3 protein constitutes 90% of the structure. The minor capsid proteins, VP1 (87 kDa) and VP2 (73 kDa) , share the same 532 C-terminal amino acids with VP3, but VP1 and VP2 have additional N-terminal sequences (Muzyczka and Berns, 2003). AAV Life Cycle The general life cycle of AAV is outlined in figure 1-2. There is a lytic life cycle with helper virus co-infection and a lat ent life cycle. The virus must first tether to the cell surface to allo w for receptor–mediated endocytosis (Summerford and Samulski, 1998). The virus moves through the cytoplasm toward the nucleus via endosomal trafficking. In the latent infection, in the absence of a helper virus, the viral genome is inserted into chromosome 19, or persists as a episome, until subsequent helper virus infection allow the genome to be excised from the cellular DNA for vi rus DNA replication (Chamberlin et al., 1998; Grimm et al., 1998; Im and Muzyczka, 1992). During a productive infection, where progeny virus parti cles are synthesized, the genome is replicated and transcribed into mRNA inside the nucleus. In the cytoplasm the virus mRNA is translated into viral protei ns including the three capsid proteins previously described and the non-structural rep proteins including Rep40, Rep52, Rep68, Rep78 required for genome replic ation and packaging. The virus


5 Figure 1-2. The latent and lytic pathways of the AAV life cycle. The virus gains entry into the host cell by first bindi ng to a host cell receptor(s) to allow for receptor mediated endocytosis. In the cytoplasm the virus moves toward the nucleus via endosomal tr afficking and the virus escapes the late endosome prior to nuclear entry. Inside the nucleus the latent pathway proceeds, in the absence of hel per virus, by insertion of the viral genome into chromosome 19. Upon subsequent helper virus infection the genome can be resc ued by excision from the host genome. Co-infection with helper virus results in a lytic infection which allows AAV to bypass the AAV genome integration into the host cell genome. The viral genome is replic ated and transcribed into mRNA, which are translated in the cytoplas m into six virus structural and replication proteins. The virus proteins are impor ted back to the nucleus. The capsid proteins assemble and the genome is packaged for the production of progeny virions.


6 proteins are imported back into the nucleus where t he T=1 icosahedral capsid (fig. 1-3A) is packaged with the newly synthesized virus genome with the help of the Rep proteins (Che janovsky and Carter, 1989). The only essential cis-active sequences in AAV are the 145 bp terminal repeats which function as origins for DNA replication, packaging sequences and integration sites (McLaughlin, S. K. et al., 1988). Recombinant AAV vectors (rAAV) are generated by retain ing the terminal repeats and replacing the internal wtAAV coding sequences with therapeutic genes. Although the natural route of infection for wtAAV2 is believ ed to be via the upper respiratory or gastrointestinal routes, rAAV is capable of achieving long term transduction in a variety of tissues, including heart, brain, liver, muscle, retina, and vasculature of experimental animals (Samul ski et al., 1998, Kern et al., 2002). In contrast to wtAAV, rAAV has often been found to persist in experimental animals as either a circular or linear episome (Afione, et al., 1996; Duan et al., 1998; Nakai et al., 1999). Helper Virus Function For the production of progeny AAV virus, a helper virus is required. The helper function of adenovirus has been t he most extensively analyzed but pseudorabies virus, herpes simplex I and II, and cytomegalovirus can all provide helper function for AAV (Atchison et al., 1965; Buller et al., 1981; Muzyczka, 1992). Deletions in several adenovirus early genes have been shown to inhibit complete helper function including E1a, E1b, E2a, E4ORF6, and virus associated (VAI) RNA (Buller and Rose, 1978; Carter et al., 1992; Handa et al., 2000; Quinn


7 Figure 1-3. Models of AAV2 capsid. T he axes of symmetry are labeled with numbers (2, 3, 5). A) The white tr iangle with the axes of symmetry marked represents the assymetric unit. The T=1 capsid is made from a combination of three capsid prot eins, VP1, VP2 and VP3, for a total of 60 coat proteins. Each grey equi lateral triangle represents a capsid protein trimer (adapted from Baker et al., 1999) B) Ribbon diagram representing the secondary structural elements of a trim er of the AAV2 VP3 protein based on the crystal stru cture of AAV2 (Xie et al., 2002). A reference monomer is in white with the positions of the acidic amino acid residues in red and the basic residues in blue, with the other two monomers in yellow. C) Close-up of the ribbon representation of the atomic model of a VP3 monomer. Axes of symmetry are labeled. Basic residues are in blue, acidic re sidues are in red, ß-strands are in broad arrows and helices are coils. D) Space filled surface rendering of the atomic model of VP3. Basic residues are in blue and acidic residues are in red. The mapped heparin binding region residues R585 and R588 are in blue and are label ed. Images of atomic models generated in Chimera (Pettersen et al., 2004).


8 and Kitchingman, 1986; Laughlin et al., 1982; Richardson and Westpul, 1984). Protein E1A is a transactivating protein that induces cells into S-phase. The E4ORF6, VAI RNA and E1B and E2A DNA bi nding proteins are all required for the accumulation of AAV mRNA. E1B and E4 may regulate AAV gene expression by transporting the AAV m RNA to the cytoplasm (Samulski and Shenk, 1988). AAV Capsid Structure The AAV capsid is responsible for interacting with cell surface molecules for endosomal uptake, protecting the genome from degradation along the pathway toward the nucleus and interacti ng with host antibodies during immune surveillance. The N-terminus of VP1 has been shown to have phospholipase activity (Girod et al., 2002) . Mutations in this region that block phospholipase active site greatly reduce the infectiv ity of the capsids causing perinuclear accumulate of the virions. This suggests that the phospholipase activity of the VP1 N-terminus is required for productive trafficking at a stage in the life cycle between endosomal trafficking and nuclear entry. The three-dimensional structures of severa l autonomous parvoviruses (Agbandje et al., 1991; Tsao et al., 1991; Ag bandje et al., 1993; Xie et al., 1996; Agbandje-McKenna, 1998; McKenna et al., 1999; Simpson et al., 200X; Kontou et al., 2005), and that of AAV2, AAV4, AAV5 and AAV8 (Xie et al., 2002; Krononberg et al., 2000; Kaludov et al., 2003; Walters et al., 2004; Lane et al., 2005; DiMattia et al., 2005), have been det ermined by X-ray crystallography and cryo-electron microscopy (cryo-EM). The N-terminal extensions of VP1 and 2 are not observed in the structures. The capsid proteins for all of the known


9 parvoviruses structures have a core eight-stranded antipara llel ß-barrel that forms the contiguous inner shell surface, with variable loop insertions between the strands forming the outer surface (fig . 1-4A and 4B). The ß-barrel strands are denoted as letters from A to I, from the N-terminus to the C-terminus, and the inter-strand loops are named with the two letters of the flanking beta strands. The major surface features include depressions at the icosahedral twofold symmetry axes and surrounding the fivefo ld axes, and protrusions at or surrounding the threefold axes. The threefol d protrusions are finger-like spikes in AAV2 (fig. 1-1C), but are more rounded mo unds in AAV4 and AAV5. Structural and mutational analysis clearly shows that parvoviral host tropism and antigenic differences arise from variations in surface amino acids (Agbandje et al., 1995). Sequence alignment of AAV1-8 reveals that there is a high degree of identity and conservation in the regions of the AAV c apsid that constitiute the core ß-barrel structure (fig. 1-4). Evolutionary changes are concentrated at the surface variable loop regions. Viral Receptors Heparan sulfate proteoglycan, the first receptor identified for an AAV virus (Summerford and Samulski, 1998), appears to f unction primarily in attachment of the AAV2 and 3 serotypes to the cell surfac e. Efficient AAV2 infection may also require a co-receptor, such as human fibrobl ast growth factor re ceptor 1 (Qing et al., 1999) or integrin v5 receptor (Sanlioglu et al ., 2000; Summerford et al., (1999).


Figure 1-4. Alignment of parvovirus caps id proteins. A) S equence alignment for the AAVs. This alignment is based on the structure of the viruses and the numbering is based on AAV2 al ong the top and AAV4 along the bottom. Regions of divergence between serotypes are labeled as roman numerals and the core ß-barrel motif is marked as black arrows. Regions of Identity are in r ed block with white lettering and conservative changes are in white block with red lettering. B) Alignment of parvovirus struct ure. Superimposition of coil representations of the VP3 monomers of parvoviruses including, AAV2 (atomic coordinates) (in red), t he pseudoatomic models of AAV4 (in blue) (Padron et al., 2005) and AAV5 (in dark green), (Walters et al., 2004), the VP2/VP3 monomers of the atomic coordinates of B19 (in pink) (Kaufmann et al., 2004), C PV (in cyan) (Xie and Chapman, 1996), FPV (in magenta) (Simpson et al., 2000), MVM (in green) (Agbandje-McKenna et al., 1998), and PPV (in brown) (Simpson et al., 2002), and the VP2 pseudoatomic coor dinates of ADV (in orange) (McKenna et al., 1999). Variable surface loop regions are again labeled I to IX. The N and C termini of the VPs are indicated. The approximate icosahedral two-, three-, and fivefold axes are indicated by the filled oval, triangle, and pentagon, respectively. ß-strands B to I are also labeled. Figure B) was g enerated with the program Bobscript (Esnouf, 1997). This figure is modi fied from Padron et al. (2005). 1999).


11 A. 220230240250260270280 290300310320330340350360370 380390 400 410 420 430 440 450 460 470 480 490 500 510 520 530 540 550 560 570 580 590 600 610 620 630 640 650 660 670 680 690 700 710 720 730 I II III V IVa IVb VIVII VIII IXJVI01692-04-Fig3.* * * * * B C D E F G H I B.


12 AAV2 then enters through receptor-mediat ed endocytosis through clathrin-coated pits; this event requires dynamin, a 100-kD a cytosolic GTPase, that selectively regulates clathrin-medi ated endocytosis (Bartlett et al., 2000). AAV1, which is ~83% identical to AAV2, does not bind to heparin sulfate as efficiently as AAV2 (Rabinowitz et al., 2002) and appears to utilize sialic acid to achieve efficient transduction in vascula r endothelial cells (Chen et al., 2005). Studies with AAV4 and AAV5 indicated that they have hemagglutination activity and require sialic acid for binding and trans duction but differ in their specificities (Kalidov et al., 2001). AAV4 requires 2,3-O-linked sialic acid and AAV5 requires 2,3-N-linked sialic acid for trans duction. In addition, the plateletderived growth factor receptors (PDGFR a and PDGFRb) have been identified as protein receptors for AAV5, and their expression correlates with transduction in vivo (Di Pasquale et al., 2003). The nature of the carbohydrate molecules utilized for cell binding and transduction by AAV7 -11 remains to be elucidated. The amino acids in AAV2 that are r equired for binding heparin sulfate were identified by mutagenesis to be argi nine pairs R484-R4 87 and R585-R588 and residue K532 (Opie et al., 2003; Kern, 2003). These residues line a channel that runs between the threefold spikes in capsid protein regions that varies somewhat between serotypes (Padron et al., 2005). This information will have a direct impact on studies designed to alter the tropism of AAV2. As yet, similar information is not available for the other AAV serotypes. A number of autonomously replicating par voviruses, including Aleutian mink disease parvovirus (ADV), canine parvo virus (CPV), minute virus of mice (MVM),


13 and bovine parvovirus have been shown to bind sialic acid, although, in the case of CPV, this interaction is not required for host cell infection. The capsid amino acids required for CPV sialic acid binding have been mapped to the wall of the dimple depression at the icosahedral twofold symmetry axis on its threedimensional structur e (Barbis et al., 1992; Cotmor e and Tattersall, 1987; Fox and Bloom, 1999; Thacker and Johnson, 1998; Tresnan et al., 1995). Structure studies of MVMp complexed with sialic acid also implicate the dimples as the receptor binding sites (Lopez-Bueno et al., 2003). Neither amino acid requirements nor capsid regions have been identified for the interactions of ADV and BPV with sialic acid. The capsid surface regions that affect host-cell receptor binding have been determined in autonomous parvoviruses C PV and B19. The ability of CPV to bind the transferrin receptor is affected by mutagenesis of capsid protein VP2 residues 93, 300, and 323 (Govindasamy et al., 2003). As determined by crystallographic studies, these three resi dues are separated from each other by 20 to 37 Ã… on two different monomers, and t herefore the receptor molecule must contact a large region of the CPV capsid. Cryo-EM and 3D reconstruction was used to determine that the glycolipid-rec eptor globoside molecules bind B19 in the depressions at the threefold axis of the capsid (Chipman et al., 1996). Binding of globoside at the threefold axis appears to translate through the capsid, resulting in a conformational change at the fivefold axis in B19. This suggests that dynamic structural changes occur in the capsid during the virus life-cycle.


14 The implication of the parvovirus twofol d interfaces in carbohydrate binding is in contrast to structural and mutational studies that identify the threefold axes of AAV2 as the most likely region of t he capsid utilized for interactions with heparin (Xie et al., 2002; Opie et al., 2003; Kern et al., 2003). In addition, cryoEM studies of the human parvovirus B 19 bound to the host-cell receptor glycolipid globoside, identified the icos ahedral threefold axes as the sugar binding site (Chipman et al., 1996). T hese observations clearly show that parvoviruses are capable of utilizing diffe rent surface regions for receptor recognition. AAV Serotype Antigenic Evolution AAV serotypes are distinguished by their unique antigenic properties. Studies by Wobus et al. (2000) have shown that monoclonal antibody A20 generated against AAV2, neutralizes infect ion of AAV2 and AAV3, but not AAV1, AAV4 or AAV5, and inhibits at a st age subsequent to host-cell receptor attachment, while antibody C37 neutraliz ed AAV2 infection by inhibiting cell attachment. Antibody C37 is not able to bind or neutralize infection of AAV1, AAV4 and AAV5 which do not bind heparin, and does not neutralize infection of AAV3 which does bind heparin. This suggests that AAV3 used a different capsid region than AAV2 for heparin binding. Antibody B1 is used to detect AAV1, AAV2, AAV3, and AAV5 by western analysis, but does not detect capsid protein by ELISA (Wobus et al., 2000). The epi tope was mapped to the 25 amino acids at the C-terminus of the capsid protein. The conformational epitope mapped for A 20 (Wobus et al., 2001) does not overlap the regions found to be involved in heparin binding (fig. 1-5), although


15 one of the loop regions can be shown by 3D modeling to be below the threefold spikes on a surface loop implicated in heparin binding. The conformational epitope for C37 clearly inhibits attachment of AAV2 to the cell by blocking the heparin binding region. Inte restingly, the D3 antibody binds all AAV serotypes tested except AAV4, a structurally distinct serotype (Padron et al., 2005). The B1 antibody epitope was also mapped in the W obus et al. (2001) study. B1, binds a linear epitope detecting denatured AAV1, AAV2, AAV3 and AAV5, but not AAV4. Phylogentic Analysis of AAV DNA and Capsid Protein Phylogentic analysis of the AAV1-5 full length genomes placed AAV1, 2 and 3 in a common evolutionary branch with AAV4 and AAV5 in their own divergent branches. Analysis of the rep ORF places AAV5 as the outlying serotype, while analysis of cap ORF places both AAV4 and AAV5 as outlying serotypes (Lukashov and Goudsmit, 2001). A phylogenetic study of specific gene regions of AAV1-8 demonstr ated that there was a hi gh degree of variability in the capsid protein gene and that this diversity arises due to homologous recombination between different serotypes in fecting the same cell. This type of molecular evolution had previously been t hought to only exist in RNA viruses (Gao et al., 2003). In a more recent study by Gao et al. (2004), newly characterized serotypes AAV7, AAV8 and AAV9 were included in the phylogenetic analysis of the primary sequence of VP1, an d again determined that AAV4 and AAV5 were


16 Figure 1-5. Neutralizing antibody epitopes for AAV2 (modified from Wobus et al., 2000). A) Peptides used for peptide scanning analysis for conformational antibodies C37 abd A20. The residue numbers for peptides are shown and the antibody and epitopes are indicated. Trimer of AAV2 capsid pr otein with mapped epitopes for conformational antibodies B) A20 and C) C37 using the atomic model based on the crystal struct ure of AAV2 VP3 (Xie et al., 2002). B) The red space-filled atoms ar e the residues of peptides that inhibited the detection of AAV2 by ELISA A20-2 and A2 0-3. Insertions (I) at black space-filled residues did not i nhibit A20 neutralization of AAV2 infection. C) The white space-f illed atoms are the location of the peptide that most strongly inhibited the ability of C37 to detect AAV2 by ELISA. Insertions at cyan residu es did not inhibit neutralization of AAV2 infection by C37. This figure was generated in Chimera (Petterson et al., 2004).


17 the only serotypes in their own distinct clades. They further showed that the human and non-human primate serotypes could be grouped into a common clade. They report that the phylogenetic grouping of the serotypes suggests a cross-species virus transmission, but th is might also suggest that AAV has been co-evolving with humans since before humans evolutionary diverged from nonhuman primates. AAV Capsid Assembly and Genome Packaging AAV viral capsids proteins assemble into icosahedral shells even when the AAV genome is not present. For ex ample, when using Adenovirus and Baculovirus expression vect or systems to express t he capsid protein the only AAV DNA present in the system is the c apsid gene. Fully assembled capsids can be produced by these methods. Th is supports the hypothesis that AAV capsid assembly is AAV genome independent and that Rep-mediated packaging occurs into a pre-assembled capsid (Dubielzig et al., 1999). The Location of Capsid Assembly The cellular location of capsid assembly into infectious particles appears to occur in the nucleus. The capsid protei ns are first synthesized in the cytoplasm and imported into the nucleus for DNA pack aging. To determine if intact capsids are imported into the nucleus, Wistuba et al. (1995) analyzed soluble and nuclear AAV2 proteins separated by sucrose cushio n gradients. They found that capsid proteins which sedimented in the peak range of 5S-30S were found in the cytoplasm suggesting that monomers, di mers, trimers and/ or pentamers may exist in the cytoplasm. The capsid prot eins localized to the nucleus sedimented


18 within the peak range of 30S-160S suggesti ng protein oligomer s, not monomers, and intact full and empty capsids were pr esent in the nuclear fraction. Previous AAV Genome Packaging Studies In in vivo pulse-chase experiments, the rate of increase for the presence of DNA containing mature virions was the sa me as the rate of decrease of empty particles. (Myers and Carter, 1980). Fo r the development of a plausible genome packaging hypothesis for AAV t here are a few criteria th at must be considered. First, in the terminal resolution step of the 5’-3’ synthesis of AAV2 genome replication there is a co valent bond between a tyrosine residue of Rep68/78 and the 5’-terminus of the ssDNA (Ryan et al., 1996; Im and Muzyczka, 1990; Snyder et al., 1990; Prasad and Trempe, 1995). This attachment may persist in the fully assembled virion. Second, the 3’-5’ helicas e activity of the small Rep proteins is required for ssDNA genome packaging (C hejanovsky and Carter, 1989). Third, for the large portion of the AAV genome that was studied (3’-149 to 4518-5’), the 3’-region is processed into the nucleus before the 5’-region (King et al., 2001) In the study by King at al., (2001) , partially packaged capsids were probed for capsid–mediated nuclease protection. They used a series of labeled-probes that were able to detect a large fragment of the AAV2 genome including nucleotides 3’-149 to 4518-5’, but did not l ook at 3’-1 to148 or 4519 to 4678-5’. They found that the 3’ s equences were protected from nuclease digestion by the capsids, and that the 5’ terminal sequenc es were nuclease sensitive, supporting a model for 3’-5’ packaging into intact capsids (King et al., 2001). This means that there is currently no assay publis hed to detect the presence of the 5’ terminal 160 nucleotides in partially assembled capsids.


19 Studies by Im and Muzyczka, (1989), clear ly show an association of the large reps with at least 125 nucleotides of the 5Â’-terminus and that this binding could be inhibited by the presence a nucl eotide inhibitor. Binding was dependent on nucleotide sequence and on the hairpin structure and was detected in samples where capsid protein was not detected (Im and Muzyczka, 1989). This suggests that the non-covalent associat ion between the 5Â’-terminal hairpin and the large Rep proteins is reversible by competition from anot her binding partner. In their study capsid protein interaction with the 5Â’-terminus is not detectable in non-denaturing gels. Prasad and Trempe (1995) repor ted that they were able to isolate a particle that sedimented at 60S, associated with Rep, that had the same capsid:DNA ratio as full particles. Nuclease diges tion revealed that only 18% of the genome was nuclease protected. These findi ngs suggest that a Rep:DNA:capsid complex may exist at the early st ages of genome packaging where a large portion of the genome is associated with t he outside of the capsid and that 60S particles are not necessarily empty (Pra sad and Trempte, 1995). This is further supported with findings by Wistuba et al . (1990), who determined that in their 30S fractions, which were treated with DNa se and RNase prior to sedimentation, there was co-immunoprecipit ation of Rep proteins and unassembled capsid oligomers, or empt y capsids, possibly associated with DNA, implicating the large Rep proteins and DNA in capsid assembly. The MVM NS1 molecule, which is analogo us to Rep78, is found covalently attached to the 5Â’-end of the virus genome before and after genome packaging.


20 The NS1 protein remains attached to t he genome by a 24 nucleotide sequence. Both the NS1 protein and the 24 nucleotide region are attached to the exterior of the packaged capsid, but during infection, and internalization, the external ssDNA and the NS1 are cleaved off, s uggesting that the attachment is not required for infection (Cotmore and Tata rsall, 1989). There is a linker repeat nucleotide sequence between the NS1:DNA complex and the encapsidated genome that allows nuclease removal the NS1:DNA complex from the infectious virion (Cotmore and Tattersall, 1989). The Rep78 binding element and covalent attachment site have been studied in AAV2. These elements lie within 40 nucleotides of the 5Â’-terminal resolution si te and covalent attachment of Rep has been suggested to be with a thymidine of the stem-loop region (Brister and Muzyczka, 1999). Genome-Mediated AAV Capsid Assembly Hypothesis The previous data of AAV capsid assembly and genome packaging suggests that Rep78 is covalently asso ciated with the 5Â’ terminus of nascent AAV ssDNA and this interaction spans 40 nucleotides of the 5Â’ terminus (Brister and Muzyczka, 1999). Unassembled capsid pr otein oligomers, as well as emptyassembled capsids can be detected in association with Rep78 and unpackaged AAV2 genome, suggesting a role for the covalently linked Rep:DNA complex in capsid assembly. For a large porti on of the AAV genome 3Â’-terminus has processed into the capsid from 3Â’ -149 to 4518-5Â’, but the 5Â’-terminal 160 nucleotide that include t he RBE and covalent linkage site have not yet been accounted for.


21 Nucleic Acid Structure in Icosahedral Viruses Ordered nucleic acid has been visualized in crystal structures and cryo-EM reconstructions in ssDNA parvoviru ses CPV, MVMi, and AAV4 (Xie and Chapman, 1996; Agbandje-McK enna et al., 1998; Padron et al., 2005) (fig. 1-6). In the crystal structures of MVMi and C PV, nucleotides were modeled into the electron density inside the capsid shell. Atomic modeling was used to determine capsid protein amino acid s associated with the icosahe drally ordered regions of the viral genome (fig. 1-7). Mutations in MVMi capsid protein amino acids associated with the DNA nuc leotides reduced infectivity and affected capsid assembly (Reguera et al., 2004), supporting the hypothesis that nucleic acid packaging improves the assembly efficien cy and capsid stability in parvoviruses, though they are not believed to require a genome for assembly. The nucleic acid of icosahedral viruses has been visualized by cryo-EM in single stranded RNA (ssRNA) viruses Parioc otavirus, Satellite tobacco necrosis virus (STNV) and Calilivirus (Tang et al., 2001; Bink and Pleij, 2002; Prasad and Matson, 1994). A dodecahedral cage at t he interior of the capsid shell was described as the ordered portions of t he packaged viral genome. The internal electron density visualized in these caps ids appears to run along the inner face of the trimeric and dimeric interfaces between adjacent monomers, but not between the monomers at pentameric interf aces. The strongest electron density, the vertices of the dodecahedr on near depressions at the i nner threefold axis of the capsid, was interpreted as rigid dupl ex nucleic acid, acting as bridge by interacting with both the capsid and the genome (Rudnick and Bruinsma, 2005).


22 Figure 1-6. Ribbon represent ation of nucleotides model ed into the X-ray crystal structure of CPV (Xie and C hapman, 1996) and MVMi (AgbandjeMcKenna et al., 1998). CPV DNA is in green and MVM DNA is in red. Electron density associated with the inte rior face of the capsid protein was determined to be ordered DNA nucle otides closely associated with the capsid shell. The regions of DNA density common to the two structures are circled with additio nal regions of DNA seen in MVM and not in CPV. Images generated usi ng the program Chimera (Pettersen et al., 2004).


23 Figure 1-7. Atomic model of MVMi amino acids within 5Ã… of the ordered nucleotides (Agbandje-McKenna 1998). The nucleotides in cyan are those that differ between the CPV and MVM structur es (fig. 1-7). The amino acids in white are labeled with the amino acid residue number and monomer chain Identifier. Im ages generated using the program Chimera (Pettersen et al., 2004).


24 A proposed model of icosahedral vi rus capsid assembly and genome packaging involves the nucleic acid density running along the internal grooves of the capsid protein between subunit di mer interfaces, serving as an adhesive growth template to assemble 12 pent amer subunits (pentagonal pyramids) with adhesive edges (Zlotnick, 1994, Rudnick an d Bruisma, 2005). The visualization of ordered viral genome at the trimeric and dimeric m onomer interfaces supports the model of genome dependent capsid assembly with pentameric building blocks of capsid proteins assembling in to the icosahedron via twofold junctions, with rigid duplex nucleic acid associated at the inner threefold grooves organizing and stabilizing the subsequently packaged genome regions (Zlotnick, 1994)., Rudnick and Bruinsma, 2005). A single pass of nucleic acid helix could reinforce dimeric adhesion between pentamers, while multiple passes of nucleic acid would be required to reinforce a trimer of pentamers, suggesting an interaction at the capsid threefold interface w ould require duplex nucleic acid. Determination of Macromolecular Structures Many biological functions are medi ated by complex molecular assemblies that are comprised of num erous components. Due to the complexity, structure determination of these macromolecular asse mblies is very difficult, but important to understand their assembly and dynami cs. X-ray crystallography is one method for determining the atomic stru cture of molecules, but relies on the crystallization of a highly pure and concentra ted sample. This can be difficult to achieve for macromolecular assemblies. Cryo-EM is a method for determining the surface electron-density envelope for a particular molecule or assembly, and


25 does not require crystallization of the s pecimen to obtain important structural information. These techniques can be us ed together to study the structure and function of molecular assemblies and complexes (Tang and Johnson, 2002) Structure Determination by X-ray Crystallography Crystallization by the hanging-drop method. Milligram quantities of highly concentrated sample, 1-10 mg/m L, are required for crystallographic studies. The hanging-drop method is a common approach for growing crystals (McPherson, 1982). Using this method, the sample drop is mixed with a precipitant solution and equilibrated in a sealed environm ent against a well solution that is more concentrated for pr ecipitants than the sa mple. Over time, the water slowly diffuses to the well solution from the sa mple. The slow dehydration of the sample causes cryst al formation. The crystallization conditions are screened by varying pH and temperature as well as precipitant concentrations. X-ray crystallographic data collection. X-rays are electromagnetic radiation of short wavelength (1Ã…) and high energy, and are absorbed, diffracted and scattered by matter. The short wavelength is needed to differentiate between the atomic distances in the mole cule. The length of a covalent bond ranges from 1-2 Ã… and hydrogen bonds from 2.5-3.5 Ã…. Like all electromagnetic waves, X-rays can be characterized by their phase and amplitude. Lenses can convert the phases and amplitudes of tr ansmitted or scattered waves into an image of the scattering object. There are lenses that exist for visible light, which has a wavelength ranging from 4,000 to 8, 000 Ã…, but lenses of this type do not exist for x-rays so other techniques mu st be used to create the image.


26 Rotating anode generators and synchrotr on sources which are both used for crystallographic studies, can produce x-rays with wavelengths ranging from 0.5-1.6 Å. The detectors of x-rays can measure the in tensity, the square of the amplitude, of the scattered waves, but cannot detect the phase information. The crystal is used to amplify the diffr action signal of the sample. Crystals are organized in repeated el ements called the unit ce ll which can be described by vectors, a,b,c, and angles, , ß, . The x-rays diffract off of the planes of electron density that dissect the unit cells in an equivalent manner. The diffraction off of each plane is contributed by all of the electrons of the unit cell. Just as the crystal and the crystallized molecule are three-dimensional, so is the diffraction of the x-rays, and therefore, each two dimensional image of the diffraction pattern must be sampled by slight ly oscillating the crystal (Chiu et al., 1997). The crystal is rotated by a set angle for subsequent images. X-ray data processing. The spots of the captured diffraction pattern are distinct because of the lattice functi on of the crystal. The spacing between diffraction spots is inversely proportional to the spacing of crystal lattice points, and so the diffraction pattern represents a sampling of reciprocal space (Chiu et al., 1997). Analysis of the spot di stances can be used to determine the symmetry organization, space group symmetry, in the unit cell of the crystal. The spots of the diffraction pattern are assigned i ndices in reciprocal space (h, k, l). The number of electrons that contribute to the diffraction is called the structure factor (F) and is dependent on the distribution of atom s in the unit cell and the


27 direction of scattering (Rossman and Arnold, 2003). This sum of electrons also takes into account the differences F(hkl) = fj cos (2 rj x S) + i fj (2 rj x S) j=1 j=1 where r is the vector distance between atoms and S is a vector perpendicular to the plane in the unit cell th at reflects the in cidence ray at angle (Drenth, 1999). The scatte ring factors, f, are propor tional to the number of electrons within a particular atom and are expressed in terms of scattering by a single electron and decreases wit h increasing scattering angle. Bragg”s Law describes diffracti on conditions as the equation, n = 2 d sin where n is an integer, is the wavelength, d is the distance between scattering planes and is the angle of diffraction. When this equation is met, constructive scattering occurs where all of the unit cells in the crystal scatter in phase and the amplitude of the scattered wave is proportional to the st ructure factor amplitude F. The missing phase information must be determined to properly describe the wave function and solve the structure of the molecule. A technique for this will be described below. The initial diffraction images can be used to determine the organization of the unit cells in the crystal lattice, t he space group. Processing the complete data set includes reducing to a set of unique reflections, and scaling to normalize the intensities. This processed data se t is used to calculate Rsym, which can indicate how well the data set fits the parameters of the assi gned space group.


28 Molecular replacement is a phase dete rmination technique that allows the use of a previously solved, homologous st ructure as a phasing model to estimate the initial phases for the unsolved stru cture, if the phase model and the unknown structure are isomorphous. The amp litudes of the unknow n structure data and the phases of the model are used in the wave function used to obtain an initial electron density map (Rossman, 1990). As the model is refined, the phase estimates are improved and extended to include more reflections from the observed data. To determine the quality of the crystal structure, Rfactor is calculated to determine how closely the re fined map fits the observed data. This statistic is improved at the model improves and is used as a guide for model refinement. Structure Determination by Cryo-Electron Microscopy Sample preparation. Determining the structure of molecules by X-ray diffraction requires a deduction of the struct ure mathematically from a diffraction pattern. This is due to the fact that t here are no lenses to focus the X-ray beam after interaction with the specimen. An electron beam can be focused by magnetic or electrostatic lenses to give an image of the object on a photographic plate or a fluorescent screen (Engel, 1991). For electron microscopy contrast is needed to see the object, therefore electron dense stains are used which can al ter the structure of the sample (fig. 18). Cryo-microscopy is the collection of transmission electron microscopy data at temperatures low enough to keep ice cr ystal contaminants from forming (Thunman-Commlike and Chiu, 2000). In cr yo-microscopy there is no stain,


29 therefore the sample images are captured slightly under focus. This allows visualization of the sample witho ut staining (Baker et al., 1999). The sample must be at reasonably high concentration and an aliquot of 2-3 µl at 0.05-5 mg/ml is applied to a grid. The sample is blotted and then quickly plunged in to a bath of liquid ethane cooled in liquid nitrogen (Baker et al., 1999). Because the cooling occurs very rapidly the water in the sample is vitrified rather that crystallized forming a thin (0.2 µm) la yer over the sample. This immobilizes the sample and retards the damage to t he sample from the intense electron beam. The sample is kept at this te mperatures below -160 ºC by constant bathing in the cryogen until the grid is in the microscope vacuum and data collection from the grid is complete. Data acquitition. The data is collected with a microscope cooled by helium or liquid nitrogen (Orlova and Saibil, 2004). The grid is first searched at low magnification (X 2000-3000) and irradiati on levels (less than 0.05 electrons/Å2/s) to judge the quality of vitrific ation, distribution of indivi dual sample particles, and to locate an appropriate sampling region (Bak er et al., 1999). Data are collected at 50,000 to 80,000 times magnification and the images are recorded by a charged coupled device photographic film. T he sample images are taken slightly underfocus because at focus, where the highest resolution data is, there is very little contrast between the sample and t he background making viewing difficult. The defocus level contributes significant ly to the contrast transfer function (Ludke, 1999).


30 A B C Figure 1-8. Negative stain electron micr ograph of AAV capsid proteins. Picture was taken at 40,000 times magnifica tion and 2% uranyl acetate was used as a stain. A) Full particles B) empty particles and C) partially packaged capsids are present in this preparation.


31 CTF and envelope function correction. The electron waves are phase shifted upon interaction with the atoms of the sample. Subsequent phase shifting occurs due to defocus and lens imperfecti ons. Electrons scattered by the object in the back of the focal pl ane of the objective lens equate to the Fourier transform of the object. The image is then magnifi ed and projected by the objective lens into the back image plane of the microscope to generate the 2D projection, which is the inverse 2D Fourier transform of the original 3D object (Henderson, 1992; Bubeck, 2005; Frank, 2005). Phase Shift of the incoming wave can be de scribed by its interaction with the objects proj ection potential (r), Â’ = i (r). If the electron is only scattered by the thin sample one time then (r) is much less than one, therefore, Â’ = [1+i (r)]. After passing through t he sample the beam is focused by the objective lens. For an ideal lens at focus the phase shifted wave could be described as a Fourier transform of that wave in the back focal plane where the Fourier transform of is a delta function, bf = F{ Â’} = (k) + iF{ (r)} (Henderson, 1992; Bubeck, 2005; Frank, 2005). But the lens is never perfect and def ocus is needed for contrast. Defocus and spherical aberration of the lens c ause an additional phase shift to the scattered electron wave which can be described as, (k) = 2 k, k = spatial frequency and =contrast transfer function. Th is modification can be detected and analyzed in the back focal plane of the objective lens where the diffraction pattern representing the 2D Fourier transform of the object can be observed. A more accurate description of the wave f unction at the back of the focal plane can


32 be described as, bf (F) = F { Â’}exp[i (k)] (Henderson, 1992; Bubeck, 2005; Frank, 2005). Due to the large depth of focus of the objective lens the 2D projection is formed in the back image plane of the micr oscope. This image is recorded by a microscope camera or CCD detector. The wave function, or image intensity (k) observed in the back image plane of t he microscope is the reverse fourier transform of the wave in the focal plane, I(k)=2 r(k)sin (k)-2 i(k)cos (k). The contrast transfer function is a mat hematical representat ion describing the additional phase shift of the scattered electron wave caused by the objective lens, ( )=2 / (-Cs 4/4) + ( f 2/2) (Henderson, 1992; Bubeck, 2005; Frank, 2005). Image analysis and 3D reconstruction. The micrographs are digitized by scanning with a step size half of the needed resolution (Ludke, 1999). The digitized micrographs are viewed and individ ual particles are boxed in a program such as RobEM (Baker et al., 1999). The boxed particles are centered in the boxes by eye. Fourier transforms of each image is regenerated and are CTF corrected. A model with the same symmetry as the sample is needed to compare with your image. A Polar Four ier transform of the asymmetric unit of the model is used to search the Fourie r transform of each image in order to estimate the orientation of each parti cle (Chang and Bake r, 1999). Programs such as PFTsearch are used to determine the orientations, , , and , of each center-estimated, x and y, image (Baker et al., 1999). The two-dimensional sample images with orientations estimat ed against the search model are used to


33 generate an initial three-dimens ional map in a program li ke EM3DR (Baker et al., 1999). The density map refinement process is iterative with several rounds of refining the origins and orientations of the images and then constructing the three-dimensional image. The newly generated map becomes the search model for the subsequent round of refinement, gradual ly reducing the contribution of the original search model to the final recons truction (Baker et al., 1999). The volume of the refined model must be scaled to the appropriate mass of the protein. This can be done by scaling the volume of the density map to the known atomic structure by altering the pixel size of t he map within a 10% error range (Baker et al., 1999). Virus:Ligand Structures of Icosahedral Viruses X-ray crystallography and cryo-EM have been used to ellucidate the structure of viruses bound to domains of their host-cell receptors, to oligosaccharide components of their recept ors, and to antibodie s (Mab and Fab). In the case of the human rhinovirus (HRV), cryo-EM and image analysis techniques were used to calculate a thr ee-dimensional reconstruction of HRV-16 bound to the DID2 immunoglubulin domai ns of the intercellular adhesion molecule-1 (ICAM-1) (Olsen et al., 1993). The structure of the virus, and the structure of a homologue of their recept or, CD4, had previously been solved by X-ray crystallography. By superimposing the known atomic structural models onto the electron density generated by cryo -EM they were able to predict the atomic interactions in the virus~receptor complex.


34 Structures of two complexes of HRV-14, with the neut ralizing antibody Mab17-IA and the Fab fragm ent Fab17, were determined by cryo-EM and X-ray crystallography (Che et al., 1998). In these studies the 4 Ã… X-ray crystal structure of HRV-14:Fab17 was compared to the cryo-EM reconstruction of the same complex to prove t hat the pseudo-atomic model constrained by the 22 Ã… cryo-EM density map was a very close approximation of the volume and details of the high resolution X-ray crystal struct ure. This evidence supports the idea that lower resolution structures can be validated and analyzed with atomic data. There are several structures in the pr otein data base of macromolecules in a complex with heparin. Heparin is a pol ymer of alternating sulfated iduronic acid and sulfated glucosamine. The cati onic residues of heparin binding proteins interact electrostatically with sulfat e and carboxylate groups of the heparin molecule. The hydroxyl groups of the heparin molecule are also involved in hydrogen bonding. There can be direct contact between the protein and heparin, or water may act as a bridge between amino acids and heparin. Arginine and lysine residues serve as the important heparin binding residues at physiological pH (Cardin and Weintraub, 1983; Fromm et al., 1995). The complex between fibroblast growth factor (FGF) and a hexasaccharide of heparin requires lysine and arginine residues and two water molecules for binding (fig. 1-9) (Faham et al., 1996). The structures of the complex between FGF:heparin:FGFr demonstrated that the oligosacchar ide helix has multiple sulfated faces that can interact with more than one molecule at a time. This ternary complex provided an explanation for the interaction between the two


35 protein molecules facilitating the dimeri zation of the receptor tyrosine kinase FGFr required for activation of signal transduction (Pellegrini et al., 2000; Schlessinger et al., 2000). In the case of the Foot and Mouth Disease Virus (FMDV), virus crystals were soaked with oligosaccharides of hepar in (Fry et al., 1999). Difference maps calculated between the complex data set and the native (uncomplexed) data set clearly showed electron density for the oligosaccharides. They were able to use the known structure of heparin to interp ret the unassigned electron density. A close look at the crystal st ructure of FMDV in a complex with heparin (fig. 1-10) reveals that surface lysine and arginine resid ues play a crucial role in the binding, but that other residues can contribut e through interactions with solvent molecules. There are several solvent mo lecules that line the interface between the virus protein capsid surface amino acids and the heparin molecule ligand (fig. 1-10A, 10B and 10E). The authors note t hat there are several conformations that the heparin probably takes on when fi tting into the binding pocket and this is apparent in their modeling. More recently, the cryo -EM structure of the Ross River Virus complexed to an oligosacc haride of heparin sulfate was determined to 22Ã… resolution. Difference map analysis comparing the complex and the uncomplexed structures revealed density fo r the heparin at the threefold tips of the virus capsid (Zhang et al., 2005).


36 Figure 1-9. Atomic model of fibrobl ast growth factor (FGF) bound to an oligosaccharide of heparin sulfate (Faham et al., 1996). The atomic model is based on the crystal structur e. A) Basic residues that make up the heparin binding region are in blue with nitrogen atoms in black. Water molecules are represented as cyan spheres. The black ribbon represents the non-heparin binding region of the FGF, while contact regions are in purple. The carbon atoms of the heparin molecule are red with the sulfur atoms in green and oxygen atoms in white. B) Close up of the sugars of the heparin and amino acids of the heparin binding region are labeled in black. These images were generated using the program Chimera (Pettersen et al., 2004).


37 Figure 1-10. Atomic model of Foot and Mouth Disease Virus bound to an oligosaccharide of heparin sulfate (F ry et al., 1999). A) The atomic model is based on the crystal structur e. Basic residues that make up the heparin binding regions are in blue and all other amino acids participating in heparin binding are in purple. Nitrogen atoms are in black. Water molecules are represented as cyan spheres. The black ribbon represents the non-hepar in binding region of the FMDV capsid. The carbon atoms of the heparin molecu le are in red, the sulfur atoms in green, and oxygen atoms in white. B) Close view of the heparin binding region. Amino acids intera cting with the oligosaccharide are labeled in black. C) Detail view of the interaction between the heparin binding region and the heparin ligand. D) The residues of the FMDV capsid that interact with heparin. The amino acid abbreviations, residue number and polypeptide chain identifiers of the amino acids involved are labeled in bla ck. E) Space filled representation of water molecules facilitating the interaction between the protein and carbohydrate ligand are in cyan. Images generated us ing the program Chimera (Pettersen et al., 2004).


38 CHAPTER 2 MATERIAL AND METHODS AAV Capsid Protein Expression Vectors Plasmids pXYZ1 (AAV1), pIM45 (AAV2), and pXYZ5 (AAV5) contain cap and rep genes for the production of reco mbinant AAV (rAAV). These plasmids are used in conjunction with plasmid p XX6 which provides adenovirus helper function in an adenovirus free environment. The cap gene was excised from these plasmids and placed into expre ssion vectors for the production of AAV capsid proteins. Adenovirus expressi ng AAV2 capsid prot ein and Baculovirus expressing AAV5 capsid were obtained as a generous gift from Dr. Sergei Zolotukhin. For construction of an adenovirus to express AAV1 capsid protein, the restriction enzyme HindIII was used to di gest the pXYZ1 template plasmids to release the entire AAV1 cap open reading frame. This gel fragment was visualized with ethidium bromide and purified from a 0.5% agarose gel. The purified HindIII fragment was ligated into a shuttle plasmid, conferring Kanamycin resistance to the transformed bacteria and placing the AAV1 capsid gene under the control of the cytomegulovirus (C MV) promoter using the Stratogene Adeasy Expression Vector System. The shuttle ve ctor contained a multiple cloning site between the CMV promoter and the SV40 po lyadenylation signal and a multiple cloning site in to which cap gene was inserted. The arm regions of the shuttle vector plasmid are the stretches of sequence homology within a second larger


39 plasmid containing the adenovirus genes. The R-ITR and L-ITR regions of the shuttle plasmid are short inverted terminal repeats (Right and Left) which participate in the replication of the vira l DNA. A recombination event in bacteria BJ5183 cells between the shuttl e vector containing the cap gene and the larger plasmid containing the E1-E3 deleted A d5 genome resulted in the generation of a new plasmid. The ampicillin resistance gene of t he larger plasmid allowed for the selection of recombinant plasmid. T he polymerase chain reaction (PCR) was used to determine the presence of the AAV1 capsid gene in the recombinant plasmid. The primer set 5’-GGT TATCTTCCAGATTGGCTCGAGG-3’ and 5’TTCGGTGGCCACAGGGTTAGTG-3’, forward and reverse primers respevtively, which are specific for the AAV1 capsid gene and cannot amplify the closely related AAV2 capsid gene, were used. The PCR was carried out with an initial denaturing step at 95º for 3 min followed by 35 rounds of the following thermocycle: 95º for 45 sec, 63º for 45 se c, 72º for 3 min. This thermocycle was followed by a final elongation step at 72º for 5 min. The new recombinant plasmid containing the AAV1 capsid gene was then lineariz ed with PmeI and transfected into human embryonic kidney (HEK293) cells using the Lipofectamine plus system to create an adenovirus capable of expressing the AAV1 capsid gene. Initial adenovirus expression vector production was detected by monitoring cytopathic effect (CPE) of the transfected cells. Transfected cells were harvested 72 hours post-transfection. Ex pression of AAV capsid protein was


40 monitored by western blot analysis of t he crude cellular lysate. The adenovirus was amplified in 293 cells which were monitored for CPE, then harvested and plaque purified to obtain a single Adenov irus clone. A single plaque was selected and amplified in a 3 cm dish of HEK 293 cells. Following the detection of CPE the media was used to seed on a 15 cm dish. The infectious media was titred by plaque assay and served as the adenovirus stock. Capsid Protein Expression AAV1 and AAV2 capsid proteins, which self-assemble into virus-like particles (VLPs) were expressed by adenovirus expression vector viruses in HEK293 cells grown in Dul becco Modified Media, suppl emented with 10% bovine calf serum (BCS) and 100U/ml penicillin and streptomycin. Ten 15 cm plates of low passage 293 cells, that were about 80% confluent, were infected at a multiplicity of infection (MOI) of 520 with recombinant a denovirus expressing AAV capsid protein in serum free conditi ons for 1 hour at which time the cells were supplemented with 10% BCS and allo wed to incubate for 48-72 hours until CPE was first detected. The cells were harvested by centrifugation 1000 X G for 10 minutes and the cell pellet was retai ned for capsid protein purification. AAV5 VLPs were expressed by a baculovirus expression vector in SF9 cell culture in an Erlenmeyer flask with Sf -900 II serum-free media rotating 135 to 150 rpm at 27ºC. Cells were grown until they reached 2 X 106 cells/ml and then subcultured for subsequent use. To make a baculovirus stock, 150 mls of SF9 cells at 2 X 106 cells/ml were infected at an MO I of 0.5 with the baculovirus containing the AAV5 cap gene. The culture was har vested at 24-48 hours postinfection when cells showed CPE and titred by plaque assay. Virus stock was


41 stored at 4ºC protected from light. AAV5 VLPs were expressed in 2 X 1 litre flasks containing 2 X 106 SF9 cells/ml in serum-free me dia. Cells were infected at an MOI of 5. The culture was harvested at 72 hours post-infection by centrifugation at 1000 X G for 10 minutes. T he cell pellet was retained for capsid protein purification. Purification of Adeno-Associ ated Virus-Like Particles Purification of AAV VLPs was previously described (Zolotukhin et al., 2002). This protocol was followed with a few modifications. Harvested cell pellets were resuspended in 30 ml lysis buffer for cells grown in 10 X 15 cm dishes, and 60 mls for cells grown in a litre suspensi on, containing 150 mM NaCl, 50 mM Tris pH 8.4. Cells were lysed by three cycles of freeze/thawing, alternating between ethanol over dry ice and a 37ºC water bath. Following the final thaw cycle Benzonase (Sigma) was added to the lysate (50U/ml) and incubated at 37ºC for 30 minutes. The lysate was clarifi ed by centrifugation at 4000 X G for 20 minutes. The supernatant was divided into 15 ml aliquots for further purification by discontinuous iodixanol (5,5’[(2-hyd roxy-1-3-propanediyl) bis [N,N’-bis(2,3 dihydroxypropyl-2,4,6triiodo-1,3-benzenecarbox-ami de]) step gradients. The layers of the iodixanol gradient we re 5 mls of 60%, 5 mls of 40%, 6 mls of 25%, and 9 mls of 15% c ontaining 1M NaCl. The io dixanol steps were made by underlaying in a 50 ml quick seal tube (B eckman) to displace , first the 15 mls of cell lysate and then the less dense layers . The iodixanol solutions were made using 60% (w/v) sterile solution of Opti Prep (Nycomed) and 1 X Tris-EDTA (TE) buffer containing 1 X PBS, 1 mM MgCl2 and 2.5 mM KCl. Sealed tubes were centrifuged in a 70 Ti rotor at 69,000 rp m for 1 hour at 18 ºC. Assembled VLPs


42 were collected from the 40:25% interface by puncturing the side of the tube with an 18 gauge needle attached to a syringe. AAV2 VLPs were further purified on a 3 ml heparin agarose Type I column (Sigma), equilibrated with 10 ml of TD bu ffer with 1M NaCl, followed by 20 ml of TD buffer. The iodixanol fr actions were loaded onto the column by gravity-flow to allow the VLPs to bind to the column . The unbound material was washed off the column with 20 ml of TD buffer. The VL Ps were eluted off of the column with TD/500mM NaCl. AAV1 and 5 VLPs were further purified using 5 ml HiTrap Q column (Pharmacia). The column was equilibrated at 5 ml/min with 25 ml of Buffer A containing 20 mM Tris and 15 mM NaCl (pH 8.0 for AAV5 and pH 7.5 for AAV1), then 25 ml of Buffer B c ontaining 20 mM Tris and 500 mM NaCl and then 25 ml of Buffer A. The iodixanol fractions were diluted 1:1 with Buffer A and then applied to the column. The unbound material was washed at least five column volumes of Buffer A. The particles were el uted off the column in 1 ml fractions of Buffer C containing 20mM Tris and 350mM NaCl. To confirm the purity a nd quality of the express ed VLPs, SDS-PAGE silver stain, western blotting with B1 anti body, negative stain (2%UA) and electron microscopy were used. Viruses we re concentrated using Amicon Ultra Centrifugal filters with a 100,000 MW cut off. Virus concentrations were determined by UV absorbance reading at =280 to determine the amount of purified virus protein in mg/ml using the conversion factors in the units mg/ml as follows: AAV1, Abs( =280)/1.665; AAV2, Abs( =280)/1.695 and AAV5,


43 Abs( =280)/1.824, using a cuvette with a pat h length of 1. The conversion factors were calculated based on extinc tion coefficients for VP3 using ProtParam tool ( ) for each serotype. The use of the conversion factor for AAV2 has been confirmed by the extinc tion coefficient calculated in Sommer et al., (2003) where the conversion fact or was found to be 1. 743. We calculated that for every milligram of capsid there are 1x1014 virus particles. Western Blot Analysis for the Detection of Capsid Protein Purified virus stocks were boiled in 1 X SDS gel loading buffer. Samples were allowed to denature for 5 minutes on a heating block. The samples were then loaded onto a 10% SDS polyacrylamid e gel. Proteins were separated on the gel by electrophoresis at 10V/cm. Sa mples were transferred from the gel to nitrocellulose by electr ophoresis at 100mA for 18 hour s. The nitrocellulose membrane was blocked in 10% non-fat dried milk/PBS/0.1% Tween 20, with rocking at room temperature for one hour or at 4º over night. Primary B1 antibody was diluted 1:3000 in blocking so lution and the membrane was exposed to the antibody, with rocking at room temper ature for one hour or at 4º over night. The membrane was rinsed in PBS/0.1% Tween 20 and then exposed to antimouse conjugated with horse radish per oxidase, diluted 1:1000 in PBS/0.1% Tween 20, with rocking for 1 hour at room temperature. T he proteins were visualized by chemiluminescence using the standard manufactu rers protocol (Amersham Life Science). Negative Stain Electron Microscopy Purified AAV VLPs were viewed us ing a Joel JEM-100CX II electron microscope (EM). A 3ml dr op of purified virus in solution at an estimated


44 concentration range of 0.12.0 mg/m l was spotted onto a 400 mesh carboncoated copper grid (Ted Pella, Inc., Redding, CA, USA) for 1-2 min before blotting with filter paper (Whatman No. 5) . The length of time the sample was allowed to settle on the grid increased as sample concentration decreased. The sample was stained with 3 ml of 2% uranyl acetate for 16 s and then blotted dry. Grids were allowed to air dry for 20 minutes prior to being viewed. SDS-PAGE and Silver Stain Analysis Purified virus stocks were boiled in 1 X SDS gel loading buffer. Samples were allowed to denature for 5 minutes on a heating block. The samples were then loaded onto a 10% SDS pol yacrylamide gel. Proteins were separated on the gel by electrophoresis at 10V/cm. The gel was washed with de ionized water for 5 min and then fixed with 10% acetic acid and 30% ethanol for 3 hours. The gel was then washed with deionized wate r for 5 min and then soaked in 10% glutaraldehyhe (Fisher G151-1) for 30 mi n, and then washed 5 times for thirty min in deionized water. The proteins embedded in the gel were then silver stained by bathing the gel in a silver nitrate solution containing 0.2g AgNo3 / 100 ml of 37% formaldehyde solution for 10 min and then washed in deionized water 2 times for 5 min. The bands were allo wed to develop for 1-5 min until the VP bands were clearly visible. The development was stopped by soaking the gel in 5% acetic acid for 5 min and then 10% ethanol for 15 min. To preserve the gel, the gel was soaked in 7% glycerol, 10% ethanol for 3 hours and then dried. Extraction of Nucleic Acid from Purified Virus-Like Particles Purified AAV2 VLPs were treated with 20 mg/ml each of RNAse A and DNAse (Sigma) to degrade any nucleic acid not protected by the capsid protein


45 shell. Nuclease treated samples were then purified on a heparin column as described above and treated wit h Proteinase K (Sigma) to degrade the capsid protein shell and release packaged nucleic acid. Phenol:chloroform extraction was used to remove protein and the remaining sample was ethanol precipitated and vi sualized on a 0.5% agarose gel with ethidium bromide. The protein:nucleic acid ratio was estimated using UV absorbance. Extracted nucleic acid wa s treated with 20 mg/ml of either RNAsefree DNAse I or DNAse-free RNAse A (Roc he) to test enzyme susceptibility. Complexing of AAV2~Heparin Oligosaccharide Capsid protein-heparin oligosacchar ide complexes were made by mixing purified virus-like particle (VLPs) with hepar in oligosaccharides (Sigma H5281) with an average molecular weight of 6 kD a. We calculated that there was an average of, approximately, 9 disaccharides per oligosaccharide. The sugar was mixed with the virus sample at room tem perature at a ratio of 1 virus-like particle per 120 heparin oligosaccharides molecule s. The sample was incubated for 30 min and then the complex sample was vitr ified on carbon-coated grids in liquid ethane cooled with liquid nitrogen. Cryo-Electron Microscopy Vitrified VLPs of AAV1, AAV2 and AAV2~heparin complex were inserted into a precooled, Gatan 626 cryotransfer holder that maintained a constant temperature of –176 °C. Samples were examined with a Philips CM200 FEG transmission electron microscope, and images recorded onto Kodak SO163 film. The size of the electron beam was adjusted to be within the limits of the 2k2 slow-scan CCD detector (0.4 µ m on the specimen) to allow images to be


46 recorded without preirradiatin g adjacent regions. A low-magnification (×4,560) survey image was captured on the CCD, and areas for high-magnification imaging were identified and selected. The images were then recorded on film at 200 kV, × 38,000 magnification, 0.5-3 µ m underfocus. Data was collected underfocus because this improved the cont rast of the unstained sample. The range of defocus levels allowed for more of the details of the capsid to be observed because different features were more clearly seen at different defocus levels. The images were scanned by us e of a Zeiss SCAI scanner with a step size of 7 µm/pixel and then twofold bin averaged to 14 µm/pixels, resulting in a pixel size of 2.98 Å at the specimen. Three-Dimensional Reconstruction of AAV Digitized micrographs were imported in to the program ROBEM (Baker et al., 1999). Individual particles were selected by boxing using a defined box diameter of 111 pixels for AAV1 and 121 pixels for AAV2 and AAV2~heparin. Background differences between individual boxed particles were normalized. The particles were compared to an icosahedral model and the centers and orientations of each image were estimated using the program Pftsearch (Baker et al., 1999). A three dimensional Pur due image format map (Pifmap) was generated using the program EM3DR and th e map displayed in ROBEM (Baker at al., 1999). Pftsearch was repeated us ing the newly generated AAV Pifmapas the orientation search model. Image rec onstruction was iterated and the density map was improved by the selection of the particles based on correlation coefficient calculations by the EMSEL program (Baker et al., 1999). The correlation coefficients for the select ed particle image were above 0.3. The


47 density map resolution was determined by di viding the particle image set into two to generate two different reconstructions. The Fourier shell correlation (FSC) coefficients and phase residuals were plott ed as a function of spatial frequency. The resolution was determined to be the in verse of the spat ial frequency where the FSC crosses 0.5 and the phase erro r becomes greater than 45º. Difference Map Analysis The difference map was calculated usi ng structure factor generated by the program Uniconvert by reverse Fourier transformation of the electron density maps of the AAV2 and AAV2-heparin comp lex. CCP4 programs were used to subtract the structure fact ors for the uncomplexed map fr om the structure factor of the complex map and to generate a density map from t he reflections that differ between the files (Collaborat ive Computational Project, 1994). The difference density was visualized in the programs Chimera (Petterson et al., 2004). The atomic coordinates of the AAV2 VP3 monomer (Xie et al., 2002) were used to build a trimer, pentamer, and an entire capsid (60 X VP3) by matrix multiplication using 3-fold, 5-fold and icosahedral symmetry operators. The difference map was superimposed onto the model of the capsid in the program Chimera (Petterson et al., 2004) and the volume of the map was scaled to the atomic model, by rendering the map vo xel size to 4.5 at a contour level of 1.9. The coordinates for a heparin hexasaccharide were taken from the coordinates of the fibroblast growth factor (FGF) ~heparin complex crystal structure (fig. 1-9) (Faham et al., 1996), and then docked into the positive difference density using the program Chimer a (Petterson et al., 2004). The fit of the heparin ligand into the difference map was first estimated by aligning basic


48 residues of FGF with that of the heparin binding region of AAV2 on the capsid surface. The position of the heparin was then manually adjusted within the difference density to maximize possible inte ractions with the residues. However, it should be noted t hat there were other possible conformations for the heparin that would enable association with the receptor binding cleft on the capsid surface including orientations where bot h FGF and AAV2 capsid can be bound to different faces of the heparin molecule at the same time. Pseudo-Atomic Model Construction The amino acid sequences of AAV1 , 2 and 5 were obtained from the National Center for Biotechnology In formation [NCBI] database and were aligned, pair-wise, in CLUSTALW (v. 1.4) (Thompson et al., 1994), with the parameters set as follows : gap opening penalty, 10; gap extension penalty, 0.5; and the BLOSUM matrix (Henikoff and Henikoff, 1992). Using the AAV2 structure as a template, the amino acid sequence of AAV1 VP3 was submitted to the online atomic model generator Swiss Model ( ) (Padron et al., 2005). The Swiss Model coordinat es, moved to a standard orientation, were superimposed into the reconstruct ed density map in the program Chimera (Petterson et al., 2004). Generation of a Surface Map Based on the AAV2 Atomic Model A low-resolution surface map of AAV2 was calculated from atomic coordinates obtained by X-ray crystallograp hic structure determinations (Xie et al., 2002) (PDB accession numbers ILP3 ). To generate a model of 60 VP3 molecules, icosahedral sy mmetry operators were applied to the VP3 monomer in a standard orientation as def ined by (Rossmann et al., 199 2). Structure factors


49 were calculated to 18 Å resolution to be compared to the cryo -EM structures of AAV2 and AAV2~heparin complex by using the CCP4 package (Collaborative Computational Project, 1994). A lo w-resolution surface density map was generated by Fourier transform of the stru cture factors (Padron et al., 2005). VLP Crystallization AAV1 Crystallization Screens AAV1 crystallization conditions were screened by varying polyethylene glycol (PEG) 8000 (0.5–4.0%) and MgCl2 (5–20 mM) concentrations at room temperature (RT) and 4º C in a buffe r containing 350 mM NaCl in 20 mM Tris– HCl at pH 7.5. The hanging drop vapo r-diffusion method (McPherson, 1982) was used in VDX 24-well plates and silicon ized cover slips (Hampton Research, Laguna Niguel, CA, USA). The sample drop contained 5 µl of VLP at 7.0 mg/ml or 4.65 mg/ml, and 5 µl precipitant soluti on equilibrated against 500 µl precipitant solution. AAV2~Heparin Co-crystallization Screens AAV2 samples at 2 mg/ml were comp lexed with 20 mg/ml heparin sodium salt (Sigma H9267). Crystallization conditions were screened by varying polyethylene glycol (PEG) 8000 (0.5–6.0% ), NaCl (300–500 mM) and glycerol (025%) concentrations at RT and 4º C in the precipitant solutions with a buffer concentration of 1 X phosphate buffer solution (PBS), 1 mM MgCl2, 2.5 mM KCl at pH 7.5. The hangin g drop vapor-diffusion method (McPherson, 1982) was used as described above for AAV1. The sample drop contained 5 µl of AAV2~heparin salt solution and 5 µl of precipitant solution equilibrated against 500 µl precipitant solution.


50 AAV5 Crystallization Screens AAV5 crystallization conditions were screened by varying the composition of the precipitant solution. We vari ed the concentration of polyethylene glycol (PEG) 8000 (0.5–2.5%), Na Cl (250 and 350 mM), MgCl2 (5–20 mM), and varying the pH range (pH 7.0–8.5) of 20 mM Tris-H Cl, at RT and 4º C. The hanging drop vapor-diffusion method (McPherson, 1982) was used for crystallization. The sample drop contained 5 µl of AAV5 VLPs at 6.5 mg/ml, and 5 µl of precipitant solution, which was equilibrated agai nst 500 µl precipitant solution. AAV5 X-ray Data Collection and Reduction AAV5 crystals were cryo-protected in 30% glycerol in the precipitant solution including 5% PEG 8000 and flashfrozen in a nitrogen stream. X-ray diffraction images were collected at the X29 beamline at the National Synchrotron Light Source (NSLS, Brookhaven National Laboratory) at = 1.100Å on an ADSC Quantum Q315 CCD detector and at the F1 station at the Cornell High Energy Synchrotron Source (CHESS, Cornell University) at = 0.924 Å on an ADSC Quantum 4 CCD detecto r. Crystals were oscill ated 0.3º at crystal-todetector distances of 300– 400 mm. The images were recorded at exposure times of 30–60 s on the X29 and 360 s on the F1 beamlines. The diffraction intensities were inde xed, integrated, scaled, and merged using the HKL2000 suite of programs (DENZO and SCALEPACK; Otwinowski & Minor, 1997). The TRUNCATE progr am from CCP4 (Collaborative Computational Project, Nu mber 4, 1994) was used to convert the intensity data set to structure factor amp litudes (Dimattia et al., 2005).


51 CHAPTER 3 CRYO-ELECTRON MICROSCOPY, THREE-DIMENSIONAL RECONSTRUCTION AND CRYSTALLIZATION OF AAV SEROTYPE 1 Introduction AAV1 as a Gene Therapy Vector Several serotypes of AAV are studied for their use as gene therapy vectors and differ in their host-cell tropism, rec eptor binding affinity and antigenic properties. These serotypes include AAV 1-8 and the more recently described serotypes 9-11 (Denby et al., 2005, Gri eger and Samulski, 2005, Mori et al., 2004; Gao et al., 2002, 2003, 2005). AAV1 has potential use as a gene delivery vector, efficiently transducing murine islets (Loiler et al., 2003) and murine skeletal muscle cells (Gao et al., 2002) . Recombinant AAV1 (rAAV1) vectors have also shown robust widespread and gr eater transduction of the gray matter than rAAV2 (Passini et al., 2003; Vite et al., 2003), and offers higher transgene expression in fetal skeletal muscle than AAV2 with intramuscular administration (Bilbao et al., 2005). In a recent study by Chen et al., (2005) rAAV1 vectors were better able than rAAV2, rAAV3, rAAV4, and rAAV5 to produce -1 antitrypsin protein, detectable by ELI SA of cellular lysate, in rat cardiac endothelial and rat cardiac smooth muscle cells. Transgene expression by rAAV1 is dram atically reduced when the aortic endothelial cells are treated with sialidase prior to infection with rAAV1, suggesting a dependence on sialic acid for the transduction of the cells by AAV1


52 (Chen et al., 2005). Transgene expression wa s not significantly reduced by preincubation of rAAV1 with soluble heparin as seen with rAAV2 and HeLa cells (Rabinozitz et al., 2002). In a simila r assay with rAAV1 a 49% reduction in transgene expression was detected when the vector was pre-incubated with soluble heparin prior to infection of simian fibroblast (COS) cells (Hauck and Xiao, 2003). Production of rAAV vectors that invo lves packaging of AAV vector genome, including a transgene inserted between A AV2 ITRs, into capsid of another serotype creates a pseudotyped rAAV (Rabinowitz et al., 2002). The transduction efficiency of pseudotyped vector s with the AAV1 capsid protein shell was not affected by the absence of hepar in sulfate proteoglucan on Chinese Hampster Ovarian (CHO) and was the bes t, when compared with serotypes 2-5, in the transduction of liver and muscle cells (Rabinowitz et al., 2002). These vectors (rAAV2/1) are able to transduce mu rine islet cells at greater efficiency than either rAAV2/2 or rAAV2/5 vector s (Zhang et al., 2005). Thus, studies comparing rAAV1 and pseudo-typed rAAV2 /1 vectors with other serotypes suggest that AAV1 is a promising candi date for gene therapy applications and, although AAV1 has a high degree of homology with AAV2, the few differences in their capsid amino acid sequence cr eate diversity between the serotypes affecting receptor binding, cell tropism. Capsid Surface Loops Associated wi th Tropism and Antigenicity An amino acid sequence alignmen t made using Clustal W (Thompson et al., shows that AAV1 is ~83% identical to AAV2, 8.83% str ongly similar, 4.48% weakly similar and 3.4% not similar, wit h a single amino acid insertion at AAV1


53 T266 (fig. 3-1). A structure-based am ino acid sequence alignment of AAV1-9, based on the atomic model of AAV2 (X ie etal., 2002) generated by crystal structures and homology-based models revealed variable surface loop regions (IIX) of the AAV capsid protein between t he highly conserved regions of the core ß-barrel motif (fig. 1-4) (Padron et al., 2005). In this analysis, the inserted AAV1 residue I266 is in divergent region I. Six of the 25 residues that are dissimilar between AAV1 and AAV2 are found in va riable loop region III and 6 others are found in variable loop region VII. Through t he course of divergent evolution the AAV capsid protein has tolerated inse rtions/deletions and amino acid substitutions that do not disrupt the proper assembly of the capsid proteins. Included in these regions are residues mapped by mutagenesis to control tropism and antigenic differences between the closely related serotypes AAV1 and AAV2 (Hauck and Xiao, 2003; Opie et al., 2002; Kern et al., 2002; Wobus et al., 2000). Monoclonal antibodies A20 and C37 can detect intact AAV2 capsid, but do not detect AAV1 by ELISA (Wistuba et al., 1997; Wobus et al., 2000). These antibodies were both found to neutralize AA V2 infection, but C37 was found to specifically inhibit cell bi nding, while A20 did not. Epitope mapping by peptide scanning (fig. 1-5) was used to identify c apsid domains involved in capsid binding for these and other monoclonal anti bodies. The epitope mapped for C37 contained residues close to divergent surface loops III, IV and V, with no divergence between AAV1 and AAV2 in the epitope contained in divergent region


Figure 3-1. Sequence alignment of t he AAV1 and AAV2 capsid protein amino acid sequence. The numbering is for AAV1, for AAV2 the numbering is n-1. Red letters represent i dentical residues, Blue are strongly similar, green are weakly similar, and black is not similar. The region from 213 to 423 has been shown to determine tropism in AAV1. Divergent regions with less than 34% identity are inderli ned. Divergent regions defined by Padron et al., ( 2005) are noted as roman numerals. The program ClustalW was used (Thompson et al., 2004).




56 V. The epitopes mapped for A20 were cont ained in divergent loop regions I, II, V and VII, with no divergence between AAV1 and AAV2 in the epitope contained in loop I. Insertion mutagenesis at re sidue 266 in AAV2 was found to alter the antigenic properties of the partially defective mutant viruses making them undetectable by ELISA with A20 antibody (Wu et al., 2000). This observation is consistant with the location of this r egion within the mapped epitope region for A20, contained in divergent loop I. AAVs are thought to infect cells by te thering to host-cell surface receptor, and then a secondary receptor is requir ed for the uptake of the virus by endocytosis. AAV1 may possibly utilize sialic acid for infection and does not appear to require heparin for cell trans duction (Chen et al., 2005). Mutagenesis studies have shown that replacing AAV2 capsid protein sequence from 213-423 with that of AAV1 slightly reduces the ability of heparin to inhibit cell transduction of AAV2 while increasing the ability of t he mutant to transduce muscle cell by 8fold when compared to AAV2 (Hauck and Xiao, 2003). T he complimentary mutant created by replacing AAV1 resi dues 213-423 with that of AAV2, reduces the ability of AAV1 to transduce muscle cells by 60%, thus demonstrating the significance of these regions in determini ng cell tropism. These findings support correlation between a reduction in heparin binding and altered tissue tropism In this study, a recombinant adenovir us was constructed to express the capsid proteins of AAV1 in HEK293 ce lls. The expressed VLPs were used to solve the structure of AAV1 by cryo-micr oscopy and 3D reconstruction to 17 Ã… resolution. Using the X-ray crystal stru cture of the AAV2 (X ie et al., 2002) VP3


57 as a template, a homology-based pseudo-atomic model was constructed for AAV1. This model was used to scale the volume of the cryo-EM density map for AAV1. Our goal was to compare the primar y sequence and structural differences between AAV1 and AAV2 in order to gain insight into the chemical basis for the diversity seen between the serotypes in receptor recognition and antigenicity. We compared the pseudo-atomic model fo r AAV1 with the atomic model for AAV2 to provide insight into residues that are structurally different between the serotypes. The pseudo-atomic model of AAV1 will be useful for future mutagenesis studies aimed at altering cellul ar tropism of the virus in order to develop targeted gene delivery systems. Results Construction of Expression Vector for AAV1 Capsid Protein In order to study the st ructure of the capsid pr otein of AAV1, milligram amounts of the protein were needed. We constructed a recombinant adenovirus that was able to express AAV1 capsid prot eins in HEK293 cells (fig. 3-2). The three major capsid proteins, VP1, VP2 and VP3 were expressed in HEK 293 cells from the single cap ORF. We were able to detect the presence of the capsid proteins by western blot anal ysis using B1 antibody that recognized denatured capsid protein (fig. 3-2C). The expressed proteins (VLPs) were purified using the techniques from Zolotukhin at al., (2002), but the pH was adjusted to 7.5 to reduce aggregation of the particles. Negative stain electron-mi croscopy was used to determine that the capsid proteins had assembled into the icosahedral VLPs (fig. 3-2D).


58 Figure 3-2. Construction of adenovirus expr essing AAV1 capsid protein. A) A HindIII restriction fragment from pl asmid pXYZ1 which includes the AAV1 capsid open reading frame A) was used to generate a recombinant plasmid B) with t he capsid gene between the CMV promoter and the SV40 polyadeny lation signal followed by the adenovirus genes. The R-ITR and L-ITR regions of the shuttle plasmid are short inverted terminal repeats (Ri ght and Left) which participate in the replication of the viral DNA. A recombination event in bacteria BJ5183 cells between a shuttle vector containing the capsid gene and a second plasmid containing the E1-E3 deleted Ad5 genome resulted in the generation the recombinant pl asmid. The recombinant plasmid was linearized with Pac1 and transfected into HEK 293 cells to produce the adenovirus expression vect or. C) Infection of HEK 293 cells with the adenovirus resulted in the expression of all three capsid proteins, which were detected by western blot analysis using the B1 antibody. D) Assembled virus-like particles were visualized by negative stain electron microscopy at a magnification of 67K.


59 Crystallization of AAV1 VLPs The purity of the VLPs (concentrat ed to 7.0 mg/ml) was demonstrated by small crystal formation in the followin g screened conditions: 1.5% polyethylene glycol (PEG) 8000, 20 mM MgCl2 20, at 25º C and 4º C. Cryo-EM Structure of AAV1 VLPs A cryo-EM reconstruction of AAV1 was obtained from micr ographs of frozen VLPs expressed in HEK293 cells. A to tal of 983 2D particle images of known centers and orientations fr om 15 micrographs were used to reconstruct the final 3D image to a 17.6 Å resolution. The AAV5 cryo-EM structure was used as the original search model. The AAV1 st ructure shows the surface features associated with parvovirus capsids, in cluding three mound-like protrusions surrounding the icosahedral threefold axes and depressions at the twoand fivefold axes. The cryo-EM map was super imposed with the icos ahedral model of the capsid made of 60 monomers in order to determine the approximate volume of the reconstruction (fig. 3-3). Threedimensional image reconstruction of AAV2 and AAV4 at 13 Å resolution (Padron et al., 2005) and the structure of AAV5 (Walters et al., 2004) at a 16 Å resolution were compared to that of AAV1 and to AAV2 calculated to 18 Å (fig. 3-4). The protrusion surrounding the fivefold pore is more pronounced and longer in AAV4 than in AAV1, 2 and 5, with AAV5 having the least pronounced, but the depression that surrounds the fivefold pr otrusion appears to be conserved for the four serotypes (fig 3-4). The twofold depression is divergent for the serotypes.


60 Figure 3-3. The pseudo-atomic model of AAV1. A) The cryo-EM reconstruction of AAV1 to 17.6 Å resolution was s uperimposed with 60 copies of the VP3 pseudo-atomic model for AAV1 generated, based on the AAV2 crystal structure. B) The monomer model of AAV1 in red superimposed with the atomic models bas ed on crystal structure of AAV4 (purple), CPV (green) and AAV2 (bl ue). Some of the ß-ribbons are labeled and the regions of sequenc e divergence are noted. The programs O (Jones et al., 1991) and Chimera (Petterson et al., 2004) were used to generate the figures.


61 In AAV4 the depression is a valley that runs longitudinally between the threefold axes of symmetry. The twofold depression in AAV1, AAV2 and AAV5 runs longitudinally between fivefold axes, but is greatly diminished in AAV5. The wall of the twofold dimple/base of t he threefold mound differs between AAV4 and the other serotypes with additional density in AAV4 altering the structure of the valley. The region between the twofol d depression and the threefold mound varies between AAV1 and AAV2 forming a broader base to the mounds in AAV1. The divergence of the cryo-EM structur e of AAV4 from AAV1, AAV2 and AAV5 at the twoand fivefold axes places AAV4 as the most structurally distinct of the serotypes. The most variation in structure between the AAV serotypes is at the mounds that surround the threefold axis of symmetry (fig 3-5). These protrusions account for the largest radial distance of the virus structure. In all of the AAV reconstructions this protruding region includes a base-like stage on which the three mounds sit, with valleys runn ing between each mound and a single depression at the cent er of the stage. The threefold protrusions characteristic of adeno-associated viruses are more pointed in AAV1 and AAV2 and rounded in AAV4 and AAV5. The size of the mounds is severely diminished in AAV5 due to a deletion within divergent loop region III in AAV5 (Walter et al., 2004). The tips of the threefold spikes ar e even more finger-like in AAV2 than in AAV1. In both AAV1 and AAV2 there is a second smaller protruding mound surrounding the threefold axis of symmetr y which appears to narrow the valley running between the larger spiked protrusi ons. This region is more pronounced


62 Figure 3-4. Cryo-EM reconstructions for AAV capsids. A) and E) are reconstruction of AAV2, B) AAV4 and C) AAV5, D) AAV1. Resolutions and axes of symmetry are noted. D. 2) and E.2) are close-up views down the twofold axis of rotational symmetry, D.3) and E.3) threefold and D.4) and E.4) fivefold. Reconstructed images were visualized and constructed using the Purdue suit e of programs. A)-C) are adapted from Padron et al. (2005), and th e figures were generated in RobEM (Baker et al., 1999).


63 in the AAV2 reconstruction than the AAV1 reconstruction and is absent, or fused with the larger protrusions in AAV4 and AAV 5 (Padron et al., 2005). These four serotypes can be placed in two distinct gr oups with respect to t he structure of the threefold region. Group1 includes AAV1 and AAV2, displaying fingerlike protrusions and a smaller second mound, and a more variable group2 includes AAV4 and AAV5 displaying a single, mound-like protrusion (fig 3-5). Antibodies to Detect Intact AAV1 Capsids The Adenovirus expressed capsids were used to generate monoclonal antibodies specific for AAV1 intact capsids. This work was done in collaboration with Dr. Colin Parrish at Cornell University in Ithaca, New York. Eight antibodies were determined by ELISA to be specific for intact AAV1 and do not react with denatured AAV1 capsid protein. These antibodies are AA7C8, AA9A8, AA7E6, AA4E4, AA8B2, AA2H2, AA5B1, and AA5H7. The hybridoma supernatants of three IgG clones AA4E4.G7, AA9A8.B12 and AA5H7.D11 are currently being characterized for serotype specificity. Discussion Correlations between Structure Differences and Tropism The greatest variation in primary amino acid sequence in VP3 between AAV1 and AAV2 lies within variable loop regions I, III and VII described in Padron et al., (2005), and in the vari able loop containing the mapped AAV2 heparin binding residues R585 and R588 (O pie et al., 2002; Kern et al., 2002; Wu et al., 2000), which all surround the th reefold axes of symmetry (fig. 4-1). The heparin binding loop does not appear to tolerate, in evolution of the


64 Figure 3-5. Detail of threefold view of cryo-EM reconstructions of AAVs. Serotype names and resolution of rec onstruction are shown. A) shows the threefold detailed of the AAV1 re construction. B) The large (3L) and small (3S) protrusions surrounding t he threefold axis characteristic of AAV2 are labeled C) Axes of sy mmetry are labeled, the threefold with a filled triangle, the twofold with a filled oval, and regions of structural divergence are labeled (Padr on et al., 2005). D) is the cryoEM map for AAV4 and E) is for AAV5. The mapped heparin binding region for AAV2 is noted with black arrow.


65 serotypes AAV1-9 (fig. 1-4), deletions or insertions as seen in the variable regions defined in Padron et al., (2005), but varies significantly even between closely related serotypes AAV1 and AAV2 with no identity from residues 585-590 (fig. 3-1). This portion of the l oop is negatively charged in AAV2 with the sequence RGNRQ. In AAV1, which does not bind heparin as strongly as AAV2 (Opie et al., 2002; Hauck and Xiao, 2002), this region of the loop has no basic residues but an acidic residue with the sequence SSSTD. The outermost tips of the threefol d mound (3L) are formed by variable region III (Padron et al., 2004) (fig. 3-4B and 3-5C). Variable regions III and the loop that contains AAV2 basic residues R585 and R588 surround the region containing variable region VI from a th reefold symmetry-related monomer to create the threefold mounds (fig. 3-5). In the cryo -EM map of AAV2 and the higher resolution X-ray structure there is second projection (3S) from the mound toward the threefold axis (Kronenberg et al., 2001; Xie et al., 2002; Padron et al., 2005). The 3S point includes the heparin bi nding residues R585 and R588 (Xie et al., 2002; Opie et al., 2002; Kern et al., 2002; Wu et al., 2000). At the threefold axis, symmetry-relat ed monomers position variable regions I, II, IV, and V immediately adjacent to the basic patch of AAV2 (fig. 3-5). This basic patch is formed when two threefold-re lated monomers inter-digitate to bring residues R484, R487, and K532, from one monomer close to residues R585 and R588 in an adjacent monomer. The valle y that runs between threefold mounds contains R532, R484, and R487, cons erved in AAV1, which are known to contribute to heparin binding in AAV2, alth ough, to a lesser extent than R585 and


66 R588. These residues may account for t he lower affinity binding of AAV1 to heparin (Opie et al., 2002; Hauck and Xiao, 2003). The AAV2 amino acid residues R585 and R588 hover above the valle y, projecting the positive charge toward the solvent from the declining face of the 3S protrusion. This creates a deeper, more enclosed valley below the spike in AAV2 than in AAV1 (fig. 3-5). The 3S protrusion is smoothed out in AAV1 with a more gradual decline toward the threefold axis (fig. 35). AAV1 also lacks the basic residues R585 and R588 that are present at this pr otrusion in AAV2 (Xie et al., 2002). The structural and charge differences between AAV1 and AAV 2, surrounding the threefold axis, control the affinity of t he viruses to bind heparin. Mapping the mutational analysis by Hauck and Xiao (2003) to the sequence alignment from Padron et al. (2005) rev eals that variable surface loop regions I and II may effect muscle cell transduction in AAV1 (Hauck and Xiao, 2003). This sequence includes the insertion at 266 for AAV1 within divergent loop I, and includes loops II, with only a single c onservative substitution I373 in AAV1 and V372 in AAV2 (Padron et al., 2005). R egions I and II cluster together forming the small protrusion lying bet ween threefold spikes surrounding the threefold axis of symmetry (fig. 3-5). This protrusion is characterist ic of the group1 structures, forming a wall, opposite to the heparin binding 3S protrusion, that lining and contouring the valley containing R 484, R487, and K532 between threefold mounds. There is structural variability of loop I as seen by superimposing the cchain of the pseudo-atomic model of AAV1 with the atomic model from the crystal structures of AAV2, AAV4 and CPV (fig. 3-3) (Xie et al., 2002;


67 Govindasamy, E. Padron, N. Kaludov, R. McKenna, N. Muzyczka, J. Chiorini, M. Agbandje-McKenna, unpublished re sults; Xie et al., 1996). The model predicts a change in conformation of loop I between A AV1 and AAV2. This is consistent with the difference seen when comparing the cryo-EM density maps of AAV1 and AAV2 at this region suggesting a role for this protrusion in tropism determination in AAV1 (fig 3-5). Higher resolution dat a would be needed to conclusively state that the envelope has diffe rences at the region. Correlations between Structure Differences and Antigenicity The antigenic sites of the AAV2 neutraliz ing antibodies, that do not detect AAV1 by ELISA, C37 and A20 (fig. 15), have been mapped to the divergent surface loops described in Padron et al., (2005) (Wobus et al., 2000). These antibodies to not detect denatured AAV2 caps id protein, but do detect intact capsids. The C37 epitope 1 (C37-1), amino acids 461-470, shows a high degree of divergence (70%) in amino acid sequence between AAV1 and AAV2. This epitope includes the C-terminal portion of divergent region III, while epitope C372, residues 493-503 lies upstr eam of the N-terminal por tion of divergent region IV. Epitope C37-3, residues 601-610 does not diverge in amino acid sequence between AAV1 and AAV2 and does not fall wit hin any of the described divergent loop regions. This suggests that the stru ctural differences seen in the cryo-EM reconstructions of AAV1 and AAV2 at the tips of the threefold spikes correlates with the amino acid sequence dive rgence between AAV1 and AAV2 and the antigenic differences between the se rotypes mapped to this region. The epitope for the A20 antibody has also been determined. This antibody neutralized the infection by AAV2 but does not inhibit cell attachment. Heparin


68 sulfate binding by the AAV2 capsid is known to facilitate cell attachment and therefore may not be disrupted by bound A20 antibody. Epitope A20-1, from residues 272-281, is identical in sequence to between AAV1 and AAV2, A20-2, from residues 369-378, 10%, and A20-3, from residues 533-542, and A20-4 from residues 566-575 are 20% divergent between AAV2 and AAV1. A20-2 and A203 lie within a divergent surface loop, II and V respectively, and A20-4 lies between the heparin binding loop and regi on VII, which forms the threefold mound from two threefold symmetry-related monomers. This suggests that there are other capsid surface regions that differ between these two serotypes that do not lie in the heparin binding region. Peptides A20-1 and A20-2 alone can both inhibit binding of A20 by ELISA and not AAV1, are also part of the region determined by Hauck and Xiao (2003) to determine AAV1 tropism and are not part of the heparin bindi ng region. These residues are at the depression surrounding the fivefold pore, a region know n to influence infectivity in stages of the virus life cycle down stream of ce ll attachment. Our findings support he hypothesis that the AAV seroty pes can utilize different regions of the capsid to interacting with the host cell.


69 CHAPTER 4 COMPLEX OF AAV2 AND HEPARIN OLIGOSACCHARIDE Introduction HSPG is the Only Known Host-Cell Receptor Required for AAV2 Cell Entry Heparin sulfate proteoglycan (HSPG) was the first cell surface receptor found to mediate AAV2 transduction by Summerford and Samulski (1998) in studies that showed a dose-dependent i nhibition of AAV2 transduction detected in competition studies with soluble heparin . They further determined that cell attachment and transduction by AAV2 were si gnificantly inhibited with CHO cells lacking HSPG, and with heparin enzymatically removed from surface of HeLa cell, while similar inhibition was not detected with the enzym atic removal of chondroitin sulfate (Summerford and Sa mulski, 1998). Their findings clearly demonstrated the ability of AAV2 to utilize HSPG for cell attachment, but did not address factors involved in viral cell entry. HSPG is an integral component of the extra-cellular matrix (ECM) and many cell surfaces which might account for the broad tropism and poor spreading from in fection site of AAV2. The direct binding to heparin sulfate is currently ex ploited in AAV2 purification by affinity chromatography (Zolotukhin et al., 1999). Fibroblast growth factor receptor (FGFr) is a receptor tyrosine kinase (RTK) requiring dimerization in FGF-mediated cell signaling and endocytosis, requiring heparan sulfate proteoglycans for dimerization (Rapraeger et al., 1991; Yayon et al., 1991; Ornitz et al., 1992; Spivak-Kroizm an et al., 1994). A model for receptor


70 dimerization, based upon the crystal st ructure of FGF2 bound to the ligand binding domain (D2–D3) of FGFR1, explai ns the interaction of FGF and heparin for the induction of FGFR dimerization and activation (Plotnikov et al., 1999). The effects of HSPG, FGF and FG Fr on AAV2 cell attachment and transgene expression were studied to deter mine if these molecules are involved in AAV2 cell entry (Qing et al., 1999) . Non-permissive MO7e cells stably transfected with plasmid expressing HSPG alone or with HSPG and FGFr were both found to equally permit rAAV2 binding and significant transgene expression, while Raji cells that do not normally expr ess HSPG or FGFr required transfection with plasmids expressing both molecules while conferring only low levels of transgene expression (Qing et al., 1999). These levels of gene expression were not affected by FGFr kinase activity inhi bitors, while competitive inhibition studies with soluble growth factors demonstrated th at FGF, but not EGF, inhibited AAV2 cell binding (Qing et al., 1999). Taken t ogether, these findings demonstrate that FGFr and HSPG, in concert, have an affe ct on rAAV2 cell binding and transgene expression, but activity of the RTK is not required for cell entry of the virus and transgene expression, suggesting yet a di fferent mechanism for activating the endocytic uptake of AAV2. Integrin receptors have many roles incl uding activation of internalization, ECM degradation, and signal transduction and are also known to facilitate virus internalization for several non-envelope d viruses including adenovirus (Dedhar and Hannigan, 1996; Bergelson et al., 1997) . A physical association between AAV2 and Vß5 integrin was determined by co -immunoprecipitation studies,


71 although the presence of an intermediaryadapter molecule to facilitate this interaction was not investigated (Su mmerford et al., 1999). Cells lacking Vß5 integrin were not able to internalize AAV2, while cells expressing Vß5 integrin on their surface were, demonstrating t he poteintial for c ooperativity between multiple surface receptors to transmi t the cell attachment signal into an internalization event (Summerf ord et al., 1999). Inhibiting Vß5 integrin with antibodies was also shown to block t he endocytosis of AAV2, but did not block cell binding, although purified Vß5 integrin was not able to directly bind to AAV2 (Sanglioglu et al., 2000; Qiu and Brown, 1999). Taken together, the previous st udies of AAV2 cell attachment and internalization support a model where AAV2 can tether to the host cell by direct binding to HSPG. The virus then relie s on HSPG to bind FGFr and associate with Vß5 integrin, possibly through cooperative binding of fibronectin, which is known to bind both HSPG and integrins while activating the integrin receptor (Mayano et al., 1999). The formation of this receptor-signaling complex is proposed to activate the cell, inducing re ceptor mediated endocyt osis of the virus via clathrin-coated pits in an event requiring dynamin, a 100-kDa cytosolic GTPase that selectively regulates clat hrin-mediated endocytosis (Duan et al., 1999; Bartlett et al., 2000). Thus, if direct binding is required to define a “receptor” then the current body of literat ure only supports HSPG as a receptor for cell attachment to faci litate AAV2 internalization. Inhibition of AAV2 cell attachment is therefore inhi biting virus-heparin binding.


72 Mapped Heparin Binding Region HSPG appears to function primarily in attachment of the AAV2 and 3 serotypes to the cell surface (Summe rford and Samulski, 1998). The basic amino acids in AAV2 that are required fo r binding heparin sulfate were identified by mutagenesis to include R585 and R588 and to utilize to a lesser extent R484, R487, and K532 (Wu et al., 2000; Opie et al., 2003; Kern, 2003). Mutagenesis analysis also determined that acidic resi dues D561-E564 were also shown to be required for heparin binding, although point mutation analysis has not been conducted to verify the nature and the extent of the contribution of the four acidic residues to heparin binding (Wu et al., 2000) . These residues are all clustered together by two threefold symmetry re lated monomers, lin ing a valley-like channel that runs between the threefold spik es of the AAV2 capsid (fig. 4-1) (Xie et al., 2002). Residues R585 and R588 form the most distal portion of the heparin binding region serving to cap off the valley. These two residues together are brought to the other residues of t he heparin binding region by a threefold symmetry-related monomer, therefore inte r-monomeric interact ions at the intratrimeric interfaces are required for heparin binding (Xie et al., 2002). Mapped Epitopes for Neutralizing Antibodies A20 and D37 Monoclonal antibodies A20 and C37 t hat detect intact capsids were assayed for AAV1-5 binding in a study by Wobus et al., (2000 ). Antibody A20 was able to detect AAV2 and AAV3 by ELIS A, and to neutralize AAV2 infection, but did not inhibit cell attachment, while antibody C37 was specific for AAV2 and


73 Figure 4-1. Basic residues surrounding AAV2 heparin binding region mapped by mutagenesis. Measuring referenc es are in angstr oms. Ribbon representation of the atomic model (Xie et al., 2002) of the capsid protein trimer created by applying a threefold ma trix to the VP3 model of the crystal structur e (1a.). Indi vidual VP3 monomer s are noted by color with space-filled representation of amino acids from mutagenesis studies (Opie et al., 2003; Kern et al., 2003). Argenine residues are in blue, lysine residues in light blue a nd histidine residues in pink. The threefold axis of symmetry is labeled with . Close-up of stick-and-ball representation of basic residues fr om mutagenesis studies (1b.).


74 neutralize infection by inhibiting cell atta chment (Wobus et al., 2000). This suggests that the footprint for A20 does not cover residues R585 and R588 of the heparin binding region and the footprin t for C37 does. This was further evidenced by mutagenesis data showing t hat an insertion at R587 inhibited detection by C37 and not A20 (Wobus et al., 2000). Structure of the AAV2 Capsid The major contributing epitope for C 37 and A20 have been mapped on the crystal structure of AAV2 (Xie et al ., 2002) (fig. 1-5). The epitope for C37 appears to cover the threefold mounds and the cleft surrounding the threefold axis. It is possible at a single FAb co vers the threefold axis blocking cell attachment in C37 neutralization. The A20 footprint appears to interact with the depression surrounding the fivefold pore. A similar interaction is seen in the cryo-EM structure of Human Rhinov irus 14 (HRV14) bound to a neutralizing antibody (Che et al., 1998). They proposed that bivalent bindi ng of the antibody around the fivefold and across the twofold axis inhibited infection by stabilizing the virus capsid, suggesting that capsid desta bilization was required for infection. The atomic model based on the X-ray cr ystal structure of AAV2 to 3 Å has also been used to visualize the cleft lined by the mapped heparin binding residue (fig. 4-1) (Xie et al., 20 02; Kern et al., 2003; Opie et al., 2003). The cryo-EM density envelope map (Kronenberg et al., 2001; Padron et al., 2005) can be used in conjunction with the atomic model to correlate the general topography of the capsid with the residues affecting the biological characteristics of the virus (Padron et al., 2005). The core of the AAV2 capsid is a ß-barrel made up of


75 strands H-I, with insertion loops named for t he ß-stands that flank them. This is a common arrangement seen for spherical ssDNA viruses. Genome-Dependent Icosahedral Capsid Assembly Hypothesis It has been proposed that nucleic acid density running along the interior groove at the twofold axis of icosahedral capsids in non-enveloped viruses may serve as an adhesive growth template, supporting a model of genome dependant assembly where pentameric building blocks of capsid protein assemble into the icosahedron via twofold junctions (Zlo tnick, 2004, Rudnick and Bruinsma, 2005). Ordered nucleic acid has also been seen at the inner face of the threefold axis and lining the twofold interface in ssDNA parvoviruses, MVMi and AAV4 (Agbandje-McKenna et al., 1998; Padron et al., 2005). Mutations in the MVM capsid amino acids at the twofold interf ace, that were most closely associated with the ordered DNA nucleotides, reduce in fectivity, capsid assembly, and VP1 externalzation (Reguera et al., 2005). In this study the structure of AAV2 complexed with a 6 kDa heparin oligosaccharide was determined to 18 Å using cryo-electron microscopy and image reconstruction. A difference map calculated by subtracting the density map of the uncomplexed AAV2, at a comparable resolution, from the AAV2~heparin complex showed the binding site of the receptor molecule at the icosahedral threefold axes of the capsid. The available crystal structures of AAV2 and that of heparin oligosacc haride were used to model heparin, constrained by the difference map, into the basic cleft of the mapped heparin binding region. A conformational change in the HI loop structure was also modeled. This conformational change affe cted the opening of the pore region at


76 the icosahedral fivefold axis in respons e to heparin binding. In addition, ordered density within the interior of AAV2 caps id, in the shape of a dodecahedron, is consistent with nucleic acid adjacent to basic and hydrophobic residues that are the general hallmark of dsDNA binding proteins. Results Density Maps The refined reconstructions of AAV2 and AAV2~heparin (fig. 4-2) were visualized in ROBEM (Baker et al., 1999) with both maps having the typical features of AAV including, threefold spikes, twofold depressions, and fivefold pore surrounded by a fivefold depression. Using the Purdue program suite (Baker at al., 1999), the complex map wa s reconstructed from 1510 selected particles from 10 micrographs and the re solution was determined to be to 18 Å. The uncomplexed map was reconstructed from 2045 selected particles from 11 micrographs and the resolution was to 13 Å (Padron et al., 2005), and was back calculated to 18 Å for comparison wit h the AAV2~heparin complex map. There is additional density in the m ap of the complex along the channel between threefold spikes toward the twof old axis, at the m apped heparin binding region. The complex map also has addition al density at the five-fold axis and in the depression surrounding the protrusion at the pore. The vertices of the regions that line the five-fold pore appear to have rotated with respect to the uncomplexed map by about 36º. A comparison of cross sections of the maps (fig. 4-3) shows that the openi ng of the five-fold pore is broader in the complex. This cross section comparison also shows that there are differences between the two reconstructions with respect to the inner surface of the capsid.


77 Figure 4-2. Three-dimensional recons tructions of adenovirus expressed AAV2 particles and the AAV2~heparin comp lex. Figures A-D) show the uncomplexed AAV2 cryo-EM density map and figures E-H) show the cryo-EM density map of the AAV2~hepar in complex. Figures A), B), E), and F) show a view down the twofol d axis which is labeled in B). C) and G) show the threefold view wh ich is labeled in C). D) and H) show the fivefold view. Some of the fivefold vertices lining the outermost region of the fivefold por e are marked with a black dot to visualize the 36º rotation diffe rence, between the two maps, surrounding the fivefold pore. These images were generated in RobEM (Baker et al., 1999).


78 Figure 4-3. Equatorial secti ons of the reconstructions. A) The axes of symmetry are marked in black numbers and so me of the capsid features are labeled. This image was generated in RobEM (Baker et al., 1999). Close-up of the equatorial section of t he fivefold pore. The blue image is the uncomplexed map and the r ed is the AAV2~heparin vomplex map. The diameters at the outermost tips of the pore opening were measured in Chimera (Petterson et al., 2004).


79 protein shell. The difference map was visualized using the program Chimera (Petterson et al., 2004), at a contour leve l ranging between 1.9 an d 2.1 (fig. 4-4) Modeling Heparin into Density at the Heparin Binding Region The difference density near the basic patch of the mapped heparin binding region (R585, R588, R484, R487, and K532) (Wu et al., 2000; Opie et al., 2003; Kern et al., 2003) runs from the threefold to the twofol d axes of symmetry (fig. 45A). An oligosaccharide helix from the crystal structure of FGF complexed with a heparin hexasaccharide fragment, in t he PDB, (Accession # 1BFC) (Faham et al., 1996) (fig. 4-5B and C) was used to in terpret the volume of the difference map in this region. The modeled olig osaccharide was superimposed onto to the atomic model of the AAV2 VP3 protein (X ie et al., 2002). The atomic model of the heparin oligosaccharide was fitted into the volume of the difference map at the mapped heparin binding region to allo w for potential interactions between heparin and the residues of the basic patch . To fill the volume of difference density extending toward the twofold axis , two additional hexasaccharides were docked into the map. Modeling Fivefold HI Loop changes A surface HI-loop change was modeled into the difference density at the depression surrounding the fivefold axis of symmetry. A comparison of positive and negative difference density maps suggest ed that the surface HI loop which lies in this depression surrounding the fivefold pore undergoes a conformational change (fig. 4-6A). The loop change in the complex suggests a conformation switching, of this loop in response to heparin binding. The loop was modeled into


80 Figure 4-4. Cross section of the diffe rence map superimposed onto the atomic model of the AAV2 capsid. T he difference density map (red) was made by subtracting the uncomplexed AAV2 map from that of the the AAV2~heparin complex. A 60 fold matrix was applied to the VP3 model of the crystal structure to ma ke a complete icosahedron (yellow ribbon) (Xie et al., 2002). Differ ence map includes an outer layer, which includes additional densit y for heparin and capsid protein surface changes, and an internal dodecahedral cage surrounded by the protein capsid shell.


81 Figure 4-5. Atomic model describing AAV2 ~heparin complex. Threefold axes ( ) and twofold axes ( 2 ) are labeled. A) Difference map was cropped and superimposed onto the atomic model of the AAV2 VP3 trimer (Xie et al., 2002). B) The basic am ino acids surrounding the mapped heparin binding region are space-fill ed. The atomic model for heparin oligosaccharide was taken from the coordinates of the crystal structure of the FGF~heparin complex (Faham et al., 1996). A heparin oligosaccharide of nine disacchari des was modeled into the difference density and a threefold symmetry wa s applied to make the symmetry related oligosaccharides models. C) Ribbon model of the VP3 trimer with heparin binding amino acids space-filled.


82 the difference density (fig. 4-6B) and the fivefold arrangement was visualized in Chimera (Petterson et al., 2004) (fig. 4-6C ). The diameter of the outermost density of the fivefold pore for the unc omplexed map was measured at ~15Ã… and the diameter of the complex was ~ 24Ã…. The results obtained from the AAV2~heparin complex cryo-EM map are similar to those from the cryo-EM structure to 22 Ã… resolution of parvovirus B19 complexed to its cellular receptor globoside, where conformational changes at the fivefold axis may also occur in response to carbohydrate binding at the threefold axis (Chi pman et al., 1996). Ordered Nucleic Acid Inside the Capsid Shell The interior of the difference map forms a dodecahedron similar to that seen for the ssRNA viruses (Zlotnick, 2004, Rudnick and Bruinsma, 2005). This network lines the interior threefold and tw ofold interfaces between capsid protein monomers. We superimposed the AAV2 dodecahedron cage onto a density map of interior of the cryo-EM structure of AAV4 (fig. 4-7A) and with the atomic model of nucleotides associated with the capsid of MVM from the crystal structure (Padron et al., 2005; Agbangje-McKenna et al ., 1998) (fig. 4-7B). The difference density superimposed with density found fo r both the AAV4 and MVM structures, illustrating the presence of a homologo us structure inside these other parvoviruses. A nucleic acid protection assay discuss ed in Chapter 2 indicated that the genetic material contained in our VL P capsids was purified as dsDNA oligonucleotides (fig. 4-8). We pu rified enough dsDNA to equal the mass of approximately a few hundred basepairs per caps id with the most prominent size DNA oligomers, 100 to 200 basepai rs (approximately less than 1012 viral


83 Figure 4-6. Fivefold HI loop conforma tional change modeled into the difference density. Positive (red) and negative (cyan) difference density maps superimposed onto a density map of the capsid based on the atomic model for AAV2 rendered to 18Ã… A) . Loop change from yellow to blue is modeled into the difference density in red mesh B) and the fivefold relationship is depicted C) to illust rate the change around the pore. Residues of the HI l oop are labeled for reference.


84 Figure 4-7. Difference density associated wit h the interior face of the parvovirus capsid shell. A) The interior r egion of the difference density, the red mesh dodecahedron, was superimpos ed onto a density map of the interior portion of the cryo-EM ma p of AAV4 (Padron et al., 2005). B) The difference density was then s uperimposed with the atomic model of ordered nucleotides, in cyan, a ssociated with the interior capsid surface of MVM based on the crystal structure of MVM (AgbandjeMcKenna, 1998). Theses image s were generated in Chimera (Petterson et al., 2004).


85 genome equivalents, of 4700 kb or 2350 bp, from approximately 1013 particles. The conclusion is that the AAV2 VLPs may be packaging genomic DNA that has been sheered due to adenovirus-mediated apoptosis. Discussion Acidic and Basic Residues of the Heparin Binding Region The heparin binding sites have been mapped by mutagenesis and have been mapped on the atomic mode l from the crystal struct ure (Kern et al., 2003; Opie et al., 2003; Xie et al., 2002). The residues involved come together from two different threefold symmetry-related monomers in the 3D crystal structure (Xie et al., 2002). Residues R585 and R588, the strong heparin binding residues, are contributed together on a single monomer and R484, R487 and K532, the weaker heparin binding resi dues, are contributed by the symmetry related monomer. Together theses residues form a basic patch in the shape of a cleft with R585 and R588 forming the wall and most distal, cap-like portion of the valley. Residues R484 and R487 are in the valley of the cleft below the cap and the wall. The other basic residue K532 lines the region of this valley running toward the twofold axis. There is an acidic patch at the base of the threefold spike forming a portion of the wall that lines the twofold depre ssion. This patch includes loop D561E564. Alanine substitutions of D561-E 564 produce non-infectious viruses that do not bind heparin while, alanine substitutions at R585 and R588 do not bind heparin, but are only partially defective fo r infectivity (Wu et al., 2000). The acidic patch includes another acidic l oop, D528-E531 interacting with loop D561-


86 Figure 4-8. Nucleic acid contained in the virus-like protein capsid shells visualized on an agarose gel with ethi dium bromide. Lane A is the molecular weight marker with the 500 bp band labeled. The sample in lane B was untreated nucleic acid extr acted from the expressed VLPs. Lane C contains sample following tr eatment with RNAse, and lane D contains sample following treatment with DNAse.


87 E564 through solvent-mediated hydrogen bo nds. In the crystal structure residues E531, D562, E563 and E564 appear to interact with the solvent together linking the two acidic loops (Xie et al., 2002) Not all of the basic residues known to effect heparin binding appear to make direct contact with the heparin mole cule, but instead are participating in charge interactions with acidic loops D561-E564 and D528-E531 (f ig 4-9). These acidic loops are near the twofold ax is, and along with residues D494, D514 and E574, they interact with R487, K507, H 509, K527, and K532. For example, nitrogen atoms from both R487 and K527 are within water-mediated hydrogen bonding distances (< 5Ã…) from the oxygen atoms of E574 and E530. This arrangement allows for each atom to participate in two water-mediated hydrogen bonds forming a closed ring of associat ion between these four amino acids, which in turn coordinates the architectu re of the heparin binding cleft valley with the cleft wall. This arrangement might c hange in response to heparin binding. The basic residues R484, R585 and R588, of the cleft wall and cap, are not involved in these charge interactions with the acidic patch, and are free to survey the solvent for receptor molecules. Viruses with capsid substitution mutation K507A are defective for cell binding and H509AAAA insertio n mutants do not bind heparin (Wu et al., 2000). K527A mutants have reduced heparin bi nding (Kern et al., 2003). K532A mutants have reduced heparin binding, but dramatically reduced infectivity by a factor of 5 logs (Kern et al., 2003; Opie et al., 2003).


88 Figure 4-9. Atomic model of charged AAV 2 residues in the heparin binding cleft. The atomic model is from the crys tal structure of AAV2 (Xie at al., 2000). Residues of the basic patc h are in blue and residues of the acidic loops are in orange. Oxy gen atoms are white and nitrogen atom black. The dashed lines connect oppos itely charged residues indicate residues within 5Ã… from each other, suggesting water-mediated hydrogen bonding. The residues are labeled for residue type, primary sequence position number and then chai n (A or B). This image was generated in Chimera (Petterson et al., 2004).


89 The mutagenesis data and the atomic m odel from the crystal structure implicates the twofold acidic loops in having a significant influence over AAV2 infectivity (Xie et al., 2002; Kern et al., 2003; Opie et al., 2003). Through charge interactions, the acidic loops can coor dinate the architectu re of the heparin binding and alter the architecture in res ponse to heparin binding. Our difference map shows density at the twofold axis, and although we fitted the density volume with heparin oligosaccharides, the densit y might be due to a conformational change in the capsid, in response to heparin binding. HI Loop Conformation Change The HI loop makes interactions between neighboring fivefold-related monomers while also interacting with the DE loop that lines the fivefold pore. The interaction between these two loops is at the core ߖbarrel where ßI and ßD are adjacent associated by a salt br idge between R310 and E683 (Xie et al., 2000). Mutants that include alanine substitu tions to either of these residues do not make capsids (Wu et al., 2000). We compared the positive difference map (AAV2~heparin minus AAV2) to the negative difference map (AAV2 minus AAV2~heparin) and detected a density mo vement between the maps at the depression surrounding the fivefold axis, wh ere the HI loop is lo cated (fig. 4-6). We modeled the density change to the HI loop Mutagenesis has been used to probe the f unction of the HI loops. Alanine substitution K665A was wild type and the FVFLI insertion at residue 664 within the HI loop was defective for capsid forma tion (Wu et al., 2000). Movement of the HI loop may be necessary to coordinat e the iris-like rotation of the DE loop


90 seen in our reconstruction of the AAV2 ~heparin complex (fig. 4-3) through conformational change at the core ߖbarre l motif (fig. 4-2). This change at the fivefold pore is correlated to heparin bindi ng at the threefold axis. All of the residues discussed that make up the basic and acidic patches of the heparin binding region lie within the long GH l oop, and therefore coul d influence the HI loop through changes to the ßH strand. The heparin binding signal at the threefold axis could be transmitted thr ough a monomer to the HI loop at the fivefold axis allowing for capsid destabilization and opening of the pore in response to heparin binding. In the crystal structure of AAV2 (Xie et al., 2002), in a VP3 pentamer the DE loop residues K321 and E322 (fig. 4-10) are arranged to form a ring of alternating charged residues with neighboring fivefold -related monomers. These residues are within a distance from each other, <5Å between fivefold-related monomers, suggesting a water-mediated charge interaction between them. This arrangement may serve to stabilize the DE loop holding the fivefold pore in the closed conformation. Disruption of thes e interactions may be necessary to open the pore. Mutational analysis has been done to probe the function of the DE loop. The double mutant K321A and E322A was heat sensitive, with no detectable capsid at 37º C. Capsid assembly was rescued at 32º C, allowing for the production of viruses with somewhat reduced infectivity and reduced transgene expression when compared to wild type rAAV2 (Wu et al., 2000). This suggests that these residues are important fo r capsid assembly. A single alanine

PAGE 100

91 Figure 4-10. Atomic model of an AAV2 VP3 pentamer (Xie et al., 2002) viewed from the fivefold rotation axis of symmetry. The fivefold axis of symmetry is labeled 5, The HI loops are labeled with solid arrow, and the tips of the DE loop (D327) of are labeled with dotted arrows. Residues E322 and K321 are label ed with dashed arrows. A) shows the side view of the pentamer and B) is viewed looking down the fivefold pore. This image was gener ated in Chimera (Petterson et al., 2004).

PAGE 101

92 substition at E322A showed a strong reduction in capsid assembly with undetectable genome packaging and infectivity, while a tyrosine substitution at nearby residue T324 increased the PLA2 activity of the VP1 unique region relative to wild type AAV2 (Blecker et al., 2005). We tested our expressed particles fo r PLA2 activity, with and without preincubation with heparin, and found a reduction in PLA2 activity in the presence of heparin (data not shown). Blecker et al ., (2005) report that reduction in PLA2 activity may correlate with a reduction in capsid stability. This suggests that heparin binding causes the HI loop change and destabilizes the capsid reducing the PLA2 activity. This supports the hypothesis that PLA2 activity is not necessary for cell entry and may be detriment al to virus infectivity at the cell surface. The fivefold pore of the AAV capsid has been implicated in genome packaging, PLA2 activity and infectivity (Bleker et al., 2005) and might also be involved in stages of the lifecycle of the virus that follow cellsurface attachment including secondary receptor activation, cellular trafficking to the nucleus and genome release. Expressed AAV2 Capsids Nuclease protect Non-Viral DNA To analyze the capsid protein inte ractions with the DNA density we superimposed the MVMi nucleotides into our dodecahedral cage and then superimposed the AAV2 atomic model (A gbandje-McKenna et al., 1998; Xie et al., 2002). Amino acid residues within 5 Ã… of the MVM nucleotides that interact at the two-fold interface include R238, R307, F420, W684, E685, L686 and Q687 (fig. 4-11). These residues are highly conserved in AAV serotypes 1-8 with substitutions in AAV4 (R238H, L686I) and AAV5 (Q687K) and are all found at the

PAGE 102

93 twofold interface within strands that make up the core ß-barrel motif, with the exception of F420, which is at the N-terminal region of the GH loop 6 residues away from the ßG strand. R238 is t he most N-terminal residue of the core ßbarrel and found on the first residue of ßB, R307 is on ßB, W684-Q687 are on Cterminal region of ßI. Mutations in AAV2 R238 and R307 and nearby D237, K309, R310, and E681-E683 inhibit the formation of capsids (Wu et al., 2005). At the threefold vertex, the dodec ahedron is associated with a threefold related basic, aromatic loop including re sidues H627, F628, H629, found in the C-terminal region of the GH loop, and are blocking the pore that exists at the capsid threefold vertex. A mutation at H627 produced a wild type virus, but mutations at F628, H629 have not yet been analyzed for functionality (Wu et al., 2000). The association of these residues with the internal difference density suggests that the capsid makes global dynamic changes in response to heparin binding. The GH loop may be responsib le for detecting the receptor attachment at the capsid surface, and for transmitting this signal to the interior of the capsid. The existence of this ordered density in the AAV2~heparin complex map, and not in the uncomplexed AAV2 map, suggests t hat capsid protein orders the nucleic acid contents in response to heparin atta chment, and in preparation for cell entry. The presence of the difference density at di mer interfaces at the interior of the capsid suggests that the nucleic acid might be involved in assembly. The DNA cage may serve to stabilize the intera ctions between assembly intermediates during capsid assembly by tethering the proteins, at dimmer interfaces,

PAGE 103

94 Figure 4-11. Capsid protein residues with the shortest radial distance from the internal DNA cage. A) The decahedron of difference density was superimposed with the atomic model of a AAV2 VP3 (Xie et al., 2002) dimer and trimer to analyze the capsid regions most closely associated with the density at the th reefold and twofold axes of symmetry. A close-up view detail B) of the inter nal difference density at the twofold and threefold axes. The potentially associated threefold loops (pink, yellow and green) and core ß-barrel st rands (blue and cyan) are noted with axes of symmetry labeled (threefold , and twofold 2). (fig. 47B), the difference density super imposed with the MVMi nucleotides (Agbandje-McKenna, 1998) was superim posed with the atomic model of AAV2 to determine which AAV2 residues C) could be found within 5Å of the MVMi nucleic acid. The residues at the twofold axis interface (blue and cyan) are within this distance. The residues at the threefold axis (pink, yellow and gr een) were determined to be the closest to the difference density map vertex at the threefold axis. Theses images were generated in Ch imera (Petterson et al., 2004).

PAGE 104

95 knitting assembly intermediates into an icosahedron by the core ß-barrel stabilizing the interactions between l oop regions. Packaging efficiency would be improved by complimentary interactions between the nucleic acid and grooves of the inner protein surface. The conservati on of amino acids at this interface would suggest that there has been selective pressure for some primary sequence specificity. These findings are not in consistent with the hypothesis that genome packaging occurs through the fivefold pore, which does not appear to be blocked by the DNA cage. There are portions of the interior cage that were not modeled with nucleic acid. This density might repr esent additional nucleic acid, or the Nterminal region of the capsid that is not seen in the crystal structure. Internal DNA Density and an Assembly and Packaging Hypothesis N-terminal amino acids R238, R307, F420, and C-terminal W684, E685, L686, and Q687 were found to be the closest to the nucleotides modeled into the interior difference density across the twofold axis. The C-terminal residues involved in these interactions are in the highly conserved ßI region (fig. 3-2) of the core barrel motif 238 is in ßB and 307 is in ßD. This suggests that capsid:DNA would stabilize the “knitting” t ogether of the core ß-barrel motif at the twofold interface promoting icosahedr al order as seen for ssRNA viruses (Rudnick and Bruisma, 2003). Al l of these residues are i dentical at this position for AAV1-9, except for a conservative substi tution in AAV4 at 238 to a histidine. Residues H627, F628 and H629, which are directly surrounding the threefold axis of symmetry, were found to be nearest to the strongest density of the dodecahedron, at the thr eefold vertex. This region of the difference density was not modeled with single-stranded nucleot ides. In the 3D structures of

PAGE 105

96 ssRNA virus that have ordered RNA, the th reefold vertex has been proposed to be a region where the duplex nucleic acid is formed because the single-stranded nucleic acid has to make multiple passes at each vertex during genome packaging (Rudnick and Bruisma, 2003). All of these residues are identical at this position for AAV1-9, except for a divergent substitution at H627 to an asparagine in AAV7-9. The thin density of the capsid protein at the threefold symmetry axis can be interpreted as a por e, but it is blocked by three F628 residues from threefold related monomer s and the internal density of the DNA, suggesting that the genome is not packaged through the threefold. There are mutagenesis data that addresse s the function of some of the amino acids nearest to the ordered internal density. Mutations at the dimeric interface which includes a R238A change and R307A change do not make capsids. Mutations from residues 681683 and from 698-693, which flank the W684-Q687 loop, made no capsids (Wu et al., 2000). The mutant with the H627A change showed wild type phenotype (Wu et al., 2000). In contrast to the twoand fivefold interfaces, residue 627 lining the threefold axis “pore” is divergent in sequence between serotypes and can tolerate a charge changing substitution as seen in the H627A substitution (Wu et al., 2000). If we assume that the internal de nsity includes ordered DNA, and assume that mutations in the associated amino acids at the twofol d are defective in capsid assembly, we might also assume that interaction of the DNA with these charged and hydrophobic residue s improve, the efficiency of capsid assembly. The linear DNA growth template would stabilize the twofold interfaces, thus

PAGE 106

97 promoting icosahedral order ing of the capsid proteins. This suggests a hypothesis about DNA-facilitat ed pentameric VP assembly (fig. 4-12). The ordered DNA density runs between twoand threefold related monomers (intradimeric and intra-trimeric interfaces), but does not line the interface between fivefold related monomers, therefore DNA is not required to stabilize pentamer formation. The N-terminus of the VP proteins is loca ted surrounding fivefold pore. The N-terminus from 217-237 runs along the interface between fivefold symmetry related monomers toward the tw ofold axis. Mutations which include E216A and D219A are only partially defecti ve (Wu et al., 2002). Mutations at both K321 and E322 are heat sensitive and essential to stabilize capsid assembly demonstrated by the fact that th is mut26 makes no capsid at 37 ºC but makes assembled, partially defective mutants at 32 ºC and E322A makes no detectable intact capsids (Bleker et al ., 2005; Wu et al., 2000). These residues 216, 219, 321 and 322 all line the fivefold por e. The data suggests that fivefold symmetry related monomer interfaces are stabilized with the N-terminal common region and do not require stabilization wit h DNA for interface interactions. Taken together, our difference map, the atomic model for AAV2 (Xie et al., 2002) and the body of mutagenesis data (Bleke r et al., 2005; Wu et al., 2000), all supports a model of nucleic acid indep endent pentamer form ation followed by dimerization of pentamers along the linear assembly template, the nucleic acid. This knits the core beta barrel motif promoting icosahedral symmetry in the formation of AAV capsid. The stabilizing in teractions at the fiveand twofold

PAGE 107

98 Figure 4-12. DNA-facilitated pentameric VP assembly. A) The atomic model of a AAV2 VP3 (Xie et al., 2002) dimer of pentamers is shown. The threefold axis is labeled as a blue tri angle and the fivefold axis with a 5. B) The dimer of pentamers would co -translationally assemble in the cytoplasm, and be imported into t he nucleus. The pentamers would dimerize on the 5Â’-end of the nasc ent DNA, adjacent to the covalently attached Rep78. A single helical turn would interact with both pentamers. Six dimers would assemble into the 2D intermediate seen in figure B). The pentamer would have to become pyramidal for circularization and would then roll al ong the DNA growth template into the C) intact capsid. Rep78 is ei ther competed off of the DNA by the protein oligomerizat ion, or by the presence of the 3Â’-end of the DNA, or remains attached to the 5Â’end at t he capsid surface. The large and small Rep complex would feed the 3Â’-end into the fivefold pore through 3Â’-helicase activity of t he small Rep proteins. 5

PAGE 108

99 interfaces allow the more complicated inter-digitation required for threefold interface interactions to occur. The VP proteins are synthesized from two mRNAs, one mRNA allows for VP3 and VP2 synthesis with a weaker start codon for VP2, and another mRNA allows for VP1 synthesis. This supports the hypothesis that a single pentamer is made up of either a single VP2 and 4 VP3 s, 5 VP1s, or 5 VP3s because of the close proximity of VP2 and VP3 protein durin g translation. A single intact capsid would be asymmetric with respect to the VP composition at fivefold axes. The VP composition would be as expected includ ing 5-10 VP1s in one to two different pentamers, and 5 VP2s in five different pent amers and the rest would be filled in with VP3 pentamer. Pack aging could occur throug h a unique pentamer and phospholipase activity could result from ex ternalization of VP1 unique at a single vertex. The charge interacting resi dues 321 and 322 N-terminal common regions that ultimately line the fivefold pore might be involved in the nucleation of pentamer formation and the destabilization of the fivefold pore. Pentamers with nuclear localization signal wo uld be selectively imported into the nucleus to the site of genome replication, and assembly of pentamers into icosahedral capsids would occur along the twofold axis in the presence of twofold interface interacting nucleic acid. Genome Uncoating Hypothesis Following the cell attachment si gnal through heparin, and subsequent internalization, the capsid would destabliz e to allow for externalization of 5 VP1 unique regions at a singly fivefold pore. All five VP1 unique regions can fit through a destabilized fivefold pore fo rming a tube-like syringe to pierce

PAGE 109

100 phospholipid membranes through phospholipase ac tivity. The capsid next ejects the genome, through the VP1 channel into t he nucleus, leaving the empty capsid shell perinuclear in the absence of helper virus. Capsid protein has been seen through microscopy localize at the outer membrane of the nucleus (Xiao et al., 2002). It will be interesting to determine the positions of the VP1 and VP2 unique regions to see how they would pa rticipate in these processes. Genome Packaging The conservation of amino acids a ssociated with the DNA suggests some specificity for optimal interactions. Evidenced by the presence of seemingly random DNA in our expressed AAV2 particl e, expressed in the absence of AAV ITRs, and by recombinant AAV technology that involves packaging of transgenes, there does not appear to be exact sequence specificity of the entire genome necessary to package DNA inside AAV capsids. But, studies of nucleotide sequence preferenc e for the optimal packaging efficency might be useful toward the developm ent of gene therapy vectors, and to determine which types of nucleotides at which positi ons along the genome allow for the most efficient genome packaging. Antigenicity and Heparin Binding Insertions of amino acids at positions 447 and 587 do not affect A20 binding, while Insertions at 261, 381, 534 and 573 inhibit A20 binding. Insertions at 534, 573 and 587 affect C37 binding, whil e an insertion at 261 does not. All of these residues can be viewed at the threefol d axis of symmetry (fig. 2-5). From this view it is clear that the C37 fo otprint covers the threefold mounds and therefore covers the heparin binding re sidues R585 and R588, neutralizing AAV2

PAGE 110

101 infection by blocking cell attachment. A20, on the other hand, does not block heparin binding, but blocks a step after internalization. The A20 epitope has been mapped to the depression surrounding the fivefold axis at the base of the threefold mounds, where the HI loop is found. It is plausible that A20 bivalently binds to the capsid at the fivefold depression, across the twofold interface, blocking the action of the acidic patch and/or the conformation change at the HI loop. This suggests roles for these two regions that influence capsid dynamics necessary for infection which must follow cell attachment, but that are critical for cell entry and infectivity. An insertion at F534 was also found to i nhibit A20 binding, although it is not included in the mapped epit ope (Wobus et al., 2000). Gi ven the close proximity of K532 to F534, it is conc eivable that A20 neutralizati on occurs by blocking the activity of K532 as well. Mutagenes is of K532 reduces heparin binding 50% relative to wild type, but with a five log reduction in infectivity. These findings suggest that residue K532 has a dramatic e ffect on infectivity at a step in the virus life cycle where a cell attachment signal is translated into a signal for a step following internalization.

PAGE 111

102 CHAPTER 5 PRODUCTION, PURIFICATION, CRYST ALLIZATION AND PRELIMINARY XRAY STRUCTURAL STUDIES OF ADEN O-ASSOCIATED VIRUS SEROTYPE 5 Introduction Evolutionary Divergent Serotype AAV5 AAV5 is the only serotype originally isolated from human tissue (BantelSchaal et al., 1984). Phylogenetic analyses based on genome sequence analysis of the AAV1-5 serotype rep and cap ORFs and primary amino acid sequence analysis of capsid protein VP1 for AAV1-9, indicate that AAV5 is the most divergent human serotype (Bantel -Schaal et al., 1999; Lukashov and Goudsmit, 2001; Gao et al., 2003; 2004). The AAV5 nucleic acid is further unique from other studied serotypes in having a Rep protein-mediated cleavage site in the ITR DNA sequence, and an al ternative intronic mRNA polyadenylation site and mRNA splicing not stimulated by Rep protein (Chiorini et al., 1999; Qiu et al., 2002; 2004). The fully assembled capsid of AAV5 has been shown to have unique antigenicity, second only to AAV4 , for specific binding of monoclonal antibodies by ELISA against intact AAV c apsids (Wobus et al., 2000). AAV has been shown to utilize clathrin -mediated endocytosis for cell entry and require helper virus for viral genome r eplication. AAV5 has been shown to be routed to the golgi with very li ttle degradation by t he proteosome, and expression of AAV5 is less dependent on helper-virus function than AAV2 (Bantel-Schaal et al., 2001; Qiu et al., 2002). Mutagenesis, cell

PAGE 112

103 binding, and transduction st udies show that the capsid protein amino acid sequence of AAV plays a central role in the observed tissue recognition and transduction efficiency disparities (Kern et al., 2003; Opie et al., 2003). These observations and a goal to enhance the efficacy of AAV gene therapy applications, by viral capsid modifications for specific cell/tissue targeting, has created a necessity for an understanding of t he capsid structure and interactions with cell surface ligands of AAV5. AAV5 as a Gene Therapy Vector Although AAV2 is the most widely studied serotype for gene therapy, AAV5 binds and infects the apical surface of hum an airway epithelia more efficiently than AAV2, by factors of 20 and 50, in vivo and in vitro , respectively (Walters et al., 2001). This property has resulted in ut ilization of recombinant AAV5 capsids, carrying the gene for the cystic fibrosis transmembrane conductance regulator, for gene therapy applications aimed at the tr eatment of cystic fibrosis. AAV5 is also a promising candidate vector for photoreceptor degeneration therapy showing higher transduction and transgene delivery than AAV2 in photoreceptor cells (Yang et al., 2002). AAV5 Host-Cell receptors The improved transduction of photorecept or cell by AAV5 may be due to the presence of PDGFR on the cell surface of rod ce lls (Lotery et al., 2003). Inhibition of this receptor tyrosine kinase blocks binding and transduction by AAV5 in cells normally permissive to AAV5 infection, while over-expression of PDGFR increases AAV5 transduction in HeLa cells, suggesting an important role for PDGFR in AAV5 cell entry (Di Pasquale et al., 2003).

PAGE 113

104 Amino acids in the overlapping VP region mediate AAV5 transduction of epithelial cells through interaction with sial ic acid binding rat her than the heparin sulfate utilized by AAV2 for broad range cellular transduction (Davidson et al., 2000; Kaludov et al., 2001; Walters et al., 2001). Studies involving neuraminadase treatment, and resialylation of cells suggest a requirement for Nlinked sialic acid for AAV5 cell binding since transduction of AAV5 was blocked by competition with soluble 2-3 and 2-6 sialic acid (Kaludov et al., 2001). Structural St udies of AAV Information based on mutagenesis studies, the cryo-Em density maps, and the atomic model based on the crystal structure of AAV2 have been used in conjunction to implicate the threef old mounds in antigenic and tropism determination for AAV2 (Xie et al, 2002; Op ie et al., 2002; Kern et al., 2002; Wobus et al., 2000; Kronenberg et al., 2001; Padron et al., 2005). The threedimensional structure of wild type AAV5 has been determined to 16Ã… by cryoelectron microscopy, and a homology -based pseudo-atomic model was constructed (Walter et al., 2004). The cr yo-EM density map (fig . 3-6) shares the typical features of the other AAV serotype known st ructures with threefold protrusions, twofold depressions and a protru sion at the fivefold axis of symmetry that surrounds a pore. In the crystal structures of AAV2 and AAV4 only the overlapping C-terminal polypeptide region is ordered, with T=1 icosahedral symmetry, and shows differences in regions that control AAV2 receptor attachment and antigenicity (Xie et al., 2002; L. Govindasamy, E. Padron, N. Kaludov, R. McKenna, N. Muzyczka, J. Chiorini, M. Agbandje-McKenna , unpublished results). To elucidate

PAGE 114

105 the structural features of AAV5 controlling its enhanced transduction of the apical surface of airway epithelial and phot oreceptor cells compared to other AAV serotypes, X-ray crystallographic studies of the viral capsid have been initiated. We have reported the producti on, purification, crystallization and preliminary crystallographic analysis of empty AAV5 VLPs. The crystals diffract X-rays to beyond 3.2 Å resolution, using synchro tron radiation, and belong to the orthorhombic space group P 212121, with unit cell param eters a=264.7, b=447.9, and c= 629.7 Å. There are four complete capsids in the unit cell and one per asymmetric unit. The orientations and positions of the capsids have been determined by rotation and translation func tions, respectively, and the structure determination is in progress. Results and Discussion Crystallization Conditions for AAV5 Virus-Like Particles The purity and integrity of the empty AAV5 viral capsids were verified by SDS-PAGE (fig. 5-1A) and negative stain EM (fig. 5-1B), respectively, prior to crystallization. Crystals of ~0.15 × 0.10 × 0.05 mm in size (fig. 5-1C) were obtained in ~2 to 3 weeks for the AAV5 viral capsids in 350 mM NaCl, 1.5% PEG-8000 at all the pHs (6.0 to 8.0) and MgCl2 concentrations screened at RT. The crystal screens at 4º C produced cr ystals of similar size after ~9 months of incubation. Crystals grown in 20 mM Tris-HCl, pH 7.5, 350 mM NaCl, 5-10 mM MgCl2, 1.5% PEG 8000 at RT and in 20 mM Tris-HCl, pH 8.5, 350 mM NaCl, 10 mM MgCl2, 1.5% PEG 8000 at 4ºC were used for data collection.

PAGE 115

106 Figure 5-1. Isolation and characterizati on of purified AAV5 empty viral capsids. A) SDS–PAGE gel of AAV5 showing the positions of VP1, VP2 and VP3, which are 80, 65 and 59 kDa in si ze, respectively. The expected positions for low-molecular-weigh t standards (in kDa; Bio-Rad, Hercules, CA, USA) are indica ted on the right-hand side. B) Transmission electron micrograph of purified AAV5 empty (no DNA) viral capsids negatively stained with 2% uranyl acetate (UA). The dark centers indicate the uptak e of UA into the empt y capsids. C) Optical photograph of AAV5 empty viral capsid crystals. Crystals are 0.15 X 0.10 X 0.05 mm in size. The photograph was taken with a Bio-Rad 1024 ES confocal microscope with an Olympus IX 70 transmission.

PAGE 116

107 X-Ray Data Collection and Processing Results The crystals diffracted X-rays to beyond 3. 2 Å resolution (fig. 5-2). The data were indexed to a primitive orthor hombic crystal system, with unit-cell parameters a=264.7, b=447.9, and c= 629.7 Å (table 5-1) . Inspection of the h00, 0k0, and 00l classes of reflections showed systematic absences for the odd reflections indicating that there were three perpendicu lar 21 screw axes and that the crystals belong to space group P212121. Separately, the diffraction data sets from crystals grown at RT and 4ºC have been processed and merged for the space group P212121. The RT cryst als data set scales with an overall R sym of 16.3 % and I/ I of 7.2 (Table 5-1). The 4ºC crystals data set did not scale with acceptable statistics (e.g. Rsym > 25 %) and was unsuit able for further analysis. Thus from here on all discussions refer to the data set for the crystals grown at RT. Only 50% of the data collected (196 images from 10 crystals) were usable due to synchrotron radiation damage of the small crystals during data collection. The statistics for the data set, utiliz ed for molecular replacement structure determination procedures, are summarized in Table 5-1. The unit cell dimensions and the molecular weight of the AAV5 empty viral capsid gave Vm values (Mathews, 1968; Kantardjieff & Rupp, 2003) of 5.2, 2.6, 1.7 and 1.3 Å3 Da-1 for 1, 2, 3 and 4 molecules per asymmetric, respectively. Packing considerations for the AAV5 space group suggests four viral capsids in the unit cell, 1 molecule per asymmetric unit. Thus assuming that t here are four viral capsids in the unit cell, the Vm is 5.2 Å3 Da-1 corresponding to solvent content of ~76%, assuming a molecular density of 1.3 g/cm3. The high Vm is likely because crystals are

PAGE 117

108 Figure 5-2. X-ray diffraction image fo r a crystal of AAV5 empty (no DNA) viral capsids. The image is a 0.3º oscillat ion diffraction pattern collected at the X29 (NSLS, BNL) beamline. The crystals diffract X-rays to beyond 3.2 Å resolution. Concentric rings depi ct the 50.0, 20.0, 10.0, 5.0 and 3.2 Å resolution shells.

PAGE 118

109 Table 5-1. Crystal data collection and processing statistics (Di Mattia et al., 2005) Values in parentheses are for the highest resolution shell. aRsym = |I . |/ |I|, where I is the intensity of an individual reflection and is the average intensity for this reflection; the summation is over all intensities.

PAGE 119

110 formed from empty viral capsids, thus containing a shell of protein surrounding solvent. If the center of the capsid was occupied by the genomic DNA the Vm value (3.6) would fall in the range for protei n-nucleic acid (Di Mattia et al., 2005). The structure determination of the AAV5 VLPs to 3.45 Ã… resolution is currently underway using the molecular r eplacement solutions and initial phases calculated with the AAV4 poly-alanine model (Govi ndasamy et al., unpublished results). Model and phase improvement will entail real-space 60-fold NCS averaging and refinement usi ng the CNS program (Brunger et al ., 1998) combined with interactive model build ing with the O program (Jones et al ., 1991). The exploitation of AAV5 and other AAV viral capsids for gene therapy applications is already underway. A compar ison of the crystal structure of AAV5 to those available for AAV2 (Xie et al ., 2002) and AAV4 (Govindasamy et al ., unpublished results) will provide the necessary information for further analyzing the structural determinants of cellular tr ansduction specificities towards improving the efficacy of rAAVs for targeted gene ther apy applications (Di Mattia et al., 2005).

PAGE 120

111 CHAPTER 6 SUMMARY AND FUTURE DIRECTIONS AAV Capsid Structures AAV1 Cryo-EM Map The cryo-electron microscopy stru cture of AAV1 to 17.6 angstroms resolution was useful for visualizing stru ctural differences between serotypes by comparing our structure to the known cryo-EM structures of AAV2, AAV4 and AAV5 (Kronenberg et al., 2000; Padron et al., 2005; Walters et al., 2004) which are known to differ in host-cell tropism and antigenicity. The AAV1 electron density envelope was used in conjunction with the atomic m odel of AAV2 and AAV4, based on the crystal structures (Xie et al., 2002; Govindasamy et al., unpublished data). We were able to generate a pseudo-atomic model for AAV1, based on homology modeling and constrained by our cryo-EM map. With previous mutagenesis data for AAV1 and AAV2 and caps id protein (Girod et al., 1999; Wu et al., 2000; Wobus et al., 2000, Opie et al., 2003; Kern et al., 2003; Hauck and Xiao, 2003) we were able to compare st ructural differences with phenotypic differences observed between serotypes and mutants. Such phenotypes include those that influence receptor and cell bi nding, internalization, DNA packaging and infectivity. Obtaining higher re solution cryo-EM and crystallographic structure data for AAV1 will be crucial to further the understanding of how the virus biological functions are dictated by the structure. T he atomic model will

PAGE 121

112 help to pinpoint residues that govern t hese processes, t herefore, it will be important to optimize the cryst allization conditions for AAV1. AAV5 X-ray Crystal Structure We were able to crystalli ze AAV5 for the first time and have since optimized the crystallization conditions. The struct ure of AAV5 is currently being refined by molecular replacement. The atomic model of this very divergent serotype will be very useful for serotype evolutiona ry studies based on AAV capsid protein structure. AAV Capsid Surface Evolution With the exception of the fivefold por e dimensions, the threefold protrusions appear to be the sites of the structural divergence between species (fig 3-4, 3-5). This divergent surface region is compris ed of a cluster of amino acid loops, as seen by superimposing the atomic m odels with the cryo-EM density envelopes (fig 3-3). These divergent sequences include amino acids that have been shown through mutagenesis and in vivo and/or in vitro assays, to influence the host cell binding, tropism and antigenic nature of AAV. Serotype amino acid sequence alignments and mutagenesis further suggest that these divergent surface loop regions, known to influence to the biology of the virus, are tolerant of amino acid insertions and changes, and may be usef ul for future mutagenesis studies directed toward specific cell targeti ng for the development of gene therapy vectors. The amino acid sequences at the threefold that tolerate change will also be useful for future phylogenetic analysis to determine the evolutionary relationships among the AAVs. Although t he appearance of the fivefold does not vary much between serotypes, we have determined in this study that the

PAGE 122

113 depression around the fivefold pore influences infectivity at a stage down stream of receptor attachment and is s ubject to conformational changes. The monoclonal antibodies that we re raised against the adenovirus expressed AAV1 capsids will be useful to accurately determine particle titres and sample concentrations for AAV1. This will allow for calculatiing particle to infectivity and particle to genome ratios of sample preparations of AAV1 for packaging efficiency and infectivity studies . This set of antibodies includes clones that are able to distinguish between intact and denatured AAV1 capsid protein. This will be useful for moni toring assembly and disassembly of AAV1 capsids and oligomeric assembly intermediates. The Heparin Density and Mutagenesis The cryo-EM structure of AAV2~heparin complex was useful for understanding which regions of the AAV2 capsid bind heparin sulfate, and to visualize these interactions and associated capsid conformational changes. As predicted by mutagenesis studies (Wu et al., 2000; Opie et al., 2003; Kern et al., 2003) R585 and R588 and R484 are availabl e to make ionic and water-mediated hydrogen bonding interactions with heparin sulfate, while the positive charge of other basic residues R487, and K532 may be par tially neutralized by interactions with residues E530 (along wit h E574 and K527) and D494 respectively (fig. 4-9). Alanine substitute mutations in E530, K527 and D494 produce viruses defective for infectivity, but positive for heparin binding. These acidic residues form negatively charged patch with acidic loops 528-531, and 561-564 of the capsid. The interactions between the acidic pat ch and residues of the basic patch may be integral for securing the heparin bindi ng region architecture in preparation for

PAGE 123

114 binding. Disruption of these interactions may also serve as a mechanism for the capsid to sense heparin binding at the cell surface by inducing conformational changes that further prepare t he capsid for cell entry. It will be interesting to see, in a high resolution X-ray crystal st ructure of the AAV2~heparin complex, if heparin competes with the acidic pat ch for contact with R487 and R532. Interactions of residue K532 with heparin appear to dramatically influence infectivity, while modestly effecting heparin binding, suggesting an important role in K532 for transmitting the cell-binding si gnal through the capsid to activate a downstream step in infection such as internalization (Opie et al., 2003). The mutagenesis data clearly supports our model of heparin interacting with residues R585 and R588 and we were able to model the difference density at the twofold as additional heparin oligosaccharide allowing fo r interactions with K532. The twofold density change may also be due to conformational changes of the twofold loops at dimer interfaces, as oppos ed to direct interaction with heparin. Higher resolution data will be needed to det ermine the true natur e of this density change and therefore optimizi ng the crystallization conditions for AAV2~heparin will be important. AAV2~Heparin Conformational Differences Changes at the fivefold pore. The mutagenesis data (Wu et al., 2000; Opie et al., 2003; Kern et al., 2003; Bl eker et al., 2005) di d not predict the structural changes, in response to heparin binding, seen at the fivefold axis and at the capsid interior running betw een twoand threefold symmetry related monomers. The changes in the diameter of the fivefold pore and in the density surrounding the pore (fig. 4-2, 4-3), in the presence of saturating amounts of

PAGE 124

115 soluble heparin, illustrated the dynamic natur e of the virus capsid in this region which is known to be associated with conformational changes in the VP2 and VP1 N-terminal regions, PLA2 activity and with genome packaging (Bleker et al., 2005). We hypothesize that the amino ac ids found most directly surrounding the pore, K321 and E322, link fivefold sy mmetry related monomers through in a water-mediated ring of alternating charge (Xie et al., 2002). Alanine substitutions on these two residues generate heat sens itive mutants t hat do not produce capsids at physiological temperature, s uggesting that these charge interactions influence conformational changes to the c apsid that dictate capsid stability and assembly (Wu et al., 2002; Bleker et al., 2005). Higher resolution crystallographic data will be needed to m ap the loop and side-chain movements that occur upon heparin binding. Changes at the capsid interior and an assembly hypothesis. Mutagenesis data did not predict the internal difference density at the interior of the capsid shell. The dodecahedron shape of the internal density was reminiscent of ordered RNA seen in ssRNA virus structure and the DNA seen in the AAV4 and MVMi structure (Rudnick and Bruisma, 2003 ; Padron et al., 2005; Agbandje-McKenna et al., 1999). We det ermined that our expressed particles did in fact contain DNA and interpreted t he internal difference as DNA that is ordered in the complex and not in the unc omplexed cryo-EM map. This density could also be the N-terminal VP1 and VP2 regions or heparin oligosaccharides that penetrated the capsid in our comp lexing reaction. Modeling the MVMi nucleotides into this region of our difference map and then superimposing the

PAGE 125

116 AAV2 capsid structure allowed us to deter mine the interior capsid residues that are closest to the difference density. Future Directions to Study AAV2 Assembly The work by Im and Muzyczka (1989), Chejanovsky and Carter (1989) King et al. (2001) Wistuba et al . (1995) include methodology that will be useful to study the 5Â’-terminus:Rep78 complex-medi ated, capsid assembly hypothesis proposed in this work. The 3Â’-labeled terminal-hairpin fragment containing the RBE could be added to the nuclear extrac t of adenovirus and AAV infected cells, or , the intact capsids could be puri fied on a heparin-agarose column and DNase treated (Im and Muzyczka, 1989; Zolotokhin et al., 2002). Next the capsids could be sedimented to separate the different populations, and then tested for the presence of the fragment in denaturing cond itions (Wistuba et al., 1995). In addition, the experiments by King et al . (2001) could be r epeated, but the 5Â’terminus should be probed. Finally, a recombinant virus could be constructed that has both rep and cap genes, but where the small rep gene has been mutated, like the rescue experiments by Chejanovsky and Carter (1989). This virus would be used to infect cells and t he purified capsids would be analyzed for the presence of 5Â’-terminal AAV DNA. Future GH Loop Mutagenesis The GH loop is a long loop that links the threefold axis with the fivefold axis. Mutagenesis of F420, which is at the dime ric interface interacting with the nucleic acid would be interesting to probe the effe ct of the GH loop on assembly at the dimer interface and genome packaging. Mutagenesis of residues H627, F628, and H629, either all together, or in comb inations, or alone would be useful to

PAGE 126

117 determine the influence of this conserved, basic/aromatic region of the GH loop. The proposed assembly hypothesis suggests that these residues do not effect the production of VP pentamers and w ould not be needed until the pentamer of dimers rolls along the DNA for circularization. If this is true, mutations in this region might result in the production of assembly intermediate (fig. 4-12) which might include, up to six dimers of pentamers, Rep78 and a few hundred nucleotides of DNA. Charged Residue Mutagenesis The acidic residues that are associ ated with heparin binding and infectivity should be analyzed in detail including D561-E564, shown to be required for heparin binding, D528-E531 which intera ct with loop D561-E564 through solventmediated hydrogen bonds and residues D494, D514, E574 which may interact with the residues that form the basic pat ch. There were two water-mediated rings of charge proposed in this study, the ri ng of interaction formed between R487, E574, K527 and E530 that is associated with the heparin binding region, and the fivefold ring formed by K321 and E322. The fivefold ring has been studied and the mutants are heat sensitiv e. The heparin binding associated ring should also be studies for its effects on heparin binding and on transmitting the cell attachment signal to t he fivefold axis.

PAGE 127

118 LIST OF REFERENCES Agbandje, M., Kajigay a, S., McKenna, R., Rossmann, M. G., Young, N. S. 1994. The structure of human parvovirus B 19 at 8 Å resolution. Virology. 203:106–115. Agbandje, M., McKenna, R., Ro ssmann, M. G., Strassheim, M. L., Parrish, C. R. 1993. Structure determination of feline panleukopenia virus empty particles. Proteins. 16:155–171. Agbandje, M., Parrish, C. R., Rossmann, M. G. 1995. The structure of parvoviruses. Semin. Virol. 6:299–309. Agbandje-McKenna, M., Llamas-Saiz, A. L., Wang, F., Tattersall, P., & Rossmann, M.G. 1998. Functional implicat ions of the structure of the murine parvovirus, minute virus of mi ce. Structure. 6:1369-1381. Baker, T. S., Olson, N. H ., Fuller, S. D. 1999. A dding the third dimension to virus life cycles: three-dimensional re construction of icosahedral viruses from cryo-electron micrographs. Micr obiol. Mol. Biol. Rev. 63:862–922. Bantel-Schaal, Zur Hausen, U., H. 1984. Characterization of the DNA of a defective human parvovirus isolated from a genital site. Virology. 134:52– 63. Bantel-Schaal, U.A., Delius, H., Schmid t, R., & Hausen, H.Z. 1999. Human Adeno-Associated Virus Type 5 Is Only Distantly Related to Other Known Primate Helper-Dependent Parvoviruses . J. Virol. 73:2, 939-947. Bantel-Schaal, U., Hub, B. & Kartenbeck, J. 2002. Endocytosis of AdenoAssociated Virus Type 5 Leads to Accumulation of Virus Particles in the Golgi Compartment. J. Virol. 76:5, 2340-2349. Barbis, D. P., Chang, S. F., Parrish, C. R. 1992. Mutati ons adjacent to the dimple of the canine parvovirus capsid st ructure affect sialic acid binding. Virology. 191:301–308.

PAGE 128

119 Becerra, S. P., Koczot, F., Fabisch, P., Rose, J. A. 1988. Synthesis of adenoassociated virus structural proteins requires both alternative mRNA splicing and alternative initiations from a single transcript. J. Virol. 62:2745–2754. Belnap, D. M., Kumar, A., Folk, J. T., Sm ith, T. J., Baker, T. S. 1999. Lowresolution density maps from atomic models: how stepping “back” can be a step “forward.” J. St ruct. Biol. 125:166–175. Bink, H.J. and Pleij, C.W.A. 2002. RNA–protein interactions in spherical viruses Brief Review. Arch Virol. 147: 2261-2279. Bossis, I., Chiorini, J. A. 2003. Cloning of an adeno-associated virus (AAAV) and generation of recombinant AAAV par ticles. J. Virol. 77:6799–6810. Böttcher, B. & Crowther, R.A. 1996. Di fference imaging reveals ordered regions of RNA in turnip yellow mosaic virus. Structure. 4:387-394. Bowman, V., Chase, E. S., Franz, A. W. E. , Chipman, P. R., Zhang, X., Perry, K. L., Baker, T. S., Smith, T. J. 2002. An anti body to the putative aphid recognition site on cucumber mosaic virus recognizes pentons but not hexons. J. Viro l. 76:12250–12258. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., GrosseKunstleve, R. W., Jiang, J. S., Ku szewski, J., Nilges, M., Pannu,N. S., Read, R. J., Rice, L. M., Simons on, T. G., Warren, L. 1998. Crystallography and NMR system: a new so ftware suite for macromolecular structure determination. Ac ta Crystallogr. D. 54:905–921. Cassinotti, P., Weitz, M., Tr atschin, J. D. 1988. Organization of the adenoassociated virus (AAV) capsid gene: mapping of a minor spliced mRNA coding for virus capsid protei n 1. Virology. 167:176–184. Chao, H., Walsh, C.E. 2004. AAV Vect ors for Hemophilia B Gene Therapy. Grand Rounds. Mt. Sinai J. Med. 71:5, 305-313. Chapman, M. S., Rossmann, M. G. 1993. Structure, sequenc e, and function correlations among parvoviruse s. Virology. 194:491–508. Chen, S., Kapturczak, M., Loiler, S.A ., Zolotukhin, S., Glushakova, O.Y., Madsen, K.M., Samulski, R.J., Hauswir th, W.W., Campbell-Thompson, M., Berns, K.I., Flotte, T.R., Atkinson, M. A., Tisher, C.C., & Agarwal, A. 2005. Efficient Transduction of Vascular E ndothelial Cells with Recombinant Adeno-Associated Virus. Hu m. G. Ther. 16:235-247. Chiorini, J.A., Wiener, S.M., Owens, R.A., Kyos tio, S.R.M., Kotin, R.M., Safer, B. 1994. Sequence Requirements for St able Binding and Function of Rep68 on the Adeno-Associated Virus Type 2 Inve rted Terminal Repeats. J. Virol. 68:11, 7448-7457.

PAGE 129

120 Chiorini, J. A., Yang, L., Li u, Y., Safer, B., Kotin, R. M. 1997. Cloning of adenoassociated virus type 4 (AAV4) and gener ation of recombinant AAV4 particles. J. Virol. 71:6823–6833. Chiorini, J.A., Afione, S. & Kotin, R.M. 1999. A deno-Associated Virus (AAV) Type 5 Rep Protein Cleaves a Unique Terminal Resolution Site Compared with other AAV Serotypes. J. Virol. 73:5, 4293-4298. Chiorini, J. A., Kim, F., Yang, L ., Kotin, R. M. 1999. Cloning and Characterization of adeno-associated virus type 5. J. Virol. 73:1309– 1319. Chipman, P.R., Agbandje-McKenna, M., Kaji gaya, S., Brown, K.E., Young, N.S., Bayer, T.S., Rossmann, M.G. 1996. Biochemistry Cryo-electron microscopy studies of empty capsid s of human parvovirus B19 complexed with its cellular receptor. Proc . Natl. Acad. Sci. 93, 7502-7506. Collaborative Computational Project. 1994. The CCP4 suite: programs for protein crystallography. Ac ta Crystallogr. D. 50:760–763. Cotmore, S.F., & Tattersall, P. 1989. A Genome-Linked Copy of the NS-1 Polypeptide Is Located on the Outside of Infectious Parvovirus Particles. J. Virol. 63:9, 3902-3911. Davidson, B.L. & Breakefie ld, X.O. 2003. Viral Vect ors For Gene Delivery To The Nervous System. Neuroscience. 4, 353. Denby, L., Nicklin, S.A. & Baker, A.H. 2005. Adeno-associated virus (AAV)-7 and -8 poorly transduce vascular endothel ial cells and are sensitive to proteasomal degradation. Gene Therpy. 12, 1534-1538. DiMattia, M., Govindasamy, L., Levy, H.C., Gurda-Whitaker, B., Kalina, A., Kohlbrenner, E., Chiorini, J.A., McK enna, R., Muzyczka, N., Zolotukhin, S., Agbandje-McKenna, M. 2005. Production, purification, crystallization and preliminary X-ray structural studies of adeno-associated virus serotype 5. Acta. Cryst. F61, 917-921. Di Pasquale, G., Davidson, B. L., Stein, C. S., Martins, I., Sc udiero, D., Monks, A., Chiorini, J. A. 2003. Identification of PDGFR as a receptor for AAV-5 transduction. Nat. Med. 9:1306–1312. Duan, D., Li, Q., Kao, A. W., Yue, Y., Pessin, J. E. , Engelhardt, J. F. 1999. Dynamin is required for recombinant adeno-associated virus type 2 infection. J. Virol. 73:10371–10376. Duan, D., Yue, Y., Yan, Z., Yang, J. , Engelhardt, J. F. 2000. Endosomal processing limits gene transfer to polar ized airway epithelia by adenoassociated virus. J. Clin. Investig. 105:1573–1587.

PAGE 130

121 Dubielzig, R., King, J. A., Weger, S., Ke rn, A., Kleinschmidt, J. A. 1999. Adenoassociated virus type 2 protein intera ctions: formation of preencapsidation complexes. J. Virol. 73:8989–8998. Esnouf, R. M. 1997. An ex tensively modified version of MolScript that includes greatly enhanced coloring capabilities. J. Mol. Graph. Model. 15:132–134. Fry, E. E., Lea, S. M., Jacks on, T., Newman, J. W., Ella rd, F. M., Blakemore, W. E., Abu-Ghazaleh, R., Sam uel, A., King, A. M., St uart, D. I. 1999. The structure and function of a foot-and-mouth diseas e virus oligosaccharide receptor complex. EMBO. J. 18:543–554. Gao, G., Vandenberghe, L.H. , Alvira, M.R., Lu, Y., Calcedo, R., Zhou, X., & Wilson, J.M. 2004. Clades of Adeno-Associated Viruses Are Widely Disseminated in Human Tissues. J. Virol. 78:12, 6381-6388. Gao, G., Alvira, M.R., Wang, L., Calcedo, R., Johnston, J., & Wilson, J.M. 2002. Novel adeno-associated viruses from rhesus monkeys as vectors for human gene therapy. PN AS. 99:18, 11854-11859. Gao, G., Alvira, M.R., So manathan, S., Lu, Y., Vandenber ghe, L.H., Rux, J.J., Calcedo, R., Sanmiguel, J., Abbas, Z., Wilson, J.M. 2003. Adenoassociated viruses undergo substantial ev olution in primates during natural infections. PNAS. 100:10, 6081-6086. Girod, A., Wobus, C. E., Zador i, Z., Ried, M., Leike, K., Tijssen, P., Kleinschmidt, J. A.,Hallek, M. 2002. The VP1 c apsid protein of adenoassociated virus type 2 is carrying a phospholipase A2 domai n required for virus infectivity. J. Gen. Virol. 83:973–978. Girod, A., Ried, M., Wobus , C., Lahm, H., Leike, K., Kl einschmidt, J., Deleage, G., Hallek, M. 1999. G enetic capsid modifications allow efficient retargeting of adeno-associated virus type 2. Nat. Med. 5:1052–1056. Erratum. 5:1438. Girod, A., Wobus, C. E., Z adori, Z., Ried, M., Leike, K., Tijssen, P., J. A. Kleinschmidt, Hallek, M. 2002. The VP1 capsid protein of adenoassociated virus type 2 is carrying a phospholipase A2 domain required for virus infectivity. J. Gen. Virol. 83:973–978. Gouet, P., Courcelle, E. D., Stuart, I., Metoz, F. 1999. ESPript: multiple sequence alignments in PostScrip t. Bioinformatics. 15:305–358. Govindasamy, L., Hueffer, K., Parrish, C.R., Agbandje-Mc Kenna, M. 2003. Structures of Host RangeControlling Regions of the Capsids of Canine and Feline Parvoviruses and Mutants. J. Virol. 77:22, 12211-12221.

PAGE 131

122 Grieger, J.C., Samulski, R.J. 2005. Packaging Capacity of Adeno-Associated Virus Serotypes: Impact of Larger G enomes on Infectivity and Postentry Steps. J. Virol. 79:15, 9933-9944. Grifman, M., Trepel, M., Speece, P., Gil bert, L. B., Arap, Pasqualini, W., R., Weitzman, M. D. 2001. Incorporati on of tumor-targeting peptides into recombinant adeno-associated virus c apsids. Mol. Ther. 3:964–975. Grimm, D., Kern, A., Rittner, K., Klein schmidt, J. A. 1998. Novel tools for production and purification of recombi nant adenoassociated virus vectors. Hum. Gene Ther. 9:2745–2760 Grimm, D., Kern, A., Pawlita, M., Ferrari, F., Sa mulski, R., Kleins chmidt, J. A. 1999. Titration of AAV-2 particles via a novel capsid ELISA: packaging of genomes can limit production of recomb inant AAV-2. Gene Ther. 6:1322– 1330. Grimm, D., Kleinschmidt, J. A. 1999. Progress in adeno-associated virus type 2 vector production: promises and prospec ts for clinical use. Hum. Gene. Ther. 10:2445–2450. Hauck, B., Xiao, W. 2003. Characterization of Tissue Tropism Determinants of Adeno-Associated Virus Type 1. J. Virol. 77:4, 2768-2774. Hauswirth, W. W., Lewin, A. S., Zolotukhin, S., Muzyczka , N. 2000. Production and purification of recombinant adeno-a ssociated virus. Methods Enzymol. 316:743–761. Heilbronn, R., Burkle, A., Stephan, S. , Zur Hausen, H. 1990. The adenoassociated virus rep gene suppresses herpes simplex virus-induced DNA amplification. J. Virol. 64:3012–3018. Henikoff, S., Henikoff, J. G. 1992. Amino acid substitution matrices from protein blocks. Proc. Natl. Acad. Sci. USA. 89:10915–10919. Hileman, R. E., Fromm, J. R., Weiler, J. M., Linhardt, R. J. 1998. Glycosaminoglycan-protein interactions: definition of consensus sites in glycosaminoglycan binding protei ns. Bioessays. 20:156–167. Hileman, R. E., Jennings, R. N., Linhardt, R. J. 1998. Thermodynamic analysis of the heparin interaction with a bas ic cyclic peptide using isothermal titration calorimetry. Biochemistry. 37:15231–15237. Hueffer, K., Govindasamy, L., Agbandj e-McKenna, M., Parrish, C. R. 2003. Combinations of two capsid region s controlling canine host range determine canine transferrin receptor binding by canine and feline parvoviruses. J. Virol. 77:10099–10105.

PAGE 132

123 Hueffer, K., Parrish, C.R. 2003. Parvovirus host range, cell tropism and evolution. Curr. Opn. Microbiol. 6:392-398. Hueffer, K., Parker, J. S., Weichert, W. S., Geisel, R. E., Sgro, J. Y., Parrish,C. R. 2003. The natural host range shi ft and subsequent evolution of canine parvovirus resulted from virus-specific binding to the canine transferrin receptor. J. Virol. 77:1718–1726. Im, D., Muzyczka, N. 1989. Factors That Bind to Adeno-Associated Virus Terminal Repeats. J. Vi rol. 63:7, 3095-3104. Jones, D. T. 1999. Protein secondar y structure prediction based on position specific scoring matrices. J. Mol. Biol. 292:195–202. Jones, T., Zou, J.-Y., Cowa n, S. W., Kjeldga ard, M. 1991. Improved methods for building protein models in elec tron density maps and the location of errors in these models. Ac ta Crystallogr. A 47:110–119. Kaludov, N., Brown, K. E., Wa lters, R. W., Zabner , J., Chiorini, J. A. 2001. Adeno-associated virus serotype 4 (AAV4) and AAV5 both require sialic acid binding for hemagglutination and e fficient transduction but differ in sialic acid linkage specific ity. J. Virol. 75:6884–6893. Kaludov, N., Padron, E., Govindasamy, L., McKenna, R., Chiorini, J. A., Agbandje-McKenna, M. 2003. Production, purific ation and preliminary Xray crystallographic studies of adeno-associated virus serotype 4. Virology. 306:1–6. Kapturczak, M.H., Chen, S., Agarwal, A. 2005. Adeno-associated virus vectormediated gene delivery to the vasculature and kidney. Acta. Biochica. Polo. Rev. 52:2, 293-299. Kaufmann, B., Simpson, A. A., Rossmann, M. G. 2004. The structure of human parvovirus B19. Proc. Natl. Acad. Sci. U SA. 101:11628–11633. Kern, A., Schmidt, K., Leder, C ., Muller, O.J., Wobus, C.E., Bettinger, K., Von der Lieth, C.W., King, J.A., Kleinshmidt, J.A. 2003. Identification of a HeparinBinding Motif on Adeno-Associated Virus Ty pe 2 Capsids. J. Virol. 77:20, 11072-11081. King, J. A., Dubielzig, R ., Grimm, D., Kleinschmidt, J. A. 2001. DNA helicasemediated packaging of adeno-associated virus type 2 genomes into preformed capsids. EMBO. J. 20:3282–3291. Klimstra, W. B., Ryman, K. D., Johnston, R. E. 1998. Adaptat ion of Sindbis virus to BHK cells selects for use of heparan sulfate as an attachment receptor. J. Virol. 72:7357–7366.

PAGE 133

124 Lin, T., Cavarelli, J., Johnson, J.E. 2003. Evidence for assembly-dependent folding of protein and RNA in an icosahedr al virus. Virology. 314, 26-33. Lotery, A.J., Yang, G.S., Mullins, R.F., Ru ssel, S.R., Schmidt, M., Stone, E.M., Lindbloom, J.D., Chiorini, J.A., Kotin, R.M., Davidson, B.L. 2003. AdenoAssociated Virus Type 5: Transduction Efficiency and Cell-Type Specificity in the Primate Retina. Hum. G. Thrpy. 14:1663-1671. Lukashov, V.V., Goudsmit, J. 2001. Evolutionary Relationships among Parvoviruses: Virus-Host Coevol ution among Autonomous Primate Parvoviruses and Links between Adeno-Associated and Avian Parvoviruses. J. VIrol. 75:6, 2729-2740. Mayor, H. D., Melnick, J. L. 1966. Small deoxyr ibonucleic acid-containing viruses (picornavirus group). Nature. 210:331–332. McKenna, R., Xia, D., Willingm ann, P., Ilag, L.L., Ross man, M.G. 1992. Structure Determination of the Bacteriophage ~oX 174. Acta Cryst. B48, 499-511. McKenna, R., Olson, N. H., Chipman, P. R., Baker, T. S ., Booth, T. F., Christensen, J., Aasted, B., Fox, J.M. , Bloom, M. E., Wolfinbarger, J. B., Agbandje-McKenna, M. 1999. Three-dimensional structure of Aleutian mink disease parvovirus: implications fo r disease pathogenicity. J. Virol. 73:6882–6891. McPherson, R.A., Rose, J.A. 1983. Struct ural Proteins of Adenovirus-Associated Virus: Subspecies and Their Relat edness. J. Virol. 46:2, 523-529. Mizukami, H., Young, N. S., Brown, K. E. 1996. Adeno-associated virus type 2 binds to a 150-kilodalton cell membrane gl ycoprotein. Virology. 217: 124– 130. Mohammadi, M., Yeh, B.K., Yayon, A., Linhardt, R.J., Ibrahimi, O.A., Eliseenkova, A.V., Schlessinger, J., Pl otnikov, A. N. 2000. Crystal Structure of a Ternary FGF-FGFR-Heparin Complex Reveals a Dual Role for Heparin in FGFR Binding and Dimer ization. Mol. Cell. 6:743-750. Mori, S., Wang, L., Takeuchi , T., Kanda, T. 2004. Tw o novel adeno-associated viruses from cynomolgus monkey: pseudotyping characterization of capsid protein. Virology. 330, 375-383. Moskalenko, M., Chen, L., Van Roey, M., Donahue, B. A ., Snyder, R. O., McArthur, J. G., Patel, S. D. 2000. Epitope mapping of human anti-adeno associated virus type 2 neutralizing antibodies: implications for gene therapy and virus structure. J. Virol. 74:1761–1766.

PAGE 134

125 Muzyczka, N., Berns, K. I. 2001. Parvoviridae: the viru ses and their replication, D. M. Knipe et al. (ed.), Fields viro logy. Lippincott, Williams and Wilkins, Philadelphia, Pa. 4:2327–2360. Padron, E., Bowman, V., Kaludov, N., Go vindasamy, L., Levy, H.C., Nick, P., McKenna, R., Muzyczka, N., Chiorini , J.A., Baker, T.S., & AgbandjeMcKenna, M. 2005. Structure of AdenoAssociated Virus Type 4. J.Virol. 79:8, 5047-5058. Park, P. W., Reizes, O., Bernfield, M. 2000. Cell surf ace heparan sulfate proteoglycans: selective regulators of li gand-receptor encounters. J. Biol. Chem. 275:29923–29926. Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M., Meng, E.C., Ferrin, T.E. 2004. "UCSF Chimera A Visualization System for Exploratory Research and Analysis. " J. Comput. Chem. 25:1605-1612 . Prasad, K.R., Trempe, J.P. 1995. T he Adeno-Associated Virus Rep78 Protein Is Covalently Linked to Viral DNA in a Preformed Virion. Virology. 214:360370. Qing, K., Mah, C., Hansen, J., Zhou, S., Dwarki, V., Srivastava, A. 1999. Human fibroblast growth factor receptor 1 is a coreceptor for infection by adenoassociated virus 2. Nat. Med. 5:71–77. Qiu, J., Brown, K. E. 1999. A 110-kDa nuclear shuttle protein, nucleolin, specifically binds to adeno-associ ated virus type 2 (AAV-2) capsid. Virology. 257:373–382. Qiu, J., Brown, K. E. 1999. Integr in alphaVbeta5 is not involved in adenoassociated virus type 2 (AAV2) infe ction. Virology. 264:436–440. Qiu, J., Mizukami, H., Br own, K. E. 1999. A deno-associated virus 2 coreceptors? Nat. Med. 5:467–468. Qiu, J., Handa, A., Kirby, M., Brown, K. E. 2000. The interaction of heparin sulfate and adeno-associated viru s 2. Virology. 269:137–147. Qiu, J., Nayak, R., Tullis, G.E. & Pintel , D.J. 2002. Characterization of the Transcription Profile of Adeno-Associ ated Virus Type 5 Reveals a Number of Unique Features Compared to Previously Characterized AdenoAssociated Viruses. J. Virol. 76:24, 12435-12447. Qiu, J., Nayak, R., Pintel, D.J. 2004. Alternative Polyadenylation of AdenoAssociated Virus Type 5 RNA within an Internal Intron Is Governed by both a Downstream Element within the In tron 3 Splice Acceptor and an Element Upstream of the P41 Initiation Si te. J. Virol. 78:1, 83-93.

PAGE 135

126 Rabinowitz, J.E., Xiao, W., Samulski, R. J. 1999. Insertional Mutagenesis of AAV2 Capsid and the Production of Recombinant Virus. Virology. 45:265, 274-285. Rabinowitz, J. E., Samul ski, R. J. 2000. Build ing a better vector: the manipulation of AAV virions. Virology. 278:301–308. Rabinowitz, J.E., Rolling, F. , Li, C., Conrath, H., Xiao, W., Xiao, X., Samulski, R.J. 2002. Cross-Packagi ng of a Single Adeno-Associated Virus (AAV) Type 2 Vector Genome into Multiple AAV Serotypes Enables Transduction with Broad Specificity. J. Virol. 76:2, 791-801. Rabinowitz, J.E., Bowles, D.E., Faust, S. M., Ledford, J.G., Cunningham, S.E., Samulski, R.J. 2004. Cross-Dressing t he Virion: the Transcapsidation of AdenoAssociated Virus Serotypes Functionally Defines Subgroups. J. Virol. 78:9, 4421-4432. Reguera, J., Grueso, E., Carreira,A., Sa´n chez-Marty´nez, C., Almendral, J.M. Mateu, M.G. 2005. Functional Rele vance of Amino Acid Residues Involved in Interactions with Ordered Nu cleic Acid in a Spherical Virus. J. Biol. Chem. 280, 17969–17977. Ried, M.U., Girod, A., Leike, K., Buning, H., Hallek, M. 2002. Adeno-Associated Virus Capsids Displaying Immuno globulin-Binding Domains Permit Antibody-Mediated Vector Retargeting to Specific Cell Surface Receptors. J. Virol. 76:9, 4559-4566. Rodney Brister, J., Muzyczka, N. 1999. Rep-Mediated Nicking of the AdenoAssociated Virus Origin Requires Two Biochemical Activities, DNA Helicase Activity and Transesterification. J. Virol. 73:11, 9325-9336. Rudnick, J., Bruinsma, R. 2005. Research Article 135 5: Icosahedral Packing of RNAViral Genomes. Phys. Rev. Lett. 94, 038101. Rutledge, E. A., Halbert, C. L., Russell, D. W. 1998. Infectious clones and vectors derived from adeno-associated virus (AAV) serotypes other than AAV type 2. J. Virol. 72:309–319. Sanlioglu, S., Benson, P. K., Yang, J., Atkinson, E. M., Reynolds, T., Engelhardt, J. F. 2000. Endocytosis and nuclear trafficking of adeno-associated virus type 2 are controlled by rac1 and phosphat idylinositol-3 kinase activation. J. Virol. 74:9184–9196.

PAGE 136

127 Shi, W., Arnold, G. S., Bartlett, J. S. 2001. Insertional mutagenesis of the adeno-associated virus type 2 (AAV2) capsid gene and generation of AAV2 vectors targeted to alternative cell-s urface receptors. Hum. Gene Ther. 12:1697–1711. Simpson, A. A., Chandrasekar, V., H ebert, B., Sullivan, G., Rossmann, M. G., Parrish, C. R. 2000. Host range and variability of calcium binding by surface loops in the capsids of canine and feline parvoviruses. J. Mol. Biol. 300:597–610. Simpson, A. A., Hebert, B., Sullivan, G. M., Parrish, C., Zadori, R. Z., Tijssen, P.Rossmann, M. G. 2002. The structur e of porcine parvovirus: comparison with related viruses. J. Mol. Biol. 315:1189–1198. Simpson, A.A., Chipman, P.R., Baker, T.S., Tijssen, P., Rossmann, M.G. DNA genome is visible in the crystal structure, but it is not as well defined as in the vertebrate parvoviruses. Structure. 6:1355-1367. Snyder, R.O., Im, D., & Muzyczka, 1990. Ev idence for Covalent Attachment of the Adeno-Associated Virus (AAV) R ep Protein to the Ends of the AAV Genome. J. Virol. 64:12, 6204-6213. Speir, J.A., Munshi, S., Wang, G., Baker, T.S., Johnson, J.E. 1995. Structures of the native and swollen forms of co wpea chlorotic mottle virus determined by X-ray crystallography and cryo-electr on microscopy. Structure. 3:63-78. Srivastava, A., Lusby, E. W., Berns, K. I. 1983. Nucleotide sequence and organization of the adeno-associated virus 2 genome. J. Virol. 45:555– 564. Summerford, C., Samulski, R. J. 1998. Membrane-associated heparin sulfate proteoglycan is a receptor for adeno-asso ciated virus type 2 virions. J. Virol. 72:1438–1445. Summerford, C., Bartlett, J. S., Samulski , R. J. 1999. Alph aVbeta5 integrin: a co-receptor for adeno-associated virus ty pe 2 infection. Nat. Med. 5:78– 82. Tihova, M., Dryden, K.A., Le, T.L., Ha rvey, S.C., Johnson, J.E., Yeager, M., Schneemann, A. 2004. Nodavirus Coat Protein Imposes Dodecahedral RNA Structure Independent of Nucleotide Sequence a nd Length. J. Virol. 78:6, 2897-2905. Trempe, J. P., Carter, B. J. 1988. Alternate mRNA splicing is required for synthesis of adeno-associated virus VP1 ca psid protein. J. Virol. 62:3356– 3363.

PAGE 137

128 Tsao, J., Chapman, M. S., A gbandje, M., Keller, W., Smit h, K., Wu, H., Luo, M., Smith, T. J., Rossmann, M. G., Copans, R. W., Parrish, C. R. 1991. The three-dimensional structure of ca nine parvovirus and its functional implications. Science. 251:1456–1464. Tsurutal, H., Reddy, V.S., Wi koff, W.R., Johnson, J.E. 1998. Imaging RNA and Dynamic Protein Segments with Low-re solution Virus Crystallography: Experimental Design, Data Processing and Implications of Electron Density Maps. J.Mol. Biol. 284(mb982231), 1439-1452. Venkataram Prasad, B.V., Ma tson, D.O., Smith, A.W. 1994. Three dimensional Structure Of Calicivirus. J. Mol. Biol. 240, 256-264. Verma, I.M., Weitzman, M.D. 2005. GENE THERAPY: Tw enty-First Century Medicine. Annu. Rev. Biochem. 74:711-738. Wagner, J. A., Reynolds, T., Moran, M. L., Moss, R. B., Wine, J. J., Flotte, T. R., Gardner, P. 1998. Efficient and pers istent gene transfer of AAV-CFTR in maxillary sinus. Lancet. 351:1702–1703. Walters, R.W., Yii, S.P., Ke shavjee, S., Brown, K.E., We lsh, M.J., Chiorini, J.A., Zabner, J. 2001. Binding of Adeno-associated Virus Type 5 to 2,3 Linked Sialic Acid Is Required for Gene Trans fer. J. Biol. C hem. 276:23, 2061020616. Walters, R. W., Pilewski, J. M., Chiorini, J. A., Zabner , J. 2002. Secreted and transmembrane mucins inhibit gene tr ansfer with AAV4 more efficiently than AAV5. J. Biol. Chem. 277:23709–23713. Walters, R.W., Agbandje-McKenna, M., Bo wman, V.D., Moninger , T.O., Olson, N.H., Seiler, M., Chiorini, J.A., Baker, T.S., Zabner, J. 2004. Structure of Adeno-Associated Virus Serotype 5. J. Virol. 78:7, 3361-3371. Wang, Z., Tong, Z., Qiao, C ., Zhou, L., Wang, B., Zhang, J., Chen, C., Li, J., Xiao, X. 2005. Adeno-associated virus stereotype 8 efficiently delivers genes to muscle and heart. Nature Biotech. 23:3, 321-328. Wistuba, A., Kern, A., Weger, S., Grimm, D., Kleinschmidt, J. A. 1997. Subcellular compartmentalizatio n of adeno-associated virus type 2 assembly. J. Virol. 71:1341–1352. Wistuba, A., Weger, S., Kern, A., Kleinsch midt, J. A. 1995. Intermediates of adeno-associated virus type 2 assembly: i dentification of soluble complexes containing Rep and Cap proteins . J. Virol. 69:5311–5319.

PAGE 138

129 Wobus, C.E., Hugle-Dorr, B., Girod, A., Petersen, G., Hallek , M., Kleinschmidt, J.A. 2000. Monoclonal Antibodies against the Adeno-Associated Virus Type 2 (AAV-2) Capsid: Epitope Mappi ng and Identification of Capsid Domains Involved in AAV-2–Cell Inte raction and Neutralization of AAV-2 Infection. J. Virol. 74:19, 9281-9293. Wu, P., Xiao, W., Conlon, T., Hughes, J., Agbandje-McKenna, M., Ferkol, T., Flotte, T., Muzyczka, N. 2000. Mutati onal analysis of the adeno-associated virus type 2 (AAV2) capsid gene and c onstruction of AAV2 vectors with altered tropism. J. Virol. 74:8635–8647. Xiao, W., Warrington, K. H., Heari ng, P., Hughes, J., Muzyczka, N. 2002. Adenovirus-facilitated nuclear transloca tion of adeno-associated virus type 2. J. Virol. 76:11505–11517. Xie, Q., Chapman, M. S. 1996. Canine parvovirus caps id structure, analyzed at 2.9 Å resolution. J. Mol. Biol. 264:497–520. Xie, Q., Bu, W., Bhatia, S., Hare, J ., Somasundaram, T., Azzi, A., Chapman, M.S. 2002. The atomic structure of adeno-associat ed virus (AAV-2), a for human gene therapy. PN AS. 99:16, 10405-10410. Xu, J., Ma, C., Bass, C., Terwilliger, E.F. 2005. A combination of mutations enhances the neurotropism of AAV-2. Virology. 341, 203-214. Yang, G.S., Schmidt, M., Yan, Z., Lindbloom, J.D., Harding, T.C., Donahue, B.A., Englehardt, J.F., Kotin, R., Davids on, B.L. 2002. Virus-Mediated Transduction of Murine Retina with AdenoAssociated Virus: Effects of Viral Capsid and Genome Size. J. Virol. 76:15, 7651-7660. Zabner, J., Seiler, M., Walter s, R., Kotin, R.M., Fulgeras, W., Davidson, B.L., Chiorini, J.A. 2000. Adeno-Associ ated Virus Type 5 (AAV5) but Not AAV2 Binds to the Apical Surfaces of Airway Epithelia and Facilitates Gene Transfer. J. Viro l. 74:8, 3852-3858. Zadori, Z., Szelei, J., Lacoste, M. C., Li, Y., Gariepy, S., Raymond, P., Allaire, M., Nabi, I. R., Tijssen, P. 2001. A viral phospholipase A2 is required for parvovirus infectivity. Dev. Cell. 1:291–302. Zlotnick, A. 1994. To build a virus capsid. An equilibrium model for self assembly of polyhedral protein comp lexes. J. Mol. Biol. 241, 59. Zolotukhin, S., Porter, M., Zolotukhin, I., Sa kai, Y., Loiler, S., Fraites, T. J., Jr., Chiodo, V. A., Phillipsberg, T., Muzyczka, N., Hauswirth , W. W., Flotte,T.R., Byrne, B. J., Snyder, R. O. 2002. Production and Purification of serotypes 1, 2 and 5 recombinant adeno-associated vi ral vectors. Methods Enzymol. 28:158–167.

PAGE 139

130 Zolotukhin, S., Byrne, B. J., Mason, E. , Zolotukhin, I., Potter, M., Chesnut, K. Summerford, C., Samulski, R. J. , Muzyczka, N. 1999. Recombinant adeno-associated virus purification using novel methods improves infectious titer and yield. Gene Ther. 6:973–985. Zolotukhin, S., Potter, M., Hauswirth, W., Guy, J., Muzyczka, N. 1996. A humanized green fluorescent protei n cDNA adapted for high-level expression in mammalian cells. J. Virol. 70:4646–4654.

PAGE 140

131 BIOGRAPHICAL SKETCH Hazel Christina Levy was born in Milwaukee, Wisconsin on July 3rd 1972. Her Mother is Shirley Murray and her older brother is Jason Levy. She graduated from Deland Senior Hi gh School Magna Cum Laude in June, 1990 and entered the University of Florida as a college freshman in August, 1990 where she studied microbiology. She gradu ated with her Bachelor of Science in May, 1996 and entered graduate school in A ugust, 1996. As a graduate in the Department of Entomology and Nematol ogy, she studied Insect Mitochondrial Genetics under Dr. James E. Maruniak and was a teaching assistant in Insect Physiology with Dr. James Nation. She married to Noah John Kaufman December 28, 1996. In December , 1998 she graduated with a Master of Science. She then began to teach Biology and Evolution at S anta Fe Community College and was a laboratory technician with Dr. Philip Laipis. In August, 2000 Hazel entered graduate school in the D epartment of Biochemistry where she pursued her dissertation research wit h Dr. Mavis Agbandje-McKenna and Dr. Nicholas Muzyczka.