Chapter 6. Discussion

The major objective of this experimental work was to develop a surgical method for organ-specific and effective gene transfer into the kidney, especially the glomeruli, aiming at targeted somatic gene therapy. The most essential disease model to be treated by this gene therapy method is Alport syndrome, which is a mostly X-chromosomally inherited disease caused by a mutation in the type IV collagen α5-chain gene, which causes structural and functional defects in the glomerular basement membrane (GBM) of the kidney. The renal manifestations of this disease are hematuria and proteinuria, which gradually lead to progressive renal failure and a need for dialysis and renal transplantation in males (Tryggvason et al. 1993). There are also some other organ manifestations in Alport syndrome, e.g. ocular lenticonus and sensory hearing loss, but these are not life-threatening. The gene mutation can be identified by DNA analysis from peripheral blood leukocytes. Theoretically, the progressive renal disease could be prevented by a transfer of a normal type IV collagen α5-chain gene into glomerular epithelial cells.

The first renal gene transfer trial was reported in 1991 by Koseki et al. Several vectors have previously been used for renal gene transfer: retrovirus, adenovirus and liposomes (Tomita et al. 1992, Woolf et al. 1993, Moullier et al. 1994, Kitamura & Fine 1997). Several methods have been reported for renal gene transfer, and the distribution of gene expression in the kidney depends on both the vector and the method used. Intra-arterial infusion of plasmid/liposome complexes has yielded transfection of renal tubular cells (Lai et al. 1997). Retrograde infusion of adenovirus or liposomes into the ureter or the cavity of the renal pelvis has resulted in similar results, transfection of mainly tubular structures, but not glomeruli. Glomerular transfection has been obtained by using ex vivo retroviral transduced cultured mesangial cells (Kitamura et al. 1997) and by intra-arterial HVJ-liposome infusion (Tomita et al. 1992). Direct intra-parenchymal injection of retroviral vectors resulted in weak transfection of tubular cells pretreated with cell proliferation inducing nephrotoxic agent (Bosch et al. 1993). Implantation of ex vivo transduced metanephric tissue into the renal parenchyma has yielded gene expression in glomerular epithelial cells, interstitial cells and the renal vasculature (Woolf et al. 1993). However, the expression efficiency in these studies has been low.

In the first part of this study, the adenoviral gene transfer (AdCMVlacZ) into cultured human endothelial and glomerular mesangial cells obtained by graded sieving (Misra 1972), resulted in intense expression of the β -galactosidase reporter gene in both cell lines, as shown by the intense blue color after staining with X-gal. This proved the usefulness of the adenovirus vector for gene transfer into glomerular cells. The adenoviral vector was chosen here, because it has proved its ability to transfect effectively various non-dividing kidney cells in several previous studies (Kozarsky & Wilson 1993, Chang et al. 1994, Moullier et al. 1994, Sukhatme et al. 1997). The endothelial and epithelial cells of the glomeruli are not highly proliferative, which makes them unsuitable targets for DNA replication requiring retrovirus vector. In addition, the adenovirus is safe and easy to handle and preserve (Wilson 1996). It yields high titers of virus stocks and is able to carry large DNA inserts, such as the human 5.3 kb α5(IV) collagen chain gene.

The in vivo intra-arterial infusions of adenoviral vectors alone in this study did not lead to any remarkable gene transfer into porcine splenic or renal cells. The findings of Moullier et al. on rat kidneys were roughly similar: transgene expression was found in scattered renal endothelial cells and in some renal tubular cells after intra-arterial infusion of the adenoviral lacZ gene (Moullier et. al. 1994). In their study, however, the glomerular cells also remained negative. On the other hand, the total number of vectors used in their study was higher in proportion to the difference in size between the rat kidney and the lower pole of the porcine kidney. The number of vectors per amount of kidney tissue in our study was lower. This may explain the somewhat better transfection efficiency with intra-arterial infusion of vectors in their study. There may also exist differences in adenoviral infectivity between species (Katkin et al. 1995, Chapellier et al. 1996). For these reasons, these studies are not fully comparable. Lai et al. similarly reported gene transfer only into renal tubular epithelial cells, but not into glomerular cells after intra-arterial infusion of liposomes in mice (Lai et al. 1997). Although plasmid-mediated gene transfer has been reported to be enhanced in the previous studies by adjuvant pharmacological agents, especially bupivacaine (Danko et al. 1994), no such enhancement was observed in this study with adenoviral vectors. Intra-arterial adenoviral gene transfer was not improved here by pharmacological agents, such as verapamil, lidocain, enalapril, papaverin, or alprostadil, infused shortly before the viral vectors. These agents were chosen because they are known to cause vasodilatation, which could theoretically improve access of the viral vectors into the glomerular structures by dilating the capillaries. However, no such effect was observed here. These findings suggest that in vivo effective adenoviral-mediated gene transfer into renal glomerular cells could not be achieved by single intra-arterial infusions of vector even in the presence of vasodilative pharmacological agents.

The efficiency of adenoviral gene transfer into mature non-dividing cells can be enhanced by prolonging the incubation time with the vectors and target cells, as it has been shown in several previous studies (Zabner et al. 1996, Teramoto et al. 1997, Chia et al. 1998). Based on the hypothesis that prolongation of the incubation time with the adenoviral vectors and glomerular cells improves transduction efficiency, the closed-circuit surgical organ perfusion method for in vivo organ-specific gene transfer was developed here. In this method, the viability of the organ during the prolonged circulation of the vector solution is guaranteed by providing a continuous oxygen supply to the cells (Hamilton et al. 1974), which is necessary especially in the kidney, which tolerates warm ischemia for 15 to 20 minutes at the most. During the cannulation in this study, warm ischemia lasted for approximately 10 minutes. The perfusion system consisted of a reservoir for the perfusate, a membrane lung, a roller pump and a heating apparatus controlled by thermostat, all connected by silicon tubing that allowed the perfusate to circulate. The technical aspects of the perfusion method were primarily tested by ex vivo closed-circuit perfusion of the kidney at room temperature in two experiments. As this did not lead to any remarkable gene transfer, the perfusion temperature was raised up to body temperature, 37°C. The absence of transgene expression at room temperature may result from various reasons: either the vector is not able to adhere to the cell membrane or it cannot pass through the cell membrane at lower temperatures, or the β -galactosidase transgene expression leading to protein synthesis requires the normal body temperature to take place. This is supported by the finding of Zeigler et al. (1996a), who studied adenoviral gene transfer into human cadaveric kidneys by renal artery infusion and incubation of adenoviral vectors for 3 hours at 4°C, after which the temperature was raised to 20°C. However, they failed to find any glomerular β -galactosidase expression after 12 hours, though some expression was seen in the tubular epithelial cells. In the present study, ex vivo closed-circuit kidney perfusion at body temperature for 12 hours resulted in highly effective glomerular gene transfer. Up to 80% of the glomeruli showed strong expression of the β -galactosidase gene. The observed essential effect of temperature on the transgene expression efficiency precluded cold perfusion of the kidney for gene transfer. Therefore, the closed-circuit warm perfusion system with a continuous oxygen supply to kidney cells is considered necessary to attain effective glomerular gene transfer.

Since the ex vivo closed-circuit organ perfusion method proved particularly effective for gene transfer into glomerular cells, this system was subsequently applied to in vivo renal gene transfer experiments with 60- and 120-minute kidney perfusions in a porcine model, which resulted in up to about 75% of the kidney glomeruli being transfected four days after the 120-minute perfusion. These results indicate the most effective in vivo glomerular gene transfer in the kidney reported so far (Kitamura & Fine 1997, Imai & Isaka 1998). Previously, Tomita et al. found 15% of the glomeruli to express the transgene after HVJ-liposome intra-arterial infusion in rats (Tomita et al. 1992). Kitamura et al. reported transgene expression in 57% of the glomeruli after retrovirus-mediated ex vivo transfection of mesangial cells following infusion of these transduced cells into the renal artery (Kitamura et al. 1994). In both of these studies, the vector used was different from that in the present study. In view of potential gene therapy for Alport syndrome, mesangial cells are not possible target cells. Gene transfer in this study using an adenoviral vector was almost selective into the glomeruli, while the other endothelial cells of the renal vascular system generally showed little expression and the tubular cells hardly any expression. The very scant excretion of the virus into urine during the perfusion can partly explain the negativity of tubular cells. No significant amounts of the adenovirus passed through the GBM, and the virus thus did not come into contact with the tubular cells.

Adenovirus-mediated gene transfer proved to be transient because of its immunogenicity on the one hand and because of the episomal location of the transgene in the cell nucleus on the other (Zsengeller et al. 1995). The transient expression obtained by adenoviral gene transfer was consequently also observed in this study. The glomerular β -galactosidase expression was intense four days after the perfusion, but hardly detectable after four weeks, and had completely disappeared after three months.

An inflammatory response was detected in paraffin-embedded kidney samples in the form of inflammatory mononuclear cell clusters, but no remarkable humoral immunological response against AdV or β -galactosidase was elicited, except in one animal, which showed slight elevation of anti-IgG AdV antibodies. Van Ginkel et al. 1995 reported stimulation of systemic IgG and IgA formation after intratracheal administration of AdV in mice. Theoretically, the absence of a humoral immunological response here could be partly explained by the locally administered viral vectors used in the perfusion method. Additionally, the excess transgene material was flushed off from the organ at the end of the perfusion. This may prevent the immunological system from “discovering“ effectively the foreign material and thus reduce the humoral immunological response. This protective mechanism of the local administration of adenoviral vectors was also suggested by Zeigler et al. (1996b), who transplanted murine kidneys transfected ex vivo by an adenovirus vector. The present animals additionally received 50 mg hydrocortisone as a single intramuscular injection immediately after the operation. However, the dose was very small in view of the possible positive effect on transduction efficiency and the anti-inflammatory effect. Besides, the role of immunosuppression has been rather related to a longer duration of expression and a better retransfection efficiency than the primary transfection efficiency of adenoviral vectors (Kaplan et al. 1997, Cassivi et al. 1999). The small number of experiments in this study does not, however, allow any definitive conclusions about the effects of the perfusion system on the immunological response. However, there were inflammatory changes and some degree of tubular necrosis in the viral perfusion-treated kidneys, while the control kidney was relatively normal. This also suggests that renal cells are destroyed by immunological mechanisms, most probably a cytotoxic response against AdV, rather than by the perfusion method itself.

The present results of effective glomerular gene transfer obtained by the novel closed-circuit warm organ perfusion method developed in this study are particularly important in view of the further development of gene therapy for glomerular diseases, such as Alport syndrome. There is a well characterized animal model for Alport syndrome -like hereditary nephritis in Samoyed dogs (Thorner et al. 1996). To find out the final step in gene therapy research for Alport syndrome, the formation of normal type IV collagen and, consequently, normal GBM in genetically defective animals and individuals has to be explored. However, the vectors and the duration of transgene expression in the target cells have to be better optimized, possibly by generating less immunogenic viral vectors or by combining viral- and liposome-mediated transfection techniques. There are also other potential applications for gene transfer into the kidney glomeruli by an organ perfusion method, such as genetic diseases, including polycystic renal disease, inflammatory and fibrotic diseases of the kidney and transplant rejection prevention (Zeigler et al. 1996a, Moullier et al. 1997, Wissing et al. 1997). Technically, the perfusion method for gene transfer could be especially suitable for the transfer of anti-inflammatory or antiproliferative genes in conjunction with organ transplantation.

As the closed-circuit perfusion method resulted in effective adenoviral gene transfer into renal glomerular cells, the suitability of this method for targeted gene transfer into other organs was also tested. To institute potential gene therapy for metabolic lysosomal diseases or systemic diseases based on an enzyme defect of genetic origin, a normal copy of the defective gene could be transferred into the spleen for corrective gene therapy. One example of such disease is Gaucher disease, which is characterized by an inherited deficiency of the glucocerebrosidase enzyme gene, resulting in glucocerebroside accumulation in visceral organs, especially the spleen and liver (Eto & Ida 1996). Medin et al. (1996) reported in vitro correction of the enzymatic defect in fibroblasts and hematologic cells derived from patients with Fabry disease by retroviral transfection of α-galactosidase A. This could be applicable to in vivo gene therapy of Fabry disease by delivering the α-gal A gene into the spleen for continuous corrective protein secretion.

The spleen is a major component of the phagocytic system and an important secondary lymphoid organ. The main functions of the spleen are the removal of effete blood cells, particulate matter and micro-organisms from the circulation as well as participation in the primary immune response to blood-borne antigens. The adult spleen is a reservoir of thrombocytes. Anatomically, the spleen is covered by a dense connective tissue capsule, from which trabeculae extend towards the interior of the organ. Splenic tissue is composed of the red and the white pulp, between which the perifollicular area is situated. The blood flow arrives through the splenic artery, which branches at the hilus into the trabecular arteries. The trabecular arteries follow the trabecular veins and the efferent lymph vessels into the white pulp, where they are called the central artery. The central arteries branch further and terminate as arterioles and arterial capillaries in the lymphatic nodules, the marginal zone, the perifollicular area and the red pulp. The vast majority of capillaries terminate in the reticular spaces of the perifollicular area or in the cords of the red pulp (Ross et al. 1989). Some of the capillaries are surrounded by macrophages. For the treatment of lysosomal storage diseases by somatic gene therapy, the therapeutic gene transfer should be targeted especially into macrophages in the splenic tissue. In this study, closed-circuit in vivo perfusion of the spleen resulted in gene transfer into splenic cells, mostly macrophages and endothelial cells.

The β -galactosidase reporter gene was transferred into the spleen by the closed-circuit organ perfusion method in eight pigs. Throughout the experiments, perfusion for 60 minutes in vivo resulted in relatively evenly distributed, effective reporter gene transduction after four days in different types of splenic cells, which were analyzed visually on histological sections after X-gal staining. The β -galactosidase reporter gene expression was seen mostly in the macrophages situated around the arterioles and arterial capillaries, which terminate in the perifollicular zone between the white and the red pulp and in the red pulp. Some smooth muscle cells and endothelial cells of the white pulp central arteries and arterioles leaving the white pulp also expressed the reporter gene.

There are only a few previous reports concerning gene transfer into the spleen (Moullier et al. 1993, Kaido et al. 1996, Ohashi et al. 1997). Kaido et al. (1996) implanted rat fibroblasts transduced ex vivo with the human hepatocyte growth factor (HGF) gene by a direct needle injection into the spleen. They found that subsequent de novo synthesis of HGF in the spleen reduced the liver injury caused by CCl₄ administration. Moullier et al. (1993) used intraperitoneally implanted collagen lattices consisting of autologous genetically modified skin fibroblasts transfected ex vivo with the β -glucuronidase enzyme gene in mucopolysaccharidosis type VII transgenic mice. They reported correction of the lysosomal storage pathology in the liver and spleen in the treated animals. An alternative method of gene delivery would be in vivo delivery directly into the spleen. Ohashi et al. (1997) used intravenous infusion of recombinant adenovirus carrying the β -glucuronidase gene and found 20% elevation of β -glucuronidase activity in the spleen and correction of pathological abnormalities in the spleen of transgenic β -glucuronidase-deficient mice. On the contrary, the previous studies by Jaffe et al. (1992) on hepatic gene transfer via portal vein infusion of adenoviral vectors failed to reveal any transgene expression in the spleen. Thierry et al. (1995) studied the distribution of marker gene transfer after a systemic single intravenous injection of plasmid/liposomes carrying the firefly luciferace reporter gene in mice. Luciferase activity was found in the liver, spleen and lung four days post-injection. They also compared intravenous, intraperitoneal and subcutaneous modes of administration and found that i.p. administration resulted in gene transfer preferentially into the spleen, while i.v. administration resulted in slightly more efficient activity in the liver and lung. In an attempt to obtain pancreatic gene transfer, Schmid et al. (1998) also reported lacZ gene transfer into the spleen and liver after intra-arterial infusion of cationic liposomes into the coeliac trunk in rats. In these study reports, the specific splenic cells expressing the transgene have not usually been determined. Additionally, quantification of the expression is difficult. In practice, the effectiveness of transfection was evaluated by measuring the de novo secreted enzyme activity or by analyzing the correction of enzyme defects.

In the present spleen gene delivery experiments, the warm perfusion method was used with an oxygen supply similarly to the kidney perfusions, even though the spleen might tolerate ischemia much better than the kidney. Splenic perfusion could probably also be done without oxygenation for 60 minutes. This possibility was not, however, examined here.

Based on several previous studies and the kidney experiments of this study, the transient nature of adenoviral gene transfer is well known, and the duration of expression was hence not assessed here with splenic gene transfer. Interestingly, when compared to the non-treated part of the spleen or totally untreated spleen, no signs of inflammation or cytotoxic effects of the perfusion or the vector were found. One explanation for this may be the natural function of the spleen as part of the immune system. Additionally, the operation was clinically well tolerated by the test animals.

The present results with adenovirus perfusion showed quite remarkable gene transfer into lung cells, including alveolar epithelial cells. The lung is an important target for gene therapy research, because this organ is often severely affected by genetic and acquired diseases, e.g. cystic fibrosis, α1-AT deficiency or pulmonary hypertension, leading to fatal complications. The lung is accessible for gene transfer through the airway or via the intravascular route. Recombinant adenovirus, liposomes and adeno-associated virus have been used as vectors in lung gene transfer (Lemarchand et al. 1994, Lee et al. 1998). Recombinant adenovirus vectors have been the most widely employed as vehicles for gene delivery into the lung because of their natural tropism for the respiratory tract. Besides, compared with cationic liposomes, the transfection efficiency of the adenoviral vectors has been better (Curiel et al. 1996).

The most logical mode of vector administration into the lung is intratracheal instillation (Jeppson et al. 1998, Halbert et al. 1998, Griesenbach 1998) or aerosol delivery (Katkin et al. 1995, Yonemitsu et al. 1997), both of which have been extensively studied using recombinant AdV. These methods have, however, yielded low efficiencies of transfection into alveolar and bronchial epithelial cells. Intravenous delivery of plasmid/luciferase cDNA have been enhanced by prolonging the incubation time with the lung and the vector in rodents (Song & Liu 1998). The quantification of expression in that study, however, is not comparable to the present study, because different marker genes were used.

The intra-arterial route for vector delivery is another method of gene transfer into the lungs. However, the overall adenoviral gene transfer efficiency in the lung, which has been studied widely in rats, has been low, with only 1–8% of lung cells expressing the transgene after intra-arterial delivery (Schachtner et al. 1995). Lemarchand et al. (1994) pointed out the possibility of adenoviral pulmonary gene transfer in sheep by a single infusion via the pulmonary artery with simultaneous arterial and venous occlusion for 15 minutes, which provides quite a short interaction time for the vectors and pulmonary cells. In their study, 76% of the test animals showed scant transgene expression in the lung cells, but the expression was low, patchy and variable. Chapelier et al. (1996) used ex vivo infusion of recombinant adenovirus into the porcine upper lobe pulmonary artery in conjunction with pulmonary allotransplantation. They incubated the virus solution in the lung lobe during cold preservation at 10°C for two hours before transplantation. Three days after transplantation, the detectable lacZ expression was low, as less than 1% of the pulmonary cells showed expression. The gene transfer in their study was detected in alveolar and bronchial epithelial cells, endothelial cells of the pulmonary vessels and a few submucosal glands.

In the present study, we were able for the first time to show the possibility of adenovirus‐mediated reporter gene transfer into the lung by prolonging the incubation time with the continuous closed-circuit warm organ perfusion method via the vascular route. Similarly to the previous studies, gene transfer into the lungs by closed-circuit perfusion was not very cell type-specific (Mastrangeli et al. 1993). It is, however, of significance that bronchial epithelial cells and alveolar epithelial cells, which are essential for the expression of the CF or surfactant protein B gene, for example, showed the most intense β -galactosidase transgene expression. This is accordant with the observation by Schachtner et al. (1995) that pulmonary arterial administration of adenoviral vectors yields primarily transfection of nonvascular cells, mostly alveolar epithelial cells, followed by bronchial epithelial cells. In addition, the transfection efficiency in the present study was reproducible and relatively even throughout the perfusion experiments lasting for 60 minutes. The expression in the bronchial epithelium obtained via the vascular route may be explained by the anastomoses between the pulmonary and bronchial circulations. Since the transient duration of expression in the pulmonary cells after adenoviral gene transfer has been pointed out in the previous studies (Muller et al. 1994, Zsengeller et al. 1995) and was also shown in the present perfusion experiments with kidney, the duration of expression in the pulmonary cells was not assessed here.

In this study, oxygenation of the perfusate was also used in lung perfusion, because the lung lobe was neither inflated nor cooled during the 60 minutes of perfusion, to prevent ischemic injury to the lung. As observed in the kidney perfusions, keeping the organ at body temperature during the viral perfusion may be essential for the transduction to succeed. This assumption is supported by the finding of Chapellier et al. (1996) on low-efficiency transfection after cold incubation of the lung with vectors. In the present study, gene transfer was only seen in the perfused middle lung lobe cells, not in the upper or lower lobes, nor in the contralateral lung cells, with the exception of the slight weak expression in the alveolar macrophages in the contralateral lung. The mechanism of low-intensity lacZ gene expression in alveolar macrophages in the contralateral untreated lung is not clear. One explanation may be intratracheal movement of transfected macrophages via cough reflexes and mucus or nonspecific endogenous expression in macrophages. Targeted gene transfer into the treated part of lung only was reported by Lee et al. (1998), who reported liposome-mediated gene transfer into the left lung of rats after 50 minutes of incubation of the vectors in a lung isolated from the circulation by clamping the vessels. In their study, the single intravenous infusion of the liposome vector resulted in distributed weak expression in the lungs, heart, liver and kidneys. In contrast to the findings of Lee et al., the lung lobe in the present study was not ischemic during the perfusion, which prevents any deleterious effects of ischemia on the lung. On the other hand, the effects of ischemia may render the cell membranes more permeable to the vector and may thus facilitate transfection.

The perfused lung lobe showed only some scant chronic inflammatory infiltrations and intra-alveolar edema in most cases and alveolar hemorrhage in one case, but neither signs of diffuse alveolar damage nor hypertrophy of pulmonary muscular arterioles, which excluded pulmonary hypertension during perfusion. The results suggest that the gene transfer procedure described here is safe and does not cause remarkable harmful reactions in lung tissue. Some chronic inflammatory cell clusters containing mainly lymphocytes were notably smaller than we found previously in the kidney after the kidney perfusion experiments.

Up to the present, we do not exactly know the requirements of gene transfer efficiency needed to correct the pulmonary manifestations of CF, for instance, but it has been estimated that 6–10% expression in pulmonary epithelial cells may be sufficient in this disease (Alton & Geddes 1995).

Gene transfer with the perfusion method, using retroviral vectors to carry hGH cDNA into the mammary gland in three goats resulted in detectable, but low human growth hormone secretion into the milk. As previously reported, the duration of retroviral gene expression was also long after these experiments, being up to five months in one experiment (Moullier et al. 1993). After one unsuccessful experiment, we later observed that the infectivity of the retroviral vectors was reduced, which probably explains the poor results in that experiment.

The limitations of the present study include the small sample size and the difficulties in the quantification of gene expression, except in the kidney, which allows easy counting of the glomeruli. However, the feasibility of gene transfer into the kidney, spleen, lung and mammary gland with the closed-circuit organ perfusion method is evident.

The prerequisite for gene transfer with this closed-circuit perfusion method is that the organ has a suitable blood circulation system. There should be at least one end artery and a suitable venous system to collect the perfusate back into the perfusion system. Such organs as the kidney, lung, spleen and possibly liver are suitable target organs for a gene transfer method of this kind. During the perfusion, monitoring of perfusion pressure is essential to prevent any pressure-related damage to the organ. The advantages of this method include the possibility to avoid administration of large amounts of foreign genetic material into an immunocompetent individual and to cut the costs, as less vector material is needed. So far, however, this novel closed-circuit perfusion method for gene transfer requires general anesthesia and major surgery. The gene transfer efficiency obtained by this method seems to be good, especially in the kidney, and the operation has been relatively well tolerated by the organs and the test animals. In the future, in conjunction with the development of more optimal vectors, less invasive methods for perfusion, such as catheterization and possibly endoscopy will be investigated. It is likely that this closed-circuit organ perfusion method will be suitable to specific gene therapy approaches, especially the treatment of genetic diseases affecting principally one organ system. This method will hence complement the variety of methods that can be used for gene transfer in the future.