Novel genomic islands and a new vanD-subtype in the first sporadic VanD-type vancomycin resistant enterococci in Norway

Background Vancomycin-resistant enterococci (VRE) represent several types of transferable vancomycin resistance gene clusters. The vanD type, associated with moderate to high level vancomycin resistance, has only sporadically been described in clinical isolates. The aim of this study was to perform a genetic characterization of the first VanD-type VRE strains detected in Norway. Methods The VanD-type VRE-strains (n = 6) from two patient cases were examined by antimicrobial susceptibility testing and whole genome sequencing (WGS) to uncover Van-phenotype, strain phylogeny, the vanD gene clusters, and their genetic surroundings. The putative transferability of vanD was examined by circularization PCR and filter mating. Results The VanD-type Enterococcus faecium (n = 4) and Enterococcus casseliflavus (n = 2) strains recovered from two cases (A and B), expressed moderate to high level vancomycin resistance (MIC 64—>256 mg/L) and various levels of teicoplanin susceptibility (MIC 2—>256 mg/L). WGS analyses revealed phylogenetically different E. faecium strains (A1, A2, and A3 of case A and B1 from case B) as well as vanD gene clusters located on different novel genomic islands (GIs). The E. casseliflavus strains (B2 and B3 of case B) were not clonally related, but harbored nearly identical novel GIs. The vanD cluster of case B strains represents a novel vanD-subtype. All the vanD-GIs were integrated at the same chromosomal site and contained genes consistent with a Clostridiales origin. Circular forms of the vanD-GIs were detected in all strains except B1. Transfer of vanD to an E. faecium recipient was unsuccessful. Conclusions We describe the first VanD-type E. casseliflavus strains, a novel vanD-subtype, and three novel vanD-GIs with a genetic content consistent with a Clostridiales order origin. Despite temporal occurrence, case A and B E. faecium strains were phylogenetically diverse and harbored different vanD subtypes and vanD-GIs.


Introduction
Vancomycin resistant enterococci (VRE) have become a global nosocomial problem three decades after the first description in the late 1980s [1]. Eight different acquired vancomycin resistance gene clusters (vanA, vanB, vanD, vanG, vanE, vanL, vanM, and vanN) have been identified [2]. The vanC gene cluster is intrinsic in E. casseliflavus and E. gallinarum [2]. In general, van gene clusters encode three groups of co-acting enzymes; 1) enzymes necessary for the synthesis of new peptidoglycan precursors, 2) enzymes that erase the inherent D-Ala-D-Ala-ending precursors, and 3) a two-component signal transduction system for inducible resistance [3]. The normal enterococcal cell wall side chain terminal residue D-Ala-D-Ala, to which vancomycin binds with high affinity, are replaced by D-Ala-D-Lac in vanA, vanB, vanD, and vanM gene clusters or D-Ala-D-Ser in the other van gene clusters [3]. Vancomycin binds to D-Ala-D-Ser with seven times lower affinity compared to D-Ala-D-Ala, causing lowlevel vancomycin resistance, while the binding affinity of vancomycin to D-Ala-D-Lac is almost 1000 times lower mediating high-level resistance [4]. The vanA and vanB clusters dominate worldwide, likely due to linkage to successful mobile genetic elements (MGEs) [5]. Although the vanA, vanB, and vanD clusters have a similar organization, the vanD gene clusters have so far only been sporadically described on chromosomal genomic islands (GIs) that have not been shown to be transferable between enterococci [6][7][8][9]. The vanD gene cluster has up till now been reported in five species of enterococci (Enterococcus faecium, Enterococcus faecalis, Enterococcus gallinarum, Enterococcus avium, and Enterococcus raffinosus) [10].
The VanD-phenotype is characterized by moderate to high level vancomycin resistance and various levels of susceptibility to teicoplanin [3,11,12]. The housekeeping ddl gene (D-Ala-D-Ala ligase) is often inactivated by mutations in vanD containing strains causing an impaired chromosomal peptidoglycan synthesis pathway and addiction to vanD-expression as the alternative peptidoglycan precursor pathway [3,7,13]. Based on sequence differences, there are five known subtypes of vanD. The sequence diversity in vanD gene cluster subtypes mostly is in the vanY D , vanH D , vanD, and vanX D genes and at the intergenic sequence between the two operons of the cluster [11]. VanD VRE are rare and have only been reported sporadically from the Netherlands, France, Canada, Japan, Sweden, Australia, the US, and Brazil during the last decades [7,8,10,[12][13][14][15][16][17][18].
In this study, we aim to determine the genetic relatedness between the first Norwegian VanD-type VRE strains, their Van-phenotype, and the putative MGEs harbouring the vanDgene cluster.
Since this study contain only limited anonymized patient data, the study was approved by the Data Protection Officer at Oslo University Hospital and the Chief of Department of Microbiology at St Olavs Hospital. The written consents of the patients were obtained to use anonymized data from their patient journal in publication of this work.

VRE strains and data collection
The first two cases of VanD-type VRE were identified in Norway in 2017. The Norwegian National Advisory Unit on Detection of Antimicrobial Resistance received the strains for further characterization (Table 2). Three VanD-positive E. faecium (VanD-type VREfm) (A1, A2, and A3) strains were isolated from case A. The strains of case A were recovered from a subphrenic abscess (A1 and A2) and through rectal screening (A3). A month later, a VanD-type

Species identification and Whole Genome Sequencing (WGS)
Strains were subcultured on blood agar to ensure pure culture. Species identification was performed by MALDI-TOF (Bruker, Billerica, USA) according to the manufacturer's instructions. Genomic DNA was extracted using DNeasy Blood and tissue kit (Qiagen, Hilden, Germany). The total DNA concentration was quantified by Qubit fluorometer (Invitrogen, Thermo Fisher Scientific). Libraries were prepared by the Nextera XT DNA library preparation kit (Illumina, San Diego, USA) and sequenced using Illumina NextSeq500 and the Mid Output 300 cycles cell.

Genomic analyses
Adapter removal and quality trimming of the raw reads were performed by trimmomatic v0.39 [22]. Later, genome assembly was done using SPAdes v3.13.0 [23] and the quality of assembled genomes was assessed using QUAST v5.0.2 [24]. The annotation of the transposons was carried out using the National Center for Biotechnology Information (NCBI) prokaryotic genome annotation pipeline (PGAP) [25]. Antimicrobial resistance (AMR) genes were

PLOS ONE
The first Norwegian vanD clusters containing VRE identified in silico from the assemblies using NCBI bacterial AMR reference gene database (PRJNA313047) [26] in ABRicate tool v0.8.7 [27]. Identification of Type IV secretion systems genes was carried out by BLASTp [28] searches against the SecReT4 database [29].

Phylogenetic analyses
To explore the phylogenetic relationship between the vanD strains and publically available genome sequences on NCBI, the global phylogenetic trees were generated based on the core genome. All closed genomes of E. faecium (n = 135) and E. casseliflavus (n = 3) from NCBI as of 04.04.2020 were retrieved and phylogenetic trees were constructed using Parsnp v1.2 [30]. Another core genome SNP tree was built for the publicly available VanD-type VREfm genome sequences together with the Norwegian vanD-type VREfm. Also, a SNP tree was generated for vanD gene cluster sequences using parsnp. Multilocus Sequence Typing (MLST) was performed using MLST tool version 2.11 [31]. For high-resolution typing, Minimum Spanning Tree was generated based on the 1423 core genes of E. faecium scheme of SeqSphere+ software V6.0.2 (Ridom GmbH, Münster, Germany [http://www.ridom.de/seqsphere/]). We used the default � 20 allelic differences as a threshold for cluster calculation and clonal relatedness [32].

Comparative genomics
The closest non-VRE strains to each of the Norwegian VanD-type VRE were selected from the global phylogenetic tree. We used Mauve [33] to sort the contigs according to the reference genomes (E1 (NZ_CP018065.1) for A1-3 strains, E4402 (NZ_LR135174) for B1 strain, and EC20 (CP004856.1) for B2-3 strains) followed by Easyfig v2.2.2 [34] for comparison. The Artemis comparison tool [35] was used to visualize the BLASTn v2.6.0 search result and to locate the mobile genetic structures containing vanD gene clusters and their insertion site in the genome. Sequences of the GIs harboring the vanD gene clusters were BLASTed against the NCBI nr database to find the homologous sequences. Pyani v0.2.7 was used to determine the average nucleotide identity (ANI) between genomes, GIs and vanD gene clusters [36]. For the novel GIs, transposon numbers were registered at the Transposon Registry [37].

Excision of putative GIs
The ability of the GIs to circularize was examined by PCR using the following pair of primers which directed outwards from the GIs ends: 5´-GCGTGAGAAGCTGACAACAA-3´and 5´-GTTTCAGCCGCCAACTATTC-3´. Subsequent Sanger sequencing of PCR products using BigDye 3.1 technology (Applied Biosystems, CA, USA) was performed to confirm the expected sequence.

Transferability of putative GIs
Transferability of vanD gene clusters was examined as described previously [38] using E. faecium BM4105-RF [39] as a recipient. To determine transfer frequency, colony forming units were counted on Brain heart infusion agar with rifampicin (30 mg/L) and fusidic acid (20 mg/ L), and/or vancomycin (8 mg/L).
Despite an increasing prevalence of VRE in Norway since 2010, only vanA and vanB have been reported until now [40]. The detection of VanD-type VRE from two different patients within two months in 2017, therefore raised a concern of facing a VanD-type VRE outbreak in Norway, although no obvious epidemiological link between the patients was identified. Thus, the pheno-and genotype of the six VanD VRE strains were examined (

AST results
The AST-results are summarized in Table 2. Briefly, all strains expressed high-level vancomycin resistance (MIC � 64 mg/L), various levels of susceptibility to teicoplanin (MIC 2 mg/L to >256 mg/L), and susceptibility to linezolid. All four E. faecium strains were ampicillin resistant and three also demonstrated high-level gentamicin resistance.
In silico analysis showed that all strains contained the vanD gene cluster integrated into their chromosome. The E. casseliflavus genomes (B2 and B3) also contained the intrinsic vanC gene cluster [2]. In the E. faecium strain B1, alignment of the housekeeping D-Ala-D-Ala ligase deduced from the ddl gene sequence showed a truncated protein of only 110 amino acids caused by a deletion resulting in a frameshift and a premature stop codon ( Table 2 and  All the other VanD-type VREfm strains showed point mutations in essential positions that presumably could lead to a non-functional Ddl ligase. In the literature, most VanD-type VRE strains described have had an impaired Ddl ligase and are thus dependent on the constitutively expressed vanD cluster to synthesise peptidoglycan [10].

The VanD E. faecium strains from the two cases were not closely related
The VanD VREfm strains from cases A and B had different MLST profiles (Table 2). A1 and A2 genomes had an identical MLST profile which was registered as the novel ST1486, a single locus (ddl allele) variant of ST117 (strain A3) belonging to the hospital associated ST78 lineage. The E. faecium strain from case B belonged to ST203 which is part of the ST17 hospital associated lineage. Population genetic modeling based on the seven MLST genes using the Bayesian Analysis of Population Structure (BAPS) software have shown that 80% of the E. faecium nosocomial strains cluster in two different groups (2-1 and 3-3) [41]. E. faecium A and B strains belonged to lineages within these different main BAPS groups (lineage ST78 to 2-1 and lineage ST17 to 3-3) [41], confirming a large phylogenetic distance. This was further shown by cgMLST analysis which revealed that A1-3 strains belonged to the same novel cluster type (CT) 3198 (Fig 1). The B1 strain belonged to another novel CT3199 and showed at least 354 allelic differences to A1-3 strains. The two ST1486 strains had only one allelic difference, while the maximum allelic differences (eight) within CT3198 were between A1 and A3. One of these allelic differences was in the ddl allele which is one of the seven MLST scheme genes. Our results show that even strains with different MLST profiles could be clonally closely related and have the same CT.
For E. casseliflavus strains, a core genome SNP tree was constructed together with publically available closed genomes. Interestingly, the two VanD strains (B2 and B3) clustered in two separate branches, showing that they were not clonally related (S3 Fig). The vancomycin susceptible E. faecium strain E1 (GCF_001886635.1) isolated from Spain in 2010, was identified as the closest genome to A1-3 strains using a core genome SNP tree of all closed E. faecium genomes in NCBI and the Norwegian VanD-type VREfm genomes (S4 Fig). Strain E1 was therefore used as a reference genome for sorting contigs and further comparative genomic analyses. Genomic comparison using Easyfig confirmed that the A1-3 genomes were very similar. The ANI between A1 and A2 was the highest (99.99%).

Comparison of case B VREfm (B1) to case A VREfm genomes, confirmed observed genomic differences (S5 Fig).
The significant phylogenetic difference between the vanD E. faecium strains from case A and B is consistent with the observed sporadic occurence of vanD-type VRE strains in contrast to the epidemic vanA/B-type VRE [7,12,13,15]. Our patient characteristics with underlying diseases and long-term antibiotic exposure including vancomycin are also consistent with previous observations in vanD VRE cases [12,17].

A novel vanD-subtype was found in strains from case B
Sequence comparison and phylogenetic analysis of complete vanD gene clusters from this study and reference sequences representing the five known vanD subtypes (vanD1-D5) [8,11,42,43], showed that the Norwegian vanD gene clusters belonged to two different vanD-subtypes. In case A, the vanD gene clusters of strains A1 and A2 were 100% identical and showed 99.96% ANI to the cluster in A3. The vanD genes of case A clustered with the vanD5 reference sequence (E. faecium strain N03-0072) (Fig 2). ANIs between the vanD5 reference sequence and A1-3 strains were >99.9%. In case B strains, B2 and B3 vanD gene clusters were 99.98% identical and the B1 vanD gene cluster showed > 99.96% ANI with them. The ANI between case A and B vanD gene clusters was around 91%. B1-3 vanD gene clusters are significantly different from the known vanD-subtypes (maximum 93.7% identity to the known subtypes) (S1 Table). Thus, we propose that the B vanD gene cluster is a new subtype termed vanD6. Identification of the novel vanD6 gene cluster in two different species of enterococci suggests interspecies genetic exchange.

Three novel vanD-containing GIs identified
Comparison alignments with non-VRE reference genomes using Artemis comparison tool showed that all vanD gene clusters in the Norwegian vanD-type VRE were part of GIs ranging between 112-126 kb ( Table 3). The GC content of the GIs was higher (44.1-44.3%) than the average GC content range of 38% of E. faecium strains [44][45][46]. For B2 and B3 E. casseliflavus strains, the genomic GC content was 42.4% and 42.3%, in contrast to 44.6 and 44.7% for their GIs, respectively. The GI Tn6711 of A1-3 strains showed identical size and had an ANI above 99.99% suggesting a common origin. The GI Tn6713 of the E. casseliflavus strains (B2 and B3) was identical in size and showed only 0.001% difference (S2 Table). The GI Tn6712 in E. faecium strain B1 was 7230 bp larger than that of E. casseliflavus GI (Tn6713), while it was 6134 bp kb shorter and showed more rearrangements compared to Tn6711 of strains A1-3 (Table 3 and Fig 3). ANIs were lowest (below 98%) between case A and B E. faecium GIs (S2 Table). Thus, the overall genetic differences between the GIs of A1-3 and B1-3, do not support a direct  spread between the two cases. However, in case B strains, we suggest one genetic event has evolved Tn6713 of E. casseliflavus to the longer Tn6712 in E. faecium or vice versa (Fig 3). All GIs lacked conjugative apparatus genes and the vanD gene cluster was the only AMR gene within the islands (GenBank Acc. No. MT951615-7). The nucleotide sequence of integrase genes in Tn6712 and Tn6713 was identical and had only one SNP compared to Tn6711. Despite the existence of the same GIs in E. casseliflavus strains (B1 and B2) of case B, the ANI between their genomes (95.1%) was too low to be clonally related. This observation strongly suggests separate acquisitions of Tn6713 in B2 and B3 strains.
Comparisons of the Norwegian vanD-GIs to those of the newly isolated VanD-type VREfm from the Netherlands and Japan with publically available WGS data revealed a high rate of identity. Two VanD-type Dutch VRE strains (E8429 and E9354) [7] contained vanD-GIs with 99.99% sequence identity to Tn6712 of B1. Moreover, the vanD5-containing GI from the Japanese E. faecium SMVRE20 [17] (AP019408.1) showed 99.98% sequence identity to Tn6711 of case A. Another Japanese vanD-GI (157 kb) from E. faecium strain AA620 (LC467712.1) showed 96% identity covering 81% of Tn6711. Although the vanD-GIs are similar between the Norwegian, Dutch, and Japanese VREfm strains, phylogenetic analyses based on SNPs suggest that the strains are not closely related (S6 Fig). The GI of the Japanese SMVRE20 has an additional gene compared to Tn6711. Likewise, Tn6712 and the Dutch GIs show only one gene in difference. Both these genes encode hypothetical proteins (Fig 3). The high identity between Tn6711 and the GIs of the Japanese VanD-type VREfm and between Tn6712 and two Dutch VanD-type VREfm GIs indicate a global spread of similar MGEs.
Due to the intrinsic vanC gene cluster of E. casseliflavus clinical strains, they already express low level resistance to vancomycin. Thus, E. casseliflavus strains often are not investigated further to see if they contain additional van clusters. In this study, we show that E. casseliflavus may be the intermediate source of the vanD type cluster containing GI (Tn6713) that spread to E. faecium (Tn6712) in case B. Based on this finding, MIC investigation of clinically important strains of E. casseliflavus should be considered to reveal possible acquired van gene clusters.

The GIs show site specific integration in E. faecium and E. casseliflavus
The insertion sites of the vanD GIs were identical for all six strains and located in the 3' end of the lysS gene which is positioned upstream of a 16S ribosomal rRNA gene. The integration resulted in a 13 bp direct repeat located 17 bp from the 3' end of the lysS gene. The left and right repeats in the different vanD-containing strains showed maximum one SNP difference. For case A GIs the imperfect direct repeats were identical. In strain B1 of case B, the repeat is identical to case A GIs but localised on opposite sides. The perfect direct repeat in B3 differed by one nucleotide compared to the other strains ( Table 3). The same integration site was also found in the recently isolated Dutch and Japanese VanD-type VREfm [7,17]. Thus, this insertion site may be a hotspot in some enterococcal species including E. faecium and E. casseliflavus.

Putative origin of vanD-containing GIs
BLAST searches revealed 89% identity with several regions of Blautia producta SCSK genome covering only 59% of the Tn6711 length. Another hit of Tn6711 BLAST showed 89% identity to Blautia coccoides YL58 with 59% coverage, spanning some small fragments that were not covered by B. producta SCSK. An even higher identity (93%) was seen between the shorter Tn6712 and Tn6713 with fragments from B. coccoides YL58 covering 59% of these GIs. Previous reports have shown that vanD-type vancomycin resistance gene clusters can be found in non-enterococcal species like Ruminococcus gauvreauii, Lachnospiraceae bacterium, and Ruthenibacterium lactatiformans [7]. The above mentioned species and Blautia genus belong to the same taxonomic order of Clostridiales and are found in both the human and animal gut microbiome [47][48][49]. Thus, anaerobic Blautia genus or other members of the Clostridiales order are possible sources for vanD GIs.

Activity and transferability of putative GIs
Mobile chromosomal genetic elements, excise and circularize before transfer [50]. Circularization PCR and amplicon sequencing confirmed that Tn6711 and Tn6713 were able to circularize supporting that they are active MGEs. Agarose gel electrophoresis of PCR products repeatedly showed stronger bands for Tn6713 in E. casseliflavus which could be due to higher activity compared to Tn6711 in E. faecium (S7 Fig). However, we were not able to transfer vanD to an E. faecium recipient in this study (detection limit 10 −10 to 10 −9 transconjugants/ donor cell) which is not surprising since a conjugation apparatus was not found in any of the GIs carrying the vanD gene clusters nor in other sites of the VanD-type VRE genomes. Type IV secretion systems play an important role in conjugation and can mediate the transfer of the conjugative plasmids and transposons. They have an impact on the spread of antimicrobial resistance among bacteria [29]. Non-conjugative MGEs can use the conjugative apparatus of other MGEs to mobilize. Thus, a mobility test can be conducted to confirm mobilization of the GIs [38,51]. However, the strains in this study already had several acquired resistance determinants that are used as markers in mobilization tests. Thus, we did not attempt to mobilize the islands.

Conclusions
We have performed a genetic characterization of the first VanD-type VRE strains recovered from two patients treated with broadspectrum antibiotics including vancomycin before VRE detection. All VanD-type VRE strains of case A were E. faecium while both vanD E. casseliflavus and E. faecium were recovered from case B. To our knowledge, this is the first two vanD E. casseliflavus strains reported. Based on our finding, we recommend MIC investigation of clinically important E. casseliflavus strains to reveal possible additional van gene clusters. In the VREfm strains of case A, we identified a unique novel ST1486, an SLV of ST117, which were phylogenetically distant from case B VREfm (ST203). Sequence analyses revealed a novel vanD-type cluster termed vanD6 subtype in case B strains. The large phylogenetic distance between the VREfm strain of the two cases, as well as differences in vanD-cluster subtypes and vanD-GIs, rejected the hypothesis of a clonal outbreak. We identified three novel similar vanD-GIs of putative Clostridiales order origin integrated at the same chromosomal site in both E. faecium and E casseliflavus.