Optimized adeno-associated virus 8 produces hepatocyte-specific Cre-mediated recombination without toxicity or affecting liver regeneration

Abstract

Engineering viral vectors to produce liver-specific protein expression may help advance understanding of hepatic regeneration and disease states. In addition to introducing genes of interest to the liver, these vectors can be adapted for gene deletion when designed to express Cre recombinase. The ability to use this system requires high, liver-restricted expression, low toxicity, and no effect on the process of interest. We developed an adeno-associated virus 8 (AAV8) with a codon-optimized Cre recombinase under a hepatocyte-specific major urinary protein (MUP) promoter (MUP-iCre-AAV8) that fulfills these requirements. A single intravenous injection of ROSA26R reporter mice, which express lacZ after Cre-mediated recombination, demonstrated homogeneous β-galactosidase expression limited to hepatocytes after only 7 days. Cre protein expression remained strong for at least 31 days. Serum liver function tests and histology demonstrated minimal liver toxicity. The presence of MUP-iCre-AAV8 did not affect hepatocyte proliferation after partial hepatectomy as measured by Ki67 staining. Conclusion: AAV8 with the MUP promoter, by virtue of its lack of hepatic toxicity or effect on liver regeneration, may be an efficient alternative to complex transgenic methodologies for studies of the mouse liver.

adeno-associated viruses (AAV) are promising vectors for liver-specific gene therapy (5, 26, 30, 36, 42). They can also be adapted to studies of basic liver physiology, the effects of interventions on normal and disease states, and liver repair after injury if it is established that they have minimal liver toxicity and no effect on the process under investigation.

Cre recombinase can be used to remove genetic elements that are flanked by specific restriction sequences (loxP). This powerful technique has been adapted to temporally and spatially alter gene expression in mice (14, 21, 35). Traditionally, this requires two strains of mice, one in which Cre recombinase is restricted to a specific tissue or cell type by choice of a promoter and a second in which the gene of interest is flanked by loxP sites (“floxed”). Temporal control can be added by using a modified Cre recombinase that is active in the presence of tamoxifen (10, 15, 18, 31, 40). The Cre/loxP system is limited by the labor and time required to produce, cross, and maintain mouse strains and reports of mosaic gene activation and tamoxifen-independent “leakage” of recombination. Furthermore, complicated breeding strategies often introduce phenotypically distinct genetic backgrounds that may confound interpretation. Finally, tamoxifen forms adducts with DNA and can independently alter liver physiology and gene expression (12, 28).

Viral delivery of Cre recombinase to target tissues has been described as a method to circumvent these problems (2, 7, 16, 17, 27, 33, 37, 38). Early generation adenoviral vectors for liver-directed delivery were limited by transient expression of Cre and cellular immune responses to viral proteins (1, 8, 22, 29, 34). More recent studies demonstrate that adenoviral-mediated gene transfer is associated with a significant immune response and liver toxicity (6, 24). Whereas protein expression levels are generally not an issue when using adenoviral vectors, achieving homogeneous Cre-mediated recombination is difficult. Systemic delivery of cytomegalovirus promoter-driven Cre in a high capacity adenoviral vector achieved recombination in only 10–20% of hepatocytes after 14 days (2) and use of a replication-deficient adenoviral vector produced a mosaic pattern of recombination after 4 days (23). Other methods of gene delivery, such as hydrodynamic delivery of plasmid DNA-encoding Cre, achieved recombination in 60% of hepatocytes after repeated treatments but were associated with hepatotoxicity as measured by liver function tests (41). Intraparenchymal liver injections of lentiviral vectors achieved local recombination and some distal recombination after 3 wk of incubation, but both hepatocytes and nonhepatocytes were transduced (27).

To address these limitations and achieve homogeneous hepatocyte-restricted induction of Cre-mediated recombination events, we used recombinant AAV8 expressing a codon-optimized Cre recombinase under a hepatocyte-specific major urinary protein promoter (MUP-iCre-AAV8). AAV8 is particularly suited for liver-directed gene delivery because of the rapid rise and subsequently high levels of expression that can be achieved after systemic administration, such as tail vein injections (5). MUP-iCre-AAV8 achieved homogeneous hepatocyte-restricted recombination after 7 days without significant hepatotoxicity or changes in hepatocyte proliferative capacity after partial hepatectomy.

MATERIALS AND METHODS

Vector construction, virus production, and purification.

A codon-optimized Cre recombinase gene (iCre) was produced by Shimshek et al (32). A plasmid containing the entire coding sequence of iCre, pBOB-CAG-iCRE-SD, was obtained from Addgene (plasmid 12336) (Cambridge, MA). The iCre gene, including the Kozak sequence, was amplified by PCR using forward primer 5′-TTCGTCGACAAGCTTGTCCACCATGGTGCC-3′ and reverse primer 5′-TTCCTGCAGTCGGGCCCCCCCTCGACTCAG-3′, which added SalI and PstI sites upstream of the translation initiation codon and downstream of the translation stop codon of iCre fragment, respectively, and removed a preexisting upstream HindIII site. The 1-kb PCR product was subcloned into pCR4-TOPO (Invitrogen, Carlsbad, CA), and the correct sequence of the iCre gene was verified. Following a SalI and PstI digestion, the iCre fragment was isolated and cloned into the same sites of pAMP.

We next constructed an AAV2 plasmid consisting of two AAV2 inverted terminal repeats flanking the MUP promoter (kindly provided by Dr. Chuck Rogler, Albert Einstein College of Medicine), a multiple cloning site, and an intron and polyadenylation signal sequence derived from SV40. Packaging of all recombinant AAV (rAAV) was carried out according to standard triple transfection protocols to create pseudotyped AAV2/8 virus. The AAV2/8 rep/cap plasmid provides the AAV2 replicase and AAV8 capsid genes. The adenoviral helper functions were provided by the pHelper plasmid (Stratagene, La Jolla, CA). Control virus was generated by packaging the empty MUP (pAMP) without the iCre gene. Briefly, AAV-293 cells were transfected with 10 μg of pHelper and 1.15 pmol each of AAV2/8 and AAV vector plasmids. The cells were harvested 48 h later, and clarified viral lysates were isolated from the cell pellets. The virus was pooled, aliquoted, and stored at −80°C. rAAV vector stocks were titered by real-time PCR using the ABI Prism 7700 Sequence Detection System from Perkin Elmer Applied Biosystems (Foster City, CA). All vectors were titered using primers and probe designed to amplify a sequence in the SV40 intron. The titers for MUP-iCre-AAV8 and empty MUP-AAV8 were 6.9 × 10¹¹ DNase-resistant particles (DRP)/ml and 1.1 × 10¹² DRP/ml, respectively.

Mice.

All procedures were approved by the animal care and use committee of Harvard Medical School and were performed in accordance with NIH guidelines. ROSA26R reporter, GtROSA, and C57BL6 mice were purchased from Jackson Laboratories (Bar Harbor, ME) (30). AlfpCre transgenic mice, which express Cre recombinase under albumin and alpha-fetoprotein regulatory elements, were obtained from K. Kaestner (University of Pennsylvania), crossed with CD-1 mice (Charles River Laboratories, Wilmington, MA), and genotyped according to reported methods (20). Adult AlfpCre mice were crossed with ROSA26R mice to obtain double hemizygous progeny.

Virus injections.

Male mice (7+ wk old) were positioned in a tail vein injection device, and the tail veins were vasodilated with warm water. Two hundred microliters of MUP-iCre-AAV8, empty MUP-AAV8, or vehicle (saline) were injected intravenously with a 27.5-gauge needle.

Partial hepatectomy, blood collection, and tissue preparation.

Partial hepatectomies were performed in the following manner. Mice were anesthetized with 60 mg/kg ketamine (Hospira, Lake Forest, IL) and 7 mg/kg xylazine (Phoenix Pharmaceutical, Burlingame, CA). A transverse incision was made inferior to the xyphoid process. The median and left lobes of the liver were eviscerated by gentle pressure on the abdomen, ligated with 4–0 silk, and excised. The abdomen was closed in two layers with running 4–0 vicryl suture (Ethicon, Piscataway, NJ). Mice were recovered under a heat lamp and monitored for pain and activity. To collect blood, mice were again anaesthetized, and the incisions were reopened. Venous blood was drawn from the inferior vena cava and allowed to clot in serum separator tubes (BD Biosciences, Sparks, MD) for at least 30 min. The tubes were centrifuged at 3,000 revolution/min for 10 min, and serum was isolated and frozen at −80°C until ready for use. PBS was perfused through the portal vein. After perfusion, organs of interest were removed, divided, and either frozen in liquid nitrogen, fixed in 10% neutral buffered formalin (American MasterTech Scientific, Lodi, CA) and embedded in paraffin, or frozen in molds in optimal cutting temperature compound (Tissue-Tek; Electron Microscopy Sciences, Hatfield, PA).

Testosterone replacement.

Selected females underwent testosterone replacement 5 days before virus injection and partial hepatectomy. Pellets (5 mg) of testosterone [21-day release (Innovative Research of America, Sarasota, FL)] were implanted subcutaneously according to the manufacturer's instructions. Testosterone reached a steady-state level within 48 h. Mice underwent virus injections after 5 days and partial hepatectomy an additional 7 days later.

Liver function testing.

Serum samples were loaded together with assay standards into a Cobas Mira Plus Chemistry Analyzer (Roche, Basel, Switzerland). Alanine aminotransferase (ALT), aspartate aminotransferase (AST), and total bilirubin were assayed using reagents from JAS Diagnostics (Miami, FL).

Histology and immunohistochemistry.

For β-galactosidase staining of tissue sections, 6-μm sections were collected on Superfrost/Plus slides (Fisher Scientific, Hampton, NH), fixed in 2% paraformaldehyde and 0.2% glutaraldehyde for 5 min, rinsed and stained overnight at 37°C (5 mM potassium ferricyanide, 5 mM ferrocyanide, 1 mg/ml 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside in 100 mM sodium phosphate, 2 mM MgCl₂, 0.01% sodium deoxycholate, and 0.02% NP-40). Slides were then rinsed in PBS, counterstained with nuclear fast red (Sigma-Aldrich, St. Louis, MO), dehydrated and mounted in Permount (Fisher Scientific). Additional sections were stained with hematoxylin and eosin. For CD31 and pancytokeratin staining, 6 μM frozen sections were fixed in acetone (Fisher Scientific) for 10 min, followed by quenching of endogenous peroxidase (peroxidase block; DAKO, Carpinteria, CA), avidin/biotin blocking (DAKO), and incubation with a rat anti-mouse CD31 antibody (1:500; BD Pharmingen, Sparks, MD), a rat anti-mouse F4/80 antibody (1:500; AbD Serotec, Raleigh, NC), or a rabbit anti-cow cytokeratin (1:1,000, DAKO) overnight at 4°C. Bound antibody was detected with biotinylated secondary antibody (Vector Laboratories, Burlingame, CA), streptavidin-horseradish peroxidase (HRP) (1:300; DAKO), and diaminobenzidine substrate solution (DAKO). Sections were counterstained, dehydrated, and mounted. Ki67 staining was performed using the same method (rat monoclonal anti-Ki67 antibody, 1:50) on paraffin-embedded liver sections that were subjected to antigen retrieval in citrate buffer (95°C, 20 min) after deparaffinization and rehydration. Microscopy was performed using the Nikon Eclipse 80i microscope and Spot Flex camera (MicroVideo Instruments, Avon, MA). Proliferation index was quantified as the average percent of labeled hepatocyte nuclei that stained with Ki67 per three high power fields (original magnification 400×).

Western blotting.

Frozen liver tissue was ground in a mortar and pestle and lysed in 50 mM Tris·HCl, pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN). Lysates were sonicated and cleared by centrifugation. Protein concentration was assayed using a modified Lowry assay (Bio-Rad, Hercules, CA). A sample (50 μg) of total protein was separated by SDS-PAGE, transferred to nitrocellulose membranes using a semidry technique, blocked in 5% milk, incubated overnight with a rabbit anti-Cre recombinase antibody (1:1,000; Covance, Cumberland, VA; or 1:500; Novagen, Madison, WI) and detected with an HRP-conjugated secondary antibody (1:10,000; Jackson ImmunoResearch Laboratories, West Grove, PA) and enhanced chemiluminescent substrate (Pierce, Rockford, IL). Membranes were incubated with 0.1% sodium azide to inactivate peroxidase activity, washed, and reincubated with a monoclonal antibody to mouse β-actin (1:10,000; Sigma) followed by an HRP-conjugated secondary antibody and developed as before.

RESULTS

MUP-iCre-AAV8 vector is functional in vivo.

To test the ability of MUP-iCre-AAV8 to mediate liver-specific recombination, tail veins of ROSA26 reporter mice were injected with 200-μl virus particles of MUP-iCre-AAV8 (1.4 × 10¹⁰ DRP/mouse) or vehicle. These mice express lacZ after successful Cre-mediated recombination. This route of administration has previously been shown to be effective in liver-specific delivery of an AAV8-MUP construct encoding green fluorescent protein (C. Bass, unpublished observations). Seven days after injection, we isolated organs from vehicle and virus-injected mice. Staining of sections for β-galactosidase activity revealed strong signal in liver and none in any other tissue tested, including kidney, heart, intestine, spleen, lung, and pancreas. No staining was observed in empty MUP-AAV8-treated livers (Fig. 1). A similar pattern of liver-specific recombination was observed 5 wk after recombination and is grossly comparable to that obtained by crossing ROSA26R mice with AlfpCre transgenic mice in which Cre recombinase is under the control of the albumin and α-fetoprotein promoters, which are liver specific (data not shown).

Fig. 1.

MUP-iCre-AAV8 achieves efficient Cre-mediated recombination in the livers of ROSA26R reporter mice. Mice were given intravenous injections of empty major urinary protein (MUP)-adeno-associated virus 8 (AAV8) (top) or MUP-iCre-AAV8 (bottom), and β-galactosidase staining of organs was assessed after 7 days. β-Galactosidase activity is limited to the livers of mice that received MUP-iCre-AAV8.

iCre directed by MUP promoter mediates recombination only in hepatocytes.

To characterize which cells in the liver are transduced by virus, consecutive tissue sections stained for β-galactosidase activity and cell-specific markers were compared. β-Galactosidase activity is present throughout the liver but spares the cells lining the sinusoids (Fig. 2A). To determine whether immunological cells, specifically Kupffer cells, are transduced, sections were stained for the macrophage marker F4/80. Although Kupffer cells are clearly present in liver sections (Fig. 2B, arrows), the close apposition of these cells to hepatocytes and the intense, homogeneous staining of hepatocytes makes it impossible to comment on recombination. To address this question, we performed similar analysis on the spleen, demonstrating no recombination in splenic macrophages, which are derived from the same cells as Kupffer cells (Fig. 2, C and D). To further evaluate the cells within the liver that are infected by the virus, consecutive sections of MUP-iCre-AAV8, GtROSA, and AlfpCre/ROSA26R mice were stained with endothelial cell (CD31) and biliary cell (pancytokeratin) markers (Fig. 2E). There is no colocalization of β-galactosidase staining with CD31 (arrows) in virus-injected mice although there is β-galactosidase staining in the endothelial cells of GtROSA mice. There is also no colocalization with pancytokeratin (arrows) in virus-injected mice although there is strong recombination in biliary cells in AlfpCre/ROSA26R mice. These data demonstrate that Cre expression is limited to hepatocytes.

Fig. 2.

MUP-iCre-AAV8 achieves hepatocyte-restricted recombination. A: β-galactosidase staining of the liver 7 days after intravenous injection of MUP-iCre-AAV8. There is homogeneous transduction of only hepatocytes. Cells occupying sinusoids and those comprising the portal mesenchyme are spared. PV, portal vein, HA, hepatic artery, BD, bile duct. Original magnification 400×. To determine whether immunological cells are transduced by the virus, consecutive sections were immunostained with Kupffer cell marker F4/80 (original magnification 400×). B: immunohistochemical staining for F4/80 in the liver. Although Kupffer cells are apparent (arrows), it was not possible to exclude β-galactosidase activity in the Kupffer cells of the liver due to high levels of staining throughout. To address this question, we performed similar analysis of the spleen. C: β-galactosidase staining of the spleen 7 days after intravenous injection of MUP-iCre-AAV8. There was no recombination in the spleen (original magnification 400×). D: immunohistochemical staining for F4/80 in the spleen. Macrophages are clearly present. E: colocalization of β-galactosidase staining in virus-injected mice with endothelial cell (CD31) and biliary cell (pancytokeratin) markers. Serial sections of virus-injected GtROSA and AlfpCre/ROSA26R mice were subjected to CD31 or pancytokeratin immunohistochemistry or β-galactosidase staining. There is no colocalization or β-galactosidase staining in virus-injected mice with either CD31 (arrowheads) or pancytokeratin (arrows) although there is clearly colocalization with CD31 in GtROSA mice and with pancytokeratin in AlfpCre/ROSA26R mice. Original magnification 400×.

Recombination efficiency of MUP-iCre-AAV8 is dose dependent.

To determine the minimum amount of MUP-iCre-AAV8 required for diffuse transduction of hepatocytes, ROSA26R reporter mice were injected with serial dilutions of the viral construct. After 1 wk, livers were harvested and analyzed by β-galactosidase staining (Fig. 3), demonstrating that maximum recombination is achieved with 1.4 × 10¹⁰ DRP/mouse (a 1:10 dilution of virus stock, Fig. 3A), with corresponding decrements in recombination as the titer decreases. 1.4 × 10⁹ DRP/mouse (1:100 dilution of virus stock, Fig. 3B) produces ∼70–80% recombination, and 0.5 × 10⁹ DRP/mouse (1:300 dilution, Fig. 3C) produces 40–50% recombination. There is no staining in empty MUP-AAV8-injected mice (Fig. 3D).

Fig. 3.

Cre-mediated recombination is dose dependent. ROSA26R reporter mice were injected with increasing dilutions of MUP-iCre-AAV8. Livers were harvested after 1 wk, and β-galactosidase staining was performed on sections. A: 1.4 × 10¹⁰ DNase-resistant particles (DRP)/mouse. B: 1.4 × 10⁹ DRP/mouse. C: 0.5 × 10⁹ DRP/mouse. D: empty MUP-AAV8 (2.2 × 10¹¹ DRP/mouse). Original magnification 400×.

Expression of Cre recombinase is sustained for at least 1 mo.

β-Galactosidase activity in individual hepatocytes persists after recombination occurs and is not an indication of sustained Cre expression. To examine protein expression levels of Cre, we performed Western blotting analysis of liver homogenates obtained from mice injected with MUP-iCre-AAV8 or vehicle. Since maximum recombination occurs with 1.4 × 10¹⁰ DRP/mouse, we compared expression levels of Cre 7 and 31 days after injection of this titer. Cre protein levels are detectable within 1 wk and increase over 1 mo after injection (Fig. 4A). There is no protein detected in vehicle or empty MUP-AAV8-injected animals. Furthermore, at 1 wk after injection, the expression level of Cre protein is already significantly greater than that in AlfpCre/ROSA26R mice (data not shown). Cre protein expression also decreases with increasing dilutions of MUP-iCre-AAV8 (Fig. 4B).

Fig. 4.

Cre protein expression is dose and time dependent. C57BL6 mice were injected with specified titers of MUP-iCre-AAV8 or vehicle. At specific points, their livers were homogenized and analyzed for expression levels of Cre recombinase (38 kDa) by Western blotting. β-Actin (37 kDa) is the loading control. A: Cre protein is detectable 1 wk after injection of 1.4 × 10¹⁰ DRP/mouse MUP-iCre-AAV8 and increases over 1 mo. There is no Cre protein detected in vehicle or empty MUP-AAV8-injected mice. B: there are serial increases in Cre protein levels in mice injected with increasing DRP of MUP-iCre-AAV8.

Administration of MUP-iCre-AAV8 and Cre-mediated recombination has minimal associated liver toxicity.

The toxicity of vector administration and Cre-mediated recombination in the liver was assessed by histological analysis of liver sections and by serum levels of liver enzymes. One month after delivery of MUP-iCre-AAV8, there was neither a significant alteration in liver cellular architecture nor an inflammatory cell infiltrate as assessed by hematoxylin and eosin staining of liver sections (Fig. 5A). One day after injection of 1.4 × 10¹⁰ DRP MUP-iCre-AAV8/mouse (the minimum dose that produces homogeneous recombination) or 2.2 × 10¹¹ empty MUP-AAV8, ALT, AST, and total bilirubin were mildly elevated but not different from vehicle-injected controls, suggesting that neither the virus nor its gene product were responsible for the elevated liver function tests. Levels return to normal within 4 days and are stable for 1 mo (Fig. 5B). Reference levels are ALT, 19–30 U/l, AST, 30–40 U/l, and total bilirubin, 0.1–0.3 mg/dl.

Fig. 5.

MUP-iCre-AAV8 delivery has little associated liver toxicity. A: hematoxylin and eosin staining of livers from vehicle-injected, empty MUP-AAV8, and MUP-iCre-AAV8 mice after 31 days demonstrates comparable cellular architecture with no inflammatory cell infiltrate. Original magnification 400×. B: serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and total bilirubin 1, 4, 7, and 31 days after injection of vehicle, 1.4 × 10¹⁰ DRP MUP-iCre-AAV8/mouse, or 2.2 × 10¹¹ DRP empty MUP-AAV8/mouse. Data are compiled from 3–4 mice in each treatment group. Reference levels are indicated in the text.

MUP-iCre-AAV8 did not alter the proliferative response of hepatocytes after partial hepatectomy.

To determine the effect of MUP-iCre-AAV8 on cellular proliferation in the liver, we performed 70% hepatectomy 1 wk after wild-type (C57BL6) mice were injected with vehicle, MUP-iCre-AAV8, or empty MUP-AAV8. As maximum recombination efficiency is obtained with 1.4 × 10¹⁰ DRP/mouse MUP-iCre-AAV8, we injected all mice with this titer of MUP-iCre-AAV8 or with 2.2 × 10¹¹ empty MUP-AAV8. Hepatocyte proliferation after partial hepatectomy, as assessed by Ki67 staining, peaks 48 h postoperatively (data not shown). We found that, in vehicle-injected mice, the mean proliferative index at 48 h was 41.8 ± 9.7%. Mice injected with MUP-iCre-AAV8 and empty MUP-AAV8 demonstrated no significant difference in mean proliferative index (41.5 ± 11.2% and 43.6 ± 0.7%, respectively) (Fig. 6).

Fig. 6.

Effect of MUP-iCre-AAV8 on hepatocyte proliferation after partial hepatectomy. Wild-type mice were injected with vehicle, 2.2 × 10¹¹ DRP empty MUP-AAV8, or 1.4 × 10¹⁰ DRP/mouse MUP-iCre-AAV8 and underwent partial hepatectomy after 1 wk. Livers were harvested 48 h postoperatively. Liver sections were stained with Ki67, and mean proliferation index was calculated. Data are compiled from 4 mice in each treatment group.

Transduction of hepatocytes by MUP-iCre-AAV8 is sex-dependent.

To determine the sex dependence of the transduction of hepatocytes by MUP-iCre-AAV8, we examined β-galactosidase expression in the native liver (excised portion) and after hepatectomy in male and female mice. Reporter mice were injected with 1.4 × 10¹⁰ DRP MUP-iCre-AAV8/mouse or 2.2 × 10¹¹ DRP empty MUP-AAV8/mouse. Livers were examined 1 wk after injection. Hepatocytes from female mice demonstrated significantly less staining than those from male mice (20% vs. 100%, Fig. 7A, top). Addition of testosterone to females increased staining to 100%, similar to the findings in males. In regenerating livers, the level of recombination in females increased to 90% only after testosterone replacement, whereas the fraction of labeled cells in male regenerating livers remained stable (Fig. 7A, bottom). To determine whether testosterone was acting by increasing protein levels of Cre, Western blot was performed before and after testosterone injections. Administration of testosterone to females increased protein expression in females to a level similar to males (Fig. 7B). Cre recombination was not observed in female reproductive organs before or after testosterone injections (data not shown).

Fig. 7.

Transduction of hepatocytes by MUP-iCre-AAV8 demonstrates sexual dimorphism. Reporter mice were injected with 1.4 × 10¹⁰ DRP/mouse MUP-iCre-AAV8 and underwent partial hepatectomy after 1 wk. A, top: the excised portion of livers of female and male mice were analyzed for β-galactosidase activity. There is more diffuse staining in the excised livers of male mice than female mice. When female mice undergo testosterone (T) replacement before MUP-iCre-AAV8 injection, recombination efficiency is equivalent to that in males. Results are representative of findings in 4 different mice of each group. Original magnification 100×. Bottom: 48 h after partial hepatectomy, the regenerating livers of the same mice were harvested and analyzed for β-galactosidase activity. Activity increases in female mice after hepatectomy. B: Western blot demonstrating differences in Cre protein expression in male and female mice. Female mice express less Cre recombinase than males, but levels are equivalent after testosterone replacement.

DISCUSSION

Liver-specific gene modification is a powerful tool for investigating the role of specific proteins in normal physiology, during repair after injury, and in disease states. We present an efficient method for achieving this using an AAV8-based vector expressing a codon-optimized Cre recombinase under the MUP promoter.

Our strategy results in homogenous and hepatocyte-selective recombination within 1 wk of administration. In contrast to transgenic strategies utilizing the Alfp promoter, which also transduces biliary epithelial cells, our virus construct is active only in hepatocytes and may be useful for separating effects observed in biliary from those in hepatocytes. Demonstration that nonhepatocytes in GtROSA mice express lacZ confirms that the specificity is due to the virus and not the reporter locus. The dose-dependent transduction of hepatocytes with our construct also suggests a way to manipulate levels of protein expression. These findings represent an advance over previous reports. Low recombination efficiencies hampered other attempts to use viral-based Cre delivery (2, 13, 27, 38). Other limitations included failure to restrict recombination to hepatocytes or hepatic tissues (1) and long incubation times ranging from 2 wk (2) to 3 wk (27). Use of a strong hepatocyte-specific promoter, MUP, and a mammalian codon-optimized Cre gene, which contains minimal CpG content and an optimal Kozak consensus sequence (32), likely accounts for the 100% recombination efficiency that we observed within 1 wk of the injection.

The efficiency of the transduction can be modeled. Mice hepatocytes have a diameter of ∼4 μm (25). Assuming a cell density close to water, there are ∼2 × 10⁸ cells in a typical adult mouse liver, which comprises 4% total body weight. If 1.4 × 10⁹ DRP/mouse are required to transduce 80% of the hepatocytes (1.6 × 10⁸ cells), and assuming a Poisson distribution, the efficiency of transduction is ∼11 DRP/cell. Virus particles may be lost to immune clearance, defective particles, or failure to effectively transduce cells.

The applicability of our strategy depends on minimal associated liver toxicity and preservation of hepatocyte proliferation in response to injury. We have achieved both these aims as demonstrated by histology and liver function tests after gene delivery and Ki67 labeling after partial hepatectomy. Peak ALT and AST after MUP-iCre-AAV8 injection were 38 and 178 U/ml, respectively, and were observed 1 day after injection, were comparable to vehicle controls, and were ∼1/5 the levels observed with other delivery methods (41). Liver function tests (ALT, AST, and bilirubin) returned to normal after 4 days. To guard against the growth-suppressive effects of AAV8 in normal and transformed cultured cells and the possibility that high levels of Cre expression in vivo might compromise cell proliferation and survival (3, 4, 11, 39), we used the lowest dose of MUP-iCre-AAV8 required to achieve homogeneous recombination. The time course of Cre expression that we observed is similar to the findings that AAV delivery of a green fluorescent protein-Cre fusion protein to the adult mouse brain can mediate a maximum level of recombination after 7–14 days without evidence of neuronal damage or loss (17).

MUP-iCre-AAV8 has other distinct advantages over previous technologies, including efficient delivery by tail vein injection. Other routes of administration, including portal vein (13) or intraparenchymal injections (27), require more technical expertise and may be more stressful to the animals.

Surprisingly, male mice demonstrated higher recombination efficiency than females. Examination of protein levels revealed lower protein levels in females, and administration of testosterone increased levels comparable to males. These results are consistent with the known hormonal control of the MUP promoter (19) or to androgen-dependent AAV transduction (9). This sex-dependent difference may limit the usefulness of the MUP-iCre-AAV8 to male mice although it may also offer the opportunity to explore sex differences in hormonal regulation of protein expression in the liver.

MUP-iCre-AAV8 can achieve efficient, dose-dependent, and hepatocyte-restricted gene deletion in vivo without liver toxicity or altering the physiological process of liver regeneration. It may be a valuable system for studying the effects of adding or deleting genes in various normal and pathological states of the liver and its response to injury.

GRANTS

This work was supported by the NIH (DK064648 to S. Karp and HL007734 to K. Ho) and the Julie Henry Fund at Beth Israel Deaconess Medical Center.