Role of oxygen in the zonation of carbohydrate metabolism and gene expression in liver. Hepatocytes around the afferent (periportal) vessels differ from those around the efferent (perivenous) vessels in their contents of key enzymes, and therefore have different metabolic capacities. Thus, the model of "metabolic zonation" proposes that the periportal cells produce glucose via glycogenosis and gluconeogenesis and that the perivenous cells utilize glucose via glycogen synthesis and glycolysis. The periportal and perivenous cells receive different signal patterns, because substrates including oxygen and hormones are degraded and products and mediators are formed during passage of blood through the liver. The different signal patterns should be important for both short-term regulation of metabolic rates and for long-term induction and maintenance of the enzyme equipments by control of gene expression. From the periportal to the perivenous zone, the concentration of the signal oxygen falls corresponding to a drop from about 13 (arterial) to 9 (mixed periportal) and then to 4 (hepatovenous) volume% gas atmosphere. For short-term regulation of metabolism, in perivenous-like cells net glucose production measured over a period of two hours was observed below 2%, net glycogen synthesis above 4%, and net lactate utilization above 6% oxygen. In periportal-like cells net glucose formation and net lactate utilization increased sharply from anoxia to 6% oxygen and then only moderately. For long-term regulation of gene expression, the glucagon (cAMP)-dependent activation of the PCK gene was modulated by oxygen. The transcriptional rate, the abundance of mRNA and the enzyme activity were increased to higher levels under arterial rather than under venous oxygen. Conversely, the insulin-dependent activation of the glucokinase gene was negatively modulated by oxygen. A heme protein appeared to be involved in oxygen sensing, since CO mimicked the effects of oxygen on the PCK gene. Hydrogen peroxide was produced by hepatocytes as a function of oxygen tension; exogenously added, it mimicked the effects of oxygen on PCK gene induction. Therefore, the heme protein containing an oxygen sensor could be a peroxide producing oxidase. It is not known at present whether the same oxygen sensor is also involved in the short-term regulation by oxygen of hepatic carbohydrate metabolism. Transfection of PCK promoter-CAT gene constructs into primary hepatocytes showed that oxygen modulated PCK gene activation in the region of −277/+73. This modulation was not mediated by isolated cAMP responsive elements.