INTRODUCTION

Leptospira interrogans is the etiologic agent of lepto-spirosis, which is a worldwide zoonosis with a much greater incidence in tropical and subtropical regions4. The recently reported genomic sequence of the L. interrogans serogroup Icterohaemorrhagiae serovar lai 5 indicated that in contrast to the two strictly parasitic spirochetes, Treponema pallidum6 and Borrelia burg-dorferi7, L. interrogans possessed much more physiologically important genes for its facultative free-living metabolism, allowing it to adapt to its diverse environment. However, the regulation of the physiology is largely unknown.

Toxin-antitoxin (TA) modules are of importance among bacterial regulatory systems8. They were firstly found on plasmids and considered prevalently to be associated with stable plasmid inheritance at cell division9, 10, 11. The well-known plasmid-encoded TA loci contained CcdA/CcdB of F factors12, 13, Kis/Kid of R114, Phd/Doc of P115, ParD/ParE of RK216, pas locus of pTF-CF2 17, ω-ε-ζ operon of pSM1903518, stb locus of pMY-SH600019 and relBE locus of P30720.

Recently, TA modules were also identified on chromosomes and considered to be associated with host stress response21. Only three TA loci families encoded by chromosomes have been identified so far, relBE2, mazEF 1, and chpIK3. Generally, TA loci are organized into operons where the first cistron encodes a small instable antitoxin while the second cistron encodes a large stable antitoxin21.

The presence of a TA module in L. interrogans was firstly reported by Picardeau et al3 by characterizing the large chromosome (CI) encoded chp locus (chpI-chpK). It was proposed that the presence of this system might provide means for bacterial adaptation to a poor nutrition environment. Further analysis of the L. interrogans genome found a mazEF locus (LA1780/LA1781) belonging to the TA system. When a pair of vapBC (virulence-associated protein) genes were cloned and expressed in Escherichia coli for functional analysis, we noticed that the characteristics of these genes and their products were similar to that of the TA systems, but unlikely to be associated with virulence. Based on their sequence homology and operon organization, a novel group of TA system was characterized.

MATERIALS AND METHODS

Materials

Plasmid pET28b (+) from Novagen (Madison, Wisconsin, USA) was used for heterogeneous gene expression. Golden Taq DNA polymerase, BigDye Terminator Cycle Sequencing Ready Reaction Kit, restriction enzymes, T4 DNA ligase, and bacterial alkaline phosphatase were obtained from Takara (China). Trizol reagent and Freund's complete and incomplete adjuvant was obtained from GBICO-BRL (USA). Vector plasmid pUCm-T, AMV Reverse Transcription System, goat anti-rabbit immuno-globulin G (Fc)-Alkaline Phosphatase conjugated and BCIP/NBT color development substrate were purchased from Promega (USA). Nitrocellulose membranes were obtained from Amersham (USA). All other reagents were of the highest purity available.

Bacterial strains and culture conditions

Leptospira strains used in this work are shown in Tab 1. The avirulent L. interrogans strain of serovar lai was kindly supplied by Prof. Saint Girons Unité de Bactériologie Moléculaire et Médicale, Institut Pasteur, France. All of the other strains were maintained by the Institute for Infectious Disease Control and Prevention (IIDC), Chinese Center for Disease Control and Prevention (CCDC), Beijing, China. The strains were cultured in EMJH(Johnson and Marris modification of the Ellinghausen and McCullough medium4) at 28°C for 3-4 d under aerobic conditions until mid-log phase was reached (about 108 bacteria/ml). Bacterial counts were determined by dark-field microscopy. E. coli strains DH5α and BL21(DE3/pLysS) used as hosts for cloning or expression were cultured in LB (Luria-Bertani) medium.

Table 1 Leptospira strains used in this study
Full size table

Databases and softwares

The complete genomic sequence of the virulent L. interrogans serogroup Icterohaemorrhagiae serovar lai was obtained from http://www.chgc.sh.cn/lep/. BLAST searching the NCBI non-redundant nucleotide database, or non-redundant protein database on SwissProt/TrEMBL were used to annotate the L. interrogans vapBC and mazEF genes. Bacterial protein sequences of mazEF, vapBC, chpIK and relBE were obtained from the genome database of KEGG by employing the TFASTA search (http://fasta.genome.ad.jp/ideas/fasta/tfasta_genome.html) with inputting VapB, VapC, MazE, MazF, ChpI and ChpK protein sequences of L. interrogans. Only those with similarity scores above e-8 were selected for further analysis. Multiple-sequence alignment was accomplished by using Bioedit. Phylogenetic trees were constructed using Mega 2.0.

Expressing vapB and vapC genes in E. coli and protein purification

PCR primers for L. interrogans vapB, vapC and vapBC genes were designed as shown in Tab 2. Forward primers incorporated an Nde I restriction site and reverse primers incorporated a Bam HI restriction site. The reverse primers were designed to be immediately behind the terminator “TTA” to ensure involvement of the terminator in the PCR product. All the genes were amplified using Leptospira spp. genomic DNA as templates, which were extracted from mid-log phase cultures of corresponding strains grown in EMJH medium22. PCR was performed using the following cycle; 95°C for 10 min, followed by 30 cycles at 94°C for 30 sec for denaturing, 54°C for 30 sec for vapB or 58°C for 1 min for vapC and vapBC for annealing, and 72°C for 1 min for reaction. The whole reaction was completed at 72°C for 10 min for elongation followed by 4°C for storage. Procedures for gene cloning, heterogeneous expression and protein purification were manipulated according to the manufacturer's manual. Plasmids constructed for this work are listed in Tab 3.

Table 2 Oligonucleotide primers specific for vapB and vapC used in the PCR and RT-PCR reactions
Full size table
Table 3 Vectors used in this study and plasmids with cloned genes
Full size table

Total RNA extraction, reverse transcription-PCR (RT-PCR)

Total RNA was extracted from Leptospira strains cultured in EMJH medium for 24, 48 (log-phase), or144 h using Trizol Reagent (GBICO-BRL). The AMV Reverse Transcription System (Promega) was used for reverse transcription experiments. RT-PCR primers were designed to simultaneously detect the 3' portion of the vapB gene together with the 5' portion of the corresponding vapC gene (as shown in Tab 2). Reverse transcription synthesized first-strand cDNA was used as the PCR template. The amount of template cDNA used was 10-15 ng per assay. The PCR protocol was as described above, except for the annealing step which was done at 54°C for 1 min.

Bacterial growth curve detection

Growth curves of the E. coli strains BL21(DE3/pLysS) containing pTB, pTC, pTBC were determined by measuring the O.D600nm. Cells were initially grown in LB overnight with appropriate antibiotics (kanamycin, 50 μg/ml) and then, transferred into fresh LB at 1% inoculums in the presence of IPTG (1 mM).

Plasmid stability test

Single colonies of E. coli DH5α carrying plasmid pUC19, pUB, pUC, or pUBC were picked from LB plates containing ampicillin (50 μg/ml) and innoculated into 5 ml of LB without ampicillin. After shaking at 37°C, 220 rpm for 12 h, cells were reinnoculated into 5 ml of fresh LB without ampicillin, but containing 1 mM IPTG (1% innoculum) for a further 12 h. This kind of reinnoculation and cultivation was continued for up to 336 h. 100 μl samples were collected every 24 h, diluted and plated onto LB-plates, with or without ampicillin (50 μg/ml). Viable colonies were counted after 12 h incubation at 37°C. The ratio of the AmpR colonies versus the total viable colony counts was used to estimate the percentage of plasmid maintained in the population.

RESULTS

The expression level of vapC in E. coli without or with the co-expression of vapB differed significantly

Among the 4,769 predicted genes of L. interrogans (based on its genomic sequence – http://www.chgc.sh.cn/lep/), a pair of closely linked CDS (LA1002/LA1001) was shown to be strikingly similar to the bacterial virulence association proteins (VapB/VapC) 23, as well as a plasmid maintenance stability determinant locus, (stb) from Shigella flexneri19.

The leptospira vapBC locus consists of an upstream 231 bp CDS (vapB) that encodes a putative protein of 77 amino acids and a downstream 399 bp CDS (vapC) that encodes a putative protein of 133 amino acids (Fig 1). The translational start codon for vapC overlaps with the last base of the translational stop codon of vapB, which is a strong indication of translational coupling. This vapBC operon structure is strikingly similar to that of the three known TA operons, relBE1, 24, mazEF1 and chp3, 25. A putative promoter region was revealed by the presence of a pair of 9 bp inverted repeats (IR). This shares the same characteristics with IRs of the E. coli chp promoter, known for specific DNA binding of the Chp/Pem protein26. Both the -10 and -35 regions for the promoter and ribosomal binding site (RBS) are typical for bacteria (Fig 2).

Figure 1
figure 1

Multiple-sequence alignment of 18 VapC-homolog proteins from L. interrogans and other bacteria with similarity scores above e-8. Only the VapC proteins with obvious upstream VapB partners were included in the alignment analysis. Identical amino acids are shown in black, while conserved amino acids are shown in grey. The alignment was accomplished using Bioedit software. Gaps introduced to maximize the similarity alignment are indicated by dashes. The bacterial abbreviations used are listed in Tab 4.

Full size image
Figure 2
figure 2

Schematic representation of the putative promoter region of the chromosome-encoded vapBC of L. interrogans. Single arrows show all of the translation directions. The overlap of the stop codon of vapB with the start codon of vapC is enlarged. The 9-bp inverted repeats (IRs) are boxed and highlighted by two head-to-head arrows. Typical -10 and -35 regions as well as the ribosome-binding site (RBS) are marked and the vapB start coden is bolded.

Full size image

In order to characterize this pair of predicted proteins, we cloned the vapB and vapC genes into pET expression vectors, either individually or together (Tab 3) and attempted to express the genes in E. coli BL21(DE3 pLysS) (Novagen). The recombinant proteins were expressed after IPTG induction and shown to have a molecular weight (MW) of 10.7 kD for VapB with his-tag and 17.1 kD for VapC with his-tag (Fig 3). The expression level of VapC was significantly lower than that of its partner VapB if expressed alone. However, when these two proteins were expressed simultaneously and coordinately from a construct containing their original operon with the characteristic coupled cistronic structure (Fig 2), the expression levels of both were high and essentially the same (Fig 3).

Figure 3
figure 3

Proteins encoded by vapB and vapC clones expressed in E.coli BL21 (DE3/pLysS). The proteins were expressed as described in Materials and Methods. The arrows indicate each expressed His-tag recombinant protein.

Full size image

The growth of the E. coli strain carrying the vapC expression plasmid was restrained but was relieved with the co-expression of the vapB gene

The growth of the E. coli carrying plasmid pTC (vapC expression plasmid) sharply decreased in the presence of IPTG compared to that of the hosts with plasmids pTB (vapB expression plasmid), pTBC (vapBC coupled expression plasmid) or pET28b (the negative control vector) (Fig 4). Quantitatively, there was no difference in their doubling times (44 min for pTC, 47 min for pTB, 45 min for pTBC and 46 min for pET28b). However, the difference was significant for the length of the lag phase (248 min for pTC but 28 min for pTB, 40 min for pTBC, and 34 min for pET28b). This difference directly leads to the difference in the speed of accumulation of biomass. At the 14th h of cultivation, the average OD600 reading of E. coli harboring pTC (1.15) was much lower than that of all the other 3 strains (1.43 for pTB, 1.49 for pTBC and 1.51 for pET28b).

Figure 4
figure 4

Effect of the over-expression of the L. interrogans vapC in the presence or absence of vapB on growth of E. coli cells harboring the expression plasmids. Symbols used for the plasmids transformed to the E. coli hosts: pET28b (□), pTB (▴), pTC (▪), pTBC (▾). Growth was determined by measuring the optical density at 600 nm.

Full size image

The vapBC locus is conserved in different species or serovars of Leptospira

The presence of the vapBC locus and the expression of this operon was tested in various kinds of pathogenic or saprophytic leptospires (Tab 1). PCR amplification of vapBC genes from genomic DNA templates of the experimental strains showed that both the 231 bp vapB fragment and the 399 bp vapC were present in all the strains tested (data not shown). Subsequent DNA sequencing results showed that the strains tested had identical gene fragments in vapB and vapC (data not shown). RT-PCR experiments employing total RNA isolated from cells grown on EMJH medium with gene specific primers (see Materials and Methods) showed that these two closely linked genes were transcribed simultaneously in all the Leptospira strains tested (data not shown).

The L. interrogans vapBC system is effective in stabilizing pUC plasmid in E. coli

The stability of a series of pUC19 derived plasmids con-taining vapB (pUB), vapC (pUC), or vapBC (pUBC) genes with the cloned genes expressed by the lac promotor of pUC19 were measured in E. coli DH5α (see Materials and Methods). As shown in Fig 6, the percentage of ampicilin resistant cells in the population of E. coli carrying pUC19 decreased. On the other hand, the majority of E. coli cells carrying pUBC with vapBC were relatively stable after prelonged incubation without any selection pressure. Neither pUC nor pUB was stable under the same conditions, although the rate of plasmid lost was slightly improved.

Figure 6
figure 6

Genetic stabilization effect of the vapBC system upon the high-copy unstable plasmid, pUC19. Plasmids used were derivatives of pUC19: pUB (carrying vapB), pUC (carrying vapC), pUBC (carrying vapBC) and pUC19 (blank control). Plasmid construction and experimental details are described in Materials and Methods.

Full size image

DISCUSSION

The vap module was first found in the genome of the strictly anaerobic bacterium, Dichelobacter nodosus (Bacteroides nodosus)18, a major pathogen implicated in foot rot (a highly contagious disease of sheep). Hybridization experiments showed that the vap module was present in all the genomes of virulent isolates, while it was absent in 67% of the benign isolates. Therefore, it was named the virulent-associated region and the hypothetical proteins encoded in this region were designated as virulent-associated proteins (vap), although their definitive functions were unknown.

Subsequently, many vapBC loci were found in other bacteria (Fig 1, Tab 4), encoded by either chromosomes or plasmids and in particular, the stb locus encoded on the large virulent plasmid pMYSH6000 of S. flexneri, are both structurally and functionally similar to the vap system. By studying the physiology of the overexpression of L. interrogans vapBC genes in E. coli, we recognized that, similar to the case of L. interrogans chpIK operon3, vapBC is a novel type of toxin-antitoxin system. In contrast to B. nodosus, the vapBC operon is widely present and expressed in leptospires. Therefore, it is more likely related to physiological regulation rather than to the virulence of the bacterium.

Table 4 The chromosome-encoded TA systems distributed in microorganisms
Full size table

Deduced amino acid sequence alignment of the VapC proteins indicated that besides the general similarity among the family members, highly conserved specific sites were observed (Fig 1). All the VapC proteins contain a common PIN domain (IPR002716, InterPro) putatively responsible for nucleic acid binding, which might infer the possibility of a common cellular target for the Vap TA system.

Further similarity searches and comparisons in the microbial genomic database with respect to TA systems revealed a total of four types encoded by bacterial chromo-somes. They are the relBE, mazEF, chpIK, and vapBC loci (Tab 4). Except for the chpIK locus, all the others are also found in plasmids (Tab 4).

It is interesting to note that several TA systems, or multiple copies of one TA system, may present in one genome (Tab 4). In this connection, L. interrogans stands out as the only bacterium that possesses three TA systems, the chpIK previously identified3, the vapBC characterized in this study, and the mazEF (LA1780/LA1781) recognized by homology searches. It is known21 that TA systems encoded by plasmids prevent plasmid loss at cell division, whereas TA loci found on chromosomes are associated with species stability or cell programmed death (bacterial apoptosis). Further studies have revealed the specific molecular targets of some of the toxins, i.e., CcdB inhibits DNA gyrase, while PemK/Kid inhibits DNA replication, presumably via interaction with the DnaB helicase27. The RelE proteins encoded by the E. coli K-12 chromosome presumably inhibit translation2, 20, but their specific target(s) within the translational machinery is not yet known. Recent studies have shown that the ChpK proteins of E.coli inhibit cell growth by cleavage of mRNA 28. Similarly, RelE toxin inhibits protein synthesis by cleavage of ribosome bound mRNA in response to nutritional stress29. These findings further support the hypothesis proposed by Gerdes21. Toxins from the TA gene families, the RelE, VapC, MazF and ChpK, may mediate inhibition of translation or DNA replication21 as part of the global cellular response to environmental stress, rather than being simply cell-killing. The presence of multiple TA systems in one particular genome, as shown in the case of L. interrogans, may imply the complexity of regulation during the transition of metabolism.

Co-evolution is usually represented by a change in genetic composition of one gene in response to the genetic change in another30. The translational coupling of two closely linked genes with a strongly interactive functional relationship to the proteins they encoded, such as the case of TA system, seems to enforce the strongest selection pressure for co-evolution. Amino acid sequence similarity and clustering analysis of each group of the TA system were carried out and the results were compared to their corresponding 16s RNA phylogenetic trees (Fig 5). In general, the co-evolution between toxins and their corresponding antitoxins is obvious, while the development of these TA systems along with their speciation route varies from case to case. It is also interesting to note that homologous analysis could spontaneously put the toxin sequences into four groups, but not for their antitoxins (data not shown). These observations indicate that quite a few of the TA systems did not originally exist in the species, i.e., horizontal gene transfer might account for significant portions of the event. Therefore, co-evolution of the toxin genes, as well as the antitoxin genes with the 16s RNA, is only observed in part of the species as individual groups, but not along the whole family. This phenomenon is clearly illustrated in the case of the MazEF family (Fig 5A) where most of the Gram-negative bacteria except Nostoc, are grouped together as that of their speciation. However, Gram-positive bacteria are grouped separately into different Gram-negative bacterial clusters. Furthermore, even for each TA system, the toxin and antitoxin may not have originated from the same ancestor, or even been acquired simultaneously. This phenomenon is particularly obvious in case where multiple copies of one TA system existed in one strain. For instance, Rhizobium meliloti has two sets of VapBC systems encoded by its genome, but the phylogenetic position VapB1 only matches with that of VapC2 and vice versa. This implies not only an independent origination, but also a possible trans-acting mechanism between two TA modules of the same category (Fig 5B). In addition, it seems that if the evolving of a TA system among bacteria was a relatively homogeneous process, e.g., without gene duplication so that only a single copy of a TA module existed in one species (as is the case of chpIK) parallel evolution between the TA module and the 16s RNA could be observed (Fig 5C). A similar situation was observed in the family of RelBE, which consists of Gram-negative bacteria only (data not shown). This fact implies that the evolvement of relBE module might be late in the evolution history and it might have no chance to be spread into bacteria other than the Gram-negative ones. Finally, we noticed that in certain cases, even the strong co-evolutionary relationship between toxin and antitoxin was distorted. For instance, antitoxin was not found in the only archeae that possessed a toxin gene (vapC), Solfolobus tokodaii. This is interesting because it suggests that the function of the toxin-like proteins found in archeae might not necessarily be the same as originally thought. Alternatively, it is equally possible that although these proteins are toxic to the host, they might not be functionally expressed under the physiological conditions of these hosts. In either case, it suggests that antitoxins are likely introduced into the system later in the evolutionary path, and that the genetic signature of this “double evolving” event is still embedded in the coding sequence that is seen today.

Figure 5
figure 5figure 5figure 5

Unrooted phylogenetic trees for the amino acid sequences of bacterial antitoxin-toxin MazEF (panel A), VapBC (panel B) and ChpIK (panel C) and 16s rRNA sequences corresponding to each kind of TA system. In panel B, S. tokodaii was drawn as the root in the VapC tree because this archeae strain is the most primitive among all the bacteria in this group. As S. tokodaii does not have a vapB gene, R. conorii, the strain closet to S. tokodaii in the tree for VapC, was drawn as the root in the VapB tree. All the other trees were unrooted. The abbreviations for the bacterial species are listed in Tab 4.

Full size image

CONCLUSIONS

Our data have demonstrated that the vapBC locus of L.interrogans belongs to the TA family rather than encoding virulence-associated proteins. The product of the vapC gene is toxic to E. coli cells, while the product of the vapB gene counteracts VapC toxicity. The presence of the vapBC operon on an unstable plasmid under the control of a lac promoter can effectively prevent plasmid loss under non-selective conditions. A pair of 9 bp IRs in the promoter region of the L. interrogans vapBC operon share the same characterization with IRs of the E. coli chpB promoter region, indicating it might be a specific DNA binding site for VapC proteins. At present, the cellular target of VapC protein remains unknown. Further study on L. interrogans VapB and VapC could help in the understanding of the cellular target of these toxins and their physiological role in bacteria.