We performed WGS analysis (Appendix 1) on 42 VREF isolates (primarily collected after April 2022) from 39 patients across 7 hospitals, designated SZ_A through SZ_F, including 2 affiliated hospitals, SZ_C1 and SZ_C2, grouped as SZ_C. We used WGS to identify STs and used phylogenetic analysis to determine ST sources in a global context. We assessed genetic diversity, indicating mutation rates during circulation, using pairwise core genome single-nucleotide polymorphism (cgSNP) distance. We characterized population structure to show divergence and emergence of novel variant populations.

Among the 42 isolates, 34 (81%) were ST80 isolates, 7 (17%) were ST80 variant isolates (ST80_GDvariant1) with ddl loci mutated from 1 to 194, and 1 (2%) was an ST78 isolate that was collected in 2023 (Appendix 1 Figure 1, panel B; Appendix 2). Unexpectedly, ST80_GDvariant1 isolates independently emerged in 5 branches and did not originate from a single mutation event (Figure 1; Appendix 1 Figure 2, panel A).

Phylogenetic analysis inferred from cgSNP of the 41 isolates (ST80 and variants) and 703 other publicly available ST80 isolates revealed that the 41 isolates from the ongoing outbreak in Guangdong are affiliated with SC11 (Appendix 1 Table 2, Figure 2, panel A) (9). Within SC11, two 2022 strains from Guangzhou, SZYSC_ZDYVRE007 and SZYSC_ZDYVRE008G, with a 5-SNP divergence, clustered adjacent to the lineage root (Figure 1; Appendix 1 Figure 2, panel A) and formed a distinct VREF sublineage (SC11-root sublineage) with a 37-SNP average divergence from other SC11 strains (bootstrap = 72) (Figure 1). In contrast, the remaining SC11 strains exhibited tight clustering (pairwise distances of <5 SNPs [36.5%] and 5–10 SNPs [49.8%]) and formed a dominant clade (SC11-outbreak sublineage) (Figure 1). Those findings suggest that the SC11-outbreak lineage originated from a single progenitor or highly related lineage, enabling rapid transmission.

Figure 2

Figure 2. Lineages detected in a study of rapid transmission and divergence of vancomycin-resistant Enterococcus faeciumST80, China, showing 2 lineages circulating in parallel. A) SC11 subpopulations labeled on pangenomic tree...

To resolve strain differentiation, we analyzed SC11-outbreak_lineage population structure by using PopPUNK (https://github.com/bacpop/PopPUNK), which integrates core and accessory genomic variation (10). We delineated 2 subpopulations, SC11-pop I and SC11-pop II (Figure 2, panel A; Appendix 1 Figure 4). SC11-pop II isolates formed tight clusters in the pangenome-based tree (Figure 2, panel A) but dispersed in the cgSNP phylogeny (Figure 1), suggesting subpopulation divergence was primarily driven by gene gain or loss.

Figure 3

Figure 3. Alignment of SC11 to determine divergence during rapid transmission and divergence of vancomycin-resistant Enterococcus faeciumST80, China. To determine the genes associated with the divergence of SC11 we aligned...

SC11-pop I and SC11-pop II circulated in parallel for >3 years (2021–2024), and SC11-pop II showed broader transmission (Figure 2, panel B) and maintained ≈15% prevalence (Figure 2, panel C). SC11-pop II showed higher prevalence than SC11-pop I in Shenzhen and other provinces (Figure 2, panel B). Genetically, SC11-pop II exhibited enhanced horizontal gene transfer activity, carrying more insertion sequences (ISs) (Appendix 1 Figure 5), plasmid-like elements (Appendix 1 Figure 6), and diverse antimicrobial resistance genes (Figure 3). Although all SC11 isolates harbored the VanA operon, 8 SC11-pop II strains uniquely acquired VanM operon (Figure 3). SC11-pop II cases showed trends of increased hospitalization and underlying conditions, including hypertension and cardiovascular, respiratory, and kidney diseases, but statistical significance was limited by the sample size (Appendix 1 Table 5). Expanded surveillance is required to clarify clinical distinctions between SC11-pop I and SC11-pop II.

To identify divergence drivers, we compared core genomic mutations and accessory gene variations between SC11-pop I and SC11-pop II. Unexpectedly, no substantial cgSNP differences emerged (Appendix 3), indicating divergence was not driven by core genome mutations. The SC11 lineage pangenome (n = 235) comprised 3,674 genes, including 2,367 core genes and 1,307 accessory genes, an accessory-to-core gene ratio of 0.55. Concerning gene gain or loss, SC11-pop II specifically acquired 152 genes (forming Recomb1 modular) and showed higher frequency of 13 other functional unknown genes (except endoribonuclease PemK) and lower frequency of 9 genes versus SC11-pop I (p<0.05) (Appendix 1 Table 6), indicating greater exogenous gene acquisition. As was observed in the hierarchical clustering of accessory genes on the pangenomic phylogeny, we identified 3 recombination gene modules (Recomb1–3) that were frequently acquired in SC11 (Figure 3, panel A). Recomb1 was exclusive to SC11-pop II (31/35 strains; 1 strain carries more than half of Recomb1 genes and is recognized as Recomb1 positive, similar to the standard for recognizing positivity in Recomb2 and Recomb3), whereas Recomb2 (68 strains) and Recomb3 (23 strains) occurred in both populations (Figure 3, panel A). The fixation of Recomb1–3 suggested their roles in emergence and adaptation of novel variants in SC11. Of note, Recomb1 contained 11 carbohydrate metabolism genes (Appendix 1 Table 4), which are known factors in E. faecium that contribute to clinical adaption and epidemics of E. faecium (12). The stable 3-year persistence of Recomb1 in SC11-pop II across regions suggests a functional importance in host adaptation and potential virulence.

Recombination surpasses mutation as the primary driver of E. faecium genetic diversity (11), and IS-mediated events occur within days during infection (13). Ten families of IS elements were found in all SC11 isolates (Appendix 1 Figure 5). Enhanced IS transposition was associated with rapid core gene mutation (Appendix 1). Recomb1–3 acquisitions were linked to IS-mediated recombination, primarily involving genes related to DNA transposition, replication, or recombination (Appendix 1 Table 4). Recomb1 contained more ISs than Recomb2 or Recomb3, and IS91 was exclusively acquired by SC11-pop II. High-frequency modular recombination in Recomb1 involved IS91, ISL3, and IS256 (specific to SC11-pop II), whereas IS200/IS605 and IS1182 occurred at lower frequencies (Table; Appendix 1 Figure 5). Recomb2 in SC11-pop I occasionally incorporated IS3 alongside ISL3, IS66, IS982, IS256, or IS30 (Table). Recomb3 exclusively associated with IS3 in all 7 SC11-pop II isolates but was absent in SC11-pop I (Table). No plasmid marker genes co-occurred with Recomb1–3, except 9 Recomb1 genes colocalized with MOBT (plasmid relaxase) on contig AXARS010000069.1 (strain SZYSC_22VRE31), suggesting that plasmids did not directly transmit Recomb1–3.

Identifying SC11’s most recent ancestor is crucial for elucidating its evolutionary mechanism and mitigating emergent threats. We hypothesize a shared ancestry between SC11-root and SC11-outbreak sublineages. Expanded surveillance of outbreak-associated hospitals and retrospective analysis of pre-2021 VREF isolates are needed to trace the origin.

In summary, we showed that increasing VREF prevalence in Shenzhen, China, constitutes part of the ongoing SC11 outbreak, likely originating from Guangzhou. Population structure analysis revealed 2 stable, circulating SC11 subpopulations, emergence of which was driven by IS-mediated recombination. Sustained surveillance of those subpopulations is essential to prevent the emergence of high-risk clones with increased transmissibility and virulence.