Intracardiac injection of a capsid-modified Ad5/35 results in decreased heart toxicity when compared to standard Ad5

Cardiovascular disease (CVD) is the leading causes of death in the western world, even after steady decline due to improved risk control and effective invasive treatment methods. Annually over 190 000 people die from CVD in the UK alone and the health care costs are over 3.2 billion pounds. As existing methods are clearly inadequate, novel treatment methods, such as gene therapy, are clearly needed.

Adenoviruses (Ads) are the most utilized gene therapy vectors. Over 50 different Ad serotypes, divided in six different subgroups (A-F) have been identified to date [1]. Viruses in different subgroups have varying tissue tropism and recognize different cellular receptors. Coxsackie-adenovirus receptor (CAR) has been identified as the primary attachment receptor for many Ads, including the most utilized serotype 5 (Ad5) [2‐4]. Group B Ads have been shown to infect cells in CAR-independent manner and the cellular receptor has been identified as CD46 for most group B viruses [1, 5, 6]. CD46 is a complement regulatory molecule ubiquitously expressed by all nucleated cells in human body [7]. Recently it was also reported that a subgroup of B adenoviruses use desmoglein 2 receptor for primary attachment [8], but serotype 35 was not included in this group.

Several preclinical gene therapy strategies have been used for cardiovascular disease using different viral vectors and non-viral gene transfer and various modes of administration [9]. Most clinical cardiovascular gene therapy trials have used the standard Ad5 vector and one of the lessons from these trials has been the key role of efficient gene transfection and delivery in achieving clinically meaningful therapeutic efficacy.Thus, it has been clear that Ad vectors must be engineered to enhance their ability to transfect the target tissue. Several methods have been developed to transductionally target Ad vectors including the use of bivalent targeting complexes and engineering genetic alterations to viral capsid proteins [10]. One targeting method which has recently been successfully used in many applications is to use hybrid adenoviruses, which are created by interchanging capsid fibers between serotypes, usually with Ad5 receiving the fiber from another serotype [11]. In this study we used such a hybrid vector, Ad5/35, which has fibers from Group B Ad serotype 35 and thus uses CD46 instead of CAR to gain entry into cells. Previous data demonstrate that Ad5/35 hybrid virus is able to efficiently transduce many cell types that are relatively resistant to Ad5 infection, including hematopoietic stem cells, tumor endothelial cells, and several cancer cell types [12‐18]. Previously we have reported an analysis of various Ad receptors in both human dilated cardiomyopathy (DCM) hearts and non-DCM hearts demonstrating significant differences in expression patterns of CAR and CD46, with CAR receptor demonstrating higher expression in both DCM and non-DCM hearts [19].

In addition to vector development, the method of gene delivery plays a crucial role in achieving efficient gene transfer to the heart. Various modes of administration have previously been used to deliver genes to the heart, including intravascular, intracoronary, intramyocardial and intramuscular routes. Ultrasound-guided intramyocardial injection of gene delivery vectors is an attractive and clinically applicable concept to directly infect cardiomyocytes. In this study we employed this method to test the hypothesis that the use of tropism-modified adenoviruses would improve transduction efficacy for cardiovascular applications and to this end evaluated the transductional capability and toxicity of a capsid-modified Ad5/35 adenoviral vector in comparison to standard Ad5 vector in the heart.

Despite the availability of several promising genes to treat cardiovascular disease, their optimal delivery to the heart remains a challenge. A number of gene delivery vectors and methods of administration have previously been used for cardiovascular applications with variable success [9]. In this paper we combined two promising approaches, hybrid adenoviral vectors and ultrasound-guided intracardiac delivery, and analyzed the differences in transduction efficiency and toxicity after intracardiac injection between Ad5 and Ad5/35 vectors. We used a transgenic mouse strain hCD46Ge, which expresses the Ad5/35 receptor, human CD46, in a similar pattern as in humans, to compare the viruses in a situation comparable to man [21, 25].

The viral vectors were injected to the left ventricle wall by using a closed-chest ultrasound guided system. With this system vectors can be administered directly into the left ventricle wall, without traumatic open-chest surgery. The injection leaves only a minor needle track and no other effects to cardiac function were observed. Safety of this injection method has been shown in healthy C57BL mice by advancing the needle 1-5 times to myocardium and also by repeated saline injections without compromising left ventricular function (unpublished data, Koskenvuo).

Transductional targeting has been widely used to improve efficacy and reduce toxicity in various gene therapy applications, most notably in cancer gene therapy [26]. Numerous reports have demonstrated marked improvements in therapeutic efficacy with the use of the hybrid adenovirus technology. For cardiovascular gene therapy, however, reports on transductional targeting applications for the myocardium have thus far been limited. We analyze here for the first time the effect of adenoviral transductional targeting on transgene expression and toxicity after intramyocardial closed-chest injection. Most previous studies involving adenoviral activity in the heart are biodistribution studies after intravenous (i.v.) administration. It has been shown previously that after systemic i.v. administration both Ad5-lacZ and Ad5/35-lacZ genomes are present at relatively high concentrations in the hearts of hCD46Ge mice as well as baboons and for Ad5/35-lacZ heart is one of the most effectively transduced tissues in both models [27, 28]. Interestingly, with the primate model no transgene activity was reported in the heart tissue for any vector despite the presence of viral genomes. Analysis of genome copies present in tissues does not discriminate between active or inactive viral particles or between intracellular genomes and viruses in extracellular space. Unfortunately, only this method of analysis was employed by Ganesh et al. (2009). Here we show also the expression and activity of viral transgene in hCD46Ge mice, indicating that transgenes from both vectors are transcribed in our murine model. In their work Ganesh et al. (2009) also analyzed vector-mediated toxicity by measuring serum IL-6 and liver enzyme levels and showed decreased overall toxicity with Ad5/35-lacZ as compared to Ad5 vector after i.v. administration. The highest serum IL-6 levels in our experiment were 87 pg/ml, only 9% of the reported serum IL-6 levels after i.v. administration (Figure 4) [27]. In addition, our analysis of organs other than the heart revealed no signs of toxicity nor adenoviral DNA, indicating that systemic toxicity is low in our intracardiac injection model. Other vector systems have previously been used for gene transfer into the heart, including adeno-associated virus (AAV). Previous studies have demonstrated high tropism for the myocardium for AAV serotypes 1,6,8 and 9, making the AAV system a potentially attractive delivery vehicle for cardiovascular gene therapy [29]. The advantages of adenovirus-based approach include higher payload capacity and long experience from clinical trials, and serotype-modification strategies can potentially alleviate the strong immune response associated with adenovirus treatment.

There are also a few previous reports using ultrasound-guided intramyocardial injection of adenovirus in mice. Huusko et al. (2010) recently used this method to inject Ad5 expressing lacZ or different isoforms of VEGF, and the inflammatory reaction was found to be moderate at 14 days post injection, and the level of tissue damage was found to be also dependent on the transgene with VEGF-A giving rise to more inflammatory reaction than lacZ [30]. Li et al. (2005) used an Ad5 vector with deletions of the E1, E2a, and E3 regions expressing inducible nitric oxide synthase and reported little inflammatory reaction after intramyocardial delivery [31].

Our previous studies on adenoviral receptor expression in human heart tissue suggested that the native Ad5 capsid configuration would be better suited for cardiac gene transfer than Ad5/35 vectors as CAR had higher levels of expression than CD46 in both normal and dilated cardiomyopathy hearts [19]. As expected, the injection of unmodified Ad5 vector resulted in significantly more efficient transgene expression than administration of hybrid vector Ad5/35 (Figure 3, Figure 4). However, although both vectors elicited an immune response, this response was markedly more severe after Ad5 administration. At 6 days post injection there was not a significant difference in toxicities between the two viruses but by 14 days the Ad5 immune reaction showed signs of irreversible tissue damage with large areas of myocyte necrosis, whereas Ad5/35 had features of resolving immune reaction. These results are in accordance with a previous report from Nanda et al., where the addition of Ad5 fiber knob to Ad35-based vectors was found to substantially increase the immunogenicity of Ad vectors [32]. LacZ transgene is also immunogenic and it is theoretically possible that the increased toxicity of Ad5 observed here could be caused by higher expression of lacZ. However, previous studies using different AAV vectors to deliver lacZ to mouse hearts have shown a mild immune response and no overt abnormalities in cardiac function despite high levels of lacZ expression, suggesting that immune response to lacZ does not play a significant role in our model [33].

The results suggest that, although Ad5 is more effective in transducing cardiac cells after intracardiac injection, this vector clearly has a worse safety profile and safer gene delivery can be achieved by using Ad5/35 hybrid vector. The definitive tissue damage after Ad5 injection will undermine the therapeutic effect of the delivered transgene and probably lead to severe side effects, although no adverse effects were evident during the 14 day post injection observation period. It must be also noted that even though Ad5 injection resulted in more efficient transgene expression, the cardiac tissue was not refractory to Ad5/35 infection and robust transgene expression was also achieved using this vector. It is notable that due to differences in viral particle/pfu ratios between the two viruses, Ad5/35-lacZ mice received 6.65 times more viral particles and still demonstrated less toxicity than A5-lacZ mice.

Results described here together with previous studies suggest that changing Ad5 fiber to those of Ad35 increases the safety profile of gene therapy vectors, although at the cost of some transduction efficiency. Ultrasound-guided intracardiac injection is a promising method for introducing genes to the heart. The increased safety with Ad5/35 vector warrants for further development of hybrid vector system for therapeutic purposes.