Chronic Ethanol Increases Adeno-Associated Viral Transgene Expression in Rat Liver via Oxidant and NFκB-Dependent Mechanisms

Chronic Ethanol Increases Adeno-Associated Viral Transgene Expression in Rat Liver via Oxidant and NF␬B-Dependent Mechanisms MICHAEL D. WHEELER,1,2 HIROSHI KONO,1 IVAN RUSYN,1 GAVIN E. ARTEEL,1 DOUGLAS MCCARTY,3 RICHARD JUDE SAMULSKI,2,3 AND RONALD G. THURMAN1,2

Recombinant adeno-associated virus (rAAV) transduction is limited in vivo, yet can be enhanced by hydroxyurea, ultravioletirradiation, or adenovirus coinfection, possibly via mechanisms involving stress in the host cell. Because chronic ethanol induces oxidative stress, it was hypothesized that chronic ethanol would increase rAAV transduction in vivo. To test this hypothesis, rAAV encoding ␤-galactosidase was given to Wistar rats that later received either ethanol diet or high-fat control diet via an enteral-feeding protocol for 3 weeks. Expression and activity of ␤-galactosidase in the liver were increased nearly 5-fold by ethanol. The increase in transgene expression was inhibited by antioxidant diphenylene iodonium (DPI), which is consistent with the hypothesis that ethanol causes an increase in rAAV transduction via oxidative stress. Ethanol increased DNA synthesis only slightly; however, it increased the nuclear transcription factor ␬B (NF␬B) 4-fold, a phenomenon also sensitive to DPI. Moreover, a 6-fold increase in rAAV transgene expression was observed in an acute ischemia-reperfusion model of oxidative stress. Transgene expression was transiently increased 24 hours after ischemia-reperfusion 3 days and 3 weeks after rAAV infection. Further, adenoviral expression of superoxide dismutase or I␬B␣ superrepressor inhibited rAAV transgene expression caused by ischemia-reperfusion. Therefore, it is concluded that ethanol increases rAAV transgene expression via mechanisms dependent on oxidative stress, and NF␬B likely through enhancement of cytomegaloviral (CMV) promoter elements. Alcoholic liver disease is an attractive target for gene therapy because consumption of ethanol could theoretically increase expression of therapeutic genes (e.g., superoxide dismutase). Moreover, this study has important implications for rAAV gene therapy and potential enhancement and regulation of transgene expression in liver. (HEPATOLOGY 2000;32:1050-1059.)

Abbreviations: AAV, adeno-associated virus; rAAV, recombinant adeno-associated virus; NF␬B, nuclear transcription factor-␬B; ip, intraperitoneally; UAC, urine alcohol concentration; EGFP, enhanced A. victorialis green fluorescent protein; CMV, cytomegaloviral; TTR, transthyritin; DTT, dithiothreitol; ONPG, O-nitrophenyl-␤-D-galactopyranoside; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; Ig, immunoglobulin; DPI, diphenylene iodonium. From the 1Laboratory of Hepatobiology and Toxicology, 2Department of Pharmacology, and 3Gene Therapy Center, University of North Carolina at Chapel Hill, Chapel Hill, NC. Received January 25, 2000; accepted August 9, 2000. Address reprint requests to: Michael D. Wheeler, Ph.D., CB# 7365 Mary Ellen Jones Building, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599. E-mail: [email protected]; fax: 919-966-1893. Supported in part by grants from the NIAAA. Copyright © 2000 by the American Association for the Study of Liver Diseases. 0270-9139/00/3205-0023$3.00/0 doi:10.1053/jhep.2000.19339

Adeno-associated virus (AAV)-mediated gene delivery is attractive because it can theoretically provide long-term, stable expression of a transgene.1,2 Recombinant AAV (rAAV) is also a useful tool for liver-directed gene delivery because the virus can infect nondividing cells, integrate specifically into the host cell genome, and is relatively nonpathogenic compared with more commonly used adenoviral vectors.1 Although rAAV provides many advantages over other vectors, its transduction efficiency in many tissues, including the liver, is limited either because of the requirement of synthesis of the second strand of DNA in the viral genome3,4 or because of the lack of transgene expression caused by decreased promoter activity (i.e., inactivation of the cytomegalovirus [CMV] promoter).5 Second-strand synthesis of the single-stranded AAV viral genome, which is required for the expression of the transgene, is hypothesized to be the limiting step in transduction; however, the cellular mechanisms for this process remain unclear. Various genotoxic agents, such as etoposide and hydroxyurea, as well as coinfection with helper viruses such as adenovirus or herpes virus, have been shown to increase rAAV transduction.3,4,6 Moreover, many agents shown to increase rAAV transduction in vitro also cause mild cellular damage and induce many cellular stress responses such as activation of nuclear transcription factor ␬B (NF␬B).7-9 An alternative hypothesis is that rAAV transduction is regulated through cellular proliferation or transcription factors rather than through an increase in cellular DNA synthesis. In fact, recent evidence showed that rAAV transduction efficiency is influenced by several factors including cell type, the proliferative stage of the host cell, or condition of the infected cell or tissue.10,11 Ethanol, a known hepatotoxicant, causes oxidative injury and mild DNA damage.12,13 Therefore, it was hypothesized that chronic enteral ethanol via an intragastric feeding protocol would increase rAAV transduction in vivo. The enteral ethanol-feeding model developed by Tsukamoto et al.14 is an animal model characterized by the development of severe liver pathology, which is characteristic of the human disease. Alcohol-induced liver injury is a chronic condition of severe oxidative stress caused by hypoxia-reoxygenation and free radical production from activated Kupffer cells, the resident hepatic macrophages.15,16 Here, it is reported that chronic ethanol increases transgene expression by rAAV about 4-fold in the liver. Also, the increase in transduction is correlated with an increase in liver injury and activation of NF␬B. These studies suggest that increased rAAV transgene expression can be potentially maximized through modulation of critical host

1050

HEPATOLOGY Vol. 32, No. 5, 2000

cell transcription factors and enhancement of transgene promoter elements. Moreover, these studies support the idea that therapeutic gene transfer approaches to alcoholic liver disease and many other diseases is feasible. MATERIALS AND METHODS Experimental Design. Male Wistar rats (⬃250 –300 g) were used in all experiments, which included 4 treatment groups (high-fat control animals, high-fat control animals infected with rAAV.CMV.lacZ, ethanol-treated animals, and ethanol-treated animals infected with rAAV.CMV.lacZ). Animals received virus intravenously 24 hours before the delivery of enteral diet and were then given ethanol-containing diet or high-fat control diet by continual enteral feeding for 3 weeks. Portal blood was collected for serum transaminase measurements, animals were killed, and livers were harvested. Sections were either snap frozen and stored at ⫺80°C or fixed in 10% neutralbuffered formalin and kept until further analysis. Tsukamoto-French Enteral Ethanol Delivery. Animals. Male Wistar rats (250 –300 g) used in this study were housed in compliance with the American Association for Accreditation of Laboratory Animal Care and institutional guidelines. An intragastric cannula was implanted into each rat as described by Tsukamoto at al.14 Briefly, cannulas were tunneled subcutaneously to the dorsal aspect of the neck and attached to infusion pumps by means of a spring-tether device and swivel, allowing complete mobility of the rats within metabolic cages. Animals were infused continuously with a high-fat liquid diet or a diet containing ethanol through the intragastric cannula for 3 weeks. For ischemia-reperfusion experiments, animals were anesthetized by using sodium pentobarbital (75 mg/kg, intraperitoneally [ip]). Livers were made ischemic for 1 hour by using portal vein clamping in vivo, as described by Colletti et al.17 Diets. The basic liquid diet was prepared according to Thompson and Reitz18 and was supplemented with lipotropes as described previously.19 The diet contained corn oil as fat (37% of total calories), protein (23%), carbohydrate (5%), minerals and vitamins, plus ethanol (35%). For the control high-fat diet, ethanol was replaced with dextrin-maltose. Animals were allowed to recover for 1 week after surgery before enteral ethanol was initiated. Ethanol levels in the diet were gradually increased to 10 to 12 g/kg/d during the first week of ethanol administration and maintained near 12 g/kg/d for the remainder of the experiment. Clinical Chemistry. Urine alcohol concentration (UAC) was determined daily by standard enzymatic alcohol analysis. UAC is reported as the average of daily UAC from samples collected during the ethanol treatment. Serum aspartate and alanine transaminase levels were determined by standard enzymatic assays (Sigma, St. Louis, MO) by using portal blood collected at death. rAAV Preparation. rAAV.CMV.lacZ and rAAV.CMV.egfp contained the reporter transgenes for bacterial ␤-galactosidase and enhanced A. victorialis green fluorescent protein (EGFP) under control of the CMV promoter and were prepared in human HEK 293 cells as described previously.20 For some experiments, the CMV promoter sequence in the rAAV.CMV.egfp vector was replaced with the transthyritin (TTR) gene promoter elements21 by using standard cloning procedures, and the rAAV.TTR.egfp vector was prepared as described earlier. The Ad vector Ad.CMV.EGFP contains the transgene for the EGFP reporter. The Ad.SOD1 (a kind gift from Dr. John Engelhardt, University of Iowa) and Ad.I␬B␣ adenoviral vectors contain the transgenes for human Cu/Zn-superoxide dismutase and the NF␬B superrepressor I␬B␣, respectively. The adenoviral vectors were prepared by the University of North Carolina Viral Vector Core as described previously.22,23 Detection of Transgene. ␤-galactosidase histochemistry. Formalin-fixed sections (8 ␮m) were stained for ␤-galactosidase. Briefly, fixed sections were incubated in phosphate-buffered saline containing 5 mmol/L potassium ferricyanide, 5 mmol/L potassium ferrocyanide, 2 mmol/L magnesium chloride, 0.02% Nonidet P-40 (wt/vol), 0.01% sodium deoxycholate (wt/vol), and 1 mg/mL 5-bromo-4-chloro-3-indolyl-␤-D-galactopyrano-

WHEELER ET AL.

1051

side (X-gal) (Sigma) for 24 to 48 hours at 37°C. Sections were then lightly counterstained with eosin. Biochemical quantitation of ␤-galactosidase activity. Whole liver tissue was homogenized in buffer A (40 mmol/L Tris, 140 mmol/L NaCl, 1 mmol/L dithiothreitol [DTT], 10 mg/mL PMSF, 1 mg/mL aprotinin, and 1 mg/mL leupeptin) and centrifuged at 900g for 10 minutes. Supernatant was diluted in buffer A in a final volume of 150 ␮L. A 150-␮L aliquot of assay buffer B (120 mmol/L Na2HPO4, 80 mmol/L NaH2PO4, 2 mmol/L MgCl2, 100 mmol/L 2-mercaptoethanol, and 1.33 mg/mL of O-nitrophenyl-␤-D-galactopyranoside [ONPG]) was added to the extract and incubated for 30 minutes at room temperature. To stop the reaction, 500 ␮L of a 2.8% sodium carbonate solution was added. The activity ␤-galactosidase was quantified by using the cleavage of ONPG to nitrophenyl that was measured spectrophotometrically at 420 nm and normalized to total protein determined with the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA). Western blot. Whole liver tissue was homogenized in buffer A (40 mmol/L Tris, 140 mmol/L NaCl, 1 mmol/L DTT, 10 mg/mL PMSF, 1 mg/mL aprotinin, and 1 mg/mL leupeptin). The homogenate was centrifuged at 900g, the supernatant was collected. Protein (10 ␮g) was diluted in buffer A to a final volume of 15 ␮L. A 15-␮L aliquot of 2⫻ Laemelli buffer was added to the sample, which was heated at 95°C for 5 minutes before loading on the gel. Samples were resolved by electrophoresis by using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Protein was transferred onto nitrocellulose by using a semidry transfer method. The membranes were incubated in a 1:2,000 dilution of mouse monoclonal anti–␤galactosidase antibody (Boehringer Mannheim, Mannheim, Germany) or a 1:1,000 dilution of rabbit polyclonal anti– green fluorescent protein (GFP) antibody (Clontech, Palo Alto, CA) for 2 hours at room temperature. Then membranes were transferred to a solution containing a 1:5,000 dilution of secondary anti-mouse or anti-rabbit immunoglobulin (Ig)G– horseradish peroxidase (HRP)- conjugated antibody (Amersham Life Sciences, Buckinghamshire, England) and incubated for 1 hour. The membranes were washed 3 times for 10 minutes in fresh triethanolamine-buffered saline/0.05% Tween-20 between each step. Finally, the membranes were washed 3 times for 15 minutes and added to ECL Western Detection Reagent (Amersham Life Sciences). Immunohistochemical staining for GFP. Formalin-fixed, paraffin-embedded sections (8 ␮m) were mounted on glass slides. Sections were deparaffinized and rehydrated and then stained with mouse antiGFP primary antibody (Clontech) for 30 minutes. The immunostaining was visualized by using the DAKO (Carpinteria, CA) immunostaining kit. Slides were counterstained with hematoxylin. Primary antibody dilutions were 1:100 in phosphate-buffered saline containing 1% Tween-20. Electromobility Shift Assay (EMSA) for NF␬B. Nuclear extracts were prepared from whole liver homogenate as previously described by Dignam et al.24 by lysing cells in Nonidet P-40 (0.25%) in 10 mmol/L HEPES, pH 7.6, containing 10 mmol/L KCl, 2 mmol/L MgCl2, 1 mmol/L DTT, 0.1 mmol/L EDTA, and small protease inhibitors. Homogenate was centrifuged at 13,000g for 10 minutes and the nuclear pellet was resuspended in 50 mL of 50 mmol/L HEPES, pH 7.6, containing 50 mmol/L KCl, 300 mmol/L NaCl, 1 mmol/L DTT, 0.1 mmol/L EDTA, small protease inhibitors, and 10% glycerol. After gentle mixing for 15 minutes, the mixture was centrifuged at 13,000g for 7 minutes. The supernatant was collected and stored at ⫺80°C. Electromobility shift assays of nuclear extracts (40 ␮g) were performed with radiolabeled oligonucleotide probes specific for NF␬B.25,26 The probe was labeled at the 5⬘-end by using T4 kinase and [␥-32P] adenosine triphosphate (3,000 Ci/mmol). Excess adenosine triphosphate was separated from the labeled probe through sequential ethanol precipitation. Protein-DNA complexes were separated through native 5% PAGE and visualized by autoradiography. The DNA–protein complex was supershifted by using antibodies specific for the p50 and p65 subunits (a kind gift from N. Rice, National Cancer Institute) of the active NF␬B complex. The densi-

1052 WHEELER ET AL.

HEPATOLOGY November 2000

TABLE 1. Summary of Treatment Groups, UAC, Serum Transaminase Levels, and Pathology Score Treatment

n

UAC

AST

ALT

Total Pathology

Control Control ⫹ AAV Ethanol Ethanol ⫹ AAV

4 4 6 6

Not determined Not determined 184 ⫾ 10 175 ⫾ 9

52 ⫾ 11 36 ⫾ 9 181 ⫾ 19* 210 ⫾ 23*

26 ⫾ 2 23 ⫾ 1 63 ⫾ 7* 74 ⫾ 3*

0.4 ⫾ 0.2 0.6 ⫾ 0.3 3.8 ⫾ 0.2* 4.2 ⫾ 0.3*

NOTE. Urine was collected from animals in all treatment groups and average UAC was determined as described in the Materials and Methods section. Serum aspartate transaminase (AST) and alanine transaminase (ALT) levels were measured by standard enzymatic procedures from blood collected from the inferior vena cava before necropsy. Pathology was scored on the basis of steatosis (0-4), inflammation (0-2), and necrosis (0-2). Data are expressed as mean ⫾ SD and are representative of 4 individual experiments. * P ⬍ .05 compared with high-fat controls, 2-way ANOVA with Tukey’s post-hoc analysis.

tometric quantification of NF␬B binding was performed by using Scion Image 1.62 software. RESULTS Routine Parameters. Ethanol was given in a high-fat diet at 12 g/kg/d via a gastric feeding tube by using an enteral feeding model. This ethanol-feeding model is characterized by severe steatosis, mild necrosis, and lymphocytic infiltration and represents the human disease more closely than other rodent models. Here, ethanol-treated animals exhibited an increase in UAC that cycled between 60 and 250 mg/dL and maintained an average UAC near 180 mg/dL (Table 1). The cyclic pattern in UAC is typical, but reasons for it are not understood. Importantly, the delivery of rAAV.CMV.lacZ had no effect on the concentration of alcohol or the cyclic pattern in the urine in this model. Serum aspartate transaminase and alanine transaminase levels were not increased in rats infected with recombinant AAV and values were not different from enzyme levels of normal animals (data not shown), indicating that rAAV by itself does not contribute to liver injury. As expected, however, transaminases were increased significantly about 4-fold in ethanol-treated animals compared with high-fat controls (Table 1). In animals that received control high-fat diet, minimal pathologic changes were observed regardless of whether or not virus was given (Fig. 1A and 1B). However, histologic analysis of ethanol-treated animals showed marked increases in panlobular fat accumulation, lymphocyte infiltration, and necrosis as expected (Fig. 1C and 1D). Chronic Ethanol Increases rAAV Transgene Expression. To test the hypothesis that ethanol would increase rAAV transduction in the liver, Wistar rats receiving rAAV encoding the bacterial reporter gene ␤-galactosidase were given ethanol via an intragastric enteral feeding protocol for 3 weeks. rAAV (1 ⫻ 1010 particles) was given intravenously to animals 24 hours before initiation of ethanol diet. After 3 weeks of ethanol delivery, the expression of ␤-galactosidase was measured by 5-bromo-4-chloro-3-indolyl-␤-D-galactoside (X-gal) staining (Fig. 2) and quantified by Western blotting by using a polyclonal antibody specific for ␤-galactosidase (Fig. 3A). X-gal staining showed ␤-galactosidase expression in periportal regions of livers of animals infected with rAAV.CMV.lacZ. Moreover, ethanol caused a significant increase in ␤-galacto-

sidase expression compared with expression observed in livers of animals fed a high-fat control diet. ␤-Galactosidase expression, measured by Western blot analysis, was increased nearly 4-fold by chronic ethanol compared with rats that received high-fat control diet (Fig. 3B). ␤-Galactosidase expression was not observed in saline-injected control animals that received either high-fat control diet or diet containing ethanol, as expected. The activity of ␤-galactosidase was measured by the enzymatic cleavage of the substrate ONPG (Fig. 3C). In rats given rAAV.CMV.lacZ and treated with ethanol for 3 weeks, ␤-galactosidase activity was increased nearly 4-fold over control animals treated with rAAV and high-fat diet alone. The increase in ␤-galactosidase activity is consistent with the increase in transgene expression detected by Western blotting, indicating that the increase in transgene activity is a direct result of rAAV transduction and expression of a functional transgene. Thus, it is concluded that rAAV transgene expression is increased dramatically by chronic ethanol. rAAV Transgene Expression Caused by Ethanol Is Dependent on Oxidative Stress. To test the hypothesis that the increase in

rAAV transduction was because of oxidative stress caused by ethanol, the antioxidant and nicotinamide-adenine dinucleotide phosphate oxidase inhibitor diphenylene iodonium (DPI) was studied. DPI has been shown to block liver injury

FIG. 1. Representative photomicrographs of livers after 3 weeks of ethanol or high-fat diet in rAAV.CMV.lacZ (1010 particles) or saline-treated rats. Histology of liver from saline- and rAAV.CMV.lacZ– treated high-fat control after 3 weeks is represented in A and B, respectively. Rat liver that received only saline and 3 weeks of enteral ethanol is represented in C. (D) Rats treated with rAAV.CMV.lacZ and enteral ethanol for 3 weeks. Original magnification, ⫻200. Typical experiments.

HEPATOLOGY Vol. 32, No. 5, 2000

WHEELER ET AL.

1053

FIG. 2. Representative photomicrographs of histochemical staining for ␤-galactosidase. Formalin-fixed sections (20 ␮m) were stained histochemically by using X-gal as described in the Materials and Methods section. Rats were given rAAV.CMV. lacZ (1010 particles) and fed either high-fat control diet (A and C) or ethanol-containing diet (B and D) for 3 weeks. (C and D) High-power magnification (⫻200). Typical experiments.

caused by ethanol given via the Tsukamoto-French protocol, suggesting that alcohol-induced liver injury is dependent on oxidative stress.27 Animals were given rAAV.CMV.lacZ (1 ⫻ 1010 particles), then ethanol, and treated daily with DPI (1 mg/kg, subcutaneously) or 5% glucose vehicle for 3 weeks, and ␤-galactosidase transgene expression was measured biochemically as described earlier. In this study, DPI treatment nearly completely prevented ethanol-induced pathologic changes as well as increases in serum transaminases, confirming earlier work (data not shown). Importantly, ethanol caused a 6-fold increase in transgene expression in animals that received vehicle (Fig. 4). However, treatment with DPI prevented the increase in ␤-galactosidase expression caused by ethanol completely. Moreover, when histochemical analy-

sis of transgene expression was performed by using X-gal, the increase in ␤-galactosidase staining caused by ethanol was attenuated by DPI treatment (data not shown). Importantly, because DPI is a potent antioxidant, these data support the hypothesis that ethanol-induced transgene expression is mediated via oxidative stress.

Role of NF␬B in Chronic Ethanol-Induced rAAV Transgene Expression. Ethanol is a potent inducer of oxidative stress and it

is well known that oxidants activate specific cellular stress responses such as NF␬B.28 Because rAAV transgene expression is increased by chronic ethanol and is sensitive to antioxidants, the question of whether NF␬B activation is involved in ethanol-induced rAAV transduction was evaluated. NF␬B was measured in nuclear extracts by electromobility shift as-

1054 WHEELER ET AL.

HEPATOLOGY November 2000

FIG. 4. DPI inhibits ethanol-induced rAAV transgene expression. Male Wistar rats were injected with rAAV.CMV.lacZ (1010 particles) before receiving an intragastric diet containing dextrin-maltose (Con) or ethanol (EtOH) via the Tsukamoto-French protocol for 3 weeks. Animals were also treated daily subcutaneously with either 5% glucose vehicle control (Vehicle) or with 1 mg/kg DPI. Liver extracts were prepared and ␤-galactosidase activity was measured by the enzymatic conversion of the colorimetric substrate ONPG to nitrophenyl. Data are expressed as mean ⫾ SEM and are representative of 4 individual experiments. (aP ⬍ .05, compared with vehicle-treated controls; bP ⬍ .05, compared with vehicle-treated animals that received ethanol, 2-way ANOVA with Tukey’s post-hoc analysis.) FIG. 3. Expression and enzymatic quantitation of ␤-galactosidase. (A) Liver extracts from animals injected with either saline or rAAV.CMV.lacZ (1010 particles) 24 hours before receiving control diet or diet containing ethanol for 3 weeks were separated by 10% SDS-PAGE and immunoblotted with an anti-␤-galactosidase monoclonal antibody. The data are representative of 3 or more experiments. (B) Image densitometry was performed to quantitate ␤-galactosidase expression determined by Western blot analysis. Data are expressed as mean ⫾ SD in arbitrary densitometric units and are representative of 3 individual experiments per group. (*P ⬍ .05, 2-way ANOVA with Tukey’s post-hoc analysis). (C) Liver extracts from animals injected with either saline or rAAV.CMV.lacZ (1010 particles) 24 hours before receiving control diet or diet containing ethanol for 3 weeks were prepared, and ␤-galactosidase activity was measured by the enzymatic conversion of the colorimetric substrate ONPG to nitrophenyl. Data are expressed as mean ⫾ SEM and are representative of 4 individual experiments. (*P ⬍ .05, 2-way ANOVA with Tukey’s post-hoc analysis).

says. Active NF␬B– oligonucleotide complex was supershifted by using antibodies against the p50 and p65 subunits of the NF␬B heterodimer, but not with nonspecific IgG antibody, and was competitively inhibited by using excess unlabeled oligonucleotide (Fig. 5A). These data confirm the specificity of NF␬B binding in these experiments and indicate that the NF␬B complex consists of the p50 and p65 subunits. There was no activation of NF␬B after 3 weeks of high-fat control diet in either saline- or rAAV.CMV.lacZ-treated animals. However, in animals treated with ethanol, NF␬B binding was increased significantly nearly 4-fold, which is consistent with other reports (Fig. 5C).29 Moreover, treatment with the antioxidant DPI blunted the activation of NF␬B (Fig. 5C).27 These data suggest that NF␬B may be responsible for transgene expression. Because rAAV transgene expression is driven by the CMV promoter, and the CMV promoter can be regulated by

HEPATOLOGY Vol. 32, No. 5, 2000

WHEELER ET AL.

1055

NF␬B and several other stress activated factors,5,30 it is reasonable to hypothesize that ethanol increases transgene expression through activation of NF␬B (see later). Acute Oxidative Stress Induces rAAV Transgene Expression.

Based on the data mentioned earlier, it is hypothesized that oxidative stress activates NF␬B, which drives transgene expression. To test this hypothesis, rAAV transgene expression and the role of NF␬B was evaluated by using a warm ischemiareperfusion model of oxidative stress. This model is used because it is an acute in vivo model in which oxidative stress and activation of NF␬B can be induced rapidly (⬃1 hour) and performed repeatedly, procedures that would be impractical in the enteral ethanol model. Ischemia reperfusion caused a 5-fold increase in ␤-galactosidase activity in the livers of animals given rAAV.CMV.lacZ (1010 particles) compared with sham-operated controls that received recombinant virus (Fig. 6A). Moreover, transgene expression induced by ischemia reperfusion was also sensitive to antioxidant treatment. These data show that oxidative stress induced by ischemia reperfusion increases rAAV transgene expression such as ethanol treatment. Oxidative Stress-Induced rAAV Transgene Expression Is Transient. In animals that received rAAV.CMV.lacZ and under-

went ischemia-reperfusion 3 days after infection, there was a 4-fold increase in ␤-galactosidase measured 24 hours after reperfusion, as in the experiments detailed earlier (Fig. 6B). However, ␤-galactosidase activity returned to control levels 3 weeks after ischemia reperfusion. Ischemia reperfusion 3 weeks after infection also transiently increased ␤-galactosidase activity by nearly 5-fold, and transgene expression, which also returned to control values in 3 weeks. Increases in rAAV Transgene Expression by Oxidative Stress Are Inhibited by Superoxide Dismutase and I␬B␣. A definitive role of

FIG. 5. Activation of NF␬B by ethanol. EMSA was performed as described earlier in the Materials and Methods section. (A) Nuclear extracts (40 mg) from animals that received ethanol were incubated with radiolabeled probe (lane 2), in the presence of nonspecific IgG antibody (lane 3), p50 antibody (lane 4), p65 antibody (lane 5), or excess unlabeled probe (lane 6). Lane 1 is radiolabeled probe alone. (B) Nuclear extract from animals treated with 5% glucose vehicle (lanes 1-4) or diphenylene iodonium (lanes 5-6) and given either high-fat control diet (lanes 1, 3, and 5) or ethanol diet (lanes 2, 4, and 6) and injected with either saline (lanes 1-2) or rAAV.CMV.lacZ (lanes 3-6) were evaluated. Data are representative of 3 or more individual experiments. (C) Image analysis was performed on 4 individual experiments in each group. (aP ⬍ .05, compared with high-fat control; bP ⬍ .05, compared with rAAV.CMV.lacZ, ethanol-treated samples, 2-way ANOVA followed up by Tukey’s post-hoc analysis.)

NF␬B in rAAV transduction has been lacking despite many studies.31 Selective inhibitors of NF␬B are limited and many of the in vitro models used to investigate NF␬B involve immortalized cell lines that have alterations in many cellular regulatory mechanisms including NF␬B. Thus, the role of NF␬B in rAAV transduction was evaluated here in vivo. Animals were infected with rAAV.CMV.lacZ (1010 particles) as well as with recombinant adenovirus containing the transgene for either enhanced green fluorescent protein (Ad.EGFP), human superoxide dismutase (Ad.SOD1), or I␬B␣ superrepressor (Ad.I␬B). Three days after infections, animals underwent ischemia reperfusion, and NF␬B activation and ␤-galactosidase activity were evaluated 2 and 24 hours later, respectively (Fig. 7A and 7B). NF␬B was activated after ischemia reperfusion in both saline and Ad.EGFP-treated animals, as expected (Fig. 7A). However, NF␬B was not activated after ischemia reperfusion in animals treated with either Ad.SOD1 or Ad.I␬B␣. Animals given rAAV.CMV.lacZ coinfected with Ad.EGFP exhibited a 6-fold increase in ␤-galactosidase activity 24 hours after ischemia reperfusion compared with shamoperated animals. In contrast, ␤-galactosidase activity was not increased in livers from animals coinfected with Ad.SOD1 or Ad.IkB. These data provide strong evidence that oxidative stress increases rAAV transgene expression in vivo. Moreover, these data are consistent with the hypothesis that NF␬B activation is required for oxidative stress-induced increases in rAAV transgene expression.

1056 WHEELER ET AL.

HEPATOLOGY November 2000

FIG. 7. Inhibition of NF␬B activation blocks rAAV transgene expression. Animals were given rAAV.CMV.lacZ (1 ⫻ 1010 particles) and recombinant adenovirus (1 ⫻ 109 pfu) containing either enhanced green fluorescent protein (Ad.EGFP), human cytosolic superoxide dismutase (Ad.SOD1), or dominant negative I␬B␣ (Ad.I␬B␣) 3 days before warm ischemia-reperfusion. Control animals underwent sham operation. (A) NF␬B was measured by electromobility shift assay. (B) ␤-galactosidase activity was measured 24 hours after ischemia-reperfusion as described in the Materials and Methods section. Data are expressed as mean ⫾ SEM and are representative of 4 individual experiments. (aP ⬍ .05 compared with sham controls; bP ⬍ .05 compared with Ad.EGFP treated animals, ANOVA with Tukey’s post-hoc analysis.)

The Role of the CMV Promoter in Oxidative Stress–Induced Increase in rAAV Transgene Expression. The CMV promoter con-

FIG. 6. Ischemia-reperfusion injury transiently induces rAAV transgene expression. (A) Animals were given diphenyliodonium (1 mg/kg, ip, daily ) or vehicle for 5 days and rAAV.CMV.lacZ (1010 particles) 24 hours before warm ischemia-reperfusion. Control animals underwent sham operation. ␤-galactosidase activity was measured 24 hours after ischemia-reperfusion as described in the Materials and Methods section. (B) Animals underwent warm ischemia either 24 hours or 3 weeks after injection of rAAV.CMV.lacZ (1010 particles). ␤-galactosidase activity was measured either 24 hours or 3 weeks after reperfusion. Data are expressed as mean ⫾ SEM and are representative of 4 individual experiments. (*P ⬍ .05, ANOVA with Tukey’s post-hoc analysis.)

tains several putative NF␬B-binding regions.5,32 Thus, to test the hypothesis that NF␬B-dependent enhancement of rAAV transgene expression caused by oxidative stress was caused by activation of the CMV promoter, studies were performed by using rAAV containing the transgene for EGFP under the control of the promoter region for the TTR gene (rAAV. TTR.egfp). The TTR promoter provides liver-specific gene expression and is devoid of NF␬B-responsive elements.21 Animals were infected with rAAV.CMV.egfp or rAAV.TTR.egfp (2 ⫻ 109 particles/animal) for 3 days and then underwent ischemia reperfusion or sham operation. Livers were harvested 24 hours later and evaluated for GFP transgene expression by Western blot and immunohistochemical staining (data not shown). In animals infected with rAAV.CMV.egfp, ischemia reperfusion increased GFP expression nearly 4-fold compared with sham-operated control animals, as expected. However, in rAAV.TTR.egfp-infected animals, transgene expression was not significantly increased after ischemia reperfusion. These data are shown with both Western analysis and

HEPATOLOGY Vol. 32, No. 5, 2000

immunohistochemistry. Moreover, these data indeed support the hypothesis that enhancement of rAAV transgene expression mediated most likely through activation of the CMV promoter. DISCUSSION Chronic Ethanol Induces rAAV Transgene Expression Through Oxidative Stress. The delivery of genes to the liver via AAV

provides many advantages: (1) it has a broad host cell range, (2) it is not immunogenic or pathogenic, (3) it provides longterm expression, and (4) it has the ability to transduce nondividing cells.1 Although these advantages make AAV an attractive gene delivery system for the liver, rAAV transduction in the liver is inefficient, transducing only approximately 3% to 5 % of the cells.33 Many agents and methods used to increase rAAV transduction and transgene expression also cause oxidative stress in vivo. Indeed, it was recently shown that oxidative stress in vitro increased rAAV transduction in immortalized cell lines.34 Ethanol, which causes oxidative stress in the liver,15 also induces rAAV transgene expression in vivo (Figs. 2 and 4).16 Interestingly, the increase in rAAV transduction is increased because of ethanol in periportal regions of the liver. There may be 2 possible explanations for this finding. First, zonal distribution of rAAV transgene expression may be a reflection of the cells that are primarily infected, suggesting that periportal hepatocytes are more permissive to rAAV. Another possibility is that ethanol-induced oxidative stress is largely mediated by activated Kupffer cells and infiltrating neutrophils, which produce superoxide largely in periportal regions. Although the mechanism for alcohol toxicity to the liver is still debated, the production of oxidants caused by ethanol clearly occurs.35 The role of oxidant production in the development of liver injury is also debatable, but it is known that oxidants activate NF␬B28,36 and subsequently increase expression of TNF␣, a critical inflammatory cytokine involved in alcohol-induced liver injury.37 In the Tsukamoto-French model, ethanol increases electron spin resonance– detectable radical production, providing direct evidence of oxidative stress.15 Moreover, oxidative stress has recently been shown to induce rAAV transduction in cell culture.31 Because treatment with the antioxidant DPI inhibited transgene expression caused by ethanol (Fig. 4), it is concluded that ethanol increases rAAV transgene expression through oxidative stressdependent mechanisms. Ethanol Enhancement of rAAV Transduction Is Not Likely Mediated Through an Increase in DNA Synthesis. Ethanol in several

models has been shown to cause mild DNA damage, possibly leading to the induction of repair mechanisms, such as nucleotide excision repair.13 However, little evidence supports the hypothesis that ethanol increases DNA synthesis. In fact, it is reported that ethanol attenuates the proliferative effects of many agents.38,39 Because it has been reported that the synthesis of the second strand of viral DNA is the limiting step in the transduction of rAAV,3 and that cellular enzymes such as PCNA may facilitate rAAV transduction,8,26 it seemed likely that ethanol increased rAAV transduction via mechanisms dependent on DNA synthesis. However, in this model, there is only a modest increase in cellular DNA synthesis that is most likely caused by viral infection and not caused by ethanol (data not shown). Moreover, the high rate of normal liver cell proliferation may be sufficient to support rAAV double-strand conversion. Evidence supporting this hypothesis is that rAAV

WHEELER ET AL.

1057

genomes nearly completely become double stranded over a period of 3 to 4 weeks in liver cells in vivo.40 Thus, the rAAV genome becomes double stranded in the absence of a transducing agent within the time frame used in these experiments, suggesting that second-strand conversion occurred in the absence of ethanol (see later). In the acute warm ischemia-reperfusion model of oxidative stress, transgene expression was transiently increased over control levels in a fashion similar to ethanol (Fig. 6A). The warm ischemia-reperfusion model was used here because it is practical and rapid. Hence, oxidative stress can be induced quickly compared with the chronic ethanol-feeding model. This model was used to address the hypothesis that oxidative stress does not influence transduction by increasing singlestrand conversion of the rAAV genome, but rather causes transient changes in transgene transcription. Indeed, transgene expression was transiently elevated 5-fold within 24 hours after reperfusion compared with sham-operated controls that received rAAV (Fig. 6B); however, it returned to control levels after 3 weeks. If ethanol caused an increase in cellular DNA synthesis leading to single-stranded conversion, transgene expression would be expected to remain elevated. Moreover, because expression of ␤-galactosidase is transiently induced by ischemia reperfusion, it is concluded that oxidative stress acts at the level of transcription, not secondstrand conversion. However, it has recently been shown that enhancement of second-strand synthesis may not necessarily contribute to sustained transgene expression and that mechanisms involved in transient expression may be different from those leading to persistent transgene expression from AAV vectors.41 Oxidative Stress–Induced rAAV Transgene Expression Involves NF␬B. Chronic ethanol causes activation of NF␬B via the pro-

duction of oxidants.27 Here, it is reported that NF␬B activation is correlated with ethanol-induced transgene expression (Fig. 5). Whether or not NF␬B activation is related to transgene expression is difficult to address experimentally because many inhibitors of NF␬B are either nonselective or have antioxidant properties, but the involvement of NF␬B is likely because it is activated by oxidants.36 Although NF␬B has also been shown to influence several factors involved in the cell cycle,42 it seems more likely that NF␬B activates transgene transcription. The role of NF␬B in rAAV transduction has been evaluated in several in vitro models, yet, its precise role remains unclear. Recently, it was shown in cell culture that expression of dominant-negative I␬B␣ blocked NF␬B activation, but had no effect on rAAV transduction (i.e., secondstrand DNA synthesis).31 However, many studies have remarkably different outcomes in vitro compared with cell culture and in vivo because immortalized cells often have compensatory signaling mechanisms and altered regulation of the cell cycle. For example, by using an in vivo model of oxidative stress such as the ischemia-reperfusion model, the increase in transgene expression was inhibited by I␬B␣ repressor protein (Fig. 7). Thus, it is concluded that NF␬B plays a critical role in rAAV transduction and transgene expression in vivo. It is also of interest that recombinant adenoviral infection did not significantly increase basal rAAV transduction because E4⫹ adenovirus provides helper function for rAAV transduction. However, the dose of recombinant adenovirus in these studies is most likely too low to significantly enhance rAAV transduction in vivo.

1058 WHEELER ET AL.

HEPATOLOGY November 2000

NF␬B is also known to contribute to CMV promoter activity in several cell types most likely by binding to 3 NF␬B consensus binding sites within the CMV promoter region.5,32 The transient increase in transgene expression after ischemia reperfusion suggest that the observed increase may be caused by transcriptional regulation (Fig. 6). Moreover, it was recently reported that rAAV transgene expression could be induced repeatedly by lipopolysaccharide administration in an arthritis model in the rat.11 These data clearly suggest that disease state or oxidative stress (i.e., by lipopolysaccharide or ethanol) increases rAAV transgene expression likely through NF␬B or other stress response transcription factors. The inhibition of NF␬B in vivo by using both antioxidant superoxide dismutase and dominant-negative I␬B␣ supports the hypothesis that oxidative stress increases NF␬B, leading to rAAV transgene expression (Fig. 7). Whether or not enhancement is completely caused by activation of the CMV promoter by NF␬B or some general feature of rAAV biology is not fully understood; however, recent data suggests that transgene enhancement caused by oxidative stress may be dependent on activation of stress-induced promoter elements (data not shown). Therapeutic Implications. The concept that rAAV may become useful in targeting liver diseases becomes more attractive in light of data that suggests that oxidative stress in the liver may be sufficient to increase transgene expression. rAAV typically transduces approximately 3% to 5% of the liver33 and provides low levels of transgene expression compared with other viral vectors. However, here, transgene expression from rAAV can be elevated to include nearly approximately 20% to 25% of the liver. As mentioned earlier, the use of ethanol or oxidative stress to increase rAAV transduction has important implications in developing techniques to increase transgene expression. Because oxidative stress has been reported to increase rAAV transduction,34 and here it is shown that oxidative stress can improve rAAV transgene expression, these findings may represent a new tool to improve gene delivery to liver with rAAV. Additionally, inducers of oxidative stress such as acute ethanol, endotoxin, or brief periods of hypoxia caused by vessel clamping (e.g., as performed in routine abdominal surgery) may also be useful to increase the efficiency of rAAV. The finding that transgene up-regulation is dependent on transcriptional activation of the CMV promoter suggests that this promoter may be useful for disease-state regulation of the transgene, especially for diseases involving oxidative stress such as alcohol-induced liver injury. Because alcohol-induced liver injury and cirrhosis is on a steady rise and therapies are lacking, a gene therapy approach is attractive. For example, rAAV-encoding antioxidant enzymes, such as superoxide dismutase and catalase, soluble TNF receptor, or dominant-negative TGF␤ receptor may be useful in preventing or treating alcoholic liver disease. Furthermore, because oxidative stress increases rAAV transgene expression, rAAV becomes even more attractive as a therapeutic tool for many conditions such as alcoholic liver disease, organ transplantation, or chemical toxicity where oxidative stress occurs. Moreover, oxidative stress induced by the disease state can, in fact, regulate the expression of therapeutic transgenes. REFERENCES 1. Rabinowitz JE, Samulski RJ. Adeno-associated virus expression system for gene transfer. Curr Opin Biotechnol 1998;9:470-475.

2. Xiao W, Berta SC, Lu MM, Moscioni AD, Tazelaar J, Wilson JM. Adenoassociated virus as a vector for liver-directed gene therapy. J Virol 1998; 72:10222-10226. 3. Ferrari FK, Samulski T, Shenk T, Samulski RJ. Second-strand synthesis is a rate-limiting step for efficient transduction by recombinant adeno-associated virus vectors. J Virol 1996;70:3227-3234. 4. Fisher KJ, Gao G-P, Weitzman MD, DeMatteo R, Burda JF, Wilson JM. Transduction with recombinant adeno-associated virus for gene therapy is limited by leading-strand synthesis. J Virol 1996;70:520-532. 5. Lo¨ser P, Jennings GS, Strauss M, Sandig V. Reactivation of the previously silenced cytomegalovirus major immediate-early promoter in the mouse liver: involvement of NF␬B. J Virol 1998;72:180-190. 6. Alexander IE, Russell DW, Miller AD. DNA-damaging agents greatly increase the transduction of nondividing cells by adeno-associated virus vectors. J Virol 1994;68:8282-8287. 7. Russell DW, Miller AD, Alexander IE. Adeno-associated virus vectors preferentially transduce cells in S phase. Proc Natl Acad Sci U S A 1994; 91:8915-8919. 8. Hu T-H, McDonald WF, Zolotukhin I, Melendy T, Waga S, Stillman B, Muzyczka N, et al. Cellular proteins required for adeno-associated virus DNA replication in the absence of adenovirus coinfection. J Virol 1998; 72:2777-2787. 9. Clesham GL, Adam PJ, Proudfoot D, Flynn PD, Efstathiou S, Weissberg PL. High adenoviral loads stimulate NF␬B-dependent gene expression in human vascular smooth muscle cells. Gene Ther 1998;5:174-180. 10. Teramoto S, Bartlett JS, McCarty D, Xiao X, Samulski RJ, Boucher RC. Factors influencing adeno-associated virus-mediated gene transfer to human cystic fibrosis airway epithelial cells: comparison with adenovirus vectors. J Virol 1998;72:8904-8912. 11. Pan RY, Xiao X, Chen SL, Li J, Lin LC, Wang HJ, Tsao YP, et al. Diseaseinducible transgene expression from a recombinant adeno-associated virus vector in a rat arthritis model. J Virol 1999;73:3410-3417. 12. Leiber CS. Biochemical and molecular basis of alcohol-induced injury to liver and other tissues. N Engl J Med 1988;319:1639-1650. 13. Brooks PJ. DNA damage, DNA repair, and alcohol toxicity-a review. Alcohol Clin Exp Res 1997;21:1073-1082. 14. Tsukamoto H, Reiderberger RD, French SW, Largman C. Long-term cannulation model for blood sampling and intragastric infusion in the rat. Am J Physiol 1984;247:R595-R599. 15. Knecht KT, Adachi Y, Bradford BU, Iimuro Y, Kadiiska M, Qun-Hui X, Thurman RG, et al. Free radical adducts in the bile of rats treated chronically with intragastric alcohol: inhibition by destruction of Kupffer cells. Mol Pharmacol 1995;47:1028-1034. 16. Knecht KT, Bradford BU, Mason RP, Thurman RG. In vivo formation of a free radical metabolite of ethanol. Mol Pharmacol 1990;38:26-30. 17. Colletti LM, Remick DG, Burtch GD, Kunkel SL, Strieter RM, Campbell DA. The role of tumor necrosis factor alpha in the pathophysiologic alterations following hepatic ischemia/reperfusion injury. J Clin Invest 1990;85:1936-1943. 18. Thompson JA, Reitz RC. Effects of ethanol ingestion and dietary fat levels on mitochondrial lipids in male and female rats. Lipid 1978;13:540-550. 19. Morimoto M, Zern MA, Hagbjork AL, Ingelman-Sundberg M, French SW. Fish oil, alcohol, and liver pathology: role of cytochrome P4502E1. Proc Soc Exp Biol Med 1994;207:197-205. 20. Samulski RJ, Chang L-S, Shenk T. Helper-free stocks of recombinant adeno-associated viruses: normal integration does not require viral gene expression. J Virol 1989;63:3822-3828. 21. Yan CC, Costa RH, Darnell JE, Chen J, Van Dyke TA. Distinct positive and negative elements control the limited hepatocyte and choroid plexus expression of transthyritin in transgenic mice. EMBO J 1990;9:869-878. 22. Engelhardt JF, Yang Y, Stratford-Perricaudet LD, Allen ED, Kozarsky K, Perricaudet M, Yankaskas JR, et al. Direct gene transfer of human CFTR into human bronchial epithelia of xenografts with E1-deleted adenoviruses. Nat Genet 1993;4:27-34. 23. Iimuro Y, Nishiura T, Hellerbrand C, Behrns KE, Schoonhoven R, Grisham JW, Brenner DA, et al. NF␬B prevents apoptosis and liver dysfunction during liver regeneration. J Clin Invest 1998;101:802-811. 24. Dignam JD, Lebovitz RM, Roeder RG. Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res 1983;11:1475-1489. 25. Rusyn I, Tsukamoto H, Thurman RG. WY-14,643 rapidly activates nuclear factor ␬B in Kupffer cells before hepatocytes. Carcinogenesis 1998; 19:1217-1222. 26. Qing K, Wang X-S, Kube DM, Ponnazhagan S, Bajpai A, Srivastava A. Role of tyrosine phosphorylation of a cellular protein in adeno-associated

HEPATOLOGY Vol. 32, No. 5, 2000

27. 28. 29. 30.

31. 32. 33.

34.

virus 2-mediated transgene expression. Proc Natl Acad Sci U S A 1997; 94:10879-10884. Kono H, Rusyn I, Thurman RG. Diphenyleneiodonium, an NADPH oxidase inhibitor, prevents early alcohol-induced liver injury in rats. Toxicologist 1999;48:257. Baldwin AS. The NF-␬B and I␬B proteins: new discoveries and insights. Annu Rev Immunol 1996;14:649-681. Lin M, Pham TV, Tsukamoto T. In vivo suppression of Kupffer cell NF-␬B activation and cytokine gene expression by iron chelator. HEPATOLOGY 1995;22:365A-365A. Bruening W, Giasson B, Mushynski W, Durham HD. Activation of stressactivated MAP protein kinases up-regulates expression of transgenes driven by the cytomegalovirus immediate/early promoter. Nucleic Acids Res 1998;26:486-489. Sanlioglu S, Engelhardt JF. Cellular redox state alters recombinant adeno-associated virus transduction through tyrosine phosphatase pathways. Gene Ther 1999;6:1427-1437. Speir E, Tomoko S, Zu-Xi Y, Victor F, Epstein SE. Role of reactive oxygen intermediates in cytomegalovirus gene expression and in the response of human smooth muscle cells to viral infection. Circ Res 1996;79:1143-1152. Koeberl DD, Alexander IE, Halbert CL, Russel DW, Miller AD. Persistent expression of human clotting factor IX from mouse liver after intravenous injection of adeno-associated virus vectors. Proc Natl Acad Sci U S A 1997;94: 1426-1431. Zwacka RM, Zhou W, Zhang Y, Darby CJ, Dudus L, Halldorson J, Oberley L, et al. Redox gene therapy for ischemia/reperfusion injury of the liver reduces AP1 and NF-␬B activation. Nat Med 1998;4:698-704.

WHEELER ET AL.

1059

35. Thurman RG. Mechanisms of hepatic toxicity. II. Alcoholic liver injury involves activation of Kupffer cells by endotoxin. Am J Physiol 1998;275: G605-G611. 36. Schreck R, Rieber P, Baeuerle PA. Reactive oxygen intermediates as apparently widely used messengers in the activation of the NF-␬B transcription factor and HIV-1. EMBO J 1991;10:2247-2258. 37. Yin M, Wheeler MD, Kono H, Bradford BU, Gallucci RM, Luster MI, Thurman RG, et al. Essential role of tumor necrosis factor a in alcoholinduced liver injury. Gastroenterology 1999;117:942-952. 38. Mohr L, Tanaka S, Wands JR. Ethanol inhibits hepatocyte proliferation in insulin receptor substrate 1 transgenic mice. Gastroenterology 1998;115: 1558-1565. 39. Chen J, Ishac E, Dent P, Kunos G, Gao B. Effects of ethanol on mitogenactivated protein kinase and stress-activated protein kinase cascades in normal and regenerating liver. Biochem J 1998;334:669-676. 40. Miao CH, Snyder RO, Schowalter DB, Patijn GA, Donahue B, Winther B, Kay MA, et al. The kinetics of rAAV integration in the liver. Nat Genet 1998;19:13-15. 41. Duan D, Sharma P, Dudus L, Zhang Y, Sanlioglu S, Yan Z, Yue Y, et al. Formation of adeno-associated virus circular genomes is differentially regulated by adenovirus E4 ORF6 and E2a gene expression. J Virol 1999; 73:161-169. 42. Joyce D, Bouzahzah B, Fu M, Albanese C, D’Amico M, Steer J, Klein JU, et al. Integration of Rac-dependent regulation of cyclin D1 transcription through a nuclear factor-␬B-dependent pathway. J Biol Chem 1999;274: 25245-25249.