Discussion 
It have been demonstrated that CsrS/R is a member of the two-component regulatory systems for regulating the multipe virulence factors of GAS, by using genetically- manipulated GAS mutants [20].
The present study demonstrate that the loss-of-functional mutations in csrS gene which were accumulated in clinically isolated GAS from patients with severe invasive infections, but not with emm-matched non-invasive strains.
The csrS mutations enhanced the expression of scpC and slo, associated with the evasion of PMN functions and in vivo virulence.
Introduction of the intact csrS gene into the severe invasive GAS restored the susceptibility to the killing by PMN and abrogated the activity for inhibition of PMN recruitment and survival, thus, demonstrating an instructional role of the loss-of-functional mutations in csrS gene for the evasion of PMN functions, providing unique pathophysiology of invasive GAS infections.
Previous studies using animal-passaged GAS have shown that mutations in both csrS and csrR gene are important for the invasive phenotype [5], [6], and mutation frequency of csrS and csrR seems to be the same [21].
However, the severe invasive isolates analyzed in this study accumulated mutations in the csrS gene but not in the csrR gene (Figure 5B and data not shown).
Then we further examined whether severe invasive isolates of other than emm49 genotype have the mutation of the csrS and the csrR genes.
The frequency of the mutation in the csrS gene is higher than that in csrR (csrS mutation:csrR mutation=59:19) (manuscript in preparation), suggesting the csrS mutation is more important in comparison with that of csrR in the clinical isolates regardless of emm genotypes.
Furthermore, the expression of some human invasive disease-associated genes [5] including slo was enhanced in the csrS mutant (Figure 5B), but not in the csrR mutant [20].
On the contrary of a dogma that CsrS/R is a definitive member of the two-component regulatory systems, which involve a coordinate pair of proteins known as the sensor kinase and the response regulator [20], CsrS may transmit a signal not only to CsrR but also to other regulators.
This dominant role of CsrS is the first important observation in this study, and its mutation is possibly more important than that of csrR in terms of etiopathogenesis of human severe invasive diseases.
Numbers of studies has pointed out virulent factors to evade host defense using genetically-manipulated GAS and animal models [18], [22], [23], although the significance of each factor to invasive infection is diverse and sometimes controversial, perhaps due to lack of proper non-invasive counterpart.
As examples, SpeB [22] and SLS [23] have been proposed as an invasive infection-associated factor by its cytotoxic effect, however, speB and sagA expression is not enhanced in any csrS-mutated severe invasive GAS isolate used in this study and others [5], [24].
Furthermore, SLS hemolytic activity of invasive GAS is significantly decreased as compared with non-invasive strains (data not shown) and SLS-deletion in invasive GAS did not affect PMN survival at all, excluding the possibility for the role of SLS in PMN necrosis seen in this study.
Extracellular deoxyribonuclease (DNase) is a virulence factor that protects emm1 type GAS against neutrophil killing by degrading the DNA framework of neutrophil extracelluar traps (NETs) [25], [26].
However, we confirmed that addition of DNase in the culture did not alter the level of PI-positive PMN, meaning bright PI staining of PMN is not due to release of NETs from PMN (Figure S1A).
DNase activity of the emm49 severe invasive GAS was lower than that of non-invasive GAS (Figure S1B), possibly due to the difference of emm type.
The expression of DNase as well as the slo and the scpC genes in emm1-genotype strains was enhanced under the csrS mutation [5].
These suggest that DNase may be important but redundant for induction of invasive diseases.
Therefore, the second important observation in the present study is that an essential requirement of csrS mutation for invasive infection is associated with increased expression of ScpC and SLO and in vitro evasion of PMN functions, though we do not exclude the possibility that other CsrS-regulating factors contribute to the escape of invasive GAS from host defense.
SLO and ScpC independently enable GAS to escape from PMN functions; Present data using clinical isolated GAS and a scpC-deletion mutant (Figure 4A and 4B) show that enhanced production of serine proteinase ScpC in virulent GAS is essential to impair PMN migration in vitro by degradation of IL-8, as others partially have demonstrated [17]-[19].
The present study also uncovers that increased activity of SLO from invasive GAS isolates induces rapid and extensive necrosis to human PMN.
SLO is a cholesterol-binding pore-forming hemolysin as well as cytotoxic for other cells [15].
A study has demonstrated SLO from invasive GAS lyse PMN [27], however this effect is likely due to complement activation by SLO [28] or PMN activation [29] but not due to cytotoxity of SLO itself as judged by their flow cytometry profiles which are distinct from ours (Figure 2A).
In the present study, we observed that SLO concentration in a short-time culture with severe invasive GAS did not reach the threshold level to kill PMN by formation of pores (data not shown) and that PMN did not undergo necrosis upon incubation with culture media of severe invasive GAS (Figure 3B), leading to the novel possibility that PMN are probably killed following encounter with invasive GAS in a contact-dependent manner.
PMN-binding GAS may make a small interface containing a high concentration of SLO between bacteria and PMN, which resembles to killing mechanism of killer cells to target cells [30].
Collaboration of SLO with other toxins may be critical to induce PMN necrosis as similarly mechanism has been reported [31], although it remains to be examined whether there exist explore interaction-associated molecules on both host and bacterial membrane is needed.
In contrast to the previous view [18], we observed that both of ScpC and SLO together, but not each of them, mediated sufficient in vivo virulence (Table 1), thus compatible with the notion that plural virulence-associated factors under the regulation of csrS abrogate PMN bactericidal functions and induce invasive diseases in in vivo animal model.
Consistently, the high mortality and histopathological findings which lacks PMN infiltration in mice tissues infected with csrS-mutated GAS (Figure 7) are similar to those seen in clinical invasive GAS infections [32].
Thus, these results suggest that the ability of incompetence for PMN functions by individual GAS strain may determine the induction and clinical outcome of invasive diseases.
Several clinical reports seem to support this hypothesis; Leukocytopenia seen in patients with STSS is more severe than that with non-STSS [33], and invasive GAS-infected patients with leukocytopenia show worse prognosis than those without leukocytopenia [33], [34].
Furthermore, predisposing factors for severe invasive GAS infection [9], such as diabetes mellitus [35], liver cirrhosis [36], and congestive heart failure [37] are known to impair PMN function.
These evidences suggest that the level of PMN function is one of the critical factors to determine the threshold for the onset of invasive GAS infection, which may be the reason for rare outbreaks of invasive GAS infections.
Thus, enhanced expression of virulence factors that could evade PMN function is a key issue at first step to cause invasive bacterial infections.
A further study in which collates clinical with bacterial/immunological data may provide with novel clues for early diagnosis and therapeutics of invasive bacterial infections.
