Correspondence: J. Tony Pembroke, Molecular and Structural Biochemistry Laboratory, Department of Chemical and Environmental Sciences, University of Limerick, Limerick, Ireland. Tel.:+353 61 202491; fax: +353 61 202568; e-mail: email@example.com
ICE R391, a prototype member of the SXT/R391 family of site-specific integrative conjugative elements (ICEs), frequently isolated from enterobacterial pathogens, exhibits an unusual, recA-dependent, UV-inducible, cell-sensitising function. This significantly decreases postirradiation cell survival rates in Escherichia coli host cells, a trait that would at first appear to be counterproductive in terms of adaptation to stress conditions. Construction and screening of a complete ICE R391 deletion library in E. coli identified three ICE R391 genes, orfs90/91, encoding a putative transcriptional enhancer, and orf43, encoding a putative type IV secretion system outer membrane-associated conjugative transfer protein, in the cell-sensitising function. Cloning and complementation of these genes confirmed their involvement in UV sensitising. Expression of both orfs90/91 and orf43 in wild-type E. coli indicated that orf43 encodes a cytotoxic gene product upon up-regulation. Deletion of the orf43 homologue in SXT, s050, also abolished its associated UV sensitisation. We hypothesise that ICE R391 and other members of the SXT/R391 family display decreased survival rates upon exposure to UV irradiation through the induction of orf43.
Integrative conjugative elements (ICEs) are a distinctive class of mobile elements characterised by an ability to mediate their own integration, excision and transfer from one host genome to another by a self-encoded mechanism of site-specific recombination, self-circularisation and conjugative transfer (Wozniak & Waldor, 2010). Currently, more than 430 ICEs have been identified in a variety of both Gram-positive and Gram-negative bacterial genomes (Bi et al., 2012). ICEs have a modular structure containing a highly conserved ‘core’ set of genes required for basic functionality: element maintenance in the host cell, integration into the host genome, excision out of the host genome and conjugative transfer between donor and recipient strains, that is interspersed with accessory gene modules that have likely been accumulated from exchanges with unrelated plasmids, prophage, transposons and other distantly related ICEs (Wozniak et al., 2009). These variable nonessential accessory genes are thought to be acquired by ICEs via insertion sequences, transposons and specific recombinases present within the core ICE genome and have been found to encode a wide range of functions beneficial to the host cell (Rice, 2002; Wozniak et al., 2009; Toleman & Walsh, 2011).
ICEs are subgrouped into families based on similarity between their core genes, specifically the integrase gene and the site of element integration on the host genome. One of the largest ICE families, the enterobacterial SXT/R391 group, has more than 74 assigned ICEs (Burrus et al., 2006; Bi et al., 2012). ICEs of the SXT/R391 family have a similar core genome with a highly homologous integrase gene, which promotes integration specifically into the 5′ end of prfC gene without gene truncation and restoring prfC gene function (McGrath & Pembroke, 2004; Burrus et al., 2006). SXT/R391 family members have been found integrated within the genome of several enterobacterial pathogens including Vibrio cholerae and Proteus mirabilis (Waldor et al., 1996; Mata et al., 2011). They commonly enhance fitness in these pathogenic strains by encoding beneficial genes including a diverse range of antibiotic and heavy metal–resistance determinants, biofilm-motility modulators and mutagenic DNA repair genes (Beaber et al., 2002; Boltner et al., 2002; Bordeleau et al., 2010). SXT/R391 family members may also function as promoters of bacterial diversity, as they are capable of mobilising nontransmissible genomic islands and virulence plasmids between hosts (Osorio et al., 2008; Daccord et al., 2010).
Unusually several ICEs of the SXT/R391 family including the prototype member ICEs R391 and SXT have also been shown to encode a unique UV-inducible cell-sensitising function, which significantly decreases postirradiation survival rates of host cells (Pembroke & Stevens, 1984; Kulaeva et al., 1995; McGrath et al., 2006). This unusual ICE-related cell-sensitising effect was found to be recA-dependent in Escherichia coli suggesting the ICE gene(s) responsible may affect a recA-controlled mechanism or be recA-inducible (Pembroke & Stevens, 1984). However, the specific ICE gene(s) involved and their possible mechanism of action have remained elusive. The ICE R391 was originally isolated from a clinical Providencia rettgeri strain in South Africa and assigned as a plasmid into a new incompatibility group, IncJ. However, on sequencing [GenBank: AY090559], it was confirmed as an ICE and to lack a plasmid replicon, be 89 kb in size and encode a mosaic structure of 96 putative genes related to plasmids, phage and transposon homologues (Boltner et al., 2002). Bioinformatic analysis of the ICE R391 genome assigned functionality to 66 of the predicted genes in integration, excision, element conjugative transfer, kanamycin resistance, mercury resistance and mutagenic DNA repair (Boltner et al., 2002). The functionality of 30 of the predicted ICE genes was cryptic with few clues as to a candidate for the sensitising function, the maintenance of which is unusual as conservation of this phenomenon is apparently detrimental to host cells under stress conditions. The maintenance of such a function, even after exposure to strong selective pressure for UV-resistant ICE mutants (P. Armshaw and J.T. Pembroke, unpublished data), is thus remarkable, suggesting that it may provide an unknown beneficial function for either the host or mobile element. Given that the specific ICE gene(s) associated with this UV-induced sensitisation have remained elusive, we generated and analysed a complete ICE R391 deletion library to identify candidate ICE genes responsible and confirmed, by cloning and complementing these genes under controlled expression, their involvement in cell sensitisation.
Materials and methods
Bacterial strains, elements and media
The bacterial strains, plasmids and mobile genetic elements used in this study are listed in Table 1. ICE R391 deletion mutants generated as part of this study are listed in Table 2. Strains were stored at − 80 °C in Luria-Bertani (LB) broth containing 50% glycerol. Media were supplemented with appropriate antimicrobial agents: nalidixic acid, 30 μg mL−1; ampicillin, 100 μg mL−1; chloramphenicol, 25 μg mL−1; kanamycin, 30 μg mL−1; streptomycin, 100 μg mL−1; sulphamethoxazole, 160 μg mL−1; trimethoprim, 32 μg mL−1; apramycin, 30 μg mL−1; mercuric chloride, 20 μg mL−1; zeocin, 25 μg mL−1 as required.
Table 1. Genotype and sources of bacterial strains, ICEs and plasmids
National Collection of Type Cultures (NCTC), Salisbury, UK
Dr R.W. Hedges, Royal Postgraduate Medical School, London, UK
StrR, SuR, TmR
Prof. Mauro M. Colombo, University of Rome La Sapienza, Rome, Italy
Ts, PBAD-gam-bet-exo cat (CmR)
Dr. P. Latour-Lambert, Institute Pasteur, 25 rue du Dr Roux, Paris, France
ApR derivative of pKOBEG
Sigma-Aldrich, Arklow, Wicklow, Ireland
Invitrogen, Bio-Sciences, Dun Laoghaire, Dublin, Ireland
CmR, p15A ori, PBAD L-arabinose inducible
NBRP E. coli strain office, National Institute of Genetics, Shizuoka 411-8540, Japan
CmR, pBAD33 containing orfs90/91 from ICE R391
CmR, pBAD33 containing orf43 from ICE R391
Table 2. Characterisation of the ICE R391 deletion library
Deletion size (kb)
orf96 could only be deleted in a ∆orfs90/91 background. Bold font indicates mutants that did not exhibit the cell-sensitising function. All mutants were assessed to determine retention of ability to transfer by conjugation in a standard conjugation assay utilising Salmonella enterica serotype enteritidis strain P125109 as the recipient. All mutants that removed the cell-sensitising function also removed the ability to transfer by conjugation. All mutant strains were AmR, the ∆28 double-deletion mutant was AmR and ZeR. For UV (+/−) and transfer (+/−), + = retained function, −= lost function, −/+ = function still observably retained but to a lesser degree than wild type.
ICE R391-specific deletions were performed as previously described (Murphy, 1998; Chaveroche et al., 2000; Datsenko & Wanner, 2000). Briefly, each ICE R391 deletion was generated by homologous recombination with transformed linear DNA that consisted of an ampicillin-resistance determinant amplified from pUC18, flanked on one side by a 40-bp sequence homologous to the 5′ end of the ICE R391 region to be deleted and on the other side by a 40-bp sequence homologous to the 3′ end of the ICE R391 region to be deleted. The linear DNA was constructed using primers with 40-bp complimentary to either the 5′- or 3′-ends of the target ICE R391 region followed by 20-bp complimentary to the 5′ or 3′-ends of the ampicillin-resistance determinant, respectively. The linear DNA was then transformed by electroporation into electrocompetent E. coli cells containing the λ Red recombination system, either integrated within the chromosome (KM32 ICE R391) or expressed on a plasmid (AB1157 ICE R391 pKOBEG). Homologous recombination between the linear DNA and the ICE R391 target region replaced the target region with the linear DNA construct in the hyper-recombination strain, thus deleting the target region from the ICE genome (Murphy, 1998). Specific deletions were verified by PCR using primers that spanned the deleted regions and subsequent sequencing of these amplicons. The resulting ICE R391 deletion mutants were screened for the retention of conjugative transfer ability, UV-inducible sensitising function or abolition of known gene functions where appropriate, for example, element excision (Fig. 1, ∆2 and ∆3 deleted xis), kanamycin (Fig. 1, ∆4 deleted aph) and mercury resistance (Fig. 1, ∆23 deleted merR/T/P/C/A). For ICE R391 double-deletion mutants such as (∆orfs90/91), (∆orf96), the second deletion was constructed using linear DNA consisting of a zeocin-resistance determinant from pcDNA3.1(+) (Table 1) rather than ampicillin resistance. The temperature-sensitive pKOBEG (carrying the λ Red recombination system) was removed from AB1157 ICE R391 mutants by overnight incubation at 37 °C before functionality screening.
Identification of mutants defective in conjugation and sensitising functions
Screening of the conjugative transfer abilities of the ICE's SXT, R391 and associated deletion mutants to recipient hosts was performed as previously described (McGrath et al., 2005). For qualitative UV assays to assess the UV-inducible cell-sensitising function, overnight LB broth cultures were subcultured and grown till an OD600 nm of 0.6 was obtained. Cells were collected by centrifugation, washed twice and resuspended in 0.85% saline. Cell suspensions were streaked across the surface of LB agar plates and allowed to air-dry. Sections of the plate were then exposed to varying doses of UV irradiation, with a primary wavelength of 254 nm, using a Griffith and George UV source. The intensity of UV irradiation was determined using a UVX radiometer (Ultra-Violet Products, San Gabriel, CA). Plates were subsequently incubated in the dark at 37 °C for 16 h (McGrath et al., 2006). For quantitative UV assays, cell suspensions were prepared as described above, left to starve for 30 min, shaken at 200 r.p.m. and incubated at 37 °C. Cell suspensions were then irradiated with UV light while being constantly agitated to prevent shielding by dead cells. After irradiation, cells were serially diluted in 0.85% saline and plated across the surface of antibiotic selective media and incubated in the dark at 37 °C for 16 h. Survival and %killing were determined as described (Pembroke & Stevens, 1984). All conjugative transfer screenings and qualitative and quantitative UV assays were performed for each mutant a minimum of three times, and results are based on the average of three separate experiments.
Cloning and expression of orfs90/91 and orf43
The orfs90/91 and orf43 were amplified by PCR from AB1157 ICE R391 genomic DNA with appropriate primers based on the ICE sequence (Boltner et al., 2002) that incorporated an Xba1 (TCTAGA) or Pst1 (CTGCAG) restriction site at the 5′ end of the forward and reverse primers, respectively. The resulting amplicons and cloning vector pBAD33 were digested with Xba1 and Pst1 and both pBAD33-orfs90/91 and pBAD33-orf43 constructed by ligating the ICE R391 orfs90/91 and orf43 regions into pBAD33, which contains an arabinose-inducible, glucose-repressible PBAD promoter (Guzman et al., 1995). Both pBAD33-orfs90/91 and pBAD33-orf43 constructs were confirmed by DNA sequencing. Qualitative and quantitative UV survival assays of pBAD33-orfs90/91 or pBAD33-orf43 strains were carried out as described, but instead of LB media, M9 minimal media with 0.4% glycerol and either 0.4% glucose or 0.02–0.2% l-arabinose were used to repress or induce expression as required. To determine the effect of induction of pBAD33-orfs90/91 or pBAD33-orf43 on host strain growth rate, each strain to be tested was grown overnight in M9 minimal media 0.4% glycerol with 0.4% glucose. Overnight cultures were washed twice with 0.85% NaCl and used as inoculate in inducing (l-arabinose) or noninducing (glucose) media as required. The optical density at 600 nm for each culture at inoculation was determined and monitored throughout incubation at 30 °C, shaking at 200 r.p.m. over required time periods.
Results and discussion
Characterisation of the ICE R391 deletion library mutants conjugation ability
A comprehensive mutational analysis of the 89-kb ICE R391 genome was carried out utilising the λ Red gene knockout system (Fig. 1, Table 2; Murphy, 1998). Mutants were generated encompassing the entire ICE R391 element, verified by PCR and sequencing and screened for loss of function. Twenty-eight directed deletion mutations were generated ranging from 0.2–17.7 kb. Sixteen deletions abolished ICE R391 conjugative transfer ability. Genes removed by these deletions, putatively thought to be required for ICE conjugative transfer, included the putative integrase gene, int (∆2 and ∆3), the putative excisionase, jef (∆2 and ∆3), the putative transcriptional enhancer genes, orfs90/91 (∆25, ∆26), and 19 putative conjugative transfer–related genes thought to be required for ICE DNA processing and mating pore formation: mobl, traI/D/J/L/E/K/B/V/A/W/U/N/F/H/G, trhF and dsbC (∆3, ∆7-∆16, ∆18, ∆23; Wozniak et al., 2009). The specific effect on conjugative transfer of the ∆28 deletion (∆orf96) could not be assessed as orf96 deletions could only be achieved in a transfer-deficient ∆orfs90/91 (∆25/∆26) deletion mutant. Deletion of any of the cryptic or predicted variable (‘noncore’) accessory ICE R391 genes did not abolish conjugative transfer (∆1, ∆4-∆6, ∆17, ∆19-∆22, ∆24, ∆27).
ICE transfer from donor to recipient cells requires excision of the ICE from the host genome, self-circularisation, a functional type IV secretion system (T4SS) to construct the mating pore and a pilus structure for element transfer and re-integration of the ICE into the recipient cell's genome (O'Halloran et al., 2007; Wozniak et al., 2009). The ∆2 and ∆3 deletions decreased the conjugative transfer rate to below detectable levels, most likely as a result of the deletion of jef and int, which would prevent excision and re-integration of the ICE from its' host genome as it has been demonstrated that deletion of the ICE R391 jef prevents detectable levels of circularised ICE R391 from being formed (O'Halloran et al., 2007). The ∆25 and ∆26 deletions likely reduced transfer rates to below detectable levels due to the deletion of the putative transcriptional enhancer complex (orfs90/91), which has previously been documented to stimulate jef expression (O'Halloran et al., 2007). The rest of the transfer-deficient mutants generated removed putative T4SS genes required to prepare ICE DNA for transfer and to mediate the physical transfer of the ICE from donor (Boltner et al., 2002) to recipient cells, and therefore, it is likely that their deletion prevented the process of conjugation itself (Lawley et al., 2003; Wozniak et al., 2009).
Characterisation of the ICE R391 deletion library mutants cell-sensitising function
All 28 ICE R391 deletion mutants created were screened for loss of the UV-inducible cell-sensitising function. Surprisingly, it was found that deletion of none of the 30 cryptic genes of ICE R391 affected the cell-sensitising function. Deletion of the putative mutagenic DNA repair–related genes rumB and rumA (Mead et al., 2007) or the putative homologous recombination promoting genes exo and bet (Chen et al., 2011) suspected as possible candidates of the cell-sensitising function, also had no effect on UV-sensitisation. Deletion of the predicted accessory ‘noncore’ genes of ICE R391, for example, ∆1 (∆orf1-orf3), ∆4 (∆orf14-orf30), ∆17 (∆orf45-orf54) or ∆23 (∆orf81-orf88), again did not affect the sensitising function, but this was not unexpected as they are part of the variable genome not present in all SXT/R391 ICEs (Boltner et al., 2002). Once again, the effect of deletion of orf96, which encodes a putative repressor (Boltner et al., 2002), could not be assessed as attempts to delete orf96 in a wild-type ICE R391 background failed. However, deletion of orf96 was achieved in a ∆orfs90/91 deletion background (Table 2, ∆25/∆26) via a double-deletion mutant.
Initially, it was found that Δ25, causing a deletion from 84279 to 87096 bp of the ICE R391, and Δ7, causing a deletion from 29575 to 42138 bp, abolished the UV-inducible sensitising effect. The Δ25 deletion removed four genes, orf90-orf94, while the Δ7 deletion removed 11 genes, orf32-orf43. The ICE R391 regions encompassed by ∆25 and ∆7 were then examined in more detail by generating small deletions of these regions. The ∆25 deletion was subdivided into ∆26 (∆orfs90/91) and ∆27 (∆orf92-orf94), while the ∆7 deletion was subdivided into ∆6 (∆orf32), ∆8 (∆orf33-orf39), Δ11 (∆orfs40/41), Δ13 (∆orf42) and Δ14 (∆orf43). Subsequent UV assay screening for the cell-sensitising function, both qualitatively (McGrath et al., 2006) and quantitatively (Pembroke & Stevens, 1984), indicated that deletion of the genes orfs90/91 (Δ26), which encode a putative transcriptional regulator related to the flhC/D transcriptional regulator of flagellar synthesis (Kutsukake et al., 1990; Boltner et al., 2002), or orf43 (Δ14), which functions in conjugative transfer and is a traV homologue (Boltner et al., 2002; Wozniak et al., 2009), abolished both the cell-sensitising and conjugative transfer functions (Table 2, Figs 2 and 3). Deletions identical to Δ14 and Δ26 were repeated a number of times and verified the involvement of these genes in the UV-inducible cell-sensitising function. Deletion of the orf43 (traV) homologue, s050, was repeated in the highly similar ICE SXT (Beaber et al., 2002) and was also found to abolish both the UV-inducible cell-sensitising and conjugative transfer functions of SXT.
Individual effect of expression of orfs90/91 and orf43 on host cells
Screening of the ICE R391 deletion library determined that the genes orfs90/91 and orf43 were involved in the cell-sensitising function. To determine whether the gene products of orfs90/91 and orf43 were both required for cytotoxicity or whether one was an activator of the other's cytotoxic function, both regions were individually cloned and the effect of their induction on appropriate ICE R391 deletion mutants and wild-type E. coli was examined. The cytotoxic effect of pBAD33-orfs90/91 and pBAD33-orf43 was assessed by monitoring the growth rate of host cells after induction with l-arabinose (pBAD). Both pBAD33-orfs90/91 and pBAD33-orf43 were shown to functionally complement for orfs90/91 and orf43 in the cell-sensitising function in appropriate deletion mutants [AB1157 R391 ∆26 (∆orfs90/91) and AB1157 R391 ∆14 (∆orf43)] by both qualitative and quantitative UV survival assays. On examining the growth rates of hosts harbouring recombinant constructs, it was shown that following induction of pBAD33-orfs90/91, only cells containing ICE R391 ∆26 (∆orfs90/91) lacking a deletion of orf43 were sensitised (Fig. 4a–c). In contrast, induction of pBAD33-orf43 expression was found to sensitise wild-type E. coli in the absence of ICE R391 (Fig. 4d) indicating that orf43 expression alone is the cytotoxic instigator and that its expression may be controlled by orfs90/91, which in turn is UV-inducible (O'Halloran et al., 2007), given their putative annotation as transcriptional enhancers. Work is currently in progress to delineate the regulation of orf43 following UV irradiation.
ICE R391 TraV (orf43) homologue has dual functions in conjugation and cell sensitisation
Deletion of orf43 (AB1157 R391 ∆14) not only abolished the cell-sensitising function but additionally abolished the ability of ICE R391 to transfer from donor to recipient cells by conjugation. Expression of pBAD33-orf43 in the orf43 deletion mutant complemented for both the cell-sensitising function and ICE R391 conjugative transfer. Comparative analysis carried out as part of this study of the orf43 homologues in all fully sequenced SXT/R391-like ICEs available in the ICEberg database (Bi et al., 2012) demonstrated that it is a highly conserved core gene (≥ 98% identity at the nucleotide level), all encoding a putative TraV family protein. TraV family proteins are thought to function in the formation and stabilisation of the T4SS mating pore within the outer membrane of bacterial cells (Lawley et al., 2003; Alvarez-Martinez & Christie, 2009; Fronzes et al., 2009). Previously characterised TraV homologues have been found to have a size range of 171–316 a.a., contain a signal sequence localising the protein to the outer membrane and have been determined to share little overall sequence similarity except for two conserved cysteine residues thought to be involved in multimer formation (Frost et al., 1985; Harris & Silverman, 2002; Lawley et al., 2003). The putative TraV homologue encoded by orf43 is 216 a.a. in size, has both predicted transmembrane spanning and signal peptide regions and contains the two conserved cysteine residues. However, unusually, even though TraV homologues are conserved across several different T4SS, the cell-sensitising function is not, indicating that the TraV homologue encoded by orf43 either is regulated differently or is structurally altered compared with other TraV homologues. The structure, mode of cytotoxicity and regulation of orf43 are currently under investigation as is its relationship with orfs90/91.
This work was funded by the Irish Research Council for Science, Engineering and Technology (IRSCET) to PA. The authors would like to thank Dr. Kenan C. Murphy for providing the hyper recombinant strain KM32, Dr. P. Latour-Lambert for providing the pKOBEG and pKOBEGApra plasmids and Drs John O'Halloran, Anna V. Piterina and Michael P. Ryan for helpful discussion.