M.J. Welsh and A.E. Smith, RAC Application 
Finally, complementation of the growth of Ad2/CFTR-1 could occur if the preparation 
were infected with low levels of wild-type virus. PCR estimates of the amount of Ad2 or 
Ad5 El DNA suggest that such contamination should be below lO 8 . Contamination at 
such a level could result from recombination between defective virus and the endogenous 
Ad5 El DNA sequences present in the 293 cells used to grow virus. Such a recombinant 
would be a Ad5ElAd2 hybrid and eventually we would expect it to overgrow the defective 
virus preparation. 
Recombination and DNA stability 
Another way in which the adenovirus genome could be unstable would result from 
recombination. Homologous recombination occurs readily within adenovirus of the same 
subgroup (105); indeed it is routinely used in the laboratory to create variant strains. 
However, the products of homologous recombination would closely resemble the starting 
viral strains, i.e., an Ad2 recombinant with newly acquired El genes would essentially be a 
wild-type virus and a CFTR cDNA containing recombinant of the serotype C strain would 
be expected to have properties similar to Ad2/CFTR-1, especially a limited ability to 
replicate. 
Non-homologous recombination could also occur but the exact structure of the products are 
less predictable. To be viable, the Ad2/CFTR-1 would at a minimum need to acquire El 
functions and delete other DNA sequences so as to enable packaging. The deleted 
sequenc es co uld involve dispensable viral genes, perhaps E3, or more likely sequences 
w ithin CFTR. The activity of such recombinants is difficult to predict, but fragments of the 
CFTR protein seem unlikely to have activity. We have no experimental data testing for 
recombination, but to date we have not observed the generation of such recombinants 
during the routine passaging of Ad2 vectors in tissue culture cells. 
B.2.b. Laboratory studies of gene transfer and expression. 
B.2.b.(l) What animal and cultured cell models were used in laboratory studies to assess 
the in vivo and in vitro efficacy of gene transfer system? In what wavs are these 
models similar to and different from the proposed human treatment? 
1. Cell lines and primary cultures 
We tested the ability of Ad2/6Gal-l and Ad2/CFTR-1 to express their respective proteins 
in several cell lines, including human HeLa cells, human 293 cells, human pancreatic 
epithelial cells (CF PAC) (130), and monkey bronchiolar epithelial cells (4MB R-5). In 
addition, we studied primary cultures of hamster tracheal epithelial cells, cynamologous 
monkey tracheal/bronchial epithelial cells, Rhesus monkey tracheal epithelial cells, and 
normal and CF human airway epithelia. These cells were all grown on culture dishes. 
Studies with these models demonstrate the ability of the vector to transfer the DNA and 
express the appropriate protein in epithelial cells. They are, however, different from the 
proposed human studies in that the cells do not form an intact epithelium. 
2. Primary cultures of human airway epithelia grown on permeable filter supports. 
In this model system, primary cultures of human airway epithelial cells are grown on 
permeable filter supports at the air-liquid interface with air on the mucosal surface (131- 
134). After seeding, these cells differentiate and form tight junctions to produce an 
electrically tight epithelial monolayer. Such cultures closely resemble the native 
epithelium, both morphologically (i.e., differentiated cells with tight junctions and a 
distinct apical and basal-lateral membrane) and functionally (i.e., low transepithelial 
Recombinant DNA Research, Volume 16 
[887] 
