Induction of stable prenatal tolerance to β-galactosidase by in utero gene transfer into preimmune sheep fetuses

To date, no postnatal hematopoietic stem cell (HSC)-based gene therapy protocol has resulted in a cure or definitive clinical improvement, with the exception of a recent case of severe combined immunodeficiency disease1 in which the transduced HSCs would be predicted to have a significant growth advantage over the host's endogenous hematopoietic cells. While transduction efficiency of HSCs from humans and large animals in vitro has improved dramatically with the development of newer generation vectors and a better understanding of the in vitro growth requirements for HSCs, the level of gene-marked cells in vivo following transplantation of ex vivo–transduced HSCs has remained well below the levels required for clinical benefit.2-6 Interestingly, several studies have demonstrated that hematopoietic progenitors grown in vitro from the patients are marked at a high enough frequency that some clinical improvement could be possible, yet the level of marked progeny in the circulation has been so low as to be nearly undetectable except by polymerase chain reaction (PCR).5,6 As yet, no explanation has been offered for this discrepancy. While it is clear that inefficient transduction of long-term repopulating HSCs and transcriptional silencing of the vector promoters likely contribute to the low levels of gene-marked cells within the circulation and the transient nature of transgene expression, a large body of evidence is emerging to suggest that host immune responses against both the vector and the vector-encoded transgene products are a major limiting factor to achieving successful genetic correction by gene therapy.7-11 Studies have now shown that it is the host immune response and not transcriptional shutdown that precludes successful long-term transgene expression following adenoviral-mediated gene transfer.7-9 Other studies employing vectors based on Moloney murine leukemia virus (MuLV), the backbone on which the vectors in most current clinical trials are based, have now shown that both the vector itself and transgenes expressed from MuLV-based vectors are capable of eliciting robust immune responses. McCormack et al showed that intramuscular injection of MuLV-based retroviral vectors into both adult mice and nonhuman primates elicited an anti-MuLV antibody response that was able to neutralize vector-mediated transduction of target cells in vitro.11 The authors then went on to demonstrate that both immunoglobulin G and cytotoxic T-lymphocyte responses developed that were specific for the vector-encoded gene products. In other studies employing an ex vivo transduction/stem cell transplantation approach,12 Dube's group demonstrated that the infusion of autologous cells transduced with retroviral vectors encoding the normal canine α-L-iduronidase (α-ID) complementary DNA into adult canines with α-ID deficiency resulted in humoral responses against α-ID and a cellular immune response against autologous α-ID–transduced cells. These immune responses limited the duration of expression of the α-ID transgene by transduced hematopoietic cells, thus preventing clinical benefit from the procedure. Similarly, transplantation of retrovirally modified lymphocytes in adult humans resulted in specific cytotoxic T-lymphocyte responses specific for the transgene product, which eventually eliminated the transduced cells in vivo.13,14 Thus, it appears that the duration of vector-encoded transgene expression following administration of either vector preparations or ex vivo–modified autologous cells to an immunocompetent recipient is limited to a large degree by host immune–mediated elimination of transgene-expressing cells. In light of these studies, it is interesting that the only unequivocally successful gene therapy trial to date was achieved in patients whose disease rendered portions of their immune system nonfunctional.1 

Performing gene therapy in utero offers the possibility of correcting genetic disorders prior to the onset of disease. In addition, there are several features of the developing fetus that suggest that it may represent a more amenable target for gene therapy than either the neonate or the adult. One of these characteristics is the immunologic naı̈veté and resultant permissive environment of the early gestational fetus. These 2 characteristics should permit the acceptance of cells and vector without the need for immunosuppression or myeloablation. In early immunologic development, before thymic processing of mature lymphocytes, the fetus appears to be receptive to foreign antigens.15,16 Furthermore, exposure to foreign antigens during this period often results in sustained tolerance, which can become permanent if the presence of antigen is maintained.17,18 We have previously reported that direct intraperitoneal injection of both low-titer and high-titer retroviral vectors into sheep fetuses during early gestation when the fetus is still largely preimmune resulted in the successful transfer of the transgene to primitive HSCs and the long-term persistence of transgene-expressing hematopoietic cells in the circulation of the in utero–transduced sheep.19,20 While in our previous studies with low-titer vectors we followed the animals for nearly 5 years after gene transfer, the follow-up studies with high-titer vector only involved a 28-month analysis of the recipients. In the present studies, we have extended the analyses of the animals receiving the high-titer G1nBgSvNa8.1 vector (titer 1 × 10⁷colony-forming units [CFU]/mL) for 40 months postinjection and evaluated whether the longevity of transgene expression in this model system was the result of induction of immune tolerance to the vector-encoded transgene product following administration of the vector during the period of immune-naı̈veté. We now demonstrate that transgene-expressing hematopoietic cells persist within the peripheral blood (PB) and bone marrow (BM) of these animals for at least 40 months posttransduction. In addition, we show that the direct injection of retroviral vectors into sheep fetuses during the period of preimmunity results in the transduction of the epithelial cells that are present in the fetal thymus and are responsible for presenting antigen and defining self during immune system development,21,22 showing that the groundwork for induction of tolerance had been laid by this approach to gene transfer. Direct evidence for induction of tolerance comes from experiments in which control sheep and in utero–transduced sheep were boosted postnatally with purified β-galactosidase (β-gal) and we showed that the PB lymphocytes from the in utero–transduced sheep exhibited a significantly lower stimulation index to autologous cells that had been transduced in vitro with the identical vector that was administered in utero than did the control animals. We also demonstrate that the in utero–transduced sheep have a reduced ability to mount an antibody response to the highly immunogenic23 vector-encoded β-gal protein when compared with age-matched normal control sheep. Collectively, our results strongly support the induction of tolerance to the vector/transgene following the administration of the vector to preimmune recipients and provide one possible explanation for the duration and stability of transgene expression seen in this model. In addition, our findings demonstrate that even subtherapeutic gene transfer in utero may induce immune tolerance to the exogenous gene and thus potentially permit the successful postnatal treatment of the recipient, if required, with relative safety.

The protocol for the gene transfer has previously been described in detail.19 20 In short, 0.2 to 0.6 mL G1nBgSvNa8.1 retrovirus supernatant (titer 1 × 10⁷ CFU/mL) was injected intraperitoneally into preimmune (55- to 60-day) fetal sheep in utero under direct visualization using the microbubble technique. Sheep were allowed to continue gestation and, following birth, blood and marrow samples were obtained at intervals for analysis of gene transfer/expression, and several of the sheep were evaluated for the induction of immune tolerance to the transgene product.

PCR analysis and Southern blotting were performed as we have previously described.19 20 Briefly, samples were amplified for 35 cycles using primers specific to either theNeo^R (neomycin resistance) or lacZgene. Following amplification, 15 μL of each reaction was electrophoresed on a 2% agarose, Tris-borate-EDTA gel, transferred to nylon membrane, and hybridized to a ³²P–end-labeled probe specific to either Neo^R or lacZ. The membrane was washed 4 times under conditions of increasing stringency and exposed to Kodak BioMax film (Eastman Kodak, Rochester, NY) for 2 to 4 hours at −75°C with one Biomax intensifying screen.

Expression of the Neo^R and lacZ transgenes in PB and BM was evaluated using immunofluorescence microscopy and flow cytometric analysis and by performing hematopoietic progenitor assays in the presence and absence of a lethal concentration of G418, as we have previously described.19 20 

Thymi were obtained from 1 normal control sheep and 3 in utero–transduced sheep at 76 days of gestation (21 days postinjection) and homogenized in Iscoves modified Dulbecco medium (Gibco Life Technologies, Rockville, MD) using a glass homogenizer to obtain a single-cell suspension. Cells from each thymus were then cultured until near confluence (about 2 weeks) in 2 gelatin-coated T75 flasks (Costar) in α-minimal essential medium (Gibco) containing 10% heat-inactivated fetal bovine serum (FBS), 0.5 μg/mL hydrocortisone, 20 ng/mL epidermal growth factor, 5 × 10⁻⁵ M 2-mercaptoethanol, and 5 × 10⁻⁹ M cholera toxin to promote selective growth of epithelial cells.24 One flask of cells from the control sheep was used as a control for the specificity of x-gal staining. The other was used to assess the susceptibility of fetal thymic epithelium to transduction with the amphotropic MuLV vector administered in utero. To this end, one subconfluent flask of control sheep thymic epithelium was transduced for 72 hours with G1nBgSvNa8.1 supernatant, changing the supernatant each 12 hours. Then, 48 hours after completion of the transduction, G418 was added to the medium at a final concentration of 500 μg/mL and the cells were grown to near confluence (about 1 week). One flask of control thymic epithelium, in vitro–transduced thymic epithelium, and thymic epithelium (grown in the absence of G418) from each of the in utero–transduced sheep was stained with a pancytokeratin antibody (Sigma Chemical, St Louis, MO) to confirm that the cultures were composed purely of epithelial cells. When this had been confirmed, another flask of thymic epithelium from each of these sources was then evaluated for the expression of the lacZ transgene using a commercially available x-gal staining system (Specialty Media, Lavallette, NJ).

Three experimental sheep and 3 control sheep were immunized subcutaneously at 6 separate sites with 50 μg of each of the following proteins: β-gal (Boehringer Mannheim, Indianapolis, IN), chicken egg albumin, and sheep albumin (both from Sigma). All primary immunizations were administered in complete Freund adjuvant, while reimmunizations (1 month after primary immunizations) employed incomplete Freund adjuvant (Sigma).

Serum samples were collected at intervals for the next 2 months. Sera were tested for anti–β-gal antibody activities using a standard enzyme-linked immunosorbent assay.25 26 Briefly, individual plates were coated (5 μg/well) with β-gal, chicken egg albumin, or sheep albumin, and nonspecific sites were blocked with 0.05% Tween 20, 2% bovine serum albumin in phosphate-buffered saline (blocking buffer). Sera were added to the coated plates in serial dilutions, incubated for 1 hour at 37°C, and rinsed 3 times with blocking buffer following incubation. Donkey antisheep immunoglobulin G antibody conjugated with peroxidase (Boehringer Mannheim) was added to each well, and the plates were incubated for an additional hour at 37°C. Following incubation and washing, peroxidase substrate, ABTS (2,2′-azino-di-[3-ethylbenzthiazoline sulfonate]) solution with enhancer (Boehringer Mannheim), was added to each well and incubated at room temperature until color (green) development was sufficient for photometric detection (5-10 minutes). Samples were measured with the Bio-Rad 3550-UV microplate reader (Bio-Rad, Hercules, CA) at 405 nm. A reference reading was also taken at 490 nm and subtracted from the readings obtained at 405 nm.

Cellular immune response was assessed as previously described12 with minor modifications. In short, at 45 months posttransduction, the 3 in utero–transduced sheep and 3 normal age-matched controls that had received subcutaneous immunizations with purified β-gal were evaluated for the ability of their PB mononuclear cells (PBMCs) to proliferate in response to either autologous BM stromal cells or autologous BM stromal cells that were transduced in vitro with the identical vector that had been administered in utero. BM stromal cells were grown by plating out whole BM cells in standard long-term marrow culture medium consisting of Iscoves modified Dulbecco medium (Gibco) with 12.5% FBS (Hyclone Laboratories, Logan UT), 12.5% horse serum (Stem Cell Technologies), and 10⁻⁶ M hydrocortisone (Sigma). At 48 hours after culture initiation, the nonadherent fraction was removed and fresh media added. The stromal cultures were maintained by twice-weekly half-volume media changes. At confluence, stromal cells were trypsinized and replated at a 1:3 dilution. Twenty-four hours later, the media were replaced with G1nBgSvNa8.1 vector supernatant supplemented with 4 μg/mL protamine sulfate, and the stromal cells were transduced for 3 days, changing the supernatant each 12 hours. Control untransduced stromal cells were manipulated similarly to the transduced stromal cells with the exception that fresh media were used in place of retroviral supernatant. At 48 hours after the completion of the transduction cycle, 500 μg/mL G418 was added to the stromal cell cultures that had been transduced. Stromal cells were maintained by 1:3 dilutions when confluent. At passage 4 to 5, the untransduced and transduced G418-selected stromal cells were plated in 100 μL RPMI with 15% FBS in 96-well plates and cultured to near confluence. Stromal cells were then treated with mitomycin C (Sigma), washed 3 times, and the media changed to RPMI with 5% autologous sheep serum. Autologous sheep PBMCs were used as responder cells and resuspended in RPMI with 5% autologous sheep serum. A total of 100 μL of responder cells was added at a ratio of 1:5 stimulator-to-responder cells in triplicate to each transduced and control stroma sample. Pokeweed mitogen (Gibco) was added to 3 of the wells with cells from each sheep to serve as a positive proliferation control, while media alone, PBMCs alone (from each individual sheep), and stromal cells alone (from each individual sheep) served as negative controls. Cells were incubated at 37°C for 72 hours, and bromodeoxyuridine (BrdU) labeling reagent was then added and the cells were incubated for an additional 24 hours. BrdU incorporation was then analyzed according to manufacturer's instructions (Boehringer Mannheim) using the Bio-Rad microplate reader at 450 nm with a reference wavelength of 690 nm.

As previously reported,20 16 preimmune sheep fetuses were injected intraperitoneally with 0.2 to 0.6 mL replication-defective MuLV-based retroviral supernatant. Fourteen of these recipients were available for analysis (2 were lost shortly after birth due to accidents at the farm). Proviral DNA and transgene expression were consistently detected in 12 of these animals throughout the 28-month course of our initial studies. At 40 months posttransduction, 9 of the 12 positive recipients from this study were still available for analysis (2 of the animals had been killed for tissue distribution studies, and 1 died as a result of tetanus). At this time, all 9 remaining in utero–transduced animals contained hematopoietic cells expressing the lacZ transgene product, β-gal, in PB (0.4%-5.1%) and BM (1.2%-5.3%) when evaluated by flow cytometry using the fluorescent β-gal substrate fluorescein di-β-D-galactopyranoside (Table 1). In addition, at 41 months posttransduction, all 8 animals that remained (another animal was killed between 40 and 41 months for analysis of vector tissue distribution) possessed neomycin phosphotransferase (NPT) activity as demonstrated by the continued ability of hematopoietic progenitors present in their BM to form significant numbers of hematopoietic colonies in the presence of a lethal concentration (2 mg/L) of G418 in vitro (10.8%-37.0% G418 resistance). Neo^R-specific PCR analysis and subsequent Southern blotting demonstrated that all G418-resistant colonies contained the bacterialNeo^R gene, confirming that vector-encoded transgene expression was responsible for growth in a lethal concentration of G418 (data not shown).

Expression of the transgene products was also demonstrated at 40 months posttransduction by fluorescence-activated cell sorter (FACS) analysis using antibodies specific for β-gal and NPT. The β-gal and NPT were detected in both PB (0.8%-6.4% and 1.2%-3.7%, respectively) and BM (2.1%-6.5% and 1.4%-5.8%, respectively) of all in utero–transduced animals, as shown in Table 1. Similar results were also obtained when blood and BM mononuclear cells were stained with these antibodies and analyzed visually on a fluorescence microscope (data not shown). In all studies, no significant staining was observed in normal control sheep.

To confirm that proviral DNA had integrated into the host genome, lacZ- and Neo^R-specific PCR analyses were performed on all in utero–transduced animals. Direct injection of retroviral vectors into the peritoneal cavity of fetal sheep resulted in the integration of proviral DNA (lacZ andNeo^R genes) into PB and BM mononuclear cells of all in utero–transduced animals that persisted for at least 40 months posttransduction, suggesting that proviral DNA integration into primitive HSCs occurred following direct injection of retroviral vectors into preimmune fetal sheep (data not shown).

Numerous reports have now established that host immune responses against the vector and/or vector-encoded gene products may limit the duration of transgene expression and result in the elimination of transduced cells in vivo.7-12 Because we administered the vectors in utero during the “preimmune” period of fetal development, we hypothesized that the longevity of transgene expression observed in our studies was the result of our having induced immune tolerance to the transgene products with this approach to gene transfer. We began addressing this issue by examining whether the epithelial cells of the fetal thymus were transduced by the vectors employed in our studies, because it is generally thought that the epithelial cells of the developing thymus are the cells responsible for presenting antigen and thus determining self during immunologic maturation. To evaluate this, we first grew populations of pure thymic epithelial cells from a control 76-day-old sheep fetus using previously described methodology.24 We then attempted to transduce these cells in vitro with the identical vector employed in our in utero gene transfer studies. As is shown in Figure1A, which is a representative x-gal staining performed on transduced thymic epithelium following 1 week of G418 selection, these pure populations of fetal thymic epithelial cells were readily transduced with the G1nBgSvNa8.1 vector supernatant and expressed high levels of the vector-encoded β-gal. We next examined whether our direct injection approach to in utero gene transfer successfully transduced these cells in vivo within the recipient sheep. To this end, we collected the thymi from 3 in utero–transduced sheep at 21 days postinjection (76 days of gestation) and again grew pure populations of epithelial cells from each thymus in the absence of G418 selection. The resultant monolayers were then subjected to x-gal staining. As can be seen in Figure 1B, which is a representative flask from one of the in utero–transduced sheep, numerous β-gal–expressing blue cells are present. No blue cells were seen in thymic epithelial cultures grown from normal control fetal sheep (data not shown). The epithelial origin of all thymic cell cultures was confirmed by staining with a pancytokeratin antibody prior to analysis for transgene expression (data not shown). These results demonstrate the successful transduction of thymic epithelium in vivo within the gene transfer recipients. In previous studies,19 20 we have demonstrated the continued presence of proviral DNA within the thymi of in utero–transduced sheep for at least 29 months postinjection by PCR. In addition, throughout the entire 4-year course of these studies we have consistently detected β-gal⁺cells when thymic sections from these sheep were analyzed by immunofluorescence with an anti–β-gal antibody (data not shown). These combined results led us to speculate that prolonged thymic expression of the transgene following our in utero approach to gene therapy may allow for induction of immunologic tolerance.

To directly assess whether tolerance was induced in the in utero–transduced animals, we examined the ability of 3 in utero–transduced sheep (nos. 698, 703, and 704) and 3 normal age-matched control sheep (nos. 7119, 7122, and 7124) to mount humoral immune responses after postnatal immunization with purified β-gal (transgene product), sheep albumin (negative control), and chicken egg albumin (positive control) in complete Freund adjuvant at 42 months posttransduction. At day 33 postimmunization, these animals were reimmunized with β-gal with incomplete Freund adjuvant in an effort to further boost the immune response against β-gal. Sera were collected from all animals before and on 11 different occasions after immunization (last samples were obtained on day 68) and evaluated for the presence of specific immunoglobulin G antibodies against β-gal by enzyme-linked immunosorbent assay as described in “Materials and methods.” As shown in Figure 2A, normal control sheep showed a significant specific antibody response against β-gal. In contrast, the ability to mount an antibody response to β-gal was clearly impaired in the in utero–transduced sheep, suggesting that these animals had been tolerized to the β-gal protein by administering the vector during the fetal period of preimmunity. Moreover, the second immunization did not elicit any additional response in the in utero–transduced sheep, while higher levels of anti–β-gal antibody (150 μg/mL) were detected in the control sheep. These results further confirm that the in utero–transduced sheep do not recognize β-gal as a foreign antigen. Antibodies against sheep albumin were never detected in any of the immunized animals, whereas antibodies against chicken egg albumin were consistently detected in all the animals from both the control group and the experimental group (Figure 2B), indicating that the in utero–transduced sheep were capable of eliciting an immune response against a foreign antigen and did not simply possess a general immune unresponsiveness. Potential immune responses against NPT were not evaluated because the only purified NPT that was commercially available (5′ Prime-3′ Prime, Boulder, CO) was suspended in a solution of G418.

Potential cellular immune responses against the transgene product/transduced cells in the in utero–transduced sheep were also evaluated by lymphocyte proliferation assays. Three experimental sheep (nos. 698, 703, and 704) and 3 normal age-matched control sheep (nos. 7119, 7122, and 7124) were evaluated at 45 months posttransduction (3 months after the first β-gal immunization). In these experiments PB lymphocytes were overlaid on autologous stromal cells that were either transduced with the G1nBgSvNa8.1 retroviral vector or mock infected with normal medium; lymphocyte proliferation under each condition was measured by incorporation of BrdU. Transduced stromal cells from both the control and experimental animals were selected in G418 (2 mg/mL); more than 90% of the transduced stromal cells expressed β-gal after selection as measured by x-gal staining. Lymphocyte proliferation was 3.1- to 5.0-fold higher when blood lymphocytes, collected from the 3 age-matched control animals, were stimulated with G1nBgSvNa8.1-transduced autologous stromal cells than untransduced stromal cells (Figure 3). In contrast, there was no significant difference between stimulation of blood lymphocytes, collected from 3 experimental sheep, to G1nBgSvNa8.1-transduced autologous stromal cells and untransduced stromal cells (stimulation index 0.7-1.1), as shown in Figure 3. Similar results were also obtained when lymphocytes were stimulated with either purified β-gal or G1nBgSvNa8.1-transduced autologous PBMCs (data not shown). All control and experimental sheep exhibited a normal proliferative response to pokeweed mitogen, with stimulation indices of 4 to 5 (data not shown), confirming that the apparent tolerance was not simply the result of general immune unresponsiveness. The results from these experiments clearly demonstrate the inability of the in utero–transduced sheep to mount a cellular immune response against the viral-encoded proteins (β-gal and NPT) in the in utero–transduced sheep, providing further confirmation that the direct injection of retroviral vectors in utero had tolerized the recipients to the vector/vector-encoded transgene products.

To confirm that the immunization with β-gal did not elicit an immune response that triggered elimination of the transduced hematopoietic cells from the circulation of the in utero–transduced animals, PBMCs were collected on the day of immunization and at day 68 postimmunization and analyzed by lacZ-specific PCR. As demonstrated in Figure 4, the lacZ gene was still detected in all of the immunized sheep at 68 days postimmunization. Additionally, β-gal–specific immunofluorescence analysis performed on blood smears collected from immunized sheep nos. 698, 703, and 704 demonstrated that β-gal was still expressed in these 3 animals (4.1%, 3.5%, and 3.0%, respectively, data not shown). These results further support the conclusion that the in utero–transduced sheep were tolerized to the β-gal protein by the direct injection of retroviral vectors in utero.

We have previously reported the safe and efficient transduction of primitive HSCs in sheep following direct injection of retroviral vectors into the peritoneal cavity of preimmune fetuses.19,20 In the present studies, we extended the time course of our analysis and show that expression of the transgene has persisted within the hematopoietic system of these in utero–transduced animals for over 40 months posttransduction without eliciting an immune response to the vector-encoded genes. The long-term expression of the immunogenic β-gal and NPT transgene products is in contrast to previous in utero and postnatal gene transfer studies in older, immunocompetent sheep fetuses,27-31 nonhuman primates,32-34 dogs,12,35rabbits,36 and mice11,37 in which immune responses against the vector-encoded proteins resulted in transient transfer or expression of the transgenes and elimination of the transduced cells in vivo. Unfortunately, the development of transgene expression-limiting immune responses following gene transfer is not merely an issue of academic interest, because it not only applies to marker gene products9,27-30,38 but also to therapeutic proteins that have never been expressed in the recipient as a result of their gene defect.12,39 40 Thus, for gene therapy to become a viable therapeutic option to individuals suffering from genetic diseases characterized by a complete absence of the affected protein, methods need to be found for evading the induction of an immune response that could result not only in failure of the gene transfer protocol but also render the patient refractory to protein replacement therapy that could have controlled the disease.

During early fetal development, there is a window of opportunity, prior to thymic processing of mature lymphocytes, during which the fetus is receptive to foreign antigens.15,16 Additionally, exposure of the fetus to foreign antigens during this window of opportunity results in sustained tolerance, which could be permanent if antigens are constantly maintained.17,18 Cellular tolerance appears to be dependent upon avidity thresholds and clonal deletion of reactive lymphocytes in the thymus, whereas the mechanism of B-lymphocyte tolerance (peripheral tolerance) appears to involve both clonal deletion and clonal suppression.21,22 41-45 The end result is an immune system that is tolerant to these specific foreign antigens.

In the present studies, we administered the MuLV-based vector supernatants into the peritoneal cavity of fetal sheep at days 55 to 60 of gestation. The corresponding period in humans is 13 to 14 weeks of gestation, during which the human fetus is considered to be immunologically naive. In sheep, the immune system reaches at least partial maturity between days 67 and 77 of gestation as demonstrated by the prolonged survival of allogeneic skin grafts placed before day 67 of gestation and vigorous rejection of grafts placed after day 77 of gestation.46 47 We reasoned that by administering the vector prior to immunocompetence, we could avoid the development of an immune response to the vector/transgene products and thus achieve long-term transgene expression. In addition, we felt that the delivery of the vector at this early stage in development coupled with the stable long-term expression of the transgenes afforded by the MuLV-based vectors we employed might have induced tolerance to the vector/transgene products. Our results demonstrate a complete absence of an immune response to the vector-encoded β-gal in the in utero–transduced sheep and show that these sheep do not mount either humoral or cellular immune responses to the transgene product, β-gal, following postnatal immunization with purified β-gal, providing strong evidence that tolerance to this exogenous protein was induced in utero.

Recent reports have demonstrated that tolerance to vectors and transgene products could be induced if lymphocytes underwent thymic selection in the presence of vector antigens48-52 and that this tolerance led to a prolongation of transgene expression. For example, DeMatteo and colleagues48,49 showed that intrathymic inoculation of neonatal mice with a recombinant adenoviral vector encoding the lacZ gene during the period prior to T-cell maturation resulted in the induction of host tolerance to the adenoviral vector and the transgene product. Furthermore, when these tolerized mice were injected intravenously as adults with the same adenoviral vector, they exhibited an impaired adenovirus-specific cytotoxic T-lymphocyte response, which allowed prolonged β-gal expression to be achieved for up to 260 days. Ilan and colleagues have also reported similar results in Gunn rats.50 In our current studies, we demonstrate that the direct injection of MuLV-based retroviral supernatants into the peritoneum of fetal sheep during the period of preimmunity results in successful transduction and transgene expression in the epithelial cells present within the thymus of the in utero–transduced sheep. Combined with our previous findings that the transgene persists within the thymus of in utero–transduced sheep for at least 4 years postinjection, we have speculated that thymic expression of the transgene may be one means by which this direct injection approach to gene transfer induced central tolerance to vector/transgene products in these sheep. The successful induction of lasting tolerance to the vector-encoded transgene products is in contrast to a recent study of similar design conducted in preimmune mice.53 As in our studies, the authors of this recent report demonstrate that administration of vector supernatant to preimmune mice allows transfer and expression without eliciting an immune response. However, upon postnatal readministration of the vector, a robust immune response to both the vector and vector-encoded luciferase occurred. The contrasting outcome of this study and ours can likely be explained by the choice of vector, because the adenoviral vector employed in the murine studies produced only transient expression, while the MuLV-based vector we employed produced stable long-term (more than 40 months) expression. Collectively, the results of this murine study and the results of our present study fit well with the existing dogma concerning the induction of stable prenatal tolerance, which requires early prenatal antigen exposure to induce tolerance and then persistence of the antigen to maintain tolerance throughout life.

In conclusion, these studies demonstrate and confirm that the direct injection of an engineered retroviral vector is a safe and effective means of delivering exogenous genes to the primitive HSCs of a developing fetus. We now also demonstrate that the long-term transgene expression obtained in this model is likely due to the induction of central tolerance to the transgene product, β-gal, following vector administration during the period of fetal preimmunity. When methods for achieving therapeutic levels of transduction have been developed, in utero gene transfer will likely become a promising approach for the treatment of genetic diseases that produce fetal damage before birth. Our findings provide a further argument for the clinical utility of in utero gene therapy, because it is now apparent that even subtherapeutic retrovirally mediated gene transfer initiated in utero during the preimmune window of opportunity may be capable of rendering the recipient tolerant to the exogenous therapeutic gene products and thus potentially permit the successful postnatal treatment of the patient, if required, with relative safety.