Variations in SXT elements in epidemic Vibrio cholerae O1 El Tor strains in China

Vibrio cholerae O1 El Tor biotype strains are responsible for three multiyear epidemics of cholera in China during the seventh ongoing pandemic. The presence of the integrative conjugative element SXT is strongly correlated with resistance to nalidixic acid, tetracycline, and trimethoprim-sulfamethoxazole in these strains. Here, we sequenced the conserved genes of the SXT element, including eex, setR, and int, from 59 V. cholerae O1 El Tor strains and extracted and assembled the intact SXT sequences from the 11 genome sequenced strains. These elements had characteristics distinct from those of previously reported integrative conjugative elements (ICEs). They could be clearly divided into two types based on the clustering of conserved genes and gene structures of the elements, showing their possibly independent derivation and evolution. These two types were present before and after 2005, respectively, demonstrating the type substitution that occurred in 2005. Four to six antibiotic-resistant genes were found on the SXT elements, including genes resistant to tetracycline, trimethoprim-sulfamethoxazole, and multiple drugs. In summary, our findings demonstrated the roles of the SXT element in the emergence of multidrug resistance in epidemic O1 El Tor V. cholerae strains in China.

Scientific RepoRts | 6:22733 | DOI: 10.1038/srep22733 Elements of the SXT/R391 family share a genetic backbone of 52 syntenic genes, which mediate their integration/excision, conjugative transfer, and regulation during the life cycle of the ICEs 2 . In addition to this conserved core scaffold, these elements also harbor variable fragments, such as antibiotic resistance genes. The variable regions range from 30 to 60 kb in length and are mostly found at five insertion hotspots (HSs), termed HS1, HS2, HS3, HS4, and HS5, and four variable regions, named VRI, VRII, VRIII, and VRIV 8 .
Determination of the sequences of the ICE of the SXT/R391 family is essential for elucidating their organization and evolution. Though, an elaborate ICE capture system on plasmids has been developed to facilitate their sequencing and has been used for determining five ICE sequences 8 , it is an arduous task due to their size and predominantly chromosomal localization. In this study, we explored the types and evolution of SXT elements in Chinese V. cholerae O1 strains by clustering the backbone genes of the SXT element. We also extracted 11 complete SXT sequences from a genomic resequencing database of V. cholerae O1 El Tor strains established in our laboratory 9 . Our SXT genomic structure analysis described the differences between two sequence types of SXT elements, positioned the antibiotic resistance genes on the SXT elements, and presented the key role of this ICE in the emergence and transmission of multidrug resistance in V. cholerae.

Results and Discussion
Clustering of SXT elements by three conserved genes. The int gene encodes the integrase enzyme of the SXT element. This enzyme is required for the site-specific integration of the SXT element into the host chromosome and its excision from the chromosome 10 . Using primers designed for the int gene, the complete CDS regions of the int genes of the tested strains were amplified and sequenced. Alignment results showed that all of the Chinese strains had completely identical int gene sequences without any mutations in nucleotides. We then compared the int gene of our isolates with 43 full-length int gene sequences (1242 bp) in the GenBank database ( Fig. 1) and found that the SXT elements of the Chinese strains were grouped together in a cluster primarily composed of SXT elements from V. cholerae. The SXT elements could be separated from those identified in Alteromonas macleodii, Shewanella, and Providencia, but not in P. alcalifaciens.
SetR and SetC/D, the backbone genes located at the extreme 3′ -and 5′ -ends of the SXT elements, respectively, are the key regulators of SXT elements. Eex and TraG are inner membrane proteins of the donor and recipient cells that mediate entry exclusion in the SXT/R391 family of ICEs and are essential for ICE transfer. Based on the eex gene, EexS (SXT) and EexR (R391), the only two exclusion groups in the large family of ICEs, are divided 11 . As shown in Figs 2 and 3, in the cluster trees for the setR and eex genes, the SXT elements of the Chinese strains were distinctively divided into two groups. Generally, the SXT elements of strains isolated before 2005 were clustered together, while the strains isolated between 2005 and 2010 clustered together in another group. The eex and setR genes of the strains in the 2005-2010 cluster showed high homology to many SXT elements isolated from other parts of the world, such as the ICEVchMex1 (Mexico, 2001), ICEVchind5 (India, 1994), and VC1786ICE (Haiti, 2010) elements. The eex genes of the strains in the pre-2005 cluster showed 100% homology only with ICEPmiUsa1 (Proteus mirabilis strain HI4320, United States of America, 1986) and ICEVchCHN1307 (V. cholerae, China, 1998), which was also identified by our laboratory. Therefore, there was an obvious type substitution for the SXT element in China, although the strains isolated in a region for several consecutive years had SXT elements with essentially the same eex and setR genes.
Genetic structure of the SXT elements. Within the 11 strains that have been completely sequenced, the attL and attR genes are located on same scaffold, and the spaces between these genes are less than 120 Kb, which is similar to the length of the SXT element. BLAST searches with the prfC gene showed that the prfC genes are adjacent to the attR site and have dozens of bases missing from the 5′ -end, consistent with the SXT recombination process. The results of these two methods were completely consistent. The 11 SXT elements were designated according to the nomenclature proposed for this family of elements (e.g., ICEVchCHN143) 4 . These elements are listed in Table 1.
The genomes of the 11 detected SXT elements were 9.25-11.03 Kb in length and had 87-102 predicted ORFs (Supplemental Table S1). The sequences of ICEVchCHN2605 and ICEVchCHN143 were found to be similar, except that ICEVchCHN2605 lacked a T at site 97,900. Furthermore, ICEVchCHN1605 was identical to ICEVchCHN143, but was missing 39 bp from positions 97,901 to 97,939. These differences were not related to the coding region; therefore, we grouped these three SXT elements as one subset of ICE.
The genetic organization of these 11 SXT ICEs was similar to that of other members of this family (Fig. 4). A total of 70 ORFs were commonly shared by these SXT elements, most of which were included in the set of previously reported conserved core genes 2 . However, an entire 17.8-kb region from s026 to s040 was missing from all 11 SXTs in our study, which was different from the ORFs of previously reported ICEs. The ORFs s084 and s083 integrated into ORF4 through point deletion of a T at site 267 and a frameshift mutation in s084. These replacements may not have an obvious effect on the SXT because deletions of core genes of unknown function, namely, s082, s083, and s084, did not influence SXT transfer frequency. Furthermore, the gene cluster from s027 to s040 is not required for SXT MO10 mobility or maintenance 8 .
Five conserved insertion hotspots are located between s043 and traL (H1), traA and s054 (H2), s073 and traF (H3), traN and s063 (H4), and s025 to traID (H5) 8 . In H1, only AHV1003 and -2255 have insertions of a cluster of four genes. In H2, all 11 SXT elements have insertions of the same two genes, s053 (Ync) and s052 (Ynd). Similarly, all 11 SXT elements have insertions in H3 and H4. Eight SXT elements, namely, ICEVchCHN143, -2605, -1605, -1627, -1909, -1944, -4210, and -956, have the same insertion at these two sites. Among these hotspots, H5 has the largest number of inserted genes and gene combinations. This spot had insertions of four types of gene clusters containing 6-10 ORFs oriented in the same direction as s025 and traI. None of the 11 SXT elements had any insertions in the four variable regions VRI-VRIV. Additionally, the main insertion site of the 11 SXT elements was between umuc and s021. Three insertion regions were formed at this spot in transposon-like structures, and these regions carried many antibiotic resistance genes. One of these regions was related to the ISCR2 element, which is an IS91-like transposable element that tends to accumulate antibiotic resistance genes.
The restriction-modification system (RM) is a system of genes encoding the functions of DNA modification, recombination, and repair 12 . RM is composed of three polypeptides: R (restriction endonuclease), which recognizes and cut specific DNA sequences; M (modification), which methylates the same sequence to inhibit DNA cleavage and protect the host cell against invasion of foreign DNA; and S (specificity), which determines the specificity of both R and M. In the ICE family, ICEVchMex1 13 and ICESpuPO1 14 encode putative type I RM systems, and ICEVspPor3 and ICEValSpa1 harbor a putative type III RM system 15 . Two ICEs in our study, AHV1003 and -2255, were found to have a completely similar type I RM system in HS5. This RM was identical to the RM of the V. cholerae O1 strain EM-1536 and the Alteromonas macleodii strain MED64. However, this RM had very low similarity to the type I RM of ICEVchMex1, with 18% protein sequence similarity in the R subunit and 14% similarity in the M subunit.
Of the 70 commonly shared ORFs of the 11 SXT elements, 1-49 showed nearly 100% homology to the ORFs between PMI2423 and PMI2482 of P. mirabilis strain HI4320 (GenBank: AM942759), which carries a conjugative transposon (PMI2423-PMI2491) similar to the ICE R391 from P. rettgeri 16 . The homologies of these ORFs were even higher than their homologies to the ICEs of V. cholerae (Supplemental Table S1). The remaining 20 ORFs shared 100% homology to nucleotide sequences of the ICEs of V. cholerae, half of which showed no homology to the P. mirabilis strain HI4320 genome. This may support the recombination of the different origins of these Chinese SXT ICEs. Based on the insertion sites of the 11 SXT elements (Fig. 4), we could easily group these elements into two distinct types. The first type included ICEVchCHN2255, ICEVchCHN57, and AHV1003. ICEVchCHN 2255 was more similar to AHV1003 than to ICEVchCHN57. The remaining SXT elements belonged to the second type, which could be further divided into three subsets: the first subset consisted of ICEVchCHN143, -1605, and -2605; the second subset consisted of ICEVchCHN1627, -4210, and -956; and the third subset consisted of ICEVchCHN1909 and -1944. This was consistent with the groups divided based on the eex and setR genes.
In addition, some backbone genes differed between these two types. For example, the length of the hypothetical protein next to TraL had a difference of six nucleotides between the two types. ORF29, which is adjacent to s063, was not present in ICEVchCHN2255, ICEVchCHN57, and AHV1003 because of a 'T' to ' A' transition at nucleotide 12 and premature termination. ideA, i.e., the backbone gene s062, is an ICE-encoded DNase that was recently shown to be necessary for inhibiting the natural transformation of Haiti outbreak strains 17 . This gene is 684 bp long and encodes a putative 227-aa-long protein in VC1786ICE. The ideA gene of ICEVchCHN57 is 100% homologous to VC1786ICE, but is absent in ICEVchCHN2255 and AHV1003. For the other nine SXT elements, the genes are 675 bp in size, with 90% nucleotide and 91% amino acid similarities to the corresponding sequences in VC1786ICE and ICEVchCHN57. Detailed information regarding the 11 SXT element sequences reconfirmed that there were two SXT types in the Chinese epidemic O1 V. cholerae strains. The major differences between them indicated that they may have different origins and would have evolved independently.
Antibiotic resistance genes in the SXT elements. The most important feature of SXT elements is that they carry multidrug resistant genes and transfer them between bacteria. Most ICE antibiotic resistance genes are embedded in a composite transposon-like element that interrupts the SXT-encoded rumAB operon. As shown in Fig. 4, the antibiotic resistance genes of the SXT elements in this study are also focused in the same region, which is partitioned into the three insertions by transposase. These antibiotic resistance genes appeared to have been obtained through repeated transposition. The genes sulI, encoding dihydropteroate synthase and sulfamethoxazole resistance (Su), and strAB, encoding streptomycin resistance (Sm), were present in all of the SXT elements analyzed in this study.
In the first insertion fragment, the SXTs of strains 1627, 4210, and 956 contained a resistance gene cluster containing tetAR, merR, and strB; this cluster conferred resistance to tetracycline, mercury, and Su. Insertion 2 contained the largest resistance genes, including the floR gene, which confers resistance to chloramphenicol, and the dhfR gene, encoding a trimethoprim-resistant dihydrofolate reductase, and was present in eight of the 11 SXT elements (absent in ICEVchCHN2255, ICEVchCHN57, and AHV1003). Furthermore, ICEVchCHN1909 and ICEVchCHN1944 possessed identical copies of the dhfR gene separated by the transposase family gene tnp. The deduced amino acid sequences of these dhfR genes had 69% identity with the folA gene of Riemerella anatipestifer RA-CH-2, encoding dihydrofolate reductase, and 16% identity with drf18 of the O139 serogroup strain MO10. Although ICEVchCHN2255, ICEVchCHN57, and AHV1003 lacked the dhfR gene at this position, they contained another copy of dihydrofolate reductase, which was identical to the dfrA1 gene in the class I integron of Klebsiella pneumoniae subsp. pneumoniae strain B-1104. The dfrA1 gene was located at an insertion of 5.022 kbp between traF and an ORF of unknown function, orf73, similar to the position of dfrA1 from V. cholerae O1 serogroup strain C10488 18 . Therefore, in the SXT elements of Chinese El Tor strains, resistance to trimethoprim was encoded either outside or inside the clusters of other resistance genes, but the homology between these two types of trimethoprim resistance determinants was only 27%, and both determinants could not simultaneously exist in one SXT element. In insertion 2, ICEVchCHN2255, ICEVchCHN57, and AHV1003 also contained a copy of the tetAR gene. This gene, which encodes a tetracycline efflux protein of class A, was 100% homologous to the ICEVspSpa1 genes of V. splendidus. The tetAR gene in insertion 1 belonged to class B and was similar to the genes of V. cholerae CP1042. However, there was only 50% homology between these two types of tetracycline-resistance genes. In insertion 3, only ICEVchCHNAHV1003 carried three resistance-related genes, namely, mphR, mrx, and mphK, which confer resistance to erythromycin.  In our previous study profiling the antibiotic resistance of Chinese O1 El Tor strains 1 , the presence of the SXT element was highly correlated with resistance to nalidixic acid, tetracycline, and trimethoprim-sulfamethoxazole. The present study confirmed that the SXT element was indeed a vector of genes conferring resistance to tetracycline and trimethoprim-sulfamethoxazole and genes causing multidrug resistance.
In conclusion, our findings showed that the SXT elements in Chinese epidemic O1 V. cholerae strains were split into two types. All of these elements had their own molecular characteristics distinguishing them from the other ICEs and exhibited obvious differences in terms of the sequences of many conserved backbone genes as well as the overall structure. However, strains isolated from a region for several consecutive years were found to have the same SXT type. In 2005, there was an obvious type substitution for the SXT element in the Chinese epidemic strains. Regardless of these differences, SXT elements of both types carried at least five drug resistance genes and played a key role in the emergence of multidrug resistance in O1 serogroup strains.

Materials and Methods
Strains. Based on our previous study 1 , 59 strains were retained from the 169 SXT element-positive strains by removal of strains having the same isolation time, region, and antibiotic resistance spectrum. These 59 V. cholerae O1 El Tor strains were isolated between 1992 and 2010 from 18 provinces in China. All strains were positive for the eex, setR, and int genes of the SXT element, and 11 of the strains were subjected to whole genome sequencing in our laboratory.
Extraction of SXT element sequences. Raw reads data of the resequenced genomes database of the Chinese O1 El Tor V. cholerae set in our laboratory were assembled de novo with SOAPdenovo (V 2.04) using optimal Kmer and minimal coverage parameters. The SXT element sequences were then extracted from the whole genome by two methods. The first was based on the consensus sequence of attL and attR. According to the alignment of the nucleotide sequences of V. cholerae chromosomes on both sides of the SXT element 5 , the consensus sequence could be determined as 'O' = ATCATCTCGCACCCTGA. Scaffolds with attL and attR with a space of less than 120 kb were selected, and the sequences between the two consensus sequences were extracted as the SXT element. The other method was BLAST with the prfC gene sequence. Sequences with attR following insertion at the 5′ -end of the prfC gene were extracted as the SXT element.