lmmunohistochemical studies first demonstrated that, both in human and rat liver, glutamine synthetase is expressed only in a narrow zone surrounding the terminal hepatic venule (Gebhardt & Mecke, 1983). This peculiar heterogeneous distribution of the enzyme protein has recently been confirmed by direct analysis of glutamine synthetase activity in periportal and perivenous cell lysates (Quistorff & Grunnet, 1987) or hepatocytes (Ugele et al., 1987; K. Lindros unpublished work) isolated by digitonin-collagenase perfusion (Lindros & Penttila, 1985). The so-called glutamine cycle encompassing periportal ammonia fixation and glutamine utilization for urea synthesis and perivenous glutamine synthesis has been established (Haussinger, 1983) and has been considered to have an important role in hepatic and systemic pH regulation. The factors governing the local expression of glutamine synthetase are, however, not known. Recently, a heterogeneous hepatic expression of another enzyme, carbonic anhydrase (CA), resembling that of glutamine synthetase, was described (Carter et al., 1987). The highand low-activity cytoplasmic forms CAI1 and CAIII, identified by specific radioimmunoassay, were shown to exhibit growth hormone-mediated sexual dimorphism in rat liver. While there was 3-4 times more CAI1 in female than in male liver, CAI11 was up to 30 times more abundant in male liver. Immunohistochemistry revealed that although weak staining was observed all over the lobule, both CAI1 and CAI11 stained intensively around the central veins. However, while livers from males exhibited much higher CAI11 activities than females in the perivenous zones, CAI1 activities were much higher in livers from females. The mitochondria1 CA isoenzyme, CAIV serves the production of urea (Dodgson et al., 1983). This process is more active in the periportal zone (Haussinger, 1983; Poso et al., 1986). Since glutamine synthetase and carbonic anhydrase are both intimately involved in regulation of pH and metabolism, and since both are differentiation markers in embryonic neural retina (Vardimon et al., 1986), their possible coexpression in rat liver was investigated. Glutamine synthetase was measured radioisotopically, by separating [ 14C]glutamate and [ 14C]glutamine by ionexchange (Pishak & Phillips, 1979). Total CA activity was measured by a pH indicator method and CAI1 and CAIII isoenzymes by specific radioimmunoassay. Periportal and perivenous hepatocytes were isolated by the digitonincollagenase perfusion technique (Lindros & Penttila, 1985; Quistorff, 1985). Hepatocytes were inoculated on culture dishes coated with rat tail collagen at a density of 6.3 x lo4/ cm2. A 22 mwbicarbonate/l5 mM-Hepes buffered 1 : 1 mixture of Waymouth MB 752/1 and Ham F-12 was ordinarily supplemented with a 5% (v/v) fetal calf serum, 5% (v/v) newborn calf serum (Gibco), insulin (10 units/l), dexamethasone (1 p ~ ) , gentamycin (10 mg/l) and nystatin ( 1 mg/l). Cultures were always supplemented with serum for the first 4 h after plating. Analysis of freshly isolated cells revealed a striking and consistent difference both in total COz hydrase activity and in CAI1 and CAI11 concentrations between periportal and