Attempts to investigate the cellular localization of keratin mRNAs by in situ hybridization with specific [35S]-labelled cDNA probes to mouse epithelia have been seriously impeded by the uncontrollable detachment of frozen tissue sections on conventionally coated glass slides (i.e. those coated with egg white, gelatin, collagen). Similarly, a variety of other coating and attachment devices have proved to be unsatisfactory or impracticable for large scale investigations. These difficulties were completely overcome and in situ hybridization was possible after a short immersion of the glass slides in a 2% solution of 3-aminopropyltriethoxysilane in acetone. This treatment provides the glass surface with aminoalkyl groups which are apparently able to react covalently with aldehyde or ketone functions of frozen tissue sections. The resulting firm adhesion of the sections enabled us to investigate the influence of different fixation and prehybridization procedures on the quality of the in situ hybridization. It was found that especially harsh prehybridization, involving hydrochloric acid, heat and proteinase K treatment, drastically reduces the morphological integrity of the sections, thus rendering a reliable assignment of the label difficult. In contrast, a mild prehybridization, consisting mainly of a rehydration of the sections in phosphate-buffered saline and equilibration in 0.1 M glycine, leaves the morphology intact and leads to a highly efficient and specific in situ hybridization.

Specific keratin cDNA probes and monospecific antikeratin antisera were used to analyze mouse epidermis and epidermal tumors for the expression of a type I 47-kDa keratin, K13, normally associated with terminal differentiation of internal stratified epithelia. We demonstrated that this keratin was virtually absent from the entire body epidermis at various stages of development. Also, it was not detected in various forms of acute and chronic epidermal hyperproliferation or in epidermal cells cultured under conditions that favored either cell proliferation or in vitro differentiation. In contrast, K13 was consistently expressed in squamous cell carcinomas of the skin induced by 7,12-dimethylbenz[a]anthracene and 12-O-tetradecanoylphorbol-13-acetate (TPA), whereas papillomas obtained by the same two-stage protocol were distinctly heterogeneous with regard to the expression of this keratin. These findings were true for two different strains of mice (NMRI and Sencar). Papillomas collected from Sencar mice after 12 wk or from NMRI mice after 15 wk of promotion with TPA were either negative for K13 or elicited variable amounts of this keratin. In all cases of positive expression of K13 in tumors, as in normal stratified internal epithelia, both the keratin protein and its mRNA invariably occurred in the differentiating cell compartments. In contrast to what we found in internal stratified epithelia, however, K13 was expressed without its commonly encountered type II 57-kDa partner, K4. Papillomas negative for the K13 protein were also devoid of K13 transcripts. This indicates that the aberrant K13 expression in tumors is regulated at the level of transcription. Our results suggest that K13 may provide a marker for malignant conversion in the mouse two-stage skin carcinogenesis model and may be especially suited for studies of gene expression regulation.

The Drosophila melanogaster tumor suppressor gene lethal(2)tumorous imaginal discs (l(2)tid) causes in homozygotes malignant growth of cells of the imaginal discs and the death of the mutant larvae at the time of puparium formation. We describe the molecular cloning of the l(2)tid+ gene and its temporal expression pattern in the wild-type and mutant alleles. Germ line rescue of the tumor phenotype was achieved with a 7.0 kb Hindlll-fragment derived from the polytene chromosome band 59F5. The l(2)tid+ gene spans approximately 2.5 kb of genomic DNA. The protein coding region, 1,696 bps long, is divided by an intron into two exons. The predicted Tid56 protein contains 518 amino acids and possesses a theoretical molecular weight of 56 kDa. It shows significant homology to all known DnaJ related proteins from bacteria, yeast, and man. The possible function of the Tid56 protein in tumor suppression is delineated.