CMV-lacZ infected grafts (panel A - Nomarsky, and panel B - fluorescence) and Ad.CB-CFTR 
infected grafts (panel C - Nomarsky, and panel D * fluorescence). Grafts exposed to the lacZ 
virus demonstrated autofluorescence below the basal laminal and low level fluorescence in basal 
cells representing endogenous CFTR expression (panels A and B). The grafts exposed to Ad.BC- 
CFTR demonstrated very high level of CFTR expression in approximately 2-10% of the cells. 
Both ciliated and secretory cells expressed the recombinant CFTR which in each case was 
localized on the apical basal lamina. Recombinant CFTR was also expressed in a number of 
intermediated cells. All types of surface epithelial cells expressed the recombinant CFTR 
(panels C and D). These experiment indicate that high levels of recombinant CFTR expression 
can be achieved in the surface epithelium. Furthermore, the recombinant protein is 
appropriately localized to the apical plasma membrane. 
Efficiency of infection of the various bronchial epithelial cell types. Quantification of the 
various cell types infected by 10 12 pfu/ml adenovirus was performed by analysis of 1,000 
transgene-expressing cells detected by X-gal staining and p-galactosidase 
immunocytochemistry. Figure 11 shows the distribution of X-gal positive cell types as 
compared to the normal distribution of that cell type within the epithelium. With a total of 
11.1% lacZ positive cells in the epithelium, 5.5% were ciliated cells, 0.3% were basal cells, 
3.8% were goblet cells and 1.5% were intermediate cells. The diffusion of X-gal precipitate 
and the apparent non-nuclear localization of a majority of the expressing cells made the 
quantification of cell types within these large, highly expressing clusters difficult. We 
performed quantitation of lacZ positive cells by immunocytochemistry using a p-galactosidase 
polyclonal antibody as confirmation of these results. In contrast to X-gal staining, which 
frequently produced a reaction product distributed throughout the cytoplasm, all positive cells 
detected by antibody staining showed a clear concentration of staining within the nucleus. The 
ability to detect p-galactosidase by immunofluorescence provided the opportunity to co-localize 
recombinant gene expression in either basal cells (cytokeratin 14) or columnar cell 
(cytokeratin 18) markers. Figure 12 shows a section of lacZ reconstituted xenograft incubated 
simultaneously with 3 antibodies: red-p-galactosidase, blue-cytokeratin 14 and green- 
cytokeratin 18. Of 1,500 lacZ positive cells counted only 1 colocalized with the basal cell 
marker. 
Expression of adenoviral proteins within human bronchial epithelium. Adenoviral protein 
expression from both wild type and recombinant adenovirus was analyzed by 
immunocytochemical techniques. Antibodies generated to purified adenovirus hexon protein 
(1056F) and purified adenovirus type 3 fiber protein (805F) have been shown to crossreact 
with all 43 human adenoviral serotypes. Both antibodies were directly labeled with FITC and 
used in double immunofluorescence studies with the polyclonal p-galactosidase antibody. Wild 
type adenoviral infection into xenografts produced immunoreactive cells with both adenoviral 
antibodies 805F (Figure 13D and 13E) and 1056F (data not shown) at 21 hours postinfection. 
In contrast, retrovirally infected grafts showed no reactivity with these antibodies (Figure 13F 
and 13G). Xenografts infected with 10 12 pfu/ml recombinant Ad.RSVIacZ adenovirus showed no 
evidence of adenoviral protein expression within 1,500 lacZ positive cells analyzed (Figure 13 
A-C). 
Recovery of adenovi rus from xenografts. Xenografts infected with Ad.RSVIacZ were irrigated 
with PBS approximately every third day and the irrigants layered on 293 cells and evaluated 
for virus using the pfu assay. Figure 14 summarizes the results obtained in three grafts. 
There is a progressive decrease in the quantity of virus recovered during a 2 week period 
following the initial infection; no virus was detected beyond 14 days. 
Summary. These experiments address several critically important issues relevant to gene 
therapy of CF with El deleted adenoviruses. First, they demonstrate potentially therapeutic 
[816] 
Recombinant DNA Research, Volume 16 
