An erythromycin-resistance gene from an erythromycin-producing strain of Arthrobacter sp

Genome Sequence of Aeromicrobium erythreum NRRL B-3381, an Erythromycin-Producing Bacterium of the Nocardioidaceae

ITALIC! Aeromicrobium erythreumNRRL B-3381 has a 3,629,239-bp circular genome that has 72% G+C content. There are at least 3,121 coding sequences (CDSs), two rRNA gene operons, and 47 tRNAs. The genome and erythromycin ( ITALIC! ery) biosynthetic gene sequences provide resources for metabolic and combinatorial engineering of polyketides.

Phylogenetic relatedness determined between antibiotic resistance and 16S rRNA genes in actinobacteria

BACKGROUND

Distribution and evolutionary history of resistance genes in environmental actinobacteria provide information on intensity of antibiosis and evolution of specific secondary metabolic pathways at a given site. To this day, actinobacteria producing biologically active compounds were isolated mostly from soil but only a limited range of soil environments were commonly sampled. Consequently, soil remains an unexplored environment in search for novel producers and related evolutionary questions.

RESULTS

Ninety actinobacteria strains isolated at contrasting soil sites were characterized phylogenetically by 16S rRNA gene, for presence of erm and ABC transporter resistance genes and antibiotic production. An analogous analysis was performed in silico with 246 and 31 strains from Integrated Microbial Genomes (JGI_IMG) database selected by the presence of ABC transporter genes and erm genes, respectively. In the isolates, distances of erm gene sequences were significantly correlated to phylogenetic distances based on 16S rRNA genes, while ABC transporter gene distances were not. The phylogenetic distance of isolates was significantly correlated to soil pH and organic matter content of isolation sites. In the analysis of JGI_IMG datasets the correlation between phylogeny of resistance genes and the strain phylogeny based on 16S rRNA genes or five housekeeping genes was observed for both the erm genes and ABC transporter genes in both actinobacteria and streptomycetes. However, in the analysis of sequences from genomes where both resistance genes occurred together the correlation was observed for both ABC transporter and erm genes in actinobacteria but in streptomycetes only in the erm gene.

CONCLUSIONS

The type of erm resistance gene sequences was influenced by linkage to 16S rRNA gene sequences and site characteristics. The phylogeny of ABC transporter gene was correlated to 16S rRNA genes mainly above the genus level. The results support the concept of new specific secondary metabolite scaffolds occurring more likely in taxonomically distant producers but suggest that the antibiotic selection of gene pools is also influenced by site conditions.

The erythromycin biosynthetic gene cluster of Aeromicrobium erythreum

The erythromycin-biosynthetic (ery) gene cluster of Aeromicrobium erythreum was cloned and characterized. The 55.4-kb cluster contains 25 ery genes. Homologues were found for each gene in the previously characterized ery gene cluster from Saccharopolyspora erythraea. In addition, four new predicted ery genes were identified. Two of the new predicted genes, coding for a phosphopantetheinyl transferase (eryP) and a type II thioesterase (eryTII), were internal to the ery cluster. The other two new genes, coding for a thymidine 5'-diphosphate-glucose synthase (eryDI) and a MarR-family transcriptional repressor (ery-ORF25), were found at the two ends of the ery cluster. A knockout in eryDI showed it to be essential for erythromycin biosynthesis. The gene order of the two ery clusters was conserved within a core region of 15 contiguous genes, with the exception of IS1136 which was not found in the A. erythreum cluster. Beyond the core region, gene shuffling had occurred between the two sides of the cluster. The flanking regions of the two ery clusters were not alike in the type of genes found.

Mutational analysis defines the roles of conserved amino acid residues in the predicted catalytic pocket of the rRNA:m6A methyltransferase ErmC'

Methyltransferases (MTases) from the Erm family catalyze S-adenosyl-L-methionine-dependent modification of a specific adenine residue in bacterial 23S rRNA, thereby conferring resistance to clinically important macrolide, lincosamide and streptogramin B antibiotics. Despite the available structural data and functional analyses on the level of the RNA substrate, still very little is known about the mechanism of rRNA:adenine-N(6) methylation. Only predictions regarding various aspects of this reaction have been made based on the analysis of the crystal structures of methyltransferase ErmC' (without the RNA) and their comparison with the crystallographic and biochemical data for better studied DNA:m(6)A MTases. To validate the structure-based predictions of presumably essential residues in the catalytic pocket of ErmC', we carried out the site-directed mutagenesis and studied the function of the mutants in vitro and in vivo. Our results indicate that the active site of rRNA:m(6)A MTases is much more tolerant to amino acid substitutions than the active site of DNA:m(6)A MTases. Only the Y104 residue implicated in stabilization of the target base was found to be indispensable. Remarkably, the N101 residue from the "catalytic" motif IV and two conserved residues that form the floor (F163) and one of the walls (N11) of the base-binding site are not essential for catalysis in ErmC'. This somewhat surprising result is discussed in the light of the available structural data and in the phylogenetic context of the Erm family.

Mechanisms of bacterial resistance to macrolide antibiotics

Macrolides have been used in the treatment of infectious diseases since the late 1950s. Since that time, a finding of antagonistic action between erythromycin and spiramycin in clinical isolates1 led to evidence of the biochemical mechanism and to the current understanding of inducible or constitutive resistance to macrolides mediated by erm genes containing, respectively, the functional regulation mechanism or constitutively mutated regulatory region. These resistant mechanisms to macrolides are recognized in clinically isolated bacteria. (1) A methylase encoded by the erm gene can transform an adenine residue at 2058 (Escherichia coli equivalent) position of 23S rRNA into an 6N, 6N-dimethyladenine. Position 2058 is known to reside either in peptidyltransferase or in the vicinity of the enzyme region of domain V. Dimethylation renders the ribosome resistant to macrolides (MLS). Moreover, another finding adduced as evidence is that a mutation in the domain plays an important role in MLS resistance: one of several mutations (transition and transversion) such as A2058G, A2058C or U, and A2059G, is usually associated with MLS resistance in a few genera of bacteria. (2) M (macrolide antibiotics)- and MS (macrolide and streptogramin type B antibiotics)- or PMS (partial macrolide and streptogramin type B antibiotics)-phenotype resistant bacteria cause decreased accumulation of macrolides, occasionally including streptogramin type B antibiotics. The decreased accumulation, probably via enhanced efflux, is usually inferred from two findings: (i) the extent of the accumulated drug in a resistant cell increases as much as that in a susceptible cell in the presence of an uncoupling agent such as carbonylcyanide-m-chlorophenylhydrazone (CCCP), 2,4-dinitrophenol (DNP), and arsenate; (ii) transporter proteins, in M-type resistants, have mutual similarity to the 12-transmembrane domain present in efflux protein driven by proton-motive force, and in MS- or PMS-type resistants, transporter proteins have mutual homology to one or two ATP-binding segments in efflux protein driven by ATP. (3) Two major macrolide mechanisms based on antibiotic inactivation are dealt with here: degradation due to hydrolysis of the macrolide lactone ring by an esterase encoded by the ere gene; and modification due to macrolide phosphorylation and lincosamide nucleotidylation mediated by the mph and lin genes, respectively. But enzymatic mechanisms that hydrolyze or modify macrolide and lincosamide antibiotics appear to be relatively rare in clinically isolated bacteria at present. (4) Important developments in macrolide antibiotics are briefly featured. On the basis of information obtained from extensive references and studies of resistance mechanisms to macrolide antibiotics, the mode of action of the drugs, as effectors, and a hypothetical explanation of the regulation of the mechanism with regard to induction of macrolide resistance are discussed.

Stereoselective production of (+)-trans-chrysanthemic acid by a microbial esterase: cloning, nucleotide sequence, and overexpression of the esterase gene of Arthrobacter globiformis in Escherichia coli

The gene coding for a novel esterase which stereoselectively hydrolyzes the (+)-trans (1R,3R) stereoisomer of ethyl chrysanthemate was cloned from Arthrobacter globiformis SC-6-98-28 and overexpressed in Escherichia coli. The cellular content of the active enzyme reached 33% of the total soluble protein in the recombinant E. coli JM105 cells and 5.6 g/liter of culture by high-density cell cultivation. The hydrolytic activity of the recombinant E. coli cells for ethyl chrysanthemate reached 605 mumol of chrysanthemic acid per min per g of dry cells, which is approximately 2,500-fold higher than that of A. globiformis cells. The stereoselective hydrolysis by the recombinant E. coli cells was efficient at substrate concentrations of up to 40% by removing the produced chrysanthemic acid by ultrafiltration. The (+)-trans-chrysanthemic acid produced had 100% optical purity. The amino acid sequence of the esterase was found to be similar to that of several class C beta-lactamases, D,D-carboxypeptidase, D-aminopeptidase, 6-aminohexanoate-dimer hydrolase, and Pseudomonas esterase. The sequence comparison also suggested that the Ser-X-X-Lys motif in the esterase was at the active site of the enzyme.

Cloning, sequencing, expression, and regulation of the structural gene for the copper/topa quinone-containing methylamine oxidase from Arthrobacter strain P1, a gram-positive facultative methylotroph

Deoxyoligonucleotides corresponding to amino acid sequences of methylamine oxidase and polyclonal anti-methylamine oxidase antibodies were used to probe Arthrobacter strain P1 plasmid and chromosomal DNA libraries. Two open reading frames, maoxI and maoxII, which are greater than 99% homologous, were cloned from the chromosomal library. The deduced amino acid sequences of the coding regions are identical except for two residues near the C termini. On the other hand, the 5'- and 3'-flanking regions of maoxI and maoxII are quite different. While either gene could code for methylamine oxidase, the dissimilarity in the 5'-flanking regions indicates that the genes are differently regulated. It was determined that maoxII alone encodes methylamine oxidase. The tyrosyl residue which is converted to topa quinone in the mature enzyme was located by comparison with amino acid sequences at the cofactor sites in other copper/topa quinone-containing amine oxidase. Transcriptional start sites and possible regulatory elements were identified in the 5' region of maoxI and maoxII, and stem-loop structures were found in the 3'-flanking regions. High levels of methylamine oxidase are produced when Arthrobacter strain P1 is grown on methylamine alone or on glucose plus methylamine, but growth on LB medium plus methylamine resulted in very low production of the enzyme. Expression of maoxII from its own promoter in Escherichia coli grown on glucose or LB medium with or without methylamine gave the same level of production of methylamine oxidase.

Molecular characterization of the enniatin synthetase gene encoding a multifunctional enzyme catalysing N-methyldepsipeptide formation in Fusarium scirpi

The gene encoding the multifunctional enzyme enniatin synthetase from Fusarium scirpi (esyn1) was isolated and characterized by transcriptional mapping and expression studies in Escherichia coli. This is the first example of a gene encoding an N-methyl peptide synthetase. The nucleotide sequence revealed an open reading frame of 9393 bp encoding a protein of 3131 amino acids (M(r) 346,900). Two domains designated EA and EB within the protein were identified which share similarity to each other and to microbial peptide synthetase domains. In contrast to the N-terminal domain EA, the carboxyl terminal domain EB is interrupted by a 434-amino-acid portion which shows local similarity to a motif apparently conserved within adenine and cytosine RNA and DNA methyltransferases and therefore seems to harbour the N-methyl-transferase function of the multienzyme.

Cloning vectors, mutagenesis, and gene disruption (ermR) for the erythromycin-producing bacterium Aeromicrobium erythreum

Genetic systems for study of Aeromicrobium erythreum, a gram-positive, G + C-rich (72%) bacterium with the capacity for erythromycin biosynthesis, are described. High-copy-number plasmids suitable as gene cloning vectors include derivatives of the Streptomyces plasmids pIJ101, pVE1, and pJV1. pIJ101 derivatives with missense substitutions at the rep gene BamHI site do not replicate in A. erythreum. Ethyl methanesulfonate treatment generated several amino acid auxotrophs and non-erythromycin-producing (Ery-) strains. Using the Ery- strain AR1807 as a recipient for plasmid-directed integrative recombination, the chromosomal ermR gene (encoding 23S rRNA methyltransferase) was disrupted. Phenotypic characterizations demonstrated that ermR is the sole determinant of macrolide antibiotic resistance in A. erythreum.

D-Xylose (D-glucose) isomerase from Arthrobacter strain N.R.R.L. B3728. Gene cloning, sequence and expression

Arthrobacter strain N.R.R.L. B3728 superproduces a D-xylose isomerase that is also a useful industrial D-glucose isomerase. The gene (xylA) that encodes it has been cloned by complementing a xylA mutant of the ancestral strain, with the use of a shuttle vector. The 5' region shows strong sequence similarity to Escherichia coli consensus promoters and ribosome-binding sequences and allows high levels of expression in E. coli. The coding sequence shows similarity to those for other D-xylose isomerases and is followed by 22 nucleotide residues with stop codons in each reading frame, a good 'consensus' ribosome-binding site and an open reading frame showing similarity to those of known D-xylulokinases (xylB). Studies on the expression of the cloned gene in Arthrobacter and in E. coli suggest that the two genes are part of a xyl operon regulated by a repressor that is defective in strain B3728. Codon usage in these two genes, and in another open reading frame (nxi) that was adventitiously isolated during early cloning attempts, shows some characteristic omissions and a strong G + C preference in redundant positions.

Bacterial resistance to macrolide, lincosamide, and streptogramin antibiotics by target modification

Images

A repeated decapeptide motif in the C-terminal domain of the ribosomal RNA methyltransferase from the erythromycin producer Saccharopolyspora erythraea

Re-analysis of the primary structure of the ribosomal RNA N-methyltransferase that confers self-resistance on the erythromycin-producing bacterium Saccharopolyspora erythraea has confirmed the presence of a C-terminal domain containing extensive repeat sequences. Nine tandem repeats can be discerned, with a decapeptide consensus sequence GGRx(H/R)GDRRT, although no single residue is wholly invariant. This highly polar, potentially flexible domain, which is predicted to adopt either a random coil or a structure with beta turns, has a counterpart in the erythromycin methyltransferase of an erythromycin-producing species of Arthrobacter. It also significantly resembles a portion of the C-terminal region of the eukaryotic protein nucleolin, which is unusually rich in dimethylarginine and glycine, and which is also predicted to behave as a random coil in solution. This resemblance, despite the very different roles of these proteins in ribosome biogenesis, strengthens the idea that in both rRNA methyltransferases and nucleolin these C-terminal sequences might contribute to rRNA binding.

Novel mechanisms of resistance to lincosamides in Staphylococcus and Arthrobacter spp

Clinical isolates of Staphylococcus and Arthrobacter spp. were screened for lincosamide resistance. Six different patterns of resistance were found. Strains designated SF27 and SF28 showed low-level resistance to lincosamides: one was susceptible to erythromycin (SF27) and the other was resistant (SF28). Analysis of ribosomes from the resistant strains in an in vitro poly(U)-dependent protein-synthesizing system showed that ribosomes of both strains were sensitive to lincomycin and clindamycin. Four patterns of high-level resistance to lincosamides were observed (strains SF4, SF19, SF30, and SF31). All of these except SF30 had ribosomes which were highly resistant in vitro to the antibiotics and showed a close correlation with results of the in vivo experiments. In vivo protein synthesis by strain SF30 was resistant to lincomycin and sensitive to clindamycin, whereas the ribosomes were sensitive when assayed in vitro. Lincosamide-inactivating enzymes were not detected in cell extracts of the six resistant strains. Strains SF19 and SF31 demonstrated two ribosome-mediated lincosamides resistance mechanisms that were not previously reported. Both strains were highly resistant to lincosamides and susceptible to erythromycin, but SF19 was also highly resistant to oleandomycin and partially resistant to various macrolides.

Evidence for natural gene transfer from gram-positive cocci to Escherichia coli

High-level resistance to macrolide-lincosamide-streptogramin type B (MLS) antibiotics in Escherichia coli BM2570 is due to the presence on the conjugative plasmid pIP1527 of the MLS resistance determinant ermBC, which is almost identical to the erm genes previously described in plasmid pAM77 from Streptococcus sanguis (ermAM) and in transposon Tn917 from Enterococcus faecalis (ermB). This gene and its regulatory region are located downstream from the insertion sequence IS1. The 23S rRNA methylase encoded by pIP1527 differs by three and six amino acids from those encoded by Tn917 and pAM77, respectively. Unlike the streptococcal elements which confer the inducible MLS phenotype, the ermBC gene is expressed constitutively in E. coli and Bacillus subtilis, due to several mutations in the regulatory region. Transcription of the ermBC gene starts from three different sites following three overlapping promoters which function in both E. coli and B. subtilis. Promoters P2 and P3 are located within the region homologous to pAM77 and Tn917, and P1 is a hybrid promoter constituted by -35 and -10 sequences located at the end of IS15 and in the streptococcal region, respectively. These results constitute evidence for the recent in vivo transfer from Streptococcus spp. to E. coli. This transfer could have been mediated by transposons such as Tn917 or Tn1545 from Streptococcus pneumoniae, which also bears an MLS determinant that is homologous to ermB. We speculate that the insertion sequences IS15 and IS1 could have played a role in the expression and dissemination of ermBC, which has been found in numerous strains of enterobacteria.

Translational attenuation control of ermSF, an inducible resistance determinant encoding rRNA N-methyltransferase from Streptomyces fradiae

An inducible resistance determinant, ermSF, from the tylosin producer Streptomyces fradiae NRRL 2338 has been cloned, sequenced, and shown to confer inducible macrolide-lincosamide-streptogramin B resistance when transferred to Streptomyces griseofuscus NRRL 23916. From mapping studies with S1 nuclease to locate the site of transcription initiation, the ermSF message contains a 385-nucleotide 5' leader sequence upstream from the 960-nucleotide major open reading frame that encodes the resistance determinant. On the basis of the potential secondary structure that the ermSF leader can assume, a translational attenuation model similar to that for ermC is proposed. The model is supported by mutational analysis involving deletions in the proposed attenuator. By analysis with restriction endonucleases, ermSF is indistinguishable from the tlrA gene described by Birmingham et al. (V. A. Birmingham, K. L. Cox, J. L. Larson, S. E. Fishman, C. L. Hershberger, and E. T. Seno, Mol. Gen. Genet. 204:532-539, 1986) which comprises one of at least three genes from S. fradiae that can confer tylosin resistance when subcloned into S. griseofuscus. When tested for inducibility, ermSF appears to be strongly induced by erythromycin, but not by tylosin.

Transformation of Arthrobacter and studies on the transcription of the Arthrobacter ermA gene in Streptomyces lividans and Escherichia coli

We report the development of a plasmid-mediated transformation system for Arthrobacter sp. NRRLB3381, using the Streptomyces cloning vector pIJ702. Our procedure gives a transformation frequency of 10(3)/micrograms of plasmid DNA. In addition we have explored the expression of the Arthrobacter ermA gene in Streptomyces lividans and Escherichia coli, and shown that the ermA promoter is recognized in S. lividans not E. coli. The relationship between Arthrobacter, Streptomyces and E. coli promoters is discussed.

Antimicrobial resistance of Staphylococcus aureus: genetic basis

Images

Extrachromosomal systems and gene transmission in anaerobic bacteria

Obligately anaerobic bacteria are important in terms of their role as medical pathogens as well as their degradative capacities in a variety of natural ecosystems. Two major anaerobic genera, Bacteroides and Clostridium, are examined in this review. Plasmid elements in both genera are reviewed within the context of conjugal transfer and drug resistance. Genetic systems that facilitate the study of these anaerobic bacteria have emerged during the past several years. In large part, these developments have been linked to work centered on extrachromosomal genetic systems in these organisms. Conjugal transfer of antibiotic resistance has been a central focus in this regard. Transposable genetic elements in the Bacteroides are discussed and the evolution and spread of resistance to lincosamide antibiotics are considered at the molecular level. Recombinant DNA systems that employ shuttle vectors which are mobilized by conjugative plasmids have been developed for use in Bacteroides and Clostridium. The application of transmission and recombinant DNA genetic systems to study these anaerobes is under way and is likely to lead to an increased understanding of this important group of procaryotes.

Cloning and analysis of ermG, a new macrolide-lincosamide-streptogramin B resistance element from Bacillus sphaericus

To analyze the regulation of a newly discovered macrolide-lincosamide-streptogramin B resistance element (ermG) found in a soil isolate of Bacillus sphaericus, we cloned this determinant and obtained its DNA sequence. Minicell analysis revealed that ermG specifies a 29,000-dalton protein, the synthesis of which is induced by erythromycin. S1 nuclease mapping was used to identify the transcriptional start site. These experiments demonstrated the presence on the ermG mRNA of a 197 to 198-base leader. Within the leader are two small open reading frames (ORFs) capable of encoding 11- and 19-amino-acid peptides. Each ORF is preceded by a suitably spaced Shine-Dalgarno sequence. The ermG protein is encoded by a large ORF that encodes a 244-amino-acid protein, in agreement with the minicell results. This protein and the 19-amino-acid peptide are highly homologous to the equivalent products of ermC and ermA. We conclude, on the basis of this homology, that ermG encodes an rRNA transmethylase. The leader of ermG can be folded into a structure that sequesters the Shine-Dalgarno sequence and start codon for the large ORF (SD3). On the basis of these data and on the observed greater responsiveness of the ermG system than of the ermC system to low concentrations of erythromycin, we propose a model for the regulation of this gene in which the stalling of a ribosome under the influence of an inducer, while reading either peptide, suffices to uncover SD3 and allow translation of the rRNA transmethylase. The evolution of ermG is discussed.

Cloning and nucleotide sequence of a carbomycin-resistance gene from Streptomyces thermotolerans

Two plasmids (pOJ158 and pOJ159) containing DNA fragments from the carbomycin(Cb)-producing strain Streptomyces thermotolerans were identified in Streptomyces griseofuscus based on their ability to confer resistance to Cb. The Cb-resistance determinants on pOJ158 and pOJ159 were designated carA and carB, respectively. In S. griseofuscus, pOJ159 also confers resistance to spiramycin, rosaramicin, lincomycin, and vernamycin B, but not to tylosin; in Streptomyces lividans, pOJ159 additionally confers resistance to erythromycin and oleandomycin. The carB gene was localized on pOJ159 to a 1.25-kb region whose nucleotide sequence was determined. The sequence has a G + C content of 68% and contains the coding sequence for carB and portions of the 5' and 3' untranslated regions. A comparison of the amino acid sequence of the protein encoded by carB (as deduced from the nucleotide sequence) with the deduced amino acid sequence of the RNA methylase from Streptomyces erythraeus (encoded by ermE) revealed extensive homology, suggesting that carB also encodes an RNA methylase. The region 5' to the coding sequence does not contain a small ORF or regions of complementarity that are commonly associated with translationally regulated macrolide-lincosamide-streptogramin B resistance genes. The 3' untranslated region contains an inverted repeat sequence that potentially can form a stable RNA stem-loop structure with a calculated delta G of -70 kcal.

Complete nucleotide sequence and transcription of ermF, a macrolide-lincosamide-streptogramin B resistance determinant from Bacteroides fragilis

DNA sequence analysis of a portion of an EcoRI fragment of the Bacteroides fragilis R plasmid pBF4 has allowed us to identify the macrolide-lincosamide-streptogramin B resistance (MLSr) gene, ermF. ermF had a relative moles percent G + C of 32, was 798 base pairs in length, and encoded a protein of approximately 30,360 daltons. Comparison between the deduced amino acid sequence of ermF and six other erm genes from gram-positive bacteria revealed striking homologies among all of these determinants, suggesting a common origin. Based on these and other data, we believe that ermF codes for an rRNA methylase. Analysis of the nucleotide sequences upstream and downstream from the ermF gene revealed the presence of directly repeated sequences, now identified as two copies of the insertion element IS4351. One of these insertion elements was only 26 base pairs from the start codon of ermF and contained the transcriptional start signal for this gene as judged by S1 nuclease mapping experiments. Additional sequence analysis of the 26 base pairs separating ermF and IS4351 disclosed strong similarities between this region and the upstream regulatory control sequences of ermC and ermA (determinants of staphylococcal origin). These results suggested that ermF was not of Bacteroides origin and are discussed in terms of the evolution of ermF and the expression of drug resistance in heterologous hosts.

N-Methyl transferase of Streptomyces erythraeus that confers resistance to the macrolide-lincosamide-streptogramin B antibiotics: amino acid sequence and its homology to cognate R-factor enzymes from pathogenic bacilli and cocci

The nucleotide sequence of a structural gene ermE for ribosomal RNA (rRNA) N6-amino adenine N-methyl transferase (NMT) of Streptomyces erythraeus, cloned by Thompson et al. [Gene 20 (1982) 51-62], has been determined. The NMT amino acid (aa) sequence deduced from the nucleotide sequence contains extensive homology to aa sequences of cognate NMTs specified by: (1) plasmid pE194 from Staphylococcus aureus, 30% G + C, ermC; (2) plasmid pAM77 from Streptococcus sanguis, 43% G + C; as well as to (3) a chromosomal determinant from Bacillus licheniformis 759, 46% G + C, ermD, cloned in a recombinant plasmid pBD90. These findings suggest that all four NMT structural genes could have evolved from a common progenitor sequence despite the wide range of % G + C of the erm genes reflecting their current respective hosts. Comparison of the four NMT sequences with respect to localized hydrophobicity averaged over a moving window of 11 aa indicates that the common features of localized hydrophobicity that characterize the C-terminal portion of the ermE and ermD proteins are distinguishable from a contrasting pattern of hydrophobicity that characterizes the ermC and pAM77-coded proteins.