Pulmonary inflammation associated with repeated, prenatal ... - Nature

Gene Therapy (1999) 6, 98–106  1999 Stockton Press All rights reserved 0969-7128/99 $12.00 http://www.stockton-press.co.uk/gt

Pulmonary inflammation associated with repeated, prenatal exposure to an E1, E3-deleted adenoviral vector in sheep HS Iwamoto1, BC Trapnell2, CJ McConnell1, C Daugherty1 and JA Whitsett1 1

Children’s Hospital Medical Center and the Department of Pediatrics, University of Cincinnati, Cincinnati, OH; and 2Genetic Therapy Inc, Gaithersburg, MD, USA

Fetal gene therapy may prove useful in treating diseases that manifest in the perinatal or early postnatal period. Adenoviruses effectively transfer gene expression to a variety of tissues but also stimulate inflammatory and immune responses. The purpose of this study was to test the hypothesis that exposure of fetal sheep to a first generation adenovirus vector encoding bacterial ␤-galactosidase, Av1nBg, before the development of the immune system, is safe, minimizes inflammatory and immune responses and induces tolerance. A total of 22 fetal sheep was studied; of these, two were born with respiratory distress, seven were

electively killed and 13 died in utero. The incidence of mortality was higher than the ⭐10% we have experienced with other fetal sheep studies and was not likely related to complications arising from surgical or anesthetic procedures. Inflammatory and fibrotic responses were observed in the lungs and may represent untoward long-term consequences of in utero adenoviral gene therapy. Tolerance to Av1nBg was not established, and repeated exposure to Av1nBg before birth was associated with significant pathology and mortality.

Keywords: gene transfer; gene therapy; fetus; fetal treatment; fetal therapy

Introduction First generation replication-deficient recombinant adenovirus vectors have been used to transfer gene expression to a wide variety of cell types in a number of species at distinct developmental stages.1–6 Gene expression is transient and appears to be limited by host inflammatory and immune responses to adenoviral proteins.2,4,7–10 Adenovirus-mediated gene transfer and expression are more effective when the ability of the recipient’s immune system to respond to vector administration is suppressed. Thus, when first generation adenovirus vectors are administered with dexamethasone, cyclosporin or interleukin inhibitors or to immunodeficient mice, gene expression outlasts that produced by adenovirus vectors administered alone to immunocompetent animals.8,9 An alternative approach is to initiate gene therapy before immune recognition and inflammatory responses have matured. In the newborn mouse, in which immune recognition mechanisms have not fully developed,11 adenovirus-mediated gene expression is prolonged compared to that in the adult mouse.12,13 Initiating gene therapy in humans during the neonatal period may not be as effective because the immune system becomes responsive by midgestation.14 Indeed, introduction of foreign substances, such as hematopoietic stem cells, results in engraftment in human fetuses only if transplantation is Correspondence: HS Iwamoto, Children’s Hospital Medical Center, Division of Neonatology, TCHRF, 3333 Burnet Avenue, Cincinnati, OH 45229-3039, USA Received 19 May 1997; accepted 27 August 1998

initiated before 16–20 weeks of gestation or if the recipient is immunocompromised.15,16 It is not known whether adenovirus-mediated gene therapy initiated before immune development in species other than the mouse can result in prolonged gene expression. The purpose of this study was to test the hypothesis that exposure of fetal sheep to a first generation adenovirus (Av1) vector (deleted in the E1 and E3 regions) before the development of the immune system, is safe, minimizes inflammatory and immune responses and prolongs gene expression due to the induction of tolerance. Sheep were used in this study because the immune system in the sheep, like that in the human, becomes responsive by midgestation.17 We and others have previously shown that intratracheal administration of an Av1 vector results in gene transfer and expression in the fetal lung, but also stimulates an inflammatory response, characterized by a dramatic increase in the number of lymphocytes and macrophages in lung fluid, in CD4+ and CD8+ cells in lung tissue and the development of fibrotic lesions.4,6 To determine whether the inflammatory and immune responses to Av1 could be avoided, we instilled Av1nBg, a vector that transfers expression of bacterial ␤galactosidase, into the amniotic cavity of fetal sheep at 2–3-week intervals beginning at 0.35–0.45 term gestation. At 0.72–0.79 term gestation, a challenging dose of Av1nBg was instilled into the fetal trachea. Lung fluid and tissue were examined for inflammatory cells and fibrotic lesions, and serum was assayed for the presence of neutralizing antibody.

Repeated adenovirus administration in fetal sheep HS Iwamoto et al

Results Overview Twenty-two fetal sheep were studied in four groups; 13 of these died in utero. The purpose of the first study was to determine whether repeated injection of Av1nBg into the amniotic cavity of anesthetized ewes under ultrasound guidance could transfer gene expression to the fetus for a prolonged period of time. Fifty percent of the fetuses died in utero (Figure 1a), possibly related to the virus administration or repeated anesthesia and injection procedures. The second study was designed to avoid the repeated anesthesia and injection procedures. Av1nBg was administered into the amniotic cavity via a chronically implanted subcutaneous port located in the ewe’s flank. Seventy-five percent of these fetuses died in utero (Figure 1b), indicating that virus administration adversely affected the fetuses. The third group of fetuses were exposed to a single dose of Av1nBg, electively killed 1–62 days after virus administration (Figure 1c) and examined to determine whether the mechanism of fetal demise could be detected histologically. In the fourth group, replication-deficient adenovirus that lacked an expression cassette, Av1Null, was administered to determine whether the high degree of mortality was related to the Av1nBg vector used in the first three studies. All of the fetuses in this group died in utero indicating that the gene product, bacterial ␤-galactosidase, was not primarily responsible for the high degree of mortality in the other groups. Two fetuses in group 1 delivered spontaneously at term gestation (Figure 1a). Fetus 1 was born at 145 days of age with respiratory distress. It did not feed well, developed pneumonia and died 1 day after birth. Fetus 3 delivered vaginally at 148 days of age and survived until elective death was performed 72 days later. This lamb had mild pneumonia and was treated with antibiotics (gentamicin and chloramphenicol) for 3 days. Seven fetuses (group 3, Figure 1c) were electively killed 1–20 days after Av1nBg administration. Ten fetuses died in utero after one to three doses of Av1nBg (Figure 1). Death occurred 11–56 days after the initial dose and 4–8 days after the final dose, when administered. This incidence of intrauterine death is much greater than the 3– 5% incidence we encountered in concurrent fetal preparations not exposed to adenovirus vectors. Most of the animals that had a Hickman port implanted subcutaneously in the ewe’s flank connected to a catheter in the amniotic fluid cavity (group 2) died in utero (5A and B, 6, 7A, and 8A and B). Two of these fetuses (5A and B) were infected and filamentous bacteria were found in the lung. Four of these fetuses (6, 7A, 8A and 8B) aborted, and it was not possible to salvage fetal tissue for analysis. Fetus 2 (group 1) also died in utero and aborted before tissue could be obtained for analysis. Tissues from fetus 7B were obtained while the ewe was in labor. This fetus was apparently dead before labor was initiated on day 150 of gestation. In this fetus, there was evidence of bacterial infection on autopsy, most likely related to placental complications secondary to intrauterine fetal death. The remaining two fetuses that died in utero, 4 and 12A, had indwelling catheters when they died. Fetal arterial pH and blood gases were within the normal range in both fetuses up to the day they died (average values: arterial pH, 7.32; PCO2 58 torr; PO2 22 torr; hemo-

globin concentration, 9.5 g/dl; blood saturation, 62.2%). These fetuses remained in utero because their twin remained alive, so it was not possible to determine the exact cause of death. Intrauterine infection was not likely to have been the cause because it usually stimulates premature delivery. The cause of death may have been related to intratracheal administration of Av1nBg, catheter placement that occurred a few days before fetal death or possibly related to multiple doses of saline. All three fetuses that received Av1Null died in utero or were aborted sometime during the 6 weeks following virus administration. It does not appear that the cause of fetal deaths was due to the proliferation of wild-type virus. Samples of fetal and maternal blood, fetal skin, lung, liver muscle and amniotic fluid were assayed for the presence of wild-type adenovirus. In no sample was actively proliferating virus measured. Av1nBg transferred gene expression to fetal membranes. Biopsies of amniotic membranes obtained intraoperatively or after death (from fetuses 1, 3, 4, 9A and B, 10A and B, 11A and B) contained blue nuclei indicative of bacterial ␤-galactosidase activity directed by Av1nBg. Skin biopsies obtained intraoperatively did not stain blue but represented a minute portion of the total skin surface area. Blue cells were present in all lobes of the lung 7 days after adenovirus administration (fetus 12B) but not after 42–55 days (fetus 7B and lamb 3). Lung fluid samples were obtained from five fetuses (1, 3, 4, 7B and 12B) before and after intratracheal instillation of Av1nBg at 104–114 days. The total number of cells in lung fluid increased dramatically over baseline within 3 days and decreased to baseline by 5–6 days after intratracheal instillation (Figure 2). Lymphocytes were the predominant cell type present in lung fluid after adenovirus administration; their abundance peaked at 3–5 days. Neutrophils began to appear in lung fluid at 6–8 days; their abundance peaked at 17–23 days. Several samples of lung tissue were examined histologically. Lung tissue was obtained from lamb 11A 20 days after a single intra-amniotic injection of 1 × 1010 p.f.u. of Av1nBg (Figure 3a). The histologic appearance reveals the normal structure of an immature lung with budding airways and mesenchyme. No evidence of inflammation was found in sections obtained from this or other fetuses (9A and B, 10A and B and 11B) treated similarly. The appearance of lung tissue obtained from lamb 12B shows evidence of an inflammatory response characteristic of responses we observed in a previous study (Figure 3b).4 Despite exposure to the virus initially at 59 days gestation, subsequent instillation of a second dose into the trachea at 114 days gestation (9 × 1010 p.f.u. of Av1nBg), was associated with the presence of necrotic cells, monocytes and neutrophils in the luminal exudate. This histological appearance of this lung correlated well with the changes in lung fluid (inverted triangles, Figure 2). Lamb 3 showed dramatic changes in histological appearance of the lung (Figure 3c). The biopsy specimen obtained at 44 days of age showed areas of peribronchial fibrosis and inflammation. Numerous lymphocytes and histiocytes were located in areas of infiltration. Fetal sheep sera lacked detectable neutralizing adenovirus antibody with two exceptions. Fetus 7B had no detectable titer until 24 days after intratracheal administration of Av1nBg when it was 130 days gestational age.

99


100

Figure 1 Experimental design and outcome: three groups of animals were studied. The length of gestation is expressed in days and as a percentage of term gestation. Fetuses (indicated by number on the left) received intraamniotic (IA) and intratracheal injections (IT) of Av1nBg or saline. Doses of Av1nBg ranged from 1 × 1010 to 1 × 1011 p.f.u. (abbreviated 1e10–1e11). Polyvinyl vascular and tracheal catheters (catheters) and Hickman catheters (Hickman port) were implanted at the times indicated.

Thereafter, the titer increased from ⭐1:10 to 1:10 at 24– 31 days and was 1:40 on the 35th day after Av1nBg administration when the fetus was 141 days’ gestation. Lamb 3 had no detectable titer at all times before birth, despite it receiving four doses of adenovirus in utero and the presence of a significant titer (1:2560, 1:5120) in the ewe. When this lamb was retested 2 days after a fifth dose of adenovirus was administered at 17 days of postnatal age, a positive titer (1:640) was detected. Lamb 3 underwent bronchoscopy at 10, 20 and 35 days of postnatal age at which time BAL and brush biopsy samples were obtained. At these times, none of the cells expressed bacterial ␤-galactosidase. Chest radiographs, obtained at 37, 44 and 58 days, demonstrated increased pulmonary markings consistent with a pattern of interstitial fibrosis. The lung biopsy obtained at 44 days of age showed peribronchial lymphocytic infiltration (Figure 3c). There was evidence of lymphogranulomas in small airways and areas of fresh infarct and scarring. There was no evidence of pleural scarring in the biopsy specimen. At 58 days of age, arterial blood saturations, assessed by pulse oximeter were 88–90% at rest and 70–75% with exertion, indicating severe respiratory compromise.

Discussion The purpose of this study was to determine whether first generation adenovirus vectors administered before the immune system matures completely was safe and could modify the inflammatory and immune responses to sub-

sequent administration of the adenovirus vector, Av1nBg, and induce tolerance. Fetal sheep were exposed to Av1nBg several times beginning as early as 0.35 term gestation. Subsequent challenge with intratracheal Av1nBg at 0.7 term gestation elicited an inflammatory response that was similar to that observed in fetuses exposed to a single dose of intratracheal Av1nBg at the same time in gestation.4 Fibrotic lesions developed following a single or multiple dose of Av1nBg. Furthermore, most of the fetuses failed to survive repeated exposure to the adenovirus vector. These results demonstrated that repeated exposure of fetal sheep to adenovirus vectors beginning at midgestation produces adverse inflammatory and immune responses. There was no evidence that suggested that repeated exposure resulted in tolerance induction and prolonged gene expression. Adenovirus vectors have been shown to transfer prolonged gene expression in young rodents in a number of studies. Administration of adenovirus vectors into the trachea, skeletal muscle, thoracic cavity or vasculature of neonatal mice or rats results in transgene expression for several weeks or months.12,13,18–22 Subsequent administration of the adenovirus vectors were tolerated and resulted in robust gene expression in some studies19,20,23 but not others.18 The relative success of this approach in rodents has been attributed to the relative immaturity of the immune system in the rodent at birth.11 It is tempting to speculate that gene transfer in neonates of other species including humans might be just as successful. However, immune function in larger species develops before


101

Figure 1 Continued.

birth. For example, in humans and sheep, the thymus becomes colonized with lymphocytes by 0.25 term,17,24 T cell receptors and class I and II MHC antigens are detectable on lymphocytes by 0.4–0.5 term.25,26 Fetal sheep tolerate exposure to allogeneic kidney and skin grafts when performed before 55–60 days of gestation (0.4 term gestation) and human fetal liver stem cells up to 66 days of gestation (0.45 term gestation).27–29 Specific antibody responses to soluble antigens develop as early as 40 days gestation in the sheep with most responses mature by 120 days.30,31 We expected that the immune system of the fetal sheep at 0.35 term gestation was sufficiently immature so that tolerance to adenovirus vectors, as has been shown for liver stem cells,28 could be induced. However, we observed no evidence that the recombinant adenovirus administered was tolerated. Initiating adenovirus therapy before 0.35 term gestation did not prevent the subsequent development of inflammatory and humoral responses that were similar to those following single exposure to first generation replication-deficit adenovirus vectors.4,6 Because immune development in humans follows a similar time course to that in the sheep, we hypothesize that the human fetal response to adenovirus vectors is likely to be more similar to that of the sheep than the mouse.

One possible reason why tolerance was not induced in these studies may be related to the route of administration. Intra-amniotic injection was selected as the route of administration because this approach is feasible and associated with low risk in humans. Additionally, in previous studies we have shown that intra-amniotic administration of the same vector to fetal sheep, mice and rats at 0.8 term gestation transferred gene expression to the fetal trachea and lungs and gastrointestinal tract, as well as to the amniotic membranes.5,6,32 We expected access to the trachea and lungs to be greater at 0.35 term gestation because amniotic fluid is less viscous and the fetal lung secretes less fluid than later in gestation.33 Because transgene expression in the lung has been shown to persist for at least 13 days following intra-amniotic administration,32 we expected that administration of adenovirus every 15 days or so would be a reasonable strategy for tolerance induction. However, this approach did not result in tolerance induction, perhaps because an insufficient amount of adenovirus reached the appropriate sites. Other routes of administration might have been more effective. Studies in neonatal mice and rats have demonstrated that priming with an initial intravenous dose of adenovirus vector resulted in tolerance to subsequent administration of the same vector.19,23 The intravenous route, which delivers


102

Figure 1 Continued.

Figure 2 Effect of intratracheal instillation of Av1nBg on lung fluid cellularity. The abundance of cells in lung fluid, expressed as the fold-increase over baseline values measured just before Av1nBg administration, increased following intratracheal administration of Av1nBg on day 0. The solid symbols denote the responses of fetuses 4 and 7B which received only an intratracheal injection. The open symbols denote responses of animals 1, 3 and 12B which received one to three intra-amniotic and one intratracheal injection.

the adenovirus directly to fetal tissues, may be a more effective route of administration. Alternatively, injection into the skeletal muscle or thoracic cavity have been successful in young rodents12,13,20,22 and may be successful in sheep. Another possible reason why we were unable to induce tolerance in the present study may be related to the dose of adenoviral vector used. We injected 1 × 1010– 1 × 1011 p.f.u. in the amniotic cavity of fetal sheep beginning at 0.35 term gestation. Because amniotic fluid volume is about 500 ml at this stage in gestation, the final concentration was around 2 × 107–2 × 108 p.f.u./ml. Administration of a higher concentration of adenovirus has been shown to transfer gene expression and to induce tolerance. Injection of 1 × 108–1 × 109 p.f.u. of adenovirus into skeletal muscle of newborn mice results in prolonged gene expression.12,21,22 Injection of similar amounts of adenovirus into the thoracic cavity or abdomen or vasculature of newborn rats or mice is sufficient to induce tolerance, prolong gene expression and reduce inflammatory and immune responses to subsequent exposure.19,20,23 Interestingly, injection of a higher dose (1 × 1010 p.f.u.) into the trachea of newborn cotton rats resulted in prolonged gene expression that was extinguished when a second vector was injected 8–10 weeks later.18 Repeated exposure to Av1nBg beginning early in gestation not only did not induce tolerance, it was associated with a high incidence of fetal death that was greater than


when Av1nBg was administered later in gestation. In the present study, 53% of the fetuses did not survive; in our previous study, 30% of the fetuses died.4 This frequency of fetal death is greater than the mortality we experienced for other studies performed concurrently (12.5% over 4 weeks of experimentation) that involved implantation of the same type of polyvinyl catheters. It is possible that adenovirus affected cell viability directly. E1, E3-deleted adenovirus vectors inhibit cell proliferation in vitro by

promoting apoptosis and slowing the cell cycle34 and inhibit cell function and precipitate a rapidly developing inflammatory response characterized by infiltrating neutrophils, focal necrosis and edema when injected into the submandibular glands of rats.35 It is possible that the very young fetuses used in the present were particularly susceptible to the harmful effects of the vectors. The high incidence of mortality may also be related to the Hickman catheter used in group 2 animals (n = 7) that all died in utero. Implantation of this catheter for 16.7 days into dogs and cats is associated with a 13% complication rate.36 Other factors may also have been responsible for the high mortality. The animals were exposed to anesthesia (group 1) or surgical procedures (fetus 12A) multiple times. Of particular concern was the fibrotic tissue damage observed in this and a previous study.4 Administration of Av1nBg stimulated a robust inflammatory response that involved macrophage, neutrophil and lymphocyte migration into the exposed area, much the same as in adult animals.2,4,7 The capacity to mount an inflammatory response develops early in the sheep as phagocytic cells are detectable in the blood and vascular endothelium by 20 days after conception (0.13 term).37 A single injection of Av1nBg into the amniotic fluid cavity had little effect on lung histology, presumably because relatively few infectious particles reached the lung (Figure 3a). However, a direct vector injection into the fetal trachea elicited inflammatory and fibrotic responses (Figure 3b) similar to those we observed in a previous study.4 The intensity of these responses is likely to be related to the dose of vector administered, although this was not tested directly in the present study. Two fetuses that delivered vaginally following intra-amniotic and intratracheal injections of Av1nBg developed respiratory distress. Radiologic markings in the lung were prominent, and there was evidence of severe tissue damage in a lung biopsy specimen. It appears that prenatal exposure to adenovirus may permanently damage the developing lung. This is consistent with the concept that the premature lung is particularly susceptible to injury and has a limited capacity for repair compared with the adult. Dexamethasone and monocrotaline treatment permanently impair normal alveolar formation when given to neonatal rats during alveolarization.38,39 Infants who develop bronchopulmonary dysplasia and survive later have chronic lung damage characterized by more frequent respiratory tract infections, episodes of wheezing, reduced airway function and

Figure 3 Histological appearance of fetal lungs following administration of Av1nBg (a) Lung tissue from animal 11B at 80 days of gestation, 20 days following Av1nBg. Histological appearance is normally immature consistent with the gestational age. Budding airways are indicated by arrows, and mesenchyme is indicated by ‘m’. Inset of area indicated by the double arrow illustrates lack of inflammation and highly glycogenated epithelium characteristic of immature lung. (b) Lung tissue from animal 12B obtained 7 days after intratracheal instillation of Av1nBg. Airways are dilated with luminal exudate (e) composed of sloughed epithelial cells, necrotic cells, mononuclear inflammatory cells, occasional neutrophils and edema fluid. In addition there are tufts of reactively proliferated epithelium (arrows). (c) Lung tissue from animal 3 obtained by biopsy at 44 days of postnatal age. Bronchi are constricted with peribronchial fibrosis (f) and inflammation (i) illustrated both at low power and on the inset. The inflammatory infiltrate consists of lymphocytes and histiocytes. Sections were stained with hematoxylin and eosin. Bars = 62.5 ␮m. Insets are 2.5 × higher magnification.

103


104

airway obstruction.40 Thus, exposure of fetal sheep to adenovirus gene transfer vectors early in development may stimulate untoward inflammatory and immune responses that could have long-term consequences. Fetal gene therapy represents an important potential application and area of investigation. Fetal gene therapy approaches may prove useful for the treatment of genetic diseases such as surfactant protein B deficiency that are fatal at birth. Diseases such as Tay–Sachs disease, Lesch– Nyhan syndrome, sickle cell anemia and cystic fibrosis have significant morbidity after birth. It would be desirable to treat these diseases early in development, before destructive disease processes cause irreversible damage. It will be necessary to improve our understanding of the development of immune and inflammatory responses, develop better gene therapy vectors and conduct appropriate preclinical studies to ensure the safety and efficacy of fetal gene therapy.

Materials and methods Adenovirus preparation The recombinant adenovirus used in this study was the same serotype 5 adenovirus-derived vector, Av1nBg, we used in previous studies.4,5 Preparations of Av1nBg were carefully screened and certified free of pathogenic organisms and evaluated for the presence of E1a-containing ‘wild-type’ virus using quantitative PCR.41 The level of sensitivity of this detection system is 1 p.f.u. of wild-type virus in 108 p.f.u. Av1nBg was administered in doses, shown to be effective in previous studies,4,5 as indicated in Figure 1. As a control we also injected Av1Null (1 × 1010 p.f.u.), a vector identical to Av1nBg in every respect except lacking an expression cassette. Animal procedures All procedures were approved by the Institutional Animal Care and Use Committee at the Children’s Hospital Medical Center. A total of 19 fetuses in 12 pregnancies were studied; five were singletons and 14 were twins (Figure 1). Animals were studied in three separate groups of pregnant ewes (four per group). The general approach in these studies was to inject Av1nBg into the amniotic cavity several times before an intratracheal challenge. Intra-amniotic injections were administered between 57– 60 days and 100–110 days of gestation. At 100–110 days, Av1nBg was instilled into the fetal trachea and responses measured as described below. Group 1: injection of Av1nBg using ultrasound guidance (Figure 1a): Animals 1–4 were sedated with intravenous sodium pentobarbital (15–20 mg/kg) and anesthetized with 1–3% halothane in oxygen. The amniotic fluid cavity was visualized using real-time ultrasound (performed by Dr Neil Johnson, Department of Radiology). A needle was directed into the amniotic cavity, and Av1nBg was injected (IA) in the amounts described in Figure 1. Three of the four fetuses also received a subsequent intratracheal (IT) injection of Av1nBg at 104–108 days gestation. To accomplish this, pregnant ewes were fasted for 24 h, sedated with intravenous sodium pentobarbital (15– 20 mg/kg) and anesthetized with 1–3% halothane in oxygen. Polyvinyl catheters were placed into the maternal descending aorta and inferior vena cava, amniotic fluid

cavity and fetal ascending aorta, superior vena cava and trachea. Av1nBg was instilled directly into the catheter inserted in the fetal trachea. All surgical incisions were sutured, and the catheters were exteriorized to the ewe’s left flank and secured in a pouch. Vascular catheters were filled with heparin saline solution (1000 U/ml) and sealed. Ampicillin (1 g), gentamycin (40 mg), and metronidazole (50 mg) were administered to the ewe and amniotic cavity intraoperatively and daily for 3 days. One fetus (No. 3) delivered vaginally and survived until elective death at 72 days of age. During that time, a fifth and final dose of Av1nBg was administered intratracheally. Serum samples were obtained from fetuses before and after tracheal administration and analyzed for the presence of neutralizing antibody. Other tests on lamb 3 included several chest radiographs (at 3, 37, 44 and 58 days of age) and lung bronchoscopies (performed at 10, 20 and 35 days of age by Dr Carlos Perez of the Division of Pulmonary Medicine). An open-chest lung biopsy was performed by Dr Brad Warner of the Department of Surgery when lamb 3 was 44 days old, and the tissue examined for evidence of gene expression and histopathology.

Group 2: injection of Av1nBg using a subcutaneous port (Figure 1b): The outcome of the first study indicated that Av1nBg administration or the repeated anesthesia and injection procedures adversely affected fetal well-being. To avoid some of these possible complications, a second group of four animals (fetuses 5–8) was studied. A subcutaneous port was placed in the ewe’s flank (Hickman subcutaneous catheter, Bard Access Systems, Salt Lake City, UT, USA) under sterile conditions. It was connected to a catheter that had its free end placed in the amniotic cavity. Av1nBg (fetuses 5A, 6, 7A and 8A) or saline (fetuses 5B, 7B and 8B) was injected repeatedly into the port at the indicated times until fetal demise or surgical implantation of vascular and tracheal catheters as described for group 1. Group 3: direct injection of Av1nBg into the amniotic cavity (Figure 1c): The outcome of the first two studies indicated that early exposure to Av1nBg or the presence of the Hickman port adversely affected the fetus. To determine whether the mechanism of early fetal demise could be detected histologically, Av1nBg was instilled directly into the amniotic cavity at the time of surgery in fetuses 9–12. Fetuses were electively killed at the indicated times. One of these fetuses also had catheters implanted into the trachea, carotid artery and superior vena cava and received one intratracheal dose of Av1nBg at 114 days’ gestation. Group 4: administration of Av1Null: To address the possibility that expression of bacterial ␤-galactosidase stimulated cellular or humoral immune responses, we administered Av1Null, a replication-deficit adenoviral vector that lacks an expression cassette, to three fetal sheep at 55–57 days gestational age. Virus was administered intra-operatively into the amniotic cavity in amounts of 1 × 1010 p.f.u. Animal monitoring: When vascular catheters were implanted, fetal arterial pH, PO2, PCO2, hemoglobin concentration and blood oxygen saturation (model 170 blood gas analyzer, Ciba Corning Diagnostics, Medfield, MA,


USA; Hemoximeter model OSM2, Radiometer, West Lake, OH, USA) were determined daily. Lung fluid was collected before and after Av1nBg injection in fetuses with tracheal catheters. At least 5 ml of lung fluid (several times the catheter dead space volume) was collected gravimetrically from the tracheal catheter and discarded. Two additional milliliters of lung fluid were then collected and stored on ice until analysis was performed.

Analytical procedures When possible, fetal tissues (esophagus, stomach, intestine, liver, spleen, trachea, lung, thymus, ovary, kidney, skin and amniotic membranes) were dissected immediately after death and fixed (4% paraformaldehyde in 0.1 m phosphate-buffered saline, pH 7.4). Using procedures described previously, tissues were analyzed for endogenous and nuclear ␤-galactosidase activity.4 Additional tissue sections were stained with hematoxylin and eosin and examined blindly for evidence of histopathology by a clinical pathologist. Total cell count per milliliter of lung fluid was measured, and specific cell types were identified as described previously.4 Maternal and fetal sera were collected before and after intratracheal Av1nBg injection to determine whether adenovirus administration stimulated the production of specific antibodies. Serum neutralizing antibody titers against wildtype adenovirus were measured using an in vitro assay.42 To explore the possibility that wild-type adenovirus, a possible contaminant of the adenoviral vector preparation, proliferated in the fetal sheep and contributed to fetal demise, samples of fetal tissue, blood and amniotic fluid from fetuses 9A and B, 10A and B, 11A and B and the fetuses that received Av1Null were assayed. Each fluid sample and supernatant from tissue homogenates were serially diluted and incubated with confluent cultures of A549 cells for 6 days at 37°C. Evidence of cytopathic effect was visualized using crystal violet.

Acknowledgements This work was supported by the Cystic Fibrosis Foundation and the ‘Center for Gene Therapy of Cystic Fibrosis and Other Lung Diseases’, NIH HL51832 (HSI and JAW). We thank Nanette Mittereder, Reiko Tanaka and Christine Kane for technical assistance. We thank Kevin Kirwin, Robert Naylor and Michael Li for their assistance with lung fluid analyses. We thank Shirley Reising for her assistance in analyzing neutralizing antibody titers and proliferation of wild-type adenovirus.

References 1 Mastrangeli A et al. Diversity of airway epithelial cell targets for in vivo recombinant adenovirus-mediated gene transfer. J Clin Invest 1993; 91: 225–234. 2 Yei SP et al. In vivo evaluation of the safety of adenovirusmediated transfer of the human cystic fibrosis transmembrane conductance regulator cDNA to the lung. Hum Gene Ther 1994; 5: 731–744. 3 Lemarchand P, Jones M, Yamada I, Crystal RG. In vivo gene transfer and expression in normal uninjured blood vessels using replication-deficient recombinant adenovirus vectors. Circ Res 1993; 72: 1132–1138. 4 Vincent M et al. Adenovirus-mediated gene transfer to the respiratory tract of fetal sheep in utero. Hum Gene Ther 1995; 6: 1019–1028.

5 Holzinger A et al. Intra-amniotic administration of an adenoviral vector for gene transfer to fetal sheep and mouse tissues. Pediatr Res 1995; 38: 1–7. 6 McCray PB Jr et al. Adenovirus-mediated gene transfer to fetal pulmonary epithelia in vitro and in vivo. J Clin Invest 1995; 95: 2620–2632. 7 Simon RH et al. Adenovirus-mediated transfer of the CFTR gene to lung of nonhuman primates: toxicity study. Hum Gene Ther 1993; 4: 771–780. 8 Yang Y et al. Cellular immunity to viral antigens limits E1deleted adenoviruses for gene therapy. Proc Natl Acad Sci USA 1994; 91: 4407–4411. 9 Zsengeller ZK et al. Persistence of replication-deficient adenovirus-mediated gene transfer in lungs of immune-deficient (nu/nu) mice. Hum Gene Ther 1995; 6: 457–467. 10 Zabner J et al. Safety and efficacy of repetitive adenovirusmediated transfer of CFTR cDNA to airway epithelia of primates and cotton rats. Nat Genet 1994; 6: 75–83. 11 Billingham RE, Brent L, Medawar PB. ‘Activity acquired tolerance’ of foreign cells. Nature 1953; 172: 603–606. 12 Quantin B, Perricaudet LD, Tajbakhsh S, Mandel J-L. Adenovirus as an expression vector in muscle cells in vivo. Proc Natl Acad Sci USA 1992; 89: 2581–2584. 13 Stratford-Perricaudet LD, Makeh I, Perricaudet M, Briand P. Widespread long-term gene transfer to mouse skeletal muscles and heart. J Clin Invest 1992; 90: 626–630. 14 Dosch H-M, Hui MF, Doherty PJ. Development of normal and abnormal immune functions in humans. In: Cooper EL, NisbetBrown E (eds). Developmental Immunology. Oxford University Press: Oxford, 1993, pp 452–470. 15 Crombleholme TM, Bianchi DW. In utero hematopoietic stem cell transplantation and gene therapy. Semin Perinatol 1994; 18: 376–384. 16 Flake AW et al. Treatment of X-linked severe combined immunodeficiency by in utero transplantation of paternal bone marrow. New Engl J Med 1996; 335: 1806–1810. 17 Morris B. The ontogeny and comportment of lymphoid cells in fetal and neonatal sheep. Immunol Rev 1986; 91: 219–233. 18 Zepeda M, Wilson JM. Neonatal cotton rats do not exhibit destructive immune responses to adenoviral vectors. Gene Therapy 1996; 3: 973–979. 19 Takahashi M et al. Long-term correction of bilirubin-UDP-glucurononsyltransferase deficiency in Gunn rats by administration of a recombinant adenovirus during the neonatal period. J Biol Chem 1996; 271: 26536–26542. 20 Kass-Eisler A et al. Circumventing the immune response to adenovirus-mediated gene therapy. Gene Therapy 1996; 3: 154– 162. 21 Acsadi G et al. A differential efficiency of adenovirus-mediated in vivo gene transfer into skeletal muscle cells of different maturity. Hum Mol Gen 1994; 3: 579–584. 22 Tripathy SK et al. Stable delivery of physiologic levels of recombinant erythropoietin to the systemic circulation by intramuscular injection of replication-defective adenovirus. Proc Natl Acad Sci USA 1994; 91: 11557–11561. 23 Walter J et al. Successful expression of human factor IX following repeat administration of an adenoviral vector in mice. Proc Natl Acad Sci USA 1996; 93: 3056–3061. 24 Haynes BF, Denning SM. Lymphopoiesis. In: Stamatoyannopoulos G, Nienhuis AW, Majerus PW, Varmus H (eds). The Molecular Basis of Blood Diseases, 2nd edn. WB Saunders: Philadelphia, 1994, pp 425–462. 25 Mackay CR, Maddox JF, Brandon MR. Thymocyte subpopulations during early fetal development in sheep. J Immunol 1986; 136: 1592–1599. 26 Haynes BF, Heinly CS. Early human T cell development: analysis of the human thymus at the time of initial entry of hematopoietic stem cells into the fetal thymic microenvironment. J Exp Med 1995; 181: 1445–1458. 27 Silverstein AM, Prendergast RA, Kraner KL. Fetal response to antigenic stimulus IV. Rejection of skin homografts by the fetal lamb. J Exp Med 1964; 119: 955–964.

105


106

28 Zanjani ED et al. Engraftment and long-term expression of human fetal hemapoietic stem cells in sheep following transplantation in utero. J Clin Invest 1992; 89: 1178–1188. 29 Neiderhuber JE, Shermeta D, Turcotte JG, Gikas PW. Kidney transplantation in the foetal lamb. Transplantation 1971; 12: 161–166. 30 Osburn BI, Hoskins RK. Antibody responses to Brucella ovis in maternal and foetal sheep. J Infect Dis 1969; 119: 267–272. 31 Silverstein AM, Uhr JW, Kraner KL. Fetal response to antigenic stimulus. II. Antibody production by the fetal lamb. J Exp Med 1963; 117: 799–812. 32 Sekhon HS, Larson JE. In utero gene transfer into the pulmonary epithelium. Nature Med 1995; 1: 1201–1203. 33 Olver RE, Schneeburger EE, Walters DV. Epithelial solute permeability, ion transport and tight junction morphology in the developing lung of the fetal lamb. J Physiol 1981; 315: 395–412. 34 Teramoto S et al. Effect of adenoviral vector infection on cell proliferation in cultured primary human airway epithelial cells. Hum Gene Ther 1995; 6: 1045–1053. 35 Adesanya MR, Redman RS, Baum BJ, O’Connell BC. Immediate inflammatory responses to adenovirus-mediated gene transfer in rat salivary glands. Hum Gene Ther 1996; 7: 1085–1093.

36 Blaiset MA, Couto CG, Evans KL, Smeak DD. Complications of indwelling, silastic central venous access catheters in dogs and cats. J Am Animal Hosp Assoc 1995; 31: 379–384. 37 Miyasaka M, Morris B. The ontogeny of the lymphoid system and immune responsiveness in sheep. Prog Vet Microbiol Immunol 1988; 4: 21–55. 38 Todd L, Mullen M, Olley PM, Rabinovitch M. Pulmonary toxicity of monocrotaline differs at critical periods of lung development. Pediatr Res 1985; 19: 731–737. 39 Massaro D et al. Postnatal development of alveoli. Regulation and evidence for a critical period in rats. J Clin Invest 1985; 76: 1297–1305. 40 Northway WH et al. Late pulmonary sequelae of bronchopulmonary dysplasia. New Engl J Med 1990; 323: 1793–1799. 41 Trapnell BC et al. Expression of the cystic fibrosis transmembrane conductance regulator gene in the respiratory tract of normal individuals and individuals with cystic fibrosis. Proc Natl Acad Sci USA 1991; 88: 6565–6569. 42 Hierholzer J, Bingham P. Vero microcultures for adenovirus neutralization tests. J Clin Micro 1978; 7: 499–506.