RocA Has Serotype-Specific Gene Regulatory and Pathogenesis Activities in Serotype M28 Group A Streptococcus

Serotype M28 group A streptococcus (GAS) is a common cause of infections such as pharyngitis (“strep throat”) and necrotizing fasciitis (“flesh-eating” disease). Relatively little is known about the molecular mechanisms underpinning M28 GAS pathogenesis.

ABSTRACT

Serotype M28 group A streptococcus (GAS) is a common cause of infections such as pharyngitis (“strep throat”) and necrotizing fasciitis (“flesh-eating” disease). Relatively little is known about the molecular mechanisms underpinning M28 GAS pathogenesis. Whole-genome sequencing studies of M28 GAS strains recovered from patients with invasive infections found an unexpectedly high number of missense (amino acid-changing) and nonsense (protein-truncating) polymorphisms in rocA (regulator of Cov), leading us to hypothesize that altered RocA activity contributes to M28 GAS molecular pathogenesis. To test this hypothesis, an isogenic rocA deletion mutant strain was created. Transcriptome sequencing (RNA-seq) analysis revealed that RocA inactivation significantly alters the level of transcripts for 427 and 323 genes at mid-exponential and early stationary growth phases, respectively, including genes for 41 transcription regulators and 21 virulence factors. In contrast, RocA transcriptomes from other GAS M protein serotypes are much smaller and include fewer transcription regulators. The rocA mutant strain had significantly increased secreted activity of multiple virulence factors and grew to significantly higher colony counts under acid stress in vitro. RocA inactivation also significantly increased GAS virulence in a mouse model of necrotizing myositis. Our results demonstrate that RocA is an important regulator of transcription regulators and virulence factors in M28 GAS and raise the possibility that naturally occurring polymorphisms in rocA in some fashion contribute to human invasive infections caused by M28 GAS strains.

INTRODUCTION

Group A streptococcus (GAS) is a human-specific pathogen that causes diseases ranging in severity from relatively innocuous pharyngitis (“strep throat”) to life-threatening necrotizing fasciitis (“flesh-eating” disease) (1, 2). In addition, GAS is responsible for postinfectious immune sequelae such as rheumatic heart disease and poststreptococcal glomerulonephritis (3). The global disease burden and economic impact of GAS disease are immense. The World Health Organization estimates that GAS causes over 700 million superficial infections, 1.78 million invasive infections, and 512,000 deaths annually (4). Despite decades of research, there is no commercially available vaccine to prevent GAS infections.

GAS strains are commonly classified by sequence variation in the emm gene, which encodes the highly polymorphic M protein virulence factor (3). Serotype M28 strains are among the more common causes of GAS pharyngitis and invasive infections in the United States and other countries (1, 5–11). Of note, serotype M28 GAS strains are strongly associated with puerperal sepsis (7–9, 12–14). Despite the importance of M28 strains in human disease, relatively little is known about the molecular pathogenesis of M28 GAS (1, 5, 12, 13). Historically, GAS pathogenesis research has focused on strains of other numerically important serotypes, such as M1 and M3 (2, 15–21).

One gene of increasingly recognized importance to GAS pathogenesis is rocA (regulator of Cov), encoding the RocA protein (22). RocA was initially identified as a positive regulator of the CovRS (control of virulence) two-component system, which is a negative regulator of virulence (22–24). Further study of rocA in multiple GAS serotypes identified a naturally occurring nonsense mutation in all serotype M18 strains that results in hyperencapsulation and increases carriage longevity in the mouse nasopharynx (25). Similarly, RocA is inactivated in all serotype M3 strains by a single nucleotide deletion that introduces a frameshift mutation, and restoration of RocA by introduction of the serotype M1 wild-type rocA allele decreases M3 GAS virulence in a mouse model of bacteremia (26–28). Deletion of rocA in serotype M1, M3, M6, M14, M18, and M89 GAS results in increased expression of virulence factors known to be regulated by the CovRS system (26–32). Although the molecular mechanism for RocA function has not been determined, RocA increases phosphorylation of the DNA binding response regulator CovR in the presence of its cognate sensor histidine kinase CovS, resulting in CovR activation (28, 30). The N-terminal transmembrane domains of RocA are crucial for the regulatory activity of RocA, suggesting that RocA functions as an accessory protein to the CovRS system (32). However, the role, if any, of RocA in pathogenesis has not been studied in M28 GAS.

Whole-genome sequencing studies of M28 GAS strains recovered from human invasive infections found an unexpectedly high number of missense (amino acid-altering) and nonsense (protein-truncating) polymorphisms in rocA (GenBank accession no. MH884522 to MH884551) (Fig. 1). Previous studies of rocA in strains of other GAS M protein serotypes identified several nonsense mutations and many frameshifting insertions and deletions (indels) that result in protein truncation (25–28, 30–32, 34–36). However, very few rocA missense mutations have been reported in other GAS serotypes, and none have been studied previously (25–28, 30–32, 34–36). The striking increase in missense mutation frequency in M28 strains led us to speculate that RocA has serotype-specific functions in M28 GAS strains. We hypothesized that RocA inactivation significantly contributes to the molecular pathogenesis of invasive infections caused by serotype M28 GAS. To test this hypothesis, we created an isogenic rocA deletion mutant strain and discovered that RocA regulates a substantial portion of the M28 GAS transcriptome, including many genes encoding transcription regulators and virulence factors. Consistent with the transcriptome data, in vitro assays showed that RocA inactivation significantly increases the secreted activity of multiple virulence factors and increases CFU under acid stress. The RocA-inactivated strain was also significantly more virulent in a mouse model of necrotizing myositis.

FIG 1.

rocA is unusually polymorphic in serotype M28 GAS strains. The affected codon and amino acid change conferred by each polymorphism are shown. For polymorphisms due to nucleotide deletion, the affected nucleotide is identified. Alleles identified in multiple isolates are indicated. Polymorphisms that result in RocA protein truncation or loss of rocA mRNA translation are shown below the protein schematic, and polymorphisms that result in amino acid changes are shown above the protein schematic. Missense polymorphisms in the predicted domains sufficient for regulatory activity (32) are colored red. Predicted domains of the RocA protein using Phyre2 are indicated (TM, transmembrane domain; HATPase, histidine kinase ATPase domain) (107). Predicted functional domains of the potential histidine kinase domain (H box, N box, F box, and G box) are identified (22). #, one strain has two polymorphisms in rocA.

RESULTS

rocA is unusually polymorphic in serotype M28 GAS.

Recent whole-genome sequencing studies of serotype M28 GAS strains recovered from human invasive infections revealed an unexpectedly high number of missense and nonsense polymorphisms in rocA (GenBank accession no. MH884522 to MH884551) (Fig. 1). We identified 29 unique polymorphisms (25 single nucleotide polymorphisms [SNPs]) and 4 insertions/deletions [indels]) among 2,101 M28 GAS strains. The frequency of polymorphisms in rocA is significantly greater than expected by chance alone (P < 0.01, Fisher's exact test). The 29 unique polymorphisms included 17 missense mutations, 11 nonsense mutations, and one six-nucleotide deletion in the upstream noncoding region that affects the presumed ribosomal binding site (30) (Fig. 1). Of note, 13/17 (76.5%) of the missense mutations occur in the 5′ end of rocA, resulting in amino acid changes in the N terminus of RocA that may be crucial for its function as an accessory protein to the CovRS system (32) (Fig. 1).

The abundance of rocA polymorphisms found in the M28 GAS strains prompted a reevaluation of data from our previously published whole-genome sequencing studies of large, comprehensive, population-based collections of other GAS M protein serotypes (15, 37, 38). We discovered that M1 and M59 GAS strains had a much lower frequency of rocA polymorphisms (see Fig. S1 in the supplemental material) (15, 37, 38). Although 29 unique rocA polymorphisms were found among 2,101 M28 GAS strains (13.8 rocA alleles per 1,000 M28 strains), only 16 unique rocA polymorphisms were identified among 3,443 M1 GAS strains (4.6 alleles per 1,000 strains) (15), and two unique rocA polymorphisms were identified among 310 M59 GAS strains (6.5 alleles per 1,000 strains) (38). In contrast, 18 unique rocA polymorphisms were found among 1,193 M89 GAS strains (15.1 alleles per 1,000 strains) (37).

The very high number of variants identified in serotype M28 GAS strains suggests that the rocA polymorphisms are selected for during human invasive infection to alter the regulatory activity of RocA. We hypothesize that altered RocA activity contributes to the molecular pathogenesis of M28 GAS.

Creation of an isogenic rocA deletion mutant strain.

As a first step toward investigating the role of RocA in serotype M28 GAS molecular pathogenesis, we created an isogenic rocA deletion mutant strain using allelic exchange (39). Wild-type (WT) strain MGAS28426 was chosen as the parental strain because it is genetically representative of serotype M28 GAS strains that commonly cause human infections, and it has a wild-type allele for all major global transcription regulatory genes, including covRS, ropB, mga, ccpA, and rocA. Whole-genome sequencing of the isogenic rocA deletion (ΔrocA) mutant strain confirmed the absence of spurious mutations. To determine if rocA deletion alters the growth phenotype of M28 GAS, the parental WT and isogenic ΔrocA mutant strains were grown in Todd-Hewitt broth supplemented with yeast extract (THY), a nutrient-rich medium. No significant difference in growth was observed (Fig. 2A) (P = not significant [NS], two-way analysis of variance [ANOVA]).

FIG 2.

Deletion of rocA significantly alters the GAS transcriptome. (A) No significant growth difference in nutrient-rich liquid medium was observed between the parental wild-type (WT) and isogenic ΔrocA mutant strains. Dashed lines represent the OD₆₀₀ of mid-exponential (ME) and early stationary (ES) growth phases for cultures that were collected for RNA-seq analysis. (B) Principal-component analysis of the WT and ΔrocA strain transcriptomes at the ME and ES growth phases. (C) Numbers of genes with significantly altered transcript levels at the ME and ES growth phases (108).

Deletion of rocA in M28 GAS strain MGAS28426 results in a substantial transcriptome change.

To test the hypothesis that rocA deletion results in altered global gene transcript levels in M28 GAS, transcriptome sequencing (RNA-seq) analysis was performed using strains grown to mid-exponential (ME) (optical density at 600 nm [OD₆₀₀] = 0.5) and early stationary (ES) (OD₆₀₀ = 1.65) growth phases (Fig. 2A). Principal-component analysis showed that rocA deletion markedly alters the global transcriptome of serotype M28 GAS at both growth phases (Fig. 2B). In total, 427 (25.8%) and 323 (19.5%) genes had significantly altered transcript levels at ME and ES growth phases, respectively (absolute transcript fold change, ≥1.5; P < 0.05 after Baggerly's test with Bonferroni's correction for multiple comparisons) (Fig. 2C). Of these genes, 109 were common to both growth phases (Fig. 2C). Many of the significantly differentially expressed genes encode transcription regulators and proven virulence factors (see below). A complete list of genes with significantly altered transcript levels is provided in Tables S1 and S2 in the supplemental material.

RocA directly or indirectly regulates transcription regulators involved in virulence in serotype M28 GAS.

The RNA-seq data demonstrated that RocA inactivation significantly altered the transcript levels of 41 transcription regulators in serotype M28 strain MGAS28426 (Table 1) (40–61). Of the 41 transcription regulators that are directly or indirectly regulated by RocA in M28 GAS, 22 have been previously studied in GAS, 11 have inferred function by homology with transcription regulators in other Streptococcus species, and 8 are of unknown function (Table 1) (40–61). One particularly interesting regulator whose expression is significantly altered by RocA inactivation in M28 GAS is mga (multiple virulence gene regulator of GAS) (57). Mga regulates the expression of multiple genes encoding proven virulence factors, including sclA (encoding a collagen binding protein) (62), fba (encoding a fibronectin binding protein) (63), scpA (encoding C5a peptidase) (64), enn (encoding IgA binding protein) (65), emm (encoding antiphagocytic M protein) (66), mrp (encoding M-related protein) (67), sfbX (encoding a fibronectin binding protein) (68), and sof (encoding serum opacity factor [SOF]) (69). Compared to the parental WT strain, the isogenic ΔrocA deletion mutant strain had significantly increased transcript levels for each gene in the Mga regulon at one or both time points (Fig. 3A and Tables S1 and S2).

TABLE 1.

GAS transcription regulator genes (proven and inferred) directly or indirectly regulated by RocA at mid-exponential and early stationary growth phases

Locus taga	Gene	Known or putative function (reference)b	Fold change relative to WTc
			ME	ES
M28_Spy0034	comRd,f	Competence (40)	−1.7
M28_Spy0104	rofAd,e,f	Regulator of fibronectin binding protein (41)	−1.9
M28_Spy0153	sgaR	Ascorbate utilization (42)	−1.5
M28_Spy0184	rivRd,e	Negative regulator of GRAB (43)	2.4
M28_Spy0189	yjdR	Multidrug resistance transporters (42)	−2.3
M28_Spy0276	nrdRd	Ribonucleotide metabolism (44)		1.8
M28_Spy0522	agaR2	Carbohydrate metabolism (42)	−1.9
M28_Spy0538	ralp3d	RofA-like transcription regulator (45)	−2.3	−1.8
M28_Spy0681	cpsYd,e,f	Resistance to opsonophagocytic killing (46)	1.6
M28_Spy0780	srtKd	Lantibiotic biosynthesis(47)		1.7
M28_Spy0872	M28_Spy0872f	GntR family transcription regulator (42)	−1.8
M28_Spy0889	nagR	N-Acetylglucosamine utilization (42)		2.8
M28_Spy0896	pdxR	Pyridoxin metabolism (42)		1.8
M28_Spy0919	ciaHd	Acid and oxidative stress (48)		2.1
M28_Spy0920	ciaRe	Acid and oxidative stress (48)		2.0
M28_Spy0963	M28_Spy0963f	Transport (42)	2.0	−1.7
M28_Spy1346	trxRd,e	Two-component system (49)	−1.9
M28_Spy1347	trxSd,e	Two-component system (49)	−2.0
M28_Spy1373	liaRd,e	Regulator of pilus proteins (50)		−1.7
M28_Spy1384	atoRd,e,f	Short-chain fatty acid metabolism (51)	1.7
M28_Spy1420	M28_Spy1420	Mga family transcription regulator (42)		−2.0
M28_Spy1445	lacR.1d	Galactose metabolism (52)		−2.3
M28_Spy1449	copYd,f	Copper toxicity (53)	−2.4
M28_Spy1501	codYd	Pleiotropic transcription regulator (54)	1.5
M28_Spy1531	scrR	Sucrose utilization (42)		−1.6
M28_Spy1545	M28_Spy1545f	XRE family transcription regulator (42)	1.7
M28_Spy1546	M28_Spy1546f	XRE family transcription regulator (42)	1.5
M28_Spy1564	srvd,e	Streptococcal regulator of virulence (55)		3.1
M28_Spy1566	M28_Spy1566f	XRE family transcription regulator (42)	−2.5
M28_Spy1569	M28_Spy1569	MerR family transcription regulator (42)	−1.6
M28_Spy1615	salRd	Lantibiotic biosynthesis (56)		1.8
M28_Spy1636	M28_Spy1636	XRE family transcription regulator (42)		−1.9
M28_Spy1704	mgad,e	Multiple gene regulator (57)	2.2	2.0
M28_Spy1708	ihkd,e,f	Polymorphonuclear leukocyte evasion (58)	1.5
M28_Spy1724	ropBd,e	Regulator of SpeB (59)	−2.3
M28_Spy1750	ctsRf	Stress and heat shock response (60)		2.7
M28_Spy1763	M28_Spy1763f	LuxR family transcription regulator (42)	−1.6
M28_Spy1769	treRf	Trehalose utilization (42)	−1.5
M28_Spy1782	spxA2d,e	Stress resistance, regulator of SpeB (61)	4.0
M28_Spy1835	ywzGf	Transport (42)	−1.5	−2.0
M28_Spy1839	pipR	Phage infection protein (42)		1.8

^aLocus tag identified in the serotype M28 reference genome MGAS6180.

^bKnown or putative function based on known role in GAS or inferred homology.

^cME, mid-exponential growth phase; ES, early stationary growth phase. Empty (blank) cell, the gene does not satisfy the P value and/or fold change requirement.

^dTranscription regulator that has been previously studied in GAS.

^eTranscription regulator with a proven role in GAS virulence.

^fTranscription regulator unique to the M28 ME RocA transcriptome compared to the M1 and M3 ME RocA transcriptomes (28, 32).

FIG 3.

Deletion of rocA significantly increases the transcript levels of genes in the Mga regulon. (A) The transcript levels of mga and eight Mga-regulated genes were significantly increased in the ΔrocA mutant strain compared to the WT strain. Genomic coordinates and fold change in transcripts are shown for each gene at mid-exponential and early stationary growth phases (P < 0.05, Baggerly test with Bonferroni correction for multiple comparisons). Points below the 1.5-fold-change cutoff did not reach statistical significance and are included for completeness. (B) Serum opacity factor (SOF) activity assay results. Data are shown as mean ± standard deviation. ***, P < 0.001; ****, P < 0.0001 (Student's t test).

To determine if the observed difference in transcript levels of genes in the Mga regulon results in an altered phenotype, SOF activity was assayed in vitro. Consistent with the RNA-seq data, the isogenic ΔrocA deletion mutant strain had significantly increased SOF activity compared to the parental WT strain (Fig. 3B).

RocA directly or indirectly regulates transcription regulators and virulence factors involved in the stress response in serotype M28 GAS.

During infection, GAS cells are exposed to oxidative and acid stress in purulent lesions (70–72). Among the 41 transcription regulators that are directly or indirectly regulated by RocA in M28 GAS, 4 are implicated in oxidative and acidic stress responses, including the CiaHR two-component system, NrdR, and SpxA2 (Table 1) (44, 48, 61). Additionally, the arcABCD operon had significantly increased transcript levels in the ΔrocA mutant strain (Fig. 4A and Table S1). The arcABCD operon encodes the ArcABCD proteins of the arginine deiminase pathway, which are also involved in the GAS response to acidic environments (73–75).

FIG 4.

Deletion of rocA significantly increases the transcript levels of genes encoding transcription regulators and proteins involved in the stress response. (A) The transcript levels of arcABCD and spxA2 were significantly increased in the ΔrocA mutant strain compared to the WT strain at mid-exponential growth phase. M28_Spy1209 encodes a putative dipeptidase, and M28_Spy1212 encodes a putative N-acetyltransferase. Genomic coordinates and fold change in transcripts are shown for each gene (P < 0.05, Baggerly test with Bonferroni correction for multiple comparisons). (B) The transcript levels of nrdR and ciaHR were significantly increased in the ΔrocA mutant strain compared to the WT strain at early stationary growth phase. Genomic coordinates and fold change in transcripts are shown for each gene (P < 0.05, Baggerly test with Bonferroni correction for multiple comparisons). (C) Growth of strains in THY buffered with HEPES (pH 7.5). (D) Growth of strains in THY buffered with 2-(N-morpholino)ethanesulfonic acid (MES) (pH 6.0). (E) CFU counts of strains grown in THY buffered with MES (pH 6.0) at 3 h. Data are shown as mean ± standard error of the mean (SEM). *, P < 0.05 (Mann-Whitney test).

Because ciaHR, nrdR, spxA2, and arcABCD had significantly increased transcript levels in the isogenic ΔrocA deletion mutant strain compared to the parental WT strain (Fig. 4A and B), we hypothesized that the isogenic ΔrocA deletion mutant strain is significantly more resistant to acidic stress. To test this hypothesis, the isogenic ΔrocA deletion mutant and parental WT strains were grown in THY alone, THY buffered to neutral conditions using HEPES (pH 7.5), and THY buffered to acidic conditions using 2-(N-morpholino)ethanesulfonic acid (MES) (pH 6.0) (29). Under neutral conditions (THY alone and HEPES, pH 7.5), the growth curves of the isogenic ΔrocA mutant and parental WT strains were nearly superimposable (Fig. 2A and 4C). Consistent with our hypothesis, when grown under acidic conditions (MES, pH 6.0), the isogenic ΔrocA deletion mutant strain had a shortened lag phase and increased slope of the exponential phase compared to the parental WT strain (Fig. 4D). After 3 h of growth under acidic conditions, significantly more CFU were present in cultures of the isogenic ΔrocA deletion mutant strain than in those of the parental WT strain (Fig. 4E).

Deletion of rocA results in differential transcript levels of multiple GAS virulence factors.

Several proven and putative virulence factors had significantly altered transcript levels in the isogenic ΔrocA deletion mutant strain compared to the parental WT strain (Table 2). Many of the differentially expressed virulence factors are known to be regulated by the CovRS two-component system, as well as RocA, in other GAS serotypes (24, 28, 32). Selected virulence factor genes with increased transcript levels in the isogenic ΔrocA mutant strain include nga (encoding NAD⁺-glycohydrolase [SPN]) (39), slo (encoding streptolysin O [SLO]) (39), spyCEP (encoding interleukin-8 [IL-8] protease) (76), mac (encoding an IgG endopeptidase and an inhibitor of reactive oxygen species generation) (77–79), and sse (encoding streptococcal secreted esterase [SSE]) (80). Selected virulence factor genes with decreased transcript levels in the isogenic ΔrocA strain include M28_Spy0109 (encoding pilin protein) (50, 81), the sag operon (carrying streptolysin S biosynthesis genes) (82), grab (encoding protein-G-related α₂-macroglobulin binding protein) (83), ska (encoding streptokinase [SKA]) (84), and speB (encoding streptococcal cysteine protease B) (85).

TABLE 2.

Selected proven and putative virulence factors of GAS regulated by rocA at mid-exponential and early stationary growth phases

Locus taga	Gene	Function	Fold change relative to WTb
			ME	ES
M28_Spy0109	M28_Spy0109	Pilin protein	−2.8
M28_Spy0137	nga	NAD+-glycohydrolase	6.3	7.6
M28_Spy0139	slo	Streptolysin O	6.0	7.2
M28_Spy0329	spyCEP	IL-8 protease	38.0
M28_Spy0540	sagA	Streptolysin S precursor	−2.7
M28_Spy0541	sagB	Streptolysin S biosynthesis protein	−3.4
M28_Spy0542	sagC	Streptolysin S biosynthesis protein	−3.2
M28_Spy0543	sagD	Streptolysin S biosynthesis protein	−3.5
M28_Spy0544	sagE	Streptolysin S self-immunity protein	−3.0
M28_Spy0545	sagF	Streptolysin S biosynthesis protein	−2.5
M28_Spy0546	sagG	Streptolysin S export ATP binding protein	−3.1
M28_Spy0547	sagH	Streptolysin S export transmembrane protein	−2.9
M28_Spy0548	sagI	Streptolysin S export transmembrane protein	−2.7
M28_Spy0649	mac	IgG endopeptidase and inhibitor of reactive oxygen species generation	38.1
M28_Spy1098	grab	Protein G-related α2-macroglobulin binding protein	−5.5	−8.7
M28_Spy1450	sse	Streptococcal secreted esterase	3.6	1.7
M28_Spy1672	ska	Streptokinase	−2.2	−7.4
M28_Spy1675	sclA	Collagen binding protein	19.1	4.1
M28_Spy1699	fba	Fibronectin binding protein	4.1	2.3
M28_Spy1700	scpA	C5a peptidase	4.1	3.7
M28_Spy1701	enn	IgA binding protein	1.6
M28_Spy1702	emm	Antiphagocytic M protein	3.5	6.0
M28_Spy1703	mrp	M-related protein		2.3
M28_Spy1715	sfbX	Fibronectin binding protein	3.4	3.0
M28_Spy1716	sof	Serum opacity factor	3.0	3.4
M28_Spy1721	speB	Streptococcal cysteine protease B		−3.2
M28_Spy1884	hasA	Hyaluronan synthase	17.2
M28_Spy1885	hasB	UDP-glucose 6-dehdrogenase	18.8
M28_Spy1886	hasC	UTP-glucose-1-phospate uridylyltransferase	17.7

^aLocus tag identified in the serotype M28 reference genome MGAS6180.

^bME, mid-exponential growth phase; ES, early stationary growth phase. Empty (blank) cells, the gene does not satisfy the P value and/or fold change requirement.

To assess the phenotypic effect of the differential transcript levels for selected virulence factors, a series of in vitro assays was performed. Compared to the parental WT strain, the isogenic ΔrocA deletion mutant strain expressed increased amounts of immunoreactive SPN and SLO proteins (Fig. 5A). Additionally, compared to the parental WT strain, the isogenic ΔrocA mutant deletion strain had significantly increased SOF, SPN, SPN, SLO, and SSE secreted activity (Fig. 3B and 5B to D) and significantly decreased SKA secreted activity (Fig. 5E). The results are consistent with the RNA-seq data and raise the possibility that deletion of rocA in M28 GAS increases virulence.

FIG 5.

Deletion of rocA significantly increases GAS virulence factor levels and activity in the culture supernatant. (A) Western immunoblot analysis of NAD⁺-glycohydrolase (SPN) and streptolysin O (SLO). (B) SPN activity. (C) SLO activity. (D) Platelet activating factor (PAF) acetylhydrolase activity. (E) SKA activity. Data are shown as mean ± standard deviation. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (Student's t test).

Deletion of rocA results in increased virulence in a mouse model of necrotizing myositis.

Next, we hypothesized that deletion of rocA significantly increases M28 GAS virulence. To test this hypothesis, the virulences of the isogenic ΔrocA deletion mutant and parental WT strains were compared using a well-established mouse model of necrotizing myositis (37, 86, 87). Compared to the parental WT strain, the isogenic ΔrocA deletion mutant strain caused significantly more mortality (Fig. 6A) and larger lesions with more tissue destruction (Fig. 6B). Also, compared to infection with the parental WT strain, significantly more CFU were recovered from mouse limbs infected with the isogenic ΔrocA deletion mutant strain (Fig. 6C). Together, these data demonstrate that deletion of rocA in serotype M28 GAS significantly increases virulence.

FIG 6.

Deletion of rocA increases GAS virulence in a mouse model of necrotizing myositis. (A) Kaplan-Meier survival curve (n = 40 mice/strain). *, P < 0.05 (log rank test). (B) Representative microscopic lesions from the inoculation site of the right lower hindlimb of mice infected with the WT or ΔrocA mutant strain on day 1 postinoculation. The necrotic lesions are encompassed by black ovals. Original magnification, ×4. (C) CFU/gram of tissue recovered on day 3 postinoculation (n = 20 mice/strain). Data are shown as mean ± SEM. *, P < 0.05 (Mann-Whitney test).

DISCUSSION

Serotype M28 GAS strains are a common cause of pharyngeal and invasive infections globally (1, 5–11). However, relatively little is known about the molecular pathogenesis of M28 strains. GAS molecular pathogenesis studies have historically used other numerically important serotypes such as M1 and M3 strains as model organisms (2). Recent whole-genome sequence analysis of 2,101 serotype M28 GAS strains recovered from large population-based studies of patients with invasive infections revealed an unexpectedly high number of missense and nonsense polymorphisms in rocA (GenBank accession no. MH884522 to MH884551) (Fig. 1), a finding that was not observed in our previous studies with M1, M59, and M89 GAS (15, 37, 38). Inactivation of rocA in a genetically representative serotype M28 GAS strain resulted in a substantial transcriptome change (38.8% of all GAS genes) (Fig. 2), including many genes encoding transcription regulators and proven or putative virulence factors. In vitro assays confirmed the RNA-seq results (Fig. 3 to 5), and the M28 RocA-inactivated strain was significantly more virulent in a mouse model of necrotizing myositis (Fig. 6). Taken together, these data suggest that RocA plays a key role in M28 GAS molecular pathogenesis in human invasive infections (Fig. 7).

FIG 7.

Model of RocA contribution to the molecular pathogenesis of serotype M28 GAS. GAS strains with a wild-type rocA gene (such as MGAS28426, left panel) have a basal level of rocA expression, RocA-regulated genes, and a wild-type virulence phenotype. GAS strains with rocA mutations (such as ΔrocA, right panel) have a substantially altered transcriptome that significantly increase virulence factor gene expression, stress response, and virulence.

We found that deletion of rocA in a genetically representative serotype M28 GAS strain resulted in a very substantial transcriptome change (Fig. 2; see Tables S1 and S2 in the supplemental material). Previous research on RocA has led to the publication of RocA transcriptomes from serotype M1 and M3 GAS strains (28, 32). The time point for collecting GAS cells, the culture media used, and the process of making RNA-seq libraries for the published M1 and M3 studies were very similar to those for our M28 RocA RNA-seq experiment (28, 32). However, the bioinformatic processes used for analysis in each study differed. Thus, to compare the three transcriptomes, we reanalyzed the publicly available M1 (accession number GSE97325) (32) and M3 (accession number GSE68277) (28) RocA RNA-seq data using a bioinformatics process identical to that for our M28 ME RocA RNA-seq data (see Materials and Methods). In contrast to the M1 and M3 RocA transcriptomes, we discovered a much higher number of genes directly or indirectly regulated by RocA in M28 GAS (see Table S3 and Fig. S2 in the supplemental material). At ME growth, 427 genes have significantly altered transcript levels in the serotype M28 isogenic ΔrocA deletion mutant strain, whereas only 357 and 224 genes have significantly altered transcript levels in the rocA deletion M1 and M3 strains, respectively (Table S3 and Fig. S2). The substantially increased size of the M28 RocA regulon compared to those of the M1 and M3 RocA regulons may be due, in part, to the number of genes encoding transcription regulators that were differentially expressed (Fig. S2). Of the 41 transcription regulators directly or indirectly regulated by RocA in serotype M28 GAS (Table 1), 26 had altered transcript levels at mid-exponential growth. In comparison, RocA altered the expression of only 24 and 15 transcription regulators in the M1 and M3 strains, respectively. That is, RocA may regulate the expression of more genes in M28 GAS due to its effect on many different transcription regulators. Another possible explanation for the much smaller RocA transcriptome in M3 GAS is the experimental strategy used. The M1 and M28 RocA transcriptomes were determined by creating isogenic deletion mutant strains lacking the rocA gene (32). In contrast, the M3 study compared an M3 GAS strain containing the naturally occurring rocA mutation to an isogenic strain containing the serotype M1 wild-type rocA allele (28). The naturally occurring RocA mutant protein in M3 GAS has partial RocA function when overexpressed, suggesting that M3 strains may natively retain some very limited amount of RocA activity (28, 32).

The RocA transcriptome has not been previously studied at early stationary growth phase (28, 32). In most GAS RNA-seq studies, regulators typically alter the expression of more genes at early stationary growth than at mid-exponential growth (24, 88–91). However, in M28 GAS, we discovered that rocA deletion significantly altered the transcript levels of fewer genes at early stationary growth phase (323 compared to 427) (Fig. 2C). Analysis of rocA expression at both growth phases identified significantly more transcripts at mid-exponential growth phase (see Fig. S3 in the supplemental material), suggesting that RocA is expressed at a higher level during early growth. Additionally, more genes encoding transcription regulators had significantly altered transcript levels at mid-exponential growth (26 at mid-exponential growth phase compared to 19 at early stationary growth phase). The two findings may, in part, explain the higher number of genes directly or indirectly regulated by RocA at mid-exponential growth phase compared to early stationary growth phase.

One possible molecular mechanism for RocA inactivation to increase M28 GAS virulence is by increasing CovR phosphorylation via the CovRS two-component regulatory system. Spontaneous mutations in covR and covS have been identified in GAS strains recovered from animals with experimental infections and humans with invasive infections (23, 92). In general, mutations in covR or covS substantially alter the GAS transcriptome and increase strain virulence (23, 24, 93). That is, covRS mutations may provide a fitness advantage in some host environments. We speculate that rocA mutations may also be selected in vivo after the initial infection is established. Although covR and covS expression was not significantly altered in the M28 isogenic rocA deletion mutant strain (Tables S1 and S2), RocA increases CovR phosphorylation, which is a key step in CovRS activation (28, 30). In support, approximately 15% of the genes that are differentially expressed in the ΔrocA M28 strain are known to be regulated by CovRS in other GAS serotypes (Tables S1 and S2) (24). For example, RocA inactivation significantly increased nga and slo expression (Fig. 5). CovRS is known to regulate nga and slo (23, 24). Possibly important to the role of RocA inactivation in M28 GAS virulence, increased expression and activity of SPN and SLO was recently implicated as the central factor underlying the emergence and global spread of epidemic clade 3 M89 strains (37, 94).

Another possible molecular mechanism for RocA inactivation to increase M28 GAS virulence is by significantly altering gene expression independently of CovRS (Table 2). That is, RocA may directly or indirectly regulate or act as an accessory protein to additional regulatory systems in GAS. For example, the isogenic ΔrocA deletion mutant strain had significantly increased transcript levels of mga and Mga-regulated genes (Fig. 3). Sanson et al. demonstrated that a naturally occurring mutation in mga increased the expression of mga and Mga-regulated genes in M59 GAS to significantly increase virulence in mice, nonhuman primates, and humans (95–97). In the rocA-inactivated serotype M1 and M3 GAS strains, mga and some genes in the Mga regulon also had increased expression (Table S3). Possibly important to the serotype-specific effect of RocA on mga expression, the genes comprising the Mga regulon differ in each of the three serotypes (57). Similarly, the stress response genes ciaHR, nrdR, and spxA2 and the arginine deiminase pathway genes arcABCD had significantly increased transcript levels in the ΔrocA M28 mutant strain (Fig. 4) (44, 48, 61, 74, 75). Consistent with the RNA-seq data, the isogenic ΔrocA strain had increased resistance to acidic stress, a condition encountered in purulent lesions (70–72). The expression of spxA2 was increased in the rocA-inactivated serotype M1 and M3 GAS strains (Table S3). In comparison, the arcABCD genes had significantly decreased expression in the rocA-inactivated serotype M1 GAS strain, whereas the arcABCD genes had significantly increased expression in the rocA-inactivated serotype M3 and M28 GAS strains (Table S3). The molecular basis for the serotype-specific regulatory activity of RocA on arcABCD is unknown.

Unexpectedly, the M28 ΔrocA isogenic deletion mutant strain had significantly decreased transcript levels of ska (Tables S1 and S2). The ska gene encodes streptokinase (SKA), an important virulence factor that disrupts the host fibrinolytic system (84, 98). SKA also leads to degradation of host extracellular matrix and basement membrane proteins. Whereas ska has significantly decreased transcript levels in the M28 ΔrocA strain, it had significantly increased transcript levels in the M1 and M3 RocA-inactivated strains (Table S3) (28, 32). Compared to that in the WT strain, SKA activity was significantly decreased in the M28 ΔrocA strain (Fig. 5E). The reason for the observed serotype-specific differences in regulation of ska is uncertain. To date, ska is known to be regulated by two different systems, FasBCAX and CovRS (43, 99–101). We detected no significant difference in expression of fasBCA or covRS in the M28 isogenic ΔrocA deletion mutant strain under the conditions studied. Our RNA-seq protocol does not capture small RNA transcripts, so fasX, a small RNA, could not be measured. One possible explanation for the serotype-specific regulation of ska is that one of the 41 differentially expressed transcription regulators, including the 14 regulators that are uniquely regulated by RocA in M28 GAS (Table 1 and Fig. S2), directly or indirectly regulates ska expression. Further studies will be needed to better understand the molecular mechanisms underlying altered expression of ska.

The difference in rocA allele frequency between M28 and other M protein serotype strains may be due to inherent genetic differences that favor a RocA-inactivated phenotype. For example, most serotype M1 and M59 GAS strains are encapsulated (15, 38), but virtually all M28 and epidemic clade 3 M89 GAS are capsule deficient (94, 102). Epidemic clade 3 M89 GAS strains lack the hasABC locus, which is required for hyaluronic acid capsule production (94). M28 GAS strains have a one-nucleotide frameshifting insertion in hasA that results in a truncated HasA protein without enzymatic activity (GenBank accession no. MH884522 to MH884551) (102). Thus, although the M28 rocA isogenic deletion mutant strain demonstrated significantly increased hasABC transcript levels (Table 2), virtually all M28 strains are incapable of capsule biosynthesis (see Fig. S4 in the supplemental material). Additionally, virtually all serotype M28 GAS strains have a missense mutation in mac, the gene encoding Mac, an IgG endopeptidase and inhibitor of reactive oxygen species generation (77–79). The missense mutation results in a loss of Mac IgG endopeptidase activity (78). The two phenotypic characteristics, possibly in combination with other, yet-unrecognized genomic factors, may predispose serotype M28, and also M89, GAS strains to select for rocA polymorphisms that increase fitness in the invasive infection environment. Additionally, the preponderance of missense mutations in the transmembrane domains of RocA in serotype M28 GAS compared to the other serotypes is of particular interest (Fig. S1). As the transmembrane domains are crucial for the regulatory activity of RocA, detailed analyses of the missense mutations located in the transmembrane domains are warranted.

In summary, our study shows that in serotype M28 GAS, RocA directly or indirectly regulates a substantial portion of the GAS transcriptome (38.8% of all GAS genes), including many transcription regulators and proven or putative virulence factors. The number of genes and transcription regulators directly or indirectly regulated by RocA in serotype M28 GAS is greater than that observed in RocA studies performed with serotype M1 and M3 GAS. The M28 RocA-inactivated strain was significantly more virulent in a mouse model of necrotizing myositis. Taken together, these data suggest that RocA plays a key role in M28 GAS molecular pathogenesis and thus may contribute to the high number of naturally occurring polymorphisms found in M28 strains recovered from human invasive infections. Our findings underscore the critical need for molecular pathogenesis research efforts to study many different microbial strains from many different genetic backgrounds (i.e., within the same serotype and across different serotypes).

MATERIALS AND METHODS

Determination of SNPs in rocA in serotype M28 GAS strains.

Whole-genome sequence analysis of the 2,101 serotype M28 GAS strains was previously performed (GenBank accession no. MH884522 to MH884551). Six strains are distant outliers and were excluded from further analysis. Among the remaining 2,095 M28 strains, 20,135 single nucleotide polymorphism (SNP) sites were identified across the core genome that spans 1,735,432 bp. Assuming that SNP sites are randomly distributed across the core genome, 1 SNP site is expected to occur every 86 bp. Given a random distribution, approximately 16 SNP sites are expected to occur in the rocA coding and upstream regulatory region that spans 1,374 bp. As previously described, Fisher's exact test was used to demonstrate that significantly more SNP sites were identified in rocA than would be expected to occur by random chance (17, 37, 59).

Construction of an isogenic rocA deletion strain.

Strain MGAS28426 was selected as the serotype M28 parental wild-type (WT) strain because it is genetically representative of serotype M28 GAS strains and has a wild-type allele for all major global transcription regulators. The isogenic rocA deletion (ΔrocA) mutant strain was constructed by deleting the entire open reading frame of rocA, as previously described (39). Sequences for primers used are listed in Table S4 in the supplemental material. Whole-genome sequence analysis of the isogenic ΔrocA mutant strain confirmed that no spurious mutations were introduced during strain construction.

RNA-seq analysis.

The WT and isogenic ΔrocA mutant strains were grown in triplicate in THY with 5% CO₂ at 37°C to mid-exponential (ME) (OD₆₀₀ = 0.5) and early stationary (ES) (OD₆₀₀ = 1.65) growth phases as previously described (95, 96). RNAprotect Bacteria Reagent (Qiagen Inc., Germantown, MD) was added (2:1), and cells were lysed by ballistic disintegration (FastPrep-96 instrument and lysing matrix B [MP Biomedicals, Santa Ana, CA]). RNA was extracted using standard methods (RNeasy kit [Qiagen Inc., Germantown, MD]), and RNA quality and quantity were assessed (Agilent 2100 Bioanalyzer [Agilent Technologies, Santa Clara, CA] and Qubit [Invitrogen, Carlsbad, CA]). RNA-seq libraries were prepared using standard methods (ScriptSeq Complete kit [Epicentre, Madison, WI]) and sequenced with an Illumina NextSeq instrument (Illumina, San Diego, CA) using the default settings.

On average, we obtained 39.5 million reads/sample for the 12 samples (WT and isogenic ΔrocA mutant strains grown in triplicate and collected at ME and ES growth phases). Reads were mapped to the genome of serotype M28 GAS reference strain MGAS6180 (12), and differential transcript analysis was performed with CLC Genomics Workbench 10.5 (Qiagen Inc., Germantown, MD) using the default settings. Genes encoding rRNA, tRNA, phage, and mobile genetic elements were excluded from analysis, as there are limitations in read mapping to repetitive sequences found within the aforementioned elements. Genes with an absolute transcript change of ≥1.5-fold and a P value of <0.05 after Baggerly's test with Bonferroni's correction for multiple comparisons were considered to be significantly differentially expressed.

RocA transcriptomes from M1 and M3 GAS strains have been published (28, 32). The time point for collecting the GAS cells at ME growth phase, the culture media used, and the process of making RNA-seq libraries in the published M1 and M3 studies were very similar to those used in our M28 RocA RNA-seq experiment conducted at ME growth phase. However, the bioinformatics processes used for analysis in each study differed. To compare the M1, M3, and M28 RocA transcriptomes, publicly available M1 and M3 RNA-seq sequencing data (28, 32) were reanalyzed with a bioinformatics process identical to that described above. That is, to compare the RocA transcriptomes of the M1, M3, and M28 GAS strains, we used a common bioinformatics process to analyze the three RNA-seq data sets. Briefly, the publicly available RNA-seq reads for the M1 (accession number GSE97325) (32) and M3 (accession number GSE68277) (28) GAS strains were downloaded and mapped to the serotype M1 MGAS5005 (103) and M3 MGAS315 (104) reference genomes, respectively. Differential transcript analysis was performed with CLC Genomics Workbench 10.5 (Qiagen Inc., Germantown, MD) using the default settings. Genes encoding rRNA, tRNA, phage, and mobile genetic elements were excluded, as there are limitations in the read mapping to repetitive sequences found within the aforementioned elements. Genes not present in the M28 reference strain also were excluded from analysis. Genes with an absolute transcript change of ≥1.5-fold and a P value of <0.05 after Baggerly's test with Bonferroni's correction for multiple comparisons were considered to be significantly differentially expressed. The differentially expressed genes were then compared across the three strains.

SOF activity assay.

Serum opacity factor (SOF) activity in the culture supernatants was assayed as previously described (69), with the modification that samples were serially diluted 2-fold in phosphate-buffered saline (PBS) with 1% sodium dodecyl sulfate (SDS) before incubation with horse serum (1:10 sample-to-horse-serum volume). PBS with 1% SDS was used as a negative control. Dilutions were determined to be positive for serum opacity factor activity at an absorbance at 405 nm of greater than 0.8 (105). Mean titers of four biological replicates were compared using Student's t test (Prism 7 [GraphPad, La Jolla, CA]), with a P value of <0.05 considered to be statistically significant.

Growth under acidic conditions.

For growth under acidic conditions, THY supplemented with 0.1 M 2-(N-morpholino)ethanesulfonic acid (MES) (pH 6.0) (Sigma-Aldrich, St. Louis, MO) was used. THY supplemented with 0.1 M HEPES (pH 7.5) (Sigma-Aldrich, St. Louis, MO) was used to determine the baseline effects of a buffered medium on GAS strain growth (29). For CFU, cultures grown in quadruplicate were harvested at 3 h, serially diluted 10-fold, and plated onto THY agar supplemented with 5% sheep blood. Colonies were counted after incubation overnight. The mean CFU of four biological replicates were compared using the Mann-Whitney test (Prism 7 [GraphPad, La Jolla, CA]), with a P value of <0.05 considered to be statistically significant.

Western immunoblot analysis of SPN and SLO in culture supernatant.

GAS strains were grown to mid-exponential growth phase and pelleted by centrifugation, and supernatants were assessed for the presence of immunoreactive NAD⁺-glycohydrolase (SPN) and streptolysin O (SLO) as described previously (39).

SPN and SLO activity assay.

GAS strains were grown to mid-exponential growth phase and pelleted by centrifugation, and supernatants were assayed for NAD⁺-glycohydrolase (SPN) and streptolysin O (SLO) activity as previously described (86). The mean titers (SPN) and activities (SLO) of three biological replicates were compared using Student's t test (Prism 7 [GraphPad, La Jolla, CA]), with P value of <0.05 considered to be statistically significant.

PAF acetylhydrolase activity assay.

Streptococcal secreted esterase (SSE) is a known secreted GAS virulence factor that hydrolyzes platelet-activating factor (PAF) (80). Thus, SSE activity in culture supernatants collected at mid-exponential growth phase was assayed with the PAF acetylhydrolase assay kit (Cayman Chemical, Ann Arbor, MI), according to the manufacturer's instructions with minor modifications (31). GAS strains were grown in triplicate in THY to mid-exponential growth phase. Supernatants were collected by centrifugation at 4,000 rpm for 10 min. The supernatant (10 μl) was added to assay buffer 2 (5 μl) and incubated with 2-thio-PAF (200 μl) at room temperature for 30 min. 5,5-Dithio-bis(2-nitrobenzoic acid) (10 μl) was added to each sample, mixed, and incubated for 1 min at room temperature. The absorbance of each sample at 412 nm was measured and used to calculate the PAF acetylhydrolase activity. Fresh THY was used as a negative control. The mean activities of three biological replicates were compared using Student's t test (Prism 7 [GraphPad, La Jolla, CA]), with a P value of <0.05 considered to be statistically significant.

SKA activity assay.

GAS strains were grown to mid-exponential and early stationary growth phases and pelleted by centrifugation, and cell-free supernatants were assayed for streptokinase (SKA) activity as previously described (100). Activity was determined as the change in absorbance over time from the initial time point to maximum absorbance at 405 nm. The mean relative activities of three biological replicates were compared using Student's t test (Prism 7 [GraphPad, La Jolla, CA]), with P value of <0.05 considered to be statistically significant.

Mouse model of necrotizing myositis.

Mouse necrotizing myositis studies were performed as previously described (37, 86, 87). Immunocompetent 4-week-old female CD1 mice (Envigo Laboratories, Houston, TX) were randomly assigned to treatment groups and inoculated in the right lower hindlimb with 5 × 10⁸ CFU of each bacterial strain in 100 μl PBS (n = 40 mice/strain). Mice were monitored at least once daily, and mortality was determined using internationally recognized criteria (106). Survival was compared using the log rank test (Prism 7 [GraphPad, La Jolla, CA]), with a P value of <0.05 considered statistically significant. For gross and histopathological evaluation, mice were sacrificed on day 1 postinoculation, and limbs were processed by standard methods (87). For quantitative culture (n = 20 mice/strain), mice were sacrificed on day 3 postinoculation. Infected limbs were amputated, placed in tared tubes containing sterile PBS, weighed, and homogenized. Tenfold serial dilutions were grown overnight on THY agar supplemented with 5% sheep blood, and CFU were counted. Mean CFU were compared by the Mann-Whitney test (Prism 7 [GraphPad, La Jolla, CA]), with a P value of <0.05 considered statistically significant. All animal experiments were approved by the Institutional Animal Care and Use Committee of the Houston Methodist Research Institute. Sample sizes for each experiment were determined using a power calculation.

Hyaluronic acid capsule assay.

GAS strains were grown in triplicate in THY to mid-exponential growth phase. Cells were collected from 10 ml of culture by centrifugation at 4,000 rpm for 10 min, and the supernatant was discarded. The cells were resuspended in a 1:1 (vol/vol) mixture of water and chloroform (1 ml) and vortexed for 30 s. The mixture was centrifuged (13,200 rpm, 10 min), and the resulting aqueous phase was used to assay for the presence of hyaluronic acid capsule using the hyaluronic acid quantitative test kit (Corgenix, Broomfield, CO) according to the manufacturer's instructions. MGAS2221, a serotype M1 GAS strain whose capsule production is well documented (39), was used as a positive control.

Accession number(s).

The serotype M28 RNA-seq sequence data have been deposited at the National Center for Biotechnology Information (NCBI) under BioProject no. PRJNA470894.