An adenoviral vector expressing human adenovirus 5 and 3 fiber proteins for targeting heterogeneous cell populations

Virology 407 (2010) 196–205

Contents lists available at ScienceDirect

Virology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y v i r o

An adenoviral vector expressing human adenovirus 5 and 3 fiber proteins for targeting heterogeneous cell populations Miho Murakami a,b, Hideyo Ugai a, Minghui Wang a, Natalya Belousova c, Paul Dent d, Paul B. Fisher e, Joel N. Glasgow a,f,g, Maaike Everts b,g, David T. Curiel a,g,⁎ a

Division of Human Gene Therapy, Departments of Medicine, Obstetrics and Gynecology, Pathology, Surgery, The University of Alabama at Birmingham, Birmingham, AL, USA Division of Molecular and Cellular Pathology, Department of Pathology, The University of Alabama at Birmingham, Birmingham, AL, USA Department of Experimental Diagnostic Imaging, MD Anderson Cancer Center, The University of Texas, Houston, TX, USA d Department of Neurosurgery, VCU Institute of Molecular Medicine and VCU Massey Cancer Center, Virginia Commonwealth University, Richmond, VA, USA e Department of Human & Molecular Genetics, VCU Institute of Molecular Medicine and VCU Massey Cancer Center, Virginia Commonwealth University, Richmond, VA, USA f Division of Cardiovascular Disease, Department of Medicine, The University of Alabama at Birmingham, Birmingham, AL, USA g Gene Therapy Center, The University of Alabama at Birmingham, Birmingham, AL, USA b c

a r t i c l e

i n f o

Article history: Received 1 May 2010 Returned to author for revision 9 June 2010 Accepted 12 August 2010 Available online 9 September 2010 Keywords: Adenovirus Mosaic Fiber Tumor Tropism expansion

a b s t r a c t Human adenovirus serotype 5 (HAdV-5) attaches to its primary receptor, the coxsackie and adenovirus receptor (CAR) as the first step of infection. However, CAR expression decreases as tumors progress, thereby diminishing the utility of HAdV-5-based vectors for cancer therapy. In contrast, many aggressive tumor cells highly express CD46, a cellular receptor for HAdV-3. We hypothesized that a mosaic HAdV vector, containing two kinds of fiber proteins, would provide extensive transduction in a heterogeneous population of tumor cells with varying expression levels of HAdV receptors. We therefore generated a fiber-mosaic HAdV vector displaying both a chimeric HAdV-3 fiber and the HAdV-5 fiber protein. We verified the structural integrity of purified viral particles and confirmed that the fiber-mosaic HAdV vector has expanded tropism. We conclude that the use of fiber-mosaic HAdV vectors is a promising approach for transducing a heterogeneous cell population with different expression levels of adenovirus receptors. Published by Elsevier Inc.

Introduction Viral vector-mediated gene therapy is a new tool currently under development for the treatment of tumors. In particular, adenoviral vectors based on human adenovirus 5 (HAdV-5) have been developed and used for cancer gene therapy (Campos and Barry, 2007; Imperiale and Kochanek, 2004). HAdV-5 has several features that are useful for cancer gene therapy applications, including a high insert capacity (up to approximately 37 kb) (Palmer and Ng, 2005) to deliver large therapeutic genes (Fisher et al., 1996; Kochanek et al., 1996), ease of production of high titers for clinical use (Wivel et al., 1999), and relative safety as the viral genome is not integrated into the host genome (Thomas et al., 2003; Verma and Weitzman, 2005). HAdV-5 efficiently transduces cells expressing the coxsackie virus and adenovirus receptor (CAR) (Bergelson et al., 1997) via its fiber protein, including various cancer cell lines in culture.

⁎ Corresponding author. Division of Human Gene Therapy, Departments of Medicine, Obstetrics and Gynecology, Pathology, Surgery, and the Gene Therapy Center, University of Alabama at Birmingham, 901 19th Street South, BMR2-502, Birmingham, AL 35294, USA. Fax: + 1 205 975 7476. E-mail address: [email protected] (D.T. Curiel). 0042-6822/$ – see front matter. Published by Elsevier Inc. doi:10.1016/j.virol.2010.08.010

However, with the exception of breast tumors, in which CAR expression is increased with increasing tumor grade (Martin et al., 2005), CAR expression levels tend to be low in the majority of advanced tumors (Anders et al., 2009; Auer et al., 2009; Korn et al., 2006; Matsumoto et al., 2005; Okegawa et al., 2001; Rauen et al., 2002; Sachs et al., 2002; Zeimet et al., 2002). In addition, CAR expression in ovarian tumors exhibits marked distributional heterogeneity (Zeimet et al., 2002). Thus, the varied, and often insufficient, expression of CAR hampers uniform HAdV-5 transduction throughout tumors, especially in advanced tumors. In contrast to CAR expression, many aggressive tumor cells highly express CD46, a membrane cofactor protein. For example, invasive carcinoma appears to preferentially retain CD46 expression (Fishelson et al., 2003; Hara et al., 1992; Kinugasa et al., 1999; Kirschfink, 2001; Murray et al., 2000; Thorsteinsson et al., 1998). Additionally, CD46 is overexpressed in cancer cells as compared to normal cells (Fishelson et al., 2003; Hara et al., 1992; Kinugasa et al., 1999; Murray et al., 2000; Thorsteinsson et al., 1998) making CD46 a promising target for cancer gene therapy. It has been demonstrated that fiber proteins for species B HAdVs (Gaggar et al., 2003; Sirena et al., 2004) bind to CD46. We have previously reported that a HAdV-5 vector incorporating a HAdV-3 fiber protein efficiently transduces cancer cell lines with low CAR expression, with transduction using the HAdV-3 fiber being more effective than

M. Murakami et al. / Virology 407 (2010) 196–205

using the HAdV-11 or the HAdV-35 fiber (Murakami et al., 2010). Although the binding affinity of HAdV-3 to CD46 is low as compared to HAdV-11 and HAdV-35 (Gaggar et al., 2003), a HAdV-5 vector with a chimeric HAdV-3 fiber protein as well as the other species B-chimeric fibers are able to effectively transduce in rodent cells stably expressing human CD46 (Fleischli et al., 2007). Accordingly, we suggest that HAdV5-based vectors that incorporate the HAdV-3 fiber protein and the native HAdV-5 fiber protein will exhibit expanded tropism, resulting in efficient transduction of cells expressing either high or low levels of CAR. We have previously developed HAdV-5 vectors that incorporate two distinct fiber proteins — called fiber-mosaic HAdV vectors — using two different methods: co-infection of cells with two different HAdV ‘parent’ vectors (Takayama et al., 2003) and genetic insertion of a second fiber gene into the HAdV-5 genome (Pereboeva et al., 2004, 2007; Tsuruta et al., 2005, 2007). However, when using the coinfection method, the possibility exists that not all viral progeny incorporate both fiber types, due to variability in the initial coinfection step. Using the second method, we genetically engineered fiber-mosaic HAdV vectors to encode a second fiber gene, including the bacteriophage T4 fibritin domain (Pereboeva et al., 2004), the bacteriophage T4 fibritin containing the 1.3S subunit of Propionibacterium shermanii transcarboxylase (Pereboeva et al., 2007), or the reovirus cellular attachment protein sigma 1 (Tsuruta et al., 2005). These additional fibers were co-expressed with either the wild type HAdV-5 fiber gene, or a chimeric HAdV-3 fiber gene in the case of sigma 1 (Tsuruta et al., 2007). These studies showed that a second fiber protein was able to function as a ligand for its respective cellular receptor and provide expanded vector tropism (Pereboeva et al., 2007; Tsuruta et al., 2005, 2007). Since tumors represent a heterogeneous cell population (Heppner, 1984, 1993), this fibermosaic system could thus be suitable for targeting multiple types of cells with distinct receptor expression profiles within tumors. On this basis, we hypothesized that a genetically generated fibermosaic HAdV-5 vector that incorporates two different types of HAdVderived fiber proteins, a chimeric fiber protein displaying the HAdV-3 shaft and knob domains, as well as the native HAdV-5 fiber protein, would expand tropism to cancer cells by utilizing distinct receptors. To this end, we first determined the optimal genomic configuration for the second fiber gene in order to gain similar levels of expression and incorporation of the two different fiber proteins. We then demonstrated that the fiber-mosaic HAdV particles contain both fiber types and that both fiber types bind their cognate receptors, resulting in cell transduction. Further, we show the capacity of the dual-fiber vector to transduce distinct cell types in a mixed cell population. Results Generation of fiber-mosaic adenoviral vectors We evaluated three different genomic configurations for insertion of a second fiber gene into the HAdV-5 genome. The recombinant shuttle plasmids, adenoviral plasmids and major cloning steps are described in Supplementary information (SFigs. 1–3). The composition of the inserted fiber genes in the fiber-mosaic HAdV genome and

197

the results of the three cloning strategies are summarized in Table 1. Strategy 1, developed in a previous study (Pereboeva et al., 2004), was used to insert the native HAdV-5, and HAdV-11 or 35 fiber genes in a head-to-tail orientation in the late region 5 (L5) transcription unit (SFig. 4). After confirming the presence of both fibers in the recombinant adenoviral genome plasmids by PCR, Ad5EGFP-F11F5 and Ad5EGFP-F35F5 were rescued and upscaled in HEK293 cells. Subsequent western blot analysis of purified viral particles showed that the HAdV-11 and HAdV-35 fiber proteins were not incorporated into the Ad5EGFP-F11F5 and Ad5EGFP-F35F5 particles (data not shown). Fiber-specific PCR analysis of the genomes from Ad5EGFPF11F5 and Ad5EGFP-F35F5 purified particles revealed that the second fiber gene was deleted (data not shown). The deletion of the second fiber gene from the adenovirus genome was likely a result of homologous recombination between the identical nucleotide sequences in the fiber tail domain (data not shown). We therefore used an alternate cloning strategy wherein the additional fiber gene was positioned in an anti-sense orientation downstream of the early region 4 (E4) transcription unit, preceded by an internal ribosome entry site (IRES) (SFig. 4, strategy 2). With the second fiber gene in anti-sense orientation with respect to the native fiber gene, homologous recombination between the HAdV-5 fiber tail sequences would result in both fiber genes being deleted from the genome, thereby ensuring that fully infectious particles with only dual-fiber genes would be produced in infected cells. Ad5EGFP-F5-IRESF11 and Ad5EGFP-F5-IRESF35 were successfully generated and propagated in HEK293 cells. However, western blot analysis of lysates extracted from infected cells showed very low expression of the second fiber whereas the native HAdV-5 fiber protein was highly expressed (data not shown). As a result, these HAdV particles contained mainly the HAdV-5 fiber protein with virtually undetectable amounts of the second fiber (data not shown). These data indicate that the second strategy using the E4 promoter with a downstream IRES to regulate the expression of a second fiber gene is not suitable for generating fiber-mosaic adenoviruses. To increase expression of the second fiber protein in infected cells, we utilized a cytomegalovirus (CMV) promoter and a polyadenylation (polyA) site at the 3′ end of the second fiber gene (SFig. 4, strategy 3). We successfully rescued and propagated AdF5-F3EGFP in HEK293 cells and confirmed the presence of both fiber genes in the genome of purified AdF5-F3EGFP particles by PCR analysis using a primer set which anneals at 100 base pairs (bp) upstream and 100 bp downstream of the HAdV-5 fiber gene locus (Figs. 1A and B). We next sought to confirm the expression profile of each fiber following infection of cells with Ad5F5-F3EGFP. HEK293 cells were infected and expression of both fiber proteins was confirmed at various time points by western blot analysis. We observed similar expression patterns for both fibers beginning at 12 h post-infection and reaching maximum expression at approximately 24 h (Fig. 1C). We next confirmed the presence of both types of fiber proteins in purified Ad5F5-F3EGFP viral particles by western blot analysis. As shown in Fig. 1D lane 3, both the HAdV-5 fiber (upper band) and HAdV-3 fiber (lower band) were detected as monomers following heat denaturation. In order to rule out the presence of heterotrimerized fibers, we performed non-denaturing western analysis on the

Table 1 Summary of the result of the three cloning strategies of additional fiber gene. Strategy

Insertion

Orientation

Promoter

Generationa

Expressionb

Incorporationc

1 2 3

L5 E4 E4

Sense Anti-sense Anti-sense

Major late E4 CMV

No Yes Yes

No Little Yes

No No Yes

a b c

Generation of mosaic adenovirus by transfection of the recombinant adenoviral plasmid. The expression level of the second fiber protein in infected cells. Incorporation of the second fiber protein along with the original fiber protein in the purified adenoviral particles.

198

M. Murakami et al. / Virology 407 (2010) 196–205

Fig. 1. Characterization of mosaic HAdV vectors. A: Schema of the fiber gene(s) of Ad5EGFP, Ad5F3EGFP or Ad5F5-F3EGFP as well as the primer binding positions for PCR analysis. The gray arrows indicate the orientation of the genes. Localization of primers is indicated by the black thin arrows. B: Validation of the presence of two fiber genes in the HAdV-5 genome by PCR; M: DNA size markers; lane 1: PCR product using DNA from purified Ad5EGFP as a template; lane 2: PCR product using DNA from purified Ad5F3EGFP as a template; lane 3: PCR product using the recombinant HAdV genome encoding two different fiber genes as a template; lane 4: PCR product using DNA from purified Ad5F5-F3EGFP as a template. C: Detection of the expression of the fiber proteins from aliquots of the purified HAdV vectors equal to 5 × 109 VP or 5 μg of total proteins extracted from virus-infected cells at various time points. D, E: Detection of fiber proteins incorporated into purified viral particles by western blot analysis. A total of 5 × 109 VP of the purified HAdV vectors were boiled and applied to a 12% SDS-PAGE gel in D, or run without boiling on a 10% SDS-PAGE in E; separated viral proteins were transferred to a PVDF membrane and probed by an antibody against the HAdV-5 fiber tail (4D2). Lane 1: Ad5EGFP; lane 2: Ad5F3EGFP; lane 3: Ad5F5-F3EGFP; protein molecular mass markers (in kDa) are indicated on the left.

viral particles. As shown in Fig. 1E lane 3, only HAdV-5 and HAdV-3 fiber homotrimers were detected. We also sought to determine whether the thermostability of the Ad5F5-F3EGFP vector was altered by incorporation of two different fiber proteins. Under all conditions tested, the thermostability of Ad5F5-F3EGFP was not different from Ad5EGFP or Ad5F3EGFP (data not shown). Moreover, one-step growth curve kinetics of Ad5F5F3EGFP was similar to that of Ad5EGFP and Ad5F3EGFP (data not shown). In the aggregate, these data demonstrate that our strategy 3 is a suitable approach for the construction of a fiber-mosaic HAdV5 vector. Analysis of viral particle morphology and incorporation of fiber proteins by transmission electron microscopy To directly examine the Ad5F5-F3EGFP viral particles and to confirm the incorporation of both types of fiber proteins into single viral particles, we performed transmission electron microscopy using the Ad5F5-F3EGFP viral particles as well as control vectors Ad5EGFP and Ad5F3EGFP, which display the HAdV-5 and HAdV-3 fibers, respectively. All viral particles, including Ad5F5-F3EGFP, had the expected icosahedral morphology (Fig. 2, upper panels). The images

of Ad5EGFP show the incorporation of the HAdV-5 fiber proteins (black arrows in Fig. 2, lower panel, left), whereas Ad5F3EGFP particles incorporated the much shorter HAdV-3 fibers (orange arrows in Fig. 2, lower panel, center). The electron micrograph of Ad5F5-F3EGFP shows the incorporation of the HAdV-5 fiber (black arrows in Fig. 2, lower panel, right) and HAdV-3 fiber (orange arrows in Fig. 2, lower panel, right) in a single viral particle. Gene transfer of fiber-mosaic adenoviral vector The binding affinity of HAdV-3 to human CD46 is lower than that of HAdV-11 and HAdV-35 (Gaggar et al., 2003). However, we have previously reported that transduction efficiency of a recombinant HAdV-5 vector expressing the HAdV-3 fiber is higher than that using the HAdV-11 fiber or the HAdV-35 fiber proteins in CD46-positive cancer cells (Murakami et al., 2010). In addition, a chimeric HAdV-3 fiber-modified HAdV-5 vector as well as the other species of B-chimeric vectors similarly transduced rodent cells stably expressing human CD46 (Fleischli et al., 2007). Therefore, we evaluated the transduction efficiency of Ad5F5-F3EGFP containing both HAdV-5 and HAdV-3 fibers in CHO cells, CHO-C2 cells (expressing the C2 isoform of CD46, but not CAR) and CHO-CAR cells (expressing human CAR). At 48 h post-

M. Murakami et al. / Virology 407 (2010) 196–205

199

Fig. 2. Electron micrograph of negatively stained viral particles of Ad5EGFP, Ad5F3EGFP and Ad5F5-F3EGFP. Black arrows indicate the incorporation of the HAdV-5 fiber. Orange arrows indicate the incorporation of the HAdV-3 fiber. Scale bar represents 100 nm.

infection (hpi), EGFP transgene expression was observed by fluorescence microscopy. As shown in Fig. 3, CHO cells were refractory to transduction by all HAdV vectors. Ad5EGFP containing the HAdV-5 fiber

provided gene delivery to CHO-CAR cells only, while Ad5F3EGFP, which displays the HAdV-3 fiber, transduced only the CD46-expressing CHOC2 cells. In contrast, the dual-fiber Ad5F5-F3EGFP vector transduced

Fig. 3. Gene transfer of Ad5EGFP, Ad5F3EGFP and Ad5F5-F3EGFP in CAR-negative CHO, human CAR-expressing CHO-CAR cells, and human CD46-expressing CHO-C2 cells. Cells were infected with adenovirus at an MOI of 100 pfu/cell for 48 h, and fluorescent signal for EGFP was observed by fluorescence microscopy.

200

M. Murakami et al. / Virology 407 (2010) 196–205

both CAR- and CD46-positive CHO cells, suggesting that Ad5F5-F3EGFP can use CAR and CD46 as cellular receptors. To further confirm these results, we also quantified the percentage of EGFP-positive cells and the mean EGFP fluorescence intensity by flow cytometry following infection with each vector. As expected, CHO cells showed the lowest percentage of EGFP-positive cells and the lowest overall fluorescence intensity in positive cells (SFig. 5A). In CHO-CAR cells, Ad5EGFP transduction resulted in nearly 100% positive cells with a concomitant 13-fold increase in EGFP fluorescence intensity compared to CHO cells. Similar increases in percentage of positive cells and fluorescence intensity were observed in Ad5F5-F3EGFP-infected CHO-CAR cells compared to CHO cells (SFig. 5B). In CHO-C2 cells, Ad5F3EGFP and Ad5F5-F3EGFP transduction resulted in approximately 80% and 75% positive cells, respectively, in contrast to only 20% positive cells with minimal fluorescence intensity with Ad5EGFP (SFig. 5C). Collectively, these gene transfer data strongly suggest that Ad5F5-F3EGFP transduction is mediated through both CAR and CD46. Gene transfer is mediated by both fiber proteins of Ad5F5-F3EGFP To confirm that Ad5F5-F3EGFP transduces cells by using both fiber proteins, we performed competitive inhibition blocking assays with recombinant fiber knob proteins in A549 cells which express both CAR and CD46 (Murakami et al., 2010). In the absence of blocking proteins, all vectors efficiently transduced A549 cells (Fig. 4, top row), with

approximately 96 to 99% of cells being EGFP-positive (SFig. 6). The presence of free HAdV-5 and HAdV-3 fiber knob proteins markedly reduced gene transfer of control vectors Ad5EGFP and AdF3EGFP, respectively (Fig. 4, rows 2 and 3). Quantification of this observation using flow cytometry showed that the HAdV-5 fiber knob protein blocked Ad5EGFP gene transfer in approximately 50% of cells (SFig. 6A) and decreased EGFP mean fluorescence intensity of the remaining positive cells by approximately 94% compared to unblocked (SFig. 6B). Likewise, the HAdV-3 fiber knob protein blocked Ad5F3EGFP gene transfer in approximately 15% of cells and decreased the mean fluorescence intensity of positive cells by approximately 50% compared to no block (SFig. 6). Therefore, these data demonstrated that the recombinant HAdV fiber proteins inhibited cell entry through the HAdV-5 and HAdV-3 fibers of Ad5EGFP and Ad5F3EGFP, respectively. HAdV-3 and HAdV-5 knob proteins used individually diminished but did not completely block transduction by Ad5F5F3EGFP (Fig. 4, rows 2 and 3). However, simultaneous addition of both fiber knob proteins blocked virtually all EGFP gene delivery provided by Ad5F5-F3EGFP (Fig. 4, bottom row). The corresponding flow cytometry analysis showed that simultaneous incubation of the HAdV-3 and HAdV-5 fiber knob proteins blocked Ad5F5-F3EGFP transduction in approximately 20% of cells (SFig. 6A) and reduced the fluorescence intensity by 80% in remaining positive cells (SFig. 6B). In contrast, flow cytometry analysis of EGFP transfer using only one type of HAdV fiber knob protein for blocking showed more limited effects

Fig. 4. Mosaic adenovirus-mediated gene transfer with blocking of HAdV-3 and HAdV-5 knobs. A549 cells were incubated with or without recombinant HAdV-5 or/and -3 knob proteins at a final concentration of 50 μg/ml at 4 °C for 1 h. After blocking using the recombinant knob proteins, A549 cells were infected with Ad5EGFP, Ad5F3EGFP and Ad5F5F3EGFP at an MOI of 100 pfu/cell for 48 h. The EGFP signal was observed by fluorescence microscopy.

M. Murakami et al. / Virology 407 (2010) 196–205

(SFig. 6). These data confirm that both HAdV-5 and HAdV-3 fiber proteins contained in the Ad5F5-F3EGFP particles are functional and can be used for cell entry. Gene transfer of Ad5F5-F3EGFP in a mixed cell population We next tested the ability of Ad5F5-F3EGFP to transduce a mixed population of cells in vitro, as a way to simulate transduction of a heterogeneous tumor-like environment. To distinguish between the two cell lines, we first labeled CHO-CAR cells with a red fluorescent protein via transduction with a lentiviral vector encoding the mCherry gene (Shaner et al., 2004) (CHO-CARmCherry). Subsequently, CARnegative, CD46-positive human prostate cancer PC-3 cells (Murakami et al., 2010) and CHO-CARmCherry cells were seeded together in wells of a six-well plate and incubated for 48 h. The mixed cells were visualized under bright field microscopy (Fig. 5, top row) and fluorescence microscopy (Fig. 5, second row). Individual co-cultures

201

were then transduced at a multiplicity of infection (MOI) 100 pfu/cell with dual-fiber Ad5F5-F3EGFP or control vectors Ad5EGFP and Ad5F3EGFP, and EGFP expression was visualized via fluorescence microscopy after 48 h (Fig. 5, third row). Consistent with our previous results (Fig. 4), Ad5EGFP transduced only CAR-positive CHO-CARmCherry cells, as determined by the cellular co-localization of the EGFP transgene and mCherry fluorescence signals, shown in yellow (Fig. 8, bottom row). Similarly, Ad5F3EGFP transduced only PC-3 cells since only the EGFP transgene could be observed and no fluorescence co-localization was noted in these cells. In contrast, both CHOCARmCherry and PC-3 cells were transduced by the dual-fiber Ad5F5F3EGFP vector, as judged by the presence of EGFP-positive PC-3 cells (green) and cells wherein the EGFP signal co-localized with the red fluorescence signal of the CHO-CARmCherry cells (yellow) (Fig. 5, bottom row). These data confirmed that the dual-fiber Ad5F5-F3EGFP transduces both cell types in a mixed cell population based on the usage of different fibers for cellular transduction.

Fig. 5. Gene transfer of Ad5EGFP, Ad5F3EGFP and Ad5F5-F3EGFP in the mixed population of PC-3 cells and CHO-CARmCherry cells. Mixed cells were infected with adenovirus at an MOI of 100 pfu/cell for 48 h. A: Illustration of color indication of fluorescent signals. B: Detection of green and red fluorescent signals in a mixed population.

202

M. Murakami et al. / Virology 407 (2010) 196–205

Discussion In this study, we generated a fiber-mosaic HAdV vector that expresses the wild type HAdV-5 fiber protein and a chimeric fiber protein comprised of the HAdV-5 tail domain and the HAdV-3 shaft and knob domains. We demonstrated that Ad5F5-F3EGFP utilizes distinct receptors using the two different fiber proteins, thus achieving tropism expansion compared to the parent single-fiber vectors. It proved challenging to achieve equal incorporation of two different types of fiber proteins in a virus population (Schoggins et al., 2003). We evaluated three strategies for positioning the second fiber gene in the viral genome. In strategy 1, we confirmed that the second fiber gene was deleted from the adenovirus genome during propagation, likely as a result of homologous recombination between the identical nucleotide sequences in the fiber tail domains. Deletion of the second fiber in the genome therefore resulted in the absence of the second fiber as determined by western blot. In strategy 2, we relied on the E4 promoter and a downstream IRES element to drive expression of the second fiber. The E4 promoter is an early promoter with maximal activity during the early stage of infection, while the major late promoter for transcription of the first fiber gene is attenuated at that time (Fessler and Young, 1998). The major late promoter is fully activated during the late phase of infection, and the transcripts for the major late genes containing the tripartite leader sequence are then selectively exported from the nucleus to the cytoplasm (Huang and Flint, 1998) and efficiently translated (Logan and Shenk, 1984). Attenuation of the E4 promoter activity in the late stage of infection appeared to result in low expression of the second fiber protein as compared to expression of the HAdV-5 fiber protein through the major late promoter. Only one of these strategies succeeded at generating a fiber-mosaic HAdV vector incorporating two different types of fiber proteins. We propose that insertion of the second fiber gene in an anti-sense orientation, as done in the successful strategy, prevents the production of vectors that encode only one fiber gene (Table 1). In addition, a CMV promoter ensured sufficiently high expression of the additional fiber protein (Fig. 1C), resulting in similar incorporation levels of the two distinct fibers in Ad5F5-F3EGFP particles (Figs. 1D and E). Demonstrating that both fibers are incorporated into a single adenovirus particle is problematic using current technologies: western blotting would show that a composite viral population contains both kinds of fiber proteins, and enzyme-linked immunosorbent assays (ELISA) or immunoprecipitation (IP)-western blot analysis is not suitable due to the formation of virus particle aggregates (Fig. 2, upper panel) in such assays. On this basis, we chose to directly visualize the fiber proteins incorporated in single viral particles by transmission electron microscopy (Fig. 2). The micrograph of Ad5F5-F3EGFP is the first to confirm incorporation of two different fiber protein types in a single viral particle that was generated by genetic modification, as distinct from similar images of wild type HAdV-41 which naturally contains two distinct fiber proteins (Favier et al., 2002). Unfortunately, this methodology did not allow us to determine the total number or the incorporation ratio of the HAdV-3 and HAdV-5 fiber proteins contained in individual particles. Since the fibers on the opposite side of the particles are not observable and the length of the fibers projecting perpendicular to the grid plane cannot be determined, only the fibers projecting laterally from the capsid can be reliably distinguished and quantitated. Our purpose for generating fiber-mosaic HAdV vectors is to efficiently infect heterogeneous cancer cell populations that may exist within tumors through the use of two different receptors. Ad5F5-F3EGFP, as developed in this study, was able to deliver a transgene into both CHO cells expressing CD46 as well as CHO cells expressing CAR, whereas the respective ‘parent’ vectors, either HAdV-5 or the chimeric HAdV-5 containing a HAdV-3-derived fiber, were only able to transduce CHO cells expressing CAR or CD46,

respectively (Fig. 3 and SFig. 5). On the other hand, co-infection using these two viruses would also achieve gene transfer in a heterogeneous population of tumor cells. However, we have shown that simultaneous co-transduction with two kinds of HAdV-5 vectors, each containing a different fiber, resulted in reduced transduction as compared with a fiber-mosaic HAdV vector containing the HAdV-5 and the HAdV-3 fibers produced by co-infection in HEK293 cells (Takayama et al., 2003), emphasizing the advantage of the fibermosaic approach. Using blocking experiments with recombinant HAdV-5 and/or HAdV-3 knob proteins, we also demonstrated that both types of fiber proteins incorporated into the Ad5F5-F3EGFP particles participated in cellular transduction (Fig. 4 and SFig. 6). Furthermore, we demonstrated that Ad5F5-F3EGFP delivered a transgene to cells that insufficiently express either CAR or CD46 in a mixed population environment, which confirmed our hypothesis that incorporating two different types of fibers would expand a vector's tropism for different cell types in a population (Fig. 5). Since tumors contain a heterogeneous population of cells (Heppner, 1984, 1993), the fiber-mosaic HAdV vector system might be suitable for achieving high transduction levels in tumor tissues. Clearly, using this technology, fiber-mosaic vectors of diverse configurations could be derived that are suited to the specific biology of target tumor cell populations. In this report we have established the technical means for the expression of a second fiber gene within the HAdV-5 genome and that both fibers retain biological function. One potential application of a genetically generated mosaic HAdV vector would be in the context of conditionally replicative adenoviral vectors (CRAds), where the progeny of the replicative HAdVs would also encode two different fiber genes, thus also utilizing distinct receptors to facilitate successive rounds of infection and cell killing within the tumor tissue. This technique should allow the co-expression of a wide variety of HAdV fiber proteins that may be required for optimal transduction of target tissues.

Materials and methods Cell lines Chinese hamster ovary (CHO) cells, human embryonic kidney (HEK) 293 (Graham et al., 1977) and human lung epithelial A549 cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA). HEK293A cells were purchased from Invitrogen (Carlsbad, CA). All cells were cultured in Dulbecco's Modified Eagle's Medium/Nutrient Mixture F-12 Ham, (DMEM/F12; Sigma-Aldrich; St. Louis, MO) containing 10% fetal bovine serum, (FBS; Hyclone; Logan, UT), 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin (Mediatech, Inc; Herndon, VA). CHO-CAR cells, stably expressing human CAR, are derivatives of CHO (Bergelson et al., 1997) and were cultured in MEM Alpha medium (MEM Alpha; Invitrogen) containing 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin. A CHO-derived cell line, CHO-C2, which stably expresses the C2 isoform of human CD46 (Oglesby et al., 1991) was cultured in Eagle's minimum essential medium (MEM; Mediatech) containing 10% FBS, 2 mM L-glutamine, 10 mM MEM non-essential amino acids (Invitrogen), 100 U/ml penicillin and 100 μg/ml streptomycin. The CHO-CAR expressing mCherry were generated by infection of these cells with lentiviral vector containing the mCherry gene (Dr. Justin C. Roth, Division of Human Gene Therapy, University of Alabama at Birmingham, unpublished vector). Briefly, the corresponding cell culture medium including the lentiviral vector and hexademethrine bromide (Sigma) at a final concentration of 8 μg/ml was added to the CHO-CAR cells grown in a 6-well plate. After 24 h incubation, the medium was replaced with fresh cell culture medium. After 48 h incubation, the cells were trypsinized, reseeded, and maintained in

M. Murakami et al. / Virology 407 (2010) 196–205

the corresponding medium with puromycin (Mediatech) at a final concentration of 10 μg/ml until the control cells were killed. All cells were propagated at 37 °C in a 5% CO2 atmosphere. The cell lines infected with the HAdV vectors were maintained using the corresponding cell culture medium but containing 2% instead of 10% FBS.

Construction of recombinant plasmids We described and schematically represented the construction methods for strategies 1 and 2 in Supplementary information and Supplementary Figs. 1 and 2 (SFigs 1 and 2). We also schematically represented the construction method for strategy 3 in Supplementary Fig. 3. In brief, the plasmid pcDNA3.1_Ad5T-3SK (Mr. Anand C. Annan, Division of Human Gene Therapy, University of Alabama at Birmingham, unpublished plasmid) is pcDNA3.1/Zeo (Invitrogen) encoding a chimeric HAdV-3 fiber gene containing the nucleotide sequences for the HAdV-5 fiber tail, the HAdV-3 fiber shaft and knob domains. The plasmid pcDNA3.1_Ad5T-3SK was digested with the NruI and PvuII, and the expression cassette carrying the cytomegalovirus (CMV) promoter, the chimeric HAdV-3 fiber gene and the bovine growth hormone polyadenylation signal (BGH polyA) was ligated into the EcoRV site of pKAN3.1F5-MCS (Dr. Hideyo Ugai, Division of Human Gene Therapy, University of Alabama at Birmingham, unpublished plasmid), which encodes the HAdV-5 fiber gene. The resulting vectors encode the HAdV-5 fiber gene in sense orientation, and the additional fiber expression in anti-sense or sense orientations. We analyzed the restriction maps of the resulting vectors and selected the plasmid encoding the second fiber expression cassette in the anti-sense orientation. This plasmid was named pKAN3.1F5CMV5T3SKpA. The nucleotide sequences of the DNA fragments cloned into all plasmids were confirmed by the UAB DNA sequencing core facility.

Construction of adenoviral genome plasmids The fiber shuttle plasmid pKAN3.1F5CMV5T3SKplA was homologously recombined with SwaI-linearized pVK900, a HAdV5 genome plasmid which encodes the expression cassette for an enhanced green fluorescence protein (EGFP) gene in the deleted E1 region (Dr. Victor Krasnykh, University of Texas MD Anderson Cancer Center, unpublished plasmid) in Escherichia coli strain BJ5183. The homologous recombination resulted in the fiber-mosaic adenoviral genome plasmid pMM900. F5CMV5T3SKpA. We also constructed the control vectors, pMM905 and pMM903, which encode the HAdV-5 fiber or the HAdV-3 fiber gene, respectively, by using fiber shuttle vectors pKAN3.1 (Borovjagin et al., 2005) and pKAN5T3SK (Murakami et al., 2010) and SwaI-linearized pVK900. The construction procedures for the other adenoviral plasmids are described in Supplementary information.

Adenovirus generation, propagation, purification, and titration The recombinant adenoviral plasmids were linearized by PacI and transfected into HEK293 cells using Lipofectamine™ LTX reagent (Invitrogen). All vectors propagated in HEK293 cells were purified by two rounds of CsCl gradient ultracentrifugation (Kanegae et al., 1994). CsCl was removed by dialysis against PBS (pH 7.4) containing 10% glycerol. The HAdV vectors were stored at −80 °C prior to use. The infectious titer (plaque forming units [pfu]/ml) of purified HAdV vectors was determined by triplicate TCID50 assays using HEK293A cells, as described elsewhere (Kanegae et al., 1994). The physical titer (viral particles [VP]/ml) were determined by Maizel's method with a conversion factor of 1.1 × 1012 VP/ml per absorbance unit at 260 nm (Maizel et al., 1968).

203

Polymerase chain reaction (PCR) analysis of the fiber region To confirm the presence of the two fiber genes in the HAdV genomes, 107 VP of purified HAdV vectors were boiled at 95 °C for 5 min and analyzed by PCR using the TaqPCR Master Kit (Qiagen Inc., Valencia, CA). Forward (Fiber −100) and reverse (Fiber + 100) primers (Murakami et al., 2010) were used to amplify the region containing the fiber loci. The following PCR conditions were applied: 1 min denaturation at 96 °C, 1 min annealing at 60 °C, and 5 min extension at 72 °C. Western blot analysis of adenoviral fiber proteins Aliquots of purified HAdV vectors equal to 5 × 109 VP or 5 μg of total protein extracted from virus-infected cells in RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% SDS, and 50 mM Tris–HCl, [pH 8.0]) at various time points (1, 3, 6, 12, 18, 24, 36 and 48 h post-infection [h.p.i]) were denatured by boiling in Laemmli sample buffer (Bio-Rad laboratories Inc., Hercules, CA) at 95 °C for 10 min. The proteins were separated by electrophoresis in sodium dodecyl sulfate 10 or 12% polyacrylamide gels (SDS-PAGE). Subsequently, western blot analysis was performed using a mouse monoclonal anti-HAdV-5 fiber tail domain antibody (4D2) (Lab Vision Corp., Fremont, CA) as described elsewhere (Murakami et al., 2010). The protein concentration was determined using the DC protein assay kit (Bio-Rad Laboratories Inc., Hercules, CA) according to the manufacturer's instructions. Pre-stained protein ladder of Kaleidoscope Standards (Bio-Rad Laboratories, Inc.) was used for identification of the virus-specific proteins in western blots. Transmission electron microscopy (TEM) A total of 1.0 × 108 VP of purified HAdV particles was transferred into a 1.5-ml tube and vortexed for 20 seconds (s) to break up aggregates. Subsequently, viral particles were fixed with an equal volume of fixative buffer (1.25% glutaraldehyde and 2% paraformaldehyde in 30 mM Hepes–NaOH buffer [pH 7.4]) and incubated at room temperature for 30 min. Ten microliters of the fixed adenoviral particles were placed on parafilm, and a glow discharged grid (400-mesh, pure carbon support films; Electron Microscopy Sciences, Hatfield, PA) was placed on the viral solution for 10 s. The carbon grid was washed with 100 μl of filtered 10 mM Tris–HCl buffer ([pH 7.4] containing 30 mM NaCl) on parafilm six times at room temperature for 30 s. The adenoviral particles on the grid were negatively stained with 1% (wt/vol) uranyl acetate at room temperature for 10 s and then air dried. Electron micrographs for the adenoviral particles were taken with FEI Tecnai F20 200 kV at the HighResolution Imaging Facility of the University of Alabama at Birmingham. Gene transfer assays Cells were grown in 6-well plates, and the cell numbers were counted to determine MOI. Cells were infected with the HAdV vectors at an MOI of 100 pfu/cell. For the gene transfer assay, we assayed triplicates for each virus and took three individual images for each well. A well contained between 5.5 × 105 and 3 × 106 cells. For blocking experiments, 1 ml of the recombinant HAdV-5 knob protein and/or HAdV-3 knob protein (Krasnykh et al., 1996) diluted to a final concentration of 50 μg/ml in FBS-free medium was added to the wells. No blocking agent was added to the control wells. Cells were incubated with the recombinant HAdV-5 knob protein and/or HAdV-3 knob protein at 4 °C for 1 h; subsequently, HAdV vectors at an MOI of 100 pfu/cell were added to the cells in 1 ml of medium containing 4% FBS. After incubation at 37 °C for 48 h, we observed the fluorescent signal for EGFP by fluorescence microscopy. The signal for the fluorescent proteins was detected at a low magnification (×200) by a fluorescent microscope.

204

M. Murakami et al. / Virology 407 (2010) 196–205

For the gene transfer experiment using a mixed population of two different cell lines, cells were grown in a 75-cm flask, and the cell numbers were counted for the subsequent seeding. PC-3 (2.25 × 105 cells per well) and CHO-CARmCherry (0.75 × 105 cells per well) cells were seeded and grown in the 6-well plates and incubated for 48 h. The mixed cells were counted to determine MOI and were infected with the HAdV vectors at an MOI of 100 pfu/cell in 2 ml of medium containing 2% FBS. After incubation at 37 °C for 48 h, we observed the fluorescent signal for EGFP by fluorescence microscopy. The signals for the fluorescent proteins were detected at a low magnification (×200) by a fluorescent microscope. Flow cytometry Cells were grown in 6-well plates, and the cell numbers were counted to determine MOI (pfu/cell). We infected cells with adenovirus at an MOI of 100 pfu/cell and incubated infected cells for 48 h. Infected cells were harvested by treatment with 0.25% trypsin–2.21 mM EDTA solution (Sigma-Aldrich) and suspended with 5 ml of medium containing 2% FBS. The cell suspension was centrifuged at 800 ×g for 5 min at 4 °C, and the cells were resuspended with 2 ml of cold corresponding medium and the cells were transferred into the tube and centrifuged at 800 ×g for 5 min at 4 °C. Subsequently, the cells washed with SM buffer (PBS [pH 7.4] containing 0.1% bovine serum albumin [BSA] and 0.01% sodium azide, stored at 4 °C) 2 times, and analyzed with a FACScan (Becton Dickinson, Mountain Vies, CA) at the UAB Cell Sorting Facility. Supplementary materials related to this article can be found online at doi:10.1016/j.virol.2010.08.010. Acknowledgments We are grateful to Drs. Jeffrey M. Bergelson (University of Pennsylvania), Victor Krasnykh (The University of Texas), Roger Y. Tsien (University of California at San Diego), Justin C. Roth, and Mr. Anand C. Annan, for providing CHO-CAR cells, unpublished plasmid pVK900, the mCherry construct, an unpublished lentiviral vector, and the unpublished plasmid pcDNA3.1_Ad5T-3SK, respectively. We thank Dr. Zhiying You for statistical analysis. We also thank Dr. Masato Yamamoto for useful advice. We gratefully thank Dr. Terje Dokland and Melissa F. Chimento for technical support with TEM at the High-Resolution Imaging Facility of the University of Alabama at Birmingham. This project was supported by award number P01CA104177 from the National Cancer Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute and the National Institutes of Health. References Anders, M., Vieth, M., Rocken, C., Ebert, M., Pross, M., Gretschel, S., Schlag, P.M., Wiedenmann, B., Kemmner, W., Hocker, M., 2009. Loss of the coxsackie and adenovirus receptor contributes to gastric cancer progression. Br. J. Cancer 100 (2), 352–359. Auer, D., Reimer, D., Porto, V., Fleischer, M., Roessler, J., Wiedemair, A., Marth, C., MullerHolzner, E., Daxenbichler, G., Zeimet, A.G., 2009. Expression of coxsackie-adenovirus receptor is related to estrogen sensitivity in breast cancer. Breast Cancer Res. Treat. 116 (1), 103–111. Bergelson, J.M., Cunningham, J.A., Droguett, G., Kurt-Jones, E.A., Krithivas, A., Hong, J.S., Horwitz, M.S., Crowell, R.L., Finberg, R.W., 1997. Isolation of a common receptor for Coxsackie B viruses and adenoviruses 2 and 5. Science 275 (5304), 1320–1323. Borovjagin, A.V., Krendelchtchikov, A., Ramesh, N., Yu, D.C., Douglas, J.T., Curiel, D.T., 2005. Complex mosaicism is a novel approach to infectivity enhancement of adenovirus type 5-based vectors. Cancer Gene Ther. 12 (5), 475–486. Campos, S.K., Barry, M.A., 2007. Current advances and future challenges in adenoviral vector biology and targeting. Curr. Gene Ther. 7 (3), 189–204. Favier, A.L., Schoehn, G., Jaquinod, M., Harsi, C., Chroboczek, J., 2002. Structural studies of human enteric adenovirus type 41. Virology 293 (1), 75–85. Fessler, S.P., Young, C.S., 1998. Control of adenovirus early gene expression during the late phase of infection. J. Virol. 72 (5), 4049–4056.

Fishelson, Z., Donin, N., Zell, S., Schultz, S., Kirschfink, M., 2003. Obstacles to cancer immunotherapy: expression of membrane complement regulatory proteins (mCRPs) in tumors. Mol. Immunol. 40 (2–4), 109–123. Fisher, K.J., Choi, H., Burda, J., Chen, S.J., Wilson, J.M., 1996. Recombinant adenovirus deleted of all viral genes for gene therapy of cystic fibrosis. Virology 217 (1), 11–22. Fleischli, C., Sirena, D., Lesage, G., Havenga, M.J., Cattaneo, R., Greber, U.F., Hemmi, S., 2007. Species B adenovirus serotypes 3, 7, 11 and 35 share similar binding sites on the membrane cofactor protein CD46 receptor. J. Gen. Virol. 88 (Pt 11), 2925–2934. Gaggar, A., Shayakhmetov, D.M., Lieber, A., 2003. CD46 is a cellular receptor for group B adenoviruses. Nat. Med. 9 (11), 1408–1412. Graham, F.L., Smiley, J., Russell, W.C., Nairn, R., 1977. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J. Gen. Virol. 36 (1), 59–74. Hara, T., Kojima, A., Fukuda, H., Masaoka, T., Fukumori, Y., Matsumoto, M., Seya, T., 1992. Levels of complement regulatory proteins, CD35 (CR1), CD46 (MCP) and CD55 (DAF) in human haematological malignancies. Br. J. Haematol. 82 (2), 368–373. Heppner, G.H., 1984. Tumor heterogeneity. Cancer Res. 44 (6), 2259–2265. Heppner, G.H., 1993. Cancer cell societies and tumor progression. Stem Cells 11 (3), 199–203. Huang, W., Flint, S.J., 1998. The tripartite leader sequence of subgroup C adenovirus major late mRNAs can increase the efficiency of mRNA export. J. Virol. 72 (1), 225–235. Imperiale, M.J., Kochanek, S., 2004. Adenovirus vectors: biology, design, and production. Curr. Top. Microbiol. Immunol. 273, 335–357. Kanegae, Y., Makimura, M., Saito, I., 1994. A simple and efficient method for purification of infectious recombinant adenovirus. Jpn J. Med. Sci. Biol. 47 (3), 157–166. Kinugasa, N., Higashi, T., Nouso, K., Nakatsukasa, H., Kobayashi, Y., Ishizaki, M., Toshikuni, N., Yoshida, K., Uematsu, S., Tsuji, T., 1999. Expression of membrane cofactor protein (MCP, CD46) in human liver diseases. Br. J. Cancer 80 (11), 1820–1825. Kirschfink, M., 2001. Targeting complement in therapy. Immunol. Rev. 180, 177–189. Kochanek, S., Clemens, P.R., Mitani, K., Chen, H.H., Chan, S., Caskey, C.T., 1996. A new adenoviral vector: replacement of all viral coding sequences with 28 kb of DNA independently expressing both full-length dystrophin and beta-galactosidase. Proc. Natl Acad. Sci. USA 93 (12), 5731–5736. Korn, W.M., Macal, M., Christian, C., Lacher, M.D., McMillan, A., Rauen, K.A., Warren, R.S., Ferrell, L., 2006. Expression of the coxsackievirus- and adenovirus receptor in gastrointestinal cancer correlates with tumor differentiation. Cancer Gene Ther. 13 (8), 792–797. Krasnykh, V.N., Mikheeva, G.V., Douglas, J.T., Curiel, D.T., 1996. Generation of recombinant adenovirus vectors with modified fibers for altering viral tropism. J. Virol. 70 (10), 6839–6846. Logan, J., Shenk, T., 1984. Adenovirus tripartite leader sequence enhances translation of mRNAs late after infection. Proc. Natl Acad. Sci. USA 81 (12), 3655–3659. Maizel Jr., J.V., White, D.O., Scharff, M.D., 1968. The polypeptides of adenovirus. I. Evidence for multiple protein components in the virion and a comparison of types 2, 7A, and 12. Virology 36 (1), 115–125. Martin, T.A., Watkins, G., Jiang, W.G., 2005. The Coxsackie-adenovirus receptor has elevated expression in human breast cancer. Clin. Exp. Med. 5 (3), 122–128. Matsumoto, K., Shariat, S.F., Ayala, G.E., Rauen, K.A., Lerner, S.P., 2005. Loss of coxsackie and adenovirus receptor expression is associated with features of aggressive bladder cancer. Urology 66 (2), 441–446. Murakami, M., Ugai, H., Belousova, N., Pereboev, A., Dent, P., Fisher, P.B., Everts, M., Curiel, D.T., 2010. Chimeric adenoviral vectors incorporating a fiber of human adenovirus 3 efficiently mediate gene transfer into prostate cancer cells. Prostate 70 (4), 362–376. Murray, K.P., Mathure, S., Kaul, R., Khan, S., Carson, L.F., Twiggs, L.B., Martens, M.G., Kaul, A., 2000. Expression of complement regulatory proteins-CD 35, CD 46, CD 55, and CD 59-in benign and malignant endometrial tissue. Gynecol. Oncol. 76 (2), 176–182. Oglesby, T.J., White, D., Tedja, I., Liszewski, K., Wright, L., Van den Bogarde, J., Atkinson, J.P., 1991. Protection of mammalian cells from complement-mediated lysis by transfection of human membrane cofactor protein and decay-accelerating factor. Trans. Assoc. Am. Physicians 104, 164–172. Okegawa, T., Pong, R.C., Li, Y., Bergelson, J.M., Sagalowsky, A.I., Hsieh, J.T., 2001. The mechanism of the growth-inhibitory effect of coxsackie and adenovirus receptor (CAR) on human bladder cancer: a functional analysis of car protein structure. Cancer Res. 61 (17), 6592–6600. Palmer, D.J., Ng, P., 2005. Helper-dependent adenoviral vectors for gene therapy. Hum. Gene Ther. 16 (1), 1–16. Pereboeva, L., Komarova, S., Mahasreshti, P.J., Curiel, D.T., 2004. Fiber-mosaic adenovirus as a novel approach to design genetically modified adenoviral vectors. Virus Res. 105 (1), 35–46. Pereboeva, L., Komarova, S., Roth, J., Ponnazhagan, S., Curiel, D.T., 2007. Targeting EGFR with metabolically biotinylated fiber-mosaic adenovirus. Gene Ther. 14 (8), 627–637. Rauen, K.A., Sudilovsky, D., Le, J.L., Chew, K.L., Hann, B., Weinberg, V., Schmitt, L.D., McCormick, F., 2002. Expression of the coxsackie adenovirus receptor in normal prostate and in primary and metastatic prostate carcinoma: potential relevance to gene therapy. Cancer Res. 62 (13), 3812–3818. Sachs, M.D., Rauen, K.A., Ramamurthy, M., Dodson, J.L., De Marzo, A.M., Putzi, M.J., Schoenberg, M.P., Rodriguez, R., 2002. Integrin alpha(v) and coxsackie adenovirus receptor expression in clinical bladder cancer. Urology 60 (3), 531–536. Schoggins, J.W., Gall, J.G., Falck-Pedersen, E., 2003. Subgroup B and F fiber chimeras eliminate normal adenovirus type 5 vector transduction in vitro and in vivo. J. Virol. 77 (2), 1039–1048. Shaner, N.C., Campbell, R.E., Steinbach, P.A., Giepmans, B.N., Palmer, A.E., Tsien, R.Y., 2004. Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nat. Biotechnol. 22 (12), 1567–1572.

M. Murakami et al. / Virology 407 (2010) 196–205 Sirena, D., Lilienfeld, B., Eisenhut, M., Kalin, S., Boucke, K., Beerli, R.R., Vogt, L., Ruedl, C., Bachmann, M.F., Greber, U.F., Hemmi, S., 2004. The human membrane cofactor CD46 is a receptor for species B adenovirus serotype 3. J. Virol. 78 (9), 4454–4462. Takayama, K., Reynolds, P.N., Short, J.J., Kawakami, Y., Adachi, Y., Glasgow, J.N., Rots, M. G., Krasnykh, V., Douglas, J.T., Curiel, D.T., 2003. A mosaic adenovirus possessing serotype Ad5 and serotype Ad3 knobs exhibits expanded tropism. Virology 309 (2), 282–293. Thomas, C.E., Ehrhardt, A., Kay, M.A., 2003. Progress and problems with the use of viral vectors for gene therapy. Nat. Rev. Genet. 4 (5), 346–358. Thorsteinsson, L., O'Dowd, G.M., Harrington, P.M., Johnson, P.M., 1998. The complement regulatory proteins CD46 and CD59, but not CD55, are highly expressed by glandular epithelium of human breast and colorectal tumour tissues. APMIS 106 (9), 869–878. Tsuruta, Y., Pereboeva, L., Glasgow, J.N., Luongo, C.L., Komarova, S., Kawakami, Y., Curiel, D.T., 2005. Reovirus sigma1 fiber incorporated into adenovirus serotype 5 enhances

205

infectivity via a CAR-independent pathway. Biochem. Biophys. Res. Commun. 335 (1), 205–214. Tsuruta, Y., Pereboeva, L., Glasgow, J.N., Rein, D.T., Kawakami, Y., Alvarez, R.D., Rocconi, R.P., Siegal, G.P., Dent, P., Fisher, P.B., Curiel, D.T., 2007. A mosaic fiber adenovirus serotype 5 vector containing reovirus sigma 1 and adenovirus serotype 3 knob fibers increases transduction in an ovarian cancer ex vivo system via a coxsackie and adenovirus receptor-independent pathway. Clin. Cancer Res. 13 (9), 2777–2783. Verma, I.M., Weitzman, M.D., 2005. Gene therapy: twenty-first century medicine. Annu. Rev. Biochem. 74, 711–738. Wivel, N.A., Gao, G., Wilson, J.M., 1999. Adenovirus vectors. In: Friedmann, T. (Ed.), The Development of Human Gene Therapy. Cold Spring Harbor Laboratory Press, New York, pp. 87–110. Zeimet, A.G., Muller-Holzner, E., Schuler, A., Hartung, G., Berger, J., Hermann, M., Widschwendter, M., Bergelson, J.M., Marth, C., 2002. Determination of molecules regulating gene delivery using adenoviral vectors in ovarian carcinomas. Gene Ther. 9 (16), 1093–1100.