Transductionally Targeted Ad Vectors for Clinical Gene Therapy Applications

As discussed above, the poor efficiency of Ad-mediated gene transfer in several human gene therapy trials has been correlated with a low level of expression of CAR by the target cells. Strategies to accomplish efficient cell-specific gene transfer by Ad vectors in vivo merely by exploiting physical methods to confine vector administration to isolated body compartments have proven inadequate. For example, locally administered Ad vectors carrying the herpes simplex virus thymidine kinase (HSV-TK) gene have been shown to disseminate, probably as a result of leakage into the bloodstream, resulting in a high level of liver-associated toxicity [119]. Substantial hepatic toxicity related to the absence of tumor cell-specific targeting has also been demonstrated in Ad-mediated transfer of the HSV-TK gene in an ascites model of human breast cancer [120]. Thus, targeted Ad vectors capable of efficient and cell-specific CAR-independent gene transfer are required for clinical gene therapy applications.

The benefits accrued in preclinical studies using tropism-modified Ad vectors provide a strong rationale for the immediate employment of these vectors in clinical trials. As discussed above, Ad vectors modified to contain the integrin-targeting RGD motif within the HI loop of the fiber knob have permitted levels of gene transfer to CAR-deficient primary cells to be enhanced more than two orders of magnitude over unmodified vectors. Based on these observations, the University of Alabama at Birmingham is currently employing this vector backbone in Phase I clinical trials for ovarian cancer and recurrent cancer of the oral cavity and oropharynx. These trials are the first to employ tropism-modified viral vectors in human patients. It is hypothesized that the tropism-modified vectors will allow augmented transfer of the herpes simplex virus thymidine kinase and cytosine deaminase genes, respectively, at lower vector doses, thereby leading to increased efficacy and reduced toxicity. These two diseases represent ideal opportunities to perform the initial studies of tropism-modified Ad vectors in the clinical context. In this regard, ovarian cancer is generally confined to the peritoneal cavity, permitting vector administration by injection into that body compartment. Cancer of the oral cavity and oropharynx is a locoregional disease accessible to direct intratumoral injection of Ad vectors. Thus, the anatomical isolation of the disease targets facilitates vector administration.

However, it is apparent that additional requirements will be imposed upon targeted Ad vectors designed for clinical use in disease settings for which systemic vector administration is mandated. It has been reported that intravenous administration of untargeted Ad5 vectors delivers more than 90% of the input virus to the liver, thereby reducing the titer of virus particles available for transduction of the target disease cells [121-123]. Importantly, several studies have shown that the intravenous administration of Ad vectors leads to liver toxicity [124, 125]. Thus, one of the barriers to intravenous delivery of Ad vectors in vivo is the high degree of sequestration by the liver.

Zinn et al. have demonstrated that the liver uptake of intravenously administered technetium (Tc)-99m-labeled recombinant Ad5 knob in mice is significantly reduced upon coinjection of unlabeled Ad5 knob, but is not affected by Ad3 knob, which recognizes a different primary receptor [126]. This indicates that the liver possesses specific receptors for the Ad5 knob, an observation supported by the subsequent reports of high levels of CAR mRNA in the liver [7, 8]. These findings seemed to suggest that successful strategies to reduce liver sequestration and achieve cell-specific targeting following intravenous injection of Ad vectors will necessitate modifications to the knob domain to prevent recognition of CAR. In support of this, Printz et al. [42] and Reynolds et al. [51] have observed significantly reduced transgene expression in the livers of mice injected with Ad5 vectors which are retargeted by bispecific conjugates which prevent binding to CAR. However, the effect of the conjugates in reducing hepatocytes transgene expression may not be due to CAR blockade alone. It is possible that the size of the antibody-complexed vector contributes to the reduction in hepatocyte transduction by effectively enlarging the vector particle such that it less readily transverses the small fenestrations of the mouse liver sinusoidal epithelium.

The reasons for hypothesizing a mechanism other than (or in addition to) CAR blockade stem from the emerging results of studies using Ad vectors whose fibers have been genetically modified so that they no longer recognize CAR. Somewhat surprisingly, Leissner et al. observed that hepatic transgene expression mediated by CAR-ablated vectors following intravenous administration into the tail vein of mice was not significantly reduced compared to unmodified vectors [127]. While the CAR-ablated vectors used in this study were suboptimal in that they did not contain a targeting ligand with specificity for an alternate receptor (and thus may eventually have accumulated in the liver "by default"), these results have called into question the notion that liver transduction by Ad vectors is purely due to the high level of CAR on hepato-cytes. We have in fact found that complexing Ad with the Fab fragment of a neutralizing anti-knob mAb was not sufficient to reduce liver transgene expression, whereas when conjugated to a ligand, the same Fab fragment achieved the desired reduction. Whether this is due to particle size or the need for an effective alternate ligand is as yet unclear. Hence, modification of Ad vectors to avoid hepatocyte transduction does not appear to be as straightforward as first thought. Additional mutations such as ablating the RGD motif in the penton base to avoid interaction with cellular integrins have been proposed and are currently under evaluation, as are studies using vectors genetically modified to increase particle size by extending shaft length with a view to diminishing viral penetration of the hepatic fenestrations. The combination of "liver untargeting" approaches with a genetically incorporated, truly specific ligand are eagerly awaited. It is clear that the development of rational strategies to facilitate the clinical application of systemically administered Ad vectors will be dependent on a better understanding of the biological basis of hepatic vector localization.

The problem of liver sequestration of Ad vectors is not an issue which relates only to hepatocytes. In this regard, Reynolds et al. showed that while transductional targeting of Ad led to a reduction in hepatic transgene expression, the biodistribution of viral DNA 1.5 h after intravenous administration was not significantly altered [51]. The authors hypothesized that this could reflect nonspecific phagocytic uptake by Kupffer cells. It has previously been shown that 90% of Ad DNA is eliminated by the liver within 24 h in an early innate immune response and does not lead to transgene expression [128]. Inhibition of Kupffer cells reduces the elimination of Ad DNA from the liver and leads to a three- to fourfold increase in hepatic transgene expression [129, 130]. This suggests that further improvements in the use of targeted Ad vectors for systemic gene delivery might necessitate strategies to mitigate against nonspecific sequestration of the vector by the reticuloendothelial system (RES). In accordance with this, Tao et al. have generated data in mice suggesting that low doses of Ad (1-3 x 1010 viral particles) are efficiently taken up by the RES/Kupffer cells, whereas high doses (1 x 1011 viral particles) saturate these cells [131].

While the hepatic sinusoids and their fenestrations constitute a highly favorable anatomic environment for Ad entry and are thus a problem in the context of liver sequestration of the vector, anatomical factors in other tissues may actually impede Ad transduction. In support of this idea, Fechner et al. have reported that expression of CAR and av integrins does not correlate with Ad vector-mediated gene delivery in vivo [132], suggesting that anatomical barriers, in particular the endothelium and the subendothelial matrix, need to be overcome in order to achieve organ-specific gene delivery. Thus, efficient gene transfer by targeted Ad vectors might require the implementation of additional methods to permeabilize anatomical barriers. In this regard, Maillard et al. have shown that pretreatment of the rabbit iliac artery with elastase could enhance Ad-mediated gene transfer to arterial smooth muscle cells after balloon abrasion [133]. In a similar approach, Kuriyama et al. have reported that the administration of proteases to degrade the fibrous proteins of the extracellular matrix prior to intratumoral injection of Ad vectors leads to increased Ad infection [134], An in vitro study by Nevo et al. demonstrated that the endothelial cell monolayer presents a physical barrier to Ad infection of myocytes, which could be partially overcome by increasing endothelial permeability with a-thrombin [135]. Protease digestion might also prove a rational strategy to increase the permeability of the basal lamina, which has been shown by Huard's group to present a physical barrier to the transduction of mature skeletal muscle by both untargeted and tropism-expanded Ad vectors [136, 137]. In a quite different approach, Cho et al. demonstrated that the efficiency of transduction of mature skeletal muscle could be enhanced by administering Ad vectors in a large solvent volume, thereby increasing the hydrostatic pressure and favoring vector egress out of the intravascular compartment [138]. In contrast to the anatomical situations described above, the "leaky" vasculature associated with solid tumors [139] is hypothetically favorable for the vascular egress of Ad tumor-targeted Ad vectors.

Thus, it is apparent that the success of systemic administration of targeted Ad vectors will depend on a greater understanding of the receptor-independent biological factors such as vascular pharmacodynamics and anatomical barriers limiting their utility, which should, in turn, facilitate the development of rational strategies whereby these obstacles might be overcome.

0 0

Post a comment