Pseudo-hydrodynamic delivery of helper-dependent adenoviral vectors into non-human primates for liver-directed gene therapy.

Helper-dependent adenoviral vectors (HDAds) are attractive for liver-directed gene therapy because they can mediate long-term, high-level transgene expression without chronic toxicity. However, systemic delivery requires high vector doses for efficient hepatic transduction, resulting in dose-dependent acute toxicity. Clearly, strategies to improve hepatic transduction with low vector doses are needed. In this regard, we have previously shown that hydrodynamic injection of helper-dependent adenoviral vectors into mice results in increased hepatic transduction, reduced systemic vector dissemination, and reduced pro-inflammatory cytokines compared with conventional injection and thus has the potential to improve dramatically the therapeutic index of helper-dependent adenoviral vectors. Unfortunately, the rapid, large-volume injection used in this method cannot be applied to larger animals. Therefore, we have developed a novel balloon occlusion catheter-based method to mimic hydrodynamic injection of helper-dependent adenoviral vectors into non-human primates that does not require rapid, large-volume injection. Using a low, clinically relevant vector dose, this minimally invasive method results in high-efficiency hepatic transduction with minimal toxicity and stable long-term transgene expression for at least 413 days.     

Vasoactive Intestinal Peptide Increases Hepatic Transduction and Reduces Innate Immune Response Following Administration of Helper-dependent Ad

Helper-dependent adenoviral vectors (HDAd) are effective tools for liver-directed gene therapy because they can mediate long-term transgene expression in the absence of chronic toxicity. However, high vector doses required for efficient hepatocyte transduction by intravascular delivery result in systemic vector dissemination and dose-dependent activation of the innate immunity. Therefore, strategies to achieve high-efficiency hepatocyte transduction using low vector doses and/or to reduce the acute elevations of proinflammatory cytokines and chemokines may have significant clinical potential. Vasoactive intestinal peptide (VIP) is an endogenous neuropeptide involved in the regulation of hepatic blood flow and plays an important role as modulator of immune functions. Here, we show that VIP pretreatment in mice is able to increase hepatocyte transduction by HDAd, decrease vector uptake by the spleen, reduce elevation of proinflammatory serum cytokines interleukin (IL)-6 and IL-12, and reduce serum levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) following intravenous HDAd injection. VIP pretreatment also resulted in a reduction in the expression of the chemokines macrophage-inflammatory protein 2 (MIP-2), monocyte chemotactic protein 1 (MCP-1), and regulated on activation normal T-cell expressed and secreted (RANTES) in the livers of mice injected with HDAd. These results suggest that VIP can improve the therapeutic index of HDAd by increasing hepatocyte transduction efficiency while reducing cytokine and chemokine expression following intravascular delivery of HDAd.     

Improved hepatic transduction, reduced systemic vector dissemination, and long-term transgene expression by delivering helper-dependent adenoviral vectors into the surgically isolated liver of nonhuman primates

Helper-dependent adenoviral vectors (HDAds) are attractive vectors for liver-directed gene therapy because they can mediate sustained, high-level transgene expression without chronic toxicity. However, high vector doses are required to achieve efficient hepatic transduction by systemic delivery because of a nonlinear dose response. Unfortunately, such high doses result in systemic vector dissemination and dose-dependent acute toxicity with potentially severe and lethal consequences. We hypothesize that the threshold to efficient hepatic transduction may be circumvented by delivering the vector into the surgically isolated liver via the portal vein. Total hepatic isolation was achieved by occluding hepatic inflow from the portal vein and hepatic artery and by occluding hepatic venous outflow at the inferior vena cava. We demonstrate in nonhuman primates that this approach resulted in significantly higher efficiency hepatic transduction with reduced systemic vector dissemination compared with systemic intravascular delivery. This method of delivery was associated with transient acute toxicity, the severity of which was variable. Importantly, stable, high levels of transgene expression were obtained for at least 665 days for one baboon and for at least 560 days for two baboons with no evidence of long-term toxicity.     

Acute toxicity after high-dose systemic injection of helper-dependent adenoviral vectors into nonhuman primates

Systemic intravascular delivery of adenoviral (Ad) vectors for liver-directed gene therapy has been widely employed because of its simplicity, noninvasiveness, and potential for high transduction. For first-generation Ad vectors (FGAd), this results in high but transient levels of transgene expression and long-term hepatotoxicity due to viral gene expression from the vector backbone. Furthermore, high doses also result in an acute innate inflammatory response with potentially lethal consequences. Unlike FGAd, helper-dependent Ad vectors (HDAd) contain no viral genes and can provide sustained, high-level transgene expression with negligible long-term toxicity. However, whether the absence of viral gene expression leads to any decrease of acute toxicity in nonhuman primates has yet to be determined. To address this, we injected one baboon with 5.6 x 10(12) HDAd viral particles (VP)/kg and a second with 1.1 x 10(13) VP/kg. Approximately 50% hepatocyte transduction, accompanied by mild and transient acute toxicity, was observed in the animal receiving the lower dose. In the animal receiving the higher dose, 100% hepatocyte transduction, accompanied by lethal acute toxicity, was observed. These results indicate that systemic delivery of HDAd, like FGAd, results in acute toxicity in baboons consistent with activation of the innate inflammatory response, the severity of which is dose dependent, and confirm the hypothesis that Ad-mediated acute toxicity is independent of viral gene expression.     

Increased hepatic transduction with reduced systemic dissemination and proinflammatory cytokines following hydrodynamic injection of helper-dependent adenoviral vectors.

Hydrodynamic injection of helper-dependent adenoviral vectors (HDAd) in mice results in increased hepatic transduction, reduced splenic and pulmonary transduction, and reduced levels of the proinflammatory cytokines IL-6 and IL-12 compared to conventional injection. These results indicate that hepatic transduction by HDAd, at least alone, does not necessarily provoke a severe innate inflammatory response. Instead, they suggest that systemic vector dissemination may play a major role in the severity of the innate inflammatory response. These results further suggest that the safety and efficacy of HDAd-mediated, liver-directed gene therapy may be improved if the vector could be preferentially, if not exclusively, targeted to liver.     

Correction of hyperbilirubinemia in gunn rats using clinically relevant low doses of helper-dependent adenoviral vectors

Crigler-Najjar syndrome type I is a severe inborn error of bilirubin metabolism caused by a complete deficiency of uridine diphospho-glucuronosyl transferase 1A1 (UGT1A1) and results in life-threatening unconjugated hyperbilirubinemia. Lifelong correction of hyperbilirubinemia by liver-directed gene therapy using a helper-dependent adenoviral (HDAd) vector has been previously reported in the Gunn rat, a model of Crigler-Najjar syndrome, but was only achieved using high doses (≥ 3 × 10(12) viral particles [vp]/kg), which are likely to elicit a severe toxic response in humans. Therefore, in this study, we investigate strategies to achieve correction of hyperbilirubinemia in the Gunn rat using clinically relevant low HDAd doses. We have found that correction of hyperbilirubinemia in the Gunn rat can be achieved with a low dose of 5 × 10(11) vp/kg by using an HDAd vector bearing a more potent UGT1A1 expression cassette. Furthermore, by using hydrodynamic injection of the improved HDAd vector, correction of hyperbilirubinemia in the Gunn rat can be achieved using an even lower dose of 5 × 10(10) vp/kg. Although hydrodynamic injection as performed in rats is not acceptable in humans, clinically attractive, minimally invasive methods have been successfully developed to mimic hydrodynamic injection of HDAd vector in non-human primates. Therefore, using an improved expression cassette combined with a more efficient method of vector delivery permits correction of hyperbilirubinemia in the Gunn rat using clinically relevant low HDAd doses and may thus pave the way to clinical application of HDAd vectors for Crigler-Najjar syndrome gene therapy.     

SR-A and SREC-I Are Kupffer and Endothelial Cell Receptors for Helper-dependent Adenoviral Vectors

Helper-dependent adenoviral (HDAd) vectors can mediate long-term, high-level transgene expression from transduced hepatocytes with no chronic toxicity. However, a toxic acute response with potentially lethal consequences has hindered their clinical applications. Liver sinusoidal endothelial cells (LSECs) and Kupffer cells are major barriers to efficient hepatocyte transduction. Understanding the mechanisms of adenoviral vector uptake by non-parenchymal cells may allow the development of strategies aimed at overcoming these important barriers and to achieve preferential hepatocyte gene transfer with reduced toxicity. Scavenger receptors on Kupffer cells bind adenoviral particles and remove them from the circulation, thus preventing hepatocyte transduction. In the present study, we show that HDAd particles interact in vitro and in vivo with scavenger receptor-A (SR-A) and with scavenger receptor expressed on endothelial cells-I (SREC-I) and we exploited this knowledge to increase the efficiency of hepatocyte transduction by HDAd vectors in vivo through blocking of SR-A and SREC-I with specific fragments antigen-binding (Fabs).     

Balloon Catheter Delivery of Helper-dependent Adenoviral Vector Results in Sustained,Therapeutic hFIX Expression in Rhesus Macaques

Hemophilia B is an excellent candidate for gene therapy because low levels of factor IX (FIX) (≥1%) result in clinically significant improvement of the bleeding diathesis. Helper-dependent adenoviral (HDAd) vectors can mediate long-term transgene expression without chronic toxicity. To determine the potential for HDAd-mediated liver-directed hemophilia B gene therapy, we administered an HDAd expressing hFIX into rhesus macaques through a novel and minimally invasive balloon occlusion catheter-based method that permits preferential, high-efficiency hepatocyte transduction with low, subtoxic vector doses. Animals given 1 × 10¹² and 1 × 10¹¹ virus particle (vp)/kg achieved therapeutic hFIX levels for the entire observation period (up to 1,029 days). At 3 × 10¹⁰ and 1 × 10¹⁰ vp/kg, only subtherapeutic hFIX levels were achieved which were not sustained long-term. Balloon occlusion administration of HDAd was well tolerated with negligible toxicity. Five of six animals developed inhibitors to hFIX. These results provide important information in assessing the clinical utility of HDAd for hemophilia B gene therapy.     

Modified infusion procedures affect recombinant adeno-associated virus vector type 2 transduction in the liver

Recombinant adeno-associated virus (rAAV) vectors have therapeutic potential for the treatment of several types of liver diseases including hepato-deficiency disorders. Most of the preclinical and clinical applications involve the use of adeno-associated vector serotype 2 (AAV-2). However, when this vector is delivered at high doses into the portal vein or hepatic artery, a relatively small number of hepatocytes are stably transduced. We elected to determine if the route of vector administration and altering the vascular delivery route within the liver influenced the relative level of transduction. First, we delivered an AAV vector expressing the human factor IX gene from a liver-specific promoter into the hepatic artery, portal vein, or general circulation of rats. Transgene expression was equal with hepatic artery and portal vein infusion, which was higher than vector administered via peripheral venous infusion. Next, we determined how localized perfusion or changing the vector dwell time affected AAV transduction in vivo. To do this, we infused an AAV vector lacking a functional expression and quantified transduction by quantifying the number of double-stranded vector DNA genomes. By increasing vector dwell time in the liver to 5 min, vector transduction was enhanced approximately 4- to 5- fold. To establish if gene transduction could be restricted to a specific anatomic location in the liver, we delivered vector into specific liver lobes by clamping the venous inflow to the middle and left liver lobes (noninfused lobes) and infusing vector into the right two liver lobes through the hepatic artery followed by vector circulation between the two right lobes and general circulation for 5 min. With this selective infusion, 40 to 120 times higher vector genome was observed in the perfused lobes than the nonperfused lobes. All the procedures described in this study were performed without detectable liver injury or toxicity. In all, the present study clearly demonstrated that hepatic arterial infusion of rAAV is effective for liver-directed gene therapy and that other parameters related to blood flow can be adjusted to further optimize gene transfer.     

Intrathecal injection of helper-dependent adenoviral vectors results in long-term transgene expression in neuroependymal cells and neurons

Helper-dependent adenoviral (HDAd) vectors are devoid of all viral genes and result in long-term transgene expression in the absence of chronic toxicity. Because of their ability to infect post-mitotic cells, including cells of the central nervous system, HDAd vectors are particularly attractive for brain-directed gene therapy. In this study, we show that intrathecal injection of HDAd results in extensive transduction of ependymal cells and sustained expression of the transgene up to 1 year post-administration. We also demonstrate, for the first time, the ability of HDAd injected by this route of delivery to transduce neuronal cells. The transduced neuroepithelial cells can be potentially used to secrete therapeutic proteins into the cerebrospinal fluid and provide them via cross-correction to nontransduced cells. Targeting of neuronal cells and long-term transgene expression make this approach attractive for the treatment of several neurologic diseases.     

Efficient gene delivery and targeted expression to hepatocytes in vivo by improved lentiviral vectors.

Safe and efficient genetic modification of liver cells could enable new therapies for a variety of hepatic and systemic diseases. Lentiviral vectors are promising tools for in vivo gene delivery. Previous data suggested that recruitment into the cell cycle was required for transduction of hepatocytes in vivo. We developed an improved vector design that enhanced nuclear translocation in target cells and significantly improved gene transfer performance. Using the new vector and a panel of internal promoters, we showed that rat hepatocytes were transduced ex vivo to high frequency without requirement for proliferation. On intravenous administration of vector into adult severe combined immunodeficient (SCID) mice, we found high levels (up to 30%) of transduction of parenchymal and nonparenchymal cells of the liver, integration of the vector genome in liver DNA and stable expression of the marker green fluorescent protein (GFP)-encoding gene without signs of toxicity. Coadministration of vectors and 5'-bromo-2'-deoxyuridine in vivo proved that cell cycling was not required for efficient transduction of hepatocytes. In addition to the liver, the spleen and the bone marrow were transduced effectively by systemic delivery of vector. GFP expression was observed in all these organs when driven by the cytomegalovirus promoter and by the phosphoglycerate kinase gene promoter. Using the promoter of the albumin gene, we could restrict expression to hepatocytes. By a single vector injection into the bloodstream of SCID mice, we achieved therapeutic-range levels of the human clotting factor IX, stable in the plasma for up to 1 year (the longest time tested), indicating the potential efficacy of improved lentiviral vectors for the gene therapy of hemophilias and other diseases.     

Development of catheter-based procedures for transducing the isolated rabbit liver with plasmid DNA

Rapid systemic injection of naked plasmid DNA (pDNA) in a large volume into a mouse tail vein has been shown to result in a high level of gene expression in the liver. However, the potential therapeutic benefit to humans embodied in hydrodynamic transfection of the liver cannot be realized until a clinically viable method for gene delivery is developed. In light of this fact, we have devised and evaluated several methods for delivering pDNA to the isolated rabbit liver using minimally invasive catheter-based techniques. Using a lobar technique, pDNA was delivered hydrodynamically to an isolated hepatic lobe using a balloon occlusion balloon catheter to occlude a selected hepatic vein. A whole organ technique was used wherein the entire hepatic venous system was isolated and the pDNA solution injected hydrodynamically into the vena cava between two balloons used to block hepatic venous outflow. Lobar delivery of a plasmid encoding a secreted alkaline phosphatase (SEAP) reporter gene resulted in significant levels of transgene product in the serum. A nonsecreted transgene product, chloramphenicol acetyltransferase (CAT), showed the highest levels of expression in the injected lobe distal to the injection site. Compared to lobar delivery, whole organ delivery yielded much higher serum levels of SEAP expression and a significantly broader hepatic parenchymal distribution of CAT expression. These preliminary studies suggest that catheter-mediated hydrodynamic delivery of pDNA to the isolated liver may provide a method for human gene therapy that is both therapeutically significant and clinically practicable.     

Sustained phenotypic correction of canine hemophilia B after systemic administration of helper-dependent adenoviral vector

We have evaluated the potential of liver-directed, helper-dependent adenoviral (HDAd) vector-mediated gene therapy in the hemophilia B dog. Two dogs were injected intravenously with HDAd (3 x 10(12) VP/kg) bearing a liver-restricted canine coagulation factor IX (FIX) expression cassette. After injection, the whole blood clotting time for both dogs declined from >60 min to 170 ng/ml for at least 256 days. For the other dog, a peak FIX level of 1258 ng/ml was achieved and stabilized at >400 ng/ml for at least 213 days. Inhibitor formation was not evident in either animal. Importantly, whereas untreated hemophilia B dogs suffer five or six spontaneous bleeds per year, the treated dogs suffered no such bleeds postinjection. Significantly, this study is the first to demonstrate long-term phenotypic correction of a genetic disorder in a large animal with HDAd. Although no evidence of chronic toxicity was observed in either animal, systemic vector administration at 3 x 10(12) VP/kg was accompanied by acute, albeit transient and variable laboratory abnormalities (alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, creatine phosphokinase, and platelet counts). The results of this study highlight both the potential benefit and the risk associated with systemic intravascular delivery of high-dose HDAd for liver-directed gene therapy.     

Factors influencing retention of adenovirus within tumours following direct intratumoural injection

Direct intratumoural (IT) administration of adenovirus is widely used, however little is known about the resulting distribution of virus particles. Here we have evaluated the influence of tumour size, volume of injectate and occlusion of injection sites (to prevent retrograde seepage) on particle biodistribution and transgene expression. In subcutaneous MDA-231 xenografts, IT injection of relatively large volumes (4 x 20% (vol/vol) injections) resulted in just 40% of the administered dose being retained in tumour tissue after 30 min, with 15% in the liver thought to reflect systemic 'overflow'. Occlusion of the injection sites using surgical adhesive increased retention of the vector to 80% in the tumour with no increase in liver levels. Spread of expression was enhanced using multiple injection sites, but not by using larger injectate volumes. In ZR75.1 breast carcinoma xenografts virus distribution was different, with no evidence of systemic overflow leading to hepatic transduction following IT injection. Typically, clinical doses employ up to 30% vol/vol IT injections. Depending on the tumour, this may give considerable systemic overflow and might account for the high frequency of fevers observed. Virus performance might be improved by tailoring volumes and frequency of IT injection for tumour biology or histotype.     

Robust spinal motor neuron transduction following intrathecal delivery of AAV9 in pigs

Adeno-associated viral vector 9 (AAV9) has recently been shown to penetrate the blood-brain barrier via intravascular administration, making it a good candidate for diffuse gene delivery. However, the potential side effects of systemic delivery are unknown. Intrathecal viral vector administration may be more invasive than intravenous injections, but it requires far less vector and it can be performed on an outpatient basis, making it an ideal route of delivery for clinical translation. A total of 12 domestic farm pigs (<20?kg) underwent a single-level lumbar laminectomy with intrathecal catheter placement for AAV9 delivery. Animals were perfused and the tissue was harvested 30 days after treatment. Gene expression was assessed by anti-green fluorescent protein immunohistochemistry. Although a single lumbar injection resulted in gene expression limited to the lumbar segment of the spinal cord, three consecutive boluses via a temporary catheter resulted in diffuse transduction of motor neurons (MNs) throughout the cervical, thoracic and lumbar spinal cords. We now present the first successful robust transduction of MNs in the spinal cord of a large animal via intrathecal gene delivery using a self-complementary AAV9. These promising results can be translated to many MN diseases requiring diffuse gene delivery.     

DRG-targeted helper-dependent adenoviruses mediate selective gene delivery for therapeutic rescue of sensory neuronopathies in mice

Dorsal root ganglion (DRG) neuron dysfunction occurs in a variety of sensory neuronopathies for which there are currently no satisfactory treatments. Here we describe the development of a strategy to target therapeutic genes to DRG neurons for the treatment of these disorders. We genetically modified an adenovirus (Ad) to generate a helper virus (HV) that was detargeted for native adenoviral tropism and contained DRG homing peptides in the adenoviral capsid fiber protein; we used this HV to generate DRG-targeted helper-dependent Ad (HDAd). In mice, intrathecal injection of this HDAd produced a 100-fold higher transduction of DRG neurons and a markedly attenuated inflammatory response compared with unmodified HDAd. We also injected HDAd encoding the β subunit of β-hexosaminidase (Hexb) into Hexb-deficient mice, a model of the neuronopathy Sandhoff disease. Delivery of the DRG-targeted HDAd reinstated neuron-specific Hexb production, reversed gangliosidosis, and ameliorated peripheral sensory dysfunction. The development of DRG neuron–targeted HDAd with proven efficacy in a preclinical model may have implications for the treatment of sensory neuronopathies of diverse etiologies.     

Expression of Apolipoprotein A-I in Rabbit Carotid Endothelium Protects Against Atherosclerosis

Expression of atheroprotective genes in the blood vessel wall is potentially an effective means of preventing or reversing atherosclerosis. Development of this approach has been hampered by lack of a suitable gene-transfer vector. We used a helper-dependent adenoviral (HDAd) vector to test whether expression of apolipoprotein A-I (apoA-I) in the artery wall could retard the development of atherosclerosis in hyperlipidemic rabbits. Carotid arteries were infused with an HDAd expressing rabbit apoA-I or a “null” HDAd and harvested 2 and 4 weeks later. ApoA-I mRNA and protein were detected only in HDAdApoAI arteries. Lesion size, lipid and macrophage content, and adhesion molecule expression were similar in both groups at 2 weeks. Between 2 and 4 weeks, most of these measures of atherosclerosis increased in HDAdNull arteries, but were stable or decreased in HDAdApoAI arteries (P ≤ 0.04 for all end points in 4-week HDAdApoAI versus HDAdNull arteries). A longer-term study in chow-fed rabbits revealed persistence of HDAd vector DNA and apoA-I expression for ≥48 weeks, with stable vector DNA content and apoA-I expression from 4 to 48 weeks. Expression of apoA-I in arterial endothelium significantly retards atherosclerosis. HDAd provides prolonged, stable expression of a therapeutic transgene in the artery wall.     

Sustained Enzymatic Correction by rAAV-Mediated Liver Gene Therapy Protects Against Induced Motor Neuropathy in Acute Porphyria Mice

Acute intermittent porphyria (AIP) is characterized by a hereditary deficiency of hepatic porphobilinogen deaminase (PBGD) activity. Clinical features are acute neurovisceral attacks accompanied by overproduction of porphyrin precursors in the liver. Recurrent life-threatening attacks can be cured only by liver transplantation. We developed recombinant adeno-associated virus (rAAV) vectors expressing human PBGD protein driven by a liver-specific promoter to provide sustained protection against induced attacks in a predictive model for AIP. Phenobarbital injections in AIP mice induced porphyrin precursor accumulation, functional block of nerve conduction, and progressive loss of large-caliber axons in the sciatic nerve. Hepatocyte transduction showed no gender variation after rAAV2/8 injection, while rAAV2/5 showed lower transduction efficiency in females than males. Full protection against induced phenobarbital-attacks was achieved in animals showing over 10% of hepatocytes expressing high amounts of PBGD. More importantly, sustained hepatic expression of hPBGD protected against loss of large-caliber axons in the sciatic nerve and disturbances in nerve conduction velocity as induced by recurrent phenobarbital administrations. These data show for the first time that porphyrin precursors generated in the liver interfere with motor function. rAAV2/5-hPBGD vector can be produced in sufficient quantity for an intended gene therapy trial in patients with recurrent life-threatening porphyria attacks.     

CpG-Depleted Plasmid DNA Vectors with Enhanced Safety and Long-Term Gene Expression in Vivo

Systemic delivery of cationic lipid–plasmid DNA (pDNA) complexes induces an acute inflammatory response with adverse hematologic changes and liver damage. Immunostimulatory CpG motifs in the pDNA are known to contribute substantially to this response. Here we constructed a pDNA vector (pGZB) that has been depleted of 80% of the CpG motifs present in the original vector. Compared with the unmodified vector, systemic administration of pGZB induced considerably fewer changes in blood parameters, reduced levels of inflammatory cytokines, and decreased liver damage. Despite the extensive sequence modifications within pGZB, there were still robust levels of transgene expression. Furthermore, in contrast to the transient expression observed from the unmodified vector, we observed sustained or increasing expression for up to 49 days from pGZB in the lung and liver of immunocompetent BALB/c mice. Studies adding CpG motifs in trans or in cis indicate that the reduced CpG content of pGZB was the major determinant for its persistent expression. This combination of decreased toxicity and sustained expression suggests that CpG-depleted pDNA vectors can greatly improve the safety and efficacy of synthetic gene delivery systems.     

Intramuscular injection of an adenoviral vector expressing hepatocyte growth factor facilitates hepatic transduction with a retroviral vector in mice

Retroviral vectors can result in therapeutic and stable levels of expression of proteins from the liver. However, most retroviral vectors transduce only dividing cells, and hepatocytes are normally quiescent. The goal of this study was to determine if an adenoviral vector could transiently express hepatocyte growth factor (HGF) in order to induce hepatocyte replication and facilitate retroviral vector transduction of the liver. Intramuscular injection of an adenoviral vector that expressed human HGF from the cytomegalovirus promoter (Ad.CMV.HGF) resulted in moderate levels of HGF in blood and liver, and replication of 3 to 12% of hepatocytes. No cytopathic effect was observed in the liver, and a control adenoviral vector induced no or lower levels of replication. When a retroviral vector expressing beta-galactosidase cDNA was injected into a peripheral vein during the peak period of hepatocyte replication induced by intramuscularly administered Ad.CMV.HGF, 8% of hepatocytes were transduced. We conclude that intramuscular injection of Ad.CMV.HGF is a safe and effective way to induce transient systemic expression of HGF and hepatocyte replication, and to facilitate transduction of hepatocytes with a retroviral vector.