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

Percutaneous transluminal in vivo gene transfer by recombinant adenovirus in normal porcine coronary arteries, atherosclerotic arteries, and two models of coronary restenosis

BACKGROUND

Gene therapy has been proposed as a possible solution to the problem of restenosis after coronary angioplasty. The current study was undertaken to assess conventional methods of gene transfer and to develop percutaneous techniques for introducing genes directly into the coronary arteries of large mammals. Since the anticipated targets of gene therapy against restenosis include atherosclerotic and previously instrumented arteries, we also evaluated gene transfer in atherosclerotic coronary arteries and in two porcine models of restenosis: one using intracoronary stents and a second using balloon overstretch angioplasty.

METHODS AND RESULTS

The conventional method of using perforated balloon catheters to deliver Lipofectin-DNA complexes directly into the coronary arteries of intact animals was applied to 18 porcine coronary arteries including normal arteries, hypercholesterolemic arteries, and those simulating restenosis. The results of this study were consistent with previously published results indicating that only low levels of luciferase gene expression could be obtained by Lipofectin-mediated gene transfer. We therefore undertook a second, parallel study to evaluate percutaneous transluminal in vivo gene transfer using a replication-deficient adenoviral vector. A comparison of the two studies revealed that the mean level of reporter gene expression in the cohort undergoing adenoviral infection was 100-fold higher than in the cohort undergoing Lipofection. Analysis of luciferase activity over time in normal arteries revealed that recombinant gene expression was half-maximal after 1 day, peaked within 1 week, was still half-maximal at 2 weeks, and declined to low levels by 4 weeks. Histochemical analysis of coronary arteries treated with a second adenovirus expressing a nuclear-localized beta-galactosidase gene demonstrated gene transfer to a limited number of cells in the media and adventitia. Immunohistochemical analysis of Ad5-infused arteries using a monoclonal antibody directed against CD44 identified a periadventitial infiltrate composed of leukocytes.

CONCLUSIONS

The recombinant adenoviral vectors proved to be far more effective than Lipofectin at delivering foreign genes directly into the coronary arteries of living mammals. Furthermore, the influences of hypercholesterolemia and arterial injury appeared to have little effect on the levels of gene expression obtained using either method. The results demonstrate that low-level recombinant gene expression, the major obstacle impeding gene therapy for the prevention of restenosis, can potentially be overcome by using adenoviral vectors to mediate coronary gene transfer in vivo. The duration of gene expression provided by these vectors and their effective deployment in atherosclerotic, balloon-overstretched, and stented coronary arteries suggest that recombinant adenovirus may have potential for evaluating gene therapy in the clinically informative porcine models of coronary restenosis.

Direct in vivo gene transfer into porcine myocardium using replication-deficient adenoviral vectors

BACKGROUND

Efficient methods of introducing genes into myocardial cells must be developed before local somatic cell gene therapy can be implemented against myocardial disease. Although adenoviral (Ad5) vectors have been used to target rodent hearts and plasmid DNA has been directly injected into the myocardium of rats and dogs, the amounts of recombinant protein produced by these procedures have not been reported, and adenoviral vectors have not been used in large mammalian hearts.

METHODS AND RESULTS

Replication-deficient recombinant adenoviral vectors carrying either the luciferase or lacZ reporter genes were injected directly into the ventricular myocardium of adult domestic swine for evaluation of reporter gene expression. This procedure did not affect regional myocardial function as assessed by systolic wall thickening using ultrasonic crystals. Luciferase activity was detected 3 days after injection, increased markedly at 7 days, and then declined progressively at 14 and 21 days. Luciferase production was comparable in the right and left ventricular walls and increased with increasing amounts of virus, reaching 61 +/- 21 ng at the highest dose examined (3.6 x 10(9) plaque-forming units). The injection of 200 micrograms of plasmid DNA (pRSVL) produced levels of luciferase comparable to 1.8 x 10(8) plaque-forming units of recombinant Ad5; however, when normalized to the number of genes injected, the adenovirus was 140,000 times more efficient than plasmid DNA. Histochemical analysis of beta-galactosidase activity produced by a second Ad5 vector demonstrated that nearly all (> 95%) of the stained cells were cardiomyocytes and that the percentage of cardiomyocytes infected by the virus could be quite high in microscopic regions adjacent to the needle track (up to 75% in fields of 60 to 70 cells); however, Ad5-infected cells were rarely observed farther than 5 mm from the injection site. Furthermore, the Ad5 vector induced pronounced leukocytic infiltration that was far in excess of that seen after injection of vehicle alone.

CONCLUSIONS

This study demonstrates for the first time that direct intramyocardial injection of replication-deficient adenovirus can program recombinant gene expression in the cardiomyocytes of a large animal species with relevance to human physiology. The efficiency of adenovirus-mediated gene transfer is far superior to that of plasmid DNA injection, and this method appears to be capable of producing more recombinant protein. However, the cell-mediated immune response to the Ad5 vector and the limited distribution of reporter gene expression suggest that less immunogenic recombinant vectors and more homogeneous administration methods will be required before Ad5 vectors can be successfully used for phenotypic modulation.

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

Human adenoviruses (Ads) are attracting considerable attention because of their potential utility for gene transfer and gene therapy, for development of live viral vectored vaccines, and for protein expression in mammalian cells. Engineering Ad vectors for these applications requires a variety of reagents in the form of Ads and bacterial plasmids containing viral DNA sequences and requires different strategies for construction of vectors for different purposes. To simplify Ad vector construction and develop a procedure with maximum flexibility, efficiency, and cloning capacity, we have developed a vector system based on use of Ad5 DNA sequences cloned in bacterial plasmids. Expanded deletions in early region 1 (3180 bp) and early region 3 (2690 or 3132 bp) can be combined in a single vector that should have a capacity for inserts of up to 8.3 kb, enough to accommodate the majority of cDNAs encoding proteins with regulatory elements. Genes can be inserted into either early region 1 or 3 or both and mutations or deletions can be readily introduced elsewhere in the viral genome. To illustrate the flexibility of the system, we have introduced a wild-type early region 3 into the vectors, and to illustrate the high capacity for inserts, we have isolated a vector with two genes totaling 7.8 kb.

Characterization of transgenic mice containing adenovirus early region 3 genomic DNA

Human adenoviruses (Ad) contain a complex transcription region (E3) which codes for proteins that interact with several arms of the immune system. However, E3 genes are not essential for replication in tissue culture. An E3-encoded 19,000-molecular-weight (19K) glycoprotein (gp19K) binds to the class I major histocompatibility complex (MHC) in the endoplasmic reticulum and prevents MHC transport to the cell surface. Three other E3 proteins are involved in the inhibition of apoptosis by tumor necrosis factor alpha. The entire E3 genomic DNA was utilized to produce transgenic mice to study the effect of the E3 proteins on pathogenesis of various infectious agents and to investigate the in vivo synthesis and processing of the multiple E3 mRNAs and proteins. There was basal expression of the E3 promoter in the thymus, kidneys, uterus, and testes and at all levels of the gastrointestinal tract. In addition, the E3 promoter of the transgene could be activated in some other organs, including the liver, by infection of these animals with an E3-deficient Ad (Ad7001) which contains a functional E1A region. Transactivation in vivo could also be demonstrated by infusion of bacterial lipopolysaccharide. There appeared to be differential ratios of expression between several of the E3 mRNAs in transgenic lung fibroblasts and primary kidney cells cultured from the transgenic animals. This observation suggested that there was differential mRNA splicing that was organ specific. These transgenic animals should provide a useful model for studying the effects of the E3 proteins on the immune system and on diseases affected either by control of MHC or by selected functions of tumor necrosis factor that are inhibitable by Ad E3 proteins.

The E1 sequence of bovine adenovirus type 3 and complementation of human adenovirus type 5 E1A function in bovine cells

The bovine adenovirus type 3 (BAV3) genome was sequenced from the left end to the HindIII site at 11%. This region comprises the entire E1 transcription unit including the open reading frames (ORF) for proteins homologous to the E1A, E1B proteins and protein IX of human adenovirus type 5 (Ad5). A portion of the BAV3 E1A protein showed significant homology with conserved region 3 (CR3), the principal transactivation region of Ad5 E1A. The BAV3 E1A protein also contains a consensus sequence known to be important for interaction with the cellular Rb protein but lacks most of the sequence corresponding to the second exon of Ad5 E1A. Promoter sequences for BAV3 E1B were not defined though the relevant region contains a 35-base pair repeat sequence. Two ORFs define the BAV3 E1B coding unit; one with regions homologous to sequences within the Ad5 E1B 19k protein, and an overlapping ORF with significant homology to the Ad5 E1B 55k protein. The encoded BAV3 E1B proteins of 157 and 420 amino acid residues (R) have predicted unmodified molecular weights of 17,393 and 46,734 respectively. Immediately following the E1B coding region there is a transcription unit containing an SP1 binding site and TATA box followed by an ORF which encodes a protein of 125R and predicted molecular weight of 13,706 with homology to protein IX of Ad5. Five concensus poly A addition sites are located in the 350 base pairs immediately following the protein IX coding region. The homology of sequences in the Ad5 E1A CR3 region and the corresponding BAV3 protein suggested that the BAV3 protein could transactivate certain Ad5 genes normally transactivated by the Ad5 E1A product. Evidence for this hypothesis was obtained in studies in which bovine cells in culture were coinfected with BAV3 and a human adenovirus type 5 (Ad5) recombinant viral vector lacking the E1A region and having a lacZ reporter gene within the E3 region dependent on E1A for its expression. Coinfection resulted in the induction of beta-galactosidase activity and the increased expression of other Ad5 early (E2A 72k) and late (hexon) proteins.

Prolonged and effective blockade of tumor necrosis factor activity through adenovirus-mediated gene transfer

A chimeric protein capable of binding and neutralizing tumor necrosis factor (TNF) and lymphotoxin was expressed in mice transduced with a replication-incompetent adenoviral vector into which a TNF inhibitor gene had been engineered. Within 3 days following the injection of 10(9) infectious particles, the TNF inhibitor concentration exceeded 1 mg/ml of plasma; this level of expression was maintained for at least 4 weeks, and detectable TNF inhibitory activity was measured 6 weeks after injection of the recombinant virus. Introduction of the artificial gene produced a phenotypic effect comparable to homozygous deletion of the 55-kDa TNF receptor, in that animals were rendered highly susceptible to infection by Listeria monocytogenes, whereas control animals receiving a replication-incompetent virus coding for beta-galactosidase were capable of resisting Listeria challenge. Adenovirus-mediated transfer of a gene encoding a TNF inhibitor offers a practical means of imposing effective, long-term blockade of TNF activity in vivo for investigational and therapeutic purposes.

Quantitative determination of adenovirus-mediated gene delivery to rat cardiac myocytes in vitro and in vivo

To optimize the use of modified adenoviruses as vectors for gene delivery to the myocardium, we have characterized infection of cultured fetal and adult rat cardiac myocytes in vitro and of adult cardiac myocytes in vivo by using a replication-defective adenovirus carrying the chloramphenicol acetyltransferase (CAT) reporter gene driven by the cytomegalovirus promoter (AdCMVCATgD). In vitro, virtually all fetal or adult cardiocytes express the CAT gene when infected with 1 plaque-forming unit of virus per cell. CAT enzymatic activity can be detected in these cells as early as 4 hr after infection, reaching near-maximal levels at 48 hr. In fetal cells, CAT expression was maintained without a loss in activity for at least 1 week. Using in vitro studies as a guide, we introduced the AdCMVCATgD virus directly into adult rat myocardium and compared the expression results obtained from virus injection with those obtained by direct injection of pAdCMVCATgD plasmid DNA. The amount of CAT activity resulting from adenovirus infection of the myocardium was orders of magnitude higher than that seen from DNA injection and was proportional to the amount of input virus. Immunostaining for CAT protein in cardiac tissue sections following adenovirus injection demonstrated large numbers of positive cells, reaching nearly 100% of the myocytes in many regions of the heart. Expression of genes introduced by adenovirus peaked at 5 days but was still detectable 55 days following infection. Adenoviruses are therefore a very useful tool for high-efficiency gene transfer into the cardiovascular system.

Adenovirus mediated expression of therapeutic plasma levels of human factor IX in mice

Gene therapy strategies designed to combat haemophilia B, caused by defects in clotting factor IX, have so far concentrated on ex vivo approaches. We have now evaluated adenoviral vector-mediated expression of human factor IX in vivo. Injection of the vector Av1H9B, which encodes human factor IX cDNA, into the tail veins of mice resulted in efficient liver transduction and plasma levels of human factor IX that would be therapeutic for haemophilia B patients. However, levels slowly declined to baseline by nine weeks and were not re-established by a second vector injection. These results address both the advantages and obstacles to the use of adenoviral vectors for treatment of haemophilia B.

Packaging capacity and stability of human adenovirus type 5 vectors

Adenovirus vectors are extensively used for high-level expression of proteins in mammalian cells and are receiving increasing attention for their potential use as live recombinant vaccines and as transducing viruses for use in gene therapy. Although it is commonly argued that one of the chief advantages of adenovirus vectors is their relative stability, this has not been thoroughly investigated. To examine the genetic stability of adenovirus type 5 vectors and in particular to examine the relationship between genetic stability and genome size, adenovirus vectors were constructed with inserts of 4.88 (herpes simplex virus type 1 gB), 4.10 (herpes simplex virus type 1 gB), or 3.82 (LacZ) kb combined with a 1.88-kb E3 deletion or with a newly generated 2.69-kb E3 deletion. The net excess of DNA over the wild-type (wt) genome size ranged from 1.13 to 3.00 kb or 3.1 to 8.3%. Analysis of these vectors during serial passage in tissue culture revealed that when the size exceeded 105% of the wt genome length by approximately 1.2 kb (4.88-kb insert combined with a 1.88-kb deletion), the resulting vector grew very poorly and underwent rapid rearrangement, resulting in loss of the insert after only a few passages. In contrast, vectors with inserts resulting in viral DNA close to or less than a net genome size of 105% of that of the wt grew well and were relatively stable. In general, viruses with genomes only slightly above 105% of that of the wt were unstable and the rapidity with which rearrangement occurred correlated with the size of the insert. These findings suggest that there is a relatively tight constraint on the amount of DNA which can be packaged into virions and that exceeding the limit results in a sharply decreased rate of virus growth. The resultant strong selection for variants which have undergone rearrangement, generating smaller genomes, is manifested as genetic instability of the virus population.