Rare Heart-Invading Streptococcus Carries a Hemolysis Gene Block Linked to Deadly Endocarditis

A rare bacterium that can quietly colonize the human mouth before turning up in one of the most dangerous infections of the heart is giving up its genetic secrets. In a new study published in Frontiers in Microbiology, researchers in Shanghai sequenced the complete genomes of three Streptococcus sinensis isolates recovered from a single patient with infective endocarditis and compared them against nearly 200 publicly available Streptococcus genomes. Their analysis revealed something striking: the clinical isolates all carried an identical, intact block of twelve genes resembling the cyl cluster that gives group B Streptococcus its blood-destroying, tissue-invading powers. The discovery offers the first detailed genomic picture of why this obscure organism keeps appearing in cases of a disease that kills tens of thousands of people each year.

The clinical story behind the study began with a 26-year-old man admitted to Changhai Hospital with fever, night sweats, a five-kilogram weight loss over two weeks, and leg pain that rest would not relieve. He was febrile at 39 degrees Celsius with a racing heart and laboratory markers of severe inflammation and anemia. Transthoracic echocardiography revealed vegetations on his mitral valve, and all four bottles from a blood culture set grew S. sinensis. Because of a documented penicillin allergy, clinicians initiated vancomycin, but the infection had already done serious damage. Surgeons found large vegetations on both mitral leaflets and a perforation of the anterior leaflet, requiring mechanical valve replacement. Notably, the organism was recovered again from the excised valvular tissue, and the patient later recovered after a 28-day hospitalization.

From that single patient, the team obtained three isolates: two from the same blood culture, designated 418A and 418B, and one from the valvular vegetation and necrotic tissue, designated 433. Whole-genome sequencing on an Illumina NovaSeq 6000 platform reached depths exceeding 100-fold for each sample. Single-nucleotide polymorphism analysis showed that the blood isolates differed from each other by zero SNPs and from the valve isolate by just one SNP. That near-clonal relationship matters epidemiologically: the three isolates represent repeated sampling of one infection rather than three independent observations, a point the authors emphasize throughout their discussion of the findings.

To place these isolates in evolutionary context, the researchers assembled a dataset of 202 Streptococcus genomes, combining their three clinical isolates with 199 public reference genomes spanning 20 named species. Orthology analysis with OrthoFinder identified 7,200 orthogroups, of which only 382 occurred in every genome, while 6,739 appeared in some but not all. The pan-genome accumulation curves kept rising as more genomes were added and never plateaued, confirming that Streptococcus species carry enormous accessory gene pools beyond a conserved backbone. This dynamic genome structure, shaped by gene gain, gene loss, and recombination, is exactly the kind of genomic backdrop in which lineage-specific virulence traits can arise and persist.

The phylogenetic reconstruction also helped settle a long-running taxonomic debate. Early analyses based on 16S rRNA and single genes had left S. sinensis precariously positioned between the Mitis, Anginosus, and Sanguinis groups of streptococci. Using alignments of core single-copy orthologous genes, the team showed that S. sinensis occupies a Mitis group-associated lineage and is most closely related to S. cristatus, followed by S. sanguinis and S. gordonii, both common oral streptococci. This placement is consistent with an oral reservoir for the species, an idea supported by clinical reports of odontogenic disease preceding endocarditis and by molecular detection of S. sinensis in subgingival plaque from patients with periodontitis.

The central discovery concerned the cyl-like region. All three clinical isolates carried an identical 11,341-base-pair block of twelve genes on Scaffold10, comprising eleven cyl-like genes and acpC, arranged as cylX-cylD-cylG-acpC-cylZ-cylA-cylB-cylE-cylF-cylI-cylJ-cylK. In group B Streptococcus, the canonical cyl operon produces granadaene, a hemolytic pigment that damages tissue barriers and allows the bacterium to penetrate the human placenta. Gene-specific BLAST verification showed 100 percent nucleotide identity and full-length coverage for every one of the twelve coding sequences relative to a reference genome, and independent read mapping confirmed 100 percent mapped-base coverage with mean depths ranging from about 210-fold to over 700-fold across the isolates. Pfam domain analysis using HMMER supported enzyme- and transporter-associated assignments for eight of the encoded proteins.

Computational structural modeling added a layer of plausibility to these annotations. Using the ESMFold protein language model, the team predicted that the CylG-like and CylZ-like proteins adopt soluble enzyme-like folds, with high confidence scores of 98.3 and 84.3 respectively on the pLDDT scale, while the CylB-like protein showed multiple predicted transmembrane helices consistent with a membrane permease, echoing the hemolysin transporter of group B Streptococcus. The authors are careful to stress that these models visualize sequence annotations only and do not demonstrate protein expression, enzymatic activity, or virulence. Synteny comparisons showed that the local genomic organization in S. sinensis broadly resembles corresponding regions in S. gallolyticus and partially in S. gordonii, while differing from the canonical cyl cluster of S. agalactiae.

Intriguingly, the cyl-like block is not universal within the species. Across the seven available S. sinensis genomes, the complete twelve-gene profile appeared only in the three near-clonal clinical isolates and one public reference genome, forming a closely related clade on the core-gene phylogeny, while the type strain HKU4 and two other reference genomes lacked the block entirely. This lineage-associated distribution could reflect ancestral inheritance, lineage-specific retention, or unresolved gain-and-loss events, but it stops well short of proving a causal link to endocarditis. With only one patient represented, the finding identifies a candidate region for future study rather than a confirmed virulence determinant of the disease.

Phenotypic experiments provided suggestive but preliminary support. On Columbia sheep-blood agar, all three isolates produced visible hemolytic zones, and in a liquid sheep erythrocyte assay their relative hemolysis signals, normalized to a detergent control, ranged from roughly 20 to 45 percent, with a clinical S. agalactiae strain as a comparator. The authors candidly note the limitations: the assay was run once with three technical tubes, lacked a bacteria-only optical blank and a cyl-negative control strain, and therefore cannot establish reproducible erythrocyte lysis or tie the phenotype to the cyl-like genes. On the antimicrobial resistance front, screening against the CARD database painted a reassuring picture: all three isolates carried only a single marker, Spne_parC_FLO, and remained phenotypically susceptible to levofloxacin and the other tested agents, including penicillin G, consistent with the generally favorable beta-lactam responsiveness reported for this species.

The study also retained classic host-interaction genes, including eno, pavA, and plr/gapA, which in other streptococci mediate plasminogen binding, fibronectin attachment, and invasion, while screening found no qualifying matches for canonical oral adhesins such as gspB, hsa, or the antigen I/II family proteins. The authors caution that such database-dependent screens cannot prove gene absence in a poorly characterized species. What the work ultimately delivers is a roadmap: independent isolates from multiple patients, complete genome assemblies, and functional experiments involving transcriptional analysis, mutagenesis, and infection models are now needed to determine whether this quiet oral colonist’s cyl-like arsenal is what allows it, in a small number of people, to breach the bloodstream and destroy a heart valve.

Subject of Research: Comparative genomics of Streptococcus sinensis isolates from infective endocarditis identifying a cyl-like genomic region

Article Title: Comparative genomics identifies a cyl-like region in Streptococcus sinensis isolates from a patient with infective endocarditis

Article References: Ma, W., Zhang, Y., Huang, X., Wan, Y., Deng, C., Lin, J., Xu, L., & Qin, Q. (2026). Comparative genomics identifies a cyl-like region in Streptococcus sinensis isolates from a patient with infective endocarditis. Frontiers in Microbiology, 17, Article 1990134. https://doi.org/10.3389/fmicb.2026.1990134

Image Credits: AI Generated

DOI: 10.3389/fmicb.2026.1990134

Keywords: Streptococcus sinensis, infective endocarditis, comparative genomics, cyl-like region, viridans group streptococci, virulence genes, hemolysis, phylogenomics, pangenome, antimicrobial resistance, whole-genome sequencing, Comparative