Contemporary Pharyngeal and Invasive emm1 and Invasive emm12 Group A Streptococcus Isolates Exhibit Similar In Vivo Selection for CovRS Mutants in Mice

Group A Streptococcus (GAS) causes diverse infections ranging from common pharyngitis to rare severe invasive infections. Invasive GAS isolates can have natural mutations in the virulence regulator CovRS, which result in enhanced expression of multiple virulence genes, suppressed the expression of the protease SpeB, and increased virulence. It is believed that CovRS mutations arise during human infections with GAS carrying wild-type CovRS and are not transmissible. CovRS mutants of invasive GAS of the emm1 genotype arise readily during experimental infection in mice. It is possible that invasive GAS arises from pharyngeal GAS through rare genetic mutations that confer the capacity of mutated GAS to acquire covRS mutations during infection. The objective of this study was to determine whether contemporary pharyngeal emm1 GAS isolates have a reduced propensity to acquire CovRS mutations in vivo compared with invasive emm1 GAS and whether emm3, emm12, and emm28 GAS acquire CovRS mutants in mouse infection. The propensity of invasive and pharyngeal emm1 and invasive emm3, emm12, and emm28 SpeBA+ isolates to acquire variants with the SpeBA- phenotype was determined during subcutaneous infection of mice. The majority of both invasive and pharyngeal emm1 SpeBA+ isolates and two of three emm12 isolates, but not emm3 and emm28 isolates, were found to acquire SpeBA- variants during skin infection in mice. All analyzed SpeBA- variants of emm1 and emm12 GAS from the mouse infection acquired covRS mutations and produced more platelet-activating factor acetylhydrolase SsE. Thus, contemporary invasive and pharyngeal emm1 GAS isolates and emm12 GAS have a similar capacity to acquire covRS mutations in vivo. The rarity of severe invasive infections caused by GAS does not appear to be attributable to a reduced ability of pharyngeal isolates to acquire CovRS mutations.

emm12, and emm28 GAS acquire CovRS mutants in mouse infection. The propensity of invasive and pharyngeal emm1 and invasive emm3, emm12, and emm28 SpeB A+ isolates to acquire variants with the SpeB Aphenotype was determined during subcutaneous infection of mice. The majority of both invasive and pharyngeal emm1 SpeB A+ isolates and two of three emm12 isolates, but not emm3 and emm28 isolates, were found to acquire SpeB Avariants during skin infection in mice. All analyzed SpeB Avariants of emm1 and emm12 GAS from the mouse infection acquired covRS mutations and produced more platelet-activating factor acetylhydrolase SsE. Thus, contemporary invasive and pharyngeal emm1 GAS isolates and emm12 GAS have a similar capacity to acquire covRS mutations in vivo. The rarity of severe invasive infections caused by GAS does not appear to be attributable to a reduced ability of pharyngeal isolates to acquire CovRS mutations.

Introduction
Group A Streptococcus (GAS) is a major human pathogen that causes both relatively mild pharyngitis and superficial skin infections and potentially lethal, severe invasive infections [1]. Severe GAS infections were frequent and often fatal in the 19 th century and reemerged in the 1980s [2]. The reemergence of severe invasive GAS infections in the 1980s is associated with the emergence of a virulent M1T1 clone of emm1 GAS [3][4][5] and virulent emm3 GAS [6]. The M1T1 clone of emm1 GAS is evolved by the acquisition of DNase Sda1-and superantigen SpeA-encoding prophages and the replacement of a 36-kb chromosomal region of pre-1980 M1 GAS with that of emm12 GAS that contains the NADase and streptolysin O genes [7][8][9]. Contemporary M3 GAS acquired a prophage that encodes the superantigen SpeK and phospholipase A2 SlaA [6]. Since 2000, M89 GAS without the genes for synthesis of the hyaluronic acid capsule has also emerged to cause severe invasive infections in the United Kingdom, as well as the United States, Finland, Iceland, and Portugal [10][11][12].
Invasive GAS isolates frequently exhibit a greater capacity to invade soft tissues and evade neutrophil responses in association with higher virulence in experimental animal infections in comparison to pharyngeal isolates. Natural acquisition of mutations of the two-component regulatory system CovRS (also known as CrsRS) leads to hypervirulence [13][14][15][16][17][18][19]. Invasive GAS isolates frequently carry CovRS mutations [20,21], which appear to arise during human infections with GAS carrying wild-type CovRS and are not transmissible [22]. CovRS negatively regulates many virulence factors, including most of those that are involved in the evasion of innate immunity [23][24][25]. As results of CovRS mutations, loss of production of the protease SpeB and enhanced production of the hyaluronic acid capsule and platelet-activating factor acetylhydrolase SsE contribute to the phenotype of hypervirulent invasive M1T1 and M3 GAS isolates [26][27][28][29][30][31][32].
The emergence of CovRS mutants of invasive M1T1 GAS has been readily demonstrated during experimental mouse infections [8,13,14,17,[31][32][33]. Not all emm1 GAS strains are subject to selection of CovRS mutations during skin and soft tissue infections [8]. The DNase sda1 and the hasA and M protein (emm) genes are reported to be required for the selection of CovRS mutants in the M1T1 strain 5448 [31,32]. However, the role of Sda1 in the selection of CovRS mutants of M1T1 GAS has not been confirmed [33]. Among the host factors important for in vivo selection of invasive M1T1 GAS CovRS mutants, neutrophils are required for selection of covRS mutants in mouse infection, and M1T1 CovRS mutants exhibit greater resistance to neutrophils in vivo [17].
Approximately 15 million cases of streptococcal pharyngitis and 10,000 cases of severe invasive GAS disease occur annually in the United States. The basis for the rarity of severe invasive GAS infections compared with common strep throat is not well understood. It is well known that isolates that cause invasive disease are the same as those circulating in the upper respiratory tracts of the community [34], and contemporary pharyngeal emm1 isolates from Finland and emm1 invasive isolates from diverse geographic areas are closely related to the sequenced M1T1 GAS strain MGAS5005 [9]. However, there are single nucleotide polymorphisms (SNP) and indels among isolates of the M1T1 clone [9,14]. It is possible that invasive GAS arises from rare genetic mutations of pharyngeal GAS that confer the capacity of mutated GAS to acquire covRS mutations during infection. MGAS2221, a pharyngeal M1T1 isolate, has been shown to acquire covRS mutations during murine infection [14], inconsistent with the possibility. However, this strain was isolated from a patient with scarlet fever and may not represent a typical pharyngitis case. Thus, parallel comparison with multiple invasive and pharyngeal emm1 isolates in the capacity to acquire covRS mutations should be made to test whether pharyngeal and invasive emm1 isolates have differential capacity to acquire covRS mutations.  [9,20,21,38]. However, emergence of CovRS mutants in experimental animal infections has been demonstrated only for emm1 GAS [8,13,14,17,[31][32][33]. CovRS mutations of an emm3 strain are not detected in muscular infection in nonhuman primates; however, GAS isolates tested might be recovered from primates at day 1 after inoculation [39], which may be too soon to accumulate CovRS mutants. Thus, it is not known whether GAS isolates of the other most dominant invasive emm genotypes other than emm1 can readily acquire CovRS mutations in experimental animal infection.
This study was designed to determine whether pharyngeal emm1 GAS isolates have less propensity to acquire CovRS mutations in vivo than invasive emm1 GAS and whether emm3, emm12, and emm28 GAS can acquire CovRS mutants in mouse infection. The emm89 GAS was excluded for the second question because invasive emm89 GAS lacks the hyaluronic acid capsule [10,11] while the capsule has been shown to be critical for in vivo selection of emm1 CovRS mutants [32]. We first identified isolates that secrete the protease SpeB (SpeB A+ for the presence of SpeB activity in culture supernatant) among 176 invasive GAS isolates collected from patients with necrotizing fasciitis (NF) and/or streptococcal toxic shock syndrome (STSS) in 2010-2013 by the CDC Streptococcus Laboratory and 50 pharyngitis isolates collected in 2014 by the Harborview Medical Center Clinical Microbiology Laboratory at University of Washington School of Medicine. We then compared the capacity of SpeB A+ invasive and pharyngeal emm1 isolates to acquire SpeB Avariants during subcutaneous infection in mice. We also examined the capacity of SpeB A+ invasive emm3, emm12, and emm28 GAS isolates to acquire SpeB Avariants during mouse infection. We found that the majority of both invasive and pharyngeal emm1 SpeB A+ isolates and two of three emm12 isolates acquired SpeB Avariants during skin infection. Sixteen analyzed SpeB Avariants of emm1 and emm12 isolates all acquired covS mutations during infection in mice. Thus, we conclude that both invasive and pharyngeal contemporary emm1 GAS isolates and emm12 GAS have a similar capacity to acquire CovRS mutations in vivo. GAS laboratory strains, clinical isolates, and M protein gene (emm) typing M1T1 strain MGAS2221 [14] and its covS deletion mutant, MGAS2221 ΔcovS [16], have been previously described. A rocA deletion mutant of MGAS2221 (MGAS2221 ΔrocA) will be described elsewhere. One hundred seventy-six invasive GAS isolates from patients with NF and/or STSS were collected from 2010-2013, and their emm genotypes were determined by the Streptococcus Laboratory at the Centers for Disease Control and Prevention (CDC) (S1 Table). These CDC isolates were from 10 diverse geographic areas in the United States. Fifty GAS isolates (HMC isolates) were isolated from throat swabs obtained from patients with pharyngitis during the period of May-August 2014 by the Harborview Medical Center Clinical Microbiology Laboratory in Seattle, Washington (S2 Table). For these invasive isolates from CDC, we simply requested in 2013 the most recent isolates from patients with severe invasive infections of necrotizing fasciitis with or without streptococcal toxic shock syndrome (STSS) and STSS without necrotizing fasciitis. For the pharyngeal isolates, the Harborview Medical Center Clinical Microbiology Laboratory collected 50 pharyngitis isolates in 2014 without any selection criteria. The emm genotypes of the HMC pharyngeal isolates were determined by emm typing according to the CDC emm typing protocol [41] and by using the CDC emm database at http:// www2a.cdc.gov/ncidod/biotech/strepblast.asp.

Bacterial growth
GAS bacteria were statically grown at 37°C in 5% CO 2 in Todd-Hewitt broth supplemented with 0.2% yeast extract (THY). Tryptose agar with 5% sheep blood and THY agar were used as solid media.

Selection of SpeB Avariants in mice
Analysis of in vivo selection of SpeB Avariants of clinical GAS isolates was performed as described previously [17]. More details regarding the experimental procedures and animal care are given below. Female, five-week old C57BL/6 mice used in the selection assay were bred at the Animal Resource Center at Montana State University (ARC) using breeding pairs of mice from the Jackson Laboratory (Bar Harbor, Maine). Adequate care of the animals used in experiments was provided by ARC in accordance with the standards incorporated in the Guide to the Care and Use of Laboratory Animals, 1996 edition (National Academy Press, Washington, D.C.). ARC, an 18,000-sq. ft facility, is fully accredited by the American Association for Accreditation of Laboratory Animal Care and is under the direction of a full-time, ACLAM-board-certified veterinarian. MSU complies with the NIH policy on animal welfare (letter of assurance filed), the Animal Welfare Act and all other applicable federal, state, and local laws. The facility is supplied with HEPA-filtered air and uses individually ventilated cages for mice.
GAS bacteria used in the selection assay were harvested from THY culture at the exponential growth phase and washed three times with pyrogen-free Dulbecco's phosphate-buffered saline (DPBS) and resuspended in DPBS. Groups of 5 or 10 mice were subcutaneously inoculated for each strain with 0.2 ml of GAS suspension in DPBS with an OD 600 of 0.9 after mice were anesthetized by isoflurane inhalation using an isoflurane vaporizer from VetEquip Inhalation Anesthesia Systems. Mice infected with GAS were monitored three times a day at about 8:00 am, 12:30 pm, and 16:30 pm. At day 4 after GAS inoculation, mice were euthanized by CO 2 inhalation, which was done with a gradual fill method at a displacement rate of 30% CO 2 of the chamber volume per minute, as recommended in The 2013 American Veterinary Medical Association Guidelines. SpeB A+ GAS isolates at the used dose caused ruffled fur but did not cause death at the endpoint of the experiment in this subcutaneous infection model.
Skin infection sites of the euthanized mice at day 4 after GAS inoculation were collected and homogenized in 1.0 ml DPBS using a Kontes pestle. The samples were then plated at 10 3 -, 10 4 -, and 10 5 -fold dilutions to obtain well isolated colonies. Forty-eight colonies for each mouse of each test strain were randomly picked, inoculated in 200 μl THY in 96-well plates, and cultured overnight at 37°C in 5% CO 2 . Three μl of 10% β-mercaptoethanol were added into each well, and the cultures were centrifuged in the plate at 3,500 rpm for 10 min. The supernatant samples, 15 μl each, were loaded into wells on casein gel plates and incubated at 37°C for 3 h. The casein gel plate with 96 wells was prepared as follows: 125 ml of heat-solubilized agarose and casein aqueous solution containing 0.38 g Tris base, 1.1 g NaCl, 0.84 g casein, and 1.25 g agarose was poured into the cover plate of a cell culture plate (Greiner Bio-One International Cat.-No. 665 180), and the tube bottom of a 96-well PCR plate was inserted into the casein solution during gel formation. The formation of a cloudy ring around wells due to casein hydrolysis was the indication of the presence of the SpeB activity in culture supernatant (SpeB A+ ), and the lack of the cloudy ring indicated lack of the SpeB activity in the samples (SpeB A-). The number of SpeB Acolonies among 48 colonies from a mouse was counted to calculate the percentage of SpeB Avariants among GAS recovered from each mouse.

Analyses for SsE production
Relative levels of SsE in culture supernatants of parent isolates and their SpeB Avariants were determined by measuring its platelet-activating factor acetylhydrolase activity using the 2-thio-PAF hydrolysis assay [42]. Briefly, 100 μl of the culture supernatant from the exponential growth phase (OD 600 =~0.4) was mixed with 100 μl of reactant solution containing 0.9 mM 2-thio-PAF and 1.3 mM 5,5'-dithiobis-(2-nitrobenzoic acid) at room temperature in wells of a 96-well plate. Absorbance change at 414 nm (ΔA 414 ) as a measure of SsE-catalyzed 2-thio-PAF hydrolysis was recorded with time using a SPECTRA Max 384 Plus spectrophotometer (Molecular Devices).

Statistical analysis
The GraphPad Prism 7 software program (Graph-Pad Software, Inc.) was used for statistical analyses. The data of in vivo selection for SpeB Avariants in Figs 1 and 2 were analyzed using one way ANOVA Tukey's multiple comparison test. In Vivo Selection of SpeB A-Variants of Pharyngeal and Invasive emm1 GAS

Clinical invasive and pharyngeal isolates
The objective of this study was to determine whether pharyngeal emm1 GAS isolates are able to acquire CovRS mutations as readily as invasive M1 isolates in vivo and whether emm3, emm12, and emm28 GAS can acquire CovRS mutants in mouse infection. To achieve this objective, we obtained 176 invasive GAS isolates from the CDC in the United States from 2010 through early 2013 (S1 Table). The 176 GAS samples included 102 isolates from patients with NF only, 60 ones from patients with STSS only, and 14 isolates with both NF and STSS. The Harborview Medical Center Clinical Microbiology Laboratory (Seattle, Washington) (HMC) collected 50 GAS isolates from patients with pharyngitis in May-August 2014 for this study (S2 Table).
The 5 most prevalent emm genotypes of these invasive isolates were emm1 (31 isolates), emm89 (21 isolates), emm3 (16 isolates), emm12 (15 isolates), and emm28 (11 isolates) ( Table 1). These were the most common emm types in the US at that time [37]. The 4 most  prevalent emm genotypes among the pharyngeal isolates were emm1 (9 isolates or 18%), emm12 (8 isolates or 16%), emm89 (5 isolates or 10%), emm28 (4 isolates or 8%) ( Table 2). Each of the emm2, emm3, and emm87 genotypes was represented by 3 pharyngeal isolates (6%). A limitation of the HMC GAS collection is its small size and origin from a single location. Nevertheless, it is reasonable to conclude that the emm1 genotype appears to be prevalent among both invasive and pharyngeal GAS isolates and that the prevalent emm genotypes in severe invasive infections are similar to the prevalent emm genotypes of circulating pharyngeal GAS strains, which is discussed in more details in Discussion.
Presence of sda1, speA, and nga 989G allele in all emm1 isolates To determine whether all the emm1 isolates in both collections belonged to the M1T1 subclone, emm1 isolates were analyzed by PCR using sda1-and speA-specific primers, and all strains generated PCR products of expected sizes. Thus, all the emm1 isolates carried the sda1 and speA genes. It has been reported that isolates of the M1T1 clone have a replacement of a 36-kb, sloand nga-containing chromosomal region of pre-1980 emm1 GAS with that of serotype emm12 GAS, which results in enhanced expression of SLO and NADase and has the 989G allele in the NADase gene nga [6][7][8][9]. We attempted to compare SLO levels in mid-exponential growth phase culture supernatants of SpeB A+ emm1 isolates, M1T1 strain MGAS2221 (positive control), and SF370 (a pre-1980 M1 strain and negative control) by Western blotting using SLOspecific antibodies. However, we could not consistently distinguish SF370 from MGAS2221 or SpeB A+ emm1 isolates on the basis of SLO levels. Thus, we determined the polymorphism at base 989 of the nga gene, and all the emm1 isolates had the 989G allele. Based on these results,  all of the emm1 isolates in the two collections are most likely similar to the contemporary Finland pharyngeal emm1 isolates that are closely related to MGAS5005 [9], a representative strain of the currently pandemic M1T1 clone.

SpeB production in clinical invasive and pharyngeal isolates
Next, SpeB protease activity in the supernatants of overnight cultures of the clinical isolates was assayed by casein hydrolysis. Overall, 37.5% of the 176 invasive isolates were SpeB A-. SpeB A+ isolate samples did not contain SpeB Abacteria as determined by checking SpeB activity of 5 or more independent colonies of each sample. Among the 5 most frequent emm genotypes of the invasive isolates, about 20% of emm1 and emm12 isolates, 60% of emm3 and emm28 isolates, and 80% of 21 invasive emm89 isolates were SpeB A-( Table 1). The SpeB Apercentage of the invasive emm1 isolates is slightly higher than reported for a larger invasive emm1 GAS collection [38]; however, the percentage of our invasive emm89 isolates that were SpeB Ais 4 times higher than observed in invasive emm89 isolates in the earlier study [38]. As our 5 emm89 pharyngeal isolates were all SpeB A+ , the association of the SpeB Aphenotype with emm89 NF and/or STSS isolates appears to be genuine. The percentages of SpeB Aisolates among NF, STSS, and NF/STSS isolates were 39.2%, 33.3%, and 42.8%, respectively. Thus, isolates from these three different forms of severe invasive infection exhibit similar percentages of isolates with the SpeB Aphenotype. Six of the 50 HMC pharyngeal isolates were SpeB A-, including 1 isolate each from the emm1, emm3, emm6, emm9, emm12, and emm22 genotypes (Table 2). Notably, all 5 emm89 isolates and 8 out of 9 emm1 isolates were SpeB A+ . Thus, pharyngeal emm1 and emm89 isolates appear to have fewer percentages of isolates with the SpeB Aphenotype than NF and/or STSS isolates.

Selection of SpeB Avariants of invasive and pharyngeal emm1 SpeB A+ isolates
The screening of invasive and pharyngeal isolates by the SpeB activity assay identified 24 invasive and 8 pharyngeal emm1 SpeB A+ isolates (Table 3). These isolates allowed us to determine whether invasive and pharyngeal emm1 GAS isolates have a different tendency to acquire CovRS mutations in vivo. The study was mainly the comparison of pharyngeal emm1 isolates with severe invasive emm1 isolates in the capacity to acquire CovRS mutants in vivo, and 19 of the 31 severe invasive emm1 isolates were from patients with necrotizing fasciitis. Thus, both pharyngeal and subcutaneous infection models should be relevant for the purpose. We had preliminary data for in vivo selection of MGAS2221 covRS mutants in intranasal infection model. However, the intranasal model is more technically challenging because of the presence of non-GAS bacteria. Thus, we chose the model of subcutaneous infection of mice. We used the SpeB activity assay to identify SpeB Avariants among isolates that were recovered from skin infection sites in mice on day 4 after inoculation of each SpeB A+ emm1 GAS isolate. SpeB Avariants were detected at skin infection sites for all 24 invasive SpeB A+ emm1 isolates (Fig 1). Twenty-one of them had mean SpeB A-% values ranging from 12.5% to 96.5%, and 3 others had low mean SpeB A-% values ranging from 3.5% to 7.9% (Table 3). SpeB Avariants were detected at skin infection sites for 7 of the 8 SpeB A+ pharyngeal emm1 isolates (Fig 2). Six had mean SpeB A-% values from 13.3% to 81.2%, and one had 1.7% SpeB Avariants (Table 3). One pharyngitis isolate had no detectable SpeB Avariants. Thus, like contemporary invasive emm1 isolates, the majority of contemporary pharyngeal emm1 GAS isolates are able to give rise to SpeB Avariants in vivo.

Significant difference among emm1 isolates in the capacity to acquire SpeB Avariants in murine subcutaneous infection
Isolates of both the invasive and pharyngeal emm1 collections had variable average SpeB A-% value in the in vivo selection for SpeB Avariants in mice. Statistical analysis of these data using one way ANOVA Tukey's multiple comparison test could divide the isolates into four groups on the basis of significant difference in the SpeB A-% data for both invasive (iG1-iG4) and pharyngeal (pG1-pG4) isolates (Table 3). The iG1 and pG1 isolates had >80% of average SpeB A-% and were significantly different in SpeB A-% from iG2 and pG2 with average SpeB A-% of <10% (P <0.0001), and the average SpeB A-% values of the iG4 and pG4 isolates ranging from 10% to 30% were significantly different from those of the iG1 and pG1 isolates (P <0.05). Both iG3 In Vivo Selection of SpeB A-Variants of Pharyngeal and Invasive emm1 GAS and pG3 had average values of SpeB A-% ranging from 30% to 60% and were not significantly different from isolates in the iG, iG2, pG1, and pG2 groups. Thus, some isolates in both invasive and pharyngeal emm1 isolates have significantly higher capacity to acquire SpeB Avariants than others in the same categories.

CovS mutations and enhanced SsE expression in emm1 SpeB Avariants selected in mouse infection
Although we have reported that the SpeB Aphenotype is a reliable indicator for CovRS mutations of MGAS2221 in subcutaneous infection of mice [17], CovRS mutations may not be the only cause of the SpeB Aphenotype. RocA has been shown to function through CovR, and rocA deletion enhances expression of CovRS-regulated virulence genes [43]. Thus, we determined whether rocA mutations are a cause of the SpeB Aphenotype by comparing the SpeB activity in culture supernatant of M1T1 strain MGAS2221 and its ΔrocA and ΔcovS mutants. MGAS2221 ΔcovS was SpeB Abut both MGAS2221 ΔrocA and MGAS2221 were SpeB A+ ( Fig  3A). Thus, rocA mutations are not a cause of the SpeB Aphenotype. Another possible common cause of the SpeB Avariants is mutations in the rgg gene [38,39,44]. We sequenced the covRS and rgg genes in 5 emm1 invasive SpeB A+ isolates and 2 SpeB Avariants of each of them selected in skin infection. The 5 clinical isolates all had same covRS and rgg sequences with MGAS2221, and all 10 SpeB Avariants had missense mutations or indels of covS but have wild-type rgg (Table 4). Being consistent with the previous findings that natural CovRS mutations enhance SsE expression [16,29,33], all the SpeB Avariants had higher levels of SsE PAF acetylhydrolase activity than their parent strains in culture supernatant at the exponential growth phase (Fig 3B). Thus, SpeB Avariants of emm1 GAS selected in subcutaneous infection in mice are caused primarily by CovRS mutations and have enhanced virulence gene expression.

Selection of CovRS mutants of emm12 GAS in subcutaneous mouse infection
To determine whether emm3, emm12, and emm28 GAS has the capacity to acquire SpeB Amutations in mouse infection, we tested 6 emm3, 3 emm12, and 4 emm28 SpeB A+ invasive isolates for selection of SpeB Avariants during cutaneous infection in mice. No SpeB Avariants were detected at day 4 after infection with the emm3 and emm28 isolates ( Fig 4A). However, 10.8% of recovered M12 isolate 1867 were SpeB A-, and isolates recovered from emm12 strain 1845 infection contained 1.7% SpeB Avariants ( Fig 4A). Four SpeB Avariants of GAS1845 and 2 SpeB Avariants of GAS1867 all had covS mutation or deletion and lacked rgg mutation compared with the parent strains (Table 4). Like the emm1 SpeB Avariants, all these emm12 SpeB Avariants had enhanced SsE PAF acetylhydrolase activity in their culture supernatant at the exponential growth phase (Fig 4B). Thus, arising of emm12 GAS CovRS mutants is demonstrated in subcutaneous infection of mouse.

Discussion
The main objective of this study was to test the hypothesis that pharyngeal emm1 GAS isolates have less propensity to acquire CovRS mutations in vivo than invasive emm1 isolates, as a possible explanation for the rarity of severe invasive GAS infections caused by pandemic M1T1 GAS. Our experimental observations do not support this hypothesis, as contemporary emm1 isolates from pharyngitis patients and patients with NF and/or STSS exhibit similar capacity to acquire CovRS mutations following subcutaneous infection of mice. Thus, the explanation for the rarity of severe invasive GAS infections remains to be fully elucidated. We also showed that contemporary emm1 isolates from both pharyngitis and severe invasive infections can have significantly different propensity to acquire CovRS mutations. In addition, we found that emm12, but not emm3 and emm28, GAS isolates can acquire CovRS mutations in subcutaneous infection of mice. The findings provides information for the understanding of the arising of GAS mutants in a hostile environment. Although contemporary emm1 isolates belong to the M1T1 clone, they have single nucleotide polymorphisms [9,14]. It is possible that some M1T1 GAS bacteria acquire rare genetic  Table 4. The SsE activity was measured using the colorimetric assay as described in the Methods. alterations that render them to selection for CovRS mutations, causing rare severe invasive infections. However, this possibility has been excluded by our finding that pharyngeal emm1 isolates have the capacity to acquire CovRS mutations during mouse infection at frequencies similar to those of invasive emm1 isolates. Nevertheless, our data are consistent with the consensus notion in the field that contemporary emm1 GAS strains cause severe invasive infections in only a small fraction of the population, suggesting that additional bacterial genetic determinants and/or host factors are involved.
As in a recent analysis of SpeB production by invasive GAS isolates, the majority (74.2%) of the CDC emm1 isolates were found to be SpeB A+ [38]. The prevalence of SpeB Aisolates is similar to that of SpeB Avariants identified on day 4 after murine infection with SpeB A+ emm1 isolates. Indeed, mixed GAS populations with or without CovRS mutations are observed in patients [45]. The likelihood to isolate a CovRS mutant depends on where and when isolates are recovered from the host [5]. Thus, the frequency of SpeB Aand SpeB A+ phenotypes among clinical invasive GAS isolates should be interpreted with caution. Mixed GAS populations with or without CovRS mutations are also consistent with the suggestion that CovRS mutations arise during human infection with GAS carrying wild-type CovRS and are not transmissible [22]. The inability of CovRS mutants to transmit suggests a "source-sink" dynamic in which GAS with wild-type CovRS has an advantage in transmission and pharyngeal infection while GAS with CovRS mutations contributes to rarer episodes of severe invasive disease in susceptible population.
The pharyngeal isolates in this study were from a single location whereas the CDC invasive isolates represent diverse geographic areas in the United States. Whether our comparison is relevant depends on whether the pharyngeal emm1 isolates in Seattle are similar to those in the other geographic areas in the United States. It is well established that contemporary emm1 GAS strains belonged to the invasive M1T1 subclone emerged in the 1980s and have essentially displaced antecedent emm1 strains across a broad geographic region [3,4,9]. Contemporary  Table 4. severe and non-severe invasive emm1 isolates in the United States are genetically related to the invasive M1T1 clone [9]. Contemporary pharyngeal emm1 isolates in the United States must be dominantly related to the invasive M1T1 clone because there are evidences for the origination of invasive GAS strains from pharyngitis strains [34,46]. This possibility is supported by the fact that contemporary pharyngeal and invasive emm1 strains in Canada and Finland were related to the invasive M1T1 clone [9]. All the pharyngeal and invasive emm1 isolates in this study carry the sda1 and speA genes and the nga 989G allele, consistent with the invasive M1T1 clone. These data and the previous findings support that emm1 pharyngeal GAS strains in the United States are dominantly related to the M1T1 subclone. Thus, the geographic limitation in our analyses unlikely invalidates our conclusion that contemporary pharyngeal and invasive emm1 isolates have a similar capacity to acquire CovRS mutations in vivo.
Invasive GAS isolates frequently have CovRS mutations resulting in a SpeB Aphenotype [20,21,38]. However, the selection of SpeB Avariants has been previously demonstrated in mouse infections only for M1T1 GAS [8,13,14,17,[31][32][33]. Neutrophils are required for in vivo selection of M1T1 GAS SpeB Avariants [17]. The capsule and M protein are critical for the selection of SpeB Avariants of M1T1 GAS [32]. However, capsule and M protein are critical virulence factors for both contemporary and pre-1980 emm1 GAS strains. Reports are conflicting as to whether Sda1 is critical for the selection of SpeB Avariants of invasive M1T1 GAS [31,33]. Thus, the critical trigger for in vivo selection of M1T1 SpeB Avariants is presently unknown. The present study shows that selection of CovRS mutants of emm12 GAS isolates can also be demonstrated in mice. We also showed that some emm1 isolates are overpopulated by emerged CovRS mutants within a few days while others accumulate small percentages of CovRS mutants at the same time point in subcutaneous infection of mice. Further characterization of the selection of CovRS mutants of emm12 GAS and elucidation of the basis for the differential propensity of contemporary emm1 GAS to acquire CovRS mutants in mouse infection may provide additional clues regarding the trigger for in vivo selection of GAS CovRS mutations.
No SpeB Avariants were detected after cutaneous infection with SpeB A+ emm3 isolates. One possible reason is that emm3 isolates have a natural rocA mutation that enhances expression of CovRS-controlled virulence factors [43] and thus alleviates the selective pressure for CovRS mutants. However, this possibility is not consistent with the observation that invasive emm3 isolates are more likely to contain CovRS mutations than pharyngeal emm3 isolates [24]. Furthermore, even in a rocA null background, CovRS mutations enhance the expression of virulence genes and critically contribute to emm3 GAS virulence [21]. An alternative possibility is that the screening of M3 GAS by the SpeB activity assay failed to pick up certain CovRS mutations. For example, the CovS G457V point mutation of invasive M3 isolate MGAS315 enhances expression of virulence genes and critically contribute to its virulence; however, this point mutation does not cause a SpeB Aphenotype in MGAS315 [19]. It is also possible that human infections have a mechanism for the selection of emm3 GAS CovRS mutants that is not mimicked in the murine skin infection model.
It is interesting that 17 of 21 invasive emm89 isolates (80%) are SpeB Awhile all 5 pharyngeal emm89 isolates are SpeB A+ . It has been reported that 20.7% of invasive emm89 isolates are SpeB A- [38], and these emm89 isolates belong to three major phylogenetic groups [11]. GAS in one of the tree groups lacks the hasABC capsule biosynthesis locus and has emerged as a frequent cause of severe invasive infections since 2000 [10][11][12]. The unusually high prevalence of the SpeB Aphenotype in CDC emm89 invasive isolates could be due to an enhanced selection pressure for acapsular emm89 GAS strains, leading to a propensity to cause necrotizing fasciitis and toxic shock syndrome.
Kazmi et al. first reported SpeB A+ -to-SpeB Aphase shift of M1T1 GAS in mouse infection [47]. A covS mutant of M1T1 GAS arisen in mice confers the SpeB Aphenotype [14]. The regulator rgg also regulates speB [48]. The SpeB Aphenotype has been detected in many clinical isolates, and clinical SpeB Aisolates are associated dominantly with mutations in rgg and covS and less frequently with covR mutations [38]. Surprisingly, we only detected mutations of covS but not rrg in SpeB Avariants of emm1 and emm12 isolates in subcutaneous infection of mice in this study. Previous studies also reported that SpeB Avariants of emm1 GAS selected in mouse infection are associated with covRS mutations [8,13,14,15,17,[31][32][33]. All these data resonate the lack of the selection of SpeB Avariants of emm3 GAS in subcutaneous infection of mice whereas covRS mutations are frequent in emm3 clinical isolates [21]. These findings suggest that different GAS strains may be subject to different stresses in different infections and that mutations of covRS and rgg confer different fitness advantage in serotype-dependent and infection-dependent ways.

Conclusions
We conclude that contemporary invasive and pharyngeal emm1 GAS isolates have a similar capacity to acquire CovRS mutations in vivo. We also conclude that emm2, but not emm3 and emm28, GAS can also readily acquire CovRS mutants in subcutaneous infection of mice.
Supporting Information S1 In Vivo Selection of SpeB A-Variants of Pharyngeal and Invasive emm1 GAS