key: cord-021532-6hmn90ac
authors: Von Seggern, Dan J.; Nemerow, Glen R.
title: ADENOVIRAL VECTORS FOR PROTEIN EXPRESSION
date: 2007-09-02
journal: Gene Expression Systems
DOI: 10.1016/b978-012253840-7/50006-7
sha: 
doc_id: 21532
cord_uid: 6hmn90ac

nan

Adenoviruses (Ad) are a large family of nonenveloped, doublestranded DNA viruses. At least 47 serotypes of human adenovirus have been isolated, as well as related animal viruses including avian, canine, and bovine types. In humans, the most common serotypes generally cause mild, self-limiting infections of the respiratory tract or gut (Horwitz, 1996) . Adenovirus has served as an important model system for studying many aspects of molecular biology, and a large body of knowledge has accumulated that allows it to be readily manipulated in the laboratory (for an extensive review of Ad biology, see Shenk, 1996) . Many features of this viral system, including the ability to infect a wide variety of nondividing cells, a large capacity for insertion of DNA, ease of production and stability of the viral particles, and the high viral titers that can be produced, make Ad-based vectors especially useful in gene transfer. Vectors based on several Ad serotypes have been used for gene expression work both in vitro and in vivo.

The adenoviral particle consists of an icosahedral protein capsid encasing a double-stranded DNA molecule of approximately 36 kb (see Fig. 1 ). The most abundant viral protein is hexon, which makes up most of the outer shell of the virus (van Oostrum and Burnett, 1985) . At each vertex is a complex composed of the penton base and fiber proteins, both of which interact with cellular receptors during the process of virus infection (Philipson et al., 1968; Bergelson et al., 1997; Tomko et al., 1997; Wickham et al., 1993) . Several minor viral proteins also contribute to the capsid and may play roles in capsid assembly and in viral chromosome packaging. Ad DNA is packaged in a complex with several viral proteins (Mirza and Weber, 1982) , and each end of the chromosome is covalently attached to a single molecule of terminal protein (TP) (Rekosh et Figure 1 Idealized cross section of the adenovirus particle. Hexon protein makes up the majority of the capsid by mass. A complex of penton base and fiber proteins is located at each vertex. The 36-kb double-stranded DNA chromosome is packaged as a complex with protein VII, and one molecule of terminal protein is attached covalently at each end. For clarity, a number of minor proteins have been omitted from the diagram. al., 1977) . TP acts as a primer for DNA synthesis (Van der Vliet, 1995) and also serves to anchor the viral chromosome to the nuclear matrix (Schaack et al., 1990) .

Most gene transfer and expression experiments have been done with adenovirus serotypes 2 and 5. The entire genomic sequences of these viruses are known (Chroboczek et al., 1992; Roberts et al., 1984) , and many years of work have been spent elucidating a welldefined program of viral gene expression and defining functions for most of the viral reading frames (Shenk, 1996) . Historically, the Ad chromosome has been divided into 100 map units (1 m.u. is roughly equivalent to 360 bp). The positions of the major transcriptional units in the virus, which can be broadly divided into two classes, are shown in Fig. 2 . Early genes, encoding mainly regulatory functions, are those transcribed shortly after infection. Late genes, including those encoding the major structural proteins, are highly expressed only after viral DNA replication begins. Both early and late Ad promoters have also been used to express heterologous genes.

Shortly after infection, transcription from viral early promoters begins. The most important regulator of viral growth is the immediate-early E la protein, a promiscuous transcriptional activator that interacts with many cellular transcription factors (reviewed by Jones, 1995) . Ela both activates further viral gene expression and modifies host cell regulation to drive the cell into S phase, which ensures that sufficient precursors for viral DNA and protein will be available. Deletion of the E la gene blocks most viral gene expression as well as DNA synthesis and therefore renders Ad replication defective. Infection by El-deleted Ad vectors often does not cause cytotoxic effects or grossly perturb the underlying cell biology; indeed, such vectors have been approved for administration to human subjects in many gene therapy protocols. These vectors are therefore useful in studies of protein function in intact cells, and most recombinant Ads now in use are E1 deleted.

E2 transcripts are spliced to produce several proteins that are involved in viral DNA synthesis. These include the viral DNA polymerase and terminal protein precursor encoded by E2b, and the The inverted terminal repeats (ITR) at each end of the chromosome are represented by black triangles, and the location of the essential packaging sequences is indicated by ~. The E1 and E3 regions can be substituted with foreign DNA as shown to generate defective and nondefective vectors, respectively. Nondefective vectors can also be produced by insertions between the right ITR and the E4 promoter. One map unit (m.u.) is equal to approximately 360 bp.

single-stranded DNA-binding protein produced from E2a. For a review of the roles of E2 products in DNA replication, see Van der Vliet (1995) . E4 contains several open reading frames (ORFs)encoding proteins that regulate processes including viral transcription, the shut off of host cell macromolecular synthesis, and correct synthesis of unit length viral chromosomes (Falgout and Ketner, 1987; Weiden and Ginsberg, 1994; Halbert et al., 1985) .

Unlike other early regions, E3 products are not required for the replication of virus in cultured cells (Ginsberg et al., 1989) . Instead, these proteins are involved in evading the host antiviral immune response. E3 encodes a 19-kDa glycoprotein that downregulates the transport of major histocompatibility complex I molecules to the surface of an infected cell and reduces the inflammatory response generated against the vector (Ginsberg et al., 1989) . Other E3 products block tumor necrosis factor-induced lysis or apoptosis of infected cells (see Wold, 1993; Tollefson, 1998) . Because the E30RF can be deleted without affecting viability, a foreign gene can be substituted into this region and expressed under control of the E3 promoter.

Late Events in Infection Viral DNA replication is initiated by about 8 hr postinfection and is followed by the activation of late viral gene expression. Genes encoding the major structural proteins, such as penton, hexon, and fiber, are all transcribed from the major late promoter (MLP), which initiates a long primary transcript spanning most of the virus genome. After the onset of viral DNA replication, the MLP is transcribed at very high levels and its transcripts are efficiently spliced into mRNAs encoding the various structural proteins (Shenk, 1996) .

Three small exons totaling 200 nucleotides (the tripartite leader, TPL) are spliced onto the 5' ends of all the late mRNAs (Chow et al., 1997; Akusj/irvi and Pettersson, 1979; Zain et al., 1979) . Late in infection, host cell protein synthesis is shut off due to inhibition of eIF-4E, a cellular translation initiation factor (Zhang et al., 1994) . The presence of the TPL confers eIF-4E independence and ensures that the viral mRNAs are translated preferentially (Dolph et al., 1988 (Dolph et al., , 1990 . This allows viral proteins to accumulate to very high levels late in infection. These viral transcriptional and translational elements have also been used in recombinant Ad to express foreign proteins at high levels and are discussed further below.

As the late gene products accumulate, virus particles are assembled in the nucleus and released by lysis of the cell, which may in-volve virus-induced apoptosis (Tollefson et al., 1996) . An infected cell can produce 1000 or more progeny virus, allowing large numbers of virions to be isolated.

Recombinant adenoviruses have been used to express foreign proteins for a number of years, and the recent explosion of interest in gene therapy has led to the development both of improved adenoviral vectors and of more efficient techniques for generating them. Adenovirus-mediated gene therapy is not the focus of this work and has previously been reviewed (Haddada et al., 1995; Kozarsky and Wilson, 1993; Ali et al., 1994) . However, technology developed for gene therapy is, in many cases, applicable to the problem of protein expression in general, and improved vectors are likely to make the adenovirus system even more versatile.

Provided that the total length of the recombinant DNA molecule does not exceed about 105% of wild type (Bett et al., 1993) , DNA can be inserted at several sites in the adenovirus chromosome to create a nondefective recombinant (see Fig. 2 ). Such vectors have been used to express foreign proteins in many studies, particularly for vaccine development studies where the ability of the virus to replicate postadministration may be advantageous. In addition to the commonly used serotypes 2 and 5, adenoviruses of other subgroups such as Ad4 and Ad7 have been used for protein expression (Imler, 1995) . A property of nondefective vectors that may be useful in some situations is their ability to shut off host cell macromolecular synthesis. This effectively enriches the expression of a recombinant protein regulated by late viral transcriptional and translational elements.

In many studies, the nonessential E3 region has been replaced by a foreign sequence of interest (Schneider et al., 1989; Morin et al., 1987; Dewar et al., 1989; Johnson et al., 1988) , and the plasmid-based systems now available for construction of recombinant viruses (see later)contain unique restriction sites to simplify cloning into the E3 locus. However, it may be advantageous to retain these sequences in viruses that are to be used for experiments in vivo because of the role of E3 in reducing the antiviral immune response (Wold, 1993; Ginsberg et al., 1989; Tollefson, 1998) . Helper-independent viruses can also be created by inserting DNA between the E4 promoter and the inverted terminal repeat at the right end of the chromosome (Saito et al., 1985; Mason et al., 1990) . However, unless another viral sequence (such as E3)is deleted, the packaging size constraint limits such insertions to about 1.8 kb.

A replication-defective adenovirus is produced by the removal of essential sequences (such as E la or E4)from the viral chromosome. As it is possible for cells to be infected simultaneously with two different adenoviruses, defective mutants can be grown by coinfection with a nondefective helper that trans complements the missing function. Pure stocks of defective vectors can also be propagated by growth in cell lines that provide the missing viral function(s). The first such cell line, 293, was constructed by the transformation of human embryonic kidney cells with sheared Ad5 DNA and contains an integrated copy of the E1 region (Graham et al., 1977) . This allows growth of viral mutants in which the E 1 a region has been replaced by foreign sequences. Complementation by 293 cells is very efficient and allows growth of El-deleted adenoviruses to near wild-type levels. A subline of 293, which was adapted for growth in suspension culture, has been described (Graham, 1987) . This should be of use in producing large quantities of recombinant protein in high-density cultures.

Regions of the viral chromosome required for replication and packaging of DNA have been identified and make up only a small part of the chromosome. The ends of the Ad chromosome are composed of 105-bp inverted terminal repeat (ITR) sequence. Binding sites for a number of proteins within the ITR are necessary and sufficient for the initiation of viral DNA synthesis (Bernstein et al., 1986; Hay, 1985) . A 164 nucleotide sequence adjacent to the left ITR is also required for the packaging of viral DNA into particles (Hearing et al., 1987) . In principle, these are the only elements required in cis, and the rest of the 36-kb chromosome could be replaced with foreign DNA.

Defective vectors lacking essentially all of the Ad genome have been described (Haecker et al., 1996; Parks et al., 1996; Kumar-Singh and Chamberlain, 1997; Kochanek et al., 1997; . These vectors are propagated by coinfection with a helper virus and have very high capacities to accept DNA inserts. In one example of this "empty" vector strategy, the Cre/Lox recombinase system was used to remove the packaging signals from the helper virus chromosome during vector growth, rendering it unpackageable (Parks et al., 1996) . However, so far there are no such vectors that are truly helper independent, and it is currently impossible to remove all the helper virus from a preparation of vector. This may be a disadvantage, depending on the intended application of the vector. As helper viruses themselves are usually replication defective, their presence at low levels may not be an issue for in vitro experiments. It also appears that the minimum allowable size of DNA for efficient packaging is about 27 kb, as shorter chromosomes are not propagated stably (Parks and Graham, 1997) .

Because E 1 a products are essential for progression through the viral life cycle, El-deleted vectors are almost completely defective for growth in most cells and can generally be expected to have only minimal effects on host cell function. This has made them valuable tools for delivering genes in cell biology studies. However, the E 1 deletion may not completely block the expression of viral genes following administration in vivo. This leads to an antivector immune response and contributes to the short-term gene expression seen in most gene therapy experiments to date (Yang et al., 1994 (Yang et al., , 1995 . This has prompted the development of vector systems deleted for additional viral genes such as E4 or the E2a DNA-binding protein, which are expected to be less immunogenic and safer in vivo (Wang et al., 1995; Brough et al., 1996; Yeh et al., 1996; Krougliak and Graham, 1995; Zhou et al., 1996; Brough et al., 1996; Gorziglia et al., 1996) . These multiply deleted vector systems also have higher capacities for insertion of a gene of interest and should be more versatile transfer vectors for protein expression.

Adenovirus DNA is infectious, in that transfection of viral DNA alone into an appropriate cell line results in its replication, viral protein synthesis, and production of complete viral particles. This process is much more efficient if the DNA-TP complex is isolated intact from virions, probably due to the role of TP in priming viral DNA synthesis (Sharp et al., 1976) . A number of methods for generating recombinant virus by the manipulation of viral DNA and retransfection have been developed (for an excellent recent review of recombinant Ad construction techniques, including protocols, see Graham and Prevec, 1995) . The large size of the adenovirus chromosome means that it contains few convenient unique restriction sites for insertion of DNA. Most protocols for the generation of recombinant viruses involve manipulation of the vector in two or more fragments, followed either by ligation in vitro or by cotransfection and recombination in host cells to regenerate a full-length chromosome. Use of the most common methods to construct an Ad recombinant with a cDNA inserted in place of the E la gene is illustrated in Fig. 3 .

Recombinant adenoviruses were originally created by isolating fragments of the viral chromosome, followed by ligation or recombination to regenerate a complete viral backbone. Generally, DNA isolated from virions is digested with a restriction enzyme that removes the left end of the chromosome, including the viral packaging signals. This renders the large DNA fragment noninfectious. The unique ClaI site located at 2.6 m.u. is commonly used for this purpose (Fig. 3A) . Undigested viral DNA and the smaller DNA fragments are often removed by sucrose gradient ultracentrifugation. The large fragment is ligated in vitro to a DNA fragment containing the left end of the chromosome and containing the desired modifications (Stow, 1981) . If the viral DNA contained in this fragment has a deletion of the E la region, the resulting recombinant Ad will be a replication-defective E1 mutant. Inclusion of unique restriction sites in the left end-containing plasmid (generally in the place of the E la gene) simplifies insertion of a cDNA. The ligation products are then transfected into host cells to regenerate infectious virus (Fig. 3A) .

In a variation of this method (Fig. 3B ), the purified large fragment is transfected into host cells along with a plasmid that contains overlapping viral DNA, including the left end (Gluzman et al., 1982; Graham and Prevec, 1995; Haj-Ahmad and Graham, 1986) . An infectious Ad chromosome can then be generated by homologous recombination between the viral fragment and the plasmid sequences, resulting in the production of infectious virus. Similar methods have been used to create viral genomes with modifications or insertions near the right end of the chromosome (e.g., insertions into the E3 region).

These methods of generating recombinant Ads may produce a background of wild-type virus due to contamination with undigested viral DNA or religation of the wild-type fragments. This increases the number of plaques that must be isolated and screened to identify the desired recombinant. Imler et al. (1995) found that the background can be reduced if the donor virus from which the large restriction fragment is derived contains a copy of the HSV thymidine kinase gene inserted into the E1 region. This parental virus cannot form plaques in the presence of gancyclovir. Growth of any residual donor virus is therefore suppressed by the drug selection, and isolation of the desired recombinant is simplified. Schaack et al. (1995a) have described a donor virus containing a f~-galactosidase gene near the left end of the genome. Plaques resulting from undigested parental DNA or religation of the two fragments in the host cells will express f~-galactosidase and can be identified by incorporation of X-Gal (a chromogenic ~-galactosidase substrate) into the agarose overlay. The cDNA of interest is inserted into one of the unique cloning sites in the plasmid pAElspla (which contains Ad DNA corresponding to 0-16 m.u. with an E1 deletion), and the resulting construct is cotransfected with a plasmid such as pBHG10, which contains an unpackageable Ad genome. A complete, packageable Ad chromosome is generated only after recombination takes place between the two DNAs. (D) The cosmid-terminal protein system, pAdexl is a cosmid containing most (0-99.3 m.u.) of the Ad5 genome and is modified to contain a unique cloning site in place of the Ela gene. Following insertion of the desired cDNA into pAdexl, it is transfected into cells along with an Ad5 DNA-terminal protein complex that has been digested by a restriction enzyme that cuts at several sites in the genome. Recombination between pAdexl and the terminal fragments of the Ad5 DNA generates a full-length chromosome with a molecule of terminal protein covalently attached at each end, which is efficiently packaged as virus.

An adenovirus chromosome, ligated into a circular Escherichia coli plasmid vector, can be excised from the plasmid and propagated as virus after transfection into an appropriate host cell line (Graham, 1984; Hanahan and Gluzman, 1984) . Berkner and Sharp (1983) also demonstrated that two or more plasmids containing overlapping adenoviral DNA fragments could undergo recombination in cells to regenerate a complete viral chromosome.

Taking advantage of these findings, a two-plasmid system ( Fig.  3C ) has been developed to simplify the construction of recombinant viral genomes (McGrory et al., 1988; Bett et al., 1994 ). An essentially complete but unpackageable viral genome is cloned into one plasmid, such as pJM17 or pBHG10. A construct such as pJM17 is noninfectious due to the insertion of plasmid sequences into the viral region, rendering it too large to package into viral particles (McGrory et al., 1988) . Deletion of the viral packaging sequences from plasmids such as pBHG10, as shown in Fig. 3C , prevents replication of the nonrecombinant plasmid (Bett et al., 1994) . A second plasmid contains the left end of the viral chromosome, including an intact packaging sequence. The E la region is generally deleted from this second plasmid (pAE1SplA in Fig. 3C ) and replaced by a linker with one or more unique restriction sites. The presence of unique cloning sites, and the smaller size of this second plasmid, greatly simplifies the process of inserting a gene of interest into the viral backbone. Another unique restriction site has been engineered into the E3 deletion of pBHG10, which allows insertion of foreign sequence here as well. After cloning of the desired sequences into either or both plasmids, they are cotransfected into a packaging cell line such as 293. Only after recombination takes place between the two plasmids will a packageable viral chromosome be generated.

This method allows propagation of all viral DNA segments in E. coli, which is technically much simpler than isolating and purifying fragments from virion DNA. The background of nonrecombinant virus produced following plasmid transfection can be very low. Using the pJM17 system, background plaques arising from rearrangement of the oversized vector plasmid have been reported (McGrory et al., 1988; Graham and Prevec, 1995) . Using a pBHG10type plasmid (which lacks packaging signals)reduces the background of nonrecombinant plaques to essentially zero.

As noted earlier, the Ad DNA-TP complex is much more infectious on a per microgram basis than plasmid DNA or proteinasetreated viral DNA alone. A method developed by Saito and coworkers (Fig. 3D ) takes advantage of this to simplify the recovery of recombinant viruses (Miyake et al., 1996) . Most of the Ad genome is contained in the cosmid pAdexl, which has been modified to contain a unique restriction site for the insertion of a foreign sequence in place of the viral E1 region. Cosmid DNA is transfected into host cells along with an adenovirus DNA-TP complex that has been digested by a restriction enzyme that cuts the viral backbone many times. Fragmentation of the viral backbone reduces background due to the carryover of donor virus or to religation of its fragments. Homologous recombination between the pAdex cosmid and the terminal fragments of the donor viral DNA regenerates a full-length chromosome that has a molecule of TP covalently attached at each end. These workers report that recovery of the desired recombinant from this system is very efficient and much simpler than using purified viral DNA alone.

Two methods have been described for generating and modifying full-length adenoviral chromosomes in lower organisms. This strategy has the theoretical advantage that regeneration of the desired Ad chromosome does not depend on any viral functions. This might simplify the recovery of poorly growing Ad mutants, which may be difficult to recover using the recombination strategies outlined earlier. Ketner et al., (1994) have succeeded in constructing a yeast artificial chromosome (YAC) containing the entire Ad2 genome. The Ad chromosome can be excised from the YAC by restriction digestion and transfected to mammalian cells. This system allows any desired alterations of the viral chromosome to be performed using the very powerful homologous recombination techniques available in the yeast system, without the need to isolate partial adenovirus clones. Methods for virus construction using recombination in E. coli have recently been described (Chartier et al., 1996; He et al., 1998) . As in the yeast system, DNA isolated from bacteria after recombination is infectious and can be used to produce virus by transfection of mammalian cells. These vector systems have not yet come into widespread use, however.

The presence of recombinant Ad following transfection of cells by appropriate DNA is readily detectable by the presence of plaques or cytopathic effect (CPE). Regardless of the method used for generating a recombinant Ad, the recovered virus should be plaque purified before use. Nondefective virus will form plaques on monolayers of essentially any cell type that is permissive for viral growth, whereas defective vectors must be grown on a complementing cell line (such as 293 for El-deleted vectors). If the cell monolayers are overlaid with culture medium containing agarose following transfection, the resulting plaques can be picked directly for further purification (for Ad plaquing and propagation protocols, see Graham and Prevec, 1995) .

Alternatively, transfected cells can be maintained under liquid medium and the wells or plates observed for the spreading cell death characteristic of an active viral infection (Zhang et al., 1993) . The virus present in a crude extract from the infected plate is then purified by plaquing. It is sometimes easier to detect the production of virus using this method, as detection of the relatively few plaques that result from transfection is dependent on the cell monolayer maintaining good morphology throughout the experiment. This is especially important if the recombinant virus grows slowly.

A major advantage of the adenovirus system is the ease with which large quantities of high-titer viral stocks can be produced. Nondefective vectors can be propagated in a number of cell lines, most commonly epithelial cells such as HeLa. Defective vectors are either grown as mixed stocks, including a helper virus, or as pure stocks using complementing cell lines. For propagation of Eldeleted Ads, the standard has been the human embryonic kidneyderived cell line 293 (Graham et al., 1977) . Additional El-complementing lines based on retinal epithelial cells or A549 lung carcinoma cells have been reported (Gorziglia et al., 1996; Fallaux et al., 1996) . The newer generation of vectors deleted for viral genes such as E2, E4, or fiber are grown in packaging lines expressing the corresponding viral proteins (Yeh et al., 1996; Gorziglia et al., 1996; Wang et al., 1995; Brough et al., 1996; Zhou et al., 1996; Langer and Schaack, 1996; Amalfitano et al., 1996; Caravokyri and Leppard, 1995; Schaack et al., 1995b; Weinberg and Ketner, 1983; Krougliak and Graham, 1995; Von Seggern et al., 1998) . In most cases, hightiter stocks of these viruses can be prepared readily.

Growth of Ad vectors requires no special facilities beyond those normally used for mammalian tissue culture (NIH Biosafety Level 2). The majority of virions produced remain associated with the cell, allowing concentration by pelleting infected cells prior to spontaneous lysis. A simple freeze-thaw lysis of infected cells is then performed, followed by banding of virus on CsC1 gradients (Graham and Prevec, 1995; Everitt et al., 1977) . For some uses, such as infection of cells in vitro to produce a stock of recombinant protein, CsC1 purification may be unnecessary. Purified virus is dialyzed into buffer containing 10% glycerol and is quite stable for long-term storage at-70 ~ (Graham and Prevec, 1995) . Storing the purified virus in 10 mM Tris, pH 8.1, 0.9% NaC1, and 10% glycerol has been found to improve stability. Titers of a standard El-deleted adenovirus in excess of 1011 plaque-forming units (PFU)/ml can be obtained readily by this method. Chromatographic methods for virus isolation have been developed that may further simplify production and scaleup (Huyghe et al., 1995) .

Virus concentration has been determined by a number of methods. For a detailed study of several methods for Ad quantitation and of factors that affect results, see Mittereder et al. (1996) . The number of virus particles in solution is often determined by absorbance at 280 nm or by a standard protein assay (1 ~g of Ad2 proteins corresponds to 4 x 10 9 viral particles). An HPLC-based method for quantifying Ad particles has also been described (Shabram et al., 1997) . However, measuring the virus particle number does not provide information about the biological activity of a preparation, and infectious titers are generally reported in PFU. Plaque assays are straightforward to carry out, but require from 10 to 14 days from start to finish, depending on the rate of growth of an individual virus. A more rapid assay is the measurement of fluorescent focus units (FFU) (Thiel and Smith, 1967) . Results of the FFU assay are available 1 to 2 days after infection of cells. Because of their high level of synthesis, viral structural proteins are good targets for immunostaining of infected cells. The authors have used polyclonal antibodies raised against either penton base or fiber to measure FFU.

The steps involved in producing a recombinant Ad vector with the standard two-plasmid system ( Fig. 3C) are (1) cloning a foreign gene or genes into the appropriate Ad plasmid(s), (2) transfection into host cells followed by recombination, and (3)purification of the resulting virus. As shown in Fig. 4 , a standard first-generation vector (e.g., insertion of a new expression cassette into the E1 region) can be generated in approximately 7-10 weeks. Much of the time involved is spent waiting for plaques or CPE to appear in the transfected cells and in the successive rounds of plaque purification, which do not require much "hands-on" work.

Proteins successfully produced from recombinant Ad (see below) have included heterologous viral proteins, intracellular enzymes, cell surface proteins, and secreted proteins such as cytokines and Table 1 . Schaffhausen et al. (1987) purified polyomavirus middle T antigen from cells infected with their Ad5 vector and reported yields of up to 100 ~g protein/100-mm dish of infected 293 cells. Zhao et al. (1993) inserted the gene encoding a protein tyrosine phosphatase (PTP1C), driven by the MLP and TPL sequences, into the E1 region of an Ad5 vector. They purified the recombinant protein from infected 293 cells and reported isolating more than 100 ~g of purified enzyme per 150-mm dish of cells (Zhao et al., 1993) . Garnier et al. (1994) used the same recombinant virus to explore scaleup of protein production using a 293 suspension cell line. They were able to produce the PTP1C protein at levels of 15% of the total cellular protein content (equivalent to 90 mg protein/liter of culture) by growing cells at high density (2 x 10 6 cells/ml). This type of strategy should be generally applicable for the large-scale production of other proteins.

An important application for recombinant adenoviruses has been the expression of heterologous viral proteins (Berkner, 1992) , either for structure/function studies or for their possible use as vaccines. A discussion of Ad-based vaccines is outside the scope of this chapter, and this topic has previously been reviewed (Imler, 1995) . However, this work illustrates some of the myriad potential strategies for Ad-mediated protein expression.

Hepatitis B virus surface antigen (HBsAg) has been produced from Ad-infected cells using a wide variety of strategies. In one case the HBsAg gene, driven by the MLP either with or without inclusion of TPL sequences, was inserted in place of the Ad5 E1 region (Davis et al., 1985) , and addition of the TPL resulted in approximately a 70-fold increase in protein synthesis. As expected for expression driven by late viral regulatory elements, the appearance of HBsAg in the medium was observed at late times (10-20 hr postinfection). HBsAg has also been placed into the E1 region downstream of the Ad5 Ela promoter (Ballay et al., 1985) and inserted into the E3 region of a nondefective Ad5 (Morin et al., 1987) . In the latter case, the hepatitis B gene was inserted without a promoter, and its expression appeared to be driven by both early (probably E3) and late (possibly the MLP) Ad promoters. Mason et al. (1990) inserted a cassette consisting of the Ad4 MLP, TPL, and HBsAg coding sequences between the right ITR and the E4 promoter of a nondefective Ad4. In addition to the TPL exons, their constructs retained varying amounts of the intronic sequence located between the first and the second TPL segments. As had been noted previously for the analogous Ad5 intron (Mansour et al., 1986) , these workers reported that sequences within this intron could dramatically upregulate mRNA levels driven by the MLP. HBsAg levels produced by infected A549 cells reached levels of almost 4 Ixg/10 6 cells in these experiments. Saito et al. (1985) also created a nondefective vector by inserting the HBsAg gene (this time driven by the HBV promoter) between E4 and the ITR of Ad5. This construct lacked a TPL and although HBV mRNA was readily detectable, HBsAg was expressed only at very low levels (less than 1 ng/10 6 cells), emphasizing the importance of translational regulation after infection with nondefective Ad vectors.

Finally, the ability of Ad vectors to accommodate multiple DNA insertions has been used to coexpress HBsAg along with other proteins. A nondefective Ad7 vector was constructed with MLP, TPL (including the first intron), and HBsAg sequences inserted between the E4 and right ITR, and either the HBV core or precore protein gene inserted in place of the E3 region and driven by the E3 promoter (Ye et al., 1991) . HBsAg was produced by these viruses with the late timing expected for MLP transcription and reached levels of 6 Ixg/10 6 cells in the medium. The E3-driven core and precore antigens were expressed by 12 hr after infection (as expected for Ad early genes) and were reported to reach levels of 1.46 and 0.33 ~g/106 cells, respectively.

Fooks and co-workers (1995)used the CMV promoter to direct expression of the measles nucleocapsid (N) protein from an Eldeleted Ad5 vector. By treating infected cells with forskolin (to increase CMV promoter transcription), they were able to produce recombinant N protein at levels representing 20% of total soluble cellular protein (Fooks et al., 1995) . These workers compared the adenovirus-produced protein to recombinant N protein produced using a baculovirus system or expressed in E. coli (Warnes et al., 1994) . The approximate level of expression of the baculovirus-and adenovirus-expressed N proteins was found to be similar and considerably higher than in the E. coli system (the yields reported in this case were 40, 25, and 3 % of total cellular protein, respectively).

The env gene of HIV-1 has been expressed using nondefective Ad5 (Dewar et al., 1989) and Ad7 (Chanda et al., 1990) vectors, with proper cleavage of the gp 160 protein into the gp 120 and gp41 fragments and syncitia formation by infected cells. Dewar et al. (1989) found that with their Ad5-based vector, gpl60 was produced in HeLa suspension cultures at approximately 1 mg/liter. Chanda et al. (1990) reported production of very high levels (2-5 mg/liter of culture) of gp160 in A549 cells infected with their Ad7 vector if the HIV-1 rev gene product was provided, either by coinfection with a rev-expressing Ad vector or from a rev gene fragment expressed from the Ad E3 promoter. A genomic fragment of SIV containing the overlapping env and rev genes was inserted into the E3 region of Ad5 (Cheng et al., 1992) , and proper splicing of the mRNA to produce both SIV proteins was observed. The presence of the rev gene upregulated env protein production in this context as well, indicating that a complex foreign transcript incorporated into an Ad vector can be spliced into two different functional mRNAs that produce appropriately interacting proteins.

The emphasis on potential therapeutic uses of adenovirus has led a number of investigators to insert cytokine or growth factor genes into Ad vectors. Biologically active interleukin (IL)-6 was expressed using a recombinant Ad that also contained the HBsAg gene (Lindley et al., 1994) . A CMV-driven human IL-2 gene was inserted into the E1 region of an Ad5 vector, and expression was evaluated in several human cell lines (Addison et al., 1995) . Secretion of IL-2 was measured at 25-110 ng/106 cells over a period of several days. The recombinant protein was active, as measured by the ability to stimulate growth of an IL-2-dependent cell line. Smith and coworkers (1996) inserted the genes for secretable ciliary neurotrophic factor (sCNTF) or for neurotrophin-3 into replication-defective Ad and reported that both viruses allowed secretion of their respective proteins into the medium at concentrations sufficient to promote neuronal survival in vitro. Other growth factors that have been expressed using Ad vectors include fibroblast growth factor-4 (HST-1/FGF-4) and vascular endothelial growth factor (VEGF) (Konishi et al., 1996; M~ihlhauser et al., 1995) .

Expression of heterodimeric proteins such as IL-12 is complicated by the need to deliver the genes encoding both subunits to the same cells in approximately equal dosages. Here, the ability to insert more than one foreign DNA segment into the adenovirus chromosome is valuable. Bramson et al. (1996) inserted a gene encoding the p35 subunit of murine IL-12 into the E1 region and that encoding the p40 subunit into E3, both driven by the strong CMV promoter. This strategy ensured that all infected cells received both genes at the same copy number. Infection of 293 cells or the human fibroblast line MRC5 with the IL-12-bearing Ad vector resulted in the secretion of biologically active IL-12 into the medium at levels up to 40,000 units/106 cells (Bramson et al., 1996) .

A major advantage of the adenovirus system is that highly efficient infection of a population of cells can be achieved in a relatively synchronous way. A protein of interest can be introduced readily into most or all cells in an experimental setting, and infection of primary cultures, biopsy samples, or cell lines that are difficult to transfect using standard methods has been reported. By using a replication-defective Ad vector, it is often possible to infect cells without grossly perturbing normal cell functions, thus allowing analysis of the functional properties of the recombinant protein. Askanas et al. (1996) used an Ad vector to express the f~-amyloid precursor protein (~-APP) in primary cells derived from biopsy specimens of normal human muscle in order to determine whether ~-APP could cause degenerative changes seen in inclusion body myositisis, a degenerative muscle disease. By using the strong CMV immediate-early promoter, this group was able to express high levels of the recombinant ~-APP in essentially all cells and to demonstrate that its overexpression caused morphological changes similar to those seen in the disease state. Huang et al. (1997) recently investigated the role of different MAP kinase isoforms in Fas-induced apoptosis by expressing several different MAP kinase molecules, as well as dominant negative, constitutively active, and chimeric forms of these proteins, in the Jurkat T-cell line. This line is refractory to transfection by standard methods, and the ability of Ad to synchronously transfect essentially all of the cells was instrumental in defining the role of a specific MAP kinase molecule in the apoptosis pathway.

In a study of glucose metabolism, an Ad vector allowed expression of hexokinase 1 in the majority of cells of isolated (but intact) rat islets of Langerhans (Becker et al., 1994) . Here, the Ad system allowed transfection of a much higher percentage of cells (60-70%) than could be transfected by nonviral methods. These investigators were able to gather information about hexokinase overexpression without the use of transgenic animals.

In endothelial cells, Ad-mediated expression of a nuclear-localized form of the NF-KB inhibitor, IK-B~, inhibited NF-KB activity and blocked expression of several NF-KB-dependent markers of cell activation in response to lipopolysaccharide. NF-KB-independent markers such as secretion of von Willebrand's factor or of prostacyclin were unaffected (Wrighton et al., 1996) . In another study, Ad vectors were used to investigate signal transduction by the type II transforming growth factor-~ (TGF-f~) receptor (Yamamoto et al., 1996) . Expression of a truncated receptor in cultured cells completely blocked both the antiproliferative and the transcriptional activation effects of TGF-f~ treatment, without affecting the response to two other growth factors.

The ability of more than one adenovirus to infect the same cell has also been used to coexpress multiple protein subunits. Herpes simplex virus 1 encodes a heterodimeric immunoglobulin receptor that is expressed on the surface of an infected cell. After cloning the genes for the two subunits into separate Ad vectors, Hanke et al. (1990) expressed either or both subunits simply by infecting target cells with either or both recombinant Ads. This allowed determination of the IgG-binding properties of the individual subunits as well as of the heterodimeric complex on the surface of the infected cells.

Ad vectors can efficiently transfect a large number of organs or tissues in vivo, including liver , kidney (Moullier et al., 1994) , skin (Setoguchi et al., 1994b) , brain (Akli et al., 1993; Doran et al., 1997) , skeletal muscle (Ragot et al., 1993) , heart , lung epithelium (Rosenfeld et al., 1991) , and several ocular tissues (Mashhour et al., 1994) (for further review, see Ali et al., 1994, or Kozarsky and Wilson, 1993) . This property allows protein function to be investigated in vivo without generating transgenic animals. Adenoviral gene delivery is particularly useful if a large number of variant proteins are to be evaluated. For example, to identify the domains of the closely related lipoprotein lipase and hepatic lipase proteins responsible for determining their substrate specificity in vivo, Kobayashi et al. (1996) made several chimeric proteins by exchanging domains between the two and cloned the resulting genes into El-deleted Ad vectors. They systemically administered these viruses to hepatic lipase-deficient mice and examined changes in serum lipids. These experiments showed that the greater ability of hepatic lipase to hydrolyze phospholipids and to reduce total serum cholesterol is largely determined by the "lid" domain of the proteins, which is thought to restrict access to the active site of the enzyme.

Cytokine and chemokine functions in vivo have been studied using Ad vectors to locally produce factors, including GM-CSF, RANTES, and IL-6 in rat lung (Xing et al., 1994 Braciak et al., 1996) . Expression of these molecules in a defined tissue has allowed investigation of their roles on inflammatory processes and recruitment of cells such as monocytes and T lymphocytes. Local expression of Ad-delivered genes may be more relevant to their normal physiological roles than systemic administration of the purified protein.

Systemically administered Ad vectors can also provide high-level temporary expression of a protein that is secreted into the bloodstream. Much of this work has focused on preclinical applications, as in expression of factor VIII for the treatment of hemophilia A (Connelly et al., 1996) or of erythropoietin to stimulate erythropoiesis (Setoguchi et al., 1994a) , but this property can also be useful in understanding the in vivo roles of secreted proteins. For example, rats treated intravenously with a viral vector directing expression of the rat leptin protein exhibited reduced food intake and weight gain, as well as a disappearance of fat deposits . Intraperitoneal administration of an Ad vector containing the gene encoding HST-1/FGF-4 to mice stimulated platelet production (Sakamoto et al., 1994) . This group then used the same viral vector in an in vitro study to demonstrate that the increased platelet count was due to FGF-4 stimulation of megakaryocyte maturation (Konishi et al., 1996) .

As noted previously, the large size of the Ad chromosome should allow for the transfer of very large genes. Vectors containing only the minimum Ad packaging sequences have been used to express full-length human (Haecker et al., 1996) or murine (Kochanek et al., 1997; Kumar-Singh and Chamberlain, 1997) dystrophin in cultured myotubes and in vivo. The 14-kb dystrophin cDNA far exceeds the capacity of other viral vectors now in widespread use.

Several groups have constructed Ad vectors that express the Cre recombinase protein of bacteriophage P1. This enzyme catalyzes recombination between 34-bp target sites (LoxP)with excision of the intervening DNA. Expression of Cre from an Ad recombinant could remove an inactivating sequence flanked by LoxP sites and switch on expression of a reporter gene either located on a second Ad coinfected along with the Ad-Cre recombinant or stably integrated into the host cell chromosome (Anton and Graham, 1995; Miyake et al., 1996; Sakai et al., 1995) . Wang et al. (1996) extended these findings in vivo and demonstrated that the Ad-Cre system was capable of causing efficient recombination between LoxP sites on the chromosomes of transgenic mice. This approach should be useful in conditionally expressing proteins or in producing gene knockouts that are spatially and/or temporally restricted.

Another application for Ad-mediated gene delivery is modulation of the receptor repertoire expressed on the surface of infected cells. Lieber et al. (1995) used an Ad vector to deliver the amphotrophic retrovirus receptor (RAM) cDNA to cells that are normally resistant to infection with the amphotropic retroviral vectors commonly used in gene therapy work. Following Ad infection, target cells were markedly more infectible with a recombinant retrovirus. An Ad recombinant was used to express human CD4 on the surface of a number of CD4-negative cell lines (Yasukawa et al., 1997) . In several of these lines, CD4 expression allowed infection by the herpesvirus HHV-7, confirming the role of CD4 in HHV-7 entry.

As replication-defective Ad vectors can be introduced to a relatively small area of tissue by local injection or application, they have utility as markers for the study of cell migration and development. When LacZ-containing Ad vectors were administered to the external surface of early stage chicken hearts, ~-galactosidase-expressing cells in some, but not all, regions of the heart were detectable at later times (Fisher and Watanabe, 1996) , providing infor-mation about the origin of the cells making up various heart structures. Similar experiments have demonstrated that Ad can be used to mark cells in the developing rat central nervous system and in preimplantation mouse embryos (Lisovoski et al., 1994; Tsukui et al., 1995) .

Insertion of foreign DNA into either or both of the E1 and E3 regions is compatible with high-titer virus production. The ability to insert two genes into the same vector chromosome has obvious advantages if a heterodimeric protein (Bramson et al., 1996) or two proteins designed to work together (Hanke et al., 1990) are to be expressed in the same cells. There are functional differences in the use of E3 vs E1 insertions, however, and one site may be preferable over the other for a particular application.

Substitution of the nonessential E3 region by foreign DNA has the advantage that replication-competent viruses can be generated readily. This property is of special interest for vaccine development, and many E3 substitution vectors have been designed for this purpose (Imler, 1995) . The most common method of generating E3 insertions in Ad5 vectors has been replacement of the XbaI fragment between 78.8 and 85.7 m.u. by foreign sequence. This removes most E3 coding sequences but leaves the E3 promoter intact, which allows it to direct expression of the transgene. A number of studies using such substitutions have reported that transgene expression was largely E3 driven, even when the inserted DNA contained its own promoter sequences (Both et al., 1993; Schneider et al., 1989; Johnson et al., 1988) . Expression of genes inserted into the E3 region of a nondefective vector may also be affected by late viral promoters, probably the MLP, late in infection (Dewar et al., 1989; Morin et al., 1987) . This viral regulation of transgene expression may be a disadvantage if cell type or temporally restricted expression is desired.

In contrast, the E1 deletions generally used remove the E la pro-moter sequences. In many cases, cellular promoters incorporated into the E1 region of adenoviral vectors have been shown to be regulated in a manner similar to their endogenous genes, allowing expression of a gene of interest in a cell-type or tissue-specific manner. As El-deleted vectors are almost completely replication defective and can have minimal effects on normal cellular processes, they have been chosen for most work involving cell biology studies or in vivo experiments.

Because viral proteins are translated preferentially at the expense of host proteins late in infection, it can be advantageous to use viral late gene control elements for applications where very high level expression is desired (for a detailed discussion of this topic, see Berkner, 1992) . As noted earlier, expression of a foreign gene placed under the control of the viral MLP and TPL can be enhanced along with that of bona fide viral gene products Alkhatib and Briedis, 1988; Berkner, 1992; Davis et al., 1985; Garnier et al., 1994) . Sequences in the first intron of the TPL have also been found to increase MLP transcription in infected cells, contributing to the very high level expression seen in some of these studies (Mansour et al., 1986; Mason et al., 1990) . The highest yields of recombinant protein reported from Ad vectors (Table 1) have been from constructs using both the MLP and the TPL for transgene expression in cells that support productive viral infections (e.g., El-deleted vectors in 293 cells or nondefective vectors in other cell types). Inclusion of the TPL in expression constructs has been reported previously to boost the production of a variety of proteins, even in uninfected cells (Sheay et al., 1993) , and TPL can increase greatly the expression of the Ad5 fiber protein in 293-based cell lines in the absence of any viral infection (Von Seggern et al., 1998) . These sequences may therefore improve protein yield even in the context of a nonpermissive cell type infected by a defective virus and may have general value in protein expression work.

As noted earlier, the E3 promoter has been used (sometimes inadvertently) to express many foreign genes. E3 is transcribed in many cell types, and transcription is activated within a few hours postinfection.

Many different promoters have been inserted into Ad vectors and in most cases have retained their activity. Widely used viral promoters include the long terminal repeat of RSV, the SV40 early promoter, and the cytomegalovirus (CMV) major immediate-early promoter. The CMV promoter provides very strong expression of a transgene in many cell types, at least in the short term (Jiang et al., 1996; Guo et al., 1996) , and has become perhaps the most widely used promoter in adenoviral vectors. Bartlett et al. (1996) have reported that the constitutively active U1 small nuclear RNA promoter, which is transcribed in essentially all cell types, has activity similar to CMV and is active when placed into an El-deleted Ad vector.

When placed in the adenovirus context, a number of tissue-specific promoters have been shown to be regulated much like their cellular counterparts. Friedman and co-workers (1986) found that the liver-specific rat albumin promoter in an Ad5 vector was expressed in rat or human hepatoma cells or in primary hepatocytes, but not in a myeloma cell line, whereas the B-cell-specific immunoglobulin heavy chain promoter was expressed strongly in myeloma but not hepatoma cells. More recently, studies of a modified mouse albumin promoter driving expression of the human factor VIII gene demonstrated specificity both for hepatocytes in vitro and for liver in vivo (Connelly et al., 1996) .

A number of groups have sought to target adenovirus-based gene expression to tumor cells. Hepatoma cells often express elevated levels of ~-fetoprotein (AFP) relative to normal liver. A recombinant Ad carrying AFP promoter and enhancer sequences driving the herpes simplex virus thymidine kinase (TK) gene expressed high levels of TK in AFP-expressing tumor cell lines such as HUH-7 or HepG2, but not in non-AFP-expressing liver cell lines such as HLF . Other workers have reported that expression of foreign genes placed under the control of the carcinoembryonic antigen (CEA), osteocalcin (OC), or DF-3 promoters in Ad vectors is specific for CEA-, OC-, or DF-3-expressing tumor cells, respectively Ko et al., 1996; Chen et al., 1995) . Hashimoto et al. (1996) found that Ad-mediated expression using two different neural cell-specific promoters was restricted to the expected cell types both in culture and after administration of the respective Ad vectors to rat brain in vivo. The ability to control the level or timing of transgene expression may be desirable in a particular experimental setting. A number of inducible or regulated promoters have been placed into the Ad backbone. For example, a synthetic construct consisting of minimal thymidine kinase promoter sequences and several retinoic acid response elements was inserted into the E 1 region of an Ad5 vector. Transcription of a reporter gene in infected cells could be induced by the addition of retinoids, both in culture and in vivo (Hayashi et al., 1994) . The mouse metallothionine promoter also retained its zinc-inducible expression in an Ad context (Yajima et al., 1996) . Varley et al. (1995) made Ad constructs containing a luciferase gene driven by either of two inflammation-responsive promoters. Cells infected with either virus were found to express high levels of luciferase in response to the cytokines present in conditioned culture medium. In vivo studies demonstrated that luciferase expression was upregulated in tissues of mice infected with these vectors and challenged with inflammatory agents such as lipopolysaccharide toxins or turpentine (Varley et al., 1995) . The pattern that emerges from these studies is that cellular promoters in an Ad chromosome often behave as would be predicted from their normal activities, although it may be worth noting that all of these specific promoter activities were generated using insertions into the viral E1 region. The ability to reproduce cell-specific patterns of gene expression after adenovirus-mediated transfer will be invaluable in targeting transgene function to specific cells and may be useful in optimizing expression of a recombinant protein in the cell type of interest.

The unique advantage of the adenoviral expression system is its versatility. Once constructed, a single vector can be used for in vitro protein expression and purification, studies of the effect of the gene product on cell biology, or in vivo studies in many tissue types and a number of different species. For these reasons, adenovirus is becoming the system of choice for work involving both in vitro and in vivo studies.

The investigator has the option of using either well-characterized viral regulatory sequences or a heterologous promoter, which should allow expression to be tailored to the requirements of the particular system of interest. The large (and growing) capacity of Ad vectors allows expression of large genes or of a combination of genes, as in expression of both subunits of a heterodimeric protein or of an antigenic protein together with a cytokine. Finally, the wide tropism and ability to infect nondividing cell types allows Ad vectors to be used in many cases where other vectors, such as retroviruses, cannot be used. The major disadvantage of this system is the complexity of vector construction. A recombinant vector for Ad-based expression of a protein can take considerably longer to produce than a plasmid for use in a prokaryotic expression system or simple transfection into eukaryotic cells. However, techniques for adenovirus construction have improved dramatically. The plasmid-based systems now available for construction of standard El-or E3-deleted vectors are quite straightforward, and the effort required for the construction of recombinant adenoviruses is similar to that needed to construct recombinant baculoviruses.

As Ad DNA does not normally integrate into host cell chromosomes, the expression of the transgene is transient. This is of little consequence in short-term or in vitro experiments, but may limit some long-term in vivo studies.

Although Ad vectors are being used successfully in vivo to study protein function, there are still problems associated with the host response to the vector itself. Following the administration of currently available Ad vectors, antivector immune responses are generated that can lead to inflammation, elimination of the infected cells, and short duration of transgene expression. This limits longterm studies in immunocompetent animals. Infection even by replication-defective Ads also affects the endogenous signal transduction pathways (Li et al., (I998) and can lead to responses in the infected cells, such as the production of cytokines. This can potentially complicate the interpretation of studies of protein expression in vivo. This type of effect can often be controlled for by administration of an Ad vector lacking an insert.

Finally, although Ad vectors have been found to infect a wide variety of cells in several species, including human, mouse, rat, and chicken, not all cells of interest may express the appropriate integrin and CAR proteins to allow infection. The repertoire of integrin and fiber receptor expression on the target cell type should therefore be considered when planning in vivo work using Ad vectors (Table 2) . 

As interest in adenovirus as a gene therapy vector remains high, further improvements in vector design and techniques for their construction can be expected. Packaging systems for the complementation of additional Ad genes such as E2, E4, terminal protein, or fiber are now available (Krougliak and Graham, 1995; Yeh et al., 1996; Caravokyri and Leppard, 1995; Schaack et al., 1995b; Gorziglia et al., 1996; Wang et al., 1995; Brough et al., 1996; Weinberg and Ketner, 1983; Amalfitano et al., 1996; Zhou et al., 1996 , Von Seggern et al., 1998 . Viral vectors deleted for the corresponding sequences will have a higher capacity for insertion of DNA and should be useful for expressing large proteins such as dystrophin or combinations of genes (perhaps multisubunit enzyme complexes). As they are expected to be less immunogenic and to persist longer in vivo, they should also improve the prospects for the use of Ad to study protein function in the context of an intact animal. Ongoing work is also extending the already impressive list of cell types that can be infected. Some cell types (most notably many hematopoietic cells) are difficult to infect using Ad. This is largely due to these cells lacking expression of either the CAR protein, which serves as the fiber receptor (Bergelson et al., 1997; Tomko et al., 1997) , or integrins appropriate for the internalization of virus (Wickham et al., 1993) . For example, in human airway organ culture, regenerating cells were found to be much more infectible than normal cells (Dupuit et al., 1995) . Another study showed that immature cells in the airway were much more susceptible to infection by Ad and that these cells expressed higher levels of the integrins needed for virus internalization (Goldman and Wilson, 1995) . In skeletal muscle, the infectibility of different cell populations has also been shown to correlate with ~v integrin expression (Acsadi et al., 1994) . Huang and colleagues have demonstrated that lymphoid cells can be infected efficiently if they are treated with growth factors that upregulate the synthesis of ~v integrins (Huang et al., 1995) and that there is an alternate pathway of infection using OL m integrins in at least some lymphoid cell types (Huang et al., 1996) .

Several groups are now developing adenoviruses that have modified tropism by altering the fiber or penton base proteins that interact with cellular receptors. Wickham et al. (1996) have described Ads with epitope tags incorporated into the penton base, allowing the virus to be redirected using a bispecific antibody approach. The modified virus was shown to infect endothelial and smooth muscle cells, which are normally difficult to infect due to low CAR expression (Wickham et al., 1996) .

Replacement of the fiber gene in a vector of one serotype by that of another has been shown to confer the cell-binding specificity associated with the new fiber serotype (Gall et al., 1996) . Chimeric fiber genes containing the receptor-binding domain of a different Ad serotype have also been used to alter viral tropism (Stevenson et al., 1997; Krasnykh et al., 1996) . Michael and colleagues added a short peptide sequence from gastrin-releasing peptide to the fiber (Michael et al., 1995) and demonstrated that this peptide epitope is accessible for antibody binding at the surface of the fiber protein.

A high-resolution structure for the receptor-binding knob domain of the fiber protein is now available, which should lead to improved strategies for redirecting fiber binding (Xia et al., 1994) . This type of work should lead to Ad vectors that can be targeted to particular cell types, extending the utility of the system for studies in vivo or using tissue samples.

Adenoviral gene transfer vectors are extremely versatile and are becoming widely used, both for protein expression in vitro and for studies of protein function in vivo. The high insert capacity of these vectors means that large genes or combinations of different genes can be transferred to cells. By using transcriptional and translational control elements from adenoviral late genes, a very high level expression of recombinant proteins has been demonstrated. Many tissue-specific or regulatable promoters have been shown to retain their function when placed on Ad chromosomes, allowing foreign genes to be expressed in a predictable way in the desired cell type.

As Ad vectors are becoming more widely used in gene therapy work, there has been much interest in improving the technology. Plasmid-based systems for vector construction have simplified the generation of recombinant Ad greatly, and the new Ad vector/packaging cell systems becoming available will provide further increases in capacity and in the effectiveness of in vivo gene transfer. Targetable vectors are being developed that will eventually allow expression of a protein of interest in specific cell types in vivo.

A differential efficiency of adenovirus-mediated in vivo gene transfer into skeletal muscle cells of different maturity

Intratumoral injection of an adenovirus expressing interleukin 2 induces regression and immunity in a murine breast cancer model

Transfer of a foreign gene into the brain using adenovirus vectors

Sequence analysis of adenovirus DNA: Complete nucleotide sequence of the spliced 5' noncoding region of adenovirus 2 hexon messenger RNA

The use of DNA viruses as vectors for gene therapy

High-level eucaryotic in vivo expression of biologically active measles virus hemagglutinin by using an adenovirus Type 5 helper-free vector system

Improved adenovirus packaging cell lines to support the growth of replication-defective gene-delivery vectors

Site-specific recombination mediated by an adenovirus vector expressing the Cre recombinase protein: A molecular swiitch for control of gene expression

Transfer of ~-amyloid precursor protein gene using adenovirus vector causes mitochondrial abnormalities in cultured normal human muscle

In vitro and in vivo synthesis of the hepatitis B virus surface antigen and of the receptor for polymerized human serum albumin from recombinant human adenoviruses

Efficient expression of protein coding genes from the murine U1 small nuclear RNA promoter

Overexpression of hexokinase I in isolated islets of Langerhans via recombinant adenovirus

Isolation of a common receptor for coxsackie B viruses and adenoviruses 2 and 5

Expression of heterologous sequences in adenoviral vectors

Generation of adenovirus by transfection of plasmids

Abundant expression of polyomavirus middle T antigen and dihydrofolate reductase in an adenovirus recombinant

Template requirements for in vivo replication of adenovirus DNA

Packaging capacity and stability of human adenovirus type 5 vectors

An efficient and flexible system for construction of adenovirus vectors with insertions or deletions in early regions 1 and 3

Protective immunity to rotavirus-induced diarrhoea is passively transferred to newborn mice from naive dams vaccinated with a single dose of a recombinant adenovirus expressing rotavirus VP7sc

Overexpression of RANTES using a recombinant adenovirus vector induces the tissue-directed recruitment of monocytes to the lung

Construction of a double recombinant adenovirus vector expressing a heterodimeric cytokine: In vitro and in vivo production of biologically active interleukin-12

A gene transfer vector-cell line system for complete functional complementation of adenovirus early regions E1 and E4

Expression and immunogenicity of the spike glycoprotein of porcine respiratory coronavirus encoded in the E3 region of adenovirus

Constitutive episomal expression of polypeptide IX (pIX) in a 293-based cell line complements the deficiency of pIX mutant adenovirus type 5

High level expression of the envelope glycoproteins of the human immunodeficiency virus type 1 in presence of rev gene using helper-independent adenovirus Type 7 recombinants

Efficient generation of recombinant adenovirus vectors by homologous recombination in Escherichia coli

Disappearance of body fat in normal rats induced by adenovirus-mediated leptin gene therapy

Breast cancer selective gene expression and therapy mediated by recombinant adenoviruses containing the DF3/MUC1 promoter

Coexpression of the simian immunodeficiency virus Env and Rev proteins by a recombinant human adenovirus host range mutant

An amazing sequence arrangement at the 5' ends of adenovirus 2 messenger RNA. Cell (Cambridge

The sequence of the genome of adenovirus type 5 and its comparison with the genome of adenovirus type 2

High-level tissue-specific expression of functional human factor VIII in mice

Expression of hepatitis B surface antigen with a recombinant adenovirus

Synthesis and processing of human immunodeficiency virus type 1 envelope proteins encoded by a recombinant human adenovirus

The adenovirus tripartite leader may eliminate the requirement for cap-binding protein complex during translation initiation

Translation by the adenovirus tripartite leader: Elements which determine independence from cap-binding protein complex

Gene expression from recombinant viral vectors in the central nervous system after blood-brain barrier disruption

Regenerating cells in human airway surface epithelium represent preferential targets for recombinant adenovirus

Synthesis and processing of the precursor to the major core protein of adenovirus type 2

Adenovirus early region 4 is required for efficient virus particle assembly

Characterization of 911: A new helper cell line for the titration and propagation of early region 1-deleted adenoviral vectors

Recombinant adenovirus deleted of all viral genes for gene therapy of cystic fibrosis

Expression of exogenous protein and analysis of morphogenesis in the developing chicken heart using an adenoviral vector

High-level expression of the measles virus nucleocapsid protein by using a replication-deficient adenovirus vector: Induction of an MHC-l-restricted CTL response and protection in a murine model

Cellular promoters incorporated into the adenovirus genome: Cell specificity of albumin and immunoglobulin expression

Adenovirus type 5 and 7 capsid chimera: Fiber replacement alters receptor tropism without affecting primary immune neutralization epitopes

Scale-up of the adenovirus expression system for the production of recombinant protein in human 293S cells

Expression of human ~l-antitrypsin using a recombinant adenovirus vector

Role of early region 3 (E3) in pathogenesis of adenovirus disease

Helper-free adenovirus type-5 vectors

Expression of ~v~5 integrin is necessary for efficient adenovirus-mediated gene transfer in the human airway

Elimination of both E1 and E2a from adenovirus vectors further improves prospects for in vivo human gene therapy

Covalently closed circles of human adenovirus DNA are infectious

Growth of 293 cells in suspension culture

Methods for construction of adenovirus vectors

Characteristics of a human cell line transformed by DNA from human adenovirus type 5

Evaluation of promoter strength for hepatic gene expression in vivo following adenovims-mediated gene transfer

Gene therapy using adenovirus vectors

In vivo expression of full-length human dystrophin from adenoviral vectors deleted of all viral genes

Development of a helper-independent human adenovirus vector and its use in the transfer of the herpes simplex virus thymidine kinase gene

Adenovirus early region 4 encodes functions required for efficient DNA replication, late gene expression, and host cell shutoff

Rescue of functional replication origins from embedded configurations in a plasmid carrying the adenovirus genome

Herpes simplex virus IgG receptors induced using recombinant adenovirus vectors expressing glycoproteins E and I

A neural cell-type-specific expression system using recombinant adenovirus vectors

The origin of adenovirus DNA replication: Minimal DNA sequence requirement in vivo

Expression of a thyroid hormone-responsive recombinant gene introduced into adult mouse liver by replication-defective adenovirus can be regulated by endogenous thyroid hormone receptor

A simplified system for generating recombinant adenoviruses

Identification of a repeated sequence element required for efficient encapsidation of the adenovirus type 5 chromosome

Adenoviruses. In "Virology

Upregulation of integrins ~vf~3 and ~vf~5 on human monocytes and T lymphocytes facilitates adenovirus-mediated gene delivery

Adenovirus interaction with distinct integrins mediates separate events in cell entry and gene delivery to hematopoietic cells

Apoptosis signaling pathway in T cells is composed of ICE/Ced-3 family proteases and MAP kinase kinase 6b

Purification of type 5 recombinant adenovirus encoding human p53 by column chromatography. Hum

Adenovirus vectors as recombinant viral vaccines

An efficient procedure to select and recover recombinant adenovirus vectors

Adenovirus-mediated in vivo gene transfer and expression in normal rat liver

Ability of adenovirus vectors containing different CFTR transcriptional cassettes to correct ion transport defects in CF cells

Abundant expression of herpes simplex virus glycoprotein gB using an adenovirus vector

Transcriptional modulation by the adenovirus E1A gene

Gene therapy for ~-fetoprotein-producing human hepatoma cells by adenovirus-mediated transfer of the herpes simplex virus thymidine kinase gene

Efficient manipulation of the human adenovirus genome as an infectious yeast artificial chromosome clone

Osteocalcin promoter-based toxic gene therapy for the treatment of osteosarcoma in experimental models

Analysis of protein structure-function in vivo

A new adenoviral vector: Replacement of all viral coding sequences with 28 kb of DNA independently expressing both full-length dystrophin and f~-galactosidase

HST-1/FGF-4 stimulates proliferation of megakaryocyte progenitors synergistically and promotes megakaryocyte maturation

Gene therapy: Adenovirus vectors

Generation of recombinant adenovirus vectors with modified fibers for altering viral tropism

Development of cell lines capable of complementing El, E4, and protein IX defective adenovirus type 5 mutants. Hum

Encapsidated adenovirus minichromosomes allow delivery and expression of a 14 kb dystrophin cDNA to muscle cells

Tumor-specific gene expression in carcinoembryonic antigen-producing gastric cancer cells using adenovirus vectors

293 cell lines that inducibly express high levels of adenovirus type 5 precursor terminal protein

Adenovirus-mediated transfer of a recombinant human ~l-antitrypsin cDNA to human endothelial cells

Adenovirus endocytosis via ~v integrins requires phosphoinositide-3-OH kinase

Adenovirus-mediated transfer of the amphotropic retrovirus receptor cDNA increases retroviral transduction in cultured cells

Construction and characterization of adenovirus co-expressing hepatitis B virus surface antigen and interleukin-6

In vivo transfer of a marker gene to study motoneuronal development

Downstream sequences affect transcription initiation from the adenovirus major late promoter

In vivo adenovirusmediated gene transfer into ocular tissues

Adenovirus vaccine vectors expressing hepatitis B surface antigen: Importance of regulatory elements in the adenovirus major late intron

A simple technique for the rescue of early region I mutations into infectious human adenovirus type 5. Virology

Addition of a short peptide ligand to the adenovirus fiber protein

Structure of adenovirus chromatin

Monitoring foreign gene expression by a human adenovirus-based vector using the firefly luciferase gene as a reporter

Evaluation of the concentration and bioactivity of adenovirus vectors for gene therapy

Efficient generation of recombinant adenoviruses using adenovirus DNA-terminal protein complex and a cosmid bearing the full-length virus genome

Recombinant adenovirus induces antibody response to hepatitis B virus surface antigen in hamsters

Adenoviral-mediated gene transfer to renal tubular cells in vivo

VEGF16s expressed by a replication-deficient recombinant adenovirus vector induces angiogenesis in vivo

A helper-dependent system for adenovirus vector production helps define a lower limit for efficient DNA packaging

A helper-dependent adenovirus vector system: Removal of helper virus by Cre-mediated excision of the viral packaging signal

Virus-receptor interaction in an adenovirus system

Efficient adenovirus-mediated transfer of a human minidystrophin gene to skeletal muscle of mdx mice

Identification of a protein linked to the ends of adenovirus DNA

DNA sequences from the adenovirus 2 genome

Adenovirus-mediated transfer of a recombinant ~l-antitrypsin gene to the lung epithelium in vivo

Construction of nondefective adenovirus type 5 bearing a 2.8-kilobase hepatitis B virus DNA near the right end of its genome

Efficient regulation of gene expression by adenovirus vector-mediated delivery of the CRE recombinase

Adenovirus-mediated transfer of the HST-1 (FGF4) gene induces increased levels of platelet count in vivo

Adenovirus terminal protein mediates both nuclear matrix association and efficient transcription of adenovirus DNA

Efficient selection of recombinant adenoviruses by vectors that express ~-galactosidase

Adenovirus type 5 precursor terminal protein-expressing 293 and HeLa cell lines

Characterization of middle T antigen expressed by using an adenovirus expression system

Expression of the glycoprotein of vesicular stomatitis virus by infectious adenovirus vectors

Stimulation of erythropoiesis by in vivo gene therapy: Physiologic consequences of transfer of the human erythropoietin gene to experimental animals using an adenovirus vector

Ex vivo and in vivo gene transfer to the skin using replication-deficient recombinant adenovirus vectors

Analytical anion-exchange HPLC of recombinant type-5 adenoviral particles

The infectivity of adenovirus 5 DNA-protein complex

Downstream insertion of the adenovirus tripartite leader sequence enhances expression in universal eukaryotic vectors

Adenoviridae: The viruses and their replication

Astroytes infected with replication-defective adenovirus containing a secreted form of CNTF or NT3 show enhanced support of neuronal populations in vitro

Selective targeting of human cells by a chimeric adenovirus vector containing a modified fiber protein

Cloning of a DNA fragment from the left-hand terminus of the adenovirus type 2 genome and its use in site-directed mutagenesis

Widespread long-term gene transfer to mouse skeletal muscles and heart

Fluorescent focus assay of viruses on cell monolayers in plastic petri plates

The E3 11.6 KDa adenovirus death protein (ADP) is required for efficient cell death: Characterization of cells infected with adp mutants

Forced degradation of Fas inhibits apoptosis in adenovirus-infected cells

HCAR and MAR: The human and mouse cellular receptors for subgroup C adenoviruses and group B coxsackieviruses

Gene transfer and expression in mouse preimplantation embryos by recombinant adenovirus vector

Adenovirus DNA replication

Molecular composition of the adenovirus type 2 virion

Inflammation-induced recombinant protein expression in vivo using promoters from acute-phase protein genes

Complementation of a fiber mutant adenovirus by packaging cell lines stably expressing the Ad5 fiber protein

A packaging cell line for propagation of recombinant adenovirus vectors containing two lethal gene-region deletions

Targeted DNA recombination in vivo using an adenovirus carrying the cre recombinase gene

Production of measles nucleoprotein in different expression systems and its use as a diagnostic reagent

Deletion of the E4 region of the genome produces adenovirus DNA concatemers

A cell line that supports the growth of a defective early region 4 deletion mutant of human adenovirus type 2

Integrins av~ 3 and avis promote adenovirus internalization but not virus attachment

Targeted adenovirus gene transfer to endothelial and smooth muscle cells by using bispecific antibodies

Constitutive and enhanced expression from the CMV major IE promoter in a defective adenovirus vector

Adenovirus genes that modulate the sensitivity of virusinfected cells to lysis by TNF

Inhibition of endothelial cell activation by adenovirus-mediated expression of IkB~, an inhibitor of the transcription factor NF-kB

Crystal structure of the receptor-binding domain of adenovirus type 5 fiber protein at 1,7 A resolution

Adenovirus-mediated cytokine gene transfer at tissue sites

Transfer of granulocyte-macrophage colony-stimulating factor gene to rat lung induces eosinophilia, monocytosis, and fibrotic reactions

Construction and characterization of a recombinant adenovirus vector carrying the human preproinsulin gene under the control of the metallothionein gene promoter

Adenovirus-mediated transfer of a truncated transforming growth factor-~ (TGF-f~) Type II receptor completely and specifically abolished diverse signaling by TGF-~ in vascular wall cells in primary culture

Cellular immunity to viral antigens limits El-deleted adenoviruses for gene therapy

Cellular and humoral immune responses to viral antigens create barriers to lung-directed gene therapy with recombinant adenoviruses

Human herpesvirus 7 infection of lymphoid and myeloid cell lines transduced with an adenovirus vector containing the CD4 gene

Co-expression of hepatitis B virus antigens by a non-defective adenovirus vaccine vector

Efficient dual transcomplementation of adenovirus E1 and E4 regions from a 293-derived cell line expressing a minimal E4 functional unit

Nucleotide sequence analysis of the leader segments in a cloned copy of adenovirus 2 fiber mRNA

Generation and identification of recombinant adenovirus by liposomemediated transfection and PCR analysis

A late adenovirus factor induces eIF-4E dephosphorylation and inhibition of cell protein synthesis

Purification and characterization of a protein tyrosine phosphatase containing SH2 domains

Development of a complementing cell line and a system for construction of adenovirus vectors with E1 and E2a deleted

The authors thank members of the Nemerow Laboratory for their assistance, Colleen McKiernan for comments on the manuscript, and Joan Gausepohl for her help with the manuscript.