Restricted transgene persistence after lentiviral vector-mediated fetal gene transfer in the...

THE JOURNAL OF GENE MEDICINE R E S E A R C H A R T I C L EJ Gene Med 2008; 10: 951–964.Published online 9 July 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jgm.1227

Restricted transgene persistence after lentiviralvector-mediated fetal gene transfer in the pregnantrabbit model

Rafael Moreno1

Marta Rosal2

Itziar Martinez1

Felip Vilardell3

Juan Ramon Gonzalez4

Jordi Petriz5

Edgard Hernandez-Andrade6

Eduard Gratacos6

Josep M. Aran1*

1Medical and Molecular GeneticsCentre, Institut d’InvestigacioBiomedica de Bellvitge, HospitalDuran i Reynals, Barcelona, Spain2Animal Care Facility, ResearchBuilding, Institut de Recerca HospitalUniversitari Vall d’Hebron, Barcelona,Spain 3The Johns Hopkins MedicalInstitutions, Pathology-GI-Liver,Department of Pathology, Baltimore,MD, USA 4Centre for Research inEnvironmental Epidemiology(CREAL) and CIBERESP, Barcelona,Spain 5Biomedical Research Unit,Institut de Recerca HospitalUniversitari Vall d’Hebron, Barcelona,Spain 6Maternal-Fetal MedicineDepartment, Hospital Clinic,University of Barcelona, Barcelona,Spain

*Correspondence to: Josep M. Aran,Centre de Genetica Medica iMolecular, Institut d’InvestigacioBiomedica de Bellvitge (IDIBELL),Hospital Duran i Reynals, Gran Vıas/n km 2,7, 08907 L’Hospitalet deLlobregat, Barcelona, Spain.E-mail: [email protected]

Received: 21 December 2007Revised: 9 May 2008Accepted: 15 May 2008

Abstract

Background Prenatal gene transfer may enable early causal interventionfor the treatment or prevention of many devastating diseases. Nevertheless,permanent correction of most inherited disorders requires a sustained levelof expression from the therapeutic transgene, which could theoretically beachieved with integrating vectors.

Methods Rabbit fetuses received 8.5 × 106 HIV-based recombinantlentivirus particles containing the enhanced green fluorescent protein (EGFP)transgene by intrahepatic, intra-amniotic or intraperitoneal injection at22 days of gestation. Provirus presence and transgene expression in rab-bit tissues were evaluated at both 1.5 and 16 weeks post-in utero interventionby polymerase chain reaction (PCR) and reverse transcriptase-PCR, respec-tively. Moreover, we assessed persistence of EGFP by immunohistochemistry.Enzyme-linked immunosorbent assays confirmed the development of anti-bodies specific against both the viral vector and the reporter protein.

Results Regardless of the route of administration employed, lentiviralvector-based in utero gene transfer was safe and reached 85% of theintervened fetuses at birth. However, the integrated provirus frequency wassignificantly reduced to 50% of that in young rabbits at 16 weeks post-treatment. In these animals, EGFP expression was evident in many tissues,including cytokeratin 5-rich basal cells from stratified and pseudostratifiedepithelia, suggesting that the lentiviral vector might have reached progenitorcells. Conversely, we identified the presence of immune-inflammatoryinfiltrates in several EGFP-expressing tissues. Moreover, almost 70% of thelentiviral vector-treated rabbits elicited a humoral immune response againstthe viral envelope and/or the EGFP.

Conclusions At two-thirds gestational age, the adaptive immune systemof the rabbit appears a relevant factor limiting transgene persistenceand expression following lentiviral vector-mediated in utero gene transfer.Copyright 2008 John Wiley & Sons, Ltd.

Keywords EGFP expression; fetal rabbit; immune response; lentiviral vector;PCR

Copyright 2008 John Wiley & Sons, Ltd.

952 R. Moreno et al.

Introduction

Although the ability to diagnose severe inherited diseases[1] has progressed rapidly, novel prenatal treatmentsthat avoid irreversible clinical manifestations still needto be elucidated. Only a few single gene disorders havebeen treated to date in utero with some success, includ-ing in those individuals undergoing metabolic therapies(congenital adrenal hyperplasia, methylmalonic acidemia,Smith–Lemli–Opitz syndrome, and 3-phosphoglyceratedehydrogenase deficiency) and cell-based therapies (rhe-sus D isoimmunization, osteogenesis imperfecta, immun-odeficiencies, and homozygous alpha thalassemia) [2].Thus, despite receiving bad press, gene therapy usinglong-lasting/integrating vectors remains a viable alterna-tive for a causal and permanent correction of most ofthese wasting diseases.

Recent clinical trials have provided a proof concepttowards the successful correction of both lymphoid [3,4]and myeloid [5] immunodeficiencies using retroviralvector-mediated gene transfer into hematopoietic stemcells. However, the recent occurrence of severe adverseevents encountered in one such trial highlighted the risksassociated with the use of integrating viruses in gene ther-apy [6]. Importantly, prenatal gene therapy entails extrascientific, medical and ethical issues due to the particulardevelopmental features of the fetus and to the fact that,in most countries, termination of pregnancy remains anoption to deal with prenatal diagnosed genetic diseases.Thus, careful optimization of both the safety and theefficacy of fetal gene transfer is required to ensure thatbenefit will be provided with a high level of certainty.

With regard to efficacy, the chosen gene transfer vectorshould provide efficient, lifelong expression of the ther-apeutic transgenic protein in the target diseased tissues.Lentiviral vectors have proven superior to the conven-tional oncoretroviral vectors in terms of gene transferefficiency due to their capability to transduce nondividingcells [7]. The first phase I clinical trial of lentiviral vector-mediated gene transfer has been successfully completed[8]. Moreover, lentiviral vectors have been employed forthe correction of different disease phenotypes in severalanimal models by in utero gene transfer [9–12]. However,integrating retroviral vectors may not provide permanenttransgene expression in the transduced fetal tissues [9].Thus, an active area of research is focused on the studyand improvement of transgene persistence, such as theability to stably transduce self-renewing progenitor cells,the capacity to avoid transgene silencing/extinction, andthe possibility of predisposition to or induction of toler-ance against the transgene product [13]. Hence, novelapproaches addressing these issues in relevant animalmodels are deemed to be extremely useful for a futureimplementation of prenatal gene therapy into clinicalpractice.

We have recently validated the pregnant rabbit as auseful model for retroviral vector-mediated gene transferin utero [14]. In the present study, we further analysed

the efficiency, persistence and overall safety of HIV-basedlentiviral vector-mediated gene transfer to fetal rabbitsthrough several routes of administration, and sought tocorrelate these parameters with the onset of adaptiveimmunity against the vector components and/or trans-genic protein.

Materials and methods

Lentiviral vector production

The replication-defective self-inactivating HIV-1-derivedlentiviral vector used throughout the study, pWPT-EGFP,has been described previously [15].

All plasmids required for assembling the recombinantlentiviruses, including the pWPT-EGFP plasmid, the pack-aging construct psPAX2 expressing the HIV-1 Gag, Pol, Tatand Rev proteins, and pMD2.G encoding the pantropicVSV-G envelope glycoprotein, were kindly provided byDidier Trono (Geneve, Switzerland). The VSV-G pseudo-typed lentiviral particles were produced by transient cal-cium phosphate-mediated three plasmid transfection onsubconfluent 293FT cells (Invitrogen, Carlsbad, CA, USA)[16]. The concentrated lentivirus stock was aliquoted inphosphate-buffered saline (PBS) and stored at −80 ◦Cuntil use.

Viral titers from the concentrated batches were deter-mined by limiting dilution upon transduction of KB3.1target cells and further analysis by flow cytometry [17].The titers were calculated as the percentage of enhancedgreen fluorescent protein (EGFP)-positive cells × 105/µlof the viral stock, and ranged between 1.2 × 108

and 2.6 × 109 TU/ml. Absence of replication competentlentivirus in our lentiviral vector supernatants/stockswas assessed by transduction of phytohaemoglutinin-stimulated human peripheral blood mononuclear cells.Two weeks after transduction, the culture medium wasassayed for the presence of the HIV p24 Gag antigen in theculture medium by enzyme-linked immunosorbent assay(ELISA) (PerkinElmer, Waltham, MA, USA).

In utero intervention

Timed pregnant White New Zealand rabbits (13–15 daysof gestation) were housed for 1 week before surgery. Thehousing was quiet, at constant room temperature andhumidity. Laparotomies were performed at 22 days of ges-tation (gestation = 31–32 days) as described previously[14]. The fetus was located and fixed by external palpationof the semitransparent uterine wall, and intraperitoneal,intra-amniotic or intrahepatic injections were performedusing 26-gauge needles in view of the common anatomicallandmarks of the E20–22 embryo [18]. Preliminary assaysusing Fast Green FCF Stain (Sigma-Aldrich, St Louis,MO, USA) fetal injections [19] estimated the accu-racy rate (intrahepatic = 85%, intraperitoneal = 70%,

Copyright 2008 John Wiley & Sons, Ltd. J Gene Med 2008; 10: 951–964.DOI: 10.1002/jgm

HIV vector fetal gene transfer and immunity 953

intra-amniotic = 90%) and the reproducibility of the tar-geted interventions. To ensure intrahepatic injection, thefetus was punctured at the lower level of the thorax,whereas intraperitoneal delivery was performed punctur-ing the lower half of the abdomen. The intra-amnioticadministration was made directly in the amniotic fluid ofthe fetal sac.

Approximately 1 h before the intervention, thelentivirus suspension was thawed and complementedwith 10 µg/ml polybrene, kept on ice, and warmed toroom temperature previously to its administration intothe individual fetuses in a volume of 100 µl/fetus. In allinstances, a total amount of 8.5 × 106 lentiviral particleswere administered per fetus. Control fetuses were injectedwith an identical volume of PBS plus polybrene. No morethan 50% of the fetuses were treated per uterine horn (thevaginal-end sacs were always excluded). Whenever pos-sible, injection of adjacent fetuses was avoided. Once theinjections were completed, the uterus was returned to theabdominal cavity and the abdomen was closed in layers.Does were kept under a warming blanket until awake andactive, and received subcutaneous injections of meloxi-cam (Metacam; Abbott Laboratories, North Chicago, IL,USA) (0.4 ml per rabbit/24 h) over 48 h as postoperativeanalgesia.

All procedures were conducted in compliance withapplicable regulations and guidelines, and were reviewedand approved by the Animal Care Committee.

Pregnancy termination and tissuecollection

On day 31 of gestation (10 days after intervention), preg-nant rabbits were sacrificed with a euthanizing doseof pentobarbital (150 mg/kg, intravenous). Doe sam-ples from gonads, uterus and liver were frozen for DNAextraction. Experimental pups were retrieved by visualidentification after Caesarean section and weighed. Thosedestined to the short-term study were sacrificed by decap-itation and tissue samples from gonads, kidney, liver,heart, lung, trachea and oesophagus were washed withPBS and quickly frozen in liquid nitrogen for subsequentnucleic acids extraction.

The pups destined for the long-term study were rean-imated, marked with a subcutaneous microchip (AVIDMicrochip ID Systems, Folsom, LA, USA) for furtheridentification, kept in an incubator C100 (Air-Shields,Hatboro, PA, USA) at 32 ◦C for approximately 1 h, andfinally allocated to foster rabbits. Sera from the samesurviving young rabbits were obtained at 6 and 16 weekspost-treatment by bleeding. At 16 weeks after birth, therabbits were sacrificed as described above. Samples weretaken from the same set of tissues as those harvested inthe short-term study, with the addition of bone marrow.A sample fraction was quickly frozen in liquid nitrogenfor nucleic acids extraction, and the remaining fractionwas fixed overnight in formalin and embedded in paraffinfor immunohistochemical evaluation.

DNA and RNA analysis

Genomic DNA was obtained from fetal tissues as describedpreviously [14]. The vector presence on the harvestedorgans was determined by polymerase chain reaction(PCR) amplification of a EGFP reporter gene fragmentcarried out in a 50-µl volume containing 50 ng of genomicDNA, 100 µM dNTPs, 5 µl of 10× Biotaq DNA polymerasebuffer, 1.50 mM MgCl2, 0.3 µl of Biotaq DNA polymerase(5 U/µl) (Bioline, Boston, MA, USA) and 20 pmol ofthe following primers: WPT-EGFP-F (5′-ACC CCG ACCACA TGA AGC AGC-3′) and WPT-EGFP-R (5′-CGT TGGGGT CTT TGC TCA GGG-3′), in a GeneAmp PCR system2400 DNA thermocycler (Perkin-Elmer, Welleslay, MA,USA). A control PCR reaction containing no DNA tomonitor contamination of PCR reagents and a positivecontrol containing vector DNA were included in each PCRanalysis. Amplification conditions involved an initial 10-min denaturation step at 94 ◦C, followed by 42 cycles(94 ◦C for 1 min; 60 ◦C for 30 s; 72 ◦C for 45 s), anda final 7-min extension at 72 ◦C, producing a 417-bpDNA fragment. A 429-bp rabbit skeletal muscle myosinheavy chain-specific DNA fragment (MIO) was amplifiedin parallel using primer pair RMMHC-F/RMMHC-R asdescribed previously [14], and served as a control forgenomic DNA isolation and loading. All samples wereevaluated through 2% agarose gel electrophoresis.

To estimate the proviral copy number in the transducedtissues, a southern blot analysis was performed usinggenomic DNA extracted from the in utero-treated samples.To increase sensitivity, samples were simultaneouslyamplified by a duplex PCR using both the WPT-EGFP-F/WPT-EGFP-R and the RMMHC-F/RMMHC-R primerpairs, and hybridized using the above-referred 417-bpEGFP fragment and 429-bp MIO fragment as probes.Image intensities were quantified through the VersadocImaging System plus Quantity One software (Bio-Rad,Hercules, CA, USA).

A standard curve was generated diluting the pWPT-EGFP plasmid with rabbit genomic DNA at different ratiosranging from 10 : 1–0.01 : 1. These standards were alsoamplified and hybridized with both probes.

Gene expression analysis was assessed in frozenfetal tissues from in utero-treated animals. Total RNAwas isolated using the Qiagen RNA purification kit(Qiagen, Valencia, CA, USA). To avoid genomic DNAcontamination, the total RNA was treated with DNA-free DNase (Ambion, Austin, TX, USA) for 1 h at37 ◦C, followed by cDNA synthesis (3 µg/sample) usingthe Omniscript reverse transcriptase (RT) kit (Qiagen).Quantification of relative EGFP mRNA levels wasperformed by real-time PCR with the LightCyclertechnology (Roche Molecular Biochemicals, Indianapolis,IN, USA) using the FastStart DNA Master SYBR GreenHot Start reaction mix (Roche), the EGFP-specific primersEGFP-F (5′-GCA GAA GAA CGG CAT CAA GGT-3′)and EGFP-R (5′-ACG AAC TCC AGC AGG ACC ATG-3′) and the housekeeping rabbit GADPH-specific primersGADPH-F (5′-TCA CCA TCT TCC AGG AGC GA-3′) and



GADPH-R (5′-CAC AAT GCC GAA GTG GTC GT-3′)for normalization. Amplification efficiencies determinedfor each primer set were similar. Positive controls andnegative controls (water) were included in each run. RT-minus reactions were also performed to assure that DNAwas removed from the samples.

Immunohistochemistry andimmunofluorescence

Formalin-fixed tissues were dehydrated with an ethanolgradient, cut and embedded in paraffin. Sliced 5-µmsections were mounted on poly L-lysine-coated glassslides. Prior to immunohistochemistry, sections weredeparaffinized and rehydrated. Immunogenic retrievalwas performed incubating the slides in 10 mM citratebuffer pH 6.0 for 20 min in a microwave oven.After performing several TBS-T (Tris-buffered salineplus 0.3% Triton X-100) washes, tissue sections wereblocked by 2 h of incubation in 10% fetal bovineserum diluted in washing buffer [TBS plus 1% bovineserum albumin (BSA) and 0.2% Tween 20] at roomtemperature by gently agitation. Further, TBS-T washedsections were incubated overnight at 4 ◦C in a humidchamber with a biotinylated goat polyclonal anti-GFPantibody (Abcam, Cambridge, UK) diluted 1 : 100 inwashing buffer. Specificity controls were performed usingbiotinylated nonimmune goat IgG antibodies (Abcam) atthe same dilution. Next day, the endogenous peroxidasewas blocked with 3% H2O2 for 10 min in the dark.Sections were subsequently washed with PBS andincubated with a biotinylated horseradish peroxidase-avidin complex (ImmunoPure ABC Peroxidase StainingKit; Pierce, Rockford, IL, USA). Finally, sections werestained with the peroxidase substrate diaminobenzidinetetrahydrochloride (Pierce) and counterstained withhematoxylin. The presence of immune-inflammatoryinfiltrates was assessed analogously using the mouse anti-rabbit T cell/neutrophil marker RPN3/57 (Santa CruzBiotechnology, Santa Cruz, CA, USA) plus a secondarybiotin-conjugated goat anti-mouse IgG antibody (Sigma-Aldrich). Moreover, each specimen was routinely stainedwith hematoxylin/eosin to evaluate tissue morphologyand the presence of inflammatory cell infiltrates by anexpert pathologist.

For immunofluorescence analysis, oesophagus and tra-chea samples (7-µm sections) from pWPT-EGFP-treated,and control PBS-treated rabbits were mounted on polyL-lysine-coated glass slides. After being deparaffinizedand rehydrated, the samples underwent antigen retrieval,washing and blocking as described above. Nonspecificbackground signal reduction was performed by incubat-ing samples with Image-iT FX signal enhancer (Invit-rogen) for 30 min at room temperature prior to theblocking step. TBS-T-rinsed sample sections were thenincubated overnight at 4 ◦C in a dark humid chamberwith the primary antibodies: goat polyclonal biotinylated

anti-GFP (Abcam) and mouse anti-human cytokeratin-5, cross-reactive with rabbit cytokeratin-5 (USBiological,Swampscott, MA, USA), both of them diluted 1 : 100 indiluting buffer. For cytokeratin-5 detection, sections wereincubated with Alexa Fluor 546 goat anti-mouse IgG(Molecular Probes, Invitrogen) diluted 1 : 300 in TBS-Tfor 45 min at 37 ◦C. EGFP was visualized by incubationwith ImmunoPure Streptavidin-FITC conjugated (Pierce)at 1 : 50 dilution in water for 30 min at 37 ◦C. Finally,after thorough TBS-T washing, samples were mountedin Vectashield antifade solution (Vector Laboratories,Inc., Burlingame, CA, USA) containing 150 ng/ml of thefluorescent nuclear stain 4′,6-diamidino-2-phenylindole(DAPI). Colocalization signals were evaluated by serialdigital image examination using an SP2 Leica confocallaser microscope (Leica Microsystems GmbH, Wetzlar,Germany).

ELISA assays

Serum from the long-term study rabbits was obtained at6 and 16 weeks of age. Blood was taken from the lateralsaphenous vein and left coagulating for approximately30 min. Samples were centrifuged twice at 3000 g for10 min and keep at −80 ◦C until use.

The presence of a rabbit humoral immune responseagainst the VSV-G-pseudotyped envelope of the lentiviralvector was determined by ELISA. Maxisorb ELISA 96-wellplates (Nunc, Roskilde, Denmark) were coated overnightat 4 ◦C with 5 × 105 viral particles/well diluted in 100 µlof 50 mM carbonate buffer (pH 9.6). After several washeswith PBS-T (PBS plus 0.05% Tween 20), plates wereblocked with 300 µl of 3% fish gelatin (Sigma-Aldrich)diluted in PBS for 2 h at room temperature. Serumsamples diluted 1 : 500 with diluting buffer (PBS-T + 3%BSA) were incubated in triplicate wells for 2 h atroom temperature. The plates were then washed againwith PBS-T and 100 µl of secondary antibody (biotin-conjugated goat anti-rabbit IgG, 1 : 200 000 dilution)(Sigma-Aldrich) was further incubated for 2 h at 37 ◦C.Subsequently, plates were washed with PBS, incubated for30 min with biotinylated horseradish peroxidase-avidincomplex (ImmunoPure ABC Peroxidase Staining Kit;Pierce) and developed using 1-Step Slow TMB (Pierce) for30 min. Finally, the reaction was stopped with 2 M H2SO4and the corresponding absorbances were measured at450 nm on an ELISA plate reader. A standard curve wasprepared by serial dilutions of an anti-VSV-G polyclonalantibody (MBL, Nagoya, Japan) in rabbit serum (pre-diluted 1 : 500 with diluting buffer) to establish the rangeof sensitivity.

The generation of a rabbit humoral immune responseagainst the EGFP reporter protein was assessed analo-gously on Maxisorb ELISA 96-well plates coated overnightat 4 ◦C with 0.1 µg/well of recombinant EGFP (BioVision,Mountain View, CA, USA) diluted in 50 µl of 50 mM

carbonate buffer (pH 9.6). To establish the range of sen-sitivity, a standard curve was prepared by serial dilutionsof a polyclonal anti-GFP antibody (MBL).



Antibody titers, against both the VSV-G-pseudotypedenvelope and the EGFP transgene, in the positive seraof the lentiviral vector-treated animals were assessed bythe above antibody capture ELISA protocol using a serialdilution schedule of the sera (1 : 100–1 : 102 400). Theestimated inflection point observed in the sigmoidal curveafter plotting absorbance (450 nm) versus the logarithmof the dilution factor was considered the serum titer ofthe test sample.

Statistical analysis

Fetal weight comparisons were performed using Student’st-test. The logarithm transformation was used to warrantnormality. The percentages of successfully treated rabbitsfound at 1.5 and 16 weeks post-treatment were comparedusing Fisher’s exact test. Differences between samplesand controls in the ELISA and neutralization assays wereevaluated using a linear mixed model in which replicateswere considered as a random factor [20].

Results

Overall safety of lentiviralvector-mediated fetal intervention inthe pregnant rabbit model

To further improve the efficiency of retroviral vector-mediated fetal gene transfer in the pregnant rabbitmodel, we assessed the performance of an HIV-basedlentiviral vector pseudotyped with the VSV-G protein,and essentially the same protocol as previously describedusing a MoMLV-based retroviral vector [14]. We assayedthree major routes of administration: intrahepatic,intraperitoneal and intra-amniotic.

In terms of survival and regardless of the route ofadministration employed, a total of 85 fetuses from20 pregnant does were intervened (64 were injectedwith recombinant lentivirus and 21 were injected withPBS) and the overall survival rate at birth was closeto 80% and similar between virus-treated and PBS-treated fetuses. Moreover, there were no significantdifferences when comparing survival rates between the

different routes of administration (data not shown). Thislikely reflects a balance between the lack of toxicity ofthe recombinant lentivirus particles employed for fetaladministration and the risk of the surgical procedure.Furthermore, no anatomical defects were detected inany of the treated fetuses. They maintained a correctweight at the termination of pregnancy according to anormal developmental pattern. Interestingly, we foundthat the median weight of the virus-treated fetuses(48.7 ± 1.3 g; n = 53) was slightly higher than that ofthe PBS-treated fetuses (43.7 ± 2.1 g; n = 15) (p = 0.08).This apparently contradictory result reflects a deliberatedbias in the fetal intervention because the virus-treatedanimals were always chosen proximal to the ovarianend of the uterine horn and the PBS-treated animalswere orientated towards cervical end positions in thesame horn. A monotonic decrease in fetal weight withincreasing position number (ovarian to cervical end) hasbeen described, likely reflecting a greater uterine spaceper fetus and a greater blood flow near the oviduct region[21].

Efficiency of lentiviral vector-mediatedfetal gene transfer: short-term study

To assess the efficiency and persistence of lentiviralvector-mediated fetal gene transfer and expression, weundertook both a short- and a long-term study. In theshort-term study we analysed the presence of the provirusin the harvested tissues from 32 injected fetuses at1.5 weeks post-intervention, just before birth. A PCR-based analysis using genomic DNA isolated from thedifferent tissues revealed significant gene transfer fromthe pantropic lentiviral vector regardless of the routeof administration (Table 1). Only four fetuses (oneafter intrahepatic administration and three after intra-amniotic administration) were found to be negativefor the presence of transgenic sequences in all tissuesanalysed. Moreover, tissues from both the control PBS-treated fetuses and the operated does were all negativefor the presence of vector sequences (data not shown). Ingeneral, the expected presence of provirus in the targetorgans coincided with the intended route of lentiviral

Table 1. Efficacy of HIV-1-derived lentiviral vector-mediated fetal gene transfer in the rabbit

Weeks post- Provirus- Lentiviral vector biodistributionb

in utero Delivery positiveintervention route (n)a animals Heart Liver Kidney Lung Trachea Oesophagus Bone Marrow Gonads

1.5 weeks IH (10) 9/10 (90%) 2/10 (20%) 7/10 (70%) 2/10 (20%) 5/10 (50%) 1/4 (25%) 2/4 (50%) – 2/10 (20%)IA (10) 7/10 (70%) 1/10 (10%) 2/10 (20%) 1/10 (10%) 4/10 (40%) 3/8 (37%) 3/8 (37.5%) – 0/9 (0%)IP (7) 7/7 (100%) 1/7 (14%) 2/7 (29%) 2/7 (29%) 5/7 (71%) 0/3 (0%) 0/3 (0%) – 3/6 (50%)

16 weeks IH (5) 3/5 (60%) 2/5 (40%) 2/5 (40%) 2/5 (40%) 1/5 (20%) 1/5 (20%) 0/4 (0%) 1/4 (25%) 1/5 (20%)IA (8) 3/8 (37%) 0/8 (0%) 0/8 (0%) 1/8 (12%) 1/8 (12%) 3/8 (37%) 0/8 (0%) 0/8 (0%) 1/8 (12%)IP (11) 6/11 (54%) 1/11 (9%) 4/11 (36%) 4/11 (36%) 1/11 (9%) 3/11 (27%) 2/11 (18%) 0/11 (0%) 1/11 (9%)

aIH, intrahepatic; IA, intra-amniotic; IP, intraperitoneal. n, number of pWPT-EGFP-treated fetuses. bThe presence of proviral DNA in the different tissuesfrom the in utero-treated fetuses were analysed after genomic DNA extraction and PCR amplification using EGFP-specific primers. For each tissue,frequencies are displayed as number of provirus-positive animals/total number of pWPT-EGFP-treated animals analysed.



vector administration (seven of ten intrahepatically-treated fetuses were positive in liver; five of ten intra-amniotically-treated fetuses were positive in at least one ofthe following tissues: lung, trachea and oesophagus; sevenof seven intraperitoneally-treated fetuses were positivein at least one of the following organs: heart, liver,kidney, and lung). Nevertheless, vector spread to othertissues was evident. Particularly, transgenic sequenceswere detected in the gonads of five fetuses (Table 1).A further genomic analysis of gonadal tissue from thesefetuses using chromosome Y-specific primers revealed thatthe lentiviral vector-mediated transduction of gonadaltissues was not related to the sex of the pups (data notshown).

Persistence of lentiviralvector-mediated fetal gene transfer:long-term study

In the long-term study, rabbit pups were recovered bycesarean extraction 1 day before birth. After reanimation,36 pups were placed to the care of a foster mother and34 survived to adulthood (24 virus-treated and ten PBS-treated). Only two casualties were documented in thevirus-treated group due to nurturer rejection. To assessthe persistence of lentiviral vector-mediated in utero genetransfer, treated young rabbits were sacrificed at 16 weeksof age and freshly extracted tissues were examined forthe presence of the provirus (Table 1). Initially, decreasedgene transfer efficiency was evident in the lentiviralvector-treated animals from the long-term study, reflect-ing a progressive extinction of the integrated proviruswith time. In this case, two of five intrahepatically-treated, five of eight intra-amniotically-treated and fiveof 11 intraperitoneally-treated animals were found to benegative in all tissues analysed. Thus, regardless of theroute of administration, the overall frequency of lentiviralvector integration in the long-term study (12/24; 50%)was significantly reduced compared to that of the short-term study (23/27; 85%). Again, all PBS-treated animalswere also completely negative for the presence of vectorsequences (data not shown). On the other hand, vectorspread followed a similar trend compared to the short-term study. Only three animals, each treated througha different route of administration, were found to bepositive for the transgene in their gonads (Table 1).

We sought to gain further insight on the persistenceof the HIV vector-mediated in utero gene transfer in therabbit model at the DNA level. Thus, we carried outa semiquantitative analysis of the proviral copy num-ber of the integrated vector at both 1.5 and 16 weekspost-transduction. Estimation of the proviral copy num-ber present in rabbit genomic DNA from the differenttissues harvested was performed by PCR amplificationand Southern blotting with an EGFP-specific probe. Ingeneral, the higher copy numbers detected coincidedwith the intended target organ(s) according to theroute of administration in each of the animals analysed

(Table S1). The liver appeared to reach and maintain thehighest mean copy number by intrahepatic administra-tion (2.9 at 1.5 weeks post-treatment; 2.7 at 16 weekspost-treatment). By contrast, lung (0.6 at 1.5 weekspost-treatment; 0.1 at 16 weeks post-treatment), tra-chea (1.9 at 1.5 weeks post-treatment; 0.01 at 16 weekspost-treatment) and oesophagus (1.5 at 1.5 weeks post-treatment; undetectable at 16 weeks post-treatment)appeared to experience a more drastic reduction inproviral copy numbers with time by intra-amniotic admin-istration.

Expression analysis of EGFP:persistence of lentiviralvector-mediated gene expression fromthe in utero-treated rabbits

Transgenic EGFP expression in the different tissues har-vested from the long-term in utero-treated animals wasdetermined both at the RNA and protein level. Real-timeRT-PCR detected EGFP transcripts while immunohisto-chemistry, using an anti-EGFP antibody to enhance EGFPvisualization and avoid the high background fluorescencepresent in some of the analysed tissues, detected EGFPprotein. The evaluation was performed in most of thePCR-positive animals containing the integrated provirus,including all those assessed by Southern blotting, andregardless of the route of administration employed. More-over, tissues from PBS-treated rabbits were also processedfor control purposes.

Overall, the EGFP mRNA transcript level in the differenttissues after intrahepatic (Figure 1), intra-amniotic andintraperitoneal administration (data not shown) corre-lated with the average DNA copy number biodistribution,and corroborated vector spread as reported in similarstudies [22].

Regarding EGFP reporter expression, in utero intrahep-atic administration of recombinant lentiviruses resultedpredominantly in patches of relatively homogeneous andunambiguous cytoplasmic EGFP staining within the hepa-tocytes (Figure 2). In utero intra-amniotic administrationled to fetal breathing and swallowing of the recombinantlentivirus particles, with concomitant EGFP expression inthe lung and trachea (Figure 2). In lung, strong stainingwas visible mainly at the surface epithelial cells from thebronchiolar tree, and to a lesser extent in the alveoli.EGFP expression was also evident in the airway epithelialcells of the trachea. In utero intraperitoneal administra-tion yielded EGFP expression in several internal organsdepending on the treated rabbit, including liver, kidney,heart, lung, trachea (data not shown) and oesophagus(Figure 2). Moreover, the estimated proviral copy num-ber seemed to correlate with the level of EGFP expression



Rel

aviv

e qu

antit

y of

EG

FP

mR

NA

n.d. n.d. n.d.

1.5 weeks

16 weeks103

104

Heart Liver Kidney Lung Trachea Esophagus Gonads100

101

102

Tissue

Figure 1. Relative level of EGFP mRNA expression after lentiviral vector-mediated intrahepatic administration into fetal rabbits.Total RNA was isolated from the indicated tissues at 1.5 weeks (white bars) and 16 weeks (black bars) post-in utero lentiviralvector administration, and the relative level of EGFP mRNA transcripts was estimated using real-time PCR. All amplifications wereperformed in triplicate in two different runs. n.d., not determined

IH

IA

Liver

Lung

Trachea

IP

Liver

Kidney

Heart

Lung

pWPT-EGFP PBS pWPT-EGFP PBS

Esophagus

pWPT-EGFP PBS

Figure 2. Long-term immunohistochemical assessment of transgenic EGFP expression after lentiviral vector-mediated in uterogene transfer. Immunostaining for EGFP-expressing cells was performed on tissues from pWPT-EGFP-in utero treated animals at16 weeks post-transduction (left paired images, ×40). PBS-treated animals were used for control purposes (right paired images,×40). Microscopic analysis of representative tissue sections according to the route of administration. EGFP-positive cells appearbrown: IH, intrahepatic administration. Hepatocyte and ductal staining; IA, intra-amniotic administration. Staining visible onlung sections (mostly in the bronchia) and trachea sections (especially evident in basal epithelial cells); IP, intraperitonealadministration. Positive staining in internal organs including liver (in hepatocytes), kidney (on tubular epithelial cells), heart (inthe striated muscle fibers), lung (around bronchioli) and oesophagus (in the basal and intermediate cell layers). All tissue sectionswere counterstained with hematoxylin



for most of the analysed tissues. Thus, low EGFP reactivitywas evident below 0.01 provirus copies/rabbit genome.

Conversely, several long-term in utero-treated animalsexhibited EGFP-expressing tissues with signs of immune-inflammatory activity. Thus, the presence of cellularinfiltrates was perceptible in some liver areas of intraperi-toneally treated rabbits (Figure S1), but also on intra-hepatically treated rabbits (data not shown) as poly-morphonuclear aggregates. Kidney sections from someintraperitoneally-treated rabbits indicated the occurrenceof discrete lymphocyte and/or polymorphonuclear infil-trates in the proximal tubules and in the glomerule, alongwith weak EGFP reactivity in the collector tubules. Inaddition, the tracheas from some intraperitoneally-treatedrabbits presented evidence of immune-inflammatory infil-trates in both the fibrous layer and the surface epithelia(data not shown). Nevertheless, despite the suggestionof a cellular immune response in several treated rab-bits, long-term expression of EGFP mRNA and reporterprotein in many tissues indicated significant lentivi-ral vector-mediated fetal gene transfer and appreciablepreservation of the transgene in the pregnant rabbitmodel.

Persistent EGFP expression is likely tobe related to lentiviral vector-mediatedtransduction of fetal progenitor cells

We focussed on oesophagus and trachea to assesswhether our lentiviral vector-mediated fetal treatmentreached self-renewing progenitor cells able to sustain thelong-term EGFP reporter expression observed in sometissues of the treated adult rabbits. Both tracheal andesophageal epithelial tissues form clearly distinguishablepseudostratified and stratified cell layers, respectively,in which cellular differentiation through different celllineages follows a vectorial process. Recent studiesindicate that basal epithelial cells are able to evolveinto/restore a fully differentiated esophageal squamousepithelium or a complete tracheal airway epithelium.These cytokeratin 5-expressing cells include subsetscapable of either multipotent or unipotent differentiationin vivo [23]. Thus, we performed an immunophenotypicidentification of the long-term EGFP-expressing cellsin tracheal and esophageal sections of the in utero-treated animals by using endogenous cytokeratin 5 asa basal cell marker. Fluorescent colocalization of both

Cytokeratin 5

EFGP

Merge

HematoxilinL

SE

BM

LP

L CE

LP

BM

CE

LP BML

SEBM

LP

P

Esophagus TracheapWPT-EGFP pWPT-EGFPPBS PBS

L

Figure 3. Immunofluorescent localization of EGFP expression within cytokeratin 5-rich basal cells of epithelial tissues afterlong-term lentiviral vector-mediated in utero gene transfer. Confocal microscopy images of representative intermediate xy sections(1 µm) showing esophageal and tracheal epithelium from pWPT-EGFP lentiviral vector-transduced rabbits (left paired images),and from PBS-treated control rabbits (right paired images). EGFP-based fluorescence is depicted in green; cytokeratin 5-basedfluorescence is depicted in red. DAPI-stained nuclei are shown in blue. Colocalization of EGFP expression and cytokeratin 5expression in the basal cells of the stratified esophageal epithelia, and the pseudo-stratified tracheal epithelia, are visualized inyellow (for details, see insets in the merged images). Fluorescent basal cells (white arrows) are located just above the basal lamina.The bottom images correspond to adjacent hematoxilin-stained sections from the above-visualized tissues. L, lumen; P, papillae;SE, squamous epithelium; CE, columnar epithelium; LP, lamina propria; BM, basement membrane. Original magnifications, ×63



a subset of EGFP immunoreactivity and cytokeratin5 immunoreactivity suggests efficient transduction andEGFP expression of cytokeratin 5-positive basal progenitorcells within the esophageal squamous epithelium and thetracheal epithelium (Figure 3).

Evidence of humoral immunity againstvector envelope and EGFP transgeneproteins after lentiviralvector-mediated fetal intervention

Regarding our previous results inferring the presenceimmune-inflammatory infiltrates in some tissues fromin utero-treated rabbits, we wished to assess whether theyoung in utero-treated rabbits also developed a humoralimmune response against the viral envelope (notably VSV-G protein) from the pseudotyped HIV-based vector and/oragainst the reporter transgene product EGFP. SpecificELISA assays to detect the presence of antibodies againstboth the viral envelope and the EGFP proteins wereperformed on sera harvested from the young rabbits at 6and 16 weeks post-in utero treatment (Figure 4). Indeed,67% of the animals (four of five through intrahepaticadministration; four of eight through intra-amnioticadministration and eight of 11 through intraperitoneal

administration) developed antibodies against either orboth the viral envelope and the EGFP reporter proteinalbeit at different titers (Table 2). In four of the remainingeight rabbits negative for both anti-viral envelope andanti-EGFP antibodies, the presence of provirus could notbe detected in any of the tissues analysed.

Interestingly, the number of animals developing anti-EGFP antibodies (15 from 24 animals analysed; 62.5%)was higher than those developing anti-viral envelope

Table 2. Antibody titers against viral envelope and EGFP proteingenerated through lentiviral vector-mediated fetal gene transfer

Titerb

Route ofdelivery (n)a 6 weeks 16 weeks

EGFP IH (4) 0 3.2 ± 0.8 (2.2–5.5)IA (4) 1.4 ± 1.4 (0–4.3) 3.4 ± 0.4 (2.2–4.3)IP (7) 0.7 ± 0.7 (0–4.2) 3.0 ± 0.2 (2.3–3.8)

Viral envelope IH (1) 0 2.9IA (1) 2.2 2.7IP (3) 1.3 ± 0.8 (0–2.7) 0.8 ± 0.8 (0–2.4)

aIH, intrahepatic; IA, intra-amniotic; IP, intraperitoneal. n, number ofantibody-positive animals. bSerial dilutions of sera from the in uteropWPT-EGFP-treated rabbits were analysed using the ELISA proceduredescribed in the Materials and Methods. Titers are given as mean ± SEM(range).

EGFP

Viral Envelope

IH 6 weeks IH 16 weeks IA 6 weeks IA 16 weeks IP 6 weeks IP 16 weeks0123456789

10

Rat

io

IH IA IP

IH 6 weeks IH 16 weeks IA 6 weeks IA 16 weeks IP 6 weeks IP 16 weeks0123456789

10

Rat

io

IH IA IP

Figure 4. Lentiviral vector-mediated in utero gene transfer evokes a humoral immune response against both the viral envelope andthe transgene product. Serum samples from the lentiviral vector-treated rabbits were collected at 6 and 16 weeks post-interventionand the presence of antibodies against both the VSV-G-pseudotyped envelope and the transgenic EGFP protein was analysedby ELISA. The horizontal lines refer to the mean ± SD serum values obtained from control, PBS-injected rabbits (n = 6). IH,intrahepatic administration; IA, intra-amniotic administration; IP, intraperitoneal administration



antibodies (five from 24 animals analysed; 21%).Moreover, the antibody titers reached against the EGFPprotein tended to be higher than those developed againstthe viral envelope proteins (Table 2). Furthermore,although the titer of the anti-viral envelope antibodiestended to increase or decrease within the time frameanalysed, depending on the treated animal, all rabbits thatmounted an antibody response against the EGFP proteinpresented sustained or mostly increased antibody titerswith time. These results likely reflect the development of amore potent humoral immune response against the steadyexpression of the EGFP immunogen than that achievedby a single pulse of the viral envelope immunogen at thetime of lentiviral transduction.

In a few in utero-treated animals, the anti-viralenvelope antibodies had neutralizing activity againstthe VSV-G pseudotyped lentiviral particles (three fromfive animals with a positive anti-viral envelope antibodytiter, one for each administration route; 60%) (data notshown). This humoral response might preclude a furtherefficient transduction of their tissues by recombinant VSV-G-pseudotyped viruses. Remarkably, the highest anti-viralenvelope titers from these animals were measured at16 weeks post-in utero treatment. None of the analysedsera from the PBS-injected rabbits showed neutralizingactivity against the recombinant lentivirus (data notshown).

Finally, the appearance of tissular immune-inflam-matory infiltrates seemed to correlate with the presenceof anti-EGFP antibodies in the analysed in utero-treatedrabbits. Relatedly, up to eight in utero-treated animals(two intrahepatically, three intra-amniotically and threeintraperitoneally) that were seropositive for antibodiesagainst EGFP were found to be negative for integratedproviral sequences in all tissues analysed. This suggeststhat, in these treated rabbits, the successfully transducedcells disappeared from the tissues: (i) because noprogenitors were transduced, resulting in loss of long-term EGFP expression due to cell turnover; (ii) bythe acquired immune response generated against thesustained expression of the foreign EGFP transgene; or(iii) for both reasons. If we add these lentiviral vector-negative, EGFP antibody-positive animals to the lentiviralvector-positive animals detected by PCR, both analysedat 16 weeks post-treatment, the corrected frequency ofgene transfer in the long-term study would reach 20/24(83%). This is close to the overall gene transfer frequencyobtained in the short-term study: 23/27 (85%) (Figure 5).

Discussion

Prenatal gene therapy represents a step ahead ofpostnatal gene therapy in terms of technical, physiologicaland biological advantages towards the prevention,amelioration or resolution of a number of genetic diseases[24,25]. In the present study, we further validate thepregnant rabbit as a useful middle-sized animal model

16 weeks 1.5 weeks

0

20

40

60

80

100

PCR(+)

PCR(+) +

PCR(-)

/Ab(

+)

PCR(+)

12/24

20/24 23/27*

% T

reat

ed R

abbi

ts

*

Figure 5. Percentages of successfully treated rabbits afterin utero lentiviral vector administration. Prenatally treated rab-bits were analysed at 1.5 weeks (black bar) for the presence ofthe proviral EGFP transgene sequence [PCR(+)], and at 16 weeks(white bars) for the presence of both the proviral EGFP transgenesequence and anti-EGFP antibodies [PCR(+) + PCR(−)/Ab(+)

groups both EGFP transgene-positive rabbits plus EGFP trans-gene-negative rabbits, which, however, developed anti-EGFPantibodies]. Frequencies resulting from each of the treatmentconditions are displayed on top of the bars. ∗No statistical dif-ference was detected between PCR(+) + PCR(−)/Ab(+) animalsanalysed at 16 weeks post-treatment and PCR(+) animals anal-ysed at 1.5 weeks post-treatment (Fisher’s exact test, p = 1.0)

for optimizing fetal gene transfer strategies. Throughseveral routes of vector administration, we show thesuperior transduction efficiency of a recombinant HIV-based lentivirus over the oncoretroviral vector used in aprevious study [14].

In terms of safety, the overall low fetal mortalityof our in utero interventions appears to confirm thelow short-term toxicity of lentiviral vectors for prenatalgene delivery. We have not detected any evidence ofmalformations or abnormal development in any of ourHIV vector-treated fetuses. A recent study has reported ahigh incidence of nonprimate lentiviral vector-associatedtumorigenesis following in utero gene transfer in mice[26]. This phenomenon might be related to the insertionalmutagenesis capability of lentiviral vectors, and to theincreased transcriptional activity of development and cellgrowth genes during fetal life. During the timeframeof our study (up to 4 months), all our in utero-treatedrabbits remained healthy and we have not found anyindication of oncogenic events on them, although wecannot exclude the potential development of tumorsat later time points. This important biosafety issueshould warrant a careful comparative analysis of lentiviralvector integration sites in a comprehensive genotoxicitystudy once the rabbit genome, already scheduled fordeep coverage by the mammalian genome project(http://www.broad.mit.edu/mammals/), becomes fully



sequenced. Nevertheless, a recent report indicates thatlentiviral vector integration in hematopoietic stemcells from a tumor-prone mouse model uncovers lowgenotoxicity, in contrast to the tumorigenic cooperatingactivity exhibited by retroviral vectors [27].

The risks of maternal tissue and fetal germ linetransgenic sequence transmission represent other safetyissues being addressed in recent in utero gene transferstudies using integrating vectors. We have been unable todocument horizontal gene transfer in any of the analysedtissues from the intervened does at the time of delivery.By contrast, we have detected the presence of transgenicsequences in the gonads of a few lentiviral vector-treatedanimals, albeit at low frequency. Although the presence ofproviral sequences within the reproductive organs of ourin utero-treated rabbits may not necessarily be the resultof transduction of the germ line, as recently suggestedin several studies [28–30], additional experiments arenecessary to discard this possibility.

Regarding the overall efficacy of the HIV vector-mediated fetal gene transfer in the rabbit model, our dataindicate significant gene transfer efficiency in most of thetissues analysed, regardless of the route of administrationemployed. Intra-amniotic administration is particularlyappropriate for reaching pulmonary, gastrointestinal andsinus apical epithelia, as we have confirmed by significanttransduction of the mucous membranes of the airwaysand the upper digestive tract. These tissues are especiallyrelevant for the treatment of congenital lung diseases,such as cystic fibrosis, where early intervention mayfavour disease prevention or avoid important obstaclesfor vector efficacy, such as the presence of mucusand inflammatory mediators after disease onset [31,32].Conversely, in utero intrahepatic transgene delivery couldbe advantageous for the treatment of a number ofinborn metabolic disorders and other conditions notaffecting the liver itself, in which a systemic delivery ofsoluble secreted proteins (clotting factors, erythropoietin,cytokines, growth factors, etc.) would alleviate or preventinjury to the diseased tissues. Alternatively, yolk sac vesselinjection might also provide a more homogeneous, lesstraumatic, and perhaps less immunogenic transduction offetal hepatocytes [9,33].

Nevertheless, certain vector delivery inconsistency(vector misdelivery and/or spread) was evident in thein utero-treated animals. This confounding factor has beenpreviously manifested [10,34,35] and could be reducedthrough more accurate targeted delivery technologies,such as ultrasound-guided fetal gene transfer [36,37], orby targeting lentiviral vectors to specific cell types in vivo,either naturally through suitable pseudotype envelopes[34], or synthetically, by incorporating a specific antibodyand a fusogenic protein as two distinct molecules into thelentiviral surface [38]. Otherwise, the use of tissue-specificand/or regulated promoters to drive transgene expressionmay also limit its ‘leakiness’ in the in utero-treated host[39,40].

We achieved 4-month prolonged expression of thetransferred transgene in about half of the young treated

rabbits. Certainly, one of the major aims of the presentstudy was the analysis of factors affecting persistenceof the integrated provirus and its expressed transgeneproduct, EGFP, in the host rabbit tissues. In the future, lessinvasive methodologies to detect transgene presence andactivity than those employed in the present study, suchas bioluminescence imaging [41], might allow a directfollow-up of in vivo transgene persistence and expressionat different time points post-treatment with adequatespecificity and sensitivity.

A necessary condition to achieve long-term, sustainedtransgene expression from integrating vectors, such asrecombinant lentiviruses, involves the transduction ofstem cells capable to both self-renew and differentiateinto specialized tissues. This is particularly evident inrapidly proliferating or in high cell turnover settingssuch as in fetal development, in adult bone marrow orin adult epithelial cells from several sources. We showthat lentiviral vector-mediated in utero treatment leadsto long-term, sustained EGFP expression in cytokeratin5-rich basal cells from the oesophagus and trachea,suggesting that recombinant lentiviruses are able to reachand successfully transduce progenitor cells from bothstratified and pseudostratified epithelia, respectively. Bothprogenitor cells from the esophageal squamous epitheliumand multipotent tracheal cells appear to be present inthe cytokeratin 5-expressing compartment. It has beenrecently shown that transgenic mice, in which EGFP wasdriven by the cytokeratin 5 promoter, expressed EGFPin most basal cells of the body, including a subset oftracheal basal cells thought to be airway epithelial stemcells [42,43].

Nevertheless, both lack of progenitor cell trans-duction and tissue turnover -particularly in epithelialtissues reached through intra-amniotic administration(Table S1)-, could significantly contribute to gradualprovirus loss during neonatal development. Although thedevelopmental status of both oesophagus and trachea inthe rabbit fetus has not yet been well characterized, arecent study in fetal rabbits has confirmed the bioavail-ability of lentiviral vectors through amniotic fluid [35].However, we have not analysed gut delivery beyond theoesophagus. Conversely, it is known that pseudostratifiedepithelia represent a major accessibility hurdle for basalcell transduction from apically applied virus [44].

Another important requirement to ensure persistenttransgene expression in the successfully transducedtissues is the maintenance of an active transcriptionalstate of the integrated lentiviral vector. However,transcriptional extinction of the lentivector in our long-term study appears to have a minor contribution ontransgene expression shut-down because virtually alltissues with integrated provirus had evidence of EGFPexpression, albeit at different levels depending on thecopy number achieved.

Lastly, the ultimate epigenetic determinant for expres-sion longevity from viral vectors is the absence of animmune response against the vector components or thetransgenic protein. Numerous reports propose that both



fetuses and neonates are immune privileged in thattheir immune systems are much less likely than theadult immune system to develop a vigorous adaptiveimmune response towards a transgenic insult [13,24,45].This is most probably due to the developmental imma-turity or naivety of the cells participating in the immuneresponse. By contrast, we report that more than two-thirds of in utero-treated rabbits from the long-term studywere capable of mounting a humoral immune responseagainst the vector envelope and/or the transgenic EGFPprotein following lentiviral vector delivery, albeit at dif-ferent degrees, perhaps reflecting the rabbit immunedevelopment borderline at the time of our fetal inter-vention. Moreover, we portray histological evidence ofthe presence of immune-inflammatory infiltrates in theEGFP-expressing tissues.

Two main aspects could explain the divergencesbetween our results and others reporting the induction oftolerance to the vector/transgene after direct injection ofretroviral, lentiviral or even adenoviral vectors into fetuses[11,46,47]. The first aspect relates to the stage of fetaldevelopment at the time of intervention. Although all ourinterventions were performed at two-thirds gestationalage (in utero application: day 22/gestation length: day32), the maturity of the rabbit immune system atthis developmental stage appears to be sufficient toelaborate a complex immune response against both asingle challenge with the viral envelope immunogen, anda continuous expression of the EGFP immunogen. Indeed,some of the antibodies raised against the viral envelopepresented neutralizing activity, which might preventfuture VSV-G-pseudotyped lentiviral vector challenge.Conversely, lentiviral vector-mediated continuous EGFPexpression did not lead to tolerization, as has beenpreviously suggested [48], but, instead, sustained avigorous immune response with time. Thus, our lentiviralvector administration was likely carried out after the‘pre-immune’ stage (which would prevent the immunehypo-responsiveness characterizing this phase of therabbit development) leading perhaps to a transienttype I interferon response and, upon reaching antigen-presenting cells (APCs), would elicit an anti-EGFPimmune response and the progressive elimination oftransduced cells. Indeed, it has been recently shown thatin vivo delivery of lentiviral vectors is able to triggerboth early, transient and cell type-specific innate immuneresponses restricting gene transfer and promoting vectorclearance [49], and further adaptive immune responsesresulting in transgene-specific immunity and clearance oftransduced cells [50,51].

The second aspect that may account for inconsisten-cies in tolerance induction is the animal model used. Aprofound increase in T lymphocytes in the human fetal cir-culation between 12 and 14 weeks of pregnancy has beenreported [52]. By contrast, adenoviral vectors adminis-tered in newborn mice achieved long-term correction oftype VII mucopolysaccharidosis [53], and neonatal genetransfer with a retroviral vector resulted in tolerance tohuman factor IX in mice and dogs [48]. Thus, it would

appear that, taking into account interspecies anatomi-cal and functional comparisons related to the immunesystem development, such as the occurrence of splenicprimordial or bone marrow hematopoiesis [54], the rab-bit would exhibit an intermediate susceptibility to immunetolerization towards transgenic proteins or vector compo-nents with respect to that observed among mice (highsusceptibility) and humans (low susceptibility) at a com-parable stage of development. Nevertheless, additionalexperiments are needed to address whether lentiviralvector-mediated in utero interventions at earlier stagesof rabbit development would lead to increased stem celltransduction and/or a higher tolerance frequency amongthe treated rabbits, and to ascertain whether species-differences in immune system development account forthe divergences observed.

Finally, the immunogenicity of the transgene productsurely plays a role in the degree of immune responsedevelopment alter fetal gene transfer. A recent fetaland neonatal gene therapy study in rats has shownthat, although all animals injected with the UDP-glucuronosyltransferase (UGT1A1)-expressing lentiviralvector developed antibodies to UGT1A1, animals injectedin parallel with EGFP-expressing lentiviral vector didnot develop antibodies to EGFP [55]. This indicatesthat tolerization depends on the immunogenicity of thetransgenic protein expressed.

In summary, the present study confirms the usefulnessof the pregnant rabbit model to optimize fetal genetransfer interventions. Thus, significant, long-lasting genetransfer and expression able to reach progenitor cellscan be achieved after lentiviral vector-mediated fetaltransduction through different routes of administration.The availability of these delivery alternatives influencesthe pattern of proviral distribution and transgeneexpression. Conversely, although retroviral and lentiviralvectors are assumed to elicit very little immune responsescompared to other viral vectors, to our knowledge,the present study is the first to document the fetalelicitation of an adaptive response against both a vectorcomponent and the transgenic protein mediated bya recombinant lentivirus as a relevant factor able toaffect transgene persistence. Hence, sustained expressionof a moderate amount of EGFP protein and/or asingle, short time exposure to the VSV-G-pseudotypedlentiviral vector in the fetal rabbit APCs were capableof boosting the fetal rabbit immune system, with thefinal outcome of gene transfer acting as a ‘vaccination’against the EGFP reporter or the viral envelope thatcould induce the progressive destruction of treated cells,limiting the persistence of the transgene and the overallefficacy of the prenatal gene transfer. Consequently,effective tolerization strategies to ensure efficient andstable fetal gene transfer and sustained transgeneexpression may arise from: (i) early intervention at a‘preimmune’ stage of fetal development, using low-immunogenicity lentiviral particles; or, alternatively, inimmunocompetent hosts (ii) by temporary administrationof neutralizing IFNαβ antibodies, IFNαβR or PRR



antagonists [49]; (iii) by restricting lentiviral transductionand/or transgene expression to nonhematopoietic cells[51,56]; or (iv) by adoptive transfer of transgene-expressing APCs [50].

Acknowledgements

This work was supported by grant 01/1417 from the Fondode Investigaciones Sanitarias (FIS), and by a fellowship from‘Associacio Catalana de Fibrosi Quıstica’ to I.M. J.M.A. issponsored by the Researchers Stabilization Program from theSNS. R.M. and J.M.A. are members of the Red de Centros delInstituto de Salud Carlos III (Ref. C03/07). We are grateful toChris Maxwell for critically reading the manuscript.

Supplementary Material

The supplementary electronic material for this paperis available in Wiley InterScience at: http://www.intersci-ence.wiley.com/jpages/1099-498X/suppmat/.

References

1. Bischoff FZ, Sinacori MK, Dang DD, et al. Cell-free fetal DNAand intact fetal cells in maternal blood circulation: implicationsfor first and second trimester non-invasive prenatal diagnosis.Hum Reprod Update 2002; 8: 493–500.

2. O’Brien B, Bianchi DW. Fetal therapy for single gene disorders.Clin Obstet Gynecol 2005; 48: 885–896.

3. Aiuti A, Slavin S, Aker M, et al. Correction of ADA-SCID by stemcell gene therapy combined with nonmyeloablative conditioning.Science 2002; 296: 2410–2413.

4. Hacein-Bey-Abina S, Le Deist F, Carlier F, et al. Sustainedcorrection of X-linked severe combined immunodeficiency byex vivo gene therapy. N Engl J Med 2002; 346: 1185–1193.

5. Ott MG, Schmidt M, Schwarzwaelder K, et al. Correction ofX-linked chronic granulomatous disease by gene therapy,augmented by insertional activation of MDS1-EVI1, PRDM16or SETBP1. Nat Med 2006; 12: 401–409.

6. Hacein-Bey-Abina S, Von Kalle C, Schmidt M, et al. LMO2-associated clonal T cell proliferation in two patients after genetherapy for SCID-X1. Science 2003; 302: 415–419.

7. Lewis P, Hensel M, Emerman M. Human immunodeficiencyvirus infection of cells arrested in the cell cycle. EMBO J 1992;11: 3053–3058.

8. Levine BL, Humeau LM, Boyer J, et al. Gene transfer in humansusing a conditionally replicating lentiviral vector. Proc Natl AcadSci USA 2006; 103: 17372–17377.

9. Han XD, Lin C, Chang J, et al. Fetal gene therapy of alpha-thalassemia in a mouse model. Proc Natl Acad Sci USA 2007;104: 9007–9011.

10. Seppen J, van der Rijt R, Looije N, et al. Long-term correction ofbilirubin UDPglucuronyltransferase deficiency in rats by in uterolentiviral gene transfer. Mol Ther 2003; 8: 593–599.

11. Waddington SN, Nivsarkar MS, Mistry AR, et al. Permanentphenotypic correction of hemophilia B in immunocompetentmice by prenatal gene therapy. Blood 2004; 104: 2714–2721.

12. Williams ML, Coleman JE, Haire SE, et al. Lentiviral expressionof retinal guanylate cyclase-1 (RetGC1) restores vision in anavian model of childhood blindness. PLoS Med 2006; 3: E201.

13. Waddington SN, Kennea NL, Buckley SM, et al. Fetal andneonatal gene therapy: benefits and pitfalls. Gene Ther 2004;11(Suppl 1): S92–S97.

14. Moreno R, Rosal M, Cabero L, et al. Feasibility of retroviralvector-mediated in utero gene transfer to the fetal rabbit. FetalDiagn Ther 2005; 20: 485–493.

15. Salmon P, Kindler V, Ducrey O, et al. High-level transgeneexpression in human hematopoietic progenitors and

differentiated blood lineages after transduction with improvedlentiviral vectors. Blood 2000; 96: 3392–3398.

16. Klages N, Zufferey R. Trono DA stable system for the high-titerproduction of multiply attenuated lentiviral vectors. Mol Ther2000; 2: 170–176.

17. Aran JM, Gottesman MM, Pastan I. Construction andcharacterization of bicistronic retroviral vectors encodingthe multidrug transporter and beta-galactosidase or greenfluorescent protein. Cancer Gene Ther 1998; 5: 195–206.

18. Beaudoin S, Barbet P, Bargy F. Developmental stages in therabbit embryo: guidelines to choose an appropriate experimentalmodel. Fetal Diagn Ther 2003; 18: 422–427.

19. Boyle MP, Enke RA, Adams RJ, et al. In utero AAV-mediatedgene transfer to rabbit pulmonary epithelium. Mol Ther 2001;4: 115–121.

20. Pinheiro J, Bates D. Mixed-Effects Models in S and S-PLUS.Springer-Verlag: New York, NY, 2000.

21. Stuckhardt JL, Brunden MN, Harris SB. Influence of intrauterineposition on fetal weight in Dutch belted rabbits. J Toxicol EnvironHealth 1981; 8: 777–786.

22. Jimenez DF, Lee CI, O’Shea CE, et al. HIV-1-derived lentiviralvectors and fetal route of administration on transgenebiodistribution and expression in rhesus monkeys. Gene Ther2005; 12: 821–830.

23. Hong KU, Reynolds SD, Watkins S, et al. In vivo differentiationpotential of tracheal basal cells: evidence for multipotent andunipotent subpopulations. Am J Physiol Lung Cell Mol Physiol2004; 286: L643–L649.

24. Coutelle C, Themis M, Waddington SN, et al. Gene therapyprogress and prospects: fetal gene therapy – first proofsof concept – some adverse effects. Gene Ther 2005; 12:1601–1607.

25. Waddington SN, Kramer MG, Hernandez-Alcoceba R, et al.In utero gene therapy: current challenges and perspectives. MolTher 2005; 11: 661–676.

26. Themis M, Waddington SN, Schmidt M, et al. Oncogenesisfollowing delivery of a nonprimate lentiviral gene therapy vectorto fetal and neonatal mice. Mol Ther 2005; 12: 763–771.

27. Montini E, Cesana D, Schmidt M, et al. Hematopoietic stem cellgene transfer in a tumor-prone mouse model uncovers lowgenotoxicity of lentiviral vector integration. Nat Biotechnol 2006;24: 687–696.

28. Lee CC, Jimenez DF, Kohn DB, et al. Fetal gene transfer usinglentiviral vectors and the potential for germ cell transduction inrhesus monkeys (Macaca mulatta). Hum Gene Ther 2005; 16:417–425.

29. Porada CD, Park PJ, Tellez J, et al. Male germ-line cells are atrisk following direct-injection retroviral-mediated gene transferin utero. Mol Ther 2005; 12: 754–762.

30. Porada CD, Tran N, Eglitis M, et al. In utero gene therapy:transfer and long-term expression of the bacterial neo(r) gene insheep after direct injection of retroviral vectors into preimmunefetuses. Hum Gene Ther 1998; 9: 1571–1585.

31. Cohen JC, Larson JE. Cystic fibrosis transmembraneconductance regulator (CFTR) dependent cytoskeletal tensionduring lung organogenesis. Dev Dyn 2006; 235: 2736–2748.

32. Ferrari S, Geddes DM, Alton EW. Barriers to and newapproaches for gene therapy and gene delivery in cystic fibrosis.Adv Drug Deliv Rev 2002; 54: 1373–1393.

33. Waddington SN, Mitrophanous KA, Ellard FM, et al. Long-termtransgene expression by administration of a lentivirus-basedvector to the fetal circulation of immuno-competent mice. GeneTher 2003; 10: 1234–1240.

34. MacKenzie TC, Kobinger GP, Kootstra NA, et al. Efficienttransduction of liver and muscle after in utero injection oflentiviral vectors with different pseudotypes. Mol Ther 2002;6: 349–358.

35. Skarsgard ED, Huang L, Reebye SC, et al. Lentiviral vector-mediated, in vivo gene transfer to the tracheobronchial treein fetal rabbits. J Pediatr Surg 2005; 40: 1817–1821.

36. Peebles D, Gregory LG, David A, et al. Widespread and efficientmarker gene expression in the airway epithelia of fetal sheepafter minimally invasive tracheal application of recombinantadenovirus in utero. Gene Ther 2004; 11: 70–78.

37. Themis M, Schneider H, Kiserud T, et al. Successful expressionof beta-galactosidase and factor IX transgenes in fetaland neonatal sheep after ultrasound-guided percutaneousadenovirus vector administration into the umbilical vein. GeneTher 1999; 6: 1239–1248.



38. Yang L, Bailey L, Baltimore D, et al. Targeting lentiviral vectorsto specific cell types in vivo. Proc Natl Acad Sci USA 2006; 103:11479–11484.

39. Galimi F, Saez E, Gall J, et al. Development of ecdysone-regulated lentiviral vectors. Mol Ther 2005; 11: 142–148.

40. Lois C, Hong EJ, Pease S, et al. Germline transmission andtissue-specific expression of transgenes delivered by lentiviralvectors. Science 2002; 295: 868–872.

41. Brovko LY, Griffiths MW. Illuminating cellular physiology:recent developments. Sci Prog 2007; 90: 129–160.

42. Borthwick DW, Shahbazian M, Krantz QT, et al. Evidence forstem-cell niches in the tracheal epithelium. Am J Respir Cell MolBiol 2001; 24: 662–670.

43. Schoch KG, Lori A, Burns KA, et al. A subset of mouse trachealepithelial basal cells generates large colonies in vitro. Am JPhysiol Lung Cell Mol Physiol 2004; 286: L631–L642.

44. Johnson LG, Olsen JC, Naldini L, et al. Pseudotyped humanlentiviral vector-mediated gene transfer to airway epitheliain vivo. Gene Ther 2000; 7: 568–574.

45. Adkins B, Williamson T, Guevara P, et al. Murine neonatallymphocytes show rapid early cell cycle entry and cell division.J Immunol 2003; 170: 4548–4556.

46. Tran ND, Porada CD, Almeida-Porada G, et al. Induction ofstable prenatal tolerance to beta-galactosidase by in uterogene transfer into preimmune sheep fetuses. Blood 2001; 97:3417–3423.

47. Waddington SN, Buckley SM, Nivsarkar M, et al. In utero genetransfer of human factor IX to fetal mice can induce postnataltolerance of the exogenous clotting factor. Blood 2003; 101:1359–1366.

48. Zhang J, Xu L, Haskins ME, et al. Neonatal gene transfer with aretroviral vector results in tolerance to human factor IX in miceand dogs. Blood 2004; 103: 143–151.

49. Brown BD, Sitia G, Annoni A, et al. In vivo administration oflentiviral vectors triggers a type I interferon response thatrestricts hepatocyte gene transfer and promotes vector clearance.Blood 2007; 109: 2797–2805.

50. Annoni A, Battaglia M, Follenzi A, et al. The immune response tolentiviral-delivered transgene is modulated in vivo by transgene-expressing antigen-presenting cells but not by CD4 + CD25+regulatory T cells. Blood 2007; 110: 1788–1796.

51. Follenzi A, Battaglia M, Lombardo A, et al. Targeting lentiviralvector expression to hepatocytes limits transgene-specificimmune response and establishes long-term expression ofhuman antihemophilic factor IX in mice. Blood 2004; 103:3700–3709.

52. Pahal GS, Jauniaux E, Kinnon C, et al. Normal developmentof human fetal hematopoiesis between eight and 17 weeks’gestation. Am J Obstet Gynecol 2000; 183: 1029–1034.

53. Kamata Y, Tanabe A, Kanaji A, et al. Long-term normalizationin the central nervous system, ocular manifestations, andskeletal deformities by a single systemic adenovirus injectioninto neonatal mice with mucopolysaccharidosis VII. Gene Ther2003; 10: 406–414.

54. Holsapple MP, West LJ, Landreth KS. Species comparison ofanatomical and functional immune system development. BirthDefects Res B Dev Reprod Toxicol 2003; 68: 321–334.

55. Seppen J, van Til NP, van der Rijt R, et al. Immune response tolentiviral bilirubin UDP-glucuronosyltransferase gene transfer infetal and neonatal rats. Gene Ther 2006; 13: 672–677.

56. Brown BD, Venneri MA, Zingale A, et al. Endogenous microRNAregulation suppresses transgene expression in hematopoieticlineages and enables stable gene transfer. Nat Med 2006; 12:585–591.


Restricted transgene persistence after lentiviral vector-mediated fetal gene transfer in the...

Documents

Transcript of Restricted transgene persistence after lentiviral vector-mediated fetal gene transfer in the...