Mapping the signal sequence-binding site on SRP reveals a significant role for the NG domain

We present evidence that the signal recognition particle (SRP) recognizes signal sequences via the NG domain on the SRP54 protein subunit. Using a recently developed cross-linking method (Fancy, D. A., and Kodadek, T. (1999) Proc. Natl. Acad. Sci. U. S. A. 96, 6020-6024; Correction (1999) Proc. Natl. Acad. Sci. U. S. A. 96, 1317), we find that signal peptides cross-link to the Escherichia coli SRP protein Ffh (the homologue of the mammalian SRP54 subunit) via the NG domain. Within the NG domain, the cross-linking site maps to the ras-like C-terminal subdomain termed the G domain. This result stands in contrast to previous studies, which concluded based on nascent chain cross-linking that the signal sequence bound to the adjacent M domain. As independent evidence of a direct binding interaction between the NG domain and the signal sequence, we find that the NG domain of Ffh binds signal peptides as an isolated entity. Our results suggest that the NG domain forms a substantial part of the binding site for the signal sequence.     

Comparative efficiencies of C-terminal signals of native glycophosphatidylinositol (GPI)-anchored proproteins in conferring GPI-anchoring.

Every protein fated to receive the glycophosphatidylinositol (GPI) anchor post-translational modification has a C-terminal GPI-anchor attachment signal sequence. This signal peptide varies with respect to length, content, and hydrophobicity. With the exception of predictions based on an upstream amino acid triplet termed omega-->omega + 2 which designates the site of GPI uptake, there is no information on how the efficiencies of different native signal sequences compare in the transamidation reaction that catalyzes the substitution of the GPI anchor for the C-terminal peptide. In this study we utilized the placental alkaline phosphatase (PLAP) minigene, miniPLAP, and replaced its native 3' end-sequence encoding omega-2 to the C-terminus with the corresponding C-terminal sequences of nine other human GPI-anchored proteins. The resulting chimeras then were fed into an in vitro processing microsomal system where the cleavages leading to mature product from the nascent preproprotein could be followed by resolution on an SDS-PAGE system after immunoprecipitation. The results showed that the native signal of each protein differed markedly with respect to transamidation efficiency, with the signals of three proteins out-performing the others in GPI-anchor addition and those of two proteins being poorer substrates for the GPI transamidase. The data additionally indicated that the hierarchical order of efficiency of transamidation did not depend solely on the combination of permissible residues at omega-->omega + 2.     

Phosphatidylinositol glycan (PI-G) anchored membrane proteins. Amino acid requirements adjacent to the site of cleavage and PI-G attachment in the COOH-terminal signal peptide.

Secreted proteins are processed from a nascent form that contains an NH2-terminal signal peptide. During processing, the latter is cleaved by a specific NH2-terminal signal peptidase. The nascent form of phosphatidylinositol glycan (PI-G) tailed proteins contain both an NH2- and a COOH-terminal signal peptide. The two signal peptides have much in common, such as size and hydrophobicity. The COOH-terminal peptide is also cleaved during processing. We propose that the amino acid in a nascent protein that ultimately combines with the PI-G moiety be designated the omega site. Amino acids adjacent and COOH-terminal to the omega site would then be omega + 1, omega + 2, etc. In previous studies, we showed that allowable substitutions at the omega site of an engineered form of placental alkaline phosphatase (miniPLAP) are limited to 6 small amino acids. In the present study, mutations were made at the omega + 1 and omega + 2 sites. At the omega + 1 site, processing to varying degrees was observed with 8 of the 9 amino acids substituted for alanine, the normal constituent. Only the proline mutant showed no processing. By contrast, the only substituents permitted at the omega + 2 site were glycine and alanine, with only trace activity observed with serine and cysteine. Thus, just as there is a -1, -3 rule for predicting cleavage by NH2-terminal signal peptidase, there appears to be a comparable omega, omega + 2 rule for predicting cleavage/PI-G addition by COOH-terminal signal transamidase.     

Intramolecular cross-linking of myosin subfragment 1 with bimane

We previously showed that the fluorescent inter-thiol cross-linker dibromobimane (DBB) [Kosower, N. S., Kosower, E. M., Newton, G. L., & Ranney, H. M. (1979) Proc. Natl. Acad. Sci. U.S.A. 76, 3382-3386] cross-links two [50 and 20 kilodaltons (kDa)] of the three major fragments of myosin subfragment 1 (S-1); on intact S-1, DBB quenches tryptophans and inhibits all ATPases [Mornet, D., Ue, K., & Morales, M. F. (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 1658-1662]. Here we characterize the modification chemically: DBB cross-links Cys-522 (50 kDa) with Cys-707 (20 kDa), thereby sealing a large preexisting heavy-chain loop containing important functionalities. Cross-linking rate is insensitive to nucleotides, but apparently sterically, either monobromobimane or DBB reduces Ca2+-ATPase to low, nonzero levels.     

Site of action of RNase I on the 50 S ribosome of Escherichia coli and the association of the enzyme with the partially degraded subunit.

The 50 S ribosome of Escherichia coli is partially degraded by RNase I in presence of a high concentration of Mg2+ (10 to 20 mM); the partially degraded subunit becomes resistant to the further action of RNase I. The latter remains latent in association with the subparticle as in case of 30 S ribosome (Neu, H.C., and Heppel, L.A. (1954) Proc. Natl. Acad. Sci. U.S.A. 51, 1267-1274). As a result of nucleolytic action, 23 S RNA is degraded to a smaller size and four proteins (L4, L10, L7/L12) are released from the subunit. From the location of these proteins, it appears that the primary site of action of RNase I is the central protuberance of the armchair model proposed for the subunit (Stoffler, G., and Whitman, H.G. (1977) in Molecular Mechanisms of Protein Biosynthesis (Weissbach, H., and Pestka, S., eds) pp. 117-144, Academic Press, New York).     

Proprotein interaction with the GPI transamidase

For characterizing how the glycosylphosphatidylinositol (GPI) transamidase complex functions, we exploited a two-step miniPLAP (placental alkaline phosphatase) in vitro translation system. With this system, rough microsomal membranes (RM) containing either [(35)S]-labeled Gaa1p or epitope-tagged Gpi8p, alternative components of the enzymatic complex, were first prepared. In a second translation, unmodified or mutant miniPLAP mRNA was used such that [(35)S]-labeled native or variant miniPLAP nascent protein was introduced. Following this, the RM were solubilized and anti-PLAP or anti-epitope immunoprecipitates were analyzed. With transamidase competent HeLa cell RM, anti-PLAP or anti-epitope antibody coprecipitated both Gaa1p and Gpi8p consistent with the assembly of the proprotein into a Gaa1p:Gpi8p-containing complex. When RM from K562 mutant K cells which lack Gpi8p were used, anti-PLAP antibody coprecipitated Gaa1p. The proprotein coprecipitation of Gaa1p increased with a nonpermissive GPI anchor addition (omega) site. In contrast, if a miniPLAP mutant devoid of its C-terminal signal was used, no coprecipitation occurred. During the transamidation reaction, a transient high Mr band forms. To definitively characterize this product, RM from K cells transfected with FLAG-tagged GPI8 were employed. Western blots of anti-FLAG bead isolates of solubilized RM from the cells showed that the high Mr band corresponded to Gpi8p covalently bound to miniPLAP. Loss of the band following hydrazinolysis demonstrated that the two components were associated in a thioester linkage. The data indicate that recognition of the proprotein involves Gaa1p, that the interaction with the complex does not depend on a permissive omega site, and that Gpi8p forms a thioester intermediate with the proprotein. The method could be useful for rapid analysis of nascent protein interactions with transamidase components, and possibly for helping to prepare a functional in vitro transamidase system.     

Studies on the phi X174 gene A protein-mediated termination of leading strand DNA synthesis

Recombinant RF (replicate form) I DNAs containing the bacteriophage phi X174 gene A protein-recognition sequence are cleaved by the phi X A protein yielding a phi X RF II X A protein complex (Zipursky, S.L., Reinberg, D., and Hurwitz, J. (1980) Proc. Natl. Acad. Sci. U.S.A. 77, 5182-5186). Such complexes support DNA synthesis in both RF I leads to SS(c) and RF I leads to RF I phi X DNA replication reactions in vitro. Two phi X A protein-recognition sequences were inserted into plasmid pBR322. Both sequences were contiguous with the same strand of the vector DNA and separated by 667 and 4275 base pairs. This recombinant plasmid (G27-4) was cleaved by the phi X A protein at either insert and both inserts support the initiation of RF leads to SS(c) DNA synthesis. This was verified by the finding that replication products were circular molecules of 667 and 4275 nucleotides. This finding is in keeping with the multifunctional activities associated with the phi X A protein; these include the site-specific nicking of RF I DNA which initiates DNA synthesis and site-specific termination resulting in the circularization of the displaced DNA strand. The phi X A protein and the Escherichia coli rep and SSb proteins catalyze the unwinding of phi X RF I DNA in vitro (Scott, J.F., Eisenberg, S., Bertsch, L.L., and Kornberg, A. (1977) Proc. Natl. Acad. Sci. U.S.A. 74, 193-197). Recombinant plasmid G27-4 RF I DNA was also unwound in vitro by this enzyme system; in this case, both circular and linear single-stranded DNA molecules of 667 and 4275 nucleotides, as well as full length circular single-stranded DNA were formed. Full length linear DNA was not detected. The two single-stranded circular DNA products formed as leading strands in RF leads to SS(c) reaction mixtures containing G27-4 RF I DNA differed in their ability to support lagging strand DNA synthesis. It was shown that the large single-stranded circular product included DNA sequences homologous to a replication factor Y effector sequence required for RF leads to RF and SS(c) leads to RF replication (Zipursky, S.L., and Marians, K.J. (1980) Proc. Natl. Acad. Sci. U.S.A. 77, 6521-6525). The 4275-nucleotide, but not the 667-nucleotide, single-stranded circular DNA product was converted to a duplex structure.     

Structure of human epoxide hydrolase reveals mechanistic inferences on bifunctional catalysis in epoxide and phosphate ester hydrolysis

The X-ray crystal structure of human soluble epoxide hydrolase (sEH) has been determined at 2.6 A resolution, revealing a domain-swapped quaternary structure identical to that observed for the murine enzyme [Argiriadi, M. A., Morisseau, C., Hammock, B. D., and Christianson, D. W. (1999) Proc. Natl. Acad. Sci. U.S.A. 96, 10637-10642]. As with the murine enzyme, the epoxide hydrolytic mechanism of the human enzyme proceeds through an alkyl-enzyme intermediate with Asp-333 in the C-terminal domain. The structure of the human sEH complex with N-cyclohexyl-N'-(iodophenyl)urea (CIU) has been determined at 2.35 A resolution. Tyr-381 and Tyr-465 donate hydrogen bonds to the alkylurea carbonyl group of CIU, consistent with the proposed roles of these residues as proton donors in the first step of catalysis. The N-terminal domain of mammalian sEH contains a 15 A deep cleft, but its biological function is unclear. Recent experiments demonstrate that the N-terminal domain of human sEH catalyzes the metal-dependent hydrolysis of phosphate esters [Cronin, A., Mowbray, S., Dürk, H., Homburg, S., Fleming, I., Fisslthaler, B., Oesch, F., and Arand, M. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 1552-1557; Newman, J. W., Morisseau, C., Harris, T. R., and Hammock, B. D. (2003) Proc. Natl. Acad. Sci. U.S.A. 100, 1558-1563]. The binding of Mg(2+)-HPO4(2-) to the N-terminal domain of human sEH in its CIU complex reveals structural features relevant to those of the enzyme-substrate complex in the phosphatase reaction.     

Complete amino acid sequence of canine cardiac calsequestrin deduced by cDNA cloning

cDNA cloning was used to deduce the complete amino acid sequence of canine cardiac calsequestrin, the principal Ca2+-binding protein of cardiac junctional sarcoplasmic reticulum. Cardiac calsequestrin contains 391 amino acid residues plus a 19-residue amino-terminal signal sequence. The molecular weight of the mature protein, excluding carbohydrate, is 45,269. Cardiac calsequestrin is highly acidic, and a striking feature is the enrichment of acidic residues (60%) within the 63 carboxyl-terminal residues. No part of the sequence contains EF hand Ca2+-binding structures. The photo-affinity probe 3-(trifluoromethyl)-3-(m-[125I]iodophenyl)diazirine was used to localize the Ca2+-regulated hydrophobic site to amino acid residues 192-223. The cardiac and skeletal muscle isoforms of calsequestrin (Fliegel, L., Ohnishi, M., Carpenter, M. R., Khanna, V. K., Reithmeier, R. A. F., and MacLennan, D. H. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 1167-1171), although the products of different genes, are 65% identical, are acidic, and share one glycosylation site. However, cardiac calsequestrin has several unique features. First, it has a 31-amino acid extension at its carboxyl terminus (residues 361-391), which contains 71% acidic residues and a second glycosylation site. Second, its mRNA contains a second open reading frame with the capacity to code for a 111-amino acid protein. Third, contrary to the restricted expression of the fast skeletal isoform, cardiac calsequestrin mRNA is present in both cardiac and slow skeletal muscle, but not in fast skeletal muscle. We conclude that the deduced amino acid sequence of cardiac calsequestrin is consistent with its ability to bind large amounts of Ca2+ (40 mol of Ca2+/mol of calsequestrin). The protein probably binds Ca2+ by acting as a charged surface rather than by presenting multiple discrete Ca2+-binding sites.     

Expression of rat liver S-adenosylhomocysteinase cDNA in Escherichia coli and mutagenesis at the putative NAD binding site

The cDNA for rat liver S-adenosylhomocysteinase has been cloned, and the nucleic acid sequence has been determined. By comparison of the deduced amino acid sequence for S-adenosylhomocysteinase with that of the dinucleotide binding region for other proteins, the sequence from amino acids 213 to 244 in rat liver S-adenosylhomocysteinase was proposed to be part of the NAD binding site (Ogawa, H., Gomi, T., Mueckler, M. M., Fujioka, M., Backlund, P. S., Jr., Aksamit, R. R., Unson, C. G., and Cantoni, G. L. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 719-723). A vector has been constructed that expresses S-adenosylhomocysteinase in Escherichia coli in the presence of isopropyl beta-D-thiogalactopyranoside by inserting the coding sequence of rat liver S-adenosylhomocysteinase cDNA downstream from the lac promoter of plasmid pUC118. The enzyme that is produced comprises as much as 10% of the soluble cellular proteins. The purified enzyme is a tetramer, contains 4 mol of tightly bound NAD, and has kinetic properties indistinguishable from those of the liver enzyme. Tryptic peptide mapping and NH2-terminal sequence analysis indicate that the recombinant enzyme is structurally identical to the liver enzyme except for the absence of the NH2-terminal blocking group. The rat liver enzyme has a blocked NH2-terminal alanine residue (Ogawa, H., Gomi, T., Mueckler, M. M., Fujioka, M., Backlund, P. S., Jr., Aksamit, R. R., Unson, C. G., and Cantoni, G. L. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 719-723). By oligonucleotide-directed mutagenesis mutant vectors have been generated that express proteins in which each of the glycines in the Gly-Xaa-Gly-Xaa-Xaa-Gly sequence of the putative NAD binding site (residues 219-224) was changed to valine. Immunoblot analysis of extracts of the cells transformed with these vectors reveals the presence of immunoreactive proteins with the subunit molecular weight of S-adenosylhomocysteinase. The mutant proteins have no catalytic activity, contain no bound NAD, and do not form the same quaternary structure as the recombinant S-adenosylhomocysteinase.     

Blocking of the initiation of protein biosynthesis by a pentanucleotide complementary to the 3' end of Escherichia coli 16 S rRNA.

Hydrogen bonding between the 3' terminus of 16 S rRNA (... C-A-C-C-U-C-C-U-U-A-OH3) and complementary sequences within the initiator region of mRNA may be a crucial event in the specific initiation of protein biosynthesis (Shine, J., and Dalgarno, L. (1974) Proc. Natl. Acad. Sci. U. S. A. 71, 1342-1346; Steitz, J. A., and Jakes, K. (1975) Proc. Natl. Acad. Sci. U. S. A. 72, 4734-4738). Using equilibrium dialysis, we have studied the binding of G-A-dG-dG-U (which is complementary to the 3' end of 16 S rRNA and which has been synthesized enzymatically) to initiation factor-free Escherichia coli ribosomes. We have also investigated the effects of the pentanucleotide on initiation reactions in E. coli ribosomes. G-A-dG-dG-U has a specific binding site on the 30 S ribosome with an association constant of 2 x 10(6) M-1 at 0 degrees C. G-A-dG-dG-U inhibits the R17 mRNA-dependent binding of fMet-tRNA by about 70%, both with 70 S ribosomes and 30 S subunits. In contrast, the A-U-G-dependent initiation reaction and the poly(U)-dependent Phe-tRNA binding was not affected by the pentanucleotide with both ribosomal species.     

Cell adhesion or integrin clustering increases phosphorylation of a focal adhesion-associated tyrosine kinase.

We have recently shown that changes in tyrosine phosphorylation of a 130-kDa protein(s) (pp130) may be involved in integrin signaling (Kornberg, L., Earp, H.S., Turner, C., Prokop, and Juliano, R. L. (1991) Proc. Natl. Acad. Sci. U.S.A. 88, 8392-8396). One component of the pp130 protein complex reacts with an antibody generated against p125fak, which is a focal contact-associated tyrosine kinase (Schaller, M.D., Borgman, C. A., Cobb, B. S., Vines, R. R., Reynolds, A. B., and Parsons, J. T. (1992) Proc. Natl. Acad. Sci. U.S.A. 89, 5192-5196). Both antibody-mediated integrin clustering and adhesion of KB cells to fibronectin leads to increased tyrosine phosphorylation of p125fak. The phosphorylation of p125fak is coincident with adhesion of cells to fibronectin and is maximal prior to cell spreading. Tyrosine phosphorylation of p125fak is induced when KB cells are allowed to adhere to fibronectin, collagen type IV, or laminin, but is not induced on polylysine. When KB cells are subjected to indirect immunofluorescence microscopy, p125fak colocalizes with talin in focal contacts. These data provide additional evidence that tyrosine kinases are involved in integrin signaling.     

Signal sequence of alkaline phosphatase of Escherichia coli.

The amino acid sequence of the signal sequence of phoA was determined by DNA sequencing by using the dideoxy chain termination technique (Sanger et al., Proc. Natl. Acad. Sci. U.S.A. 74:5463-5467, 1977). The template used was single-stranded DNA obtained from M13 on f1 phage derivatives carrying phoA, constructed by in vitro recombination. The results confirm the sequence of the first five amino acids determined by Sarthy et al. (J. Bacteriol. 139:932-939, 1979) and extend the sequence in the same reading frame into the amino terminal region of the mature alkaline phosphatase (Bradshaw et al., Proc. Natl. Acad. Sci. U.S.A., 78:3473-3477, 1981). As was predicted (Inouye and Beckwith, Proc. Natl. Acad. Sci. U.S.A. 74:1440-1444, 1977), the signal sequence was highly hydrophobic. The alteration of DNA sequence was identified for a promoter mutation that results in the expression of phoA independent of the positive control gene phoB and in insensitivity to high phosphate.     

Characterization of an Escherichia coli mdoB mutant strain unable to transfer sn-1-phosphoglycerol to membrane-derived oligosaccharides

A procedure for the isolation of mutants affected in components containing glycerol derived from phospholipids yielded two mutant strains that contain membrane-derived oligosaccharides (MDO) devoid of glycerol (Rotering, H., Fiedler, W., Rollinger, W., and Braun, V. (1984) FEMS Microbiol. Lett. 22, 61-68). MDO are found in the periplasmic space of Escherichia coli and other Gram-negative bacteria, and they may comprise up to 7% of the cells dry weight. The biosynthesis of MDO is osmoregulated (Kennedy, E. P. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 1092-1095) and linked to the metabolism of phospholipids (van Golde, L. M. G., Schulman, H., and Kennedy, E. P. (1973) Proc. Natl. Acad. Sci. U. S. A. 70, 1368-1372). This leads to substitution of MDO with sn-1-phosphoglycerol and phosphoethanolamine (Kennedy, E. P., Rumley, M. K., Schulman, P., and van Golde, L. M. G. (1976) J. Biol. Chem. 251, 4208-4213). MDO also contain succinate in O-ester linkage. We now report that one mutant strain lacks phosphoglycerol transferase I activity and thus is unable to transfer sn-1-phosphoglycerol residues from phosphatidylglycerol to MDO. The mdoB gene affected in this mutant has been located at 99.2 min on the E. coli chromosome. The ethanolamine content of MDO isolated from the mutant strain is elevated, whereas the number of succinate residues is not affected. The only phenotype of mdoB mutants we found is a dramatic reduction of the diglyceride content observed in dgk mdoB double mutants when the beta-glucoside arbutin is present in the growth medium.     

Light-induced interactions between rhodopsin and the GTP-binding protein. Relation with phosphodiesterase activation

The existence of rapid light-induced changes of light scattering in suspensions of bovine rod outer segment membranes has been described previously [H. Kühn et al. (1981) Proc. Natl Acad. Sci. USA, 78, 6873-6877]. The signal observed in the presence of GTP has been interpreted as being related to the rhodopsin-catalyzed exchange of GTP for GDP bound to the GTP-binding protein, i.e. to the formation of the activator of the cGMP phosphodiesterase [B.K.K. Fung et al. (1981) Proc. Natl Acad. Sci. USA, 78, 152-156]. We have tested this interpretation in the present paper by investigating the relation between the light-scattering signal and the activity of the phosphodiesterase using rapid recording techniques for both processes. All the results obtained are consistent with the above hypothesis. The amplitude of the light-scattering signal and the activity of the phosphodiesterase are shown to present the same dependence upon the flash intensity and upon the concentration of GTP or its analog guanosine 5'-[beta, gamma--imido]triphosphate (p[NH]ppG). The results suggest that the GTP-binding protein possesses one high-affinity p[NH]ppG-binding site (Kd much less than 0.1 microM). At high concentrations of GTP or p[NH]ppG the phosphodiesterase is activated in the dark and the light-scattering signal is correspondingly reduced; both effects are prevented by previous incubation with guanosine 5'-[beta-thio]diphosphate (p[S]pG).     

Unusual COOH-terminal structure of staphylococcal protease

The extracellular enzyme, staphylococcal protease, carries a COOH-terminal tryptic peptide of 43 amino acid residues most of which are aspartic acid, asparagine, and proline. This peptide might have a function equivalent to that of a similar segment previously observed at the NH2-terminal end of the membrane-bound penicillinase precursor of Bacillus licheniformis (Yamamoto, S., and Lampen, J. O. (1976) Proc. Natl. Acad. Sci. U. S. A. 73, 1457-1461). These observations would suggest that bacterial exoproteins which are secreted in the form of precursors differ from extracellular proteins by the presence of an extra segment at their NH2- and/or COOH-terminal ends.     

Identification of the in vivo and in vitro phosphorylation sites of rat liver fructose 1,6-bisphosphatase

Rat liver fructose 1,6-bisphosphatase appears to be unique in that it extends 24-26 residues beyond the COOH-terminal amino acid of other mammalian fructose 1,6-bisphosphatases and this extension contains phosphorylation sites. Using as a frame of reference the 335-residue sequence of pig kidney fructose 1,6-bisphosphatase (Marcus, F., Edelstein, I., Reardon, I., and Heinrikson, R. L. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 7161-7165), the rat liver enzyme would extend to residue 361. Limited proteolysis in the COOH-terminal region of the molecule with chymotrypsin, trypsin, or both sequentially, led us to establish that the phosphorylation sites are located at Ser residues 341 and 356. The in vitro phosphorylation of purified rat liver fructose 1,6-bisphosphatase by the catalytic subunit of cyclic AMP-dependent protein kinase results in modification at both residues, although the major site of phosphorylation (61%) is at Ser-341. In contrast, rat liver fructose 1,6-bisphosphatase purified from animals that had been injected with [32P] phosphate contains most of the label (81%) at Ser-356.     

ATP-driven exchange of Na+ and K+ ions by Streptococcus faecalis

We describe the characterization of KtrII, a novel potassium transport system of Streptococcus faecalis, first discovered by H. Kobayashi [1982) J. Bacteriol. 150, 506-511). KtrII requires sodium ions and mediates the stoichiometric exchange of internal Na+ for external K+. Potassium accumulation is not energized by the electrochemical potentials of either H+ or Na+; the energy source is probably ATP. Two lines of evidence indicate that KtrII is a manifestation of the sodium-stimulated ATPase reported earlier (Heefner, D. L., and Harold, F. M. (1982) Proc. Natl. Acad. Sci. U.S.A. 79, 2798-2802). (i) Mutants that lack the ATPase also lack KtrII, and revertants recover both in parallel. (ii) KtrII and the Na+-ATPase are induced in parallel when cells are grown on media rich in sodium, particularly under conditions that limit the generation of a proton potential. KtrII is not induced in response to K+ deprivation. We propose that the Na+-ATPase exchanges Na+ for K+ ions.