Evidence for Horizontal Gene Transfer in Evolution of Elongation Factor Tu in Enterococci

Abstract

The elongation factor Tu, encoded by tuf genes, is a GTP binding protein that plays a central role in protein synthesis. One to three tuf genes per genome are present, depending on the bacterial species. Most low-G+C-content gram-positive bacteria carry only one tuf gene. We have designed degenerate PCR primers derived from consensus sequences of the tuf gene to amplify partial tuf sequences from 17 enterococcal species and other phylogenetically related species. The amplified DNA fragments were sequenced either by direct sequencing or by sequencing cloned inserts containing putative amplicons. Two different tuf genes (tufA and tufB) were found in 11 enterococcal species, including Enterococcus avium, Enterococcus casseliflavus, Enterococcus dispar, Enterococcus durans, Enterococcus faecium, Enterococcus gallinarum, Enterococcus hirae, Enterococcus malodoratus, Enterococcus mundtii, Enterococcus pseudoavium, and Enterococcus raffinosus. For the other six enterococcal species (Enterococcus cecorum, Enterococcus columbae, Enterococcus faecalis, Enterococcus sulfureus, Enterococcus saccharolyticus, and Enterococcus solitarius), only the tufA gene was present. Based on 16S rRNA gene sequence analysis, the 11 species having two tuf genes all have a common ancestor, while the six species having only one copy diverged from the enterococcal lineage before that common ancestor. The presence of one or two copies of the tuf gene in enterococci was confirmed by Southern hybridization. Phylogenetic analysis of tuf sequences demonstrated that the enterococcal tufA gene branches with the Bacillus, Listeria, and Staphylococcus genera, while the enterococcal tufB gene clusters with the genera Streptococcus and Lactococcus. Primary structure analysis showed that four amino acid residues encoded within the sequenced regions are conserved and unique to the enterococcal tufB genes and the tuf genes of streptococci and Lactococcus lactis. The data suggest that an ancestral streptococcus or a streptococcus-related species may have horizontally transferred a tuf gene to the common ancestor of the 11 enterococcal species which now carry two tuf genes.

The elongation factor Tu (EF-Tu) is a GTP binding protein playing a central role in protein synthesis. It mediates the recognition and transport of aminoacyl-tRNAs and their positioning to the A site of the ribosome (20). The highly conserved function and ubiquitous distribution render the elongation factor a valuable phylogenetic marker among eubacteria and even throughout the archaebacterial and eukaryotic kingdoms (3, 31). The tuf genes encoding EF-Tu are present in various copy numbers per bacterial genome. Most gram-negative bacteria contain two tuf genes (5, 15, 19, 39, 41, 43). As found in Escherichia coli, the two genes, while being almost identical in sequence, are located in different parts of the bacterial chromosome (15, 20, 41). However, recently completed maps of microbial genomes revealed that only one tuf gene is found in Helicobacter pylori as well as in some obligate parasitic bacteria, such as Borrelia burgdorferi, Rickettsia prowazekii, and Treponema pallidum, and in some cyanobacteria (16, 18, 24, 32, 41, 44). In most gram-positive bacteria studied so far, only one tuf gene was found (8, 14, 17, 22, 28–30, 32, 35, 39). However, Southern hybridization showed that there are two tuf genes in some clostridia (39) as well as in Streptomyces coelicolor and Streptomyces lividans (46, 47). Up to three tuf-like genes have been identified in Streptomyces ramocissimus (48).

Although massive prokaryotic gene transfer is suggested to be one of the factors responsible for the evolution of bacterial genomes (12, 27, 42), the genes encoding components of the translation machinery are thought to be highly conserved and difficult to transfer horizontally due to the complexity of their interactions (23). However, a few recent studies demonstrated evidence that horizontal gene transfer has also occurred in the evolution of some genes coding for the translation apparatus, namely, 16S rRNA and some aminoacyl-tRNA synthetases (6, 27, 45, 48, 49). No further data suggest that such a mechanism is involved in the evolution of the elongation factors. Previous studies concluded that the two copies of tuf genes in the genomes of some bacteria resulted from an ancient event of gene duplication (10, 39). Moreover, a study of the tuf gene in R. prowazekii suggested that intrachromosomal recombination has taken place in the evolution of the genome of this organism (41).

To date, little is known about the tuf genes of enterococcal species. In this study, we analyzed partial sequences of tuf genes in 17 enterococcal species, namely, Enterococcus avium, E. casseliflavus, E. cecorum, E. columbae, E. dispar, E. durans, E. faecalis, E. faecium, E. gallinarum, E. hirae, E. malodoratus, E. mundtii, E. pseudoavium, E. raffinosus, E. saccharolyticus, E. solitarius, and E. sulfureus. We report here the presence of two divergent copies of tuf genes in 11 of these enterococcal species. The six other species carried a single tuf gene. The evolutionary implications are discussed.

(This study was presented in part at the 100th General Meeting of the American Society for Microbiology, Los Angeles, Calif., 21 to 25 May 2000.)

MATERIALS AND METHODS

Bacterial strains.

Seventeen enterococcal strains and other gram-positive bacterial strains obtained from the American Type Culture Collection (ATCC; Manassas, Va.) were used in this study (Table 1). All strains were grown on sheep blood agar or in brain heart infusion broth prior to DNA isolation.

TABLE 1.

tuf gene sequences obtained in our laboratory

Species	Strain	Gene(s)	GenBank accession no.
Abiotrophia adiacens	ATCC 49175	tuf	AF124224
Enterococcus avium	ATCC 14025	tufA	AF124220
		tufB	AF274715
Enterococcus casseliflavus	ATCC 25788	tufA	AF274716
		tufB	AF274717
Enterococcus cecorum	ATCC 43198	tuf	AF274718
Enterococcus columbae	ATCC 51263	tuf	AF274719
Enterococcus dispar	ATCC 51266	tufA	AF274720
		tufB	AF274721
Enterococcus durans	ATCC 19432	tufA	AF274722
		tufB	AF274723
Enterococcus faecalis	ATCC 29212	tuf	AF124221
Enterococcus faecium	ATCC 19434	tufA	AF124222
		tufB	AF274724
Enterococcus gallinarum	ATCC 49573	tufA	AF124223
		tufB	AF274725
Enterococcus hirae	ATCC 8043	tufA	AF274726
		tufB	AF274727
Enterococcus malodoratus	ATCC 43197	tufA	AF274728
		tufB	AF274729
Enterococcus mundtii	ATCC 43186	tufA	AF274730
		tufB	AF274731
Enterococcus pseudoavium	ATCC 49372	tufA	AF274732
		tufB	AF274733
Enterococcus raffinosus	ATCC 49427	tufA	AF274734
		tufB	AF274735
Enterococcus saccharolyticus	ATCC 43076	tuf	AF274736
Enterococcus solitarius	ATCC 49428	tuf	AF274737
Enterococcus sulfureus	ATCC 49903	tuf	AF274738
Lactococcus lactis	ATCC 11154	tuf	AF274745
Listeria monocytogenes	ATCC 15313	tuf	AF274746
Listeria seeligeri	ATCC 35967	tuf	AF274747
Staphylococcus aureus	ATCC 25923	tuf	AF274739
Staphylococcus epidermidis	ATCC 14990	tuf	AF274740
Streptococcus mutans	ATCC 25175	tuf	AF274741
Streptococcus pneumoniae	ATCC 6303	tuf	AF274742
Streptococcus pyogenes	ATCC 19615	tuf	AF274743
Streptococcus suis	ATCC 43765	tuf	AF274744

DNA isolation.

Bacterial DNAs were prepared using the G NOME DNA extraction kit (Bio101, Vista, Calif.) as previously described (25).

Sequencing of putative tuf genes.

In order to obtain the tuf gene sequences of enterococci and other gram-positive bacteria, two sequencing approaches were used: (i) sequencing of cloned PCR products and (ii) direct sequencing of PCR products. A pair of degenerate primers (U1, 5′-AAYATGATIACIGGIGCIGCICARATGGA-3′, and U3, 5′-CCIACIGTICKICCRCCYTCRCG-3′) were used to amplify an 886-bp portion of the tuf genes from enterococcal species and other gram-positive bacteria as previously described (25). For E. avium, E. casseliflavus, E. dispar, E. durans, E. faecium, E. gallinarum, E. hirae, E. mundtii, E. pseudoavium, and E. raffinosus, the amplicons were cloned using the Original TA cloning kit (Invitrogen, Carlsbad, Calif.) as previously described (25). Five clones for each species were selected for sequencing. For E. cecorum, E. faecalis, E. saccharolyticus, and E. solitarius as well as the other gram-positive bacteria, the sequences of the 886-bp amplicons were obtained by direct sequencing. Based on the results obtained from the earlier rounds of sequencing, two pairs of primers were designed for obtaining the partial tuf sequences from the other enterococcal species by direct sequencing. One pair of primers (EntA1, 5′-ATCTTAGTAGTTTCTGCTGCTGA-3′, and EntA2, 5′-GTAGAATTCAGGACGGTAGTTAG-3′) was used to amplify the enterococcal tuf gene fragments from E. columbae, E. malodoratus, and E. sulfureus. Another pair of primers (U1 and EntB, 5′-GTAGAAYTGTGGWCGATARTTRT-3′) was used to amplify the second tuf gene fragments from E. avium, E. malodoratus, and E. pseudoavium.

Prior to direct sequencing, PCR products were electrophoresed on a 1% agarose gel at 120 V for 2 h. The gel was then stained with 0.02% methylene blue for 30 min and washed twice with autoclaved distilled water for 15 min. The gel slices containing PCR products of the expected sizes were cut out and purified with the QIAquick gel extraction kit (QIAgen Inc., Mississauga, Ontario, Canada) according to the manufacturer's instructions. PCR mixtures for sequencing were prepared as described previously (25). DNA sequencing was carried out with the Big Dye Terminator Ready Reaction cycle sequencing kit using a 377 DNA sequencer (PE Applied Biosystems, Foster City, Calif.). Both strands of the amplified DNA were sequenced. The sequence data were verified using the Sequencher 3.0 software (Gene Codes Corp., Ann Arbor, Mich.).

Sequence analysis and phylogenetic study.

Nucleotide sequences of the tuf genes and their respective flanking regions in E. faecalis, Staphylococcus aureus, and Streptococcus pneumoniae were retrieved from The Institute for Genomic Research (http://www.tigr.org) microbial genome database, and sequences of Streptococcus pyogenes were obtained from the University of Oklahoma database (http://www.genome.ou.edu/strep.html). DNA sequences and deduced protein sequences obtained in this study were compared with those in all publicly available databases by using the BLAST (2) and FASTA programs. Unless specified, sequence analysis was conducted with the programs from the GCG package (version 10; Genetics Computer Group, Madison, Wis.). Sequence alignment of the tuf genes from 74 species representing all three kingdoms of life (Tables 1 and 2) was carried out by use of Pileup and was corrected upon visual analysis. The N- and C-terminus extremities of the sequences were trimmed to yield a common block of 201 amino acids, and equivocal residues were removed. Phylogenetic analysis was performed with the aid of PAUP 4.0b4, written by David L. Swofford (Sinauer Associates, Inc., Publishers, Sunderland, Mass.). The distance matrix and maximum parsimony were used to generate phylogenetic trees, and bootstrap resampling procedures were performed, using 500 and 100 replications in each analysis, respectively.

TABLE 2.

tuf gene sequences selected from databases for this study

Species	Gene(s)	Accession no.a
Agrobacterium tumefaciens	tufA	X99673
	tufB	X99674
Anacystis nidulans	tuf	X17442
Aquifex aeolicus	tufA	AE000657
	tufB	AE000657
Bacillus stearothermophilus	tuf	AJ000260
Bacillus subtilis	tuf	AL009126
Bacteroides fragilis	tuf	P33165
Borrelia burgdorferi	tuf	AE000783
Brevibacterium linens	tuf	X76863
Burkholderia cepacia	tuf	P33167
Campylobacter jejuni	tufB	Y17167
Chlamydia pneumoniae	tuf	AE001363
Chlamydia trachomatis	tuf	M74221
Corynebacterium glutamicum	tuf	X77034
Cytophaga lytica	tuf	X77035
Deinococcus radiodurans	tuf	AE000513
Escherichia coli	tufA	J01690
	tufB	J01717
Fervidobacterium islandicum	tuf	Y15788
Haemophilus influenzae	tufA	L42023
	tufB	L42023
Helicobacter pylori	tuf	AE000511
Homo sapiens (human)	EF-1α	X03558
Methanococcus jannaschii	EF-1α	U67486
Mycobacterium leprae	tuf	D13869
Mycobacterium tuberculosis	tuf	X63539
Mycoplasma genitalium	tuf	L43967
Mycoplasma pneumoniae	tuf	U00089
Neisseria gonorrhoeae	tufA	L36380
Nicotiana tabacum (tobacco)	EF-1α	U04632
Peptococcus niger	tuf	X76869
Planobispora rosea	tuf1	U67308
Saccharomyces cerevisiae (yeast)	EF-1α	X00779
Salmonella enterica serovar Typhimurium	tufA	X55116
	tufB	X55117
Shewanella putrefaciens	tuf	P33169
Spirochaeta aurantia	tuf	X76874
Spirulina platensis	tufA	X15646
Streptomyces aureofaciens	tuf1	AF007125
Streptomyces cinnamoneus	tuf1	X98831
Streptomyces coelicolor	tuf1	X77039
	tuf3	X77040
Streptomyces collinus	tuf1	S79408
Streptomyces ramocissimus	tuf1	X67057
	tuf2	X67058
	tuf3	X67059
Synechocystis sp.	tuf	AB001339
Taxeobacter ocellatus	tuf	X77036
Thermotoga maritima	tuf	AE000512
Thermus aquaticus	tuf	X66322
Thermus thermophilus	tuf	X06657
Thiobacillus cuprinus	tuf	U78300
Treponema pallidum	tuf	AE000520
Wolinella succinogenes	tuf	X76872

^aSequence data were obtained from GenBank, EMBL, and SWISSPROT databases. Genes were designated as they appear in the references. 

Protein structure analysis.

The crystal structures of (i) Thermus aquaticus EF-Tu in complex with Phe-tRNA^Phe and a GTP analog (34) and (ii) E. coli EF-Tu in complex with GDP (40) served as templates for constructing the equivalent models for enterococcal EF-Tu. Homology modeling of protein structure was performed using the SWISS-MODEL server and inspected using the SWISS-PDB viewer version 3.1 (21).

Southern hybridization.

In a previous study (25), we amplified and cloned an 803-bp PCR product of the tuf gene fragment from E. faecium. Two divergent sequences of the inserts, which we assumed to be tufA and tufB genes, were obtained. The recombinant plasmid carrying either tufA or tufB sequence was used to generate two probes labeled with digoxigenin (DIG)-11-dUTP by PCR incorporation following the instructions of the manufacturer (Boehringer Mannheim, Laval, Québec, Canada). Enterococcal genomic DNA samples (1 to 2 μg) were digested to completion with restriction endonucleases BglII and XbaI as recommended by the supplier (Amersham Pharmacia Biotech, Mississauga, Ontario, Canada). These restriction enzymes were chosen because no restriction sites were observed within the amplified tuf gene fragments of most enterococci. Southern blotting and filter hybridization were performed using positively charged nylon membranes (Boehringer Mannheim) and QuikHyb hybridization solution (Stratagene Cloning Systems, La Jolla, Calif.) according to the manufacturers' instructions, with modifications. Twenty microliters of each digest was electrophoresed for 2 h at 120 V on a 0.8% agarose gel. The DNA fragments were denatured with 0.5 M NaOH and transferred by Southern blotting onto a positively charged nylon membrane (Boehringer Mannheim). The filters were prehybridized for 15 min and then were hybridized for 2 h in the QuikHyb solution at 68°C with either DIG-labeled probe. Posthybridization washings were performed twice with 0.5× SSC–1% sodium dodecyl sulfate (SDS) at room temperature for 15 min and twice in the same solution at 60°C for 15 min (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). Detection of bound probes was achieved using disodium 3-(4-methoxyspiro(1,2-dioxetane-3,2′-(5′-chloro)tricyclo(3,3.1.1^3.7)decan)-4-yl) phenyl phosphate (CSPD) (Boehringer Mannheim) as specified by the manufacturer.

Nucleotide sequence accession numbers.

The GenBank accession numbers for partial tuf gene sequences generated in this study are given in Table 1. Sequences were assigned accession no. AF124220 to AF124224 and AF274715 to AF274747.

RESULTS

Sequencing and nucleotide sequence analysis.

In this study, all gram-positive bacteria other than enterococci yielded a single tuf sequence of 886 bp using primers U1 and U3 (Table 1). Each of the four enterococcal species E. cecorum, E. faecalis, E. saccharolyticus, and E. solitarius also yielded one 886-bp tuf sequence. On the other hand, for E. avium, E. casseliflavus, E. dispar, E. durans, E. faecium, E. gallinarum, E. hirae, E. mundtii, E. pseudoavium, and E. raffinosus, direct sequencing of the 886-bp fragments revealed overlapping peaks according to their sequence chromatograms, suggesting the presence of additional copies of the tuf gene. Therefore, the tuf gene fragments of these 10 species were cloned first and then sequenced. Sequencing data revealed that two different types of tuf sequences (tufA and tufB) are found in eight of these species, namely, E. casseliflavus, E. dispar, E. durans, E. faecium, E. gallinarum, E. hirae, E. mundtii, and E. raffinosus. Five clones of both E. avium and E. pseudoavium yielded only a single tuf sequence. These new sequence data allowed the design of new primers specific for the enterococcal tufA or tufB sequences. Primers EntA1 and EntA2 were designed to amplify only enterococcal tufA sequences, and a 694-bp fragment was amplified from all 17 enterococcal species. The 694-bp sequences of tufA genes from E. columbae, E. malodoratus, and E. sulfureus were obtained by direct sequencing using these primers. Primers U1 and EntB were designed for the amplification of 730-bp portion of tufB genes and yielded the expected fragments from 11 enterococcal species, including E. malodoratus and the 10 enterococcal species in which heterogeneous tuf sequences were initially found. The sequences of the tufB fragments for E. avium, E. malodoratus, and E. pseudoavium were determined by direct sequencing using the primers U1 and EntB. Overall, tufA gene fragments were obtained from all 17 enterococcal species but tufB gene fragments were obtained from only 11 enterococcal species (Table 1).

The identities between tufA and tufB for each enterococcal species were 68 to 79% at the nucleotide level and 81 to 89% at the amino acid level. The tufA gene is highly conserved among all enterococcal species, with identities ranging from 87 to 99% for DNA and 93 to 99% for amino acid sequences, while the identities among tufB genes of enterococci ranged from 77 to 92% for DNA and from 91 to 99% for amino acid sequences, indicating their different origins and evolution (Table 3). Since E. solitarius has been transferred to the genus Tetragenococcus (13), which is also a low-G+C-content gram-positive bacterium, our sequence comparison did not include this species as an enterococcus. The G+C content of enterococcal tufA sequences ranged from 40.8 to 43.1%, while that of enterococcal tufB sequences ranged from 37.8 to 46.3%. Based on amino acid sequence comparison, the enterococcal tufA gene products shared higher identities with those of Abiotrophia adiacens, Bacillus subtilis, Listeria monocytogenes, S. aureus, and Staphylococcus epidermidis. On the other hand, the enterococcal tufB gene products shared higher percentages of amino acid identity with the tuf genes of S. pneumoniae, S. pyogenes, and Lactococcus lactis (Table 3).

TABLE 3.

Nucleotide and amino acid sequence identities of EF-Tu between different enterococci and other low-G+C-content gram-positive bacteria^a

^aThe data are percent sequence identities. The data in the upper right triangle represent the deduced amino acid sequence identities of EF-Tu of gram-positive bacteria, while the data in the lower left triangle represent the DNA sequence identities of the corresponding tuf genes. The sequence identities between different enterococcal tufA genes are boxed, while those between enterococcal tufB genes are shaded. 

In order to elucidate whether the two enterococcal tuf sequences encode genuine EF-Tu, the deduced amino acid sequences of both genes were aligned with other EF-Tu sequences available in SWISSPROT (release 38). Sequence alignment demonstrated that both gene products are highly conserved and carry all conserved residues present in this portion of prokaryotic EF-Tu (Fig. 1). Therefore, it appears that both gene products could fulfill the function of EF-Tu. The partial tuf gene sequences encode the portion of EF-Tu from residues 117 to 317, according to E. coli numbering (40). This portion makes up of the last four α-helices and two β-strands of domain I, the entire domain II, and the N-terminal part of domain III on the basis of the determined structures of E. coli EF-Tu (40).

FIG. 1.

Abridged multiple amino acid sequence alignment of the partial tuf gene products from selected species by the program Alscript (4). Residues highly conserved in bacteria (34) are boxed in grey and gaps are represented with dots. Residues in reverse print are unique to the enterococcal tufB gene as well as to streptococcal and lactococcal tuf gene products. Numbering is based on E. coli EF-Tu, and secondary structure elements of E. coli EF-Tu are represented by cylinders (α-helices) and arrows (β-strands) (40).

Based on the deduced amino acid sequences, the enterococcal tufB genes have unique conserved residues, Lys129, Leu140, Ser230, and Asp234 (E. coli numbering), that are also conserved in streptococci and L. lactis, but not in the other bacteria (Fig. 1). All these residues are located in loops except for Ser230. In other bacteria the residue Ser230 is replaced by highly conserved Thr, which is the fifth residue of the third β-strand of domain II. This region is partially responsible for the interaction between the EF-Tu and aminoacyl-tRNA by the formation of a deep pocket for any of the 20 naturally occurring amino acids (34, 40). According to our three-dimensional model (data not illustrated), the substitution Thr230→Ser in domain II of EF-Tu may have little impact on the ability of the pocket to accommodate any amino acid. However, the high conservation of Thr230 compared to the unique Ser substitution found only in streptococci and 11 enterococci could suggest a subtle functional role for this residue.

The tuf gene sequences obtained for E. faecalis, S. aureus, S. pneumoniae, and S. pyogenes were compared with their respective incomplete genome sequences (http://www.tigr.org/tdb/mdb/mdbinprogress.html). Contigs with greater than 99% identity were identified. Analysis of the E. faecalis genome data revealed that the single E. faecalis tuf gene is located within an str operon in which tuf is preceded by fus, which encodes the elongation factor G. This str operon is present in S. aureus and B. subtilis but not in the two streptococcal genomes examined. The 700-bp or so sequence upstream of the S. pneumoniae tuf gene has no homology with any known gene sequences. In S. pyogenes, the gene upstream of tuf is similar to a cell division gene, ftsW, suggesting that the tuf genes in streptococci are not arranged in an str operon.

Phylogenetic analysis.

Phylogenetic analysis of the tuf amino acid sequences with representatives of eubacteria, archaebacteria, and eukaryotes using neighbor-joining and maximum parsimony methods showed three major clusters representing the three kingdoms of life. Both methods yielded similar topologies consistent with the rRNA gene data (data not shown). Within the bacterial clade, the tree is polyphyletic, but tufA genes from all enterococcal species always clustered with those from other low-G+C-content gram-positive bacteria (except for streptococci and lactococci), while the tufB genes of the 11 enterococcal species form a distinct cluster with streptococci and L. lactis (Fig. 2). Duplicated genes from the same organism did not cluster together, thereby not suggesting evolution by recent gene duplication.

FIG. 2.

Distance matrix tree of bacterial EF-Tu based on amino acid sequence homology. The tree was constructed by the neighbor-joining method. The tree was rooted using archaeal and eukaryotic EF-1α genes as the outgroup. The scale bar represents 5% changes in amino acid sequence, as determined by taking the sum of all of the horizontal lines connecting two species.

Southern hybridization.

Southern hybridization of BglII-XbaI-digested genomic DNA from 12 enterococcal species tested with the tufA probe (DIG-labeled tufA fragment from E. faecium) yielded two bands of different sizes in nine species, which also carried two divergent tuf sequences according to their sequencing data. For E. faecalis and E. solitarius, a single band was observed, indicating that one tuf gene is present (Fig. 3). A single band was also found when digested genomic DNAs from S. aureus, S. pneumoniae, and S. pyogenes were hybridized with the tufA probe (data not shown). For E. faecium, the presence of three bands can be explained by the existence of an XbaI restriction site in the middle of the tufA sequence, which was confirmed by sequencing data. Hybridization with the tufB probe (DIG-labeled tufB fragment of E. faecium) showed a banding profile similar to the one obtained with the tufA probe (data not shown).

FIG. 3.

Southern hybridization of BglII-XbaI-digested genomic DNAs of some enterococci (except for E. casseliflavus and E. gallinarum, whose genomic DNA was digested with BamHI-PvuII) using the tufA gene fragment of E. faecium as a probe. The sizes of hybridizing fragments are shown in kilobases. Strains tested are listed in Table 1.

DISCUSSION

In this study, we have shown that two divergent copies of genes encoding EF-Tu are present in some enterococcal species. Sequence data revealed that both genes are highly conserved at the amino acid level. One copy (tufA) is present in all enterococcal species, while the other (tufB) is present in only 11 of the 17 enterococcal species studied. Based on 16S rRNA sequence analysis, these 11 species are members of three different enterococcal subgroups (E. avium, E. faecium, and E. gallinarum species groups) and a distinct species (E. dispar). Moreover, 16S rDNA phylogeny suggests that the 11 species that possess two tuf genes all have a common ancestor from which they evolved further to become the current species (36). Since the six other species having only one copy diverged from the enterococcal lineage before that common ancestor, it appears that the presence of one tuf gene in these six species is not attributable to gene loss.

Two clusters of low-G+C-content gram-positive bacteria were observed in the phylogenetic tree of the tuf genes: one contained a majority of low-G+C-content gram-positive bacteria and the other contained lactococci and streptococci. This is similar to a previous finding based on phylogenetic analysis of the 16S rRNA gene and the hrcA gene coding for a unique heat shock regulatory protein (1). The enterococcal tufA genes branched with most of the low-G+C-content gram-positive bacteria, suggesting that they originated from a common ancestor. On the other hand, the enterococcal tufB genes branched with the genera Streptococcus and Lactococcus, which form a distinct lineage separated from other low-G+C-content gram-positive bacteria (Fig. 2). The finding that these EF-Tu proteins share some conserved amino acid residues unique to this branch also supports the idea that they may have a common ancestor. Although these conserved residues might result from convergent evolution upon a specialized function, such convergence at the sequence level, even for a few residues, seems to be rare, making it an unlikely event. Moreover, no currently known selective pressure, if any, would account for keeping one versus two tuf genes in bacteria. The G+C contents of enterococcal tufA and tufB sequences are similar, indicating that they both originated from low-G+C-content gram-positive bacteria, in accordance with the phylogenetic analysis.

The tuf genes are present in various copy numbers in different bacteria. Furthermore, the two tuf genes are normally associated with characteristic flanking genes (10). The two tuf gene copies commonly encountered within gram-negative bacteria are part of either the bacterial str operon or the tRNA-tufB operon (5, 10, 41). The arrangement of tufA in the str operon was also found in a variety of bacteria, including Thermotoga maritima, the earliest divergent bacterium sequenced so far (33), Aquifex aeolicus (11), cyanobacteria (7, 24), Bacillus spp. (28, 29), Micrococcus luteus (35), Mycobacterium tuberculosis (9), and Streptomyces spp. (46, 47). Furthermore, the tRNA-tufB operon has also been identified in A. aeolicus (11), Thermus thermophilus (38), and Chlamydia trachomatis (10). The two widespread tuf gene arrangements argue in favor of their ancient origins (10). It is noteworthy that most obligate intracellular parasites, such as Mycoplasma spp. (17, 22), R. prowazekii (41), B. burgdorferi (16), and T. pallidum (18), contain only one tuf gene. Their flanking sequences are distinct from the two conserved patterns as a result of selection for effective propagation by an extensive reduction in genome size by intragenomic recombination and rearrangement (10, 16, 18, 41).

Most gram-positive bacteria with low G+C content that have been sequenced to date contain only a single copy of the tuf gene as a part of the str operon. This is the case for B. subtilis, S. aureus, and E. faecalis. PCR amplification using a primer targeting a conserved region of the fus gene and the tufA-specific primer EntA2, but not the tufB-specific primer EntB, yielded the expected amplicons for all 17 enterococcal species tested, indicating the presence of the fus-tuf organization in all enterococci (data not shown). However, in the genomes of S. pneumoniae and S. pyogenes, the sequences flanking the tuf genes differ, although the tuf gene itself remains highly conserved. The enterococcal tufB genes are clustered with those of streptococci, but at present we do not have enough data to identify the genes flanking the enterococcal tufB genes. Furthermore, the functional role of the enterococcal tufB genes remains unknown. One can only postulate that the two divergent gene copies are expressed under different conditions.

The amino acid sequence identities between the enterococcal tufA and tufB genes are lower than either of (i) those between the enterococcal tufA and the tuf genes from other low-G+C-content gram-positive bacteria (streptococci and lactococci excluded) or (ii) those between the enterococcal tufB and streptococcal and lactococcal tuf genes. These findings suggest that the enterococcal tufA genes have a common ancestor with other low-G+C-content gram-positive bacteria via the simple scheme of vertical evolution, while the enterococcal tufB genes are more closely related to those of streptococci and lactococci. The facts that some enterococci possess an additional tuf gene and that the single streptococcal tuf gene is not clustered with those of other low-G+C-content gram-positive bacteria cannot be explained by the mechanism of gene duplication or intrachromosomal recombination. According to sequence and phylogenetic analysis, we propose that the presence of the additional copy of the tuf gene in 11 enterococcal species is due to horizontal gene transfer. The common ancestor of the 11 enterococcal species now carrying tufB genes acquired a tuf gene from an ancestral streptococcus or a streptococcus-related species through gene transfer during enterococcal evolution before the diversification of modern enterococci. Further study of the flanking regions of the gene may provide more clues to the origin and function of this gene in enterococci.

Recent studies of genes and genomes have demonstrated that considerable horizontal transfer occurred in the evolution of aminoacyl-tRNA synthetases in all three kingdoms of life (6, 26, 48). The heterogeneity of 16S rRNA is also attributable to horizontal gene transfer in some bacteria, such as Streptomyces, Thermomonospora chromogena, and Mycobacterium celatum (37, 45, 49). In this study, we provide the first example in support of a likely horizontal transfer of the tuf gene encoding EF-Tu. This may be an exception since stringent functional constraints do not allow for frequent horizontal transfer of the tuf gene as with other genes. However, enterococcal tuf genes should not be the only such exception as we have noticed that the phylogeny of Streptomyces tuf genes is at least as complex as that of enterococci. For example, the three tuf-like genes in one high-G+C-content gram-positive bacterium, S. ramocissimus, branched with the tuf genes of phylogenetically divergent groups of bacteria (Fig. 2). Another example may be the tuf genes in clostridia, which represent a phylogenetically very broad range of organisms and form a plethora of lines and groups of various complexities and depths. Four species belonging to three different clusters within the genus Clostridium have been shown by Southern hybridization to carry two copies of the tuf gene (39). Further sequence data and phylogenetic analysis may help in interpreting the evolution of EF-Tu in these gram-positive bacteria. Since the tuf genes and 16S rRNA genes are often used for phylogenetic study, the existence of duplicate genes originating from horizontal gene transfer may alter the phylogeny of microorganisms when the laterally acquired copy of the gene is used for such analyses. Hence, caution should be taken in interpreting phylogenetic data. In addition, the two tuf genes in enterococci have evolved separately and are distantly related to each other phylogenetically. The enterococcal tufB genes are less conserved and unique to the 11 enterococcal species. We previously demonstrated that the enterococcal tufA genes could serve as a target to develop a DNA-based assay for identification of enterococci (25). The enterococcal tufB genes would also be useful in the identification of these 11 enterococcal species.