Conjugate-Based Targeting of Recombinant Adeno-Associated Virus Type 2 Vectors by Using Avidin-Linked Ligands

Abstract

The development of targeted vectors, capable of tissue-specific transduction, remains one of the important aspects of vector modification for gene therapy applications. Recombinant adeno-associated virus type 2 (rAAV-2)-based vectors are nonpathogenic, have relatively low immunogenicity, and are capable of long-term transgene expression. AAV-2 vectors bind primarily to heparan sulfate proteoglycan (HSPG), a receptor that is present in many tissues and cell types. Because of the widespread expression of HSPG on many tissues, targeted transduction in vivo appears to be limited with AAV-2 vectors. Thus, development of strategies to achieve transductional targeting will have a profound benefit in the future application of these vectors. We report here a novel conjugate-based targeting method to enhance tissue-specific transduction of AAV-2-based vectors. The present report utilized a high-affinity biotin-avidin interaction as a molecular bridge to cross-link purified targeting ligands, produced genetically as fusion proteins to core-streptavidin, in a prokaryotic expression system. Conjugation of the bispecific targeting protein to the vector was achieved by biotinylating purified rAAV-2 without abolishing the capsid structure, internalization, and subsequent transgene expression. The tropism-modified vectors, targeted via epidermal growth factor receptor (EGFR) or fibroblast growth factor 1α receptor (FGFR1α), resulted in a significant increase in transduction efficiency of EGFR-positive SKOV3.ip1 cells and FGFR1α-positive M07e cells, respectively. Further optimization of this method of targeting should enhance the potential of AAV-2 vectors in ex vivo and in vivo gene therapy and may form the basis for developing targeting methods for other AAV serotype capsids.

Adeno-associated virus (AAV)-based vectors are becoming increasingly popular for gene therapy of human diseases (3, 16, 22, 33, 34). Despite the potential for long-term expression, genomic integration, and low immunogenicity, the transduction efficiency of AAV type 2 (AAV-2) vectors varies significantly among cell types. For example, the cells of muscle and brain are highly permissive for AAV-2 infection. However, certain human megakaryocytic cell lines and primary cells of hematopoietic origin show inconsistencies in infection (8, 15, 19, 20, 23), which may be due to either insufficient receptor numbers, an absence of optimal amounts of intracellular factors necessary for AAV second-strand synthesis, or a combination of both (2, 6, 7, 10, 30). Thus, improvements to overcome these limitations will have a positive impact on the application of AAV vectors in gene therapy. Whereas recent studies on intracellular events conducive for AAV transgene expression have identified the roles of several host cell factors (10, 14, 30), in order to overcome the limitations of receptor-mediated AAV infection, both genetic and nongenetic approaches are currently being developed (19, 24, 26, 37, 38). Although genetic modification of AAV capsid may be a preferred means of achieving vector retargeting, the size limitation of AAV capsid as well as limited availability of targeting ligands pose major concerns. Thus, development of high-efficiency and stable methods of retargeting will broaden the potential utility of AAV-based vectors in the future.

The avidin-biotin complex represents the highest-affinity interaction between a protein and a ligand known in nature. This property formed the basis for the establishment of many diagnostic, biotechnological, and therapeutic applications using avidin-biotin conjugates (36). In the present work, we report a novel conjugate-based retargeting of AAV-2 vector to cells by using a high-affinity avidin-biotin molecular bridge. A recombinant bispecific protein containing sequences of either human epidermal growth factor (EGF) or human fibroblast growth factor 1α (FGF1α) as a target cell ligand was genetically fused to core-streptavidin and affinity purified following production in a prokaryotic expression system. The purified protein was conjugated to biotinylated recombinant AAV-2 (rAAV-2), encoding luciferase, and used to infect either EGF receptor (EGFR)-positive SKOV3.ip1, a human ovarian cancer cell line, or EGFR-negative MB-453, a human breast cancer cell line. The results showed a significant enhancement of transgene expression only in SKOV3.ip1 cells, indicating the target cell-specific transduction of rAAV-2 through an alternate receptor. Validation of the strategy by using FGF1α also resulted in the transduction of tropism-modified vector in AAV-2 infection-resistant M07e cells which had been stably transfected with FGF1α receptor (FGFR1α) (25). Further development of this high-affinity, conjugate-based retargeting of AAV may prove beneficial in both ex vivo and in vivo human gene therapy.

MATERIALS AND METHODS

Cells and viruses.

The human embryonic kidney cell line 293 was obtained from the American Type Culture Collection. The EGFR-positive human ovarian cancer cell line SKOV3.ip1 was obtained from David Curiel (The University of Alabama at Birmingham). The EGFR-negative MB-453 cell line was obtained from Allen Wells (University of Pittsburgh, Pittsburgh, Pa.). The FGFR1α-positive human megakaryocytic leukemia cell line, M07e was obtained from Arun Srivastava, Indiana University School of Medicine, Indianapolis. Construction of the rAAV-2 plasmid encoding luciferase and production of high-titer virus have been described earlier (19, 23).

Enzymes, antibodies, and other molecular biology reagents.

Restriction endonucleases and other DNA-modifying enzymes were purchased from New England Biolabs (Beverly, Mass.) or Promega Corporation (Madison, Wis.). Mouse anti-human EGF monoclonal antibody, rabbit polyclonal anti-human FGF1α antibody, and a mouse monoclonal antibody recognizing the c-Myc epitope were purchased from Sigma Chemical Co. (St. Louis, Mo.). Horseradish peroxidase (HRP)-conjugated secondary antibodies were purchased from Pharmacia, and an ECL detection kit was purchased from Amersham. N-Hydroxysuccinimide ester (NHS)-water-soluble biotin and biotinylated β-galactosidase (β-gal) were purchased from Vector Laboratories (Burlingame, Calif.).

Construction of prokaryotic expression vectors encoding bispecific fusion protein.

A prokaryotic expression vector containing the coding sequences of EGF, fused to core-streptavidin, was constructed by PCR amplification of the EGF sequence from the plasmid pFB1CArs1EGF (obtained from David Curiel, The University of Alabama at Birmingham) by using the forward primer 5′-AGTTCAGCTGCAGAATAGTGACTCTGAATGTCCCC-3′ and reverse primer 5′-CACCGGATCCTGCACCTCCGCGCAGTTCCCACCA-3′. The amplified product was digested with restriction enzymes PvuII and BamHI and cloned in the vector pSTE2-215 yol (GenBank accession number Y18290; kindly provided by Stefan Dübel, University of Heidelberg, Heidelberg, Germany). The resultant plasmid, pEGF-ST, contained a periplasmic leader sequence upstream of the EGF gene under the control of the T7 promoter. Sequences of a 9E10 epitope recognized by a c-Myc antibody and a His₅ portion were retained at the 3′ end of the EGF-streptavidin (EGF-ST) fusion gene for immunodetection and affinity purification, respectively. Similarly, the coding sequence of FGF1α was amplified by PCR and cloned as an FGF-streptavidin (FGF-ST) fusion protein. The primer sequences for the amplification of FGF1α were as follows: forward primer,5′-CAGCTCAGCCGGCCATGGCGCAAGTTCAGCTGCAGAATGCTAATTACAAGAAC-3′; reverse primer, 5′-TCCAGCGGCCGCCCGATCAGAAGAGACTGG-3′. Following ligation, the DNA was transformed into Escherichia coli JM109, and positive clones were identified by restriction digestion and expression of fusion proteins.

Expression and purification of recombinant fusion protein.

E. coli JM109 cells were transformed with the recombinant expression plasmid, and the culture was grown overnight in Luria-Bertani medium containing 100 mM glucose. For large-scale production of the fusion protein, the overnight culture was inoculated into fresh Luria-Bertani medium (1:20 by volume) containing 100 mM glucose and grown at 37°C for 3 to 4 h till the optical density at 600 nm (OD₆₀₀) reached 0.6 to 0.8. Next, the cultures were induced with 20 μM IPTG (isopropyl-β-d-thiogalactopyranoside) and grown for an additional 5 h at 30°C. The cells were collected by centrifugation at 3,500 × g for 15 min at 4°C. Subsequent purification steps were performed at 4°C. The medium was completely removed, and the pellet was resuspended in 1/100 volume (to that of the original culture) of a buffer containing 50 mM Tris-HCl and 20% sucrose (pH 8.0). The suspension was left on ice with occasional mixing for 30 min and then centrifuged for 30 min at 10,000 rpm. The supernatant was saved, and the pellet was resuspended with same volume of 5 mM MgSO₄ and left on ice with occasional mixing for an additional 30 min. The suspension was centrifuged at 13,800 × g, and the clear supernatant was added to that obtained above and stored as soluble periplasmic extract. The pellet was resuspended in 1/50 volume (to that of the original culture) of a buffer containing 6 M guanidine-HCl and 100 mM Tris (pH 7.0) and left rotating overnight. The next day, following centrifugation at 13,000 rpm for 30 min, the supernatant containing the fusion protein was affinity purified with an Ni-nitrilotriacetic acid (Ni-NTA) column (Qiagen). The column was equilibrated with a buffer containing 50 mM Tris-Cl, 100 mM NaCl, and 20 mM imidazole (pH 7.0). Washing of the column was done with 3 void volumes of the same buffer, and elution of the bound fusion protein was achieved by increasing the concentration of imidazole to 250 mM. The eluted fusion protein was dialyzed against a buffer containing 100 mM Tris and 400 mM l-arginine for at least 16 h with two changes of the dialysis buffer. The dialyzed protein was centrifuged at 13,000 rpm to remove insoluble material, and the clear supernatant was stored frozen in aliquots at −20°C. The soluble periplasmic extract and the renatured protein, obtained from the inclusion bodies, were not mixed.

Western blot analysis.

Approximately 10 μl of the purified fusion protein was separated in either reducing or nonreducing 12% polyacrylamide gel as described before (12). The separated proteins were either stained using Coomassie blue or transferred to polyvinylidene difluoride membranes. Immunodetection of the fusion proteins was carried out with either a mouse monoclonal antibody for human EGF (for EGF-ST) or a rabbit polyclonal antibody (for FGF-ST). Secondary antibodies were goat anti-mouse or rat anti-rabbit antibodies, respectively, conjugated to HRP. Detection of the bands was done using ECL chemiluminescent substrate. Western blot analysis of the biotinylated AAV was performed using a streptavidin-conjugated antibody coupled to HRP.

Production of tropism-modified rAAV-2.

Packaging of rAAV-2 encoding luciferase were done by plasmid transfection in 293 cells and gradient purification of the virions as described earlier (19, 20, 23). The particle titer of the purified virus was determined by genomic slot blot analysis (19, 20). Vector biotinylation was performed by incubating 10¹¹ particles of rAAV-2-luc in HEPES buffer (pH 7.3) containing NHS-biotin. The ratio of the capsid protein to biotin concentrations was maintained at 10:1, and the approximate concentration of AAV capsid was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis with known amounts of bovine serum albumin. The vector was initially resuspended in HEPES buffer, and the biotin solution was added slowly. The mixture was incubated at room temperature for 45 min with occasional stirring. Following this, free biotin was removed by spin dialysis with Centricon 30 filters. Biotinylation was confirmed by both immunoblot analysis and electron microscopy as described below.

Electron microscopy.

To visualize the biotinylation of rAAV-2, following dialysis, the biotinylated virus was conjugated to 5-nm-diameter colloidal gold-labeled streptavidin in a buffer containing 50 mM Tris, 0.1% bovine serum albumin, and 125 mM NaCl (pH 7.4) at 4°C overnight. The conjugated virus was then dialyzed against phosphate-buffered saline (PBS), and approximately 10⁹ particles of the vector were negatively stained with 2% phosphotungstic acid and observed under a transmission electron microscope (Philips 400).

Conjugation of bispecific fusion protein with biotinylated β-gal, receptor-mediated endocytosis, and enzyme assays.

Ten micrograms of biotinylated β-gal was incubated with 2.5 to 25 μg of purified, refolded EGF-ST or FGF-ST at room temperature in HEPES buffer for 1 h. Endocytosis of EGF-ST- or FGF-ST-conjugated biotinylated β-gal to EGFR or FGFR-positive and -negative cells was determined by incubating the cells with the conjugate at 37°C in a minimal amount of Opti-MEM (100 μl/well in a 24-well tissue culture plate) for 30 min. The cells were washed, supplemented with the respective complete medium containing fetal calf serum and growth factors, and cultured for additional 16 h prior to in situ detection as described earlier (19, 21).

EGFR- or FGFR-positive and -negative cells were plated in 24-well plates, and infection of tropism-modified or unmodified rAAV-2-luc was carried out in 100 μl of Opti-MEM for 2 to 3 h at 37°C in a CO₂ incubator. Transduction of M07e cells was performed in sterile 5-ml round-bottom polypropylene tubes. Following transduction, the cells were washed four times with PBS, resuspended in the respective complete medium, and cultured for 48 h. Luciferase activity was determined from cell lysates and expressed as relative light units (RLU), normalized to protein content of each sample as described previously (23).

RESULTS

Construction of prokaryotic expression vectors encoding genetic fusions of targeting ligands.

For efficient production of a fusion protein with binding affinity to biotin and a target cell-specific receptor, we constructed a plasmid containing sequences of core-streptavidin, genetically fused to the EGF-coding region, in a prokaryotic expression vector under the control of the T7 promoter to yield a plasmid, pEGF-ST (Fig. 1). We have chosen EGF as a prototypic targeting ligand because previous studies using adenoviral vectors demonstrated the efficacy of targeting with EGFR as an alternate receptor (5). Further, EGF is a small protein of only 51 amino acids, which will minimize complexities of folding. Core-streptavidin has also been successfully used in targeting, and studies have shown the affinity of core-streptavidin to be similar to those of the full-length molecule in biotin binding and tetramerization (11). To achieve proper folding of the two proteins, which is vital for the function of each moiety, an eight-amino acid linker (Gly₄, Ala₃, and Ser₁) was included between EGF and core-streptavidin. At the amino-terminal end of the fusion gene, a leader sequence was incorporated in frame for secretion of the fusion protein into the periplasmic space, and at the carboxy-terminal, a His₅ tag was included for affinity purification. An additional prokaryotic expression vector, containing a fusion of core-streptavidin with human FGF1α, was similarly constructed by replacing the EGF sequences from the plasmid pEGF-ST. The coding region of the FGF1α was amplified by PCR with primers containing restriction enzyme sites to facilitate direct subcloning. The initiation codon, ATG, was changed to AAT (Asn) in order to avoid adjacent open reading frames. Selection of positive clones expressing the fusion protein was performed by transforming the recombinant plasmids into bacterial strain JM109 and growing them as 5-ml cultures. For each type of fusion, 12 to 14 clones were initially screened by miniculture analysis. When the OD₆₀₀ reached 0.6, cultures from each clone were divided into two; one portion was induced with 100 μM IPTG, and the other was maintained as uninduced. Both the induced and uninduced cultures were grown for an additional 4 h at 30°C. Following this, the bacterial pellets from both uninduced and induced cultures were collected and lysed with Tris-Cl buffer containing 8 M urea. The crude lysates were separated by SDS-PAGE and stained with Coomassie blue. From the positive clones expressing the fusion protein, DNA was isolated and sequenced by automated sequencing.

FIG. 1.

Strategy outlining targeting of rAAV-2 by using avidin-linked ligands. A targeting ligand (EGF or FGF1) is cloned in the 5′ region of the core-streptavidin gene in frame by using a linker for proper folding. A His₅ tag is added, also in frame, at the 3′ end of the fusion protein for affinity purification. Expression of the fusion protein is achieved by IPTG induction in E. coli, and the fusion protein is purified through Ni-NTA column. The purified bispecific fusion protein is conjugated to biotinylated rAAV-2 containing a transgene and used in targeting of a specific cell type(s).

Expression and purification of bispecific targeting ligands in a bacterial system.

Large-scale production of the fusion proteins was performed by growing 2- to 4-liter cultures of each. Inoculation of the positive clones for 50- to 100-ml overnight cultures was done from glycerol stocks of the bacteria containing the correct plasmid rather than freshly transforming DNA each time, to avoid mutations. The large-scale cultures were grown by inoculating 1/20 volume of the overnight culture and grown initially at 37°C until the OD₆₀₀ reached 0.5 to 0.8, following which IPTG was added and the cultures were grown at 30°C for 5 h more. IPTG induction was done at a concentration of 20 to 30 μM to maximize periplasmic secretion of the fusion protein and minimize the aggregation of inclusion bodies. Harvesting and subsequent steps were carried out at 4°C. The periplasmic extract was obtained by osmotic lysis with Tris buffer and MgSO₄. The soluble fraction was separated by high-speed centrifugation and pooled, following which the pellet containing the inclusion bodies was resuspended in Tris buffer containing 6 M guanidine-HCl overnight. After high-speed centrifugation, the supernatant was checked by nondenaturing PAGE with known concentrations of bovine serum albumin in adjacent lanes to empirically determine the amount of fusion protein present in each fraction. Based on the presence of His tag at the carboxy-terminal region of the fusion protein, affinity purification of the fusion protein was performed with Ni-NTA resin in a column. Approximately 5 to 8 mg of fusion protein was obtained in total (both periplasmic fraction and inclusion body) from a 2-liter culture. At least 3 volumes of Ni-NTA resin, necessary for binding with the amount of fusion protein, were used for each purification. Elimination of nonspecific binding and increases in specific binding of the His tag protein were achieved by including 20 mM imidazole in the binding buffer. After washing the of unbound material, the bound fusion protein was eluted by increasing the imidazole concentration to 250 mM. The recovery of purified fusion protein was determined by testing the fractions before and after the column purification, which normally ranged between 40 and 60%. The purified fusion protein was dialyzed overnight in Tris buffer containing 400 mM l-arginine, with at least two changes of the buffer. Immunoblot analysis of the renatured fraction indicated that approximately 30% of the dialyzed protein was in tetrameric form, 10% was in dimeric form, and the remaining was in monomeric form. A representative analysis of the purified bispecific fusion protein (EGF-ST) is given in Fig. 2. Similar results were obtained with the fusion protein FGF-ST (data not shown). The purified fractions were filtered through sterile 0.2-μm-pore-size filters and stored in aliquots at −20°C.

FIG. 2.

Affinity purification of a streptavidin-EGF fusion protein. A prokaryotic vector containing the core streptavidin-EGF fusion gene was expressed in bacteria, and the fusion protein was affinity purified and further solubilized for optimal refolding. (A) SDS-PAGE analysis of the total cell lysate (lane 2) and affinity-purified, refolded streptavidin-EGF (lane 3) by Coomassie blue staining. Lane 1, marker proteins. (B) Western blot analysis of the purified streptavidin-EGF with a monoclonal antibody against human EGF. Approximately 30% of the solubilized fusion protein was found to be in a characteristic tetrameric form (t). Monomeric and dimeric forms of the purified fusion protein are designated m and d, respectively.

Functional characterization of bispecific targeting ligand in situ.

Prior to using the purified bispecific fusion protein for vector retargeting, the biological activities of both of the targeting moieties were determined in an in situ assay. To this end, biotinylated β-gal was preincubated with different concentrations (1 to 10 μg) of purified EGF-ST fusion protein at room temperature for 45 min. The conjugate was then incubated with EGFR-positive SKOV3.ip1 cells (5 × 10⁵ cells) in a volume of 200 μl of Opti-MEM at 37°C for 45 min, following which the cells were washed five times in PBS and grown in culture overnight. Control cells were incubated with either the fusion protein alone, biotinylated β-gal alone, or biotinylated β-gal, conjugated to a nonspecific fusion protein containing core-streptavidin to determine the specificity of receptor-mediated endocytosis. The results, shown in Fig. 3, demonstrate that both the core-streptavidin and EGF portions of the purified fusion protein were functional. No endocytosis of biotinylated β-gal was observed when a nonspecific targeting ligand, cloned as a fusion protein with core-streptavidin, was used in the conjugation step (data not shown).

FIG. 3.

Biological activity of the genetic conjugate for EGFR-mediated targeting. EGFR-overexpressing human ovarian cancer cells (SKOV3.ip1) were either mock treated (A), incubated with 10 μg or biotinylated β-gal alone (B), incubated with ∼10 μg of Ni-NTA-purified streptavidin-EGF fusion protein alone (C), or incubated with 10 μg of biotinylated β-gal conjugated with ∼10 μg of purified streptavidin-EGF (D) for 2 h at 37°C. Following the treatments, the cells were washed with PBS and incubated for 24 h at 37°C. In situ detection of β-gal was performed with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside).

Biotinylation of rAAV-2.

Next, we tested if biotinylation of rAAV-2 capsid will allow binding to a targeting ligand through an avidin bridge without compromising the trafficking of the vector through an alternate receptor and subsequent intracellular events, including endosomal transport, nuclear localization, and uncoating, needed for optimal transgene expression. Unlike genetic modification of the capsid structure to include a targeting ligand in specific domains of AAV VP1, -2, or -3 protein, biotinylation of the capsid will result in modification at several domains, which will allow high-affinity conjugation of the avidin-linked targeting ligand. Biotinylation of rAAV-2 was achieved by incubating rAAV-2 with NHS-water-soluble biotin at room temperature for 45 min, following which free biotin was removed by dialysis. The efficiency of biotinylation was confirmed both by Western blot analysis and electron microscopy. The results of Western blot analysis, shown in Fig. 4, confirmed that biotinylation occurred in all three capsid proteins (Fig. 4, lanes 2), and a modest increase in the molecular masses of all three capsid proteins upon biotinylation further suggested stable biotinylation. To determine if random biotinylation of AAV results in the disruption of the icosahedral structure of the capsid and if biotinylation occurred homogenously in the vector particles, we performed electron microscopy. Visualization of gold-streptavidin particles under the electron microscope following binding to biotinylated AAV indicated uniform biotinylation as well as retention of the icosahedral morphology of the capsid compared to unmodified rAAV-2 (Fig. 5). We also carried out an infection assay to determine if biotinylation abolishes infection of the rAAV-2 through the native heparan sulfate proteoglycan (HSPG) pathway. To this end, 10⁸ particles of unmodified or biotinylated rAAV-2-luc were used to infect 293 cells in a 12-well plate for 1 h. Free virus was removed by washing with PBS, and luciferase activity was determined 48 h after transduction as a measure of transgene expression. The results, shown in Fig. 6, clearly indicated a comparable amount of luciferase activity, suggesting that biotinylation of the vector did not abolish infection through the native receptor.

FIG. 4.

Western blot analysis of biotinylated AAV-luc. Approximately 5 × 10⁹ particles of unmodified (lanes 1) or biotinylated (lanes 2) rAAV-2-luc were denatured, separated by PAGE, and transferred to Immobilon filters. Immunodetection was performed using either streptavidin-conjugated HRP alone (A) or an anti-AAV capsid primary antibody followed by an anti-mouse secondary antibody and streptavidin-conjugated HRP (B) prior to the detection step. VP-1, -2, and -3, AAV capsid proteins.

FIG. 5.

Electron micrographs of control (A) and biotinylated (B) AAV-luc. rAAV-2 (10¹⁰ particles) was either mock treated or incubated with 5 μg of NHS-water-soluble biotin per ml in a volume of 50 μl HEPES buffer for 2 h at room temperature. Free biotin was removed by using a Centricon-30 filter, and the virus particles were further incubated with 5-nm-diameter colloidal gold-streptavidin, diluted in Tris-buffered saline, overnight at 4°C. Five microliters was used for visualization under the electron microscope following negative staining with 2% phosphotungstic acid. Magnification, ×53,000. Bar, 50 nm.

FIG. 6.

Luciferase activity in SKOV3.ip1 and MB-453 cells following transduction of unmodified, biotinylated, and EGFR-targeted AAV-luc. SKOV3.ip1 and MB-453 cells were either mock infected (▪) or infected with 10⁸ particles of AAV-luc that was unmodified (▩), biotinylated alone ( ), or biotinylated and conjugated to EGF-streptavidin fusion protein ( ) for 1 h at 37°C. Following infection, free virus was removed by washing with PBS and the cells were incubated for 48 h. Luciferase activity was determined by lysing the cells and expressed as RLU, normalized to protein content of each cell lysate.

EGFR-mediated transduction of tropism-modified rAAV-2.

The next set of experiments was performed to determine the efficacy of conjugate-mediated targeting of rAAV-2 in EGFR-positive and EGFR-negative cells. As a positive cell line, SKOV3.ip1, which is known to express high levels of EGFR, was used, and as a negative cell line, MDA-MB453 cells were chosen (5). Approximately 10⁸ particles of rAAV-2 luciferase, which was either biotinylated or unmodified, were incubated with 2.5 to 25 μg of affinity-purified EGF-ST fusion protein for 45 min at room temperature. Unconjugated fusion protein was removed by centrifugation in 100-kDa-cutoff filters. Infections of SKOV3.ip1 and MB-453 cells with the modified viruses were done in 12-well plates for 45 min at 37°C, and free virus was removed by washing with PBS. Following infection, the cells were incubated for an additional 48 h, lysed, and assayed for luciferase activity. From the results given in Fig. 6, it is apparent that targeted transduction of rAAV-2 by conjugation through a biotin-avidin linker results in significant enhancement of vector transduction only in EGFR-positive SKOV3.ip1 cells and not in EGFR-negative MDA-MB453 cells. The increase in the transduction efficiency was more than 100-fold in SKOV3.ip1 cells. Preincubation of cells with soluble heparin did not abolish infection of EGF-ST-conjugated AAV-luc in SKOV3.ip1 cells alone (data not shown). Both the unmodified AAV-luc and the biotin-EGF-ST-conjugated vector resulted in comparable levels of luc activity in EGFR-negative MB-453 cells, suggesting that biotinylation and subsequent conjugation steps did not abolish natural tropism of the vector to HSPG. Similarly, both unmodified AAV-luc and biotinylated AAV-luc showed comparable luciferase activity, suggesting either that biotinylation of the vector does not result in elimination of native tropism to heparan sulfate or that a significant amount of unbiotinylated vector is present in the preparation. Nonetheless, the results clearly indicate that significant transduction enhancement could be achieved by this method of retargeting.

FGFR-mediated transduction of tropism-modified rAAV-2 in transduction-resistant M07e cells.

To demonstrate further that the conjugate-based approach for retargeting will allow transduction of cells that are resistant to AAV infection, we used FGFR1α-positive M07e cells, which we identified earlier to be resistant to AAV infection (19, 25). A bispecific fusion protein containing FGF1α and core-streptavidin was similarly purified and conjugated to biotinylated rAAV-2-luc. Upon transduction of the tropism-modified virus into FGFR1α-positive M07e cells, significant transduction was observed, based on the luciferase expression (Fig. 7). As expected, there was no luciferase activity in mock-transduced or biotinylated AAV-luc-transduced cells, since these cells are deficient in HSPG receptor.

FIG. 7.

Luciferase activity in FGFR1α-positive M07e cells. Approximately 5 × 10⁴ FGFR-positive M07e cells were either mock infected or infected with 10⁹ particles of AAV-luc that was unmodified, or of biotinylated AAV-luc, or of biotinylated AAV-luc conjugated to 1 μg of FGF-streptavidin fusion protein for 1 h at 37°C. Following infection, free virus was removed by washing with PBS and the cells were incubated for 48 h. Luciferase activity was determined by lysing the cells and expressed as RLU.

DISCUSSION

Strategies to develop targeted AAV-2 vectors have been attempted, so far with modest success. Recently, studies have identified several putative domains on AAV-2 capsid which are amenable for inclusion of chimeric sequences (9, 26, 37, 38). Although these studies have indicated the possibility of developing genetically modified AAV-2 capsids, a major concern is the size limitation. Since AAV is a parvovirus with an icosahedral capsid of ∼25 nm in diameter, incorporation of larger targeting ligands within the capsid could impair the structural configuration of the capsid, which in turn may affect the efficiency of packaging and/or infectivity. To date, only a few smaller peptides have been reported to be cell-specific ligands for targeting, which can be genetically incorporated within AAV capsid. As a nongenetic approach, an immunological targeting performed by chemical conjugation of two antibodies involved lengthy procedures for production and purification of antibodies in large amounts (1). Conjugation of two monoclonal antibodies for bispecific targeting may also involve greater stability problems for in vivo applications. Thus, development of newer methods of targeting is needed to overcome the existing limitations.

Targeted gene transfers using molecular conjugates to specific cell types have been reported to include conjugated plasmids and retrovirus and adenovirus vectors (4). In molecular conjugate vectors, targeting moieties have been attached to vectors mainly through electrostatic interactions or through bispecific monoclonal antibody conjugates. Although these methods have produced efficacious results in vitro, their realistic application in vivo is limited due to lack of stability of the vector-conjugate complex. Similar limitations in monoclonal antibody therapy for tumor patients have been overcome by using avidin-biotin interactions (27). Thus, expanding the utility of high-affinity, stable interaction of the avidin-biotin system to achieve modifications in vector tropism may provide greater efficacy gains. In this regard, recently Smith et al. reported high-efficiency transduction of a biotinylated recombinant adenovirus, conjugated to a biotinylated stem cell factor through an avidin bridge, into c-Kit receptor-positive hematopoietic stem cells, indicating the feasibility of this approach (32).

The targeting strategy that we developed is based on high-affinity streptavidin-biotin interaction. Streptavidin is a tetrameric protein of ∼60 kDa which binds biotin with exceptional affinity (K_d = 10⁻¹⁵ M). The high affinity of biotin for streptavidin has made this pair of molecules very useful for many in vitro and in vivo applications (35). The majority of in vivo avidin-biotin therapeutic applications are presently used for tumor targeting (17, 18, 28, 31, 35). Development of pretargeting strategies by using avidin-biotin interactions of therapeutic molecules and antibodies resulted in increased efficacy gains and greatly minimized the required dose of the therapeutic molecules (39). Targeted immunotherapy of colon adenocarcinoma using a biotinylated anti-CEA monoclonal antibody and a biotinylated drug, neocarzinostatin, resulted in a fivefold increase in the therapeutic efficacy (17). High-efficiency targeting of tumor cells by administration of a monoclonal antibody and a radiolabel by using avidin-biotin interactions has been reported not only in preclinical studies but also in clinical trials (29). This method of tumor targeting has also resulted in high tumor-to-nontumor targeting ratios in addition to reducing background radioactivity of the directly labeled antibody (29). An avidin-biotin system has also been effectively utilized in the delivery of nerve growth factor to brain cells by using transferrin receptors to overcome the blood-brain barrier (13). Thus, the potential utility of avidin-biotin interaction is, in different therapeutic contexts, capable of modulating efficacy, safety, and specificity of targeting.

The results of the present study prove that conjugate-based targeting of rAAV-2 by using genetically produced avidin-linked ligands was efficacious in increasing target cell-specific transduction. It is noteworthy that by this method of targeting, it was not only possible to enhance transduction efficiency of EGFR-targeted AAV to EGFR-positive cells but also to use an FGFR1α-targeted AAV to transduce M07e cells, which are otherwise resistant to AAV infection (19). Previous studies indicated that EGFR abundance and activation inhibited expression from AAV transgenes, possibly through phosphorylation of a single-stranded binding protein (ssD-BP) (14, 24). However, in the present studies, targeting through EGFR does not appear to inhibit AAV-2 transduction. It remains possible that intracellular signaling events following EGFR-mediated entry of the vector are different from those of heparin sulfate receptor-mediated entry. Additionally, variation in the cell types used in these studies may account for such events. The steps in bispecific targeting conjugate production are simple and feasible for large-scale production. Thus, developing targeting strategies based on this interaction should allow in vivo stability of the modified vectors. Although we have not attempted to produce the bispecific targeting protein in a mammalian or baculovirus system, it may be necessary for certain cellular ligands to have posttranslational modifications such as glycosylation for optimal function.

Despite the efficacy of transduction enhancements in a target cell-specific manner, based on the results obtained, it is apparent that under the biotinylation conditions used, the modified rAAV-2 still possesses tropism for the natural receptor. For future in vivo studies, it may be important to block the heparin binding of AAV-2, since several tissues express HSPG in significant proportions. Recent studies have identified heparin binding domains on AAV-2 capsids, and Wu et al. have generated heparin binding-deficient mutants without compromising the packaging ability, genomic content, or titer (37). Utilizing these mutant capsids will provide a significant advantage in overcoming the limitations of HSPG binding. We are presently focusing our efforts to validate the potential of these mutant capsids in this context. These limitations, however, should not affect the potential application of this strategy in ex vivo gene therapy protocols involving isolation and/or expansion of hematopoietic and nonhematopoietic stem cells and transduction of the vector in vitro prior to autologous transfer. In fact, we and others have reported that transduction of AAV-2 vectors to hematopoietic stem cells shows high levels of variation and inconsistency, possibly due to the absence or low levels of HSPG receptors in addition to other possible limitations, including defects in intracellular trafficking and second-strand synthesis (2, 6,7, 10, 30). Nonetheless, the strategy outlined here should be readily usable in the ex vivo protocols, which do not involve the possibility of nontarget cells being exposed directly to the vector. Further, this conjugate-based strategy should be readily adaptable for different targets by cloning new ligands into the hybrid construct to increase the versatility.