As with studies on the Na + / K + pump24, early results on Na+ /Ca 2 + exchange appeared inconsistent with consecutive ion exchange models (see ref. 5 for review). Our work provides strong support for a consecutive Na +/Ca2+ exchange mechanism a nd should facilitate structure-function studies ofthe cloned Na+ /Ca2+ exchanger. The results of both ion jump experiments and steady-state I ~aCa measurements are inconsistent with Ca2+ translocation involving net negative cha rge movement 8• We conclude from our site-density estimates and I NaCa measurements that max imum exchanger turnover rates are about 5,000 S -I . As alread y proposed for the Na + / K + pumpl 2-17 and rod out er segment Na + / Ca2+ , K + excha nger 2 ,22, voltage-dependence must reside in the binding and release of extracellular Na+ or a closely associated occlusion/ deocclusion reaction. THE al and a2 domains of major histocompatibility complex (MHC) class I molecules function in the binding and presentation of foreign peptides to the T-cell antigen receptor and control both negative and positive selection of the T-ce)) repertoire l -3 , Although the a3 domain of class I is not involved in peptide binding, it does interact with the T-cell accessory molecule, CD8 (refs 4, 5). CD8 is important in the selection of T cells as anti-CD8 antibody injected into perinatal mice interfers with this process 6 • We previously used a hybrid class I molecule with the a 11 a2 domains from L d and the a3 domain from Q7 b and showed that this molecule 718 binds an Ld-restricted peptide but does not interact with CD8-dependent cytotoxic T lymphocytes 7• Expression of this molecule in transgenic mice faUs to negatively select a subpopulation of anti-L d cytotoxic T Iymphocytes. In addition, positive selection of virus-specific L d -restricted cytotoxic T lymphocytes does not occur. We conclude that besides the a1la2 domains of class I, the a3 domain plays an important part in both positive and negative selection of antigen-specific cells.To examine the role of the a3 domain of class I molecules in the responses of alloreactive and antigen-specific cytotoxic T Iymphocytes (CTL), we generated two transgenic mouse strains, C3H.L d and C3H.L Q 3, derived from C3H/ HeJ (H_2 k ) mi ce. C3H .L d express inta ct L d , whereas C3H.LQ3 ex press a n LU molecule whose a3 domain is switched with Q7 b . The Q7 b a 3 domain differs from LU at live residues and in addition has a unique three-amino-acid insert at positions 275-277 (ref. 8). We showed that this molecule (L Q3) binds the same viral peptide as L d but is not recognized by peptide-specific CD8-dependent CTL (ref. 7). Similar observations have been reported with a 3 mutant class I molecules not recognized by CD8-dependent CTL, the defect being due to an alteration in the binding site for C D8 (refs 4,5). 80th L d and L Q3 genes contain the same Qj .0

We previously demonstrated that C3H/HeJ transgenic (TG) mice that express a laboratory-engineered class I molecule, Q1O/L, exclusively on liver parenchymal cells show no evidence of hepatic disease even after deliberate immunization. Nevertheless, these animals demonstrate cytotoxic T-lymphocyte (CTL) activity specific for Q1O/L, although it is less than that obtained from non-TG littermates. We now show that this decrease in CTL activity is not a reflection of a decrease in precursors, since both TG and normal animals have similar numbers. When non-TG C3H mice are primed with H-2Ld and H-2KbmI antigens, which extensively crossreact with Q1O/L, their specific in vitro CTL activity directed against H-2Ld, H-2Kbml, and Q1O/L is increased 10-to 20-fold, as expected. Although primed TG mice show similar increases in in vitro CTL activity directed against H-2Ld and H-2Kbm'l, they display no increase in anti-Q10/L activity. Whereas anti-H-2Ld spleen cells from non-TG mice readily generate CTL lines and clones specific for H-2Ld and Q1O/L, TG cells give rise to anti-H-2Ld lines or clones only. These data indicate that the tolerance in TG mice is accounted for by the inactivation or deletion of an important CTL subpopulation having the capability of recognizing the peripheral antigen in situ. To determine whether tolerance would persist in the absence of Q1O/L, TG cells were transferred into non-TG recipients. Three weeks later Q1O/L-specific lytic activity generated in in vitro bulk cultures remained reduced compared to non-TG cells, indicating that the tolerant phenotype was stable during this interval.

A hybrid murine class I gene, Q10/L, was injected into C3H/HeJ fertilized ova to produce transgenic (TG) mice. This fusion gene contained 414 bp of Q10 promoter sequences which was sufficient to direct liver-specific expression in two lines of animals. Animals from these lines did not have Q10/L mRNA in 10 nonhepatic tissues examined including thymus, spleen, and bone marrow. The ontogeny of Q10/Ld expression in both liver and yolk sac paralleled expression of endogenous Q10. Analysis of liver cells from these lines by flow cytometry and immunofluorescence demonstrated the presence of the Q10/L Ag solely on hepatocytes. TG animals showed no signs of hepatic disease as evidenced by an absence of cellular infiltrates in the liver and a normal profile of serum enzymes that are elevated in association with hepatic disease. When spleen cells from TG animals were cocultured with splenocytes that express Ag cross-reactive with Q10/L, CTL were generated that recognized and lysed L cells which express Q10/L. However, the extent of lysis was less than that generated from non-TG control littermates. That these cross-reactive T cells were physiologically significant was demonstrated by adoptive transfer of in vivo primed T cell enriched spleen cells which produced a mononuclear infiltration of the liver of TG recipients. However, inoculation of Q10/L L cells or splenocytes expressing Q10/L cross-reactive Ag into TG mice did not induce cellular infiltration or overt hepatic disease. Whereas inoculation of normal C3H mice with these cells led to priming of Q10/L reactive CTL, anti-Q10/L CTL could not be primed in TG mice. This suggests that Ag expression solely on hepatocytes can lead to inactivation of specific CTL clones and thus account for the observed in vivo tolerance.