Research Reports

Marie-Jeanne T.F.D. VranckenPeeters, Andre Lieber, JamesPerkins and Mark A. KayUniversity of Washington, Seattle, WA, USA

ABSTRACT

For many preclinical studies, the mousehas been an invaluable model. For hepaticstudies, including gene therapy, the use ofthe mouse has been limited because of theinability to obtain long-term portal vein ac-cess. In this study, we have developed a sur-gical cannula model that allows for repeatportal vein infusion in a noninvasive man-ner. We have used this model to establishthat the tissue distribution of recombinantadenoviral vectors is similar after portalvein or peripheral vein infusion. The major-ity of the vector was present in the liver,ranging from 14 to 28 copies per hepato-cyte. The second most prevalent tissueswere the spleen and lung with 1/10 less ade-noviral DNA. The brain and ovaries had theleast DNA, 1/1000 less than the liver. Addi-tional studies were performed to study theeffects of secondary adenovirus infusionthrough the portal vein cannula. Permanentportal vein access in a mouse model will beinvaluable for a large number of medicalstudies, including the development of newtechnologies for hepatic gene transfer.

INTRODUCTION

The liver is an attractive target organfor gene therapy since a wide variety ofproteins are manufactured in the liverand many metabolic diseases resultfrom an absence or deficiency of thesehepatocyte-derived gene products. Sofar, two major approaches have beenused for liver-related gene therapy inanimal models in an attempt to correctgenetic deficiencies. Ex vivo gene ther-apy involves transplantation of autolo-gous retrovirus-transfected hepatocytes(3,6). To date this is the only method ofhepatic gene therapy that has been usedin clinical trials (5).

Strategies for direct in vivo genetransfer include nonviral and viral vec-tors for delivery into the portal or pe-ripheral vasculature (10). Recombinantretroviral vectors and recombinant ade-noviral vectors are most widely usedfor hepatic gene transfer. The retrovirusis limited by the low titer, and thus, vol-ume restriction limits the number oftransducing particles that can be deliv-ered (2,9). Conversely, recombinantadenoviral vectors can be concentratedto high titer and can transduce non-dividing cells at very high efficiencies(11). There are two major disadvan-tages of adenoviral vectors. The first isthe associated immune response thatlimits the persistence of gene expres-sion and the ability to perform secon-dary gene transfer (1,17,20). A secondpotential problem is that after intrave-nous infusion, adenovirus, while highlyhepatotrophic, infects most other tis-sues to varying degrees (4,8, 11,17).

The mouse is a valuable animalmodel for studying adenovirus-mediat-ed gene therapy since it is small, andthere are a large number of different in-

bred and mutant strains for studyingdifferent metabolic genetic diseases(15). The use of the mouse as an animalmodel for hepatic gene transfer, how-ever, has one disadvantage: the limitedaccess to the portal vasculature. Theanatomy of this small animal makes itdifficult to perfuse the portal vein, andbecause of adhesions after portal veininjection, it has not been possible toperform multiple injections. Addition-ally, access to the portal vein is limitedby ethical considerations of repeatedsurvival procedures in animal researchstudies. The ability to perfuse the portalvasculature in a noninvasive mannerwould be useful for developing newtechnologies to increase the efficiencyof both in vivo and ex vivo gene thera-py protocols.

In this report we describe a newmethod that allows multiple portal veininjections in a mouse model by theplacement of a permanent cannula.Subsequently, two questions were ad-dressed using this new model. The firstwas to quantitate recombinant aden-ovirus within different tissues after ei-ther portal vein or peripheral tail veininfusion. Smith et al. (17) demonstratedsimilar levels of a plasma protein afteradministration of an adenoviral vectordirectly into the liver parenchyma ortail vein. They were able to quantitatethe amount of adenovirus DNA in vari-ous tissues after peripheral vein admin-istration of adenovirus. Other studieshave used a less quantitative poly-merase chain reaction (PCR) assay todetermine a range of adenoviral DNAin different tissues after portal vein in-fusion (4). None of these studies direct-ly compared adenovirus distribution intissues after portal and peripheral veininfusion.

278 BioTechniques Vol. 20, No. 2 (1996)

Method for Multiple Portal Vein Infusions inMice: Quantitation of Adenovirus-Mediated Hepatic Gene TransferBioTechniques 20:278-285 (February 1996)

The second question we addressedto our new model was whether theblock to secondary gene transfer byproduction of neutralizing antibodiescould be overcome by comparing trans-gene expression after a second directadenovirus infusion into the portal vas-culature vs. the peripheral vein.

MATERIALS AND METHODS

Animals

Female C57Bl/6 mice (The JacksonLaboratory, Bar Harbor, ME, USA),ages 5–6 weeks, were used in the de-scribed experiments. All studies wereperformed in accordance with the insti-tutional guidelines at the University ofWashington.

Placement of a Permanent PortalVein Cannula

Mice were anesthetized with anintraperitoneal administration of 0.5mL of 20 mg/mL Avertin (2,2,2-tribro-moethanol; Aldrich Chemical, Milwau-kee, WI, USA). A midline abdominalincision was made, and the skin wasseparated from the abdominal wall tocreate a subcutaneous pocket. The ab-dominal wall was opened at the lineaalba, and the portal vein was exposedby displacing the intestinal duct.

Two different gas-sterilized tubes

were used for the cannulation. An 8-mm-long PE10 (0.011 in. i.d., 0.024 in.o.d.) polyethylene tube (Clay Adams,Parsippany, NJ, USA) was inserted intoa silicone tube (0.02 in. i.d., 0.037 in.o.d.; Scientific Products Medical GradeSilicone Tubing; Baxter, IL, USA). Thepolyethylene tube was advanced abouthalfway (4 mm) past the orifice of thesilicone tube and cut at the tip (withscissors) at about a 45 degree angle.Using a forceps, the sharp edge of thePE10 tube was used to penetrate theportal vein about 5 mm proximal to itsbifurcation (Figure 1A). The catheterwas gently advanced 4 mm (length ofexposed PE 10 tube) and the forceps re-moved. The cannula was perfused withheparinized saline (1 U/mL) and thenfixed in place with a drop of an adhe-sive (Histoacryl Blau; Braun Melsun-gen AG, Melsungen, Germany).

The intestinal duct was placed backto its original position, whereafter thecannula was tunneled through the ab-dominal wall, avoiding any blood ves-sels, and fixed with a 4.0 silk suture(David and Geck, Inc., AmericanCyanamid, Manati, PR, USA). Afterclosure of the abdominal wall (continu-ous suture, 4.0 silk), the cannula wastied off at the distal end and placed sub-cutaneously in the previously createdpocket (Figure 1B). Finally the skin

was closed using a 4.0 silk suture.After opening the skin at the proxi-

mal site of the already existing abdomi-nal scar, the cannula can easily bepulled out and connected to a syringepump to infuse either adenoviral vec-tors, retroviral vectors or in vitro-trans-duced hepatocytes.

Adenovirus Production and Admin-istration

Two recombinant adenoviral vectorswere used, the Ad.RSV-β-gal (18) andthe Ad/RSV-hAAT (7). The formervirus produces E. coli nuclear localizedβ-galactosidase (β-gal) and the latterproduces human α1-antitrypsin fromtransduced cells. The adenoviruseswere propagated, purified, titered andscreened for the absence of helper virusas described (1). The purified virus wasstored in 10 mM Tris-HCl pH 8.0, 1mM MgCl2, 10% glycerol in aliquots at-80°C and diluted freshly in serum-freeDulbecco’s modified Eagle media(DMEM) (Life Technologies, Gaithers-burg, MD, USA) prior to infusion.Mice received injections in either thetail vein or the portal vein by the portalvein cannula. For tail vein injections, 5× 109 plaque-forming units (pfu) of thepurified adenovirus were diluted to 100µL and injected into the tail vein inabout 5 s. For portal vein infusions, 5 ×109 pfu were diluted to 200 µL and in-fused over 5 to 10 min. ONPG and DNA Blot Analysis

Animals that received the Ad.RSV-β-gal were sacrificed, and fresh liver,spleen, lung, heart, brain and ovarywere analyzed for β-gal activity byo-nitrophenyl β-D-galactopyranoside(ONPG) assay, as previously described(1). A β-gal enzyme standard (SigmaChemical, St. Louis, MO, USA) thatwas mixed with the lysate from corre-sponding tissues of a normal animalwas used as a reference. The ONPG as-say had a linear range from 0 to 2500units. The β-gal activity from the liversof each mouse was analyzed individu-ally. For the other tissues, equalamounts of tissue from each of the 4mice were combined and analyzed to-gether. For whole tissue DNA prepara-tion, 0.1 g of tissue was homogenizedin 3 mL 10 mM Tris-HCl, 400 mMNaCl, 2 mM EDTA, 0.5% sodium do-decyl sulfate (SDS), pH 8.2. The cell

Vol. 20, No. 2 (1996) BioTechniques 279

Figure 1. Surgical placement of a permanent silicone cannula into the portal vein in mice. (A)Demonstration of insertion of the proximal tip in the portal vein; (B) the distal end of the cannula isplaced subcutaneously.

Research Reports

lysate was incubated overnight with200 U/mL proteinase K. After phenolchloroform extraction, 1.1 mL of satu-rated NaCl were added to the lysate.The supernatant was removed aftercentrifugation (30 min, 10 000× g atroom temperature) and the DNA pre-cipitated with ethanol. The DNA wasresuspended in 1× TE (10 mM Tris-HCl, 1 mM EDTA, pH 8.0).

Ten micrograms of DNA were di-gested with HindIII. The DNA wasloaded onto a 0.7% agarose gel and,following electrophoresis, transferredto a Hybond (Amersham, Indianapo-lis, IN, USA) filter. The blot was hy-bridized sequentially with a random-primed [α-32P]CTP-labeled 3.4-kbβ-gal DNA fragment from plasmidpCMV-β-gal (13) and the 2-kb HindIII/EcoRI mouse metallothionein I genefragment from plasmid mMMT-1 (16)in rapid hybridization buffer (Amer-sham) according to the manufacturer’sprotocol. The blot was washed in 0.1×standard saline citrate (SSC) at 65°Cand exposed to X-ray film and/or ana-lyzed by a Model 400S Phosphor-Imager (Molecular Dynamics, Sun-nyvale, CA, USA).

Biochemical Analysis

From mice that received theAd/RSV-hAAT adenovirus, blood sam-ples were obtained from the retro-orbital plexus. Serum concentrations ofhuman α1-antitrypsin were determinedby an enzyme-linked immunosorbentassay (ELISA), using a human-specificantibody as previously described (9).

For the adenoviral neutralizing anti-body assay, mouse serum samples wereheat-inactivated at 55°C for 40 min andanalyzed, following the protocol as de-scribed previously (1).

RESULTS

Permanent Access into the Portal Vein

A safe method for placement of apermanent portal vein cannula in micewas developed as described in themethods. The first series of experi-ments were designed to determine thesafety and function of the cannula. In-sertion of the catheter is shown inFigure 1. In an initial experiment, 20mice received a cannula. In experi-

enced hands, mortality rate was lessthan 5% in this and subsequent experi-ments. In some animals, methyleneblue was infused to demonstrate paten-cy and flow through the portal vascula-ture. In all cases the cannula remainedpatent for injection for at least 2 weeksafter placement without perfusion withheparinized saline. Cannulas have beenleft in mice for up to a year withoutcausing any detectable complications.Histological liver sections, obtainedfrom 4 mice, 12 weeks after placementof the cannula were normal (notshown).

Distribution of Adenovirus

To determine the tissue distributionof recombinant adenovirus, mice wereinjected with 5 × 109 pfu of Ad.RSV-β-gal into the tail vein (n = 4) or the portalvein (n = 4), through the cannula. Thisdose of adenovirus has previously beenshown to transduce 80% to 90% of he-patocytes (11). Three days after injec-tion of adenovirus, animals were sacri-ficed and different tissues (liver, spleen,lung, heart, brain and ovary) were ana-lyzed for β-gal enzymatic activity.

A high endogenous β-gal activitypresent in the spleen of normal miceprecluded analysis in this tissue. Al-though there was some variation be-tween animals, there was no significantdifference in β-gal enzymatic activity

280 BioTechniques Vol. 20, No. 2 (1996)

Figure 2. Recombinant adenoviral DNA detec-tion in mice transduced with Ad.RSV-ββ-gal.Three days after administration of 0.5 × 1010 pfuof Ad.RSV-β-gal by portal vein or tail vein injec-tion, total tissue DNAs were analyzed by South-ern blot. 10 µg of total DNA were digested withHindIII and loaded into each lane. The blot washybridized with a 32P-labeled β-gal gene frag-ment. DNA from the livers of individual animalswere analyzed: (A) control mouse, lane 1; tailvein infusion, lanes 2–5; portal vein infusion,lanes 6–9. DNA analysis of other tissues was per-formed after pooling an equal tissue mass fromeach of 4 animals; (B) tail vein infusions, lanes1–5; portal vein infusions, lanes 6–10; heart,lanes 1 and 6; ovary, lanes 2 and 7; lung, lanes 3and 8; spleen, lanes 4 and 9; brain, lanes 5 and 10.The specific β-gal hybridization band is the lowerone at 3.4 kb. There is a nonspecific hybridizationsignal that is slightly larger that is also present inthe control lane. Note the lower band in (B), lane4, is due to a small crack in the original gel.

ONPG Activity in U/g Tissue

Portal Vein Tail VeinOrgan Injection Injection Control

Liver #1 >2500 1004 0

#2 2104 >2500

#3 1764 1220

#4 456 656

Lung * 198 305 0

Heart * 50.7 128.6 0

Brain * 1.16 1.58 0

Ovary * 0.89 5.27 0

Animals were infused with 0.5 × 1010 pfu of Ad.RSV-β-gal and their organs wereremoved after 3 days. Liver tissue was analyzed individually in each of the 4mice.

*Mean activity obtained from equal amounts of tissue from each of the 4 mice.

Table 1. ββ-gal Enzyme Analysis in Different Tissues After Administration of Ad.RSV-ββ-galAdenovirus

in the livers of animals infused by tailvein or portal vein. The activity foundin the liver of 4 different mice rangedbetween 450 and >2500 U/g tissue(Table 1). Because of the relativelylower amount of gene expression inother tissues relative to the liver, eachtissue from the 4 animals was pooledfor analysis. The β-gal activity was de-tectable in all tissues with the lungshaving the second greatest amount ofactivity, followed by the heart. Both thebrain and the ovaries showed activitieswhich were 1/1000 of that obtained fromthe liver (Table 1). The amount of geneexpression did not appear to be depen-dent on the route of adenovirus admin-istration.

It is highly probable that variation inβ-gal activity between tissues may inpart result because of differences ingene regulation and may not directlyreflect adenovirus copy number. Thus,to more accurately quantitate genetransfer to various tissues, the relativeamount of DNA corresponding to theadenoviral genome was determined bygenomic Southern blot analysis (Figure2). Digestion of adenoviral DNA withHindIII produced the expected 3.4-kbβ-gal DNA fragment. The relativeamount of adenoviral DNA in all the li-

ver samples was similar (Figure 2A),whereas marked differences were de-tected between different tissues (Figure2B).

To quantitate the average number ofadenoviral genome copies per diploidgenome, samples containing varyinggenome equivalents of β-gal plasmidDNAs mixed with normal mouse ge-nomic DNA were used as standards inthe Southern blot (not shown). The blotwas reprobed with a mouse metallo-thionen exon I DNA probe to adjust forsmall variations in DNA loading andtransfer between lanes. A phosphor im-ager was used to quantitate the relativeβ-gal hybridization signal in each lane.The results are summarized in Table 2.The amount of adenoviral DNA aver-aged 14–28 copies per hepatocyte peranimal. The hepatic adenovirus ge-nome copies were 1/10 to 1/1000 lessthan in other tissues, and the relativequantity per tissue was in approximateagreement with the amount of β-gal en-zyme activity. The lung and spleen hadthe second greatest amount of adenovi-ral DNA that was 1/10 less than the liv-er. Importantly, no significant differ-ences in tissue distribution were foundafter administration of adenovirus byportal vein or tail vein.

Vol. 20, No. 2 (1996) BioTechniques 281

DNA Copy per Cell

Portal Vein Tail VeinOrgan Injection Injection

Liver #1 20 14#2 22 28#3 24 24#4 28 24

Spleen * 1.8 2.8Lung * 3.0 2.0Heart * 0.12 0.08Brain * 0.02 0.01Ovary * 0.02 0.08

The β-gal hybridization signals from the Southern blot (see Figure 2) werequantitated and presented as genome copies per cell. For concentration mark-ers, 5, 30 and 120 pg of pCMV-β-gal plasmid (30 pg represents 1 genomeequivalent) mixed with 10 µg genomic DNA from a noninjected animal weredigested with HindIII. After quantification of signals by a phosphor imager, theblot was rehybridized with a mouse metallothionein probe. This was used as aninternal control to adjust for the quantity of DNA in each lane and prior to calcu-lating the number of adenovirus copies per cell.

Table 2. Adenoviral Genome Quantitation in Tissues

Research Reports

Repeat Adenovirus Injection

The inability to transduce hepato-cytes a second time with adenovirus ismost likely related to humoral immuni-ty (1,17); thus, to determine whether in-fusion of virus just proximal to the liverin the portal vasculature would lead tosecondary gene transfer, the permanentportal vein cannula was used for sec-ondary adenovirus injection. Mice re-ceived a primary infusion of 5 × 109

pfu of Ad.RSV-β-gal into either the tailvein (n = 4) or the portal vein cannula(n = 4). All animals developed high-titer neutralizing antibodies directedagainst the adenovirus independent ofthe route of adenovirus administrationthat ranged between 1/64 and 1/128.Three weeks hereafter, mice were ex-posed to a second administration ofadenovirus. Ad/RSV-hAAT (5 × 109

pfu) virus was injected into the mice bythe same route as the primary infusion,by the tail vein or the portal vein cannu-

la. The antibody titers in all the animalsmeasured 3 days later were greater than1/1024.

To assess the success of the secondadenovirus injection, blood sampleswere obtained from the retro-orbitalplexus and assayed for hAAT proteinby ELISA. No hAAT (<100 ng/mL)was detected in the serum from any ofthe 8 mice that had previously been ex-posed to Ad.RSV-β-gal. In contrast,serum hAAT levels in previously naivecontrol mice injected by means of thecannula or tail vein were in the range of5000–10 000 ng/mL. Thus, even directsecondary infusion of virus into theproximal portion of the portal vein wasnot sufficient to achieve secondary ade-novirus-mediated gene transfer.

DISCUSSION

We report a new surgical techniqueto make the portal vasculature more ac-cessible for hepatic gene therapy in the

mouse by creating permanent access tothe portal vein. This method allows formultiple injections directly into theportal vein without a second invasiveprocedure. All injections can be per-formed slowly (10–50 min) by connect-ing the cannula to a syringe pump, al-lowing a greater volume to be perfused.

Using this new model, we addressedtwo questions for using adenoviral vec-tors for hepatic gene therapy. Adminis-tration of adenovirus into either the tailvein, inferior vena cava or the portalvein results in similar levels of gene ex-pression (7,17). In some previous stud-ies, amount of transgene product wasmeasured as a serum/plasma protein,and the proportion produced in the livercould not be quantitated. RecombinantDNA quantitation is a more accuratereflection of gene transfer, and in thisstudy, 14–28 adenoviral DNA copies ofadenovirus were detected per hepato-cyte. This is in agreement with Smith etal. (17), who demonstrated 55 copies

Research Reports

per cell when a twofold larger dose wasadministered by tail vein. Conversely,Fang et al. (4) reported 0.7 to 7 copiesper mouse hepatocyte with viral deliv-ery to the portal vein using a less quan-titative PCR assay to establish adeno-virus copy number. The lower copynumber may have been due to the ana-lysis performed one week after genetransfer, by which time gene expressionhad decreased. Assuming that there are108 hepatocytes per liver and that thereare 20 adenoviral genomes per cell,then we estimate 40% of the adenoviruswas taken up by liver. This is probablyan underestimate since mouse hepato-cytes may be polyploid (19). Our studyclearly demonstrates that the amount ofadenovirus in the liver is the same re-gardless of the route of administration.

The tissue distribution of adenovirusis an important safety considerationwhen such vectors are considered forclinical protocols. This study demon-strates that the distribution of vector

into other tissues is not influenced bythe route of administration. Both tailvein and portal vein injection result intransduction of other organs, includingheart, brain and ovary, but with 1/100 to1/1000 less vector per cell than liver.This study is in agreement with othersthat demonstrate that the spleen andlung contain the second greatestamount of vector, about 1/10 less thanliver (4,17).

Adenovirus-mediated gene transferleads to transient gene expression; forlong-term treatment of genetic metabo-lic disorders, multiple injections are re-quired. Previous studies reported thatrepeated tail vein injection failed to re-sult in gene expression (1,17). We nowdemonstrate that repeat injections intothe portal vein did not result in detecta-ble gene expression either. In this studyboth tail vein injection and portal veininjection led to the development of neu-tralizing antibodies against adenovirusthat inhibited reinfection of hepato-

cytes. Injection of a high dose of aden-ovirus into the portal vein, so that thevirus reaches the liver parenchyma im-mediately, did not circumvent the im-munological block against adenovirus.The reason for this result is not knownbut could be related to issues of first-pass uptake by the liver. Further studiesare needed to address the mechanism.

Long-term access to the portal veinwill be an attractive means by which todevelop new methods for improvingthe efficiency of both nonvirus- (e.g., li-posomes) and retrovirus-mediated invivo/ex vivo hepatic gene therapy.Retrovirus-mediated in vivo gene ther-apy has the resulting advantage of per-manent gene expression. However, thisapproach is limited by the amount ofvirus particles that one can infuse. Themaximal titer of standard retrovirus is106–107 colony-forming units (cfu) permL. The volume that can be injected inthe mouse is limited to 1 mL (about 2/3of the total blood volume of a mouse)

and results at best in a transduction ef-ficiency of only 1%–2% (2,9). Recentlywe have developed new technologiesusing the portal vein cannula to givemultiple doses of retrovirus (12).

For hepatocyte transplantation in themouse as part of an ex vivo approach,the amount of hepatocytes that can beinjected in the portal vein is limited.When injecting more than 5 × 105 cellsin the portal vein, the mortality rate is100% mainly due to pulmonary emboliand portal vein thrombosis. Less than1% of the total liver mass can be re-placed by transplanting this maximalamount of hepatocytes (14). By givingslow infusions of 2 × 106 hepatocytes24 hours apart, it has been possible totransplant up to 4 × 106 hepatocytes(unpublished data). This will increasethe range of transplantation studies thatcan be performed in mice.

The surgical implantation of a per-manent cannula into the portal vein ofmice will have applications for study-ing a number of different metabolic pa-rameters outside of gene therapy thatrequire portal vein access for multipleprocedures.

ACKNOWLEDGMENTS

This work was supported in part byNIH Grant, DK 49022 and the LucilleP. Markey Charitable Trust. We thankR. Garcia for his technical advice. M.-J.T.F.D. V.P. was supported by the DutchNWO Grant 901-01-096.

REFERENCES

1.Barr, D., J. Tubb, D. Ferguson, A. Lieber,A.J. Perkins and M.A. Kay. 1995. Strain re-lated variations in adenovirally-mediatedtransgene expression in mouse hepatocytes invivo: comparisons between immunocompe-tent and immunodeficient inbred strains. GeneTher. 2:151-155.

2.Branchereau S., D. Calise and N. Ferry.1994. Factors influencing retroviral-mediatedgene transfer into hepatocytes in vivo. Hum.Gene Ther. 5:803-808.

3.Chowdhury, J.R., M. Grossman, S. Gupta,N.R. Chowdhury, J.R. Baker and J.M. Wil-son. 1991. Long-term improvement of hyper-cholesterolaemia after ex vivo gene therapy inLDLR-deficient rabbits. Science 254:1802-1805.

4.Fang, B., R.C. Eisensmith, X.H.C. Li, M.J.Finegold, A. Shedlovsky, W. Dove andS.L.C. Woo. 1994. Gene therapy for phenyl-ketonuria: phenotypic correction in a geneti-cally deficient mouse model by adenovirus-mediated hepatic gene transfer. Gene Ther.

1:247-254.5.Grossman, M., S.E. Raper, K. Kozarsky,

E.A. Stein, J.F. Engelhardt, D. Muller, P.J.Lupien and J.M. Wilson. 1994. Successfulex vivo gene therapy directed to liver in a pa-tient with familial hypercholesterolaemia.Nat. Genet. 6:335-341.

6.Kay, M.A., P. Baley, S. Rothenberg, F. Le-land, L. Fleming, K. Parker Ponder, T. Liu,M. Finegold, G. Darlington, W. Pokornyand S.L.C. Woo. 1992. Expresson of humanα1-antitrypsin in dogs after autologous trans-plantation of retroviral transduced hepato-cytes. Proc. Natl. Acad. Sci. USA 89:89-93.

7.Kay M.A., F. Graham, F. Leland and S.L.C.Woo. 1995. Therapeutic serum concentrationsof human alpha-1-antitrypsin after adenovrialmediated gene transfer into mouse hepato-cytes. Hepatology 21:815-819.

8.Kay, M.A., C.N. Landen, S. Rothenberg,L.A. Taylor, F. Leland, S. Wiehle, B. Fang,D. Bellinger, M. Finegold, A.R. Thompson,M. Reed, K.M. Brinkhous and S.L.C.Woo.1994. In vivo hepatic gene therapy: completealbeit transient correction of factor IX defi-ciency in hemophilia B dogs. Proc. Natl.Acad. Sci. USA 91:2353-2357.

9.Kay, M.A., Q. Li, T. Liu, F. Leland, C.Toman, M. Finegold and S.LC. Woo. 1992.Hepatic gene therapy: persistent expression ofhuman α1-antitrypsin in mice after directgene delivery in vivo. Hum. Gene Ther. 3:641-647.

10.Kay, M.A. and S.L.C. Woo. 1994. Gene ther-apy for metabolic disorders. Trends Genet.10:253-257.

11.Li, Q., M.A. Kay, M. Finegold, L.D. Strat-ford-Perricaudet and S.L.C. Woo. 1993. As-sessment of recombinant adenoviral vectorsfor hepatic gene therapy. Hum. Gene Ther.4:403-409.

12.Lieber, A., M.T.F.D. Vrancken Peeters, L.Meuse, N. Fausto, J. Perkins and M.A. Kay.1995. Adenovirus-mediated urokinase genetransfer induces liver regeneration and allowsfor efficient retrovirus transduction of hepato-cytes in vivo. Proc. Natl. Acad. Sci. USA92:6210-6214.

13.Ponder K., R. P. Dunbar, D.R. Wilson, G.Darlington and S.L.C Woo. 1991. Evalua-tion of relative promoter strength in primaryhepatocytes using optimized lipofection.Hum. Gene Ther. 2:41-52.

14.Ponder, K., S. Gupta, F. Leland, G.Darling-ton, M. Finegold, J. DeMayo, F.D. Ledley,J.R. Chowdhury and S.L.C. Woo. 1991.Mouse hepatocytes migrate to liver parenchy-ma and function indefinitely after intrasplenictransplantation. Proc. Natl. Acad. Sci. USA88:1217-1221.

15.Porteous, D.J. and J.R. Dorin. 1993. Howrelevant are mouse models for human diseaseto somatic gene therapy? Trends Biotechnol.11:173-181.

16.Searle, P.F., B.L. Davison, G.W. Stuart,T.M. Wilkie, G. Norstedt and R.D. Palmi-ter. 1984. Regulation, linkage, and sequenceof mouse metallothionein I and II genes. Mol.Cell. Biol. 4:1221-1230.

17.Smith, T.A., M.G. Mehaffey, D.B. Kayda,J.M. Saunders, S. Yei, B.C. Trapnell, A.McClelland and M. Kaleko. 1993. Aden-

ovirus mediated expression of therapeuticplasma levels of human factor IX in mice. Nat.Genet. 5:397-402.

18.Stratford-Perricaudet, L.D., I. Makeh, M.Perricaudet and P. Briand. 1992. Wide-spread long-term gene transfer to mouseskeletal muscles and heart J. Clin. Invest.90:626-630.

19.Webber, E.M., J.C. Wu, J.C. Wang, G.Merlino and N. Fausto. 1994. Overexpres-sion of transforming growth factor alpha caus-es liver enlargement and increased hepatocyteproliferation in transgenic mice. Am. J.Pathol. 145:398-408.

20.Yang, Y., H.C.J. Ertl and J.M. Wilson.1994. MHC class I-restricted cytotoxic T lym-phocytes to viral antigens destroy hepatocytesin mice infected with E1-deleted recombinantadenoviruses. Immunity 1:433-442.

Received 24 May 1995; accepted 2August 1995.

Address correspondence to:Mark A. KayUniversity of WashingtonDivision of Medical GeneticsCampus Box 357720/M1959 N.E. Pacific StreetSeattle, WA 98195, USAInternet: mkay@u.washington.edu

Vol. 20, No. 2 (1996) BioTechniques 285