Population Biology of the Human Restricted Pathogen, Streptococcus pyogenes

Abstract

S. pyogenes, also referred to as β-hemolytic group A streptococci, are strictly human pathogens with a global distribution and high prevalence of infection. The organisms are characterized by high levels of genetic recombination, extensive strain diversity, and a narrow habitat. This review highlights many key features of the population genetics and molecular epidemiology of this biologically diverse bacterial species, with special emphasis on ecological subdivisions and tissue-specific infections, strain diversity and population dynamics in communities, selection pressures arising from the specific host immune response and antibiotic exposure, and within-host selection during the course of invasive disease.

The Streptococcus genus contains > 50 named species that inhabit a broad range of hosts, including humans and domesticated animals, where they often colonize as part of the normal flora and/or cause infection. The most clinically significant of the human pathogens are Streptococcus pyogenes, the causative agent of “strep” throat, and S. pneumoniae, a leading cause of bacterial pneumonia. S. pyogenes and S. pneumoniae are exclusive to humans, existing in a carrier state or as pathogens, whereby both natural immunity and therapeutic or preventive interventions (egs., antibiotics, vaccines) have the potential to shape their population biology. The evolution of S. pneumoniae clones was recently described in an excellent review (Henriques-Normark et al., 2008). This article focuses on S. pyogenes, which lie within the pyogenic branch of the 16S rRNA-based phylogenetic tree, along with other streptococcal species displaying a β-hemolytic pattern of growth on blood agar, such as the human and animal pathogen S. agalactiae (Facklam, 2002). S. pyogenes also contain the Lancefield serogroup A carbohydrate on their cell surface, and are often referred to as the large colony-forming β-hemolytic group A streptococci.

1. Overview of key genome features

S. pyogenes contain a single circular chromosome of ~ 1.9 Mb; the complete genome sequence has been determined for > 10 independent isolates (Ferretti et al., 2001, Beres et al., 2002, McShan et al., 2008, Smoot et al., 2002, Banks et al., 2003, Green et al., 2005, Beres & Musser, 2007, Holden et al., 2007, Nakagawa et al., 2003). A striking feature of the S. pyogenes genome is the abundance of exogenous genetic elements, including prophage and integrated conjugative elements (ICE), occupying ~ 10% of the chromosome [reviewed in (Beres & Musser, 2007, Banks et al., 2002)]. That these elements undergo horizontal exchange is strongly supported by their mosaic structure and unequal distribution across strains. Recombination between prophage genes can also lead to large-scale chromosomal inversions (Nakagawa et al., 2003). Genes encoding virulence factors and antibiotic resistance are often found in association with the exogenous genetic elements, however, unlike many bacterial pathogens, S. pyogenes appear to be completely devoid of pathogenicity islands (Schmidt & Hensel, 2004).

Several genome-wide analyses on the evolution of S. pyogenes have been performed. About half of S. agalactiae genes have homologs in both S. pyogenes and S. pneumoniae, with conservation of gene synteny being more pronounced between S. agalactiae and S. pyogenes (Tettelin et al., 2002). A comparison of numerous genomes of S. agalactiae and S. pyogenes reveals that S. pyogenes has a smaller pan-genome and more recombination in its core-genome, perhaps reflecting the narrower habitat diversity of S. pyogenes (Lefebure & Stanhope, 2007). Many individual strains of S. pyogenes appear to have acquired a variety of genes from related streptococcal species, such as S. agalactiae and S. dysgalactiae subspecies equisimilis, via lateral gene transfer (Stalhammar-Carlemalm et al., 1999, Franken et al., 2001, Kalia & Bessen, 2004, Bessen et al., 2005, Beres & Musser, 2007). In fact, up to 50% of the ~ 200 S. pyogenes virulence genes examined were present in S. dysgalactiae ssp. equisimilis isolates evaluated by comparative genomic hybridization (Davies et al., 2007), indicative of a close genetic relationship between S. pyogenes and this largely commensal species.

2. Disease, transmission and ecology

S. pyogenes are responsible for a minimally estimated 616 million cases of throat infection (pharyngitis, tonsillitis) worldwide per year, and 111 million cases of skin infection (primarily non-bullous impetigo) in children of less developed countries (Carapetis et al., 2005). Streptococcal pharyngitis and impetigo are superficial, self-limiting infections that usually cause only a mild illness. A clinically inapparent oropharyngeal infection can also occur, in which the host mounts a significant immune response, yet lacks obvious clinical symptoms of illness. Infections at the throat and skin are self-limiting, and typically resolve within 2 weeks, coincident with the rise of specific host immune defenses.

Most morbidity and mortality due to S. pyogenes arises from invasive and autoimmune disease, although each is somewhat rare in occurrence when compared to the mild infections at the throat and skin. Rheumatic fever is an autoimmune disorder that can have serious cardiac manifestations. Invasive disease is the result of streptococcal infection of deep tissue and can be moderate to quite severe, such as in cases of toxic shock syndrome and necrotizing fasciitis.

S. pyogenes is a free-living organism, however, its ecological niche appears to be quite narrow: Its’ only known biological host is the human, and like many other streptococci, lacks an environmental reservoir of known importance. Thus, direct human-to-human transmission, which mostly occurs via respiratory droplets or skin contact, is critical for its maintenance. Streptococcal organisms tend to reside extracellularly, but can also be found within the mammalian host cells of the respiratory tract (Osterlund et al., 1997, Cleary et al., 1998). In addition to causing clinical infection at the throat and skin, S. pyogenes can persist in a quiescent carrier state; it is not uncommon for throat carriage rates to exceed 15–20% within communities of school-aged children during seasonal peaks (Bisno & Stevens, 2005). The carrier state elicits little or no immune response to streptococcal antigens, and throat carriage can persist for weeks or months (Kaplan, 1980). Secondary transmission of streptococci from an impetigo lesion to the oropharynx is well-documented, and will typically result in a carrier state (Bisno & Stevens, 2005, Ferrieri et al., 1972); carriage at the skin is less well understood, and may be only transient. Importantly, the mucosal epithelium of the oropharynx and the epidermal layer of the skin constitute the primary ecological niches of S. pyogenes. It is at these two tissue sites that the organism most often enters and exits its biological host.

Reproductive success of S. pyogenes at the host epithelium probably follows at least 2 different dynamics. During infection, the generation of new progeny is likely to be high; in an experimental model for impetigo, a > 5,000-fold increase in colony forming units (cfu) is observed (Scaramuzzino et al., 2000, Lizano et al., 2007). An exponential rise in the number of new progeny leads to larger infective doses, which in turn, increases the chance for successful transmission. Yet the window of opportunity for transmission may be narrow due to rapid clearance of the infection by the mounting host immune response. During carriage, the organisms may be in a persistent dormant state (Wood et al., 2005, Leonard et al., 1998), with a low yield of new progeny and a feeble host immune response, but with a much longer time frame for potential transmission events. The physiological state of the organism during its exposure to a new host may also impact transmission success, providing yet another possible factor shaping the distinct dynamic processes for infection versus carriage.

The relative incidence of disease caused by S. pyogenes varies throughout the world, in accordance with both season and locale (Carapetis et al., 1999). In temperate regions, pharyngitis is highly prevalent during the winter months, perhaps as a consequence of indoor crowding; impetigo, although less common, is most often encountered during warm and humid weather conditions. Impetigo caused by S. pyogenes is far more prevalent than pharyngeal infection in many tropical regions. Thus, climate is an important risk factor for the different forms of streptococcal disease, which in turn, creates spatial and temporal distances between the organisms causing throat versus skin infections.

3. Strain typing based on surface structures

Epidemiological markers are useful for investigating outbreaks of disease and providing a reference point for deciphering the genetic organization of a bacterial population. During the 1920’s, Dr. Rebecca Lancefield began work aimed at understanding the basis for protective immunity to S. pyogenes infection. Antiserum raised to extractable surface antigens known as M proteins led to opsonophagocytosis and killing of the strain from which the M protein was derived (Lancefield, 1962); however, antiserum directed to the M protein of one organism often failed to protect against many other isolates. A serological typing scheme arose through the development of antiserum directed to M proteins of different isolates. More than 80 distinct M types were identified, and strong protective immunity is M type-specific.

M proteins form hair-like fibrils extending about 60 nm from the surface of the bacterial cell (Fischetti, 1989). The determinants of serological type lie at the amino-termini, which correspond to the distal fibril tips. The serologically-based M protein typing scheme was replaced about 12 years with a highly correlated nucleotide (nt) sequence-based emm typing scheme (Beall, 2008, Beall et al., 1996). emm type is defined by the 5′ end region of the emm gene, and genes assigned the same emm type have > 92% nt sequence identity. More than 200 emm types are currently listed in the online C.D.C. database (Beall, 2008).

S. pyogenes display heterogeneity in the number and arrangement of 4 basic forms of emm and emm-like genes that are defined by differences in the region encoding the cell wall-spanning domain (Hollingshead et al., 1994, Hollingshead et al., 1993). A strain can have up to 3 distinct emm or emm-like genes, lying in a contiguous stretch. There are 5 recognized arrangements of emm and emm-like genes that are denoted emm patterns. Together, the 5 emm patterns comprise 3 main groups known as pattern A–C, D and E, with emm type residing in either the first (pattern A–C) or middle (patterns D and E) gene. Nearly all emm types examined are restricted to a single emm pattern grouping (McGregor et al., 2004b). So while there may be horizontal movement of the emm gene between strains, any one emm type seems to be largely restricted to the same emm pattern group.

A second serological typing scheme for S. pyogenes is based on the T antigen, originally defined by its resistance to trypsin. There appears to be a far greater number of distinct M serotypes than T serotypes (Johnson et al., 2006). The tee gene of a T6 type strain was initially mapped to the highly recombinatorial FCT region of the S. pyogenes genome, positioned ~ 300 kb from the emm region (Schneewind et al., 1990, Bessen & Kalia, 2002). More recently, T antigens have been characterized as forming elongated pili that function as adhesins (Mora et al., 2005, Kreikemeyer et al., 2005, Manetti et al., 2007, Abbot et al., 2007). A new approach for nt sequence-based typing of the genes encoding pilus structural proteins was recently introduced (Falugi et al., 2008).

Serum opacity factor (SOF) is a protein produced by ~ 50 % of S. pyogenes isolates, both in a surface-bound and secreted form. The amino-terminal region binds apolipoprotein present in serum, leading to its opacification, whereas the carboxy-terminal region functions as an adhesin via binding to fibronectin (Gillen et al., 2008). A serological typing scheme was developed in which SOF type-specific antiserum inhibits serum opacity activity. sof-positive strains and emm pattern E generalists comprise nearly perfect overlapping groups.

S. pyogenes also possess a surface polysaccharide capsule that functions as an important virulence factor. The quantity of capsule produced by S. pyogenes strains can be elevated in association with increased levels of transmission via a respiratory route (Stollerman & Dale, 2008). However, the capsular material is uniformly composed of hyaluronic acid, a poor immunogen. The other streptococcal species causing significant disease in humans - S. pneumoniae and S. agalactiae – differ from S. pyogenes in having a variety of capsular polysaccharide antigenic forms that provide the basis for serological typing as well as targets for protective immunity.

4. Tissue tropisms among S. pyogenes

With serological typing schemes in place, decades of epidemiological studies revealed that there are serotypes of S. pyogenes having a strong tendency to cause throat infection, but not skin infection, and similarly, there are other serotypes often seen with impetigo, but much less so with pharyngitis (Bisno & Stevens, 2005, Wannamaker, 1970). This observation gave rise to the concept of distinct throat and skin types, suggesting that there exists a degree of specialization among strains belonging to this species. Indeed, the emm pattern genotypes have proven to be fairly strong markers for preferred tissue site of infection among S. pyogenes strains.

Several recent reports on population-based surveillance for S. pyogenes associated with pharyngitis or impetigo infections, when taken together, yield emm typing data for > 3,700 isolates collected from all 6 major continents (Bessen et al., 2000, Dicuonzo et al., 2001, Sakota et al., 2006, Shulman, 2004, Smeesters et al., 2006, Tewodros & Kronvall, 2005, Bessen et al., 2008, Alberti et al., 2003, Espinosa et al., 2003, Brandt et al., 2001, McDonald et al., 2007b). Since isolates sharing the same emm type almost always belong to the same emm pattern group (McGregor et al., 2004b), emm pattern can be inferred with a high degree of accuracy, based on knowledge of the emm type (Table 1).

Table 1.

The association of emm pattern with tissue site of infection

Location of population based surveillance	Pharyngitis (or tonsillitis)				Impetigo				Reference for study
	# of isolates	% of isolates that are emm pattern:			# of isolates	% of isolatesthat are emm pattern
		A–C	D	E		A–C	D	E
Australia, tropical	NR				125	13	46	41	Bessen et al., 2000
Rome (Italy)	114	50	1	48	n.d.				Dicuonzo et al., 2001
Germany	216	51	0	49	n.d.				Brandt et al., 2001
Spain	520	32	1	68	n.d.				Alberti et al., 2003
Mexico	282	54	1	44	n.d.				Espinosa et al., 2003
USA	> 1,900	53	1	47	n.d.				Shulman et al., 2004
Ethiopia	57	26	28	35	47	0	32	47	Tewodros et al., 2005
Brazil	52	17	21	62	58	3	57	40	Smeesters et al. 2006
Brussels (Belgium)	163	55	0	45	NR				Smeesters et al. 2006
Nepal	NR				53	19	30	51	Sakota et al., 2006
Australia, tropical	NR				129	13	53	35	McDonald et al., 2007b
average		42	7	50		10	44	43
s.d.		15	11	10		8	12	6
paired t-test, A–C versus D	0.005					0.008

NR, < 10 isolates recovered; n.d., not determined

The combined population analysis reveals a strong trend whereby the emm pattern AC group has a strong predilection to cause infection at the throat, and emm pattern D strains have a strong tendency to cause impetigo. The relationship between emm pattern and infection site is statistically significant when the data for each of the surveys is combined (Table 1); the differences in the distribution of emm pattern A–C versus D strains are highly significant for collections of both pharyngitis (t = 0.005; paired t-test, 2-tailed) and impetigo (t = 0.008) isolates. The emm pattern A–C strains are considered throat specialists, whereas emm pattern D strains are skin specialists. As a group, the pattern E strains are designated generalists, readily causing infections at both tissue sites. Thus, S. pyogenes strains can be divided into 3 clinically and ecologically relevant groups based on emm pattern genotype.

Conceivably, tissue site preferences for infection at the throat versus skin might be explained by tissue-specific colonization factors, such as those encoded by the FCT region genes that give rise to surface pili. There is some correlative data to support this idea (summarized in Table 2). The cpa gene, encoding a pilus accessory protein, is present in nearly all emm pattern D strains, but absent from the majority of emm pattern A–C strains (Kratovac et al., 2007); the well-characterized Cpa protein of an M49 strain displays high affinity-binding to human collagen type I (Kreikemeyer et al., 2005), a form that is abundant in the dermis of human skin. Cpa is essential for streptococcal virulence in an experimental model for impetigo (Lizano et al., 2007).

Table 2.

Summary of genotype and phenotype correlates of emm pattern

Genotype or phenotype	Equivalent genome position (kb) *	emm pattern group
		A–C	D	E
Preferred tissue site of infection	n/a	throat (specialist)	skin (specialist)	throat and skin (generalist)
RofA/Nra, transcriptional regulator	115	nearly all have rofA lineage alleles	nearly all have nra lineage alleles	nearly all have rofA lineage alleles
prtF1, encoding fibronectin-binding protein	117	most positive	most negative	most positive
cpa, encoding collagen-binding protein of pilus	120	most negative	most positive	most positive
Streptokinase, plasminogen activator	1674	all have (sub)cluster 1 or 2a alleles	many have subcluster 2b alleles	nearly all have cluster 1 alleles
PAM, plasminogen-binding M protein	1711	all negative	many positive	all negative
Mga, transcriptional regulator	1714	nearly all have lineage 1 alleles	all have lineage 2 alleles	all have lineage 2 alleles
sof, encoding fibronectin-binding protein	1728	nearly all negative	nearly all negative	all positive

^*Based on the 1.90 Mb genome of M28 strain MGAS6180

The FCT region gene prtF1 encoding a fibronectin-binding protein shows an inverse relationship to that of cpa, being present in most emm pattern A–C strains and absent from most emm pattern D strains (Kratovac et al., 2007). The emm pattern E generalists mostly harbor both cpa and prtF1, perhaps reflecting their tropism for both the skin and throat. However, emm pattern D organisms can readily undergo secondary transmission to the throat suggesting that they can colonize both tissues, even though disease is largely restricted to the skin. Furthermore, in addition to cpa and prtF1, S. pyogenes possess other adhesins, including ≥ 5 other fibronectin-binding proteins (encoded by prtF2, sof, sfbX, fbaA and fbp54), most of which have a differential presence among strains (Kratovac et al., 2007, Jeng et al., 2003, Terao et al., 2001, Courtney et al., 1996, Jaffe et al., 1996), and at least one of which (prtF2) contributes to virulence in experimental skin infection (Lizano et al., 2007). Thus, FCT region-derived colonization factors may facilitate tissue-specific infection, yet additional factors appear to be at play as well. Of the > 10 S. pyogenes isolates whose complete genome sequence has been published, 10 emm types are represented, but none of the strains belong to emm pattern group D; therefore, it is plausible that new tissue-specific colonization factors remain to be discovered.

The M proteins are mosaics of functional domains that bind a variety of host factors present in plasma and tissue. The high-affinity plasminogen-binding domain of M protein (Sanderson-Smith et al., 2007, Berge & Sjöbring, 1993) is restricted to a subset of emm pattern D strains (Svensson et al., 1999) (Table 2). Plasminogen-binding M proteins, also known as PAM, capture host plasminogen on the bacterial cell surface. Bacterial surface-bound plasmin activity, due to a broad spectrum protease, arises via interaction of the PAM-bound form of plasminogen with the streptococcal exoprotein streptokinase (Ringdahl et al., 1998). Streptokinase is encoded by the ska locus, positioned ~ 35 kb from the emm chromosomal region. The ska alleles of S. pyogenes form 2 distinct phylogenetic lineages, and all PAM-positive emm pattern D strains examined share sub-cluster 2b alleles (Kalia & Bessen, 2004) (Table 2); ska is also essential for full virulence in an experimental model for streptococcal impetigo (Svensson et al., 2002). Striking functional differences have been described for streptokinase encoded by alleles of different lineages (McArthur et al., 2008, Nordstrand et al., 2000). Additional phenotypic correlates of the emm pattern D genotype include elevated levels of secreted cysteine protease activity due to the streptococcal exoprotein SpeB, which is also a key virulence factor in experimental impetigo (Svensson et al., 2000).

The mga gene encodes a transcriptional activator of emm and emm-like genes, and lies immediately upstream of the emm region [reviewed in (Hondorp & McIver, 2007)]. All S. pyogenes strains examined have a mga locus, but the mga alleles comprise 2 discrete phylogenetic lineages of ~ 24% nt sequence divergence. Nearly all emm pattern A–C strains harbor a mga-1 lineage allele whereas pattern D and E strains have mga-2 lineage alleles (Bessen et al., 2005) (Table 2). Mga acts as a global regulator of transcription, and it is plausible that mga-1 and mga-2 products differentially regulate genes that act as key determinants of tissue-specific infection.

Two mutually exclusive lineages of alleles, having ~ 33% nt sequence divergence, are present in a second locus (rofA/nra) that encodes transcription regulatory proteins, for which FCT region genes are a target (Bessen et al., 2005, Kreikemeyer et al., 2003, Podbielski et al., 1999, Fogg et al., 1994). Nearly all emm pattern A–C and E strains have rofA, whereas most emm pattern D strains harbor nra (Bessen et al., 2005) (Table 2). The association of rofA and nra with emm pattern group differs from the distribution of mga alleles, in that the pattern E generalists tend to share rofA with pattern A–C strains, but share mga-2 with pattern D strains. Tissue-specific infection by S. pyogenes may require a particular set of virulence factor genes, combined with the differential expression of multiple virulence factors, acting in concert.

To gain further insight on the evolution of S. pyogenes, the gain or loss of genes which correlate with emm pattern genotypes was assessed by maximum parsimony. The genes were assigned a character state (presence or absence, or lineage) at 6 loci of the emm or FCT regions, for > 100 S. pyogenes isolates representing different emm types (Figure 1). The phylogenetic tree shows that the majority of isolates cluster in accordance with their emm pattern group. Most strains belonging to the emm pattern E group of generalists are positioned in between the throat and skin specialists, consistent with the concept that specialists tend to emerge from generalists (Elena & Sanjuan, 2003).

Figure 1.

Maximum parsimony tree based on emm and FCT region genotypes. FCT region genes (rofA/nra, prtF1, cpa, prtF2) and emm region genes (mga, sof) provide the 6 characters for the unrooted tree generated by maximum parsimony. Character states correspond to gene presence/absence or lineage. Taxa (N = 11), representing ≥ 1% of the > 100 S. pyogenes strains evaluated, were used to generate the tree by the branch-and-bound search method. Each circle at the nodes represents an individual strain, and is further defined as belonging to emm pattern group A–C (open), D (filled) or E (hatched). The starting set of 114 strains, from which the 11 taxa are derived, is also depicted in Figure 3.

Not only are the throat and skin specialists physically separated by their ecological niches, but there is geographic partitioning in terms of the prevalence of throat versus skin infection in different parts of the world (Carapetis et al., 1999). The non-overlapping seasonal peak incidence for pharyngitis and impetigo introduces even more distance. The spatial and temporal separation might impose barriers to lateral gene exchange which, in turn, raises the question of whether emm pattern A–C and D strains are at the beginning stages of forming new species.

5. Overview of mechanisms underlying genetic change

The mechanisms underlying genetic change in S. pyogenes presumably include both specialized and generalized transduction via bacteriophage and conjugation via ICE. Replicative plasmids have also been described for some isolates (Woodbury et al., 2008). S. pyogenes lack the full complement of competence genes mediating transformation that are found in S. pneumoniae (Martin et al., 2006), however, some S. pyogenes strains have homologs (sil locus) involved in quorum sensing and DNA transfer (Eran et al., 2007, Hidalgo-Grass et al., 2002).

There is evidence for differences among strains in their potential for genetic change, rooted in their genetic machinery. The presence of clustered regularly interspaced short palindromic repeat (CRISPR) regions within the genomes of some S. pyogenes strains may result in selective blocking of bacterial cell infection by new phage via an RNA interference mechanism (McShan et al., 2008, Barrangou et al., 2007). ICE harboring genes that encode type II restriction-modification cassettes may prevent the uptake of foreign DNA originating from certain sources (Euler et al., 2007). The mutator phenotype, leading to high rates of mutation in some strains, can arise from a defect in DNA mismatch repair that is controlled by prophage excision and re-integration at a chromosomal site upstream from the mutL locus (Scott et al., 2008). Thus, individual strains may differ in the quality and frequency of genetic change.

Based on whole-genome alignments derived from 12 S. pyogenes isolates, genetic flux was found to be dominated by gain and loss of prophage genes. Estimation of the rates of genetic flux indicate that, at least for some strains, prophage integration may have accelerated in recent time (Didelot et al., 2008). Another study yielded similar findings, showing that the within-species branches of a phylogenetic tree constructed from isolates of several Streptococcus species, including S. pyogenes, have a higher rate of gene gain or loss than the branches leading to a species (Marri et al., 2006). Furthermore, a higher proportion of genes that underwent recent lateral transfer show evidence for positive selection, when compared to ancient genes, indicating that some of the newly acquired genes probably serve an adaptive function. Yet in another phylogenomic study taking a slightly different approach, the proportion of S. pyogenes genes exhibiting signs of positive selection were about equal for accessory and core genes (Anisimova et al., 2007). Species-specific genes identified for S. pyogenes, such as those encoding bacteriocins, may have conferred a survival advantage (Marri et al., 2006).

The impact of homologous recombination and point mutation on the population genetic structure of S. pyogenes was estimated using a multilocus sequencing typing (MLST) approach, as described below.

6. Population genetic structure defined by MLST

MLST based on neutral housekeeping genes is routinely used to define clones of S. pyogenes with greater precision (Enright et al., 2001). The MLST data posted at www.mlst.net currently lists 463 sequence types (ST) of S. pyogenes, based on allelic profiles at 7 loci (Aanensen, 2008). Numerous investigators have generously contributed to this rich data set.

Using the MLST data, a population snapshot generated by the eBURST program (Feil et al., 2004) reveals 463 STs (dots) and 71 clonal complexes (CCs), in which the connected STs (lines) are single locus variants (SLVs) sharing 6 of the 7 housekeeping alleles (Figure 2A). A high proportion of STs (43.8%) differ from all others by ≥ 2 alleles (singletons), and the largest CC contains only 3.02% of the total STs. This places the population structure of S. pyogenes at the far end of the spectrum for bacteria, approaching that of the gut pathogen Helicobacter pylori. Organisms falling in this part of the scale are characterized by high levels of genetic diversification, due to both point mutation and homologous recombination (Turner et al., 2007). The population snapshot of S. pyogenes is strikingly different from that of its close genetic relative S. agalactiae (Figure 2B) (Jolley, 2008), which has a high % of STs in large straggly (and possibly unreliable) eBURST groups (Turner et al., 2007).

Figure 2.

Population snapshots generated by eBURST. MLST data from online sources (Aanensen, 2008, Jolley, 2008) is displayed as single eBURST diagrams, by setting the group definition to 0 of 7 shared housekeeping alleles. STs that are single locus variants (SLVs) are connected by a line. (A), S. pyogenes, and (B), S. agalactiae.

Recombination and mutation can be envisioned as two opposing forces in shaping the population genetic structure of a bacterial species (Hanage et al., 2006b). There are several methods for assessing the relative amount of recombination and mutation that contributes to genetic change. Statistical tests were used to measure the congruence between pair wise combinations of the phylogenetic trees corresponding to each of the 7 housekeeping genes used for MLST of S. pyogenes, based on representative strains from each deep branch (Feil et al., 2001). Of the 42 possible pair wise tree comparisons, no significant congruence between trees was observed. The lack of congruence is indicative of relatively high levels of recombination within S. pyogenes.

Using MLST data for an expanded sample set of S. pyogenes isolates and the multilocus infinite alleles model, high rates of recombination are predicted once again (Hanage et al., 2006a). The value for the estimated rate of mutation (θ) is 7.1 for S. pyogenes, which is similar to that calculated for S. pneumoniae (θ= 7.4). The estimated rate of recombination (ρ) is even higher for S. pyogenes (51.2) than for S. pneumoniae (ρ= 29.6). S. pneumoniae is similar to S. pyogenes in its lack of congruency among the topologies of phylogenetic trees based on housekeeping gene sequences (Feil et al., 2001). Both sets of measures are also consistent with population snapshot findings on S. pneumoniae showing a slightly higher % of STs associated with the largest CC (Turner et al., 2007). Together, these data support the notion that S. pyogenes exhibits a slightly higher level of recombination, as compared to S. pneumoniae.

An empirical estimate of the number or alleles changed by recombination compared to mutation, based on SLVs generated by eBURST analysis, yields a ratio of 8.9 to 1 for S. pneumoniae (Feil et al., 2001). Empirical estimates for the recombination to mutation ratio in S. pyogenes have been more difficult to attain because of the few SLVs observed in this species. In a recent analysis of a highly diverse set of nearly 600 S. pyogenes isolates collected from > 25 countries worldwide, representing 156 emm types and 259 STs, only 56 SLVs were detected by eBURST (Bessen et al., 2008). The genetic mechanism of descent was estimated as recombination for ≥ 33 of the 56 SLVs, yielding a recombination to mutation ratio of ≥ 1.4, which is perhaps a bit lower than expected in light of the other types of measurements. However, when stratified according to emm pattern group, the ratio of recombination to mutational events is 0.14, 4.0 and 3.0 for emm patterns A–C, D and E, respectively. Thus, diversification by mutation seems to predominate among the group of throat specialists, whereas skin specialists and generalists tend to diversify by recombination.

Analysis of the population genetic structure of emm pattern-defined groups of S. pyogenes isolates are based on small sample sets and data needs to be interpreted with caution. Nonetheless, emm pattern A–C strains show the highest level of congruence for housekeeping gene tree topologies, indicative of relatively lower levels of recombination, whereas emm pattern D strains displayed the lowest congruency, indicating higher recombination (Kalia et al., 2002). Therefore, the different methodologies point to the same general trend, whereby the throat specialists have the highest tendency to diversify by mutation. Differences among the emm pattern-defined groups in terms of their predominant mechanism for genetic change point to different dynamics in shaping the population structures of the ecologically distinct organisms. Conceivably, this may be a result of higher incidences of co-colonization or co-infection with multiple strains for skin specialists and generalists (Carapetis et al., 1995), increasing the chances for lateral gene transfer, and/or differences in the machinery underlying genetic change.

Ecological barriers can give rise to allopatric speciation. It was of interest to ascertain whether there are signatures of early stages of speciation within the S. pyogenes population. Numerous analyses on housekeeping alleles - including phylogenetic trees of concatenated alleles, splits graphs, fixed nt differences, distribution of shared alleles - all provide strong support for extensive recombination involving housekeeping genes of the 3 emm pattern groupings (Kalia et al., 2002). These findings extend to isolates known to be recovered from the upper respiratory tract versus skin lesions, regardless of emm pattern group (Kalia et al., 2002). Figure 3 shows a phylogenetic tree based on concatenated nt sequences of the 7 housekeeping genes used in MLST, for 114 strains representing 114 STs and 113 distinct emm types. There are only a few branches showing strong bootstrap support, and the nt diversity (π) is very low. Strains assigned to emm pattern groups A–C, D and E are scattered along the length of the tree, and fail to exhibit strong clustering. When depicted as a circular tree, the concatenated sequences display a “bush-like” structure (data not shown), consistent with rapid genetic exchange. A subset of these 114 strains are represented in Figure 1, showing ecological congruence based on several adaptive loci.

Figure 3.

Phylogenetic tree based on concatenated housekeeping gene sequences. The nucleotide sequences of the 7 housekeeping genes used in MLST were concatenated (N = 3,134 sites) and a tree constructed by the minimum evolution method using the Kimura 2-parameter substitution model. Bootstrap analysis (500 replicates) was performed. Included are data from 114 S. pyogenes isolates representing 114 distinct STs and 113 distinct emm types, as described in (Bessen et al., 2005); a subset of these strains are depicted in Figure 1. The emm pattern group of each strain is indicated by bars positioned to the right of each taxon: A–C (open), D (filled) and E (hatched). Nucleotide diversity (π) equals 0.00756.

In summary, there is a lack of clonal congruence among strains corresponding to the 3 emm pattern-defined groups, as demonstrated with housekeeping genes. Thus, despite some niche separation created by distinct epidemiologic trends (temporal and spatial) and innate tissue tropisms, there is no evidence for housekeeping gene sequence divergence between strains of differing emm patterns, or between isolates known to be recovered from the upper respiratory tract versus skin lesions. This finding provides further support for ecological congruence among isolates of each emm pattern group, whereby loci exhibiting a high degree of linkage with emm pattern genotypes (Figure 1, Table 2), and which also map to distal positions on the genome, likely play a direct role in the adaptations leading to throat or skin infection.

7. Strain diversity and dynamics within defined host populations

Surveillance studies conducted on small spatial scales provide evidence for ample migration of S. pyogenes. In a remote tropical island community of aboriginal Australians, periodic surveillance was conducted over 2 years by collecting specimens from the throat and skin lesions (Bessen et al., 2000, McGregor et al., 2004a). Of ~ 500 patient visits involving 224 individual subjects, 16 S. pyogenes were recovered in association with throat carriage and 121 isolates were obtained from skin lesions; there were no cases of pharyngitis. Together, the 137 S. pyogenes isolates represent 32 emm types, 35 STs and 35 unique emm type-ST combinations, with only 2 CCs. Using the emm type-ST combination to define clone, the Simpson’s diversity index (Grundmann et al., 2001), D, is measured as 0.9589 (95% CI, 0.9505, 0.9673), indicative of a highly diverse bacterial population.

The S. pyogenes strains afflicting this remote island community most likely migrated in from elsewhere. At least 90% of the emm types were recovered from places outside of Australia (McGregor et al., 2004a). There is no known reservoir for S. pyogenes aside from the human host. Also, the genomes of the many S. pyogenes strains examined carry genetic material for only a single emm type (and perhaps a few minor variants), as opposed to a broad genetic repertoire for assembling the complete emm type set.

Similarly high levels of S. pyogenes strain diversity have been observed in other remote human populations sampled over relatively short periods. The combined data for 3 remote Australian aboriginal communities on the mainland yielded 350 S. pyogenes isolates from 49 households and 1173 individuals, periodically screened over 1 or 2 years (McDonald et al., 2007b, McDonald et al., 2007a); 43 emm types were found among these isolates, yet long-term throat carriage of the same emm type was uncommon. In far-western Nepal, 120 S. pyogenes isolates were collected from a single screening of 60 children in each of 8 villages, yielding 45 emm types and 51 STs (Sakota et al., 2006). Thus, despite rather limited outside contacts, many distinct strains of S. pyogenes circulate within remote human populations over a short time frame.

The 3 studies cited above were dominated by cases of superficial skin infection, with few or no instances of pharyngitis, although asymptomatic throat carriage was evident. A large multiple site surveillance study of streptococcal pharyngitis in children was recently conducted for several urban and suburban areas in the United States and Canada, yielding nearly 2,000 isolates over 2 years (Shulman, 2004). In year 1 and year 2, the number of distinct emm types recovered was 29 and 31, respectively, with 23 emm types shared between both years. Thus, a highly diverse set of S. pyogenes strains are also responsible for oropharyngeal disease, yielding a Simpson’s diversity index D of 0.8907 (95% CI, 0.8848, 0.8966) when emm type is used to define clone. Interestingly, the diversity value for the multisite pharyngitis study is somewhat less than that for the remote aboriginal Australian island population (adjusted D = 0.9379, when clone is defined by emm type only; 95% CI, 0.9296, 0.9462) (Bessen et al., 2000). This might be explained by differences in the sampling approaches and/or the diversity of throat versus skin specialist strains.

The pharyngitis survey also reveals striking differences between the multiple study sites in terms of the distribution of predominant emm types, although the predominant emm types fluctuate only slightly from year to year (Shulman, 2004). Yet, a comparison of the M type distribution among pharyngitis isolates obtained from pediatric patients in Chicago during the 1960s versus 40 years later shows marked changes in the predominating M types (Shulman et al., 2006). The factors underlying these trends are probably complex, and may include longer term cyclical changes in herd immunity. Age-related changes in emm type distribution are observed in childhood cases of pharyngitis, with older children showing increased infection by uncommon emm types, and a corresponding decrease in common emm types (Jaggi et al., 2005). Conceivably, these shifts are due to the acquisition of protective immunity following exposure to highly prevalent strains earlier in life.

8. emm type as a target of host immune selection

For the many S. pyogenes strains examined, protective immunity is emm type-specific, whereby M type-specific antibodies mediate opsonization and overcome the antiphagocytic property of M protein (Lancefield, 1962, Beachey et al., 1981). Because they elicit a highly protective host immune response, the M type determinants are the targets of a promising S. pyogenes vaccine currently under development (Bisno et al., 2005, Dale et al., 2005, McNeil et al., 2005). However, given the large number of emm types present throughout the world, the practicality of an M type-specific vaccine for the prevention of S. pyogenes disease has been the subject of debate.

The C.D.C. website currently lists ~ 217 emm-types identified among S. pyogenes, represented by ~ 883 partial emm alleles (i.e., subtypes) as defined by the 150 nt encoding the 50 amino-terminal residues of the mature M protein (Beall, 2008). The emm type-determining regions of alleles assigned to the same emm type give high quality alignments, and small indels and nt substitutions are readily apparent (Bessen et al., 2008). However, alignments of the emm type-determining regions derived from multiple emm types are often characterized by numerous gaps and high levels of sequence heterogeneity, and therefore it becomes more difficult to establish the phylogenetic relationships between different emm types (Wertz et al., 2007). Trees based on amino acid sequence alignments that include a large portion of the signal peptide show a clear segregation of emm types derived from SOF-positive versus SOF-negative strains (Sakota et al., 2006), which closely correspond to emm patterns E versus A–C and D groups, respectively.

Immune selection pressures imposed by the host may drive diversification of the emm type-specific region, leading to immune escape. Experimental findings provide evidence for amino acid changes in the M type-specific region that allow for immune escape (Jones et al., 1988, Beres et al., 2006, Eriksson et al., 2001), and such changes may explain the recent emergence of an important M3 type clone. A signature of selection lies in the ratio of nonsynonymous substitutions per nonsynonymous site (Ka) and synonymous substitutions per synonymous site (Ks), whereby nonsynonymous substitutions lead to a change in amino acid residue. The Ka and Ks values were calculated for 105 alignments of partial emm alleles (i.e., subtypes) corresponding to 105 emm types, derived from a global collection of > 500 S. pyogenes isolates (Bessen et al., 2008). The ratio of the mean Ka to mean Ks value for the 105 alignments was 1.96, indicative of positive diversifying selection acting on the type-specific region. For the emm pattern A–C group of throat specialists, this ratio value was ~ 3- to 4-fold higher than for skin specialists and generalists, suggestive of stronger immune selection pressures acting on the M type-specific region in throat strains.

Despite the large number of distinct emm types and emm allele subtypes that exist, there also appear to be functional constraints which put the brakes on genetic diversification. The M type-specific regions of most S. pyogenes strains bind the complement regulator C4b-binding protein (C4BP). C4BP binding is achieved in the absence of a shared amino acid sequence motif and even though substitutions can introduce antigenic change without altering C4BP binding activity (Persson et al., 2006), it seems likely that there are functional constraints on sequence variation. Of probable relevance is the finding that most isolates lacking C4BP binding activity have emm types characteristic of pattern A–C strains (egs., M types 1, 3, 5, 6, 12, 19, 24, 26, 30, 39) (Persson et al., 2006, McGregor et al., 2004b). When combined with the Ka to Ks ratio data (Bessen et al., 2008), the findings are suggestive of higher levels of purifying (negative) selection on pattern D and E emm types in order to preserve C4BP binding activity.

Immune escape might also arise following a change in emm type mediated by recombinational replacement of all or part of the emm gene. Among a genetically diverse set of nearly 600 S. pyogenes isolates, a small proportion of the STs examined (5.4%) were found in association with a substantially larger fraction (> 20%) of the total emm types evaluated; nearly all of the emm-variable STs were restricted to emm pattern groups A–C and D (Bessen et al., 2008). There are also numerous instances of the same emm type (~ 50%) associated with distant STs differing at ≥ 5 of the 7 housekeeping alleles. Horizontal transfer of an emm type to a distant ST was observed primarily among the emm pattern D and E strains. Thus, relationships between emm type and genetic background differ among the 3 host tissue-related groups; any differences in selection pressures and mechanisms for genetic change may have important implications for vaccine design. This analysis also shows that emm type is a reasonably good marker for ST or CC among the throat specialist strains, but a rather poor marker for clone among skin specialists and generalists.

A given M or emm type can be found in association with numerous SOF types (or sof alleles) and T types, as summarized in an analysis of > 40,000 isolates collected worldwide over a 50 year period (Johnson et al., 2006). These findings provide further evidence that emm type undergoes horizontal exchange with other strains. There are numerous examples of emm types associated with > 1 SOF or sof type, suggestive of independent lateral transfer events involving the sof locus, which is positioned ~ 16 kb upstream of emm. A 450 nt region at the 5′ end of the sof gene is used to define the partial sof allele and it strongly correlates with SOF serological types (Beall et al., 2000). Attempts to generate multiple sequence alignments of the 5’ end region of sof alleles has been thwarted by extensive sequence heterogeneity and evidence for past recombination that likely involved short tandem duplications and inverted repeats (Wertz et al., 2007).

The number of S. pyogenes “strains” or “clones” – as defined by unique combinations of emm type, sof type, T or pilus gene type and ST – remains to be established. In addition to emm, sof and tee (i.e., FCT region) genes, there may be additional immunodominant surface antigens that shape the population structure of this species.

9. Antibiotic resistance

β-lactams such as penicillin are the drug of choice for treating upper respiratory tract infections caused by S. pyogenes. Fortunately, resistance to β-lactams has not emerged in S. pyogenes, perhaps due to the organisms’ failure to acquire plasmids harboring a β-lactamase gene, or a high biological cost is incurred when penicillin-binding proteins involved in cell wall synthesis undergo mutations that lower their affinity for penicillin (Horn et al., 1998). However, S. pyogenes have evolved resistance to other antibiotics, which can adversely affect the clinical course of disease and shape the population biology of the species by causing shifts in fitness when placed in an antibiotic infused environment.

Macrolides and lincosamides are the primary treatment for GAS infections in patients with β-lactam hypersensitivity or chronic, recurrent pharyngitis due to prior treatment failure; clindamycin (a lincosamide) is the first choice drug for patients with life-threatening soft tissue infections, such as necrotizing fasciitis, because it halts exotoxin production. Studies in Japan, Finland and elsewhere show a strong correlation between macrolide consumption and resistance in S. pyogenes (Freeman & Shulman, 2002, Fujita et al., 1994, Garcia-Rey et al., 2002, Perez-Trallero et al., 2001, Seppala et al., 1997). In a recent 3-year longitudinal surveillance study in Pittsburgh, 100% of S. pyogenes isolates recovered from children were macrolide-sensitive until the third year of study, when macrolide resistance shot up to 48% of all isolates; all resistant isolates were M6 (Martin et al., 2002). Thus, a macrolide resistant clone can have a significant survival advantage under certain conditions.

At least 3 genes confer resistance to macrolides in S. pyogenes: erm(A) and erm(B), leading to ribosomal modification, and mef(A), which promotes drug efflux. In a survey of 212 macrolide-resistant isolates obtained from throughout the world, at least 49 independent acquisitions of macrolide resistance were estimated, based on unique combinations of emm type, ST and the resistance gene type (Robinson et al., 2006). Numerous genetic elements harboring resistance genes have been defined, and include plasmids, ICEs and non-conjugative elements containing features of prophage, plasmids and/or transposons, some of which are shared with related species such as S. pneumoniae (Varaldo et al., 2008, Banks et al., 2003). Some elements carry determinants for resistance to both macrolides and tetracyclines, the latter often being conferred by tet(M) or tet(O). Resistance to tetracyclines is estimated to have been acquired via ≥ 80 independent horizontal gene transfer events (Ayer et al., 2007). Considering the higher global prevalence of tetracycline resistant S. pyogenes, it is plausible that tetracycline usage drives the acquisition of macrolide resistance to some extent, via genetic elements harboring both resistance genes.

Some strains of S. pyogenes (eg., M6 isolates) are resistant to fluoroquinolones; this phenotype can be found in isolates recovered prior to the development of these synthetic antibiotics (Orscheln et al., 2005). Non-therapeutic antibiotics to which at least some S. pyogenes strains are susceptible include bacteriocins (lantibiotics), produced by S. pyogenes or by other bacterial species that are part of the normal flora of the oral cavity (Upton et al., 2001, Wescombe et al., 2006, Walls et al., 2003). Operons encoding lantibiotic synthesis genes are listed among the prominent species-specific genes of S. pyogenes (Marri et al., 2006). These natural compounds have the potential to profoundly shape the ecology of a microbial community.

10. Population biology considerations in invasive disease

One of the more serious consequences of S. pyogenes infection is invasive disease, wherein the organism gains access to normally sterile tissue. A transient bacteremia probably occurs in the ~ 50% of patients who develop invasive infections without an obvious portal of entry (Bisno & Stevens, 2005). Host factors that modulate the immune response are key determinants of the severity of invasive streptococcal disease (Abdeltawab et al., 2008). Several clones which are highly correlated with invasive disease cases have been identified. However, there is strong evidence for strain prevalence in the community, rather than innate virulence potential, as a major factor in the tight association between certain strains, as defined by emm type, and invasive disease (Rogers et al., 2007, Shulman, 2004, Cockerill et al., 1997).

The M1T1 clone is among the most significant of S. pyogenes strains, displaying both a high degree of lethality due to invasive disease and a widespread prevalence over the past 25 years (Aziz & Kotb, 2008). Injection of a pharyngeal-derived M1 isolate into mice led to the recovery of organisms from deeper tissue having a mutation in genes encoding the 2-component signal transduction system CovRS, resulting in an altered transcriptome (Sumby et al., 2006). Thus, genetic change among one or a few of the infecting organisms may be a prerequisite for gaining access to and/or surviving in deep tissue. The findings would appear to imply that bacteria must reproduce at the upper respiratory tract in sufficient numbers in order to increase the probability of generating mutants having increased fitness for invasive disease; however, the pharyngeal infection that precedes invasive disease is typically asymptomatic. Perhaps an unusual ability of M1 strains to induce clinically inapparent infection at the throat is an evolved trait that confers a survival advantage to the organism because the patient fails to seek antibiotic treatment.

Unique genetic features of the hypervirulent M1T1 clone also include the presence of prophage carrying genes for the pyrogenic exotoxin SpeA and SdaI nuclease; these genes are absent from the M1 strains which have been in circulation decades before emergence of the hypervirulent M1T1 clone (Aziz et al., 2005, Aziz et al., 2004). SpeA is a superantigen which recognizes subsets of human T lymphocytes in a manner that eventually leads to an outpouring of cytokines; the relative lack of speA in earlier S. pyogenes isolates may have resulted in low levels of neutralizing antibody to SpeA, thereby contributing to the increased incidence of severe invasive disease in recent years (Aziz & Kotb, 2008). SdaI is degraded by the SpeB cysteine protease whose expression, in turn, is controlled by CovRS. In the absence of SpeB, the DNA degrading activity of SdaI is able to facilitate the escape of bacteria from neutrophil extracellular traps and in doing so, SdaI may act as a selective force that promotes the emergence and invasive spread of covRS mutants (Walker et al., 2007). This important new concept - that selection pressure exerted by the innate immune response of the host generates hypervirulent genetic variants - may also be broadly applicable to invasive disease caused by other S. pyogenes strains, as well as by other bacterial pathogens.

One aspect of the epithelium-to-invasive genetic transition that remains to be more firmly established is the epidemiological association between the pharyngeal genotype (or pharyngeal transcriptome profile) and recovery from the oropharynx, and the invasive genotype (or invasive transcriptome profile) and isolation from deep tissue. To what extent is the pharyngeal genotype recovered from deep tissue, or the invasive genotype recovered from the epithelium? Is the invasive genotype essential for deep tissue infection having a clear portal of entry? Is the invasive genotype capable of causing primary infection at the epithelial surface and transmitting to new hosts? This information should provide added insights into the epidemiology and population biology of severe S. pyogenes disease.

11. Future directions

Despite our extensive knowledge, there remain many unanswered questions on the population biology of S. pyogenes. Although identification of new emm types has slowed somewhat in recent years, it is difficult to predict the total number of extant emm types; some parts of the world, notably much of Africa and portions of Asia and South America, have yet to be intensively sampled. Because sequence alignment of many emm types is generally poor, one needs to wonder whether there are intermediate genetic forms in circulation (but not yet sampled), or if they no longer exist, why did they become extinct, whether by random genetic drift or selection. How best to define strains of S. pyogenes also remains to be established; “strain” can be defined in epidemiologic terms by the loci that affect its transmission, rather than its entire genetic constitution (Gupta et al., 1996). The type-specific determinants of M protein satisfy this criteria because of the role played by M type as a target of protective immunity, yet is remains unclear as to which additional factors affect transmission (egs., T antigens, SOF). The number of STs as defined by MLST is bound to grow with more intensive sampling and as a consequence, perhaps measures of the relative rates of recombination to mutation will become more accurate.

Many of the dynamic qualities of S. pyogenes population biology can only be roughly estimated at this time. Little is known about strain migration, beyond that it appears to occur at a high rate. Herd immunity undoubtedly influences the transmission success of individual strains by establishing the number of susceptible hosts. The reason for why there exist > 200 emm types in nature is probably explained in part by strain migration, herd immunity, and the size and density of the human host population. Human genetic loci that confer susceptibility to invasive disease and autoimmune sequelae are recognized (Abdeltawab et al., 2008, Guilherme & Kalil, 2007), but less is know about the potential role of host genetics in the ability to cause mild infections at the throat and skin.

Defining a collection of bacteria as a species can be challenging because the boundaries are often fuzzy for highly recombinogenic organisms (Hanage et al., 2005). The complex microbial communities in which S. pyogenes often resides - at the human throat and skin – typically include other streptococcal species and gram positive organisms whose gene sequences and genetic machinery are occasionally compatible. Gaining a more complete understanding of the co-inhabitants may, in turn, sharpen our understanding of S. pyogenes by placing it in a broader context.