in vitro studies of the effects of rhinovirus on human airway epithelium have used cells grown under conditions known to produce low levels of differentiation. The relevance of the results to native epithelium is questionable. Here we grew primary cultures of human tracheal or nasal epithelium under three conditions. One condition produced pseudostratified, mucociliary cells virtually indistinguishable from native epithelium. The other two conditions produced undifferentiated squamous cells lacking cilia. Cells were infected for 6 h with rhinovirus-16. After a 24-h incubation period, we determined levels of viral RNA in the cells, numbers of infectious viral particles released in the mucosal medium, expression of a variety of epithelial cytokines and other proteins, release of IL-6 and IL-8, and transepithelial electrical resistance and voltage. After infection, levels of viral RNA in the poorly differentiated cells were 30 or 130 times those in the differentiated. Furthermore, expression of mRNA for inflammatory cytokines, release of infectious particles, and release of IL-6 and IL-8 were closely correlated with the degree of viral infection. Thus well-differentiated cells are much more resistant to viral infection and its functional consequences than are poorly differentiated cells from the same source. ion transport; porous-bottomed inserts; short-circuit current; transepithelial electrical resistance RHINOVIRUS (RV) causes severalfold increases in the output of a variety inflammatory cytokines from human airway epithelial cell lines (29,41) and from primary cultures of human airway epithelium (18,23,31,39). However, the cell lines used (A549 and BEAS-2B) generally do not form tight junctions (17,25), and the primary cultures were grown on solid supports, an approach known to produce highly undifferentiated cells (36). We questioned whether results obtained on such cultures were really representative of viral infection of native epithelium. Accordingly, we have grown cells from the same tracheas and nasal scrapings under several different culture conditions to achieve a range of differentiation from a squamous to a full-blown pseudostratified, mucociliary phenotype. We then compared the susceptibility of these various cultures to viral infection. METHODS

We used scanning electron microscopy to count the number of mucous gland openings in the tracheae and lower portion of the larynges of the rat, guinea pig, hamster, mouse and rabbit. Cells of the airway surface epithelium were removed by protease digestion better to visualise the gland openings. The distribution of glands was further studied by conventional histology and by PAS\Alcian blue staining of whole mounts. In all rodent species, gland openings in the larynx occurred with a frequency of 1-2 per mm#. Mice had no gland openings in their tracheae, and hamsters, only a handful. Rat tracheae contained 126p42 gland openings (p..; n l 6) at a frequency of " 0n6 per mm# at the top of the trachea and " 0n15 per mm# at the bottom. Guinea pig tracheae contained 153p90 gland openings (p..; n l 5), with 54 % being in the top 40 % of the trachea. In both rat and guinea pig, tracheal glands were found in the ventral aspect between the cartilaginous rings, and were absent from the dorsal membranous portion. Gland openings in most species were simple circles of " 50 µm diameter. However, glands in the rat trachea generally opened obliquely into shallow (" 20 µm deep) oval troughs (" 150i75 µm), which had their long axes oriented from head to tail. In the rabbit, there was no evidence of tracheal or laryngeal glands histologically. However, the tracheal and laryngeal surfaces contained numerous pits (" 30 µm diameter) distributed evenly over and between cartilages at a frequency of " 4 per mm#. These may correspond to the ' nests ' of goblet cells described by others.