Retrovirus Molecular Conjugates

Two significant barriers limit the use of amphotropic retrovirus for human gene transfer protocols: 1) low transduction efficiency in cells with low receptor expression and 2) safety concerns originating from the risk of formation and propagation of replication competent virus in vivo. In principle, if ecotropic retrovirus, which is incapable of infecting human cells, could be transiently modified to effectively transduce human cells, this safety risk could be alleviated. Here we demonstrate that formation of amphotropic retrovirus polylysine molecular conjugates (aMMLV-PL) enhanced gene transfer up to 10-fold in a variety of human cell lines over the equivalent of unconjugated vector (aMMLV). The polylysine modification and formation of ecotropic retrovirus molecular conjugates (eMMLV-PL) permitted effective and stable transduction of different human cell lines as well as primary human bone marrow stroma cells at frequencies of greater than 80%. It is conceivable that this novel ecotropic-based conjugate retrovirus vector could also potentially provide enhanced safety characteristics not only over amphotropic retrovirus vectors but also over genetically tropism-modified recombinant ecotropic vectors. In contrast to genetic modifications, physical or chemical modifications are not propagated. Thus, formation of replication competent eMMLV from conjugates would be self-limited and would not result in virus propagation in humans.

Two significant barriers limit the use of amphotropic retrovirus for human gene transfer protocols: 1) low transduction efficiency in cells with low receptor expression and 2) safety concerns originating from the risk of formation and propagation of replication competent virus in vivo. In principle, if ecotropic retrovirus, which is incapable of infecting human cells, could be transiently modified to effectively transduce human cells, this safety risk could be alleviated. Here we demonstrate that formation of amphotropic retrovirus polylysine molecular conjugates (aMMLV-PL) enhanced gene transfer up to 10-fold in a variety of human cell lines over the equivalent of unconjugated vector (aMMLV). The polylysine modification and formation of ecotropic retrovirus molecular conjugates (eMMLV-PL) permitted effective and stable transduction of different human cell lines as well as primary human bone marrow stroma cells at frequencies of greater than 80%. It is conceivable that this novel ecotropic-based conjugate retrovirus vector could also potentially provide enhanced safety characteristics not only over amphotropic retrovirus vectors but also over genetically tropism-modified recombinant ecotropic vectors. In contrast to genetic modifications, physical or chemical modifications are not propagated. Thus, formation of replication competent eMMLV from conjugates would be self-limited and would not result in virus propagation in humans.
Gene therapy is emerging as a promising novel therapeutic approach to treat human disease (1)(2)(3)(4)(5). However, technical barriers related to the biology of vector cell interaction and patient safety remain the most critical end points in clinical trials. Therefore, the success of human gene therapy depends on effective transfer and expression of the desired genes into the target cell at the lowest possible risk for the individual. Gene therapy approaches most commonly utilize recombinant viral vectors, in which both cellular uptake and transgene expression depend on interactions between the viral vector and the target cell (3,4,6). The best established vector system for achieving stable transduction in human cells is the amphotropic retrovirus vector, with its infectivity or cellular uptake at least in part determined by the presence of natural receptors on the target cells. Since the use of retrovirus vectors is restricted to specific receptor-expressing cells, transduction efficiency is largely determined by cell surface receptor density (7)(8)(9). Due to the lack of ecotropic receptors on human cells, unmodified ecotropic retrovirus cannot be used for transduction of human cells.
Currently, all clinical approaches for stable transduction of human cells rely on amphotropic retrovirus. The rare event of formation of replication-competent amphotropic retrovirus could result in systemic viremia of a treated individual, as amphotropic retrovirus can infect and replicate in most human cells. If technology would permit the effective use of ecotropic retrovirus for human gene therapy protocols, this crucial safety concern could be alleviated. Uncontrolled virus replication in humans with ecotropic virus would be unlikely, since the lack of natural receptors prevents infection of human cells. Here we have developed and evaluated a strategy to accomplish this goal by synthesizing ecotropic retrovirus molecular conjugates (eMMLV-PL). 1 To prove feasibility for this concept, the first step would require physical or chemical modifications of ecotropic retrovirus, not only permitting infection of human cells but also the efficient cellular transduction with the transferred gene. The development of strategies to overcome ecotropic/ amphotropic receptor-dependent vector uptake and enhancement of cellular transduction efficiency through circumventing the natural pathway of virus to cell entry has become a major goal of gene therapy research.
We previously determined that recombinant adenovirus molecular conglomerate vectors substantially enhanced vector uptake and transgene expression after formation of polylysinebased conglomerates over recombinant adenovirus particles alone (10). Here, we investigate the hypothesis that polylysineconjugated retrovirus results in increased uptake and gene expression over the equivalent of unconjugated virus (Fig. 1A). Furthermore, we hypothesized that through this procedure, the tropism restriction of ecotropic virus could be overcome, and human cells would become amenable to transduction with ecotropic retrovirus vectors (Fig. 1B). If successful, this conjugate ecotropic vector would have an additional significant safety advantage over currently employed amphotropic systems.

MATERIALS AND METHODS
Cells and Culture Conditions-The human embryonic kidney cell 293 line was provided by Dr. Robert Gerard, University of Texas Southwestern, Dallas, TX (also available from ATCC ((Manassas, Virginia))) (11). The human retinoblastoma cell line 911 was provided by Dr. Bout (12). The human pulmonary epithelial carcinoma-derived cell line A549 was purchased from ATCC. All cell lines were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc) supplemented with 10% fetal bovine serum (Life Technologies, Inc.) and penicillin/streptomycin (Cellgro Mediatech, Herndon, VA) and maintained at 37°C under a 5% CO 2 atmosphere in a humidified incubator (standard cell culture conditions). Primary human bone marrow cells were obtained by standard aspiration from the iliac crest of healthy voluntary donors and collected in MyeloCult H5100 medium (StemCell Technologies, Vancouver, BC, Canada) supplemented with 250 units/ml heparin. The institutional review board at Louisiana State University Health Sciences Center of New Orleans, LA approved this study involving human subjects. For erythrocyte lysis, specimens were incubated for 1 min in 0.747% NH 4 Cl solution. After two washing steps with phosphate-buffered saline (PBS), 1ϫ10 7 cells were seeded in 10-cm dishes (Nunc, Naperville, IL) and cultured in Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum and penicillin/streptomycin and maintained under standard conditions. Non-adherent cells were decanted with media exchange 2-3 times/week. A near confluent bone marrow stroma layer of fibroblast morphology was established after 2-3 weeks. Gene transfer experiments in these cells were performed at 70 -80% confluence.
Retroviral Vectors-The amphotropic retroviral helper cell line PA317 or the ecotropic retroviral packaging cell line GPϩE86 were transfected with the plasmid pLZ12 using calcium phosphate (13,14). Plasmid pLZ12 was described by Galileo et al. (15) and is a modification from plasmids originally made by Casadaban et al. (16). Plasmid pLZ121 encodes the lacZ reporter gene under control of the Rous sarcoma virus promoter and has a 5Ј nuclear localization signal (Fig. 1).
The retroviral helper cell line was cultured under standard conditions (14). Supernatants were collected from stable vector-producing cells, and retrovirus purification was performed after a procedure previously described by Akatsuka et al. (17). Briefly, a 30% (w/w) stock solution of polyethylene glycol 8000 (Sigma) was prepared in doubledistilled water and stored in aliquots at 4°C. Viral supernatants were gently mixed with polyethylene glycol in 250-ml polystyrene tubes to achieve a final 8% polyethylene glycol solution. The mixture was maintained overnight at 4°C. After centrifugation at 1500 ϫ g for 45 min, the precipitate was dissolved in 3 ml of TES buffer (10 mM Tris-HCl, pH 7.2, 2 mM EDTA, 150 mM NaCl). The titer of virus preparations was determined as previously described on NIH3T3 monolayers (18;19).
Chemical Linkage of Retrovirus to Polylysine-Retrovirus was covalently linked to poly-L-lysine (PL) (Sigma), molecular mass 30 -70 kDa, using the heterobifunctional chemical linker 1-ethyl-3(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) (Pierce). Briefly, 3.6 ml of retrovirus (7 ϫ 10 7 colony-forming units/ml) was resuspended in Hepes-buffered saline (HBS; 5 mM HEPES pH 7.8, 150 mM NaCl) and 0.4 ml of PL (6 mg/ml in HBS). 0.04 ml of freshly made EDC solution (250 mg/ml in dH 2 O) was added, and the reaction mixture was incubated on ice for 4 h. This ratio of reagents was found to be optimal, as determined in several pilot experiments in different human and murine cell lines (data not shown). The product was then dialyzed overnight at 4°C against a large volume of TES using 12-14-kDa cutoff membrane tubing (Spectra/Por, Laguna Hills, CA). The final product, retroviruspolylysine (eMMLV-PL) was separated into aliquots in TES and stored at Ϫ80°C.
Chemical Linkage of Avidin to Polylysine-NeutrAvidin (NA) (Pierce) was covalently linked to PL, molecular mass 30 -70 kDa, using the heterobifunctional chemical linker EDC. Briefly, 10 mg/ml NA was resuspended in HBS. 1.5 ml of NA solution was gently mixed with 0.5 ml of PL (6 mg/ml in HBS) before 0.04 ml of freshly made EDC solution (250 mg/ml in dH 2 O) was added. The reaction mixture was incubated on ice for 4 h. The product was then dialyzed overnight at 4°C against a large volume of PBS using 12-14-kDa cutoff membrane tubing (Spectra/Por, Laguna Hills, CA). The final product, avidin-polylysine was separated into aliquots in PBS and stored at Ϫ20°C.
Biotinylation of Retrovirus-Retrovirus was covalently linked to sulfo-N-hydroxysuccinimide-biotin (biotin) following the instructions of the manufacturer (Pierce). Briefly, 0.2 ml of biotin (10 mg/ml in Me 2 SO) was mixed with 2.8 ml of retrovirus in PBS (7ϫ10 7 colony-forming units/ml). The reaction mixture was incubated on ice for 2 h. The product (eMMLV-B) was then exchanged in TES buffer (0.01 M Tris-HCl, pH 7.2, 0.002 M EDTA, 0.15 M NaCl) over EP 10 DG columns (Bio-Rad). EMMLV-B was separated into aliquots and stored in TES at Ϫ80°C. The titer of eMMLV-B was determined as previously described on NIH3T3 monolayers (20).
Gene Transfer Analysis-LacZ staining was performed as previously described (21). Briefly, cells were fixed in 0.05% glutaraldehyde phosphate buffer (150 mM sodium chloride, 15 mM sodium phosphate, pH 7.3) (Sigma) for 10 min at room temperature and rinsed with PBS. Cells were stained for lacZ expression overnight at 37°C using freshly made X-gal solution buffer (1 mg/ml X-gal, 20 mM potassium ferrocyanide, 20 mM potassium ferricyanide, and 2 mM MgCl 2 in PBS). After rinsing with 70% ethanol, cells were analyzed by microscopy for blue cells.
Polymerase Chain Reactions (PCR)-Genomic DNA was extracted from cell pellets using a DNA extraction kit (Qiagen, Valencia, CA). A unique and specific 1,036-base pair sequence of the Escherichia coli lacZ gene was amplified using the primer pair 5Ј-GCC-GAC-CGC-ACG-CCG-CAT-CCA-GC-3Ј (sense) and 5Ј-GCG-GCG-CGG-TGA-CCA-CAC-CC-GG-3Ј (antisense) using Taq polymerase (Promega, Madison, WI). Amplification conditions consisted of a 1-min denaturing cycle at 94°C, a primer annealing step at 65°C for 1 min, and a primer extension step at 72°C for 2 min for a total of 35 cycles. A final elongation cycle was performed for 10 min at 72°C (MJ Research, Peltier Thermal Cycler, Watertown, MA). The product was analyzed after electrophoresis on a 1% agarose gel after ethidium bromide staining under UV fluorescence.
For quantitative REAL-time PCR, a unique and specific 84-base pair sequence of the E. coli lacZ gene was amplified using the primer pair E. coli B-gal-f 5Ј-ATG-CGC-CCA-TCT-ACA-CCA-A-3Ј (sense) and E. coli B-gal-rev 5Ј-AGA-AAC-AAC-CCG-TCG-GAT-TCT-C-3Ј (antisense). Product labeling was performed using the fluorochrome SYBR® Green following the instructions recommended by the manufacturer (PE Biosystems, Foster City, CA). Amplification conditions consisted of a 10min AmpliTaq Gold polymerase activation step at 95°C followed by a denaturing step at 95°C for 15 s and a combined annealing and extension step at 60°C for 1 min in a thermocycler (GeneAmp®5700 sequence detection system, PE Applied Biosystems). A total of 40 cycles of sequential denaturing/annealing and extension was performed. A standard curve was established by concomitant amplification of DNA extracted from a retrovirally stably transduced NIH3T3 clone. The samples were analyzed with the 5700 software detection system and corrected for total DNA content using a spectrophotometer (DU640B, Beckman Instruments). An R of Ͼ0.99 was required for standard curves. The sensitivity of the assay was determined to be 10 copies/ sample. Copy number per cell was determined by determining sample LacZ copy from a standard curve.
Southern Blot Analysis-Genomic DNA was extracted form different lacZ-expressing 911 single cell clones previously transduced with eMMLV-PL using a commercial DNA extraction kit (Qiagen, Chatsworth, CA). 40 g of restriction-digested genomic DNA and standards established with the lacZ gene derived from the NotI-digested control plasmid pACCMVLacZ (21) were loaded on a 0.8% agarose gel and separated by electrophoresis following previously described procedures (22). The DNA was transferred to positively charged nylon membranes (MSI, Westborough, MA), which was subsequently hybridized with a 32 P-radiolabeled probe specific for the lacZ sequence (5Ј-GCC-GAC-CGC-ACG-CCG-CAT-CCA-GC-3Ј). After removal of nonhybridized probe by washing, the membrane was exposed to radiographic film and developed. Restriction digestion of pACCMVLacZ with EcoRI linearized the plasmid, yielding a 12-kb fragment; digestion with NotI released a 4.3-kb cassette encoding the lacZ gene (21). For demonstration of integration by junction fragments, genomic DNA was digested with XbaI. There is a unique XbaI site in the retrovirus construct just 3Ј of the lacZ reporter gene (15,16). The frequency of XbaI restriction sites in human genomic is approximately 1/4 kb. For evaluation of lacZ copy number integrated into host cells, genomic DNA was digested with XhoI and XbaI to release a 4.4-kb fragment encoding the lacZ gene within the retrovirus construct (15,16). Genomic XhoI-and XbaI-digested DNA was analyzed by Southern blot together with different dilutions of purified NotI-digested pACCMVLacZ plasmid DNA fragments to establish a standard. The membranes were probed with the above-described lacZ-specific probe, and the radiographic intensities were determined using a chemiluminescent 4000 low light imaging system (Alpha Innotech Corp., San Leandro, CA). Copy number per cell was calculated using previously published formulas where the band intensity of the transduced clones were compared with intensities from the standard curve (22)(23)(24).  (Fig. 2B).

Formation of Polylysine-based Retrovirus Molecular Conglomerates (aMMLV-PL) Results in Enhanced Gene
Formation of Polylysine-based Retrovirus Molecular Con-

glomerates (eMMLV-PL) Overcomes Tropism Restriction of Ecotropic Retrovirus-
To study the hypothesis that formation of polylysine-based conglomerates overcome the tropism restriction of ecotropic retroviruses, biotinylated eMMLV encoding the lacZ reporter gene (eMMLV-B) was joined to PL-NA to form eMMLV-PL. Similar to aMMLV-B, the biotin modification did not significantly decrease infectivity of eMMLV (data not shown). Human embryonal kidney 293 cells were infected with different vector and control constructs at an m.o.i. of 0.4 or 1.5 and analyzed for reporter gene expression. As expected, no transgene expression was observed with unmodified ecotropic virus (eMMLV) or eMMLV-B. However, physical linkage of eMMLV-B to PL-NA with conglomerate formation of eMMLV-PL resulted in significant transgene expression of 25 and 84%, respectively (Fig. 3). Retrovirus Molecular Conglomerates (eMMLV-PL) Integrate into the Genome of Human Cells at One Copy per Cell-Integration of lacZ gene into the host genome was demonstrated by genomic DNA restriction digestion of five 911 single cell clones using the enzyme XbaI. Different sized junction fragments were identified in all clones examined for the presence of lacZ (Fig. 6). The copy number of the transferred reporter gene in transduced 911 cells was measured by REAL-time PCR and by Southern blot. A copy number of 0.91/cell (0.87-0.95) was quantified by PCR, and a copy number of 0.83/cell was quantified by Southern blot (Fig. 7, A and B). The combined evidence for these experiments indicates that eMMLV-PL results in integration of the transgene at 1 copy/cell. . For quantitative Southern blot analysis, genomic DNA was restriction-digested with XhoI and XbaI to release the 4.4-kb lacZ cassette from the integrated retrovirus construct. Plasmid pACCMVLacZ was restriction-digested with the enzyme NotI, releasing the 4.3-kb lacZ gene to establish a standard. The membrane was hybridized using a probe specific for the lacZ gene as outlined under Materials and Methods." Signal intensity was used to calculate copy number per cell (B).

DISCUSSION
Our findings demonstrate that ecotropic retrovirus could be modified with a conjugate technique to accomplish effective transduction of human cells. If successful, this procedure could potentially provide additional safety features for use in humans. Ecotropic virus is chemically linked within the conjugate, so it remains genetically unaltered. The theoretical risk of in vivo propagation is substantially less for a transient tropism modification compared with a genetically tropism-modified ecotropic retrovirus vector such as vesicular stomatitis viruspseudotyped envelopes. In the event that transiently modified ecotropic virus becomes replication competent, its progeny would be ecotropic retrovirus, which is incapable of propagating in humans. On the other hand, progeny arising from a genetically engineered, tropism-modified virus could continue to maintain its tropism-modified form and, thus, would remain infectious to human cells, posing a significant safety risk.
First, we had to proof that the polylysine conjugation does indeed enhance the transduction efficiency of retrovirus. Although previous studies showed in principle feasibility of retrovirus retargeting via alternate cell surface receptors, several retargeting studies were associated with significant reduction or even disappearance of cellular transduction (25) (26,27,28). In contrast, greatly enhanced transduction efficiency was reported for retrovirus vectors redirected by pseudotyping with the vesicular stomatitis virus envelope (29,30).
To investigate the hypothesis that polylysine conjugation would enhance retrovirus transduction, retrovirus molecular conglomerates were synthesized by biotinylation and linking to avidinylated polylysine. Our data demonstrate that at low m.o.i., formation of molecular conjugates synthesized with amphotropic virus (aMMLV-PL) substantially enhances transgene expression over the equivalent amount of free amphotropic retrovirus (aMMLV-B). The human cell lines tested were permissive to infection with amphotropic retrovirus; thus, increasing m.o.i. eventually resulted in nearly 100% transduction efficiency with both virus vectors. Therefore, the advantage in this particular experimental setting was limited to a transduction enhancement observed at low m.o.i.
The mechanism of the increased transduction efficiency of amphotropic conjugates could be explained through enhancement of virus uptake via their natural receptors or the utilization of an alternate pathway for virus uptake. The former concept has been proposed for fibronectin-meditated retrovirus transduction augmentation (31,32). This model proposes that fibronectin facilitates colocalization of retrovirus particles and target cells with a large increase in local virus titer presented to the cell (31,33,34). Similarly, the adjunct use of the positively charged polycations such as Polybrene in retroviral gene transfer procedures has long been known to nonspecifically promote retrovirus transduction efficiency and is now used for many protocols (35).
To proof that the entry mechanisms of retrovirus-polylysine conjugates are independent of natural receptors and are entirely polylysine-mediated, experiments were designed using ecotropic retrovirus constructs in human cells. We found that only the physical binding of eMMLV to polylysine accomplished cellular transduction with efficiencies of 80 -90% in different human cell lines and primary cells. Although polycations have previously been reported to facilitate infectivity and to enhance transduction efficiency (36), our data indicate that this effect can not only be greatly enhanced by chemical or physical linkage of virus and polycation, but this linkage allows an ability to completely overcome receptor-dependent virus uptake. This phenomenon could be explained by the ability of positively charged polylysine to nonspecifi-cally penetrate cell surface membranes and enter the cytosol (37, 38). Compared with individually retargeted retrovirus particles, conglomerates consist of multiple virus particles bound to polylysine, thus providing a higher likelihood of delivering one intact virus to the nucleus. This explanation of an added survival benefit is supported by studies where DNA bound to polylysine was found to have significantly enhanced resistance to intracellular degradation over free DNA (39). Similar to DNA, the retrovirus particles may be relatively protected from nuclease digestion within the polylysine conglomerate formation. Therefore, the improved transduction efficiency could result as the combined effect of both polylysine-mediated enhanced uptake and also improved retrovirus survival from cytosolic degradation.
Transduction with eMMLV-PL in human cells resulted in a similar genomic copy number as eMMLV in the murine cell line NIH3T3. This implies that the formation of retrovirus conglomerates does not change the fundamental principal established for retrovirus/cell interactions of only a single integration event with one retroviral copy per cell (40).
In summary, this novel retrovirus-based gene transfer system has the potential to improve transduction efficiency that comes from low receptor expression, such as for instance in hematopoietic cells (8,9). Moreover, this system would provide additional safety features over amphotropic retrovirus vectors currently used in human trials. Even the case of formation of replication-competent ecotropic virus would be theoretically a self-limited problem, as progeny of transiently modified ecotropic virus could not further propagate in humans due to its tropism restriction. Last, the polylysine-based retrovirus system may enable specific receptor retargeting by formation of receptor-targeted retrovirus molecular conglomerate vectors (10,41).