Ad5-based vectors containing the fibers of Ad serotypes 11 and 35 have shown great promise, as they are able to transduce many human cell types, including human hematopoietic progenitor cells, human DCs, and primary human tumor cells, with greater efficiency than parental unmodified Ad5 vectors (4
). While chimeric Ad5/35 or Ad5/11 vectors are beneficial for ex vivo transduction, their in vivo application is limited by Ad5 antibodies, which are found in most humans (23
). Anti-Ad antibodies interfere with the infection of target cells and increase the toxicity of systemically applied Ad5 vectors (67
). A large portion of anti-Ad5 antibodies are directed against the hexon protein (66
), which is still present in Ad5/35 and Ad5/11 vectors. This problem can be addressed by the construction of vectors completely derived from Ad35 or Ad11. While other investigators have reported on the development of Ad35 vectors (14
), we focused on Ad11 as a basis for a new vector system. Our work on Ad11 vectors was initiated based on the following two findings. First, the prevalence of Ad11-neutralizing antibodies in serum from nonimmunocompromised donors and cancer patients was much lower than the prevalence of Ad5-neutralizing antibodies, indicating that Ad11 vectors may be more applicable for widespread use in humans than Ad5 vectors. In agreement with our observations, low prevalence of neutralizing antibodies to other B group Ad serotypes, including Ad11 and Ad35, in humans has been seen in other studies (10
). Second, vectors containing Ad11 fibers were more efficient than Ad5/35 vectors in transduction of HSCs (61
), indicating the existence of differences in tropism between these vectors, which could be beneficial for gene therapy applications. This prompted us to study in more detail the receptor usage of Ad5/11 and Ad5/35 vectors. We discovered that Ad5/11 appeared to use CD46 as well as a yet unidentified, new receptor(s) and that this receptor(s) cannot be used by vectors containing the fiber from serotype Ad35. It is unlikely that the differences in infection of HeLa and A549 cells by Ad5/11-CMV-GFP and Ad5/35-CMV-GFP in the presence of Ad11p and Ad35 fiber knobs are due to differences in the affinities of Ad11p and Ad35 fibers for CD46, because infection of CHO-C2 cells (which express CD46) by Ad5/11-CMV-GFP and Ad5/35-CMV-GFP is inhibited at similar levels by Ad11p and Ad35 fiber knobs (Fig. ). Other lines of evidence also support the existence of an additional receptor(s) for Ad11, in particular, Ad11p, which, unlike CD46, cannot be used by all B group Ads. First, the amino acid sequences of B group Ad fiber knobs are highly divergent (63
). Second, although Ad11 is typically associated with kidney and urinary tract infections and is commonly found in immunocompromised patients (18
), unlike Ad35, it has also been associated with both respiratory infections and conjunctivitis (1
). Third, radiolabeled Ad11a and Ad11p demonstrate different levels of attachment to a number of human cell lines (35
). Fourth, radiolabeled Ad11p is able to bind to human CD34-positive cells with greater efficiency than Ad3 and Ad35 (61
). Fifth, antiserum to Ad11p fiber, but not Ad11a or Ad35 fiber, can completely block Ad11p binding to A549 cells (35
). Sixth, Ad11p fiber knob can completely block binding of wild-type Ad35 to A549 cells, while recombinant Ad35 fiber knob cannot completely block Ad11p binding (34
). These data support our observations with Ad5/11 and Ad5/35 vectors and provided a rationale to initiate development of Ad vectors based entirely on Ad11p with the aim of developing a vector system with a more beneficial cell tropism than vectors targeted through Ad5 or Ad35 fibers.
Using either eukaryotic homologous recombination or E. coli
recombination, we were able to successfully rescue E1-deleted Ad11 vector genomes that are able to replicate in cells expressing Ad5 E1 and E4 proteins or cells expressing Ad5 E1 proteins and the Ad11-E1B55K protein. Replication of E1-deleted vectors based on the B group serotype Ad35, which shows >98% homology in all ORFs except hexon and fiber (63
), has also been seen in similar cell lines (48
). Since no regions of homology are present between Ad11 vector genomes and the complementing Ad11-E1B55K sequence in 293-Ad11-E1B55K cells, it is unlikely that replication-competent Ad11 vectors would emerge in culture. Although Ad5 E1 sequences from 293-Ad11-E1B55K cells could theoretically be rescued into the E1-deleted Ad11 backbone, it is doubtful that an Ad11(Ad5E1+) vector would emerge in culture since the Ad5 and Ad11 E1 regions show less than 60% DNA homology and since 293 cells expressing Ad5 E1 are not able to support efficient replication of E1-deleted Ad11 vectors. By using the plasmids described in this report, it should be possible to make vectors containing inserts up to ~4.4 kb based on the observation that Ad capsids can package genomes of ~105% the wild-type length. We are currently developing the Ad11 vector system to introduce deletions in E3 that will enable the insertion of larger cassettes. To date, the largest vector genome we have rescued is ~101% the wild-type Ad11 genome length. In 293-Ad11-E1B55K cells the genome-to-PFU ratio of Ad11 vectors was considerably higher than for Ad5 and Ad5/11 vectors, which have genome-to-PFU ratios of ~20:1, which is consistent with previous observations for Ad35 vectors (51
). Although the reason for this higher ratio is not known, it may be a cell-specific observation caused by differences in trafficking of Ad5, Ad5/11, and Ad11 vectors.
One of our concerns over the use of Ad11 as a gene transfer vector was the mild oncogenicity of wild-type B group Ads previously seen in rodents (15
). Although no link between human cancer and Ad7 or Ad11 DNA was found in a previous study with radiolabeled Ad7 and Ad11 DNA and normal and tumor DNA from multiple organs (71
), we wanted to compare the transforming activity of Ad11 with Ad5, which is known to be non oncogenic in rodents (7
). In our studies no transforming activity was seen with E4+ or E1+E4+ Ad11 genome plasmids. (For human Ads the E1 and E4 regions are known to contain transforming genes.) Furthermore, the highest levels of transforming activity were seen with E1+E4+ genome plasmids from nononcogenic Ad5. These observations allay any concerns over the potential oncogenicity of Ad11-based vectors.
A currently accepted paradigm in Ad biology is that the interaction between the Ad fiber and the primary attachment receptor determine the tropism and infection efficiency of Ads. Like all wild-type B group Ads (12
), Ad5/11 and Ad11 vectors are able to utilize CD46 as a cellular receptor; however, although Ad11 and Ad5/11 possess the same fiber, they transduce K562 cells and CD34-positive hematopoietic progenitor cells with different efficiencies and also show different biodistributions upon intravenous injection into CD46 transgenic mice. We speculate that capsid proteins other than fiber (which are different for Ad5/11 and Ad11 vectors) influence virus attachment, internalization, and/or intracellular trafficking in a cell type-specific manner. Alternatively, differences in capsid net charge between Ad5 and Ad11 hexon or penton proteins may affect the interaction between the Ad11 fiber and cellular receptors (56
In order to characterize the in vivo characteristics of Ad11 vectors, biodistribution studies were carried out in transgenic mice that express human CD46 in a similar pattern to humans (25
). These mice express both CAR and CD46 and enable the comparison of Ad5, Ad5/11, and Ad11 vectors following tail vein infusion. Thirty minutes after intravenous delivery, Ad genomes were found in the liver for all three vectors. Although higher levels of Ad5/11 genomes are shown in Fig. , this is not representative of all animals. Similar levels of Ad5, Ad5/11, and Ad11 genomes were found at this time point in other animals (data not shown), and this corroborates our previous observation that similar levels of long-shafted CAR-interacting Ad5/9L and short-shafted CD46 interacting-Ad5/35 genomes are present in liver at 30 min postinjection (53
). At 72 h postinfection, fewer Ad5/11 than Ad5 vector genomes were present in the liver and, as the GFP expression data suggest, these genomes represent vector particles that have transduced hepatocytes. This finding is in agreement with studies in wild-type mice with Ad5/35 vectors (53
). Neither Ad11 genomes nor Ad11-mediated GFP expression was found in livers at 72 h postinfection, indicating that Ad11 particles are efficiently cleared from the liver and do not transduce hepatocytes. A potential explanation for the absence of hepatocyte transduction might be the rapid clearance of Ad11. Alternatively, Ad11 vectors might not be able to transduce hepatocytes via blood factors, as was recently shown for Ad5 and Ad5/35 vectors (57a
). In the lung and kidney, more Ad11 genomes than Ad5 and Ad5/11 genomes were found at 30 min postinfection; however, all vectors were cleared from these organs by 72 h. Notably, it is unlikely that vector signals in Southern blots originated from contaminating blood cells because blood was flushed from all organs before harvesting.
Ad particle trapping in liver, lung, and spleen and subsequent degradation are apparently more pronounced for Ad11 vectors than Ad5 or Ad5/11 vectors. The mechanism of Ad11 vector clearance from these tissues is unclear. We speculate that it involves resident macrophages. Differences between Ad5/11 and Ad11 indicate a role for Ad11 capsid proteins other than fiber in these processes.
While CD46 is a major receptor that interacts with the Ad11 fiber, our in vitro studies suggest the existence of an additional Ad11 receptor. The influence of this yet unknown receptor on in vivo Ad11 tropism remains unclear. Notably, preliminary studies between Ad5/11 and Ad5/35 (Ad35 fiber cannot interact with the new Ad11 receptor) in CD46 transgenic mice (11
) and baboons (S. Ni, K. Bernt, A. Gaggar, and A. Lieber, unpublished data) (baboons express CD46 at similar levels to humans [21
]) did not reveal marked differences in the biodistributions of these two vectors, thus raising a question about the dominant influence of the new Ad11 receptor on in vivo tropism of Ad11 vectors.
Analysis of vector genomes in blood indicates that a large fraction (10 to 20%) of incoming Ad11 particles bind to blood cells (data not shown) and that this phenomenon is more pronounced than for Ad5 and Ad5/11 vectors. Blood cell binding is a problem for all Ad vectors (9
), and the higher levels seen with Ad11 vectors give us an easier model to study this phenomenon. We are currently investigating the blood cell types involved in Ad11 binding, which apparently represent a major trap responsible for clearance not only of Ad11 particles but also Ad5 and Ad5/11 particles.
We have successfully developed a system for generating E1-deleted Ad11 vectors. These vectors are nononcogenic, are less likely to be neutralized in the presence of human serum, and efficiently transduce important gene therapy target cells. Our study, particularly the comparison of Ad11 and Ad5/11 tropism, has also uncovered important differences in in vitro and in vivo transduction between these vectors. Our data indicate that differences in Ad5 and Ad11 capsid proteins other than fiber can indirectly (for example, through electrostatic repulsion) influence the interaction between Ad11 fiber attachment receptors and/or that capsid proteins other than fiber can mediate Ad11 attachment and infection. Our data suggest that the currently accepted model that postulates that the interaction between the Ad fiber and a primary attachment receptor determines Ad infectivity and tropism should be revised. This study also points towards the necessity of careful evaluation of tropism-modified Ad vectors in adequate animal models before their clinical application can be considered.