The rat intestinal fatty acid binding protein (I-FABP) gene exhibits cell-specific as well as regional differences in its expression within the continuously regenerating small intestinal epithelium. To investigate the underlying mechanisms, we linked portions of its 5' nontranscribed domain to the human growth hormone (hGH) gene and analyzed expression of the hGH reporter in transgenic mice by RNA blot, solution hybridization, and immunocytochemical techniques. Sequences located within 277 nucleotides of the start site of I-FABP transcription are sufficient to limit hGH expression to the intestine. Although the absolute levels of hGH mRNA in the duodenum and proximal jejunum of these transgenic mice were similar to those of I-FABP mRNA, steady-state hGH mRNA concentrations were approximately 100 times lower in their distal small intestine. Addition of nucleotides -278 to -1178 of the I-FABP gene "restored" hGH mRNA concentrations in the distal jejunum and ileum to levels comparable to murine I-FABP mRNA. Serum hGH levels were 1000 times lower in the "short promoter" transgenic mice compared to animals with the "long promoter" transgene, indicating that efficient distal small intestinal hGH expression is required to produce elevated hGH concentrations in serum. The distribution of hGH in villus-associated enterocytes and goblet cells and its lack of expression in the crypts of Lieberkuhn mimicked that of the endogenous I-FABP gene product in all transgenic pedigrees. However, bands of hGH-negative cells extending from the base to the tips of villi were frequently observed in mice that were heterozygous for the short promoter transgene. This mosaic staining was not observed for I-FABP. These data suggest that (i) different cis-acting sequences may be required for complete expression of proximal-distal I-FABP gradients than for recapitulation of its normal crypt-villus tip distribution; (ii) differences may exist in the export pathways of secreted proteins within enterocytes located in various regions of the small intestine; and (iii) there may be subtle genetic differences among various crypt stem cells that can be detected in vivo by observing mosaic patterns of transgene expression along the villus epithelium.