Adenovirus-mediated ex vivo gene transfer of basic fibroblast ... - Nature

Gene Therapy (2001) 8, 837–845  2001 Nature Publishing Group All rights reserved 0969-7128/01 $15.00 www.nature.com/gt

RESEARCH ARTICLE

Adenovirus-mediated ex vivo gene transfer of basic fibroblast growth factor promotes collateral development in a rabbit model of hind limb ischemia N Ohara1, H Koyama1, T Miyata1, H Hamada2, S-I Miyatake3, M Akimoto4 and H Shigematsu1 1

Division of Vascular Surgery, Department of Surgery, Graduate School of Medicine, The University of Tokyo, Tokyo; 2Department of Molecular Medicine, Sapporo Medical University, Hokkaido; 3Department of Neurosurgery, Osaka Medical College, Osaka; and 4 Department of Ophthalmology, Shinshu University, Nagano, Japan.

Adenovirus-mediated ex vivo gene transfer of basic fibroblast growth factor (bFGF), a new strategy for the treatment of chronic vascular occlusive disease, was examined in a rabbit model of hind limb ischemia. The left femoral artery was completely excised to induce an ischemic state in the hind limb of male rabbits. Simultaneously, a skin section was resected from the wound, and host fibroblasts were cultured. The cultured fibroblasts were infected with adenovirus vector containing modified human bFGF cDNA with the secretory signal sequence (AxCAMAssbFGF) or LacZ cDNA (AxCALacZ). At 21 days after femoral artery excision, the gene-transduced fibroblasts were administered through the left internal iliac artery. The fibroblasts significantly accumu-

lated in the ischemic hind limb, and the AxCAMAssbFGFtreated cells secreted bFGF for less than 14 days without elevation of systemic bFGF level. At 28 days after cell administration, calf blood pressure ratio, angiographic score, capillary density of muscle tissue and blood flow of the left internal iliac artery were determined, and animals with AxCAMAssbFGF-treated cells showed significantly greater development of collateral vessels, as compared with those with AxCALacZ-treated cells. These findings suggest that adenovirus-mediated ex vivo gene transfer of bFGF was effective for improvement of chronic limb ischemia. Gene Therapy (2001) 8, 837–845.

Keywords: adenovirus; basic fibroblast growth factor; ex vivo gene transfer; chronic ischemia; collateral vessel

Introduction Induction of vasculogenesis, angiogenesis and arteriogenesis to enhance collateral vessel development and tissue perfusion is expected to open a new strategy in the treatment of chronic vascular occlusive disease,1,2 and this concept was termed ‘therapeutic angiogenesis’.3,4 Vasculogenesis, angiogenesis and arteriogenesis are known to be promoted by some growth factors, such as bFGF and vascular endothelial growth factor (VEGF).2 In order to develop a therapy to enhance collateral development, these growth factors have been tested in vivo with several delivery techniques.4–7 Direct administration of growth factor protein was initially investigated in animal models of ischemia, which showed significant development of collateral vessels and improvement of the ischemic state.4,5,8 However, high-dose bolus delivery has a high potential risk of systemic toxicity through cytokine diffusion into the systemic circulation. Gene transfer of angiogenic growth factors is another approach for therapeutic enhancement of collateral development. In recent studies, a growth factor gene was Correspondence: T Miyata, Division of Vascular Surgery, Department of Surgery, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113–8655, Japan *The first two authors contributed equally to this paper Received 25 October 2000; accepted 29 March 2001

transferred into arterial wall cells,6 or muscle cells7 of ischemic tissue. Gene therapy directed to arterial wall cells, however, has limitations in the delivery to complicated arterial lesions and efficiency of gene transfer.6 Meanwhile, Tsurumi et al7 transferred the VEGF gene to muscle cells by intramuscular injection of naked DNA, and collateral vessel development was significantly augmented. This method utilized the unique profile of striated muscle cells that potentially take up and express a foreign gene transferred in the form of naked plasmid DNA. Since the DNA was injected intramuscularly, there was less limitation in its application to human ischemic extremities. In this study, we developed an alternative strategy for gene transfer to ischemic limb tissue in a rabbit model. One distinct feature of the strategy was that the angiogenic growth factor gene was introduced into ischemic tissue via an ex vivo method. The ex vivo method is defined as auto-transplantation of host cells with transfer of a certain gene. We used skin fibroblasts harvested from the host animal as the target for gene transfer, and infused the fibroblasts through an intra-arterial catheter after gene transfer. Another unique point was that the gene of modified human bFGF fused with the secretory signal sequence of interleukin-2 (IL-2) was transferred to the fibroblasts, since the original bFGF gene lacks the signal sequence for secretion.9,10 To transfer the bFGF gene to host cells, an adenovirus vector was used in the

Ex vivo gene transfer promotes collateral development N Ohara et al

838

present study. Transduced gene expression was high, but transient with the adenovirus vector.11 We think that these characteristics are important for the safe application of this therapy, since long-term expression of growth factor might initiate unexpected side-effects.

Results In vitro study To assess secretion and expression of the infected cells, cultured rabbit fibroblasts were infected with adenovirus vector containing modified human bFGF cDNA with (AxCAMAssbFGF, secretory group) or without the signal sequence (AxCAJSbFGF, native group). Western blot analysis showed the time course of bFGF expression in both the culture medium and the cell lysate. Two forms of bFGF (18 and 22 kDa) were observed in the medium of the secretory group with a maximum at 4–10 days after infection, though no bFGF was detected in the native group medium (Figure 1a). Another form (24 kDa) of

bFGF was detected in the cell lysate (Figure 1b). The enzyme-linked immunosorbent assay (ELISA) data showed that the secreted bFGF level in the medium of the secretory group was significantly higher than that in the native group from 1 day to 21 days after infection, and that the ratio of the secretory group value to the native group value at each time point was 4.85–36.2 (Figure 1c). Although the expressed bFGF in the cell lysate of the secretory group was also significantly higher than that in the native group from day 1 to day 28, the ratio of the secretory group value to the native group value at each time point was 1.86–3.68, which was lower than that of secreted bFGF values in the medium (Figure 1d). The DNA synthesis activity of secreted bFGF in the medium was quantified in cultured fibroblasts by incorporation of 3H-thymidine. The uptake of 3H-thymidine by rabbit fibroblasts increased when the conditioned medium of the secretory group was added, and this uptake was significantly higher than that in the native group until 28 days after infection (Figure 1e).

Ex vivo gene transfer and distribution of administered cells For evaluation of the angiogenic response in vivo, the left femoral artery of the rabbit was completely excised (ischemic model). At 21 days after femoral artery excision, 5 × 106 fibroblasts, infected with AxCALacZ (control group, n = 12) or AxCAMAssbFGF (bFGF group, n = 11), were injected as a bolus via the left internal iliac artery (Figure 2a). Before the experiment, 111In-labeled fibroblasts were injected into rabbits in the same manner, and the distribution of administered cells in organs and tissues was assessed. The distribution of 111In-labeled fibroblasts revealed significant accumulation of cells in above and below knee muscles of the left hind limb (Figure 3a). Although no significant accumulation was detected in other organs and tissues, 5.4% and 2.7% (mean) of the labeled cells were detected in the lung and liver, respectively. To determine whether the significant cell accumulation in the left limb was specific to ischemic tissue, we injected both 111In-labeled fibroblasts and 51Cr-labeled microspheres into the left ventricle of the same rabbit, and compared the cell distribution with regional blood flow calculated from the microsphere data. Cell distribution in the bilateral hind limbs was highly correlated with their regional blood flow (Figure 3b), and both cell distribution and regional blood flow in muscles of the right hind limb were significantly higher than those of the left ischemic hind limb (Figure 3c).

Figure 1 Western blot probed with anti-bFGF antibody shows the time course of bFGF expression in the culture medium (a) and cell lysate (b) of infected rabbit fibroblasts. ELISA shows bFGF level in the culture medium (c) and cell lysate (d). Italic figures indicate ratio of the secretory group value to native group value. (e) The DNA synthesis activity of secreted bFGF in the medium, which was measured by counting incorporation of 3H-thymidine. Values are shown as mean ± s.d. (*P ⬍ 0.01, **P ⬍ 0.05). PC, positive control; C, control sample; d, days after infection; N, native group; S, secretory group. Gene Therapy

Calf blood pressure ratio In the study using the ischemic model, the ratio of left calf systolic pressure to right calf systolic pressure (calf blood pressure ratio) showed no significant difference between the control group and bFGF group before cell administration. At 28 days after infected cell injection, calf blood pressure ratio in the bFGF group was significantly higher than that in the control group (Figure 4). To evaluate the influence of intra-arterial injection of 5 × 106 fibroblasts, calf blood pressure ratio was measured immediately after injection, and no significant difference was detected between the pressure ratio immediately before and after cell administration (Figure 4). We also examined the effect of collateral development in non-


839

Figure 2 Experimental protocols of ischemic model (a) and non-ischemic model (b).

ischemic tissue; 5 × 106 fibroblasts infected with AxCALacZ (n = 5) or AxCAMAssbFGF (n = 5) were injected through the left internal iliac artery of normal rabbit (non-ischemic model, Figure 2b). In the non-ischemic model, the ex vivo gene transfer induced no effect on calf blood pressure ratio at 28 days after cell administration (Figure 4).

Angiographic score At 28 days after administration of infected fibroblasts to rabbits of the ischemic model, angiographs showed few collateral arteries in the control group (Figure 5a). In contrast, many collateral vessels had developed in the bFGF group (Figure 5b). Angiographic score in the bFGF group demonstrated a significant increase of collateral vessels as compared with that in the control group, while no significant difference was detected in the study using the non-ischemic model (Figure 5c).

Figure 3 (a) Distribution of 111In-labeled fibroblasts in organs and tissues at 5 h after cell injection into the left internal iliac artery. Data are presented as percentage of radioactivity distributed in each tissue relative to total radioactivity of administered cells. (b) Correlation between cell distribution in bilateral hind limb muscles after cell injection into the left ventricle and their regional blood flow measured with microspheres. 111Inlabeled fibroblasts and 51Cr-labeled microspheres were injected into the left ventricle. These two values were significantly correlated by linear regression. (c) Cell distribution after administration into the left ventricle and regional blood flow in bilateral hind limb muscles. Values are shown as mean ± s.d. (*P ⬍ 0.01, **P ⬍ 0.05). AK, above-knee muscles; BK, below-knee muscles; closed circles, values of above-knee muscles; open circles, values of below-knee muscles; L, left hind limb; R, right hind limb; open bars, cell distribution; hatched bars, regional blood flow.

Capillary density, arterial diameter, and in vivo blood flow In the bFGF group of the ischemic model, capillary density of the left semimembranous muscle was significantly higher, diameter of the left caudal gluteal artery was sigGene Therapy


840

Figure 4 Calf blood pressure ratio immediately after femoral artery excision, immediately before, immediately after and 28 days after cell administration. Values are shown as mean ± s.d. (*P ⬍ 0.01). NS, no significant difference.

nificantly larger, and blood flow at rest and maximum blood flow of the left internal iliac artery were also significantly higher than those in the control group (Figure 6a, b and c). On the contrary, in the non-ischemic model, no significant difference between the two groups was observed in capillary density and blood flow (Figure 6a and c).

Systemic bFGF level and anti-adenovirus antibody titer ANOVA analysis detected no significant change in the time course of systemic bFGF level after injection of AxCAMAssbFGF-treated cells into the rabbit ischemic model (Figure 7a). Anti-adenovirus antibody titer was significantly lower in animals with infected cell administration at all time points as compared with that in rabbits with intravenous injection of AxCAMAssbFGF (positive control) (P ⬍ 0.01) (Figure 7b). Fate of gene-transduced fibroblasts in vivo To evaluate the fate of the administered cells and their influence on host tissues, rabbits of the ischemic model were killed at various time-points after injection of AxCAMAssbFGF-treated fibroblasts. Immunostaining for bFGF showed that a large number of bFGF-positive cells were scattered in the left adductor muscle at 1, 4 and 7 days after injection of the infected cells, while bFGFpositive cells were few in control slides (Figure 8a, b and c). The bFGF-positive cells subsequently decreased, and the number of cells after day 14 was almost equal to that of control (Figure 8d and e). In internal organs, no significant change in the number of bFGF-positive cells was detected in the time course study (data not shown). Hematoxylin/eosin staining and Elastica van Gieson staining revealed neither fibrosis nor other changes in all tissues until 28 days after cell administration (data not shown).

Figure 5 Selective internal iliac angiograms of AxCALacZ-treated (a, angiographic score = 0.36) and AxCAMAssbFGF-treated (b, angiographic score = 0.71) rabbit at 28 days after injection of infected fibroblasts. Arrow indicates internal iliac artery. Development of collateral vessels was quantified by the angiographic score 28 days after injection (c). Values are shown as mean ± s.d. (*P ⬍ 0.01). Cont, control group; bFGF, bFGF group; NS, no significant difference.

Local bFGF accumulation in vivo Local accumulation of bFGF in tissues was analyzed by Western blot after concentration using heparinsepharose. In the left adductor muscle, bFGF accumulation was increased from 1 day after cell injection, and abundant bFGF was observed at day 4 and 7 (Figure 9a).

Discussion

Gene Therapy

The protein level decreased after that, though a slightly high amount of bFGF was detected until 28 days after cell administration as compared with that of control. In lung and liver tissues, the time course of bFGF accumulation showed no meaningful increase above the control level (Figure 9b and c).

In the present study, we showed an alternative strategy for gene transfer to limb tissue of chronic ischemia, in


841

Figure 6 (a) Capillary density of left semimembranous muscles was measured in tissue sections stained by indoxy-tetrazolium method. (b) Diameter of proximal left gluteal artery. (c) In vivo blood flow of left internal iliac artery. APV was recorded at rest, maximum APV was measured after papaverine injection, and blood flow was calculated. Blood flow values in the non-ischemic model were lower than those in the ischemic model, since the left femoral artery was preserved in the non-ischemic model. Values are shown as mean ± s.d. (*P ⬍ 0.01, **P ⬍ 0.05). Cont, control group; bFGF, bFGF group; NS, no significant difference.

which auto-fibroblasts, adenovirally transduced with modified bFGF gene, were infused into the ischemic limb through an intra-arterial catheter. This ex vivo method promoted collateral vessel development in a rabbit model of hind limb ischemia, without elevation of systemic bFGF level. Further, this protocol induced low-level antiadenovirus antibody, indicating minimal contamination of host tissues with adenovirus. The most important finding in the present study was that the fibroblasts significantly accumulated in muscle tissues of the left hind limb after trans-catheter administration into the left internal iliac artery. The significant cell accumulation was not specific to ischemic tissue, because the distribution of fibroblasts that were injected into the left ventricle was highly correlated with regional blood flow measured using microspheres. We, therefore, believe that the injected cells were trapped mechanically in small diameter arteries and capillaries of the muscle tissue. Indeed, immunohistological examination of the left adductor muscle showed that bFGF-positive cells were scattered between myocytes, where small vessels and capillaries were located. The entrapment of administered cells in small arteries and capillaries might be desirable for inducing collateral vessel development, since

Figure 7 (a) Time-course of systemic bFGF level measured by ELISA. (b) Time course of anti-adenovirus antibody titer quantified by neutralizing test. Titers are shown as dilution ratio, and titers less than 1:4 were assigned a value of 1. Values are shown as mean ± s.d. C, control sample; d, days after infection.

elongation of collateral arteries starts from the peripheral part, and enlargement occurs in small diameter arteries. Namely, the gene-transduced cells in the present study secreted bFGF at an appropriate site where development of collateral vessels occurred. Although there is a possibility that the trapped fibroblasts may reduce the peripheral blood flow, no significant decrease of calf blood pressure ratio was detected immediately after cell administration. Recently, Arras et al12 reported a similar procedure using microspheres to deliver angiogenic factor protein into the heart. In the same study, they injected microspheres adsorbed by bFGF protein into the coronary artery and observed in vivo release of bFGF within 1 week. Meanwhile, the secreted growth factor might continue to function in the ischemic tissue for a few weeks, since bFGF could bind to heparan sulfate proteoglycan of the extracellular matrix.13 In the present study, a slightly increased amount of bFGF was detected in the lysate of the left adductor muscle until 28 days after cell administration, while the number of bFGF-positive cells in the same muscle had returned to the control level at day 14, suggesting the existence of bFGF bound to the matrix after day 14. It is also an advantage of the ex vivo method that this protocol could be easily applied as therapy for ischemic heart disease, since trans-catheter administration is possible. In the direct application method for recombinant Gene Therapy


842

a

b

d

c

e

Figure 8 Microphotographs of left adductor muscle; control (a), 4 days (b), 7 days (c), 14 days (d) and 28 days (e) after injection of AxCAMAssbFGF-treated fibroblasts. Sections were immunostained with antibFGF antibody and counterstained with hematoxylin. bFGF-positive cells are stained brown. Bar, 50 ␮m.

Figure 9 Time-course of bFGF accumulation in left adductor muscle (a), lung (b) and liver (c). bFGF in each lysate was concentrated with heparinsepharose and analyzed by Western blot using anti-bFGF antibody. PC, positive control; C, control sample; d, days after infection.

growth factor protein, an intra-arterial catheter was the major delivery device.4,5,8 In contrast, it is complicated to apply naked DNA to the myocardium by intramuscular injection. Losordo et al14 injected naked DNA of VEGF into the ventricular wall via mini-thoracotomy. Although Gene Therapy

mini-thoracotomy needs only basic surgical techniques, the trans-catheter approach might be simpler and safer than injection into the heart via mini-thoracotomy. The ex vivo method has another distinct feature in using adenovirus vector. Since the host animal induces no humoral immune response to the gene-transduced host cells, it is possible to repeat the administration of gene-transduced cells several times. In contrast, infection of adenovirus vector induces a cytotoxic immune reaction by T cells, and the cytotoxicity decreases the lifetime of the infected cells, suggesting relatively short-term function of the transduced gene.15,16 Indeed, immunostaining in the present study showed that bFGF-positive cells in the left adductor muscle had diminished by 14 days after cell injection. Regarding the short-term function and the possibility of repeated administration, we believe that gene expression was highly ‘controllable’ with the ex vivo method. We now speculate that the number of injected fibroblasts is also a critical factor for regulating gene expression. Control of gene expression is indispensable for appropriate collateral vessel development and prevention of side-effects due to excessive growth factor. To enhance collateral vessel development in ischemic tissue, recombinant protein of bFGF has been administered in a variety of experimental systems, and significant effects were observed.5 Many recent studies, however, selected VEGF as an angiogenic growth factor to induce collateral vessels in ischemic tissue. Isner and colleagues4,7,8,17 presented several studies using VEGF in a rabbit model of hind limb ischemia and also in clinical trial, and showed favorable effects in improvement of the ischemic state. One reason for using VEGF is that VEGF receptors are upregulated on endothelial cells of newly formed blood vessels in vivo.18 In contrast, it is not yet clear how FGF receptors are expressed on endothelial cells in vivo.18 Another reason is that VEGF has a signal sequence for secretion, which bFGF lacks.9 In the present study, we used modified bFGF gene with the added signal sequence of IL-2 for protein secretion. The in vitro data of the present study showed that the added signal sequence was effective for the secretion of bFGF protein, indicating that AxCAMAssbFGF was available for angiogenic gene therapy. Furthermore, animal studies of retinal angiopathy showed that bFGF induced less neovascularization of the retina than VEGF.19 Worsening of diabetic retinopathy is thought to be a severe potential side-effect of collateral enhancement therapy, and its indication in diabetic patients has been restricted in clinical trials of VEGF. Thus, there is a possibility that bFGF could widen the clinical application of the therapy. A crucial issue of this protocol is its influences on other non-ischemic tissue and organs. Although systemic bFGF level did not increase in this study, the distribution study showed low level accumulation of labeled cells in other organs. These data indicate that some gene-transduced cells entered the venous system and systemic circulation, suggesting a possibility of unexpected side-effects in other organs. However, morphological analysis showed no abnormal findings in all organs and tissues up to 28 days after cell injection. One possible explanation for the negative findings is that the bFGF level expressed in other organs was too low to induce any histological changes. The present study showed no increase of bFGFpositive cells in all specimens except the left adductor


muscle and no change of bFGF expression in extracts of both liver and lung after cell administration. This might be because the number of accumulated cells was relatively low for the whole volume of each organ. Another possible explanation for no histological change is low responsiveness of non-ischemic tissues to bFGF stimulation. This concept was supported by the findings that AxCAMAssbFGF-treated cells induced no significant angiogenic response in non-ischemic normal hind limb, and suggested high clinical relevance, since it might preclude unexpected effects of bFGF in non-ischemic tissues. In contrast, Gowdak et al20 showed that adenovirusmediated VEGF gene-transfer stimulated angiogenesis in normoperfused skeletal muscle. In addition, we must be aware of the vasodilation effect caused by overexpressed bFGF in the analysis of angiographs and blood flow data.21 Indeed, arterial diameter measured at the caudal gluteal artery increased in the bFGF group, though there was a possibility that the increase of blood flow widened the diameter. In the present study, nitroglycerin was used for angiography, and papaverine for measuring maximal blood flow, to reduce the influence of vasodilation. In conclusion, we demonstrated an alternative strategy for gene transfer to chronic ischemia tissue, and examined it in a rabbit model of hind limb ischemia. The ex vivo method was applied in this strategy, and modified bFGF gene with the secretion signal sequence of IL-2 was transferred into host fibroblasts by adenovirus vector. The gene-transduced fibroblasts secreted bFGF in the ischemic limb tissue for less than 14 days, and induced a significant development of collateral vessels by 28 days after cell administration.

Materials and methods Adenovirus vectors Replication-deficient recombinant adenovirus vectors containing human bFGF gene with (AxCAMAssbFGF) or without the signal sequence (AxCAJSbFGF) were constructed according to the method described previously.22,23 The cDNA of the original human bFGF and modified human bFGF fused with the secretory signal sequence of IL-2 were obtained from plasmid pTB62724 and pTB107910 (donated by Dr K Igarashi, Biotechnology Research Laboratories, Takeda Chemical Industries, Osaka, Japan). The control vector for the animal experiment was AxCALacZ, which varies from the Escherichia coli LacZ cDNA in the same expression unit described above. In vitro study Cultured rabbit fibroblasts confluent in 100 mm dishes (passage 3) were infected with AxCAJSbFGF (native group) or AxCAMAssbFGF (secretory group) at 20 plaque-forming units (p.f.u.)/cell in 3 ml Dulbecco’s modified Eagle’s minimum essential medium (DMEM, Gibco BRL, Grand Island, NY, USA) with 2% FBS (DMEM-2%).25 After 1-h incubation, the cells were rinsed twice with PBS and cultured in 10 ml DMEM-2%. The culture medium was changed at 24-h intervals, and the medium and cells were harvested at 1, 4, 7, 10, 14, 21 and 28 days after infection for Western blot analysis, ELISA and 3H-thymidine incorporation assay. Another group of

cells not infected was used as control. A preliminary experiment of AxCALacZ transduction to fibroblasts at 20 p.f.u./cell revealed 100% efficacy by X-gal staining, and we excluded the contamination of the samples with residual virus by applying them to 293 cells and observing them for 14 days. Each culture medium (50 ␮l) and cell lysate (10 ␮g total protein) were subjected to Western blot analysis using mouse monoclonal antibody against bovine bFGF (1:500, Upstate Biotechnology, Lake Placid, NY, USA) according to the protocol described previously.26 Recombinant human bFGF (2 ng, Upstate Biotechnology) was used as a positive control. For quantitative analysis of bFGF in culture medium and cell lysate, ELISA was carried out, using bFGF ELISA kit (R&D Systems) according to the supplier’s instructions. Before the ELISA analysis, each cell lysate was diluted with PBS (1:100) to minimize the influence of detergent. The DNA synthesis activity of secreted bFGF was evaluated by incorporation of 3H-thymidine. Rabbit fibroblasts (passage 3) were plated at a density of 4 × 104 cells/well in a 24-well tray and cultured in DMEM with 10% FBS (DMEM-10%). After the cells reached confluence, they were made quiescent by incubation in 1 ml of medium containing 1% FBS for 4 days. Then, 100 ␮l of medium samples was added to the prepared fibroblasts for 20 h in culture. Subsequently, 1 ␮Ci 3H-thymidine (Amersham Pharmacia Biotech, Little Chalfont, UK) was added to each well, and after 2 h, incorporation into DNA of the cells was measured by a liquid scintillation counter, as described previously.26 These in vitro analyses were repeated at least twice.

843

Animal model of chronic hind limb ischemia (ischemic model) We used a rabbit model of hind limb ischemia for evaluation of in vivo angiogenic responses.4,7,27 All protocols were approved by the Guide for Animal Experimentation, The University of Tokyo. Male Japanese White rabbits weighing 3.0–3.5 kg (Saitama Rabbitry, Saitama, Japan) were anesthetized with an intramuscular injection of a mixture of ketamine (50 mg/kg) and xylazine (2.5 mg/kg), and the left femoral artery was completely excised from its proximal origin to the bifurcation formed by the saphenous and popliteal arteries as described previously.4,7 Consequently, blood flow to the ischemic hind limb was dependent on collateral vessels issuing from the left internal iliac artery. Before closure of the wound, a 10 × 10 mm section of skin was obtained for primary culture of fibroblasts (Figure 2a). Ex vivo gene transfer Fibroblasts were cultured from the resected skin tissue, and grown to confluence in 100-mm dishes in DMEM10%. It needed 10–19 days and three passages to obtain the desired number (more than 5 × 106) of fibroblasts. At 20 days after removal of the femoral artery, the fibroblasts were infected in vitro with AxCALacZ or AxCAMAssbFGF at 20 p.f.u./cell and incubated for 24 h. At 21 days after femoral artery excision, a 3-French end-hole infusion catheter (Rapid Transit, Cordis, Miami, FL, USA) was introduced into the left common carotid artery and positioned in the left internal iliac artery under fluoroscopic guidance. Then 5 × 106 fibroblasts, infected with AxCALacZ (control group, n = 12) or AxCAMAssbFGF (bFGF group, n = 11), were suspended in 1.5 ml DMEMGene Therapy


844

2% after double washing with PBS, and injected as a bolus through the infusion catheter (Figure 2a). Heparin and other anticoagulants were not used during all procedures.

Distribution of administered cells At 21 days after removal of the left femoral artery, 1 × 107 cultured fibroblasts were suspended in 0.75 ml DMEM-2%. 111In-oxine solution (250 ␮Ci, Amersham Pharmacia Biotech) was added to the suspension and incubated for 30 min at 37°C. After centrifugation, the supernatant was removed. The labeled fibroblasts were incubated in 1 ml DMEM-2% for an additional 30 min at 37°C, washed twice, and resuspended in 3 ml DMEM2%. Then, 1.5 ml of the suspension containing 5 × 106 labeled cells was injected through the left internal iliac artery, and 5 h later, the treated rabbits (n = 5) were killed. Above-knee muscles, below-knee muscles and skin were taken from the bilateral hind limbs, and internal organs were resected. Each tissue or organ was weighed (WWHOLE), minced and homogenized. A certain volume of sample was taken from each of the homogenized specimens, weighed (WSAMPLE), and counted with a gamma-counter (CIn-SAMPLE), while radioactivity of the remaining suspension containing 5 × 106 labeled cells was counted (CIn-CELLS). The distribution of administered cells (%) in each tissue was calculated as 100 × (CIn-SAMPLE/CInCELLS) × (WWHOLE/WSAMPLE). 111 In-labeled fibroblasts (5 × 106) were also injected into the left ventricle at 21 days after removal of the left femoral artery, and the cell distribution in the bilateral hind limb muscles was compared with regional blood flow to the same muscle tissues. First, 51Cr-labeled microspheres (3 × 106, 15 ␮m in diameter, NEN Life Science Products, Boston, MA, USA) were injected into the left ventricle through a 3-French end-hole catheter via the right common carotid artery. Another catheter was inserted into the descending aorta via the left common carotid artery, and a reference blood sample was withdrawn beginning 30 s before microsphere injection and then continued for 2 min at a rate of 2.0 ml/min. Immediately after the microsphere administration, 111In-labeled fibroblasts were injected into the left ventricle of the same rabbit. Then, the treated animals (n = 5) were killed, and aboveknee muscles and below-knee muscles were taken from the bilateral hind limbs. The samples were prepared in the same manner as above, and radioactivity of 111In and 51 Cr (CCr-SAMPLE) were counted for each sample. The cell distribution was calculated as described above, and regional blood flow (ml/min/whole tissue) was calculated as (withdrawal rate) × (CCr-SAMPLE/CCr-BLOOD) × (WWHOLE/WSAMPLE), where CCr-BLOOD was the count of reference blood sample. Non-ischemic hind limb model (non-ischemic model) To assess the angiogenic effect in non-ischemic tissues, skin was resected from normal rabbit and fibroblasts were cultured. Twenty-one days later, 5 × 106 fibroblasts infected with AxCALacZ (n = 5) or AxCAMAssbFGF (n = 5) were injected though the left internal iliac artery in the same manner as described above (Figure 2b). Calf blood pressure ratio Calf blood pressure was measured in both hind limbs using a Doppler probe as previously described at four Gene Therapy

time-points as follows: immediately after femoral artery excision, and immediately before, immediately after and 28 days after cell administration.4,7 The calf blood pressure ratio was defined as the ratio of left systolic pressure to right systolic pressure.

Angiographic score Selective internal iliac arteriography was performed 28 days after the injection of infected fibroblasts. A 3-French end-hole catheter was introduced into the left internal iliac artery through the right common carotid artery; the catheter tip was positioned at the level of the middle part of the first sacral vertebra. Contrast medium (5 ml, Iopamiron 370, Schering, Berlin, Germany) was injected at a rate of 1 ml/s after intra-arterial administration of nitroglycerin (0.25 mg, Kyowa, Tokyo, Japan), and angiographic score was calculated on the 4-s angiograph as described previously.7 The same film was also scanned with a transmissive scanner, and the diameter of the proximal part of the left caudal gluteal artery was measured with NIH image software (version 1.61) to evaluate dilatation of the distal artery. Capillary density At 28 days after the administration of gene-transduced cells, tissue specimens were obtained as transverse sections from the left semimembranous muscle of each animal and embedded in OCT compound (Miles, Naparville, NJ, USA) without fixation. Frozen sections (5 ␮m) were cut from two specimens and capillary endothelial cells were stained by the indoxy-tetrazolium method.4,7 A total of 20 different fields from the two muscle sections were randomly selected, and capillaries were counted to determine the capillary density. Resting and maximum blood flow Before selective internal iliac arteriography, a 0.014-in Doppler guide wire (Endo Sonics, Rancho Cordva, CA, USA) was introduced via the 3-French infusion catheter to the proximal segment of the left internal iliac artery. Average peak velocity (APV) was measured at rest, and maximum APV was quantified immediately after bolus intra-arterial injection of 2 mg papaverine (Dainippon Pharmaceutical, Osaka, Japan).7,8 Diameter (d) at the site of the Doppler sample volume was measured on the arteriograph, and blood flow was calculated as (␲d2/4) × 0.5 × APV. Fate of gene-transduced fibroblasts in vivo time-course study For the time-course study, rabbits of the ischemic model were killed at 1, 4, 7, 14, 21 and 28 days after the injection of AxCAMAssbFGF-treated cells (n ⭓ 5 at each time point), and various samples were taken, which were bilateral adductor muscle, lung, heart, liver, kidney, spleen, testis and serum. One group of rabbits (n = 6) was killed at 21 days after femoral artery excision as control. The bFGF concentration of serum was measured by ELISA for analysis of systemic bFGF level, and serum anti-adenovirus antibody level was also quantified by neutralization test (type 5, SRL, Tokyo, Japan). AxCAMAssbFGF (1 × 108 p.f.u.) was injected intravenously into normal rabbits (n = 5), and anti-adenovirus antibody was measured intermittently as positive control. To identify the gene-transduced cells in each tissue, fresh frozen sec-


tions (7 ␮m) were immunostained for bFGF. The monoclonal antibody against bovine bFGF (1:100, Upstate Biotechnology) was applied after hydrogen peroxide treatment and blocking by 1.5% goat serum. Subsequent incubation with biotinylated goat anti-mouse IgG (1:100, Vector Laboratories, Burlingame, CA, USA) and ABC Elite kit (Vector Laboratories) was performed. Control slides were stained with normal mouse IgG instead of the primary antibody. Each tissue was fixed in 4% paraformaldehyde and embedded in paraffin. Sections (4 ␮m) were cut and stained by hematoxylin/eosin or Elastica van Gieson method. To assess the bFGF level contained in tissue samples, left adductor muscles (1 g), lung (0.5 g) and liver (0.25 g) were lysed in 1 ml minimum essential medium (Gibco BRL), and bFGF in each lysate was concentrated with heparin-sepharose CL-6B (Amersham Pharmacia Biotech) and analyzed by Western blot as described previously.28

Statistical analysis Results were expressed as mean ± s.d. Statistical significance was evaluated using unpaired Student’s t test for comparisons between two means. Kruskal–Wallis followed by Scheffe’s F was used for analysis of cell distribution, and ANOVA for analysis of the time course of systemic bFGF level. Correlation between cell distribution and regional blood flow was examined by Pearson correlation coefficient. A value of P ⬍ 0.05 was interpreted to denote statistical significance.

Acknowledgements This study was supported by a Grant-in Aid for Scientific Research (C) from the Ministry of Education, Science, and Culture of Japan (10671103). The studies using 111In and 51Cr were carried out at Radioisotope Center, The University of Tokyo.

References 1 Sasayama S, Fujita M. Recent insight into coronary collateral circulation. Circulation 1992; 85: 1197–1204. 2 Carmeliet P. Mechanisms of angiogenesis and arteriogenesis. Nat Med 2000; 6: 389–395. 3 Hockel M et al. Therapeutic angiogenesis. Arch Surg 1993; 128: 423–429. 4 Takeshita S et al. Therapeutic angiogenesis. A single intraarterial bolus of vascular endothelial growth factor augments revascularization in a rabbit ischemic hind limb model. J Clin Invest 1994; 93: 662–670. 5 Unger EF et al. Basic fibroblast growth factor enhances myocardial collateral flow in a canine model. Am J Physiol 1994; 266: H1588–H1595. 6 Feldman LJ et al. Low-efficacy of percutaneous adenovirusmediated arterial gene transfer in the atherosclerotic rabbit. J Clin Invest 1995; 95: 2662–2671. 7 Tsurumi Y et al. Direct intramuscular gene transfer of naked DNA encoding vascular endothelial growth factor augments collateral development and tissue perfusion. Circulation 1996; 94: 3281–3290.

8 Bauters C et al. Physiological assessment of augmented vascularity induced by VEGF in ischemic rabbit hindlimb. Am J Physiol 1994; 267: H1263–H1271. 9 Abraham JA et al. Nucleotide sequence of a bovine clone encoding the angiogenic protein, basic fibroblast growth factor. Science 1986; 233: 545–548. 10 Sasada R, Seno M, Watanabe T, Igarashi K. Expression of modified bFGF cDNAs in mammalian cells. Ann NY Acad Sci 1991; 638: 149–160. 11 Kass-Eisler A et al. Quantitative determination of adenovirusmediated gene delivery to rat cardiac myocytes in vitro and in vivo. Proc Natl Acad Sci USA 1993; 90: 11498–11502. 12 Arras M et al. The delivery of angiogenic factors to the heart by microsphere therapy. Nat Biotechnol 1998; 16: 159–162. 13 Friesel RE, Maciag T. Molecular mechanisms of angiogenesis: fibroblast growth factor signal transduction. FASEB J 1995; 9: 919–925. 14 Losordo DW et al. Gene therapy for myocardial angiogenesis: initial clinical results with direct myocardial injection of phVFGF165 as sole therapy for myocardial ischemia. Circulation 1998; 98: 2800–2804. 15 Wilson JM. Adenoviruses as gene-delivery vehicles. New Engl J Med 1996; 334: 1185–1187. 16 Yang Y et al. Cellular immunity to viral antigens limits E1deleted adenoviruses for gene therapy. Proc Natl Acad Sci USA 1994; 91: 4407–4411. 17 Baumgartner I et al. Constitutive expression of phVEGF165 after intramuscular gene transfer promotes collateral vessel development in patients with critical limb ischemia. Circulation 1998; 97: 1114–1123. 18 Risau W. What, if anything, is an angiogenic factor? Cancer Metast Rev 1996; 15: 149–151. 19 Ozaki H et al. Basic fibroblast growth factor is neither necessary nor sufficient for the development of retinal neovascularization. Am J Pathol 1998; 153: 757–765. 20 Gowdak LHW et al. Adenovirus-mediated VEGF121 gene transfer stimulates angiogenesis in normoperfused skeletal muscle and preserves tissue perfusion after induction of ischemia. Circulation 2000; 102: 565–571. 21 Wu HM, Yuan Y, McCarthy M, Granger HJ. Acidic and basic FGFs dilate arterioles of skeletal muscle through a NO-dependent mechanism. Am J Physiol 1996; 271: H1087–H1094. 22 Miyake S et al. Effective generation of recombinant adenoviruses using adenovirus DNA–terminal protein complex and a cosmid bearing the full-length virus genome. Proc Natl Acad Sci USA 1996; 93: 1320–1324. 23 Akimoto M et al. Adenovirally expressed basic fibroblast growth factor rescues photoreceptor cells in RCS rats. Invest Ophthalmol Vis Sci 1999; 40: 273–279. 24 Kurokawa T, Sasada R, Iwane M, Igarashi K. Cloning and expression of cDNA encoding human basic fibroblast growth factor. FEBS 1987; 213: 189–194. 25 Takahashi JC et al. Adenovirus-mediated gene transfer of basic fibroblast growth factor induces in vitro angiogenesis. Atherosclerosis 1997; 132: 199–205. 26 Koyama H, Reidy MA. Reinjury of arterial lesions induces intimal smooth muscle cell replication that is not controlled by fibroblast growth factor 2. Circ Res 1997; 80: 408–417. 27 Pu LQ et al. A persistent hind limb ischemia model in the rabbit. J Invest Surg 1994; 7: 49–60. 28 Kardami E, Fandrich RR. Basic fibroblast growth factor in atria and ventricles of the vertebrate heart. J Cell Biol 1989; 109: 1865–1875.

845

Gene Therapy