Gene therapy applications in gastroenterology and hepatology.

Advantages and disadvantages of viral vectors and nonviral vectors for gene delivery to digestive organs are reviewed. Advances in systems for the introduction of new gene expression are described, including self-deleting retroviral transfer vectors, chimeric viruses and chimeric oligonucleotides. Systems for inhibition of gene expression are discussed, including antisense oligonucleotides, ribozymes and dominant-negative genes.


Recent advances in retroviral vectors:
The long terminal repeat (LTR) regions of recombinant retroviral vectors contain viral enhancers and promoters that may activate host proto-oncogenes upon integration. Also, the enhancers can interfere with other promoters on the vector, making it difficult to manipulate the expression of the target gene. However, deletion of the LTR enhancer-promoter region reduces the virus titre greatly. Hwang et al (1) replaced the 3¢-LTR enhancer-promoter region with a tetracycline operator. Viral expression was activated by providing a transactivator supplied from a plasmid in packaging cells, producing virus in high titre lacking the LTR.
In another approach, a self-deleting retroviral vector was prepared using a Cre-loxP system in which enhancers are self-deleted after integration (2). Recent examples of in vivo gastrointestinal models for gene therapy: Shaked et al (3) transfected donor hepatocytes ex vivo by incubating them with retroviral vector encoding neomycin phosphotransferase or interleukin-7, resulting in sustained expression of target gene in hepatocytes following subsequent liver transplantation. In an in vivo transduction animal model, Kay et al (4) demonstrated that transfusion of recombinant retroviral vector encoding human factor IX into factor IX-deficient dogs via the portal vein resulted in persistent expression of factor IX for as long as five months. Noel et al (5) showed that retroviral vectors could transduce an intestine-derived cell line and that the efficiency was proportional to the amount of virus administered. Lau et al (6) investigated gene transfer in rat intestine by a retroviral vector encoding a reporter gene into the lumen of a segment of rat intestine and found that the reporter gene was expressed in both the crypts and villi. Clinical trials: Grossman et al (7) studied a patient with familial hypercholesterolemia due to defective low density lipoprotein (LDL) receptors. A recombinant retroviral vector containing a normal gene for the LDL receptor was transduced into hepatocytes of the patient ex vivo and then transferred back into the body via portal vein infusion. The target gene was expressed for up to four months, with a significant decrease in LDL level and a decrease in the LDL to HDL ratio from 10 to 13 before treatment to 5 to 8 after treatment. These effects were maintained for at least 18 months.

Adenoviral Vectors Strategy for use of adenoviruses as vectors:
The main advantages of adenoviruses are that they do not require dividing cells and can infect a wide variety of tissues with high efficiency. In addition, the virus has a propensity for uptake by the liver. The genome generally remains in the nucleus as an extrachromosomal entity. However, expression of adenoviral vectors is usually transient. Early prototype adenoviral vectors caused inflammation and immune response upon repeated administration. Recent advances in adenoviral vectors: Prototype adenoviral vectors caused an immune response against infected cells and could only transfer exogenous genes up to 8.5 kb in size. To address these problems, Haecker et al (8) developed a new adenoviral delivery system with deletion of all viral genes. A helper virus was used to supply all essential viral genes for viral replication and packaging. The new vector successfully transferred target genes into myotubes in vitro and muscle fibres in vivo. Recent examples of in vivo gastrointestinal models for gene therapy: Adenoviral vectors have been used to transfer target genes into primary human or murine hepatocytes via portal vein injection (9). A cDNA of human lipase was transferred by an adenovirus vector into gallbladder epithelial cell lines and also into isolated sheep gallbladder cells (10). In immunocompromised mice, intravenous injection of adenovirus-encoding reporter gene resulted in expression in colonic epithelium for six months (11). Also, there is evidence that adenoviral vectors can transfer a reporter gene into rat small intestine epithelial cells via an intestinal feeding tube (12).

Chimeric Adenoretroviral Vectors
A chimeric adenoretroviral vector combining the favourable aspects of adenoviruses and retroviruses was described by Bilbao et al (13). Two chimeric adenoretroviral vectors were prepared, by inserting sequences encoding retroviral packaging (LTR) and retroviral vector replication (retroviral proteins gag/pol/env) into separate adenoviral vectors. These two vectors were used to transfect target cells in vivo, converting the target cells into producer cells, which can generate progeny retroviral particles that infect neighbouring cells, with integration into the host genome and stable expression.

AAV Vectors
Strategy for use of AAVs as vectors: AAV vectors are prepared like adenoviruses in that the gene of interest is substituted for critical viral genes. However, preparation is more complicated, usually requiring co-infection of helper virus such as adenovirus or herpesvirus. Major advantages of this system are that the virus is nonpathogenic, does not require dividing cells and has a broad infection specificity, and that the wild-type AAV can specifically integrate into the host genome on chromosome 19. This has the potential of generating vectors with persistent expression through a predictable innocuous integration. However, recombinant AAV have been reported to integrate less specifically. Recent examples of in vivo gastrointestinal models for gene therapy: Ponnazhagan et al (14) showed that intravenous injection of AAV resulted in a predominant uptake by the liver. Snyder et al (15) designed an AAV vector encoding the gene for human factor IX. The vector was prepared in a mutant adenovirus helper virus in which a missing early gene was supplied from a plasmid. A single infusion of AAV particles via portal vein into mice induced expression of human factor IX in the blood up to 2000 ng/mL for 36 weeks.

NONVIRAL VECTOR SYSTEMS
There are several nonviral vector delivery systems: direct injection of naked DNA, electroporation, particle bombard-ment, calcium phosphate precipitation, liposome transfer and receptor-mediated transfer. Compared with viral methods, nonviral vectors are noninfectious, are generally less immunogenic, are less toxic to cells and can be modified to contain targeting components, but are also frequently less efficient.

LIPOSOMES Strategy for use of liposomes as vectors:
The most successful system consists of two elements -cationic lipid and negative-charged DNA fragment. The amphiphilic cationic lipid couples to negatively charged phosphate groups of DNA backbone, neutralizes the DNA charge and condenses the complex. Various kinds of lipids are applied in this system (16). Recent examples of in vivo gastrointestinal models for gene therapy: Korba et al (17) used liposome-mediated phosphatidyl-ddG (2¢,3¢-dideoxyguanosine, lamivudine) to inhibit the replication of hepatitis virus. Detected by HBV DNA, liposome-treated groups achieved a much lower than 90% effective concentration. Daily injection of such liposomes intraperitoneally into woodchuck hepatitis virus (WHV)-infected woodchucks for four weeks decreased serum WHV DNA by 23-to 46-fold compared with the group treated with free ddG. Clinical trials: A phase I trial using a liposome system for in vivo hepatic delivery has been reported. Rubin et al (18) directly injected liposomes containing a human leukocyte antigen (HLA)-B7 gene with beta-2 microglobulin into hepatic metastatic lesions of 15 colorectal cancer patients. The target DNA was detected in 93% of patients, mRNA detected in 33%, and the HLA-B7 was detected in 50% to 63% (by fluorescence-activated cell sorting or immunohistochemistry, respectively). No serious toxicity was observed.

Ligand Receptor-mediated Gene Delivery Systems
In ligand receptor-mediated gene delivery systems, natural cell surface receptors are used to internalize nucleic acids by coupling an appropriate ligand with the DNA of interest. Depending on the specificity of the receptor for host cells, genes can be targeted to specific cell types. For example, asialoglycoproteins have been used to target genes to the liver by binding to the asialoglycoprotein receptor on the cell surface of hepatocytes. In an analogous manner, insulin and transferrin (19) have also been used to target genes to other cell types.

Asialoglycoproteins as vectors:
The asialoglycoprotein receptor is selectively distributed on the plasma membrane of mammalian hepatocytes. After binding with the receptor, the asialoglycoprotein is internalized by endocytosis, entering an endosomal compartment, and ultimately lysosomes. By linking DNA noncovalently to an asialoglycoprotein carrier, DNA may be targeted to hepatocytes resulting in expression of the gene of interest.
Wu et al (20) showed that injection of an albumin-encoding plasmid complexed to an asialoglycoprotein increased albumin levels in analbuminemic rats up to 37 µg/mL. Simi-larly, an asialoglycoprotein/DNA complex was used to deliver the gene for a normal LDL receptor into Watanabe rabbit hepatocytes by peripheral venous injection. The cholesterol level in this animal model of familial hypercholesterolemia was decreased by 25% in serum, but the effect only lasted four to five days after which levels returned to previously high values (21).
To avoid intracellular degradation, approaches have been devised to incorporate agents that can disrupt endosomal vesicles, allowing DNA to escape before fusion with lysosomes. Viral fusogenic proteins such as influenza HA and vesicular stomatitis virus G protein (22,23) have been employed. Cristiano et al (24) co-infected primary cultures of murine hepatocytes with a replication-defective adenovirus together with an asialoglycoprotein-complexed reporter gene. All of the cells co-infected with adenovirus expressed the reporter gene, while only 0.1% of the cells without adenovirus were transfected. Similar results were also obtained in vivo. To ensure colocalization of the endosomolytic component and DNA complex, asialoglycoprotein was chemically linked to an adenovirus (25). This resulted in a modified adenovirus particle that had lost its natural infectivity but retained its infectivity for cells that possessed asialoglycoprotein receptors. Because of the known requirement of migration of endosomal vesicles on microtubules, a microtubule-disrupting drug, colchicine, was administered to block lysosomal degradation of targeted DNA. The gene for normal glucuronyl transferase in a targetable complex was injected intravenously following a single colchicine injection into Gunn rats with defective glucuronyl transferase. This resulted in substantial decreases in serum bilirubin and production of bilirubin conjugates in bile that lasted for at least 60 days (26).

MOLECULAR STRATEGIES FOR INHIBITION OF PRE-EXISTING GENE EXPRESSION Chimeric Molecules
An RNA/DNA hybrid capable of correcting a genetic mutation in situ has been described. These hybrids contained RNA nucleotides that were modified by 2¢-O-methylation on the ribose to prevent degradation by nuclease. An RNA/DNA chimeric was successfully introduced into the Huh-7 hepatoma cell line to correct a mutation in an alkaline phosphatase gene with reported efficiency of 30% (27). More recently, using a chimeric oligonucleotide, a mutation was produced in a factor IX gene in cell culture and in vivo. The latter resulted in a prolongation of the activated partial thromboplastin time (28).

Ribozymes
Ribozymes are catalytic RNA molecules that can specifically bind and cleave a target RNA (29). Chemical modification can enhance the stability of ribozymes. Hammerhead ribozymes have been prepared against the hepatitis C virus (HCV) targeting its 5¢-noncoding and core domains (30). In vitro assays showed that those ribozymes specifically cleaved target sequences of HCV and suppressed the expression of a reporter gene by up to 71%. Ribozymes have been introduced into hepatitis B virus (HBV)-, HCV-infected cells (31,32) to block the transcription of viral proteins.
The 5¢, 3¢ region around the core catalytic domain of ribozymes may affect its activity. Ruiz et al (33) created a multimeric ribozyme molecule to solve this problem. A monomer of ribozyme was designed in which target regions to the hepatitis B virus were placed 5¢ and 3¢ to the catalytic domain of the ribozyme. The construct was then cloned as tandem into an expression vector. Upon transfection, selfcleavage occurred, producing multiple monomeric ribozymes, lacking any extra sequences around its catalytic region.
Because ribozymes are composed of RNA, they are quite susceptible to nuclease degradation. This property can limit the effectiveness in cells. A DNA equivalent of ribozymes that will cleave at the direct repeat I and polyadenylation signal regions of the HBV mRNA was prepared. This enzyme was shown to inhibit intracellular HBV gene expression substantially, and can be modified to increase stability while maintaining activity (34).

Antisense Oligonucleotides
Antisense oligonucleotides are synthetic molecules designed to hybridize with a target mRNA sequence and block its translation. Antisense oligonucleotides in the form of asialoglycoprotein targetable complexes when introduced into HBV-producing cells significantly suppressed the viral gene expression and replication in vitro. Furthermore, pretreatment of Huh-7 hepatoma cells with asialoglycoprotein-based complexes of an antisense against the HBV poly A signal and 5¢ upstream region, blocked the subsequent HBV production after transfection of a plasmid that contained the complete HBV genome. Viral RNA was decreased by 60%, and the expression of hepatitis B surface antigen decreased by 97% (35).
Antisense oligodeoxynucleotides against duck HBV (DHBV) were prepared by Offensperger et al (36). Primary hepatocytes from DHBV-infected ducklings were exposed to these antisense oligonucleotides, resulting in strong inhibition of DHBV replication in vitro. Assays on two-week DHBV postinfected ducks showed that intravenous infusion of antisense oligodeoxynucleotides at a daily dose of 20 µg/g body weight for 10 consecutive days almost totally abolished DHBV replication. Also, injection of such antisense oligonucleotides into newborn healthy ducklings prevented subsequent infection by DHBV when challenged with infectious virus.
Using another model, Bartholomew et al (37) showed that targeted delivery of an antisense sequence against the poly A signal and 5¢ upstream region of WHV using a complex with an asialoglycoprotein-based conjugate in a woodchuck hepatitis model significantly decreased the virus particles in the bloodstream by five-to 10-fold, and this decrease was maintained over two weeks.
The 5¢ nontranslated region (NTR) of HCV contains important elements that control HCV translation. Wu and Wu (38) prepared antisense oligonucleotides directed against a sequence in the internal ribosomal binding site of the NTR, and a portion of the NTR overlapping the core protein translational start site of HCV. In transient transfections of a plasmid containing a luciferase gene immediately downstream from an HCV NTR insert, oligonucleotides in the form of asialoglycoprotein-polylysine complexes were administered to Huh-7 cells, and luciferase activity generated by cytomegalovirus HCVluc measured. Antisense complexes inhibited luciferase activity by 75% and 99% at 0.01 µM and 0.1 µM, respectively. In cell lines stably transfected with cytomegalovirus HCVluc plasmid, complexed antisense complexes inhibited luciferase activity in Huh-7 cells by 20% at 10 µM and 85% at 60 µM, and were competable by an excess asialoglycoprotein (38).

Dominant Negative Gene Products
A dominant negative gene product is a mutant protein that interacts and interferes with the function of its normal counterpart. This phenomenon was discovered as a natural mutant of hepadnaviral core protein, which inhibited the function of wild-type core protein and inhibited replication of the virus. Subsequently, intentional core mutants of human HBV and WHV were shown to block wild-type viral replication by 95%. By using retroviral or adenoviral vectors, Scaglioni et al (39) transferred HBV-or DHBV-dominant negative core mutants into hepatoma cell lines that had either constitutive or transient hepatitis infection, viral replication was greatly inhibited.