Cloning of trg, a gene for a sensory transducer in Escherichia coli.

Clones of trg, a gene which codes for a chemotactic transducer, were isolated linked to ColE1 and pBR322 vectors. Studies with the hybrid plasmids demonstrated unequivocally that trg is the structural gene for methyl-accepting chemotaxis protein III. The Trg protein was found to be structurally complex, electrophoresing as a series of seven bands on high-resolution sodium dodecyl sulfate-polyacrylamide gels. The multiplicity of bands is a function of the activity of cheR, which codes for a methyltransferase, and of cheB, which codes for a demethylase. It appears that Trg, a quantitatively minor transducer, resembles the two major transducer proteins, Tsr and Tar, in that all three are multiply methylated and also multiply modified in a second way which requires an active cheB gene. However, preliminary analysis of the Trg protein indicated that it is significantly less related structurally to the Tsr or Tar protein than those two transducers are to each other. This implies that the features of multiple methylation and cheB-dependent modification are likely to be critical for the common physiological functions in chemotactic excitation and adaptation performed by all three transducers.     

Limited homology between trg and the other transducer proteins of Escherichia coli.

Transducers are transmembrane proteins that are central to the chemotactic system of Escherichia coli. The proteins transduce ligand recognition into an excitatory signal and function in adaptation as methyl-accepting proteins. The transducer genes tsr, tar, and tap have extensive homology with each other. However, previous studies revealed little indication of homology between those three transducer genes and a fourth gene, trg. We investigated the relationship between trg and the other genes by blot-hybridization experiments and the relationship between Trg and the other transducer proteins by immune precipitation and experiments with an antiserum raised to purified Trg protein. In experiments in which 35% mismatch would be tolerated, weak hybridization of trg was detected to a DNA fragment containing tar and tap but not to a fragment containing tsr. In experiments in which only 30% mismatch would be tolerated, no trg hybridization was apparent either to total chromosomal DNA or to DNA from hybrid plasmids carrying the other transducer genes. An anti-Trg serum formed immune precipitates with the Tsr and Tar proteins as well as with the Trg protein to which it was raised. We conclude that there is homology between Trg and the other transducer, but the homology is more limited than that shared among the other transducers. Furthermore, we found no indication of additional transducer genes closely related to trg. Thus, the trg gene is a somewhat distant cousin within a single transducer gene family of E. coli.     

High-level synthesis in Escherichia coli of the SV40 small-t antigen under control of the bacteriophage lambda pL promoter

Several plasmids were constructed in which the SV40 small-t antigen gene was inserted in close proximity downstream from the thermoinducible leftward promoter (pL) of bacteriophage lambda. Upon temperature induction the best of our constructions expressed a small-t-related 19 000-dalton polypeptide in an amount corresponding to approx. 2.5% of total de novo protein synthesis. This 19 000-dalton protein was identified as small-t by specific immunoprecipitation with anti-T serum and by two-dimensional fingerprint analysis. In addition to the 19 000-dalton product, representative plasmids expressed fairly large amounts (up to 7% of total de novo protein synthesis) of a protein with an apparent Mr of 14 500. This 14 500-dalton polypeptide was shown to be related to authentic small-t. Presumably the secondary structure of the mRNA starting at pL is such that translation initiation at an internal AUG codon of the small-t gene is favored over initiation at the true initiating codon.     

Effects of glutamines and glutamates at sites of covalent modification of a methyl-accepting transducer.

Chemotactic transducer proteins of Escherichia coli contain four or five methyl-accepting glutamates that are crucial for sensory adaptation and gradient sensing. Two residues arise from posttranslational deamidation of glutamines to yield methyl-accepting glutamates. We addressed the significance of this arrangement by creating two mutated trg genes: trg(5E), coding for a transducer in which all five modification sites were synthesized as glutamates, and trg(5Q), in which all five were glutamines. We found that the normal (3E,2Q) configuration was not an absolute requirement for synthesis, assembly, or stable maintenance of transducers. Both mutant proteins were methylated, although Trg(5Q) had a reduced number of methyl-accepting sites because two glutamines at adjacent residues were blocked for deamidation and thus could not become methyl-accepting glutamates. The glutamine-glutamate balance had striking effects on signaling state. Trg(5E) was in a strong counterclockwise signaling configuration, and Trg(5Q) was in a strong clockwise signaling induced by ligand binding, and alanines substituted at modification sites had an intermediate effect. Chemotactic migration by growing cells containing trg(5E) or trg(5Q) exhibited reduced effectiveness, probably reflecting perturbations of the counterclockwise/clockwise ratio caused by newly synthesized transducers not modified rapidly enough to produce a balanced signaling state during growth. These defects were evident for cells in which other transducers were not available to contribute to balanced signaling or were present at lower levels than the mutant proteins.     

Polypeptides encoded by the mer operon.

HgCl2-induced polypeptides synthesized by Escherichia coli minicells containing recombinant or natural HgR plasmids were labeled with [35S]methionine and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. All plasmids examined encoded two heavily labeled, HgCl2-inducible polypeptides of 69,000 and 12,000 daltons. Most plasmids also encoded two additional HgCl2-inducible proteins in the 14,000- to 17,000-dalton range. Antiserum prepared against a purified mercuric ion reductase reacts with the 69,000-dalton polypeptide and a minor 66,000-dalton protein seen in several different HgR minicells. Recombinant plasmids constructed from portions of mer DNA from the IncFII plasmid NR1 were also analyzed in the minicell system. Five HgCl2-inducible polypeptides (69,000, 66,000, 15,100, 14,000, and 12,000 daltons) were synthesized in minicells carrying pRR130, a recombinant derivative containing the EcoRI-H and EcoRI-I restriction fragments of NR1. The EcoRI-H fragment of NR1 encodes the three small mer proteins of 15,100, 14,000, and 12,000 daltons and the amino-terminal 40,000 daltons of the mercuric ion reductase monomer.     

Hepatic lipase. Synthesis, processing, and secretion by isolated rat hepatocytes

Hepatic lipase, a glycoprotein synthesized and secreted by the hepatocyte, binds to sinusoidal endothelium where it is involved in metabolism of lipoprotein phospholipid and triglyceride. To better understand the regulation of hepatic lipase, we investigated the synthesis, post-translational processing, and secretion of the enzyme by isolated rat hepatocytes. Metabolically labeled [35S]methionine hepatic lipase protein, produced by the collagenase-dispersed hepatocytes, was immunoisolated from detergent-solubilized cells and incubation medium at designated times, using a polyclonal rabbit anti-rat hepatic lipase antibody raised against hepatic lipase purified to homogeneity from rat liver post-heparin perfusates. Following polyacrylamide gel electrophoresis and fluorography, radiolabeled hepatic lipase was quantitated by densitometry. Newly synthesized hepatic lipase was rapidly secreted and accumulated in the medium as a 59,000-dalton protein in a manner consistent with a constitutive process. An intracellular 53,000-dalton precursor of the mature 59,000-dalton hepatic lipase was identified by immunoprecipitation. The 53,000-dalton form could also be generated by endoglycosidase digestion of the secreted 59,000-dalton protein. In pulse-chase experiments, the 53,000-dalton protein was converted into the 59,000-dalton form. A 47,000-dalton form of hepatic lipase was immunoisolated from cell lysates only after tunicamycin treatment and could be generated from the secreted 59,000-dalton enzyme by prolonged endoglycosidase digestion. These data show that hepatic lipase is synthesized and rapidly secreted by isolated rat hepatocytes. Further, an intracellular 47,000-dalton precursor peptide can be identified after tunicamycin treatment, which may represent the hepatic lipase polypeptide, presumably after removal of its signal sequence; a 53,000-dalton partially glycosylated peptide exists as a major precursor form in the cell; and the mature 59,000-dalton hepatic lipase is present in the hepatocyte, but it is rapidly secreted.     

Multiple covalent modifications of Trg, a sensory transducer of Escherichia coli

The sensory transducers of Escherichia coli are integral membrane proteins that mediate the tactic response of cells to chemical stimuli. Adaptation to environmental stimuli is correlated with methylation of the transducer proteins. Two transducer genes, tsr and tar, exhibit extensive homologies while no homology has been detected between a third transducer, trg, and those genes. The Tsr and Tar proteins have been shown to contain multiple sites for methylation as well as two sites for another modification that requires an active cheB gene product and is designated the CheB-dependent modification. In this study, covalent modifications of the Trg protein were characterized by analysis of tryptic peptides. We found that methylation occurred at several sites on the Trg protein and that the protein contained at least three sites for CheB-dependent modification, two of which were located on a tryptic peptide that contains both methionine and lysine. This tryptic peptide is analogous to the methionine- and lysine-containing methyl-accepting peptides isolated from the Tsr and Tar proteins and like those peptides may contain several methyl-accepting sites. We estimated the pKa of the group created by the CheB-dependent modification on the methionine- and lysine-containing peptide of Trg to be between pH 2.2 and 5.8. This result supports the idea that the CheB-dependent modification is an enzymatic deamidation of glutamine to glutamic acid.     

Identification of a protein of 15,000 daltons related to isoleucine-valine biosynthesis in Escherichia coli K-12.

The effect of the ilvG671, ilvG468, and ilvG603 mutations (phenotype, IlvG+ Valr; formerly ilvO) upon proteins synthesized was determined by infection of irradiated Escherichia coli K-12 cells, using specifically constructed derivatives of lambda dilv phage. These ilvG alleles are similar to the previously studied ilvG2096(Valr) allele in that they activate the latent ilvG gene which is present in the wild-type strain, leading to the synthesis of a 62,000-dalton protein. In addition, all of these ilvG (Valr) alleles increase the synthesis of a 15,000-dalton protein. To localize the gene coding for the 15,000-dalton protein, the proteins produced in maxicells containing plasmids with specific deletions of ilv and rrnX DNA segments were analyzed. The gene coding for the 15,000-dalton protein was located within a region about 1,000 base pairs long between ilv and trpT. The function of the 15,000-dalton protein is not known.     

Molecular cloning and characterization of the genes (pbpA and rodA) responsible for the rod shape of Escherichia coli K-12: analysis of gene expression with transposon Tn5 mutagenesis and protein synthesis directed by constructed plasmids.

Two cell shape-determining genes of Escherichia coli K-12, pbpA, the structural gene for penicillin-binding protein 2, and rodA, whose protein is unknown, were subcloned into plasmid vectors from the transducing phage lambda MAd lip24, which carries the lip-leuS region of the E. coli chromosome. Plasmids with restriction enzyme-created deletions or transposon Tn5 insertions were isolated, and studies of genetic complementation of these plasmids with chromosomal mutations were carried out. Thus, a physical and genetic map of the rodA-pbpA region was established. The genes rodA and pbpA lie side by side within a 4.4-kilobase-pair region. The size of the rodA gene has been shown to be between 0.86 and 1.6 kilobase pairs; such DNA would encode a protein with a molecular weight between 32,000 and 59,000. Since Tn5 mutagenesis of the rodA gene did not affect the expression of the pbpA gene and vice versa, the genes rodA and pbpA seem to have independent promoters. Analysis of the proteins synthesized from the constructed plasmids in maxicells revealed that the plasmid carrying the pbpA gene encoded penicillin-binding protein 2 and amplification of the protein occurred. The product of the rodA gene was not identified.     

Mutations specifically affecting ligand interaction of the Trg chemosensory transducer.

The Trg transducer mediates chemotactic response to galactose and ribose by interacting, respectively, with sugar-occupied galactose- and ribose-binding proteins. Adaptation is linked to methylation of specific glutamyl residues of the Trg protein. This study characterized two trg mutations that affect interaction with binding protein ligands but do not affect methylation or adaptation. The mutant phenotypes indicated that the steady-state activity of methyl-accepting sites is independent of ligand-binding activity. The mutation trg-8 changed arginine 85 to histidine, and trg-19 changed glycine 151 to aspartate. The locations of the mutational changes provided direct evidence for functioning of the amino-terminal domain of Trg in ligand recognition. Cross-inhibition of tactic sensitivity by the two Trg-linked attractants implies competition for a common site on Trg. However, the single amino acid substitution caused by trg-19 greatly reduced the response to galactose but left unperturbed the response to ribose. Thus Trg must recognize the two sugar-binding proteins at nonidentical sites, and the complementary sites on the respective binding proteins should differ. trg-8 mutants were substantially defective in the response to both galactose and ribose. An increase in cellular content of Trg-8 protein improved the response to galactose but not to ribose. It appears that Trg-8 protein is defective in the generation of the putative conformational change induced by ligand interaction. The asymmetry of the mutational defect implies that functional separation of interaction sites could persist beyond the initial stage of ligand binding.     

A structural model of human erythrocyte protein 4.1

Limited proteolysis and specific chemical cleavage methods have enabled a detailed structural characterization of human erythrocyte protein 4.1. This protein is composed of two chemically very similar polypeptide chains (a and b) with apparent molecular masses of 80,000 and 78,000 daltons. Cleavage of protein 4.1 at cysteine residues by 2-nitro-5-thiocyanobenzoic acid produces a series of doublets which differ by approximately 2,000 daltons and have identical peptide maps. Alignment of these peptides by mapping analysis has localized 4 cysteine residues within a 17,000-dalton segment on both a and b polypeptides. Mild chymotryptic treatment at 0 degrees C cleaves protein 4.1 primarily in three central locations and generates two families of unrelated peptides. Analysis of these fragments in two-dimensional gels and by peptide mapping reveals an unusual polarity in protein 4.1 structure in that each polypeptide chain contains two segments, one relatively acidic the other basic, that are segregated at opposite ends of the molecule. The basic region is digested into a cysteine-rich 30,000-dalton domain which resists further breakdown while the acidic region is readily degraded into smaller fragments. The peptides derived from the acidic region all appear as doublets suggesting that protein 4.1 a and b polypeptides differ close to the terminus of the acidic end. Similar phosphorylation sites occur on both polypeptides within a segment some 24,000-34,000 daltons from the acidic terminus.     

Expression and mutagenesis of thrombospondin.

Thrombospondin is a 420,000-dalton adhesive glycoprotein that is composed of three subunits of equivalent molecular weight. When the cDNA for the complete coding region of the human endothelial cell thrombospondin subunit is expressed in mouse NIH 3T3 cells, a 420,000-dalton protein is synthesized and secreted. The expressed protein comigrates with human platelet thrombospondin both in the presence and in the absence of a reducing agent. The expressed protein binds to a monoclonal anti-thrombospondin antibody, heparin, and calcium. In addition to the 420,000-dalton protein, the transfected cell lines also express a variable amount of a 140,000-dalton polypeptide. When the culture supernatants that are produced by cells that are expressing thrombospondin are applied to heparin-Sepharose, the 420,000-dalton and the 140,000-dalton proteins are bound to the column and are eluted with buffer containing 0.55 and 0.3 M NaCl, respectively. The 140,000-dalton protein only binds to heparin-Sepharose in the presence of calcium. Deletion of the region of homology with procollagen results in defective assembly of the trimer. Deletion of the type 1 or type 2 repeats results in decreased stability of the subunit with the predominant polypeptides that are expressed having molecular weights of 127,000 and 130,000, respectively. These polypeptides retain low-affinity heparin-binding activity. High-affinity heparin binding is markedly diminished by mutations in either of two sequence motifs that include clusters of lysines and arginines.(ABSTRACT TRUNCATED AT 250 WORDS)     

Purification of receptor protein Trg by exploiting a property common to chemotactic transducers of Escherichia coli

The methyl-accepting chemotactic transducers of Escherichia coli were found to bind strongly to Cibacron blue-Sepharose. Among potential elutants tested, only S-adenosylmethionine at moderate concentrations and NaCl at concentrations greater than 1.5 M caused dissociation of these detergent-solubilized transmembrane proteins from the dye. Release by S-adenosylmethionine may be a generalized effect rather than the result of a specific binding site for that compound on transducers. A truncated trg gene was created that coded for the carboxyl-terminal three-fifths of the transducer, which constitutes the cytoplasmic domain common to all four transducers in E. coli. This domain bound to Cibacron blue-Sepharose and was eluted in a pattern similar to that exhibited by intact Trg, indicating that interaction with the dye occurred in this conserved domain. Adherence to Cibacron blue and elution by high salt formed the core of an efficient purification scheme, developed for Trg but applicable to all transducers in E. coli and perhaps to methyl-accepting chemotaxis proteins in other species. Determination of the amino acid sequence at the beginning of purified Trg confirmed that it contained a longer hydrophilic segment at its amino terminus than other transducers of E. coli. The initial methionine of Trg is neither cleaved nor modified, in contrast to the Tar transducer in which the amino terminus was found previously to be blocked. Circular dichroic measurements of purified Trg indicated that the secondary structural organization of the protein is predominantly alpha-helix.     

Isolation of the GSY1 gene encoding yeast glycogen synthase and evidence for the existence of a second gene

Glycogen synthase preparations from Saccharomyces cerevisiae contained two polypeptides of molecular weights 85,000 and 77,000. Oligonucleotides based on protein sequence were utilized to clone a S. cerevisiae glycogen synthase gene, GSY1. The gene would encode a protein of 707 residues, molecular mass 80,501 daltons, with 50% overall identity to mammalian muscle glycogen synthases. The amino-terminal sequence obtained from the 85,000-dalton species matched the NH2 terminus predicted by the GSY1 sequence. Disruption of the GSY1 gene resulted in a viable haploid with glycogen synthase activity, and purification of glycogen synthase from this mutant strain resulted in an enzyme that contained the 77,000-dalton polypeptide. Southern hybridization of genomic DNA using the GSY1 coding sequence as a probe revealed a second weakly hybridizing fragment, present also in the strain with the GSY1 gene disrupted. However, the sequences of several tryptic peptides derived from the 77,000-dalton polypeptide were identical or similar to the sequence predicted by the GSY1 gene. The data are explained if S. cerevisiae has two glycogen synthase genes encoding proteins with significant sequence similarity The protein sequence predicted by the GSY1 gene lacks the extreme NH2-terminal phosphorylation sites of the mammalian enzymes. The COOH-terminal phosphorylated region of the mammalian enzyme over-all displayed low identity to the yeast COOH terminus, but there was homology in the region of the mammalian phosphorylation sites 3 and 4. Three potential cyclic AMP-dependent protein kinase sites are located in this region of the yeast enzyme. The region of glycogen synthase likely to be involved in covalent regulation are thus more variable than the catalytic center of the molecule.     

Comparison of the major polypeptides of the erythrocyte nuclear envelope

The three most abundant nonhistone polypeptides (molecular weights 75,000, 71,000 and 61,000) of the avian erythrocyte nucleus have previously been isolated in the nuclear envelope fraction. They have been separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis and peptide-mapped after limited enzymatic digestion. Three enzymes–chymotrypsin, papain and Staphylococcus aureus protease–were used. Results obtained with each enzyme indicate strong similarities between the three nuclear envelope polypeptides. The amino acid compositions of the two most abundant polypeptides (P75 and P71) have been determined and found to be similar. Further, they readily yield large fragments upon brief alkaline hydrolysis. For both P75 and P71 the degree and the pattern of alkaline fragmentation are almost identical. A 61,000-dalton polypeptide which appears to be P61 is obtained from P75 and P71 by mild acid hydrolysis. These results establish the close chemical similarity of these predominant polypeptides in the erythrocyte nucleus and suggest that they serve related functions.     

Characterization of an outer membrane mannanase from Bacteroides ovatus

Bacteroides ovatus utilizes guar gum, a high-molecular-weight branched galactomannanan, as a sole source of carbohydrate. No extracellular activity was detectable. Approximately 30% of the total cell-associated mannanase activity partitioned with cell membranes. When inner and outer membranes of B. ovatus were separated on sucrose gradients, the mannanase activity was associated mainly with fractions containing outer membranes. Enzyme activity was solubilized by 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) or by Triton X-100 at a detergent-to-protein ratio of 1:1. The enzyme was stable for only 4 h at 37 degrees C and for 50 to 60 h at 4 degrees C. Analysis of the products of the CHAPS-solubilized mannanase on Bio-Gel A-5M and Bio-Gel P-10 gel filtration columns indicated that the enzyme breaks guar gum into high-molecular-weight fragments. The CHAPS-solubilized mannanase was partially purified by chromatography on a FPLC Mono Q column. The partially purified mannanase preparation contained three major polypeptides (Mr 94,500, 61,000, and 43,000) and several minor ones. High mannanase activity was seen only when B. ovatus was grown on guar gum. Cross-absorbed antiserum detected two other guar gum-associated outer membrane proteins: a CHAPS-extractable 49,000-dalton polypeptide and a 120,000-dalton polypeptide that was not solubilized by CHAPS. Neither of these polypeptides was detectable in the partially purified mannanase preparation. These results indicate that there are at least two guar gum-associated outer membrane polypeptides other than the mannanase.     

Identification of polypeptides encoded by cloned pJM1 iron uptake DNA isolated from Vibrio anguillarum 775.

The XhoI fragment containing much of the iron uptake region of plasmid pJM1 was isolated from Vibrio anguillarum 775 and cloned into plasmid pBR322. Plasmid-encoded polypeptides were examined in maxicells of Escherichia coli, and transposon mutagenesis was used to map insertion mutations in the structural DNA encoding the OM2 polypeptide. Tn1000 insertions that mapped within OM2 and blocked maxicell expression of OM2 resulted in the loss of ferric iron-anguibactin receptor function when plasmids containing OM2:: Tn1000 insertions were introduced into V. anguillarum cells. Two iron-regulated polypeptides were identified in maxicell polypeptide profiles of E. coli SS201. A 20,000-dalton polypeptide was expressed in maxicells of SS201 grown under conditions of iron limitation but was barely detectable in profiles of SS201 cells that were grown under high-iron conditions. DNA encoding the 20,000-dalton polypeptide mapped downstream of and adjacent to the gene encoding OM2. DNA sequences required for production of a 46,000-dalton polypeptide mapped 4.5 kilobases downstream of the OM2 structural gene. The 46,000-dalton polypeptide was synthesized at high levels in E. coli SS201 maxicells grown under high-iron conditions, but synthesis of the protein was severely repressed under conditions of iron limitation. Iron-regulated expression of both proteins in maxicells of SS201 was relieved upon deletion of a 4.9-kilobase SalI-XhoI fragment of pJM1 DNA, which indicated that pJM1 DNA sequences present in the deleted fragment are required for regulated expression of both proteins in E. coli. Maxicells of SS201 harboring these deletion derivatives synthesized the 20,000-dalton polypeptide at very low constitutive levels and the 46,000-dalton polypeptide at high constitutive levels, regardless of the iron concentration of the growth medium. The observed regulation of the 20,000-dalton protein suggested that it might play a role either in siderophore biosynthesis or in the functional expression of OM2. The opposite regulatory pattern observed for the 46,000-dalton polypeptide suggested that it does not play a structural role in siderophore or OM2 biosynthesis, but the observed regulatory pattern might be expected if the 46,000-dalton protein played a negative regulatory role in siderophore biosynthesis.     

Proteolysis of smooth muscle myosin light chain kinase. Formation of inactive and calmodulin-independent fragments

Proteolysis by trypsin of gizzard myosin light chain kinase in the absence of Ca2+-calmodulin causes a biphasic effect on kinase activity. During the initial phase of proteolysis, Ca2+-calmodulin-dependent kinase activity is reduced over a thousand-fold. Further proteolysis, in the second phase, causes an increase in activity that is independent of Ca2+-calmodulin. Loss of activity is associated with the formation of a 64,000-dalton fragment. Calmodulin-independent activity is associated with the formation of a 61,000-dalton fragment. Procedures for the isolation of each fragment are outlined. Tryptic hydrolysis of the isolated 64,000-dalton peptide generates the 61,000-dalton peptide and increases calmodulin-independent activity. Km values for ATP and light chains for the native kinase and two fragments are the same, i.e. approximately 100 and 5 microM, respectively. Neither fragment binds to F-actin. Amino acid analyses of both fragments are given. Synthetic peptides corresponding to the calmodulin-binding regions of the smooth and skeletal muscle kinases are potent inhibitors of the 61,000-dalton fragment. These data demonstrate the existence of an inhibitory region that is suggested to be located between the active site and the calmodulin-binding site. Whether it is distinct from or at the N-terminal end of the calmodulin-binding site cannot be determined from these data.     

Unidirectional theta replication of the structurally stable Enterococcus faecalis plasmid pAM beta 1.

Numerous bacterial replicons remain poorly characterized due to difficulties in localization of the replication origin. We have circumvented this problem in the characterization and fine mapping of the origin of plasmid pAM beta 1 by exploiting the Bacillus subtilis termination signal, terC. In terC-containing derivatives, theta-form molecules with two invariant endpoints accumulate. The endpoints, which correspond to plasmid origin and terC, were mapped with single-nucleotide precision. Analysis of the replication intermediates of wild-type molecules by two-dimensional gel electrophoresis confirmed the location of the plasmid origin. Our results demonstrate that pAM beta 1 replication proceeds unidirectionally by a theta mechanism. This work confirms the use of termination signals to localize origins, suggests that termination in B. subtilis occurs by a mechanism similar to that of Escherichia coli and establishes that in addition to rolling circle replicating plasmids, Gram positive bacteria harbour plasmids which replicate by a theta mechanism.