《Nature Medicine》：用艾滋病病毒（HIV）追踪癌细胞

近日美国科学家说，他们让一种艾滋病病毒“改头换面”，追踪恶性黑色素瘤细胞随血液扩散的动向，成为对付癌症的“**”。

美国加利福尼亚大学洛杉矶分校的研究人员在新一期《Nature Medicine》杂志网络版上发表论文说，这一方法可以用来输送治疗癌症的基因，成为癌症治疗中的利器。

研究人员首先使艾滋病病毒脱毒，使其成为一种“特洛伊木马”，这样它可以将治疗物输送到肺部等癌细胞经常扩散的地方，而不会引发艾滋病。随后，他们将艾滋病病毒的蛋白质外壳更换成辛德毕斯病毒的外壳，艾滋病病毒的蛋白质外壳能引导病毒粘附免疫Ｔ细胞，而辛德毕斯病毒的蛋白质外壳则是以Ｐ糖蛋白为目标的。

科学家在论文中解释说，Ｐ糖蛋白存在于恶性黑色素瘤细胞等许多癌细胞的表面，是癌症化疗中的主要障碍之一，它能像足球比赛的守门员一样，把治疗药物挡在细胞之外。而艾滋病病毒攻击力强，辛德毕斯病毒的蛋白质外壳又以Ｐ糖蛋白为目标，就有可能合成专门针对癌细胞的载体病毒。

为了验证这种载体病毒的效能，研究人员将萤火虫身上提取出的荧光素粘附在病毒上，并将病毒注入实验鼠体内。实验鼠经过改造，在肺部移植了恶性黑色素瘤。研究人员用特殊的照相镜头观察载体病毒在实验鼠血液中的运动，结果发现病毒在血液中直奔向实验鼠肺部的肿瘤，发出的荧光照亮了癌细胞。

科学家表示，尽管在实验鼠身上取得了初步成功，但这种病毒要用于人类，在安全性和特异性方面还必须更多地改进。下一步，他们将试验这种病毒是否能将基因治癌药物精确输送到肿瘤部位

Published online: 13 February 2005; | doi:10.1038/nm1192

Lentiviral vector retargeting to P-glycoprotein on metastatic melanoma through intravenous injection

Kouki Morizono1, 2, Yiming Xie1, 2, Gene-Errol Ringpis1, 2, Mai Johnson3, Hoorig Nassanian1, Benhur Lee1, 4, Lily Wu3 & Irvin S Y Chen1, 2, 5
1 Department of Microbiology, Immunology and Molecular Genetics, University of California, 10833 Le Conte Avenue, Los Angeles, California 90095, USA.

2 UCLA AIDS Institute, University of California, 10833 Le Conte Avenue, Los Angeles, California 90095, USA.

3 Department of Urology, University of California, 10833 Le Conte Avenue, Los Angeles, California 90095, USA.

4 Department of Pathology and Laboratory Medicine, University of California, 10833 Le Conte Avenue, Los Angeles, California 90095, USA.

5 Department of Medicine, David Geffen School of Medicine, University of California, 10833 Le Conte Avenue, Los Angeles, California 90095, USA.

Correspondence should be addressed to Irvin S Y Chen syuchen@mednet.ucla.edu

Targeted gene transduction to specific tissues and organs through intravenous injection would be the ultimate preferred method of gene delivery. Here, we report successful targeting in a living animal through intravenous injection of a lentiviral vector pseudotyped with a modified chimeric Sindbis virus envelope (termed m168). m168 pseudotypes have high titer and high targeting specificity and, unlike other retroviral pseudotypes, have low nonspecific infectivity in liver and spleen. A mouse cancer model of metastatic melanoma was used to test intravenous targeting with m168. Human P-glycoprotein was ectopically expressed on the surface of melanoma cells and targeted by the m168 pseudotyped lentiviral vector conjugated with antibody specific for P-glycoprotein. m168 pseudotypes successfully targeted metastatic melanoma cells growing in the lung after systemic administration by tail vein injection. Further development of this targeting technology should result in applications not only for cancers but also for genetic, infectious and immune diseases.

Clinically effective gene therapy protocols for various diseases would ideally use procedures for efficient and specific targeting of therapeutic genes to affected cells while maintaining stable transduction and long-term expression. This would be accomplished by direct injection into the bloodstream followed by homing of the vector to the desired target cells or organs. Thus, there have been many attempts to develop targeted gene transduction systems based upon various viral vectors. Adenovirus and adeno-associated virus vectors have been used in targeted gene delivery strategies because of their simple binding and entry mechanisms1. Although these vectors have been used successfully in vitro for targeting to specific cells, their usefulness in vivo has been limited by their natural tropism2, especially to liver cells3. Reducing nonspecific infection of untargeted tissues is one of the important aspects for the translation of these vectors to an in vivo setting.

Oncoretroviral- and lentiviral-based vectors have several properties that make them ideal for use in gene therapy4. Efficient integration of retroviral DNA into the host genome enables stable long-term transgene expression. Unlike oncoretroviral vectors, lentiviral vectors are capable of transducing nondividing cells. The application of specific targeting with retroviral vectors has been problematic and the few studies of retroviral vector targeting in living animals are not efficient5, 6. Inserting ligands, peptides or single-chain antibodies into the retroviral receptor binding envelope subunit has been the most common approach used to alter or restrict the host range of retroviral vectors5, 6, 7, 8, 9, 10, 11. Another approach is bridging virus vector and cells by antibodies or ligands12, 13. In general, most strategies have suffered from inconsistent specificity and low viral titers as a result of modification of the retroviral envelope7, 8, 9, 10, 11, 14. The modified envelope proteins seem to have specific binding activity but low fusion activity, resulting in inefficient entry into cells15. In the absence of specific targeting, current strategies depend upon direct injection to a localized site16 or, as in the case of the only successful treatment of heritable diseases, X-linked severe combined immunodeficiency (SCID) and adenosine deaminase (ADA)−deficient SCID (ADA-SCID), ex vivo isolation, purification and transduction of target hematopoietic cells17, 18.

We previously found that the envelope of the alphavirus Sindbis is able to pseudotype oncoretroviruses and lentiviruses19. The two integral membrane glycoproteins, E1 and E2, form a heterodimer and function as a unit. E2 binds to the host cell receptor. E1 mediates membrane fusion in a low pH−dependent fashion. E3 works as a signal sequence peptide for E2 protein. We reported an oncoretroviral and lentiviral gene−targeting system based on antibody-mediated specific binding of a modified chimeric Sindbis virus envelope (ZZ SINDBIS) that encoded the ZZ domain of protein A. We showed that monoclonal antibodies directed to cell-surface antigens can be used to redirect the target specificity of these vectors when pseudotyped with the modified Sindbis envelope. Of particular note, the vectors maintained high viral titers, which could be further increased by simple ultracentrifugation.

Although effective for in vitro targeting, intravenous injection of ZZ SINDBIS pseudotypes into mice resulted in higher levels of infectivity in liver and spleen cells. The high-affinity laminin receptor20 and heparin sulfate are among the known receptors21 of Sindbis virus. Their wide distribution and highly conserved nature may be in part responsible for the residual nonspecific tropism observed with the ZZ SINDBIS pseudotyped vector. We identified several mutants of E2 that reduced the endogenous tropism of the Sindbis envelope. We utilized the modified ZZ SINDBIS envelope, designated m168, and a lentiviral reporter vector to target P-glycoprotein (P-gp)−expressing melanoma cells in the lungs of a mouse model of metastatic melanoma. Specific targeting of metastatic tumor cells was shown through direct injection of the vector into the bloodstream.

Results
Strategy to assess targeted gene transduction
Our goal is to use targeting vectors for direct targeting of gene therapy vectors to specific target cells through direct injection in the bloodstream. Thus, we designed a series of experiments to determine the background level of infection in a mouse model and determine the optimal conditions for targeting lentiviral vectors to specific human cancers in a tumor transplant mouse model.

We used lentiviral vectors bearing distinct reporter genes to assess transduction efficiency (Supplementary Fig. 1 online). The EGFP-expressing virus vector allowed a quantitative assessment of infectivity in vitro as monitored by flow cytometry22. The firefly and Renilla luciferase−expressing virus vectors were assayed by a noninvasive cooled charged-coupled device (CCCD) imaging to quantify the level of specific targeting of vectors in live mice23.

HIV (ZZ SINDBIS) has nonspecific infectivity in vivo
Figure 1a shows the typical enhancement in infectivity of the ZZ SINDBIS pseudotyped virus vector in vitro using a monoclonal antibody directed to human leukocyte antigen (HLA). We observed approximately a 30-fold enhancement of cells positive for enhanced green fluorescent protein (EGFP) in the presence of HLA-specific antibody relative to that seen in the absence of monoclonal antibody. As a comparison, VSV-G envelope−pseudotyped virus infected cells at a high level in the absence of antibody. The titer of ZZ SINDBIS virus is usually about 20% of VSV-G-pseudotyped virus but, like VSV-G pseudotypes, can be further concentrated at least 100-fold by ultracentrifugation.

Figure 1. HIV vector pseudotyped by ZZ SINDBIS has nonspecific infectivity in the absence of target-specific antibody in vitro and in vivo.

(a) 293T cells (1 105) were infected with TRIP GFP (ZZ SINDBIS) (34 ng of HIV p24) with or without anti-HLA (1 g/ml). For comparison of titers, cells were infected with TRIP GFP (VSV-G) (8 ng of HIV p24). We analyzed EGFP expression by flow cytometry. (b) FUhLucW (VSV-G) (1.5 g HIV p24), FUhLucW (Sindbis) (3 g HIV p24) and FUhLucW (ZZ SINDBIS) (3 g HIV p24) were injected intravenously through the tail vein. Five days after injection, the reporter gene (firefly luciferase) expression was imaged using a CCCD camera. p/s/cm2/sr, photons/sec/cm2/steridian.

Full Figure and legend (129K)

ZZ SINDBIS−pseudotyped virus vectors expressing firefly luciferase were used to quantify the level and specificity of targeting in transduced cells in the organs of live mice. ZZ SINDBIS virus vectors containing the broadly expressed ubiquitin-C promoter24 were injected into the tail vein in the absence of monoclonal antibody and expression was monitored by luciferase expression using CCCD imaging (Fig. 1b). Lentiviral vectors pseudotyped with each of the envelopes, VSV-G, wild-type Sindbis and ZZ SINDBIS, infected both liver and spleen. Although ZZ SINDBIS gave a weaker signal, consistent with its infectivity in vitro, there was still clear expression in liver and spleen. Injection of recombinant luciferase did not show a signal in major organs, indicating that the signal observed with ZZ SINDBIS was the result of infection of cells in the organs (data not shown). We further verified infection of Sindbis and ZZ SINDBIS in the liver and spleen by quantitative DNA PCR analysis (Table 1).

Table 1. Copy number of lentiviral vector/104 cells

Full Table

Nonspecific infectivity results from Sindbis envelope domain
We investigated the nature of the nonspecific background infectivity of ZZ SINDBIS pseudotypes. A mouse polyclonal antibody that neutralizes wild-type Sindbis virus infectivity was used to show that the background infectivity was the result of Sindbis virus domains and not the ZZ protein A sequences (Supplementary Fig. 2 online). We determined that the level of EGFP+ cells in the wild-type and ZZ SINDBIS virus pseudotypes was substantially reduced in the presence of the antibody specific for Sindbis. Infectivity of VSV-G pseudotypes was not blocked, nor was infection with the control antibody. These results indicate that Sindbis virus domains within the Sindbis virus envelope are responsible for the nonspecific infectivity of ZZ SINDBIS pseudotypes.

Identification of a mutant with enhanced specificity
We further ablated residual infectivity by targeting E2 domains previously reported to affect binding to target cells, block epitopes for neutralizing antibody and function in Sindbis virus tropism (Fig. 2a). The infectivity of mutants was tested on two different cell types, 293T, a human kidney cell line used for standard titration of virus stocks, and HepG2 cells, derived from a human hepatocellular carcinoma (Supplementary Table 1 online). We identified several E2 mutants with reduced levels of nonspecific infectivity and, thus, an enhanced selectivity for targeting. Because some of these mutations also reduced the titer of viruses produced, we combined the mutations conferring enhanced selectivity with other mutations that enhanced infectivity. The combination of mutations of m1, m6 and m8 in domains R1, R2 and R4, respectively, resulted in a pseudotyped virus, termed m168, that had enhanced selectivity on 293T and HepG2 liver cells with maintenance of viral titers of about 106 EGFP+ units/ml in unconcentrated supernatants and stability after 100-fold concentration by ultracentrifugation. Our data are consistent with previous studies that show the role of the R1 and R2 domains for heparin sulfate binding and the R4 domain for rescue of the reduced titer of an R1 mutant in replication-competent Sindbis virus25. Although the absolute selectivity index varies in different cell types, the selectivity index of m168 was consistently greater in all human (293T, HepG2, Jurkat, HeLa, LNCaP) and mouse (NIH3T3, NB43) cell lines tested.

Figure 2. Generation of ZZ SINDBIS mutants to reduce nonspecific infection.

(a) Schematic representation of mutated domains and mutants. The two integral membrane glycoproteins, E1 and E2, form a heterohexamer and function as a unit. E2 binds to the host cell receptor. E1 mediates membrane fusion in a low pH−dependent fashion. E3 works as a signal sequence peptide for E2 protein. (b) HepG2 cells were infected with unconcentrated TRIP GFP (ZZ SINDBIS) (21 ng HIV p24) or TRIP GFP (m168) (10 ng HIV p24) with or without anti-HLA (1 g/ml). EGFP expression was analyzed by flow cytometry. (c) SDS-PAGE and western blotting of ZZ SINDBIS and m168 pseudotyped virions. The band at approximately 65 kDa in the ZZ SINDBIS lane is the chimeric E2 protein. The band at approximately 75 kDa in the m168 lane is chimeric E2 protein with uncleaved E3 protein.

Full Figure and legend (72K)

A representative experiment illustrating enhanced specificity of m168 infection in HepG2 cells in the presence of an HLA monoclonal antibody is shown in Fig. 2b. In the absence of antibody the background level of infectivity is reduced when compared to the ZZ SINDBIS virus and the levels of infectivity and stability are maintained. Note that the m168 envelope protein is larger as a result of mutation m1, which prevents cleavage of E2 and E3 (Fig. 2c).

M168 virus shows reduced nonspecific infectivity
The ultimate goal of these studies is to develop a gene-transfer vector capable of delivery directly into the bloodstream to target specific tissues or cells. To this end, we tested the ability of the genetically modified m168 lentiviral pseudotypes to infect cells in live mice. We injected viruses bearing luciferase reporter genes, without targeting antibody, through the tail vein into SCID mice and assessed the location of the infectivity using a CCCD camera to determine the level of luciferase expression in the mice. Consistent with the in vitro results, the modified ZZ SINDBIS m168−pseudotyped virus showed a substantially lower infectivity in the liver and spleen of inoculated animals, relative to the parental ZZ SINDBIS−pseudotyped virus (Fig. 3a). Because the intensity of the CCCD imaging for luciferase expression can be influenced by a number of variables, such as depth of tissue and positioning of the animal during imaging, we confirmed these results by isolation of organs and PCR analysis for vector DNA sequences (Table 1). Of note, we did not detect infection in ovaries, indicating that transduction of our vector into germ line cells was unlikely to occur.

Figure 3. The m168 pseudotyped lentiviral vector has reduced nonspecific infectivity and mediates antibody-directed targeted gene transduction after systemic injection into mice.

(a) FUhLucW (ZZ SINDBIS) and FUhLucW (m168) were each injected into three mice through the tail vein. Five days after injection, the reporter gene (firefly luciferase) expression was imaged. (b) B16F10MDR5 cells expressing Renilla luciferase were injected into mice through the tail vein. We injected FUhLucW (ZZ SINDBIS) or FUhLucW (m168), to which we had added anti-P-glycoprotein monoclonal antibody or isotype control antibody, into the tail vein 30 min later. Ten days later, the level of metastasis in the lungs of B16F10MDR5-treated mice was determined by imaging the level of Renilla luciferase expression. Twelve days after cell and virus injection, virus infection was determined by imaging the level of firefly luciferase reporter gene expression. (c) B16F10MDR5 cells were injected into mice through the tail vein. Twelve days later, FUhLucW (m168), to which we had added anti-P-glycoprotein monoclonal antibody or isotype control antibody, or FUhLucW (VSV-G) was injected into the tail vein. Three days after virus injection, virus infection was determined by imaging the level of firefly luciferase reporter gene expression. (d) Immediately after whole-body imaging each organ was isolated to image luciferase expression.

Full Figure and legend (278K)

Targeting of P-gp expressing micrometastatic melanoma
Malignant melanoma is an aggressive human tumor that metastasizes to multiple tissues, including remote skin, soft tissue, lymph node and lung26. Increased expression of the gene involved in multidrug resistance (ABCB1) causes overproduction of the transmembrane transport protein P-glycoprotein (P-gp)27, rendering tumor cells resistant to natural product amphiphilic anticancer drugs. One study showed that 33−76% of the melanoma cell lines derived from primary tumors or metastases of untreated patients scored positive for P-gp28. We selected a mouse z model for human malignant metastatic melanoma in which the mouse B16F10 tumor cells migrate through the bloodstream to engraft and form tumors in the lungs. We engineered the mouse melanoma cells to express the human ABCB1 gene (B16F10MDR5 cells) to provide a physiologically relevant cell-surface molecule for targeting (Supplementary Fig. 1 online).

First, the specificity of the targeting for the P-gp−expressing melanoma cells was shown in vitro (selectivity index of >90 versus 10.8 for parental ZZ SINDBIS; Supplementary Fig. 3 online). Thus, this combination of tumor cells and m168-pseudotyped transducing vector was used for vector targeting in live mice.

The tumor cells were first marked in vitro by transducing with a vector expressing Renilla luciferase. In this manner, the localization of vector expression and tumor cells in the mice was differentially visualized using different substrates for the two luciferase genes, firefly and Renilla, respectively (Fig. 3b). Following injection, the tumor cells migrate to the lungs and were visualized by CCCD imaging for Renilla luciferase. We injected the m168 virus bearing antibody to P-gp through the tail vein 30 min after tumor cell inoculation. Vectors injected in the absence of antibody to P-gp show no signal for firefly luciferase. In contrast, when virus is pseudotyped with m168 bearing antibody specific for P-Gp, expression of firefly luciferase (m168 virus vector) colocalizes in the lung with that of Renilla luciferase (tumor melanoma cells).

The colocalization in the lungs of firefly and Renilla luciferase expression, representing tumor and virus infection, respectively, was confirmed to be a result of infection of the melanoma cells by the targeting vector. Melanoma tumor cells were isolated from lung tissue by allowing the cells to rapidly outgrow other cells in culture (Supplementary Fig. 4 online). We observed firefly luciferase activity, representing expression from the vector, predominately in those tumor cells isolated from mice in which virus targeted to antibody specific for P-gp was used for the infection (Table 2). Quantitative real-time PCR analysis confirmed these results. Of note, m168 pseudotype vectors show a higher level of infection in melanoma cells in vivo compared to the parental ZZ SINDBIS, probably the result of less nonspecific transduction of other mouse tissues.

Table 2. Luciferase assay of melanomas
Full Table

We also visualized the targeting of micrometastatic tumors by immunohistochemistry (Supplementary Fig. 5 online). We observed tumor micronodules in the lungs at day 6 after transplant with morphologic characteristics of melanoma and positive for melanoma antigen S-100 (ref. 29). About 1% of these nodules were also positive for EGFP, consistent with the previous PCR analysis (Table 1).

We identified rare transduced cells in the spleen and liver as macrophages and related Kupffer cells, respectively, by flow cytometry and immunohistochemistry (Supplementary Fig. 6 online).

Targeting of established melanoma tumors
In another set of experiments, we tested the ability of the m168 vector to target established tumors. We used the same protocols as above, except that tumors were allowed to form for 12 d before intravenous injection of the targeting vector. In this model system, visible tumors are evident at 8 d after inoculation and death occurs within 16 d after inoculation. We visualized mice using CCCD imaging 3 d after intravenous injection for both the location of tumors and the specificity of targeting. Specific targeting to the tumors in the lungs of the animals was observed only after intravenous injection with m168 pseudotypes plus antibody specific for P-gp (Fig. 3c and Table 2). We observed no nonspecific infection in the lungs in the absence of tumor and the infection to tumors in the lungs was dependent upon the presence of antibody specific for P-gp. The transduction of organs was confirmed by isolation of specific organs after killing the mice (Fig. 3d). The reduced signal intensity in lungs of this experiment relative to targeting of micrometastatic cells (compare mice of Fig. 3b with Fig. 3c) results from limited growth of transduced cells (3 d versus 12 d after transduction) before imaging. For comparison, intravenous injection of VSV-G pseudotypes show infection of a broad number of tissues without specificity for the tumors consistent with previously published studies30.

Top
Discussion
The potential applications of a specific targeting gene therapy vector are broad and limited only by one's ability to identify specific molecules on the surface of targeted cells. The target molecules could be protein antigens such as those described here. Alternatively, they could be receptors for specific ligands or even unknown molecules identified by peptide library screening approaches expressed selectively on the desired cell or tissue types.

Early treatment of metastatic cells is of considerable therapeutic value. In theory, using the strategies described here, metastases could be targeted well before they grow to a size to be visualized by current technologies. Residual tumor cells that remain at the tumor site after localized treatment with radiation and/or surgery could also be targeted and eliminated. Targeting P-gp as a tumor antigen can be useful not only for metastatic melanoma but also for many other tumors that express the gene that encodes it and are thus rendered resistant to multiple chemotherapeutic drugs27. P-gp is also expressed on some normal cells, thus the targeting of tumor cells that express P-gp would be most successful in those situations in which P-gp was substantially overexpressed in the tumor cells. The effectiveness of targeting tumor cells will be dependent upon factors such as the ratio of vector to tumor cells and accessibility of tumor cells to the vector.

One limitation of the targeting vector system described here is the use of monoclonal antibodies. In an animal or person with an immunocompetent humoral immune system, the presence of circulating antibodies could compete for the monoclonal antibodies of the targeting vector. Thus, future clinical application would probably be more effective with chimeric, recombinant, single-chain antibody sequences or specific ligand or peptide sequences. Human chorionic gonadotropin sequences have been successfully recombined into chimeric Sindbis envelope to target choriocarcinoma cells in Sindbis virus−based vectors30. Therefore, it should be readily feasible to develop similar chimeric envelope proteins to pseudotype retroviral vectors. Host natural or adaptive immune responses to the vector or antibody or ligand also need to be considered in future therapeutic applications.

Direct injection into the bloodstream and infection to specified cells would facilitate the application of gene therapy to many diseases. Because most diseases, both acquired and hereditary, either originate in specific cells or manifest their clinical phenotypes in specific tissues, the gene therapeutic vectors could be delivered where they would be most effective. It is conceivable that direct introduction of therapeutic genes into the stem cells in bone marrow and systemic injection after progenitor cell mobilization could be achieved. Targeting vectors that circulate and home to specific cells could allow early therapeutic intervention in the case of diseases such as cancer, and to eliminate residual cells from chronic or latent infections by infectious agents. Finally, targeting of specific cells and tissues would greatly enhance the safety of gene therapeutic applications by reducing inadvertent infection of irrelevant cells or tissues that has resulted in serious adverse effects in some clinical trials31. Insertional mutagenesis of retroviral vectors, if it does occur, would be limited to a much smaller subset of cells.6

Future enhancements in vector specificity and further understanding of vector trafficking and host response will be important for successful clinical applications.

Top
Methods
Plasmid construction.
All mutants of pIntronZZ SINDBIS were generated using a site-directed mutagenesis kit (Stratagene) with various oligonucleotides corresponding to the mutations (Supplementary Table 1 online). CCRMDRsc1 was constructed from pRRL-cPPT-CMV-X-PRE (provided by W. Osborne, University of Washington)32 and ha-MDRsc (provided by B. Sorrentino, St. Jude Children's Research Hospital)33. FUhLucW was constructed from FUGW (provided by D. Baltimore, California Institute for Technology)24 and pGL3-Basic (Promega). FUIntronRW was constructed from FUGW and phRL-CMV (Promega).

Virus production.
We produced all lentivirus vectors by calcium phosphate−mediated transient transfection of 293T cells (Supplementary Methods online)

In vitro infection of cells.
We incubated 293T cells (1 105) and HepG2 cells (1 105) with 200 l of unconcentrated HIV vector (TRIP GFP)22 pseudotyped by ZZ SINDBIS or its mutants with or without antibody to HLA (1 g/ml; Sigma) for 2 h at 37 °C with 5% CO2. Three days after infection, we analyzed the cells by flow cytometry for EGFP expression. The titers of TRIP GFP (ZZ SINBIS) and TRIP GFP (VSV-G) in the presence of anti-HLA antibody (1 mg/ml) were 3 105 and 1.3 106 EGFP transduction units/100 ng HIV p24, respectively.

Immunoblot assay.
We subjected the HIV vectors (FUhLucW) pseudotyped with ZZ SINDBIS or m168 to electrophoresis through an sodium dodecyl sulfate polyacrylamide gel (10%). The amount of virus for each sample was normalized by the amount of HIV p24-antigen (5 ng/sample). Their envelope proteins were detected by using anti-Sindbis virus ascites fluid (ATCC) and horseradish peroxidase−conjugated antibody specific for mouse IgG (Santa Cruz Biotechnology) as described previously19 (Supplementary Methods online).

In vivo analysis of background infection.
We cared for mice and performed procedures in accordance with the University of California Animal Research Committee guidelines. HIV vector (FUhLucW) pseudotyped by VSV-G, Sindbis virus envelope, ZZ SINDBIS or m168 were injected into the tail vein of 6-week-old female nonobese diabetic (NOD) SCID mice. The number of infectious units per microgram p24 from FUhLucW (VSV-G), FUhLucW (Sindbis), FUhLucW (ZZ SINDBIS) with antibody to HLA and FUhLucW (m168) with antibody to HLA is 1 108, 2 107, 2 107 and 1 107, respectively, titrated on 293T cells. The amount injected for each virus was normalized to the amount of HIV p24 (3 g of HIV p24 in 300 l phosphate-buffered saline (PBS)). We anesthetized mice 5 d after injection and injected them intraperitoneally with D-luciferin (3 mg/mouse; Xenogen). CCCD images were obtained using a cooled IVIS CCD camera (Xenogen), and analyzed with IGOR-PRO Living Image Software. We performed the data acquisition for 1 min, 20 min after D-luciferin injection. Mice were killed by CO2 narcosis after CCCD imaging. We excised the organs from each mouse and isolated genomic DNA using a DNeasy kit (QIAGEN) following the manufacturer's protocol. Quantification of the vector copy number and cell number in the DNA isolate was performed by using SYBRgreen real-time PCR kit (QIAGEN) and an ABI PRISM 7700 sequence detector (Perkin Elmer; Supplementary Methods online).

Targeted infection of micrometastatic melanoma cells in vivo.
To express Renilla luciferase as a marker, the human P-gp−expressing mouse cell line, B16F10MDR5, was transduced by the lentiviral vector FUIntronRW (VSV-G). One day before subsequent cell and virus injection, TMbeta-1 (1 mg; provided by M. Miyasaka, Osaka University) was injected into 6-week-old female NOD/SCID mice. We injected Renilla luciferase expressing B16F10MDR5 (2 105 cells in 150 l of PBS) into mice through the tail vein. Thirty minutes later, FUhLucW (ZZ SINDBIS) or FUhLucW (m168), to which we had added anti-P-glycoprotein monoclonal antibody (Kamiya Biomedical Company) or isotype (IgG2a) control antibody (10 g/ml) (eBioscience) was injected into the tail vein. The amount of each virus used for injection was normalized to the amount of HIV p24 (3.6 g of HIV p24 in 150 l PBS). We determined lung metastasis of B16F10MDR5 10 d after cell and virus injection by imaging the Renilla luciferase. Mice were anesthetized and coelenterazine (20 g; Prolume) was injected through the tail vein. We performed data acquisition directly after coelenterazine injection for 1 min. We determined virus infection 12 d after cell and virus injection by imaging the expression of firefly luciferase reporter gene. Mice were anesthetized and injected intraperitoneally with D-luciferin (6 mg/mouse). Imaging was performed as previously described. Mice were killed after imaging using D-Luciferin. We analyzed isolated tumor cells for the firefly luciferase activity and copy number of firefly luciferase gene (Supplementary Methods online).

Targeted gene transduction of established tumor.
B16F10MDR5 cells (2 105) were injected into 6-week old female NOD-SCID mice through the tail vein. We injected FUhLucW (m168) to which we had added anti-P-glycoprotein monoclonal antibody or isotype control antibody or FUhLucW (VSV-G) into the tail vein 12 d later. The amount of each virus used for injection was normalized to the amount of HIV p24 (2.5 g of HIV p24 in 250 l PBS). Fifteen days after virus injection, virus infection was determined by imaging the level of firefly luciferase reporter gene expression as described previously.

Note: Supplementary information is available on the Nature Medicine website.

Top
Received 16 January 2004; Accepted 10 December 2004; Published online: 13 February 2005.

REFERENCES
Nicklin, S.A. & Baker, A.H. Tropism-modified adenoviral and adeno-associated viral vectors for gene therapy. Curr. Gene Ther. 2, 273–293 (2002).
Muller, O.J. et al. Random peptide libraries displayed on adeno-associated virus to select for targeted gene therapy vectors. Nat. Biotechnol. 21, 1040–1046 (2003). | Article |
Martin, K. et al. Simultaneous CAR- and alpha V integrin-binding ablation fails to reduce Ad5 liver tropism. Mol. Ther. 8, 485–494 (2003). | Article |
Sandrin, V., Russell, S.J. & Cosset, F.L. Targeting retroviral and lentiviral vectors. Curr. Top. Microbiol. Immunol. 281, 137–178 (2003).
Martin, F., Chowdhury, S., Neil, S., Phillipps, N. & Collins, M.K. Envelope-targeted retrovirus vectors transduce melanoma xenografts but not spleen or liver. Mol. Ther. 5, 269–274 (2002). | Article |
Jiang, A. & Dornburg, R. In vivo cell type-specific gene delivery with retroviral vectors that display single chain antibodies. Gene Ther. 6, 1982–1987 (1999). | Article |
Han, X., Kasahara, N. & Kan, Y.W. Ligand-directed retroviral targeting of human breast cancer cells. Proc. Natl. Acad. Sci. USA 92, 9747–9751 (1995).
Marin, M. et al. Targeted infection of human cells via major histocompatibility complex class I molecules by Moloney murine leukemia virus-derived viruses displaying single-chain antibody fragment-envelope fusion proteins. J. Virol. 70, 2957–2962 (1996).
Nilson, B.H., Morling, F.J., Cosset, F.L. & Russell, S.J. Targeting of retroviral vectors through protease-substrate interactions. Gene Ther. 3, 280–286 (1996).
Somia, N.V., Zoppe, M. & Verma, I.M. Generation of targeted retroviral vectors by using single-chain variable fragment: an approach to in vivo gene delivery. Proc. Natl. Acad. Sci. USA 92, 7570–7574 (1995).
Valsesia-Wittmann, S. et al. Modifications in the binding domain of avian retrovirus envelope protein to redirect the host range of retroviral vectors. J. Virol. 68, 4609–4619 (1994).
Boerger, A.L., Snitkovsky, S. & Young, J.A. Retroviral vectors preloaded with a viral receptor-ligand bridge protein are targeted to specific cell types. Proc. Natl. Acad. Sci. USA 96, 9867–9872 (1999). | Article |
Roux, P., Jeanteur, P. & Piechaczyk, M. A versatile and potentially general approach to the targeting of specific cell types by retroviruses: application to the infection of human cells by means of major histocompatibility complex class I and class II antigens by mouse ecotropic murine leukemia virus-derived viruses. Proc. Natl. Acad. Sci. USA 86, 9079–9083 (1989).
Kasahara, N., Dozy, A.M. & Kan, Y.W. Tissue-specific targeting of retroviral vectors through ligand-receptor interactions. Science 266, 1373–1376 (1994).
Zhao, Y. et al. Identification of the block in targeted retroviral-mediated gene transfer. Proc. Natl. Acad. Sci. USA 96, 4005–4010 (1999). | Article |
Akporiaye, E.T. & Hersh, E. Clinical aspects of intratumoral gene therapy. Curr. Opin. Mol. Ther. 1, 443–453 (1999).
Cavazzana-Calvo, M. et al. Gene therapy of human severe combined immunodeficiency (SCID)-X1 disease [see comments]. Science 288, 669–672 (2000). | Article |
Aiuti, A. et al. Correction of ADA-SCID by stem cell gene therapy combined with nonmyeloablative conditioning. Science 296, 2410–2413 (2002). | Article |
Morizono, K., Bristol, G., Xie, Y.M., Kung, S.K. & Chen, I.S. Antibody-directed targeting of retroviral vectors via cell surface antigens. J. Virol. 75, 8016–8020 (2001). | Article |
Wang, K.S., Kuhn, R.J., Strauss, E.G., Ou, S. & Strauss, J.H. High-affinity laminin receptor is a receptor for Sindbis virus in mammalian cells. J. Virol. 66, 4992–5001 (1992).
Klimstra, W.B., Ryman, K.D. & Johnston, R.E. Adaptation of Sindbis virus to BHK cells selects for use of heparan sulfate as an attachment receptor. J. Virol. 72, 7357–7366 (1998).
Zennou, V. et al. HIV-1 genome nuclear import is mediated by a central DNA flap. Cell 101, 173–185 (2000). | Article |
Bhaumik, S. & Gambhir, S.S. Optical imaging of Renilla luciferase reporter gene expression in living mice. Proc. Natl. Acad. Sci. USA 99, 377–382 (2002). | Article |
Lois, C., Hong, E.J., Pease, S., Brown, E.J. & Baltimore, D. Germline transmission and tissue-specific expression of transgenes delivered by lentiviral vectors. Science 295, 868–872 (2002). | Article |
Heidner, H.W., McKnight, K.L., Davis, N.L. & Johnston, R.E. Lethality of PE2 incorporation into Sindbis virus can be suppressed by second-site mutations in E3 and E2. J. Virol. 68, 2683–2692 (1994).
Allen, P.J. & Coit, D.G. The role of surgery for patients with metastatic melanoma. Curr. Opin. Oncol. 14, 221–226 (2002). | Article |
Ambudkar, S.V., Kimchi-Sarfaty, C., Sauna, Z.E. & Gottesman, M.M. P-glycoprotein: from genomics to mechanism. Oncogene 22, 7468–7485 (2003). | Article |
Berger, W. et al. Intrinsic MDR-1 gene and P-glycoprotein expression in human melanoma cell lines. Int. J. Cancer 59, 717–723 (1994).
Cameron, M.D. et al. Temporal progression of metastasis in lung: cell survival, dormancy, and location dependence of metastatic inefficiency. Cancer Res. 60, 2541–2546 (2000).
Sawai, K. & Meruelo, D. Cell-specific transfection of choriocarcinoma cells by using Sindbis virus hCG expressing chimeric vector. Biochem. Biophys. Res. Commun. 248, 315–323 (1998). | Article |
Lehrman, S. Virus treatment questioned after gene therapy death. Nature 401, 517–518 (1999). | Article |
Barry, S.C. et al. Lentivirus vectors encoding both central polypurine tract and posttranscriptional regulatory element provide enhanced transduction and transgene expression. Hum. Gene Ther. 12, 1103–1108 (2001). | Article |
Bunting, K.D., Galipeau, J., Topham, D., Benaim, E. & Sorrentino, B.P. Transduction of murine bone marrow cells with an MDR1 vector enables ex vivo stem cell expansion, but these expanded grafts cause a myeloproliferative syndrome in transplanted mice. Blood 92, 2269–2279 (1998).