A Tn7-Like Transposon Is Present in the glmUS Region of the Obligately Chemoautolithotrophic Bacterium Thiobacillus ferrooxidans

Abstract

The region downstream of the Thiobacillus ferrooxidans ATCC 33020 atp operon was examined, and the genes encoding N-acetylglucosamine-1-uridyltransferase (glmU) and glucosamine synthetase (glmS) were found. This atpEFHAGDC-glmUS gene order is identical to that of Escherichia coli. The T. ferrooxidans glmS gene was shown to complement E. coli glmS mutants for growth on minimal medium lacking glucosamine. A Tn7-like transposon, Tn5468, was found inserted into the region immediately downstream of the glmS gene in a manner similar to the site-specific insertion of transposon Tn7 within the termination region of the E. coli glmS gene. Tn5468 was sequenced, and Tn7-like terminal repeat sequences as well as several open reading frames which are related to the Tn7 transposition genes tnsA, tnsB, tnsC, and tnsD were found. Tn5468 is the closest relative of Tn7 to have been characterized to date. Southern blot hybridization indicated that a similar or identical transposon was present in three T. ferrooxidans strains isolated from different parts of the world but not in two Thiobacillus thiooxidans strains or a Leptospirillum ferrooxidans strain. Since T. ferrooxidans is an obligately acidophilic autotroph and E. coli is a heterotroph, ancestors of the Tn7-like transposons must have been active in a variety of physiologically different bacteria so that their descendants are now found in bacteria that occupy very different ecological niches.

Thiobacillus ferrooxidans is an acidophilic (optimum pH, 1.6 to 2.2), obligately chemoautolithotroph and is a member of a consortium of physiologically related bacteria used in processes to recover metals such as copper, uranium, and gold from ores (4, 17, 21). Because of its interesting physiology and industrial use there has been considerable interest in this bacterium. The T. ferrooxidans atp genes had previously been cloned and sequenced (5). Limited sequencing of the region downstream of the atp operon (unpublished) had indicated that the gene order atpEFHAGDC-glmUS as found in Escherichia coli (19, 29) was the same for T. ferrooxidans. In E. coli the pst operon, which encodes a phosphate-specific transport system, lies downstream of the glmUS genes. Phosphate nutrition of T. ferrooxidans is of particular interest, as phosphate is one of the few nutrients added to industrial gold-bearing arsenopyrite ore treatment processes (21). Since the gene order of E. coli and T. ferrooxidans was conserved in this region, we wished to determine whether this conservation extended to the pst operon.

Isolation of T. ferrooxidans genomic DNA downstream of the atp operon.

Cosmid p818.1 (Fig. 1) was isolated from a cosmid gene library of T. ferrooxidans ATCC 33020 by its ability to complement an E. coli uncD mutant for growth on minimal medium plus succinate according to the procedure reported by Brown et al. (5). The cosmid was mapped for restriction endonuclease sites and appeared to extend approximately 35 kb downstream of the T. ferrooxidans atp operon. In order to demonstrate that the insert in cosmid p818.1 represented unrearranged chromosomal DNA from T. ferrooxidans, the insert from subclone p818.52 (Fig. 1) was labelled by using a digoxigenin (DIG) DNA labelling kit (Boehringer Mannheim) and used in a Southern blot hybridization experiment (Fig. 2). Hybridizations were performed at 68°C for 18 h followed by two washings at 22°C in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) plus 0.1% sodium dodecyl sulfate and two washings at 65°C in 0.1× SSC plus 0.1% sodium dodecyl sulfate. Cosmid p818.1 and T. ferrooxidans genomic DNA gave positive hybridization signals to fragments of identical size when digested with the same restriction enzyme. BamHI digests gave signals at 2.5 and 2.8 kb (Fig. 2, lanes 1 and 2), HindIII digests gave signals at about 10 kb (lanes 3 and 4), and BglII digests gave signals at 5.7 kb (lanes 5 and 6). The additional weak signals in the p818.1 lanes arose because the KpnI-SalI probe from p818.52 had a small quantity of contaminating vector (pBluescript KS) which has regions of homology to the cosmid vector (pHC79). This confirmed that the region extending from the BglII site shown at 37 kb to the HindIII site at 25.5 kb of p818.1 was unrearranged chromosomal DNA from T. ferrooxidans ATCC 33020.

FIG. 1.

A map of cosmid p818.1 and subclones showing the location of Tn5468 as well as the direction of the genes and ORFs downstream of the atp operon. Abbreviations for restriction endonuclease sites are as follows: A, ApaI; Ba, BamHI; Bg, BglII; E, EcoRI; H, HindIII; K, KpnI; and S, SalI. The region sequenced in both directions extended from H* at 25.5 kb to Ba* at 38.5 kb. The pound sign and open arrow denote the direction of limited single-strand sequencing which identified the position and direction of the spoT gene. R, right.

FIG. 2.

Hybridization of DIG-labelled p818.52 to cosmid p818.1 and total DNA from strains of T. ferrooxidans, T. thiooxidans, and L. ferrooxidans. DNA in lanes 1 and 2 was digested with BamHI; that in lanes 3 and 4 was digested with HindIII, and that in lanes 5 to 11 was digested with BglII.

Location of glmUS genes and glmS gene complementation.

The 3.5-kb BamHI-BamHI fragment downstream of the T. ferrooxidans operon was cloned into vector pUCBM21 to give p818.16 (Fig. 1). This fragment was sequenced in both directions by the dideoxy chain termination method by using a Sequenase DNA sequencing kit (U.S. Biochemical Corporation, Cleveland, Ohio). This region contained a partial open reading frame (ORF) of 182 amino acids and complete ORFs of 711 and 277 amino acids. The partial ORF had a high percent amino acid sequence identity to the C terminus of the glmU gene products of E. coli (59.1%; SWISS-PROT accession no. P17114), Haemophilus influenzae (58.0%; SWISS-PROT accession no. P43889), Neisseria meningitidis (50.8% [27]), and Bacillus subtilis (43.5%; SWISS-PROT accession no. P14192). The first complete ORF had amino acid sequence identity with the glmS gene products of E. coli (47.8%; SWISS-PROT accession no. P17169), H. influenzae (51.9%; SWISS-PROT accession no. P44708), B. subtilis (39.1%; SWISS-PROT accession no. P39754), Saccharomyces cerevisiae (39.4%; SWISS-PROT accession no. P14742), and Mycobacterium leprae (42.2%; SWISS-PROT accession no. P40831) as well as to the nodulation protein M sequences of Rhizobium meliloti (43.6%; SWISS-PROT accession no. P25195) and Rhizobium leguminosarum (44.0%; SWISS-PROT accession no. P08633).

Glucosamine-6-phosphate synthase (encoded by the glmS gene) and N-acetylglucosamine-1-uridyltransferase (encoded by the glmU gene) are two of the enzymes required for the synthesis of UDP-N-acetylglucosamine, a major precursor of the bacterial cell wall peptidoglycan. E. coli glmS mutants such as strain CGSC5392 (provided by B. Bachmann; originally isolated as E110 by Wu and Wu [30]) are not capable of growth in the absence of added glucosamine. To test whether the T. ferrooxidans glmS gene was functional in E. coli, plasmid p818.16 (Fig. 1), which has the T. ferrooxidans glmS gene cloned behind a pUCBM21 vector promoter, was transformed into competent E. coli CGSC5392 cells and plated onto Luria agar plus ampicillin (100 μg/ml). E. coli CGSC5392(p818.16) cells produced medium-sized colonies on Luria agar, whereas E. coli CGSC5392(pUCBM21) cells, which contained vector but no insert, produced pinprick-sized colonies. Large colonies were produced by both types of cells when d-glucosamine (200 μg/ml) was added to the plates. E. coli(p818.20) cells, which contained 10 kb of T. ferrooxidans DNA, including the glmU and glmS genes (Fig. 1), failed to produce large colonies in the absence of added d-glucosamine. This suggests that the T. ferrooxidans glmS gene is not expressed from its own promoter in E. coli.

Location of Tn5468.

The region from BamHI (35.2 kb) to HindIII (25.5 kb) (Fig. 1) was determined by double-stranded DNA automated sequencing using fluorescently labelled nucleotides; all regions were sequenced multiple times from both strands. The region from SalI (near tnsA) to EcoRI (in the N terminus of tnsD) was dissected by using a primer walking strategy. The region from EcoRI (tnsD) to HindIII was determined by generating a number of mini-Tn7 element insertions in this area by using an in vitro transposition system (23). After transformation, individual mini-Tn7 insertions were isolated and the sequences of the Tn5468 DNA flanking the newly inserted mini-Tn7 was determined with primers specific for the ends of Tn7.

Analysis of the Tn7-like transposon, Tn5468.

Analysis of the newly sequenced region also revealed a DNA segment, now designated Tn5468 (Fig. 1), immediately downstream of the T. ferrooxidans glmS gene, that has considerable similarity to Tn7. Tn7 is a bacterial transposon (6, 7) which inserts site specifically downstream of glmS in other bacteria such as E. coli that are physiologically very distantly related to T. ferrooxidans. The finding of closely related transposable elements in the distinct ecological niches of T. ferrooxidans and E. coli raises interesting questions about the evolution and dispersal of transposable elements (see below).

Site-specific insertion of Tn7 downstream of glmS proceeds via recognition of a particular DNA sequence in the COOH region of glmS (3, 11); the actual point of element insertion is 23 bp downstream of the glmS termination codon. Such a target selection strategy allows the transposon to use information within the highly conserved glmS gene to identify a particularly favorable target site; however, this strategy also avoids disruption of glmS through the insertion of Tn7 of that gene itself. Such a site-specific insertion provides a safe haven for Tn7 in bacterial chromosomes (6, 7). The biochemical roles of the Tn7 transposition proteins (Tns proteins) have been established. TnsB specifically recognizes the ends of Tn7 by binding to multiple repeats of its recognition sequence that are present in each end (1, 25). TnsA and TnsB then collaborate in an interdependent fashion to form the transposase, which executes the chemical steps of strand exchange (18, 22). The DNA breakage and joining activity of the Tn7 TnsA+B transposase is not constitutive; rather, transposase activity depends upon activation of TnsA+B by TnsC, an ATP-dependent DNA binding protein (9, 24). The transposase activation activity of TnsC is, in turn, modulated by other Tns proteins that interact with target DNA (3, 23, 24). For example, the interaction of TnsC with TnsD, a protein that binds specifically to DNA sequences in glmS, mediates the site-specific insertion of Tn7 just downstream of this gene (3).

Tn5468 closely resembles Tn7 in structure and organization (Fig. 3), encoding four ORFs that are clearly similar to the Tn7 transposition proteins TnsA, TnsB, TnsC, and TnsD; also present in Tn5468 is another ORF that is not obviously related to sequences currently present in databases. In the right end of both Tn7 and Tn5468, there are four closely juxtaposed repeats that are separated by 8 bp from the extreme tip of the transposon (Fig. 4); these repeats in Tn7 have been shown to be DNA sequences necessary in cis for Tn7 transposition (2, 8). It has also been shown that in Tn7 the identity of the terminal 5′-ACA-3′ is a key to breakage and joining activities (9a, 10). There is a very similar pattern of repeats and terminal sequences in the right end of Tn5468, providing evidence that Tn5468 is indeed a Tn7-like transposon (Fig. 4). In Tn7, the left end also contains this 8-bp terminal sequence and multiple repeats (2); however, no such sequences are evident in the region between the T. ferrooxidans recG and the Tn5468 ORF, suggesting that this element may have become truncated in some way. The repeat sequences at the ends of Tn7 are binding sites for TnsB protein (1, 25). The sequence similarity of repeats (Fig. 4) and of Tn7 and Tn5468 TnsB’s (Fig. 3) suggest that these T. ferrooxidans repeats are binding sites for T. ferrooxidans TnsB and thereby define the functional terminus of Tn5468. Tn5468 TnsB also conserves several acidic residues which have been shown to be essential to the transposase activity of Tn7 TnsB (22). In Tn7, TnsA and TnsB collaborate interdependently to form the transposase (18, 22); these TnsB proteins are also related to the transposases of a variety of mobile elements, as exemplified by comparison to the transposase of Tn5053, another element distantly related to Tn7, which is found in some species of Xanthomonas and Klebsiella (14, 20). Tn5468 also encodes a protein highly related to Tn7 TnsA, in which key residues (D134 and E169 [E. coli numbering]) that have been shown to be essential to transposase activity are conserved (18, 22). These TnsB’s and TnsA’s are all likely members of the human immunodeficiency virus integrase superfamily (14, 20, 22). Tn5468 also contains ORFs that are related to TsnC, containing conserved purine nucleotide binding motifs (9), and to TnsD, which binds specifically to particular sequences in glmS (3) and thus mediates site-specific insertion (Fig. 5). The similarities support the view that Tn5468 is a Tn7-like element that at some time transposed site specifically to a position just downstream of T. ferrooxidans glmS by a mechanism highly related to the Tn7 transposition mechanism.

FIG. 3.

Comparison of Tn5468 and Tn7, indicating the sizes and relatedness of the predicted gene products. Tn5468 is shown in its position within the T. ferrooxidans ATCC 33020 chromosome, with the right border of Tn5468 indicated by an arrow and the truncated, left-hand region indicated by oblique parallel lines. Percent amino acid (aa) identities and similarities were calculated according to the same criteria as those used by the BLAST subroutine (NCBI database), with the length of the smaller protein as the denominator. ∗, the similarities between tnsE of Tn7 and the unidentified ORF of Tn5468 were apparent only within the N-terminal halves of the proteins.

FIG. 4.

The sequences at the right (R) ends of several Tn7-like elements, Tn5468 from T. ferrooxidans, Tn7 from E. coli, and TnE-15, a partial element also found in T. ferrooxidans, are shown. Inspection of the sequences reveals the presence of a 20-bp sequence present in four closely juxtaposed copies in the right end of each element. In Tn7, these right-end repeats and similar repeats in the left end of Tn7 are required for transposition; the Tn7 20-bp repeat is a binding site for the element-encoded TnsB protein. The terminus of each element consists of an 8-bp segment ending in ACA-3′; the terminal A is the 3′ end of the transposon. Analysis of this segment in Tn7 suggests that the ACA sequence is critical to provoke DNA breakage and joining; the other 5-bp segment likely serves as a spacer between the B binding site and the ACA sequence at the terminus (9a). Lowercase letters in the consensus sequence represent the following: r, purine; y, pyrimidine; w, A or T; m, A or C.

FIG. 5.

The locations of the right (R) ends of Tn5468 and Tn7 with respect to the chromosomal glmS genes are shown. In both cases, the element has been inserted about 20 bp downstream from the COOH end of the glmS gene. The E. coli sequence (lower line) is numbered, with the center of the 5 bp duplicated upon Tn7 insertion designated 0. Also shown is the E. coli sequence necessary for TnsD binding to attTn7 (3). Although the glmS sequences, i.e., the TnsD binding sites, are highly conserved between E. coli and T. ferrooxidans, the sequences between the point of element insertion and the end of glmS are not. Other analysis of attTn7 in E. coli has shown that sequence alterations in this region, even at the point of Tn7 insertion itself, have little effect on attTn7 target activity (3).

It should be noted that Tn7 can also use a different class of target sites when transposition is mediated by a slightly different ensemble of Tns proteins, TnsABC+E, TnsE being an alternative target selector (15, 28). Inspection of the Tn5468 sequences has not revealed a gene like that encoding TnsE. An interesting hypothesis is that the unidentified ORF of Tn5468 provides yet another target selection protein that could promote insertion into other types of sites.

Distribution of Tn7-like transposons among other acidophilic chemoautolithotrophs.

Evidence of the activity of another Tn7-like transposon has been found in T. ferrooxidans E-15 (12). In this strain there appears to have been a duplication of the merC and merR genes, which are separated by a region with Tn7-like terminal repeat sequences and a partial ORF with strong similarity to the Tn7 tnsA gene. The insertion of the Tn7-like transposon in T. ferrooxidans E-15 is not associated with the glmS termination region. Although both the T. ferrooxidans ATCC 33020 and E-15 strains have Tn7-like transposon sequences, they are clearly different from each other and Tn7. The four repeated sequences of the three transposons are compared in Fig. 4.

We wished to determine whether transposons related to Tn5468 were present in other T. ferrooxidans isolates and in other bacteria which grow in a similar environment. Thiobacillus thiooxidans and Leptospirillum ferrooxidans are physiologically similar to T. ferrooxidans in that they are all gram-negative, obligately autotrophic, acidophilic, chemolithotrophic bacteria (21). Chromosomal DNA was prepared from three strains of T. ferrooxidans, ATCC 33020, ATCC 19859, and ATCC 22370 (type strain); two strains of T. thiooxidans, DSM 504 and ATCC 19377 (type strain); and L. ferrooxidans DSM 2705 (type strain). This genomic DNA was digested with BglII, and the fragments were separated on an agarose gel, Southern blotted onto a Hybond N⁺ nylon membrane, and hybridized to DIG-labelled p818.52 probe DNA as described earlier. The probe hybridized to a 5.7-kb BglII fragment from all three of the T. ferrooxidans isolates (Fig. 2, lanes 6 to 8) but not to either of the T. thiooxidans or the L. ferrooxidans strains (Fig. 2, lanes 9 to 11). The T. ferrooxidans ATCC 33020 strain was isolated in Japan and genomic DNA was prepared from cells grown in iron medium in a laboratory in South Africa, while the ATCC 19859 and ATCC 22370 strains originated from Canada and the United States, respectively, and DNA was prepared from cells grown on tetrathionate medium in a laboratory in Sweden. This eliminates the possibility that the T. ferrooxidans strains or chromosomal DNA had become mixed during growth or preparation. The glmS region of the T. ferrooxidans chromosome appears to be highly conserved and a transposon closely related to Tn5468 appears to be widely distributed as a single-copy insert in the genome of T. ferrooxidans. No sequences with similarity to the Tn5468 tnsBC sequence were detected in the isolates of T. thiooxidans or L. ferrooxidans tested.

Region downstream of Tn5468.

An ORF of 652 amino acids was present in the region downstream of Tn5468 in the opposite orientation to the Tn5468 genes (Fig. 1). Comparison of the predicted amino sequence with the GenBank database by using the BLAST subroutine (NCBI database) indicated the highest sequence similarities, 34 and 32% sequence identity, to the recG gene products of E. coli and H. influenzae, respectively. Limited sequencing further downstream indicated a region of high similarity, 64 and 59% amino acid identity (over 85 amino acids), to the spoT gene products from E. coli and H. influenzae, respectively. In both E. coli and H. influenzae, the recG (RecG is an ATP-dependent DNA helicase) and spoT (SpoT is a guanosine-3′,5′-bis diphosphate 3′-pyrophosphatase) genes are linked and together with spoU and spoS form the spo operon (13, 16). In T. ferrooxidans the situation appears different as the spoT and recG genes, although physically linked, are orientated in opposite directions and are therefore divergently translated.

No region with sequence similarity to the pst operon of E. coli was found, and there was insufficient space between the apparent end of Tn5468 and the RecG-like ORF of T. ferrooxidans to accommodate the spo genes.

Comparison of T. ferrooxidans atp-glm-Tn5468-spo region with other bacteria.

Given very different physiologies and ecological niches, E. coli and T. ferrooxidans show remarkable similarity in this region of their chromosome. The glmU and glmS genes are linked and located immediately downstream of the atp (unc) operon. In contrast, in H. influenzae and Helicobacter pylori (the only other bacteria in which the location of both atp and glmS genes have been reported), these genes are not physically linked on the chromosome (http://www.tigr.org).

The similarity between E. coli and T. ferrooxidans is further extended by the discovery of the Tn7-like transposon in the region downstream of the T. ferrooxidans glmS gene. Although specific Tn7 insertion sites have been identified in the chromosomes of a variety of bacteria as divergent as Agrobacterium tumefaciens, Caulobacter crescentus, Klebsiella pneumoniae, Pseudomonas aeruginosa, Rhizobium meliloti, Rhodopseudomonas capsulata, and Xanthomonas campestris (6), all these bacteria were deliberately exposed to Tn7 in the laboratory. In contrast, a transposon that is Tn7-like but clearly different from Tn7 has been active in natural isolates of T. ferrooxidans from different parts of the world. T. ferrooxidans is an obligately autotrophic bacterium whose growth is inhibited by low concentrations of most organic compounds, such as small amounts of free galactose in agar (26). As a result, much of the evolutionary development of T. ferrooxidans would be expected to have taken place with little physical contact or genetic exchange between it and typical heterotrophic bacteria such as those listed above. Tn7, Tn5468, and the Tn7-like transposon of T. ferrooxidans E-15 (12) must have originated from a common ancestor which was active in bacteria long before they became as physiologically divergent as T. ferrooxidans and E. coli are today.

Nucleotide sequence accession number.

The 13-kb sequence between the BamHI and HindIII sites indicated by asterisks (Fig. 1) has been assigned GenBank accession no. AFO32884.