Since the first clinical trial of gene therapy for the treatment of ischemic limbs (which used a plasmid vector expressing the VEGF gene) by Jeffrey Isner in 1996 
, more than 150 gene therapy trials have been initiated, and about half of them are still opened (http://www.wiley.co.uk/genetherapy/clinical/
). The VEGF gene has been widely used in preclinical and clinical assays to treat ischemic diseases because it is a pleiotropic factor involved in many cellular activities, most of which are closely related to angiogenesis. In most of the animal studies, vectors expressing VEGF have been injected directly into the ischemic tissue, and this was found to improve blood flux and transiently augment vessel density 
. However, we and other groups have demonstrated that long periods of high expression of this gene in ischemic tissue can lead to deleterious effects, such as a decrease in capillary density, the loss of muscle mass and force and the augmentation of limb necrosis 
. It is likely that the high concentration of VEGF provided by the ischemic tissues together with the transfected cells promotes fast endothelial cell proliferation and vessel formation, which are not followed by adequate vessel maturation. These premature vessels in muscles can be easily disassembled by muscular movements, potentially leading to edema and cell death. Using VEGF together with either the stem cell mobilizing factor G-CSF (granulocyte colony stimulating factor) or the arteriogenic and vasculogenic factor GM-CSF (granulocyte macrophage-colony stimulating factor) to treat mouse ischemic limbs has resulted in much better outcomes than treatment with VEGF alone 
. Our interpretation of these results is that, to make a stable and functional vessel, it is necessary to recruit more growth factors and cells in an adequate time frame and at adequate concentrations, as happens physiologically.
In angiogenic gene therapy assays, plasmid and adenoviral vectors are primarily used because they provide efficient gene transfer to muscles in vivo and express transgenes transiently. Most (if not all) of these vectors are designed to express transgenes highly and continuously using strong constitutive promoters, with the expectation that the secreted angiogenic factors will be spread around the whole ischemic area. However, most angiogenic factors like VEGF contain a heparin-binding domain in their structure 
, which causes these factors to be retained around the production area. Consequently, an area with cells transfected with VEGF produces this factor continuously irrespective of the local production due to ischemia, making the local concentration higher than necessary after a certain period of time. Such a condition usually leads to the formation of unstable, immature and hypofunctional vessels 
. The best way of expressing VEGF for therapy is ideally by using vectors that respond to the requirements of the local tissue, i.e.
, vectors that are regulated by the local degree of ischemia, as normal cells are.
Physiological VEGF gene expression is modulated by the local oxygen concentration. In ischemic tissue, the oxygen supply is limited, and oxygen distribution occurs mostly by passive diffusion from arteries to tissues 
. This is one of the reasons that limbs distant from the obstructed artery are more affected than proximal limbs. Therefore, it is expected that the degree of ischemia, or the degree of oxygenation, will be variable in different parts of ischemic tissue. In this manner, VEGF production should correlate with either the degree of ischemia or the oxygen concentration. In humans, to overcome the ischemic condition, the ischemic tissues naturally express angiogenic factors based on the local oxygen concentration, such as VEGF, which is monitored by HIF-1 
. HIF-1α translocates to the nucleus during hypoxia, where it associates with HIF-1β to make a dimer, which in turn activates genes containing HRE (hypoxia responsive element) sequences. Genes involved in survival under hypoxic conditions, such as those that regulate angiogenesis, vessel dilation, erythropoiesis and glycolysis, are regulated by the binding of HIF to the HRE 
To make a VEGF-expressing vector responsive to hypoxia, we used nine repeats of the HRE sequence, which is responsive to HIF. A vector with nine repeats was used because it functioned better than vectors with fewer repeats 
. In addition, with the goal of making an integrative vector capable of providing long-term VEGF expression, the phiC31 integrase system was used. This system allows integration of the vector in one direction, i.e
., once the vector is integrated into the host genome it cannot be removed enzymatically 
. Using this system, our expectation was that one treatment of an ischemic limb with the vector (pVHAVI) would be sufficient for a long period, and VEGF gene expression would be regulated by the vector itself.
To demonstrate that the pVHAVI system was functioning correctly, these vectors were tested in vitro using HEK293T cells and myoblast cell line C2C12 and in vivo using a mouse ischemic limb model. In our in vitro study, HEK293T cells were transfected with several vectors (), and VEGF gene expression was followed for more than 90 days. The pVHAVI system was the only one that responded to hypoxia, which was induced by CoCl2 to stabilize HIF1α from degradation. It is important to note that hypoxia was induced at three different times and the cells modified with pVHAVI responded precisely to the hypoxic signal each time. A very similar result was also seen in C2C12 cell line. These results demonstrate clearly that VEGF expression can be activated at any moment by hypoxia, as we predicted.
Correct functioning of the pVHAVI system, as indicated by the induction of VEGF gene expression by hypoxia, was observed for a month. In this study, we opted to use the limb ischemia model rather than administration of CoCl2
because the first model is much more similar to human ischemic disease, and it is a well-established animal model 
. Additionally, we have not found a method for using CoCl2
to induce ischemia in vivo or any other method that can induce limb ischemia repeatedly without significantly affecting the animal's physiology. As a result, we could not induce hypoxia repeatedly in animals to evaluate the functioning of the pVHAVI system over a long period as we could for the in vitro model. Consequently, we chose to evaluate the system indirectly after gene therapy by examining therapeutic parameters such as alterations in muscle mass, force and histology and by visually assessing muscle necrosis. For in vivo gene transference, electroporation was used because in our previous studies, we demonstrated that this method is reproducible and results in high levels of transfection 
The first important step in our in vivo study was to demonstrate that the pVHAVI and pVAVI systems were integrated into muscle cells by phiC31 integrase after one month, and this was proven by PCR (). It is also very important to note that both systems produced VEGF at a similar level (), but the physiological response was quite different between them, as follows: 1) most of the animals treated with pVHAVI showed no necrosis, whereas half of those treated with pVAVI had some degree of necrosis (); ) muscle weight and force were higher in animals treated with the pVHAVI system () and ) the degree of angiogenesis and fibrosis was better with the pVHAVI system than the pVAVI system ().
Even though the VEGF concentration in muscles was similar in both systems, it is important to note that this concentration was determined from the whole muscle only once, at the 4th week, due to limitations of the method. Therefore, whether any variation in VEGF gene expression occurred in different parts of the muscle after transfection with either vector system during the 4 weeks is unknown. However, it is reasonable to assume that the pVHAVI system, which mainly expresses VEGF during hypoxia, acted differently than the pVAVI system, which expresses VEGF constitutively. We have no direct evidence to support this assumption, but the in vitro data and the improvement of animals treated with the pVHAVI system allow us to make this interpretation.
In this study, the viral CMV promoter was used to construct pVHAVI, which promoted hVEGF expression for three months in vitro and one month in vivo, at least. However, it is a well-known phenomenon that viral promoters, used to express mammalian genes, are frequently silenced during long-term studies. The use of muscle specific promoters like MCK (muscle creatine kinase) in construction of pVHAVI can bring better benefit to PAD patients in long-term treatment.
In conclusion, we demonstrated that a plasmid vector with an HRE sequence incorporated into it provides hypoxia-inducible VEGF expression. This vector, produced with the phiC31 integrative system, allowed for long-term VEGF gene expression, which was only activated under hypoxic conditions. For the treatment of mouse limb ischemia, the hypoxia-sensitive vector pVHAVI ameliorated the symptoms much better than the hypoxia-insensitive vector pVAVI. This last result corroborates the idea that the presence of high concentrations of VEGF in ischemic tissue is not beneficial or is less beneficial than maintaining a lower but sufficient and long-term concentration of VEGF locally.