Construction of a novel expression cassette for increasing transgene expression in vivo in endothelial cells of large blood vessels.

The success of gene therapy hinges on achievement of adequate transgene expression. To ensure high transgene expression, many gene-therapy vectors include highly active virus-derived transcriptional elements. Other vectors include tissue-specific eukaryotic transcriptional elements, intended to limit transgene expression to specific cell types, avoid toxicity and prevent immune responses. Unfortunately, tissue specificity is often accompanied by lower transgene expression. Here, we use eukaryotic (murine) transcriptional elements and a virus-derived posttranscriptional element to build cassettes designed to express a potentially therapeutic gene (interleukin (IL)-10) in large-vessel endothelial cells (ECs) at levels as high as obtained with the cytomegalovirus (CMV) immediate early promoter, while retaining EC specificity. The cassettes were tested by incorporation into helper-dependent adenoviral vectors, and transduction into bovine aortic EC in vitro and rabbit carotid EC in vivo. The murine endothelin-1 promoter showed EC specificity, but expressed only 3% as much IL-10 mRNA as CMV. Inclusion of precisely four copies of an EC-specific enhancer and a posttranscriptional regulatory element increased IL-10 expression to a level at or above the CMV promoter in vivo, while retaining--and possibly enhancing--EC specificity, as measured in vitro. The cassette reported here will likely be useful for maximizing transgene expression in large-vessel EC, while minimizing systemic effects.

Local adenoviral-mediated expression of recombinant hirudin reduces neointima formation after arterial injury.

Catalytically active thrombin, acting locally, is thought to mediate neointima formation after arterial injury. We constructed an adenovirus vector, AdHV-1.2, containing a complementary DNA for the thrombin inhibitor hirudin. AdHV-1.2 directed the synthesis and secretion of biologically active hirudin from vascular cells in vitro. In vivo gene transfer of hirudin into smooth muscle cells of injured rat carotid arteries resulted in peak secretion of at least 34+/-23 pg hirudin per vessel per 24 hours, and resulted in a significant (P<0.05) 35% reduction in neointima formation. Systemic partial thromboplastin times were not affected by local hirudin expression. These results support the hypothesis that local thrombin activity contributes to neointima formation after arterial injury and suggest that local delivery of a highly specific antithrombin may constitute an effective intervention for arterial proliferative disease.

Fas ligand expression in islets of Langerhans does not confer immune privilege and instead targets them for rapid destruction.

Fas ligand is believed to mediate immune privilege in a variety of tissues, including the eye, testis, and a subset of tumors. We tested whether expression of Fas ligand on pancreatic islets either following adenoviral or germline gene transfer could confer immune privilege after transplantation. Islets were infected with an adenoviral vector containing the murine Fas ligand cDNA (AdFasL), and were transplanted into allogenic diabetic hosts. Paradoxically, AdFasL-infected islets underwent accelerated neutrophilic rejection. The rejection was T cell and B cell independent and required Fas protein expression by host cells, but not on islets. Similarly, transgenic mice expressing Fas ligand in pancreatic beta cells developed massive neutrophilic infiltrates and diabetes at a young age. Thus, Fas ligand expression on pancreatic islets results in neutrophilic infiltration and islet destruction. These results have important implications for the development of Fas ligand-based immunotherapies.

Established immunity precludes adenovirus-mediated gene transfer in rat carotid arteries. Potential for immunosuppression and vector engineering to overcome barriers of immunity.

Preclinical arterial gene transfer studies with adenoviral vectors are typically performed in laboratory animals that lack immunity to adenovirus. However, human patients are likely to have prior exposures to adenovirus that might affect: (a) the success of arterial gene transfer; (b) the duration of recombinant gene expression; and (c) the likelihood of a destructive immune response to transduced cells. We confirmed a high prevalence (57%) in adult humans of neutralizing antibodies to adenovirus type 5. We then used a rat model to establish a central role for the immune system in determining the success as well as the duration of recombinant gene expression after adenovirus-mediated gene transfer into isolated arterial segments. Vector-mediated recombinant gene expression, which was successful in naive rats and prolonged by immunosuppression, was unsuccessful in the presence of established immunity to adenovirus. 4 d of immunosuppressive therapy permitted arterial gene transfer and expression in immune rats, but at decreased levels. Ultraviolet-irradiated adenoviral vectors, which mimic advanced-generation vectors (reduced viral gene expression and relatively preserved capsid function), were less immunogenic than were nonirradiated vectors. A primary exposure to ultraviolet-irradiated (but not nonirradiated) vectors permitted expression of a recombinant gene after redelivery of the same vector. In conclusion, arterial gene transfer with current type 5 adenoviral vectors is unlikely to result in significant levels of gene expression in the majority of humans. Both immunosuppression and further engineering of the vector genome to decrease expression of viral genes show promise in circumventing barriers to adenovirus-mediated arterial gene transfer.

Adenovirus-mediated gene transfer into normal rabbit arteries results in prolonged vascular cell activation, inflammation, and neointimal hyperplasia.

Adenovirus vectors are capable of high efficiency in vivo arterial gene transfer, and are currently in use as therapeutic agents in animal models of vascular disease. However, despite substantial data on the ability of viruses to cause vascular inflammation and proliferation, and the presence in current adenovirus vectors of viral open reading frames that are translated in vivo, no study has examined the effect of adenovirus vectors alone on the arterial phenotype. In a rabbit model of gene transfer into a normal artery, we examined potential vascular cell activation, inflammation, and neointimal proliferation resulting from exposure to replication-defective adenovirus. Exposure of normal arteries to adenovirus vectors resulted in: (a) pronounced infiltration of T cells throughout the artery wall; (b) upregulation of intercellular adhesion molecule-1 and vascular cell adhesion molecule-1 in arterial smooth muscle cells; (c) neointimal hyperplasia. These findings were present both 10 and 30 d after gene transfer, with no evidence of a decline in severity over time. Adenovirus vectors have pleiotropic effects on the arterial wall and cause significant pathology. Interpretation of experimental protocols that use adenovirus vectors to address either biological or therapeutic issues should take these observations into account. These observations should also prompt the design of more inert gene transfer vectors.

A mouse model of arterial gene transfer: antigen-specific immunity is a minor determinant of the early loss of adenovirus-mediated transgene expression.

We developed a murine model of arterial gene transfer and used it to test the role of antigen-specific immunity in the loss of adenovirus-mediated transgene expression. Adenoviral vectors encoding either beta-galactosidase (beta-gal) or green fluorescent protein were infused to the lumen of normal common carotids of CD-1 and C57BL/6 mice and atherosclerotic carotids of Apoe(-/-) mice. At 3 days after gene transfer, significant reporter gene expression was detected in all strains. Transgene expression was transient, with expression undetectable at 14 days. Next, a beta-gal-expressing vector was infused into carotids of ROSA26 mice (transgenic for, and therefore tolerant of, beta-gal) and RAG-2(-/-) mice (deficient in recombinase-activating gene [RAG]-2 and therefore lacking in antigen-specific immunity). beta-Gal expression was again high at 3 days but declined substantially (>90%) by 14 days. In vivo labeling with bromodeoxyuridine revealed that carotid endothelial proliferation was increased dramatically by the gene-transfer procedure alone, likely leading to the loss of episomal adenoviral DNA. Gene transfer to normal and atherosclerotic mouse carotids can be accomplished; however, elimination of antigen-specific immune responses does not prevent the early loss of adenovirus-mediated transgene expression. Efforts to prolong adenovirus-mediated transgene expression in the artery wall must be redirected. These efforts will likely include strategies to avoid the consequences of increased cell turnover. Nevertheless, despite the brevity of expression, this mouse model of gene transfer to normal and severely atherosclerotic arteries will likely be useful for investigating the genetic basis of vascular disease and for developing gene therapies.

Cardiovascular overexpression of transforming growth factor-beta(1) causes abnormal yolk sac vasculogenesis and early embryonic death.

Transforming growth factor-beta(1) (TGF-beta(1)) is expressed in the adult and embryonic vasculature; however, the biological consequences of increased vascular TGF-beta(1) expression remain controversial. To establish an experimental setting for investigating the role of increased TGF-beta(1) in vascular development and disease, we generated transgenic mice in which a cDNA encoding a constitutively active form of TGF-beta(1) is expressed from the SM22alpha promoter. This promoter fragment directs transgene expression to smooth muscle cells of large arteries in late-term embryos and postnatal mice. We confirmed the anticipated pattern of SM22alpha-directed transgene expression (heart, somites, and vasculature of the embryo and yolk sac) in embryos carrying an SM22alpha-beta-galactosidase transgene. SM22alpha- beta-galactosidase transgenic mice were born at the expected frequency (13%); however, nearly all SM22alpha-TGF-beta(1) transgenic mice died before E11.5. SM22alpha-TGF-beta(1) transgenic embryos identified at E8.5 to E10.5 had growth retardation and both gross and microscopic abnormalities of the yolk sac vasculature. Overexpression of TGF-beta(1) from the SM22alpha promoter is lethal at E8.5 to E10.5, most likely because of yolk sac insufficiency. Investigation of the consequences of increased vascular TGF-beta(1) expression in adults may require a conditional transgenic approach. Moreover, because the SM22alpha promoter drives transgene expression in the yolk sac vasculature at a time when embryonic survival is dependent on yolk sac function, use of the SM22alpha promoter to drive expression of "vasculoactive" transgenes may be particularly likely to cause embryonic death.

Plasminogen activator inhibitor type 1 increases neointima formation in balloon-injured rat carotid arteries.

BACKGROUND: Elevated plasma levels of plasminogen activator inhibitor type 1 (PAI-1) are associated with myocardial infarction, atherosclerosis, and restenosis. PAI-1 is increased in atherosclerotic arteries and failed vein grafts. No experimental data, however, support a causal relationship between elevated PAI-1 expression and vascular lesions. Paradoxically, data generated in PAI-1 knockout mice suggest that PAI-1 might decrease lesion formation after arterial injury and that PAI-1 gene transfer might prevent restenosis. METHODS AND RESULTS: Using the rat carotid balloon injury model and a PAI-1-expressing adenoviral vector, we tested whether elevated arterial PAI-1 expression would alter neointima formation. Compared with control-transduced arteries, neointima formation in PAI-1-transduced arteries was initially retarded. By 14 days, however, the intimas of PAI-1-transduced arteries were significantly larger than intimas of control-transduced arteries (1.6+/-0.1x10(5) versus 1.2+/-0.1x10(5) micrometer(2), n=18 to 19, P<0.03). PAI-1 expression in individual arteries correlated with increased cell proliferation at 4 and 8 days after injury (R=0.6, P<0.02 and P<0.006). PAI-1 expression also correlated with fibrin(ogen) accumulation (R=0.77, P<0.001), and fibrin(ogen) accumulation correlated strongly with proliferation (R=0.86, P<0.00001). CONCLUSIONS: Increased expression of PAI-1 in the artery wall promotes neointima growth after balloon injury. Therefore, despite encouraging data generated in other animal models, PAI-1 is not a promising agent for gene therapy to prevent restenosis. Moreover, our data associate elevated PAI-1 expression with fibrin(ogen) accumulation and increased cell proliferation. These data suggest a mechanism to explain the association between elevated PAI-1 expression and the progression of arterial disease.

Inclusion of the E3 region in an adenoviral vector decreases inflammation and neointima formation after arterial gene transfer.

Adenoviral vectors are promising agents for vascular gene transfer. Their use, however, is limited by inflammatory host responses, neointima formation, and brevity of transgene expression. Inclusion of the immunomodulatory adenoviral E3 genes in a vector might prevent inflammation and neointima formation and prolong transgene expression. We compared 2 adenoviral vectors in a model of in vivo gene transfer to rabbit arteries. Both vectors expressed a luciferase reporter gene. One vector (AdE3Luc) contained the adenovirus early 3 (E3) region and the other (AdRSVLuc) lacked E3. Expression of E3 genes by AdE3Luc was confirmed in vitro and in vivo. Arteries transduced with AdE3Luc had substantially and significantly less inflammation (fewer T cells and lower levels of vascular cell adhesion molecule-1 and intercellular adhesion molecule 1 expression) and decreased neointima formation 14 days after gene transfer. Luciferase expression from the 2 vectors was equivalent, however, at both 3 and 14 days after gene transfer. Expression of E3 had no systemic immunosuppressive effects, as measured by peripheral blood counts and by assays for serum antibodies to adenovirus. We conclude that expression of E3 significantly decreases adenovirus-induced arterial wall inflammation and neointima formation. Because inflammation and neointima formation are major barriers to the clinical application of adenoviral vectors, use of E3-containing vectors improves the promise of adenovirus-mediated arterial gene transfer.

Control of clot lysis by gene transfer.

Intravascular clot formation is a local process that can result in serious clinical consequences, including limb loss and death. Gene transfer and expression of recombinant plasminogen activators in the endothelial cells of the vessel wall offer an attractive approach to the enhancement of local fibrinolytic activity. In vitro studies have demonstrated that endothelial cell fibrinolytic activity can be increased by gene transfer of either secreted or cell surface-anchored plasminogen activators. Future work will define the ability of gene transfer to facilitate clot lysis in vivo.

Gene therapy for arterial thrombosis.

Conventional antithrombotic treatments with antiplatelet, anticoagulant or fibrinolytic drugs are not uniformly successful and are associated with hemorrhagic side effects. Thus, new approaches to the prevention and treatment of arterial thrombosis are desirable. The gene transfer approach is particularly attractive because of its unique ability to express an antithrombotic gene at selected sites of the vessel wall (where thrombosis is threatened) while avoiding systemic anticoagulation. Clinical conditions potentially amenable to antithrombotic gene therapy include coronary artery bypass grafting, percutaneous transluminal coronary angioplasty, peripheral artery angioplasty or thrombectomy, intravascular stenting, and vascular graft prostheses. Gene therapy may prove effective in preventing subacute thrombosis in these settings and, eventually, may play an adjuvant role to systemic thrombolysis in the treatment of acute arterial occlusion. The introduction of an antithrombotic gene into the arterial wall can be achieved either by direct in vivo gene transfer (e.g., by luminal administration of a viral vector) or by in vitro genetic manipulation of cells before their seeding onto vascular grafts, stents, or denuded arteries. The direct gene transfer approach has been used to deliver antithrombotic genes to animal arteries in vivo. Antithrombotic genes used to date include those encoding enzymes of the prostacyclin synthetic pathway, nitric oxide synthase, the thrombin inhibitor hirudin, and thrombomodulin. The in vitro gene transfer approach has been used to enhance the fibrinolytic activity of vascular grafts by overexpressing plasminogen activators. If the initial successes of gene therapy for thrombotic disease in animal models are confirmed by longer-term experiments, and if new vectors are developed which permit prolonged transgene expression without inflammation, human studies can be initiated.

Cationic liposomes enhance adenovirus entry via a pathway independent of the fiber receptor and alpha(v)-integrins.

The ability of adenoviral vectors to mediate efficient gene delivery both in vitro and in vivo is limited by the availability of specific cell surface receptors and alpha(v)-containing integrins. We tested whether this limitation could be overcome by enhancing viral entry with cationic liposomes. In cultured vascular smooth muscle cells, delivery of adenoviral vectors in the presence of cationic liposomes increased vector-encoded transgene expression up to 20-fold. The increase in transgene expression was associated with the formation of adenovirus-lipid aggregates and an increase in the amount of vector DNA in the cells, suggesting that enhanced viral entry was responsible for the increase in gene expression. Treatment of the cells with an RGD-containing peptide or adenovirus type 5 fiber protein did not diminish liposome enhancement of transgene expression, indicating that liposomes increase viral entry via a pathway independent of the fiber receptor and of alpha(v) integrin-assisted endocytosis. Liposomes also significantly enhanced transgene expression from adenoviral vectors delivered to cells deficient in alpha(v)-containing integrins. The magnitude of liposome enhancement of transgene expression in cultured smooth muscle cells was greatest during brief periods of virus-cell contact and at low concentrations of virus. Despite these promising in vitro results, addition of liposomes did not improve in vivo adenoviral gene delivery into injured rat carotid arteries. Liposomes can improve adenoviral gene delivery in vitro; however, application of this observation to accomplish improved in vivo gene delivery remains a challenge.

Optimizing vascular gene transfer of plasminogen activator inhibitor 1.

The vessel wall fibrinolytic system plays an important role in maintaining the arterial phenotype and in regulating the arterial response to injury. Plasminogen activator inhibitor type 1 (PAI-1) regulates tissue fibrinolysis and is expressed in arterial tissue; however, its biological role remains uncertain. To help elucidate the role of PAI-1 in the artery wall, and to begin to clarify whether manipulation of vascular PAI-1 expression might be a target for gene therapy, we used adenoviral vectors to increase expression of rat PAI-1 in rat carotid arteries. Infusion of an adenoviral vector in which PAI-1 expression was driven by a promoter derived from the Rous sarcoma virus (RSV) did not increase PAI-1 expression above endogenous levels. To improve PAI-1 expression, we modified the vector by (1) truncating the 3' untranslated region of PAI-1 to increase the mRNA half-life, (2) substituting the SRalpha or the cytomegalovirus (CMV) promoter for the RSV promoter, (3) including an intron in the expression cassette, and (4) altering the direction of transcription of the transgene cassette. The optimal expression vector, revealed by in vitro studies, contained the CMV promoter, an intron, and a truncated PAI-1 mRNA. This vector increased PAI-1 expression by 30-fold over control levels in vitro and by 1.6 to 2-fold over endogenous levels in vivo. This vector will be useful for elucidating the role of PAI-1 in arterial pathobiology. Because genes that are important in maintaining the vascular phenotype are likely to be expressed in the vasculature, the technical issues of how to increase in vivo expression of endogenous genes are highly relevant to the development of genetic therapies for vascular disease.

Adenoviral gene transfer in arteries of hypercholesterolemic nonhuman primates.

Arterial gene transfer with adenoviral vectors is a promising approach for the treatment and prevention of vascular disorders. However, in small animals such as rats and rabbits adenoviral vectors can have deleterious effects on the artery wall. The effects of adenovirus in primate arteries have not been studied. AdRSVn-LacZ, a replication-defective adenoviral vector, was delivered to the left brachial arteries of six hypercholesterolemic cynomolgus monkeys; right brachial arteries received vehicle only. Serum was collected before gene transfer and at vessel harvest 9 or 10 days later. Recombinant gene expression was present in occasional endothelial cells of transduced arteries, and all animals generated neutralizing antibodies. In transduced arteries, immunostaining revealed a fourfold increase in intimal and medial macrophage accumulation (p < 0.05); intimal cellularity was also significantly increased (twofold; p < 0.05). T cell density and total cellular proliferation (determined by bromodeoxyuridine labeling) were unaffected. In hypercholesterolemic nonhuman primates, adenoviral vectors increase vessel wall inflammation and promote the progression of early atherosclerotic lesions. The long-term consequences of these observations remain unclear; however, a better understanding of host responses to specific vector systems appears necessary for the development of safe and effective approaches to human vascular gene therapy.

Adenoviral vector-mediated gene transfer into sheep arteries using a double-balloon catheter.

The potential for catheter-based in vivo delivery of genetic material to the arterial wall is incompletely explored. We evaluated the level of recombinant protein production as well as the anatomic distribution and duration of gene expression following adenoviral vector-mediated gene transfer into sheep arteries via a double balloon catheter. Catheters were positioned in the carotid or femoral arteries of 20 sheep via a combined percutaneous and surgical approach, and virions infused over a 30-min period. Three days later, recombinant gene expression was identified in approximately 30% (range 0-80%) of the luminal endothelial cells within the targeted area of the artery. Persistent recombinant protein expression was identified histochemically for up to 4 weeks, although the number of positive cells decreased steadily. High levels of both beta-galactosidase (beta-Gal) activity and protein (mean 20 mU and 44 ng per vessel) were measured in vessel extracts 3 days after gene transfer, again decreasing significantly over a 4-week period. Transgene expression was limited almost entirely to the intima and adventitia; adventitial gene transfer occurred virtually exclusively along the vasa vasorum. In comparison to previous studies of catheter-based gene transfer, adenoviral vectors delivered by double balloon catheter resulted in a particularly high efficiency of endothelial cell gene transfer. The efficiency and amount of recombinant gene expression achieved in this study suggest that catheter-based gene delivery may eventually be applicable to the treatment of focal human arterial disease.

A non-cleavable mutant of Fas ligand does not prevent neutrophilic destruction of islet transplants.

BACKGROUND: Fas ligand (FasL) mediates apoptosis of susceptible Fas-expressing lymphocytes, and may contribute to the maintenance of peripheral tolerance. In transplantation models, however, artificial expression of FasL on cellular as well as islet transplants results in accelerated rejection by neutrophils. The mechanism of the neutrophilic response to FasL expression is unknown. FasL, like other members of the tumor necrosis factor family, is cleaved to a soluble form by metalloproteases. We tested the hypothesis that soluble FasL (sFasL) was responsible for neutrophil migration by creating a non-cleavable mutant of FasL. METHODS: Three mutants of FasL with serial deletions in the putative proteolytic cleavage site of human FasL were made using inverse polymerase chain reaction. The relative fractions of sFasL and membrane-bound FasL were assessed by Western blot and immunoprecipitation, as well as by cytotoxicity assay using Fas-expressing target cells. The fully non-cleavable mutant was transduced into murine islets as well as myoblasts and tumor cell lines, and tested in a murine transplantation model. RESULTS: Serial deletions in the putative metalloprotease site of FasL resulted in a fully non-cleavable mutant of FasL (ncFasL). Expression of ncFasL in tumor lines induced higher levels of apoptosis in Fas bearing targets than wild-type FasL. Transplantation of ncFasL-expressing islets under the kidney capsule of allogenic mice resulted in accelerated rejection identical to that seen with wild-type Fas ligand-expressing islets. Myoblasts and tumor cell lines expressing ncFasL also induced neutrophil infiltration. CONCLUSIONS: Membrane-bound Fas ligand is fully capable of inducing a neutrophilic response to transplants, suggesting an activation by Fas ligand of neutrophil chemotactic factors.

Retroviral vector-mediated gene transfer into endothelial cells.

Over the last few years several groups have used retroviral vectors to achieve stable gene transfer into endothelial cells. In vitro experiments include transduction of cultured cells with genes of potential therapeutic interest, such as growth hormone and tissue plasminogen activator (t-PA). Animal studies have demonstrated the feasibility of in vivo recombinant gene expression from transduced endothelial cells, but have thus far been accomplished only with the lacZ marker gene. All studies to date have been oriented primarily toward the use of transduced endothelial cells to provide gene therapy. Numerous issues remain to be addressed with experimental data prior to the initiation of a clinical protocol using transduced endothelial cells. These issues include the introduction of larger numbers of transduced cells into the vasculature and the achievement of appropriate regulation of transgene expression. The use of retroviral vectors to study basic endothelial cell biology has been relatively ignored. The tool of retroviral vector-mediated gene transfer is available for use in answering both therapeutic and pathophysiological questions in endothelial cell biology.

Gene therapy for restenosis: are we ready?

The application of gene therapy techniques to the clinical problem of coronary restenosis has generated tremendous attention and enthusiasm. Use of gene transfer technology to prevent a common intractable illness would represent a watershed event for human gene therapy. However, the time is not yet right to initiate gene therapy trials for restenosis. The biology of restenosis is incompletely understood, catheter-based gene delivery is poorly adapted to the coronary circulation, and current gene transfer vectors are ill-suited for safe and effective gene delivery to the coronary artery wall. Basic research designed to overcome these obstacles is currently more appropriate than the initiation of clinical trials.

Characterization of recombinant plasminogen activator production by primate endothelial cells transduced with retroviral vectors.

Retroviral vector-mediated expression of plasminogen activators (PAs) from endothelial cells (ECs) has been proposed as a potential therapeutic approach for intravascular thrombosis. To define the potential for gene transfer to increase fibrinolytic activity in a primate system, baboon ECs were transduced with retroviral vectors expressing wild-type and glycosylphosphatidylinositol-anchored urokinase, as well as wild-type and serpin-resistant tissue PA (t-PA). Expression of either t-PA or urokinase was increased by one log over baseline levels. There was no specific effect of either t-PA or urokinase overexpression on endogenous t-PA, urokinase, or PA inhibitor 1 (PAI-1) expression. Recombinant urokinase could be anchored to the cell surface at a level eight-fold above that of receptor-bound urokinase. The majority of secreted urokinase accumulated in conditioned medium as a free proenzyme, whereas both wild-type and serpin-resistant t-PA accumulated almost exclusively in complexes with PAI-1. In most but not all of the assays, the urokinase vectors conferred PA activity above that of the t-PA vectors. These data show that PA synthesis and activity are specifically increased subsequent to retroviral vector-mediated gene transfer in primate ECs. However, definition of an optimal PA vector will require in vivo experimentation.