Introduction
The group of Mitis streptococci [
1,
2] encompasses species that significantly differ in their pathogenic potential. The best-known representative of the group,
Streptococcus pneumoniae (pneumococcus), is responsible for a high burden of respiratory tract and invasive infections, especially in children and the elderly [
3]. However, species such as
Streptococcus pseudopneumoniae,
Streptococcus mitis,
Streptococcus oralis and others are typically non-pathogenic residents of the human nasopharynx, although invasive infections caused by these bacteria are also occasionally observed [
2,
4]. The conventional microbiological identification of
S. pneumoniae in clinical practice relies on two features: (1) susceptibility to optochin; (2) and solubility in sodium deoxycholate (bile), together with a specific morphology of colonies producing alpha-hemolysis on blood agar [
5]. Reactivity with specific antibodies recognizing polysaccharide capsule (latex agglutination tests and the quellung reaction, i.e. capsular swelling) is another important feature of pneumococci; however, isolates of other closely related species occasionally also demonstrate these properties [
6‐
11]. To complicate species identification issues, a small number of ‘true’ pneumococcal isolates present as optochin resistant and/or poorly soluble in bile [
12‐
14]. Moreover, so-called rough pneumococci do not produce polysaccharide capsule, due to either mutations in the
cps locus or the complete lack of this locus in certain lineages [
15]. Misidentification may not only delay the proper treatment of a patient, but it also influences the correct estimation of the burden of disease caused by pneumococci and other viridians streptococci. In addition, as particular species in the Mitis group differ in the prevalence of resistance to antimicrobials, misidentification results in biased reporting of susceptibility levels in
S. pneumoniae [
11,
13,
16].
With the development of molecular techniques, DNA-based methods have been proposed to improve the identification of Mitis streptococci, mostly focusing on confirming or excluding identification of an isolate as
S. pneumoniae. Initially, certain targets such as genes encoding pneumolysin (
ply), autolysin (
lytA), pneumococcal surface antigen A (
psaA), conserved genes in the
cps locus and spn9802 and spn9828 loci of unknown function were proposed as specific for
S. pneumoniae [
17‐
21]. However, their specificity was later questioned due to the presence of counterparts of some of these ‘pneumococcal’ genes in other Mitis streptococci [
22,
23]. More specific approaches, which employed PCR-RFLP of
lytA and the 16S rRNA genes [
24,
25] and partial sequencing of
sodA and
rpoB [
26,
27], proved to be more reliable for the purpose of identification. The correct selection of appropriate target gene(s) is important not only for the correct identification of isolates but also for culture-free detection of pneumococci in clinical materials, which often relies on PCR-based methods [
28,
29]. Multilocus sequence typing (MLST), based on sequencing of seven loci encoding housekeeping genes of
S. pneumoniae and identification of alleles and sequence types (STs) from allelic profiles with the web-accessible database (
https://pubmlst.org/spneumoniae/), is considered to provide an unambiguous identification of an isolate as pneumococcus [
8]. Multilocus sequence analysis (MLSA), based on the same principle but using different target loci, allows identification of species within the Mitis and other viridians streptococci [
30]. These two approaches, however, are available only for specialized laboratories, both in terms of equipment and software as well as adequately trained personnel. The recent advances of whole-genome sequencing (WGS) technologies and increasing availability of WGS for laboratories have opened new possibilities also for identification purposes. In particular, ribosomal MLST (rMLST) that indexes variation of the 53 genes encoding bacterial ribosomal proteins has been proposed as a universal identification tool [
31].
The National Reference Center for Bacterial Meningitis (NRCBM), Poland, has performed a systematic, country-wide, voluntary-based surveillance of invasive and respiratory tract infections caused by
S. pneumoniae in Poland since 1997 (
http://koroun.edu.pl/) and possesses an archival collection of pneumococcal isolates starting from the early 1990s. During its activity, the NRCBM occasionally received isolates identified as
S. pneumoniae in clinical laboratories, which at the NRCBM were classified as other Mitis streptococci, usually negative in serotyping and/or presenting MLST profiles composed of new alleles, divergent from those characteristic for pneumococci. The aim of our study was to investigate these isolates using a genomic approach in order to better understand their mutual genetic relationships and position within the Mitis group, especially in the relation to
S. pneumoniae.
Discussion
Correct species identification and understanding of the phylogenetic relationships within the Mitis group of streptococci still poses a challenge, despite several approaches addressing this issue over the years. This problem is associated with unusual mitis strains that are misidentified as
S. pneumoniae (here proposed to be named misID streptococci). To our knowledge, this is the first attempt to characterize misID streptococci using genomic approaches at such scale. The analysed collection has certain features that make it especially interesting for the conducted analyses. Most isolates (56, i.e. 89%) were derived from clinically relevant materials, such as BAL, sputum and in a single-case blood, and they were collected over a relatively long time and localisation span. In this manner, risk of a potential bias due to repeated isolations was reduced, and high diversity of studied material was assured. All misID streptococci presented at least one and typically both phenotypic features used for differentiation of
S. pneumoniae from other streptococci in microbiological laboratories, such as optochin susceptibility and bile solubility. While some strains of pneumococci are known to be optochin-resistant [
12,
14], bile solubility is considered a principal characteristic of
S. pneumoniae, and such observation in our study and other studies [
48,
49] underlines the difficulty posed by such isolates for correct identification.
Nonsusceptibility to antimicrobials of main classes was very common in the analysed group and exceeded levels observed for
S. pneumoniae in Poland ([
50‐
53] and unpublished NRCBM data). Streptococci from the Mitis group are considered a reservoir and potential source of resistance genes for
S. pneumoniae as indicated for the chromosomal
pbp2b and
pbp2x genes [
54‐
56] and
parC [
57]. Also, the acquired resistance genes, such as
erm(B),
mef(A) and
tet(M), are the same as major erythromycin and tetracycline resistance determinants in
S. pneumoniae [
58‐
60].
Several gene targets have been proposed as the basis for
S. pneumoniae identification and detection in clinical material. In our collection, all misID streptococci harboured a 6-bp deletion in the
lytA gene [
24], indicating a very good performance of this test. However, both
S. pseudopneumoniae with pneumococcal
lytA and
S. pneumoniae with
lytA characteristic for Mitis streptococci have been observed [
61]. Cytosine at nucleotide position 203 in the 16S rRNA genes is considered specific for the vast majority of pneumococci as it is replaced by adenosine in all other Mitis streptococci [
25]. This test reliably distinguished misID streptococci in our study, with the exception of two isolates with mixed bases at this position. Several bacterial species carry more than one copy of the rRNA operon, and heterogeneity of copies of the 16S rRNA gene was observed earlier in
S. oralis [
62] and other species [
63]. Among other proposed targets,
psaA and
ply were very common among the misID streptococci, and this observation was also made by others [
22,
23,
64]. The spn9808 and spn9828 loci occurred ubiquitously among misID representing
S. pseudopneumoniae but were absent among
S. mitis-related isolates and thus could indeed exclude them as pneumococci.
Multilocus sequence-based approaches such as MLST, MLSA and rMLST were also evaluated as tools for distinguishing pneumococci and other Mitis streptococci. MLST following the
S. pneumoniae scheme has been proposed as a method to reliably include or exclude isolates as pneumococcus [
8]; however, in the current study, several isolates, especially among
S. pseudopneumoniae, carried alleles’ characteristic for
S. pneumoniae, up to a complete identification of two isolates as ST11884. Importantly, however, the single isolate NTP 138 representing this ST in the
S. pneumoniae MLST database appears to be
S. pneudopneumoniae in the core-genome analysis (Fig.
2) and as such should be removed from the database. In phylogenetic networks and trees, both MLSA and rMLST clearly separated the misID isolates from
S. pneumoniae. The misID isolates showed the presence of several alleles from the rMLST scheme found also in pneumococci. This is different from the
Neisseria genus, where some species shared some of rMLST alleles but not MLST alleles, and it was hypothesised that while the rMLST genes undergo recombination, metabolic genes from the MLST scheme evolve to specialize to particular niches [
65]. The presence of a shared pool of both rMLST and MLST alleles in the misID streptococci and
S. pneumoniae is consistent with a similarity of niches’ characteristic for both these groups. It may also indicate a relatively frequent horizontal transfer of genes between the misID streptococci and pneumococci [
22] due to a natural competence common in the Mitis group [
66].
The core-genome analysis, applied to investigate relatedness of misID streptococci to
S. mitis,
S. pseudopneumoniae, the 596553 strain and pneumococci, is considered the most reliable approach for such purposes for Mitis streptococci [
67,
68]. The misID streptococci did not form a single cluster in the core-genome analysis but they were associated with
S. pseudopneumoniae and with several branches of
S. mitis. A similar grouping was also observed in a study on streptococcal isolates from respiratory tract and invasive infections, first considered atypical pneumococci [
64]. In this study, with the use of MLSA, 61
S. pseudopneumoniae and 13
S. mitis were identified while 24 isolates could not be classified due to incomplete MLSA profiles. All these isolates, although collected in a single country (Spain), showed a remarkable diversity, similar to our observations. The core-genome analysis performed in our study revealed a clustering of the majority of misID streptococci associated with
S. mitis into a separate group together with the 596553 strain. This strain was proposed to represent a novel species of streptococcus, based on a lack of clustering with the B6 strain of
S. mitis, the IS7493 strain of
S. pseudopneumoniae and the SPN032672 of
S. pneumoniae genomes in single-nucleotide polymorphism (SNP) analysis and on 81% similarity to the most closely related species,
S. pseudopneumoniae, revealed by protein-by-protein analysis [
36]. The fact that we observed several epidemiologically independent isolates, clustering with the 596553 strain, further supports the idea that indeed this group might represent a novel species.
Two contrasting hypotheses were proposed concerning the evolution with the Mitis group. According to one, the common ancestor of Mitis streptococci was most similar to the current
S. pneumoniae, and other representatives of the group adapted to a more commensal lifestyle by the loss of certain virulence-associated traits, resulting in a genome reduction [
62]. Such reduction was not, however, apparent in our study. In contrast, misID genomes tended to be slightly larger (on average 2,152,263 ± 27,211 kb in comparison with 2,110,084 ± 29,809 kb observed for reference strains of
S. pneumoniae used in this study; CI = 99%). The other hypothesis assumes that the
S. pneumoniae species is relatively young and evolves due to its genome plasticity and acquisition of adaptive traits [
68]; this hypothesis is in agreement with short branches and relative compactness of
S. pneumoniae cluster in comparison with other Mitis streptococci. Such structure of the core-genome-based AML tree was observed here. It is important to note that misID streptococci did not show any clustering close to
S. pneumoniae that might suggest their recent diversification from this species. However, features such as optochin susceptibility and bile solubility might have indeed been characteristic for a common ancestor of Mitis streptococci. While these properties have been lost by most of the members of this group, they have been preserved in a few lineages, in particular in
S. pneumoniae and a few others, such as misID streptococci, which nowadays cause identification problems in clinical laboratories. It appears that the diversity of such organisms remains in a significant part unexplored, and more data are necessary to fully understand the relationships within this very particular group of bacteria.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.