During January 1, 2015–November 1, 2018, a total of 5,480 ABCs GAS isolates were subjected to WGS. We identified 35 atypical gacI-positive isolates; each yielded 1 of the M protein gene (emm) subtypes stG245.0, stG485.0, or stG652.0 commonly associated with SE (4–6). These 35 isolates lacked multilocus sequence types (MLSTs) inclusive of known S. pyogenes allelic designations. Lancefield grouping (7) and MLST (https://pubmlst.org/sdysgalactiae) (6) revealed the 35 isolates were serologically group A and MLST sequence type (ST) 128 (GAS/ST128/SE). We received 13 additional SE isolates recovered through ABCs GAS surveillance during this period that were found to be non–group A isolates (9 group G, 2 group C, and 2 group L) with MLSTs unrelated to ST128 (Figure 1). According to our normal protocol, these 13 non–group A SE isolates and 2 group G S. canis isolates that we also received were removed from the ABCs GAS database.

The Lancefield group A carbohydrate consists of a polyrhamnose chain with an immunodominant N-acetylglucosamine side chain (9) that functions in GAS pathogenesis (1). The group C carbohydrate also has a polyrhamnose backbone; however, its immunodominant side chain is the dissaccharide N-acetylglucosaminosyl-N-acetylglucosamine (9). Genomic comparison of the 12 gene group A carbohydrate synthetic cluster gacA-L (1) from S. pyogenes with the corresponding regions of the 35 GAS/ST128/SE revealed an upstream crossover point within the S. pyogenes gacE ABC transporter gene and a downstream crossover point within ebsA (Figure 2). The ancestral recipient SE strain was implicated as group C S. equisimilis (GCS/SE) by virtue of the near-identical sequence of the 1,363-bp sequence within GAS/ST128/SE encompassing sections of gacD and gacE homologs (designated as gccD and gccE) (Figure 2) with GCS and the marked divergence of this 1,363-bp sequence from group G SE (data not shown). This sequence is immediately adjacent to the upstream crossover point shown between S. pyogenes and GCS/SE (SP-5005 and SE-7136; Figure 2). We also found these same crossover points within the group carbohydrate gene cluster of the available genomic sequence from the previously described invasive GAS/SE strain AC-2713 recovered in 1999 (3). Subsequent genomic analysis revealed AC-2713 to be ST128 and emm type stG485.0. Phylogenetic analysis revealed that AC-2713 differed by 126 single-nucleotide polymorphisms from a pair of genetically indistinguishable GAS/ST128/SE recovered within the East Bay area of San Francisco, California, USA (Figure 1). These 2 isolates were from recurrent invasive GAS infections within the same patient that occurred 1.5 months apart.

Comparison of the S. pyogenes gacA-L cluster with the corresponding gcc loci from group C SE strains (SE-7136; Figure 2) revealed that GCS/SE genes shared homology with all 12 gacA-L genes (56%–89% sequence identity). The weakest conservation was observed between the gac/gccIJK genes (56%–69% identity), consistent with the requirement of gacIJK for the group A immunodominant N-acetylglucosamine side chain but not for synthesis of the polyrhamnose core (1). Two additional genes, designated gccM (glycotransferase gene) and gccN (UDP-monosaccharide epimerase gene), were evident within the gcc gene cluster. In the ancestral recombination event, an 11,527-bp GAS (S. pyogenes) chromosomal segment composed of the gacE 3′ portion, along with the gacF-L genes and a 5′ portion of ebsA, replaced the corresponding 13,813 bp of a GCS/ST128/SE strain, resulting in the recombinant GAS/ST128/SE lineage (Figure 2). This fragment encompasses the intact 7-gene gacF–gacL segment; each gene shared 99.4%–99.7% sequence identity with counterparts in S. pyogenes. The evident functionality of the hybrid gac/gcc cluster within the GAS/ST128/SE progeny lineage is consistent with identical roles of the first 3 genes of the cluster (gac/gccA–C) in the biosynthesis of the polyrhamnose core (1) that is present within the groups A, C, and G carbohydrates (9). Each of these 3 genes are also required for S. pyogenes viability (1).

The occurrence of multiple emm types within the same MLST is common in SE (5,6) and differs from emm/MLST associations within S. pyogenes, where an MLST is nearly always definitive of a single emm type (2,10). The presence of 3 different emm types and 8 macrolide-resistant isolates within GAS/ST128/SE (Figure 1) is indicative of a long-standing successful lineage. A single isolate of this lineage was positive for the exotoxin gene speC (Figure 1) that was carried on a prophage highly similar to a previously described speC-positive S. pyogenes strain (8). The relative genomic positions of the prophages are exactly conserved between the 2 species, inserted within the pepD gene in the genomic region that lies between the bacterial cell division genes ftsE/ftsX and the ribosomal protein gene rpsL31B (Figure 1). The number of single-nucleotide polymorphism differences between individual GAS/ST128/SE core genomes ranged from 0 to 613 (Figure 1). The GAS/ST128/SE strain AC-2713 recovered 20 years ago (3) is also indicative of a long-established lineage.

The 34 GAS/ST128/SE isolates for which information was available (32 from blood, 1 from a joint, and 1 from a surgical wound) recovered in ABCs since January 1, 2015, were recovered from older adults (age range 22–93 years; mean age 63 years) from 8 ABCs sites; most (85%) patients were men. Most patients had underlying medical conditions (data not shown), including 16 with diabetes, 15 with cellulitis (including 1 who had necrotizing fasciitis), 8 with pneumonia, and 6 with septic shock. One patient with bacteremia died.

ABCs identifies invasive infections caused by GAS without identification of isolates to the species level. Since 2015, when we implemented WGS as our primary platform for GAS characterization, we have identified rarely occurring non–S. pyogenes isolates through our bioinformatics pipeline automated MLST function rather than previously employed phenotypic testing. Of ≈16,000 GAS isolates recovered from ABCs during 1994–2014, only 11 had emm types characteristic of SE. All 11 were collected during 2011–2014 and were of the 3 emm types found among the 35 GAS/ST128/SE isolates from this study. Genomic analysis verified the GAS/ST128/SE lineage of these 11 older isolates (data not shown). Finding 35 additional invasive isolates of this lineage recovered during January 1, 2015–November 1, 2018, through ABCs suggests a level of expansion attributable to strain adaptation and fitness or to a more susceptible population. Attempts to identify circulating ST128/SE strains of the original group C have been unsuccessful, including an examination of a population-based sampling of SE (5).

Because group A SE is suspected to be rare, these findings raise the question of whether invasive disease attributable to SE of groups C, G, and L is also increasing. A 2-year population-based study of β-hemolytic streptococcal disease attributable to Lancefield groups other than A and B within 2 ABCs sites during 2002–2004 revealed that 80% of such isolates were SE (11), with clinical manifestations and targeted susceptible populations similar to S. pyogenes. Incidence of invasive disease attributable to non–group A SE during this period was estimated at 2.5 cases/100,000 population, similar to the incidence of GAS infections (2.89 cases/100,000 population) in these same 2 sites. The incidence of overall invasive GAS disease in the United States has also markedly increased during recent years, from 3.4 cases/100,000 population in 2012 to 7.2 cases/100,000 population in 2017 (https://www.cdc.gov/abcs/reports-findings/survreports/gas17.html).

Additional information about emergent invasive group A Streptococcus dysgalactiae subsp. equisimilis, United States, 2015–2018.