Tn916-like genetic elements: a diverse group of modular mobile elements conferring antibiotic resistance

Abstract

Antibiotic-resistant Gram-positive bacteria are responsible for morbidity and mortality in healthcare environments. Enterococcus faecium, Enterococcus faecalis, Staphylococcus aureus and Streptococcus pneumoniae can all exhibit clinically relevant multidrug resistance phenotypes due to acquired resistance genes on mobile genetic elements. It is possible that clinically relevant multidrug-resistant Clostridium difficile strains will appear in the future, as the organism is adept at acquiring mobile genetic elements (plasmids and transposons). Conjugative transposons of the Tn916/Tn1545 family, which carry major antibiotic resistance determinants, are transmissible between these different bacteria by a conjugative mechanism during which the elements are excised by a staggered cut from donor cells, converted to a circular form, transferred by cell−cell contact and inserted into recipient cells by a site-specific recombinase. The ability of these conjugative transposons to acquire additional, clinically relevant antibiotic resistance genes importantly contributes to the emergence of multidrug resistance.

Enterococcus, Clostridium difficile, Staphylococcus aureus, Streptococcus pneumoniae, Tn916, antibiotic resistance

Introduction

The emergence of antibiotic resistance among bacterial pathogens is a major problem in the treatment of infectious disease in both the community and in healthcare settings throughout the world. In industrialized nations, there has been a steady rise in the incidence of high-profile healthcare-associated infections that have become resistant to one or more antibacterial agents making treatment increasingly difficult. These include, but are not limited to, methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE) and multidrug-resistant Streptococcus pneumonia (Rossolini et al., 2010). In addition, there are pathogens, such as Clostridium difficile, which have risen to global prominence over the last few years and have the ability to acquire mobile genetic elements from enterococci, indicating the potential to acquire resistance to the last line of defence antibiotics, for example vancomycin and other glycopeptides (Cartman et al., 2010; Jasni et al., 2010). These resistances are commonly acquired on mobile genetic elements such as conjugative plasmids and conjugative transposons, which are capable of broad host range transfer between pathogens (Weigel et al., 2003; Jasni et al., 2010) and between commensal and pathogenic bacteria. For example, methicillin resistance, which is mediated by the product of the mecA gene and is present in MRSA strains, most likely originates from Staphylococcus fleurettii, an animal commensal (Tsubakishita et al., 2010).

The Tn916/Tn1545 family is responsible for a large proportion of the antibiotic resistance in these different pathogens. These conjugative elements are responsible for the dissemination of many antimicrobial resistance genes (usually resistance to tetracyclines, but also macrolides, lincosamides and streptogramins, kanamycin and mercury) to some of the most important Gram-positive pathogens. Their properties are reviewed in detail here.

Enterococcus spp., a reservoir of antibiotic resistance elements in hospital environments

Thiercelin used the name ‘entérocoque’ in 1899 to emphasize the intestinal origin of organisms isolated from patients with enteritis (Murray, 1990). The name Streptococcus faecalis was first used in 1906 (Andrews & Horder, 1906) to describe a faecal microorganism, and Streptococcus faecium was described in 1919 (Orla-Jenson, 1919). Subsequently, the term ‘enterococcal group’ for streptococci capable of growing under certain conditions was used in 1937 (Sherman et al., 1937) and an official proposal for the Enterococcus genus followed in 1970 (Kalina, 1970), but was largely ignored until genetic evidence of a genera-specific difference was provided in 1984 (Schleifer & Kilpper-Balz, 1984). Now the genus Enterococcus is accepted and a great deal of work has been carried out on it, primarily due to the fact that both Enterococcus faecalis and Enterococcus faecium, as well as to a lesser extent some of the other members of the genus such as Enterococcus casseliflavus, are responsible for a large number of hospital-associated infections (Woodford & Livermore, 2009). The global spread of a single, multilocus sequence typing-based clonal complex (CC) of E. faecium (i.e. E. faecium CC17) is responsible for the majority of hospital outbreaks and clinical infections across five continents (Leavis et al., 2006). Similar data exist for E. faecalis and in some clonal complexes, such as E. faecalis CC2; comparative genomics have shown significant enrichment of mobile genetic elements in these genomes (Solheim et al., 2011). Antibiotic resistance among Enterococcus spp. is mediated by genes acquired on both plasmids and transposons (reviewed in Weaver et al., 2002; Hegstad et al., 2010; Palmer et al., 2010). Enterococcus spp. have a propensity to acquire resistance genes located on conjugative transposons, which has contributed to the spread of resistance to different antibiotics including vancomycin and has led for example to the emergence of VRE by the acquisition of vanB2 on the conjugative transposons Tn5382/Tn1549 (Rice et al., 2001).

Clostridium difficile emerges as an important host of transmissible antibiotic resistance elements

Clostridium difficile is an anaerobic spore-forming organism that was initially discovered in 1935 in the stools of newborns (Hall & O'Toole, 1935). In 1978, it was identified as the main causative agent of antibiotic-associated diarrhoea and pseudomembranous colitis, a severe necrotizing disease of the colon (Bartlett et al., 1978). In the last 10 years, there has been an increase in C. difficile infection (CDI) with the emergence of epidemic strains (Bartlett et al., 2006) initially from Canada (Pépin et al., 2004) and subsequently from the United States (McDonald et al., 2005). These outbreaks were caused by a single strain, which was designated NAP1 ribotype 027 (Loo et al., 2005). Related epidemic strains of C. difficile have now been identified as the cause of hospital outbreaks in the United Kingdom (Smith, 2005) and in 15 other European countries (Kuijper et al., 2008). This ribotype has also been identified in Japan (Kato et al., 2007) and Australia (Riley et al., 2009). In addition, there are other ‘hypervirulent’ strains emerging, such as ribotypes 001, 012, 017 and 078 (Dawson et al., 2009; Huang et al., 2009), indicating that the epidemiology of C. difficile infections worldwide is complex and may not be determined primarily by the ability of a strain to cause disease. In addition, while cases in North America and Europe are predicted to decrease in the near future, CDI is predicted to rise in Asia, Africa and South America over the next 10 years (Gerding et al., 2010). In Asia, ribotype 017 appears to be locally prevalent. In a recent study, 105 out of 408 C. difficile isolates (25.7%) from South Korea were shown to be toxin A⁻B⁺ strains and were all ribotype 017 (Kim et al., 2010). A high incidence of PCR ribotype 017 isolates (18.7% of strains tested) was also seen in a study on isolates from Shanghai (Huang et al., 2009). It has been hypothesized that differences in ribotype prevalence in different parts of the world may be linked to the antibiotic resistances possessed by different ribotypes and the local treatment strategies used; however, this requires more confirmatory work.

Treatment with antibiotics plays a central role in the development of CDI as this disrupts the normal microbiota, which is believed to present a barrier against C. difficile colonization. Once this colonization resistance is impaired, C. difficile can colonize the gut and cause disease, which is mediated by the major virulence factors toxins A and B (Lyras et al., 2009; Kuehne et al., 2010). The usual treatment for CDI is either oral metronidazole or vancomycin and although decreased susceptibility to these agents has been reported, resistance has not yet been observed. Antibiotic resistance in C. difficile is varied among different strains with some resistance associated with different ribotypes (Huang et al., 2009). Mobile genetic elements, especially conjugative transposons (which frequently contain antibiotic resistance genes), are common in C. difficile (Sebaihia et al., 2006).

Staphylococcus aureus: its resistance to methicillin and β-lactams is found in clonal complexes

Staphylococcus aureus was discovered in Scotland in 1880 by Sir Alexander Ogston in pus from surgical abscesses. Staphylococcus aureus typically lives as a commensal of the human nose in 30–70% of the population (Peacock et al., 2001); however, it is also responsible for a wide range of infections including minor skin abscesses to more serious pneumonia, toxic shock syndrome, bloodstream infections and endocarditis. The emergence of MRSA followed soon after the introduction of methicillin in 1960 (Jevons, 1961), and since then it has become extremely prevalent worldwide with rates often exceeding 40% of S. aureus isolates (Grundmann et al., 2006). One interesting aspect of S. aureus epidemiology is that the global expansion of the population has occurred by the expansion of certain strains as clonal complexes (Feil et al., 2003). One property that has emerged from the many genome sequences of this organism is that the genome is made up of many different types of mobile genetic elements (Malachowa & DeLeo, 2010), the most clinically significant probably being staphylococcal cassette chromosomes (SCC) (Gould, 2009). SCC is a genomic island ranging in size from approximately 24 to >50 kb, which is always found inserted in a hotspot within the orfX gene. All MRSA strains carry this element, which confers resistance to methicillin and all β-lactams. Recent reviews cover this topic comprehensively (Chambers & DeLeo, 2009; Lindsay, 2010; Malachowa & DeLeo, 2010).

Streptococcus pneumoniae: emergence of multidrug resistance is a serious clinical problem

Streptococcus pneumoniae was first isolated in 1881 on opposite sides of the Atlantic ocean by Louis Pasteur in France and George Sternberg in the United States in independent studies on saliva. It was originally described as Microbe septicémique de la salive by Pasteur and as Micrococcus pasteuri by Sternberg (Watson et al., 1993). It was subsequently referred to as Pneumococcus and was renamed Diplococcus pneumoniae in 1920 (Winslow et al., 1920). It was finally given the name S. pneumoniae in 1974 (Deibel & Seeley, 1974). It is the most commonly identified bacterial cause of community-acquired pneumonia, meningitis and otitis media. It is also a frequent cause of bacteraemia, and accounts for significant morbidity and mortality. The highest incidence of pneumococcal disease is observed in children <2 years and in adults >65 years of age. A recent review of many studies have shown that six to 11 serotypes of S. pneumoniae account for ≥70% of invasive pneumococcal disease in children (Johnson et al., 2010). Treatment of these infections with antibiotics is increasingly becoming problematic due to increased resistance to penicillin, macrolides and other antibiotics (Fuller et al., 2005; Jones et al., 2010). The first report of multidrug-resistant (resistant to more than three different antimicrobial classes) S. pneumoniae came from South Africa in 1978 (Jacobs et al., 1978) and now between 15% and 30% of S. pneumoniae isolates are multidrug resistant (Lynch & Zhanel, 2009). Resistance to many of the drugs are due to genes present on conjugative transposons.

Transmission of antibiotic resistance determinants by mobile genetic elements in Gram-positive pathogens

There are numerous recent reviews of specific mobile elements in S. aureus (e.g. SCCmec; Chambers & DeLeo, 2009), Enterococcus spp. (Weaver et al., 2002; Hegstad et al., 2010; Palmer et al., 2010), Streptococcus spp. (Roberts & Mullany, 2009; Varaldo et al., 2009) and C. difficile (Adams et al., 2002). The subject of the remainder of this review will be those elements that are found in all of the organisms in question. Members of the Tn916/Tn1545 family of conjugative transposons have been found in all four of the subject organisms (Table 1), and so it is likely that transfer between these important pathogens is occurring, either directly or via other transfer intermediates: for example the transfer of Tn5397 between C. difficile and E. faecalis has recently been demonstrated in vitro (Jasni et al., 2010).

Table 1

Details of Tn916-like elements discussed in this review

Name of CTn/ICE	Antibiotic resistance [gene]	Approximate size (kb)	Comments	Reference(s)
Tn916	Tetracyclines [tet(M)]	18	First member of this family discovered	Franke & Clewell (1981)
Tn918	Tetracyclines [tet(M)]	18	Very similar to Tn916	Clewell et al. (1985)
Tn925	Tetracyclines [tet(M)]	18	Very similar to Tn916	Christie et al. (1987)
Tn1545	Tetracyclines [tet(M)], MLS (two copies of [erm(B)], kanamycin [aphA-3]	26.5	Contains a copy of IS1239	Courvalin & Carlier (1986), Cochetti et al. (2008)
Tn2009	Tetracyclines [tet(M)] Macrolides [mef(E)]	23.5	mef(E) is located on the MEGA element	Del Grosso et al. (2004)
Tn2010	Tetracyclines [tet(M)] Macrolides [mef(E)] MLS [erm(B)]	26.3	Contains the erm(B) element of Tn6002 and the MEGA element of Tn2009	Del Grosso et al. (2006), Li et al. (2011)
Tn2017	Tetracyclines [tet(M)] Macrolides [mef(E)] MLS [erm(B)]	28.5	Contains the MEGA element and Tn917 (containing erm(B))	Del Grosso et al. (2009)
Tn3702	Tetracyclines [tet(M)],	18	Very similar to Tn916	Horaud et al. (1990)
Tn3872	Tetracyclines [tet(M)], MLS [erm(B)]	24	Contains Tn917 inserted into orf9	McDougal et al. (1998)
Tn5031	Tetracyclines [tet(M)]	18	Very similar to Tn916	Fletcher et al. (1989)
Tn5251	Tetracyclines [tet(M)]	18	Very similar to Tn916, this element has inserted into Tn5252 forming the composite element Tn5253	Ayoubi et al. (1991)
Tn5381	Tetracyclines [tet(M)]	18	Very similar to Tn916	Rice et al. (1992)
Tn5383	Tetracyclines [tet(M)]	18	Very similar to Tn916	Rice et al. (1992)
Tn5386	Lantibiotic resistance	29.5	Contains lantibiotic immunity genes in place of tet(M) Contains a group II intron Also encodes putative cell surface-associated protein upstream of the conjugation module	Rice et al. (2005), Rice et al. (2007)
Tn5397	Tetracyclines [tet(M)]	20.6	Contains a group II intron This element was the first Tn916-like element to be identified which uses a large serine recombinase as opposed to a tyrosine integrase for excision and insertion reactions	Mullany et al. (1996), Roberts et al. (2001)
Tn5801	Tetracyclines [tet(M)]	25	Identified by genome sequencing, similar to Tn916 contains additional genes upstream of the conjugation module	Kuroda et al. (2001)
Tn6000	Tetracyclines [tet(S)]	33.2	The tyrosine integrase is more related to that of staphylococcal pathogenicity islands Contains a group II intron Encodes a helicase from vrl from Dichelobacter nodosus Encodes additional restriction modification proteins	Roberts et al. (2006), Brouwer et al. (2010)
Tn6002	Tetracyclines [tet(M)] MLS [erm(B)]	20.8	Tn916 containing an erm(B) containing cassette within orf20	Warburton et al. (2007)
Tn6003	Tetracyclines [tet(M)], MLS [two copies of erm(B)], kanamycin [aphA-3]	25	Tn6002 containing another erm(B) containing cassette known as MAS	Cochetti et al. (2008)
Tn6009	Tetracyclines [tet(M)] Inorganic mercury [merA] Organic mercury resistance [merB]	23	Contains a staphylococcal mercury resistance operon located upstream of the conjugation module	Soge et al. (2008)
Tn6014	Tetracyclines [tet(M)]	25	This is the first Tn5801-like transposon shown to transfer	De Vries et al. (2009)
Tn6084	Tetracyclines [tet(M)]	22	Contains a group II inton in the same place as Tn6085 a and b and a copy of ISEfa11	Rice et al. (2010)
Tn6085a and b	Tetracyclines [tet(M)]	20.7	Identical elements in the same isolate as Tn6084 above Contains a group II intron	Rice et al. (2010)
CTn1	Unknown	28	Encodes a putative ABC transporter within unknown substrate Also encodes putative cell surface-associated protein upstream of the conjugation module	Sebaihia et al. (2006)
CTn6	Unknown	23	Encodes a putative Mg2+-transporting P-type ATPase Encodes putative cell surface-associated protein upstream of the conjugation module	Sebaihia et al. (2006)
CTn7	None evident	26	Accessory genes are hypothetical	Sebaihia et al. (2006)

Nomenclature of conjugative transposons and integrative conjugative elements

The issue of nomenclature for these types of elements has been the subject of much lively debate. The originally discovered member of this family, Tn916 (Franke & Clewell, 1981), was termed a conjugative transposon and designated a Tn number according to the rules published for transposable elements (Campbell et al., 1977, 1979a, b). The allocation of these numbers was carried out by Dr Esther Lederberg from Stanford University Medical School, CA. Lists of allocated Tn numbers up to Tn4685 were subsequently published (Lederberg et al., 1981, 1987). However, this system stopped with the retirement of Dr Lederberg and subsequently a variety of rules were adopted for naming newly discovered transposons and nowhere in the field were these rules more varied than with the conjugative elements.

In 1999, Hochhut and Waldor published details of a 62-kb self-transmissible conjugative element from Vibrio cholerae, which encodes multiple antibiotic resistances (Hochhut & Waldor, 1999). This element, designated SXT, was called a CONSTIN, an acronym for a conjugative, self-transmissible, integrating element. This term is still used today although only in relation to the SXT element in Vibrio sp. (e.g. Goel et al., 2010). Later Burrus et al. (2002) proposed to use the prefix ICE, an acronym for integrative and conjugative element, followed by the initials of the name of the bacterial genus and species and a number corresponding to the rank of the discovery (e.g. ICESt1 for the first ICE found in Streptococcus thermophilus) (Burrus et al., 2000). Subsequently, Burrus et al. (2006a) further proposed inclusion of three letters to identify the country of origin, followed by a number to distinguish between different isolates of the same species and country. For example, the first ICE of the SXT/R391 family found in V. cholerae isolated in Mexico was designated ICEVchMex1 (Burrus et al., 2006b).

In 2008, after 2 years of discussions a consensus was generally agreed upon whereby the naming of new transposable elements of any family would be carried out using a reimplementation of the previous Tn system. The registry for Tn numbers, hosted by University College London (http://www.ucl.ac.uk/eastman/tn/) (Roberts et al., 2008), now makes it possible to assign names in a logical way to any transposable element including those found in the increasingly large amounts of metagenomic data becoming available, for example Tn6032 (Suenaga et al., 2009). It was also decided that the terms conjugative transposon and ICE were interchangeable (Roberts et al., 2008; Wozniak & Waldor, 2010) and therefore an ICE could be assigned a number from the registry, for example ICE6013 (Smyth & Robinson, 2009). Alternatively, any transposable element that is also conjugative can be given the prefix CTn. This system now appears robust and able to cope with any newly identified transposable element for which there is not already a suitable nomenclature system in place.

The Tn916/Tn1545 family of conjugative transposons

Originally discovered in the late 1970s in E. faecalis, Tn916 was the first conjugative transposon encoding antibiotic resistance to be reported (Franke & Clewell, 1981). It is the smallest member of the Tn916/Tn1545 family and contains 24 ORFs organized into functional modules involved in conjugal transfer, recombination (excision and insertion reactions), transcriptional regulation and accessory functions (antibiotic resistance) (Fig. 1a) (Senghas et al., 1988; Flannagan et al., 1994; Roberts & Mullany, 2009). Tn916 has been detected in or transferred into over 35 different genera of bacteria, often in multiple species of a single genus (Clewell et al., 1995; Rice et al., 1998; Roberts & Mullany, 2009). Tn1545 is homologous to the entire Tn916; however, it also contains some important insertions: there is an insertion of an erm(B) and aphA-3 containing cassette within the 3′ region of orf20 and a copy of IS1239 upstream of orf12 (Cochetti et al., 2008).

Figure 1

Schematic representation of Tn916 and related elements. The arrows represent the individual orfs pointing in the probable direction of transcription and labelled above. The colour coding for the functionality of the predicted proteins is as follows: blue, conjugation; green, transcriptional regulation; red, accessory genes (in the case of Tn916 this is tetracycline resistance); grey, insertion and excision (recombination). Genes not present in Tn916 are shown in orange on the other elements. (a) Tn916: the arrow points to the position of the experimentally determined oriT (Jaworski & Clewell, 1995). The modules are indicated by the brackets underneath and are labelled with the relevant function. (b) Tn6000: the major differences are the presence of restriction/modification genes before the conjugation module, the presence of an insertion of DNA that shares homology to the vrl of Dichelobacter nodosus and the integrase and excisionase genes that share more homology to staphylococcal pathogenicity island genes than they do to Tn916 integrase and excisionase genes. The closest homologue to various regions of DNA are indicated below the figure by brackets (adapted from Brouwer et al., 2010). (c) Tn5397: the major differences are the presence of tndX in Tn5397 as opposed to xisTn and intTn in Tn916, the presence of a self-splicing group II intron in Tn5397 and an 88-bp deletion in the regulatory region resulting in the removal of orf12 and the generation of orf25 and orf26. (d) Tn5801 from Staphylococcus aureus: the major differences are the ORFs located before the conjugation module. Most of the additional genes are hypothetical; however, sav415 is predicted to encode a transposase (de Vries et al., 2009).

An increasingly reported observation in newly discovered Tn916-like elements is the association of other, often mobilizable, transposons within the Tn916-like backbone (Table 1) (Cochetti et al., 2007, 2008; Roberts & Mullany, 2009; Varaldo et al., 2009).

The genetic consequences of carriage of Tn916/Tn1545-like elements go beyond acquisition of resistance

Due to the mechanistic details of their movement, which are varied and dependent on the recombinase enzyme (Mullany et al., 2002), there are a variety of ways in which conjugative transposons can cause heritable changes to the genome of their host, the most obvious being acquisition of the element itself. Conjugative transposons were not, until recently, thought to be capable of independent replication, although recent studies of the conjugative transposon ICEBs1 from Bacillus subtilis suggest that replication of the circular form may be a general phenomenon (Lee et al., 2010). Despite the potential for replication, these elements usually need to enter the host cells' genome in order to survive in the cell. They do this using their own site-specific recombinases (Mullany et al., 2002); these reactions can result in changes in the DNA around the target site of the element. The tyrosine integrase found in most reported Tn916-like elements cuts the DNA in a 5′–3′ staggered endonucleolytic cleavage, resulting in a circular molecule containing a heteroduplex at the joint of the circular form, which is known as the coupling sequence (Caparon & Scott, 1989). This means that in transconjugants the joint of the circular form will be from the donor's genome (Fig. 2). This will result in a genetic variation upon which natural selection can act. Likewise upon excision from the host replicon, the element will remove flanking DNA, potentially leading to further genetic variation within a bacterial population (these processes are illustrated in Fig. 2).

Figure 2

Schematic representation showing the recombination reactions of Tn916 excision and insertion. The thick lines represent Tn916 and the thin lines are the flanking genomic DNA. Nucleotides in yellow represent those originally associated with the inserted copy of Tn916. Nucleotides in green indicate those associated with the new target site. Nucleotides in red indicated those that are the product of DNA repair through replication or mismatch repair. (a) Excision: a staggered cleavage at the coupling sequence (represented by a hexamer of single nucleotides) results in the excision of Tn916 with noncomplementary overhangs, which are ligated together forming a heteroduplex. The heteroduplex is resolved in the vacated target site by DNA replication. (b) Insertion: staggered cleavage at the 3′ end of the new target site and the heteroduplex at the joint of the circular intermediate generates molecules that are subsequently ligated together, resulting in an inserted Tn916 with heteroduplexes at each end. Again, these are resolved by DNA replication resulting in two copies, each flanked by different DNA sequences (adapted from Caparon & Scott, 1989).

Another heritable change can occur if the element inserts into an ORF and generates in an insertional mutation. Sometimes insertion events will form gene fusions resulting in new proteins. For example, in C. difficile strain 630 the conjugative transposon CTn5 has inserted into gene CD1844 encoding a putative surface associated-protein; this insertion event has led to the possible production of a fusion protein (Sebaihia et al., 2006). Outward reading promoters from integrated elements can lead to differential expression of host genes. This has been observed experimentally for Tn916 in B. subtilis (Celli & Trieu-Cuot, 1998). Furthermore, in E. faecalis, insertion of Tn916 upstream of a haemolysin gene can lead to a hyper-haemolytic phenotype (Ike et al., 1992). The final way in which these elements can change the DNA of the host is by trans-acting interaction with other elements, such as that observed between Tn916 and Tn5386 (Rice et al., 2005, 2007) and mobilization of other nonconjugative plasmids and transposons leading to their transfer to new hosts (Flannagan & Clewell, 1991). The effect that a conjugative transposon has on its host is variable and extends beyond the simple provision of a new phenotype such as antibiotic resistance.

Tn916-like elements in Enterococcus spp.

A clinical isolate, E. faecalis strain DS16 (initially described as S. faecalis ssp. zymogenes strain DS16; Tomich et al., 1979), harbours the conjugative haemolysin-bacteriocin plasmid pAD1 (60 kb) and the nonconjugative plasmid pAD2 determining resistance to streptomycin, kanamycin and erythromycin, in addition to the chromosomally located Tn916 (Franke & Clewell, 1981). In this paper, it was shown that when strain DS16 was mated with the plasmid-free E. faecalis strain JH2-2, some transconjugants resistant to tetracycline contained the Tn916 determinant linked to pAD1. In addition, derivatives of DS16 devoid of pAD1 were capable of transferring tetracycline resistance to recipient strains. Transconjugants (plasmid-free) from such matings could subsequently act as donors in the transfer of tetracycline resistance. Both transposition and transfer were found to be Rec-independent. Since then detailed and extensive investigations have revealed how these elements are able to move (Fig. 2).

A large number of Tn916-like elements have been detected in enterococci (tabulated in Hegstad et al., 2010) isolated from various environments including humans and pigs (De Leener et al., 2004; Agersø et al., 2006), wild boars (Poeta et al., 2007), broilers (Cauwerts et al., 2007), house flies from food settings (fast-food restaurants in Kansas) (Macovei & Zurek, 2006), cats and dogs (De Leener et al., 2005) and food (Huys et al., 2004). Many of these studies detected the Tn916 integrase gene, (intTn), by PCR to determine the presence of Tn916. This approach however leads to more questions than answers, as it does not give an indication of the genetic linkage between the various resistance genes that have been detected or between the resistance genes and the integrase. However, many other Tn916-like elements have been investigated in detail and shown to encode only resistance to tetracycline via the tet(M) gene (Table 1), including but not limited to Tn916 (Franke & Clewell, 1981), Tn918 (Clewell et al., 1985), Tn925 (Christie et al., 1987), Tn3702 (Horaud et al., 1990), Tn5031-3 (Fletcher et al., 1989), Tn5381 and Tn5383 (Rice et al., 1992), Tn6084, Tn6085a and Tn6085b (Rice et al., 2010). Some of these have not yet been fully sequenced and, therefore, may represent elements that are very similar or identical to Tn916 (e.g. Tn3702 and Tn5031). Interestingly, the recent report of Tn6084, Tn6085a and Tn6085b show that these three conjugative transposons reside in the same cell and all contain intratransposon insertions. All three of these elements contain a group II intron inserted in a gene involved in conjugation (Tn916orf-06). In addition, Tn6084 also contains an insertion of ISEfa11 56 nucleotides upstream of the start codon of tet(M), which is in the middle of the sequence involved in the formation of one of the terminator structures. These are thought to regulate the expression of the element (Su et al., 1992). It would be interesting to determine whether this insertion has an effect on the element's response to tetracycline (Rice et al., 2010).

Of particular interest is the chimerical conjugative transposon Tn6000 (Fig. 1b). This element shares homology with diverse proven and putative elements from C. difficile (Tn5397), S. aureus (SPIbov1), Lactococcus lactis (pK214) and Dichelobacter nodosus (virulence-related locus; vrl). It confers tetracycline resistance by Tet(S) as opposed to Tet(M). While Tn6000 does not contain any additional resistance genes, it does contain a region upstream of the conjugation module that appears to be involved in restriction/modification (Brouwer et al., 2010). Although the function of these predicted proteins is still to be proven, it is tempting to speculate that antirestriction proteins (Orf18 and Orf25) and methyltransferases (Orf26 and Orf29) will protect the incoming element from any host endonucleases. Following successful integration into the host chromosome, the putative restriction enzyme encoded by the element could act like a molecular vaccine for the cell, protecting it from any other incoming DNA (Kobayashi, 2001). The restriction/modification genes in Tn6000 are located upstream of the conjugation module (Fig. 1b). Tn6009, a Tn916-like conjugative transposon originally discovered in Klebsiella and subsequently detected in E. faecalis, contains a functional staphylococcal mercury resistance operon also located upstream of the conjugation module (Soge et al., 2008), in the same place as the restriction/modification module in Tn6000; this site is often used for the acquisition of additional accessory genes. Tn5386 from E. faecium strain D344R (Rice et al., 2007) and CTn1 and CTn7 from C. difficile 630 (Sebaihia et al., 2006) all contain genes predicted to encode putative cell surface-associated proteins in this region. These genes may have been picked up by Tn916-like elements by erroneous, or variable, excision reactions, which are comparable to the erroneous excision of a specialized transducing phage genome from the host replicon before transduction. This may be a common mechanism allowing Tn916-like elements to acquire new genes; however, as we are likely to see only the successful elements that have been selected for by a suitable environmental selective pressure, the impact of this mechanism of acquisition of new genetic material may be undervalued.

Tn916-like elements in C. difficile

The first resistance-encoding conjugative transposon investigated in any depth in C. difficile was Tn5397 (Mullany et al., 1996). This element encodes tetracycline resistance via the tet(M) gene product. Tn5397 was originally shown, by Southern hybridization, to share homology to Tn916 (Hächler et al., 1987) and subsequent sequence determination confirmed that it was a member of the Tn916 family with some fundamental differences (Roberts et al., 2001). The tyrosine integrase and excisionase gene of Tn916 is absent in Tn5397, and instead there is a gene for a large serine recombinase, which is responsible for the excision and insertion of the element. The protein, TndX, is most closely related to TnpX from the mobilizable transposons Tn4453a and b from C. difficile (Lyras et al., 1998) and Tn4451 from Clostridium perfringens (Abraham & Rood, 1987). These transposons encode resistance to chloramphenicol and TnpX has been shown to be the only protein required for transposition of the element (Lyras et al., 2004). Also in contrast to Tn916, Tn5397 contains a group II intron, which has been shown to splice from its host gene (orf14). However, it has also been shown that splicing of the mutant intron is not necessary for transfer, presumably because the intron has inserted so close to the 3′ end of the gene (Roberts et al., 2001). These differences are summarized in Fig. 1c.

Tn5397 has a preferred target site in all C. difficile strains in which it has been investigated (Wang et al., 2000). The element can be transferred to B. subtilis where it can insert into the genome without any obvious site preference. However, if the preferred C. difficile target site is introduced into the B. subtilis chromosome, the element will always enter that site, indicating a strong site preference. Additionally, Tn5397 transfers between C. difficile strains (Mullany et al., 1996) and between C. difficile and E. faecalis in vitro (Jasni et al., 2010). Transfer between C. difficile and E. faecalis is clinically important because the two organisms inhabit the human gut and concomitant infection has been reported widely (Ray & Donskey, 2003). Moreover E. faecalis, and more frequently E. faecium, strains harbour a conjugative transposon Tn1549, which contains the vanB2 gene conferring vancomycin resistance (Tsvetkova et al., 2010). Vancomycin is one of the two antibiotics currently recommended for treatment of CDI (the other is metronidazole), and therefore the transfer of vancomycin resistance from E. faecalis to C. difficile would be a problem. In addition, Tn1549 has been acquired by a variety of other anaerobic bacteria including other Clostridium spp. (Ballard et al., 2005); therefore, transfer of Tn1549 from other donor bacteria to C. difficile may be possible. It may just be a matter of time until resistance to vancomycin emerges in this pathogen.

Genome sequencing of C. difficile 630 (ribotype 012) demonstrated that it contained many mobile elements, primarily conjugative transposons (Sebaihia et al., 2006). Tn5397 (termed CTn3 in Sebaihia et al., 2006) is one of seven putative conjugative transposons identified in the C. difficile 630 genome. Four of these, CTn1, Tn5397 (CTn3), CTn6 and CTn7, are related to Tn916 and differ primarily in their accessory genes (it is noted here that CTn1, Tn5397 [CTn3], CTn6 and CTn7 have also been called CTnCD1, CTnCD3, CTnCD6 and CTnCD7 recently) (He et al., 2010). While Tn5397 contains tet(M) conferring tetracycline resistance, the others contain a putative ABC transporter (CTn1), Mg²⁺ transporting P-type ATPase (CTn7) and hypothetical proteins (CTn6). Also, CTn7 specifies a large serine recombinase similar to that of Tn5397, whereas CTn1 and CTn6 specify a tyrosine integrase that is predicted to catalyse integration and excision. Tn5397 has also been shown to be present in strains from around the world and is associated with certain ribotypes (Bakker et al., 2010). Tetracycline resistance in ribotype 012 and 046 strains is associated with Tn5397 elements, whereas resistance in ribotypes 017 and 078 is associated with Tn916-like elements (Bakker et al., 2010). The association of these mobile genetic elements with certain groups of strains can give insights into the evolution of the so-called ‘hypervirulent’ strains, which have recently spread across the globe (Stabler et al., 2009), and demonstrate how strains are related epidemiologically through time. Regions of ribotype 027-specific mobile genetic elements seem to have been acquired in the last 16 years of the organism's evolution and, when analysed in depth, could help us understand the epidemic nature of the spread of these strains (Stabler et al., 2009).

Additional differences have been reported in Tn916-like elements in C. difficile. The presence of Tn916-like conjugative transposons was demonstrated in many clinical strains (Spigaglia et al., 2005). Moreover, an analysis of strains isolated in Italy from 1997 demonstrates that Tn916 is more common than Tn5397 (Spigaglia et al., 2006). Whether this is due to an epidemiological shift in the population dynamics of these elements or whether it is simply because it was never before looked for is unclear. What is clear is that there have been interactions between different resistance elements in C. difficile as exemplified by the discovery of the genetic linkage of tet(M) and erm(B) in C. difficile strain cd1911 (ribotype R), both upstream of a Tn916 integrase gene, and therefore presumably on a Tn916-like conjugative transposon (Spigaglia et al., 2007). However, all attempts to transfer both resistances together from this strain were unsuccessful.

Unlike Tn5397, Tn916 behaves differently in different strains of C. difficile. In strain CD37 (ribotype undetermined) it has a conserved attachment or target site (attB) (Mullany et al., 1991) whereas in other strains such as CD196 (ribotype 027) and 42373 (ribotype undetermined) the element enters the genome in multiple sites (Roberts et al., 2003; Hussain et al., 2005). The reasons for this disparity in target site selection are currently not understood, but are likely to be due to host factors, as the same element acts differently in different strains.

Tn916-like elements in S. aureus

There are many resistance genes and virulence factors in S. aureus that are encoded on mobile genetic elements, the majority of which have been reviewed recently (Malachowa & DeLeo, 2010). However, little mention has been made of conjugative transposons. This is most likely because their clinical importance in terms of conferred resistance is overshadowed by the SCCmec elements, which confer methicillin resistance. However, Tn916 and related elements have been reported in S. aureus. Transfer of plasmid-located and chromosomal copies of Tn916 from E. faecalis ISP1047 to S. aureus has been demonstrated (Jones et al., 1987). Additionally a survey of 37 tet(M) containing staphylococcal strains demonstrated, by dot blot hybridization, that two contained intTn (Poyart-Salmeron et al., 1991). Sequencing of the methicillin-resistant strain Mu50 revealed the presence of a putative conjugative transposon that was related to Tn916 and designated Tn5801 (Kuroda et al., 2001) (Fig. 1d). A comprehensive survey of 205 tetracycline-resistant S. aureus strains isolated from humans and animals (cattle, lamb and pigs) were screened for the presence of tet(M) and the integrase of either Tn916 or Tn5801 (which share 38.6% identity at the nucleotide level). Tn916-like elements were found in isolates from both humans and animals, whereas Tn5801 was found only in isolates from humans (de Vries et al., 2009). Additionally, one of these Tn5801-like elements was shown to transfer and designated Tn6014 (de Vries et al., 2009). There are multiple alleles of tet(M), each of which is usually associated with one type of mobile element; for example there is a tet(M) allele associated with Tn916 and another tet(M) allele associated with Tn5801 (de Vries et al., 2009).

Tn916-like elements in S. pneumoniae

Tn1545 was initially discovered in S. pneumoniae strain BM4200. The element is essentially Tn916 with an insertion of the erm(B) gene encoding macrolide, lincosamide and streptogramin (MLS) resistance and the kanamycin resistance gene aphA-3 (Courvalin & Carlier, 1986; Cochetti et al., 2008). Many clinical strains of S. pneumoniae carry tet(M), which is usually resident on Tn916/Tn1545-like elements; eight out of the 36 pneumococcal genomes currently sequenced contain one of these elements (Santoro et al., 2010). Since then various Tn916/Tn1545-like elements have been detected and characterized in S. pneumoniae or found in other streptococci. Early work aimed at detecting Tn1545-like elements by identification of the various resistance genes present in isolates by dot blot (Seral et al., 2001) showed a range of combinations of the genes present on Tn1545. Additionally, 63 out of 65 S. pneumoniae strains showed the presence of intTn demonstrating, that elements from the Tn916/Tn1545 group were likely to be common in this organism (Montanari et al., 2003). It is likely that this early work was in fact detecting some of the more recently characterized conjugative transposons such as Tn6002 and Tn6003. Tn6002 was initially characterized in Streptococcus cristatus from a clinical sample taken from a periodontal patient (Warburton et al., 2007). The element is essentially Tn916 with an insertion in orf20 (Fig. 3). The insertion contains five genes, one of which is erm(B) conferring the MLS phenotype upon its host (Warburton et al., 2007). Subsequently, Tn6002-like elements were found in many strains of S. pnuemoniae collected in Italy between 2000 and 2002 (Cochetti et al., 2007). These authors also found a derivative of Tn6002, which contained an additional insert of the macrolide–aminoglycoside–streptothricin (MAS) element within the erm(B) gene cluster of Tn6002 (Cochetti et al., 2007) (Fig. 3). Further analysis of the BM4002 reference strain yielded the full sequence of Tn1545, which was shown to be essentially Tn6003 with an IS1239 insertion upstream of orf12 (Fig. 3) (Cochetti et al., 2008).

Figure 3

Schematic representation of the diversity of mobile elements associated with Tn916 in Streptococcus pneumoniae. The colours are as in Fig. 1. The additional resistance genes are labelled vertically. The elements in which each insertion is found is labelled near the insertion; note that some insertions are present in more than one element and some elements, for example Tn2010, contain more than one insertion.

Another Tn916-like element in S. pneumoniae is Tn3872, which contains an insertion of Tn917 (Shaw & Clewell, 1985). Tn917 is a nonconjugative Tn3-like transposon which is responsible for the spread of erythromycin resistance in Japan (Okitsu et al., 2005) and Italy (Cochetti et al., 2008). Tn917 is found inserted into a number of different Tn916-like elements (Table 1 and Fig. 3).

Another gene encoding resistance to macrolides via the efflux mechanism is mef(E). This gene has been found to be associated with the Tn916-like elements Tn2009 (Del Grosso et al., 2004), Tn2010 (Del Grosso et al., 2006; Li et al., 2011) and Tn2017 (Del Grosso et al., 2009) (Table 1 and Fig. 3). The mef(E) gene is present on the macrolide efflux genetic assembly (MEGA) element (Santagati et al., 2000; Gay & Stephens, 2001).

Composite genetic elements have been found in S. pneumoniae, for example Tn5253 has a Tn916-like element, Tn5251, inserted into a larger element Tn5252 (Ayoubi et al., 1991). Previously known as ω(cat tet) BM6001 (Shoemaker et al., 1979), it also encodes resistance to chloramphenicol and is present in many common pandemic S. pneumoniae clones (Henderson-Begg et al., 2009). Tn5251 can excise from Tn5252 (Provvedi et al., 1996). More recently, the complete sequence of the element has been published (accession number FJ11160), which shows it is >99% identical to Tn916 (Santoro et al., 2010).

A genetic continuum and conclusions

When considering any mobile genetic element in any host, it must be remembered that due to horizontal gene transfer and the interactions between different elements it is likely that many new combinations of genes can be generated. One of the major players in the generation of these new elements are the Tn916/Tn1545 family of conjugative transposons. The host range of Tn916-like elements is very broad among Gram-positive bacteria and this means that the elements have the opportunity to acquire new genetic material from diverse sources. It is also worth noting that there are many Tn916-like elements in many species of bacteria that do not encode any obvious antibiotic resistance but instead contain alternative, and often cryptic, accessory genes.

What we know about the resistance genes on Tn916-like elements to date is not the end of the story; it is more of a snapshot in time. It is highly likely that new variants of Tn916, which contain other clinically important resistance genes capable of broad host range transfer, will be discovered. The limit to the number of genes carried by these elements may be directly related to their biological cost to the host cell. As Tn916-like elements have evolved intricate mechanisms of transcriptional regulation that likely result in a minimal biological cost to the host, it is plausible to imagine the emergence of a ‘super-transposon’ on which many clinically relevant antimicrobial resistance genes could exist. A worrying thought considering the lack of new antimicrobials that are currently coming to market!

Acknowledgements

Research in the authors' laboratories is carried out with financial support from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 241446 (ANTIRESDEV) and under grant agreement no. 223585 (Hyperdiff) and the Medical Research Council (grant no. G0601176).

Editor: Teresa Coque