Streptococcus pyogenes upregulates arginine catabolism to exert its pathogenesis on the skin surface

The arginine deiminase (ADI) pathway has been found in many kinds of bacteria and functions to supplement energy production and provide protection against acid stress. The Streptococcus pyogenes ADI pathway is upregulated upon exposure to various environmental stresses, including glucose starvation. However, there are several unclear points about the advantages to the organism for upregulating arginine catabolism. We show that the ADI pathway contributes to bacterial viability and pathogenesis under low-glucose conditions. S. pyogenes changes global gene expression, including upregulation of virulence genes, by catabolizing arginine. In a murine model of epicutaneous infection, S. pyogenes uses the ADI pathway to augment its pathogenicity by increasing the expression of virulence genes, including those encoding the exotoxins. We also find that arginine from stratum-corneum-derived filaggrin is a key substrate for the ADI pathway. In summary, arginine is a nutrient source that promotes the pathogenicity of S. pyogenes on the skin.

In brief Hirose et al. show that S. pyogenes upregulates arginine catabolism to exert its pathogenesis under low-glucose conditions. S. pyogenes changes global gene expression, including upregulation of virulence genes, by catabolizing arginine. S. pyogenes acquires arginine from stratum-corneum-derived filaggrin to adapt to glucose starvation on the skin.

INTRODUCTION
One of the most important human bacterial pathogens of skin is Streptococcus pyogenes, which can produce superficial impetigo or more deep-seated cellulitis but also more severe invasive infections such as sepsis, necrotizing fasciitis, and streptococcal toxic shock syndrome (Cunningham, 2000;Walker et al., 2014). S. pyogenes typing based on M protein and T antigen (pilus major subunit) antigenicity (Falugi et al., 2008) confirms several serotypes are capable of causing severe infections, but in recent decades, one subclone of the M1 serotype, the globally disseminated clonal M1T1 clone (Chatellier et al., 2000), has persisted uninterruptedly as the most frequently isolated S. pyogenes strains from both invasive and noninvasive infections (Lynskey et al., 2019;Walker et al., 2014).
Human skin is an inhospitable environment that includes the physical barrier of the stratum corneum (SC), an acidic surface pH, and the active synthesis of innate defense factors such as antimicrobial peptides, proteases, lysozymes, cytokines, and chemokines that together serve to recruit immune cells and prime adaptive immune responses (Cogen et al., 2008;Proksch, 2018). To successfully colonize or establish infection in the skin, pathogens must possess virulence determinants for evasion of these immune factors and also for acquisition of nutrients, whose availability the host may restrict under the concept of nutritional immunity. Elucidation of such bacterial metabolic pathways that are essential for in vivo survival can reveal unique targets for novel therapeutics.
The arginine deiminase (ADI) pathway is a metabolic pathway found in many kinds of bacteria that serves to supplement energy production and provide protection against acid stress in vitro (Abdelal, 1979;Cotter and Hill, 2003). The ADI pathway of S. pyogenes has been studied in a limited manner and determined to antagonize nitric oxide (NO) production by macrophages and contribute to the asymptomatic colonization of the murine vaginal mucosa (Cusumano et al., 2014). The S. pyogenes ADI pathway is negatively regulated by the following three virulence-related transcriptional control systems: the control Report ll of virulence (CovRS) two-component gene regulatory system, catabolite control protein CcpA, and regulator gene of glucosyltransferase Rgg (Dmitriev et al., 2006;Shelburne et al., 2010). CovRS mediates a general stress response to changing temperature, pH, and osmolarity (Dalton and Scott, 2004); and CcpA and Rgg are directly linked to environmental glucose deprivation (Dmitriev et al., 2006;Shelburne et al., 2008). Although CovR deletion induces a several-fold increase of ADI pathway genes, CcpA or Rgg deletions markedly induce several-log-fold increases in the pathway (Dmitriev et al., 2006;Shelburne et al., 2010). These findings suggest that the ADI pathway is especially important to S. pyogenes under low-glucose conditions. Notably, the glucose concentration in the SC of skin is much lower than that present in blood (Sylvestre et al., 2010). In this study, we investigated whether the S. pyogenes ADI pathway contributes to the viability and pathogenesis of an M1T1 strain under low-glucose conditions and in the skin.

S. pyogenes ADI pathway contributes to its viability and virulence
The S. pyogenes ADI pathway is comprised of ArcA, B, C, and D (Figure S1A;Cusumano et al., 2014). ArcA, an arginine deiminase, is the first enzyme of the ADI pathway and catalyzes the irreversible hydrolysis of arginine to citrulline and ammonia. We constructed a markerless complete arcA deletion mutant (DarcA) by using a double crossover homologous recombination technique, preserving as a control a wild-type (WT) revertant strain (Wr) from the single crossover step back to the integrity of arcA. We first observed the time course of pH change of bacterial cultures grown under arginine-rich conditions, using phenol red as an indicator of elevated pH ( Figure 1A). The WT S. pyogenes parent strain induced pH elevation in the culture medium beginning in stationary phase, whereas the S. pyogenes DarcA did not, confirming that a loss of arcA renders the bacterium's arginine catabolism dysfunctional. Deletion of arcA led to a decrease in S. pyogenes viability during the decline phase of stationary growth in Todd-Hewitt broth supplemented with 0.2% yeast extract (THY broth) ( Figure 1B). In contrast, the WT S. pyogenes strain exhibited a strong increase in arcA gene expression and ammonium ion production occurring during stationary phase following glucose starvation ( Figures 1C, 1D, and 1E). Although a pH elevation was not detected during the stationary phase in THY broth, pH levels in phenol red broth supplemented with arginine were elevated above neutral pH for both the WT and Wr cultures (Figures S1B and S1C). Both WT and Wr showed the potential for long-term survival in neutral pH for at least 40 days during the decline phase, whereas the mutant strain lost viability after only 2 days ( Figure S1B). These findings indicate that S. pyogenes neutralizes excess protons during stationary phase by synthesizing ammonia in an ADI-dependent manner, thus promoting long-term viability during the decline phase.
S. pyogenes cultured in THY broth experienced glucose starvation beginning in stationary phase. To investigate whether arginine catabolism contributed to S. pyogenes virulence phenotypes under such glucose-starved conditions, we prepared chemically defined medium (CDM ; Table S1) without glucose, phenol red, or arginine. As a first surrogate marker of virulence, we evaluated S. pyogenes cytotoxicity against the cultured human keratinocyte cell line HaCaT. We supplemented 1,000 mM arginine in the CDM to match the physiological concentration present in human muscle tissue (Canepa et al., 2002). At 20 h after infection of HaCaT cells under arginine-supplemented conditions, WT and Wr showed a dose-dependent increase in cytotoxic potential compared to the DarcA mutant ( Figures  1F and S1D), linking arginine catabolism to this virulence phenotype.
By metabolizing arginine, S. pyogenes acquires adenosine triphosphate (ATP) as an energy source, coupled to discharge of ammonium ion ( Figure S1A). Intracellular acidification may affect the growth and cell viability of streptococci (Dashper and Reynolds, 2000). We probed the contribution of arginine catabolism to S. pyogenes intracellular pH and ATP levels under low-glucose conditions using CDM. No effects on the intracellular pH of S. pyogenes were observed 1 h, 5 h, and 15 h postincubation in CDM (Figure S1E). Conversely, at 1 h and 5 h post-incubation, arginine catabolism contributed to the acquisition of ATP under the no-glucose condition (Figure 1G), which correlated with viability of S. pyogenes at 5 h and 15 h post-incubation in CDM ( Figure 1H). S. pyogenes WT in CDM at 15 h had greater viability when cultured in the presence of HaCaT cells than in their absence (Figure 1H), suggesting that ADI-dependent S. pyogenes cytotoxicity induces release of nutrition from damaged keratinocytes, promoting S. pyogenes survival. (B) Bacterial growth ability and viability in THY broth. Error bars indicated the mean + SE (n = 4). Representative data obtained from at least three independent experiments are shown. **p < 0.01. (C) The arcA expression of WT S. pyogenes at each growth phase within THY broth. Data obtained by combining three independent experiments are displayed with a box-whisker plot (n = 9). **p < 0.01. (D and E) Glucose (D) and ammonium ion (E) levels in culture supernatant of WT (n = 9) and DarcA (n = 4) during growth in THY broth. Error bars indicated the mean + SE. **p < 0.01. (F) Arginine-dependent cytotoxicity of S. pyogenes. HaCaT cells cultured in CDM were infected with S. pyogenes (MOI = 500) for 20 h. Viable cells and dead cells are indicated green and red, respectively. LDH, lactate dehydrogenase. Values are presented as the mean of four wells from one of three independent experiments. Error bars indicated the mean + SE. **p < 0.01. p < 0.01. (G) Intracellular ATP levels of S. pyogenes. Representative data obtained from at least three independent experiments are shown. Vertical lines represent the mean + SE (n = 4). **p < 0.01. (H) Bacterial viability in only CDM or CDM with cultured HaCaT cells. Error bars indicated the mean + SE (n = 4). Representative data obtained from at least three independent experiments are shown. **p < 0.01.
Cell Reports 34, 108924, March 30, 2021 3 Report ll OPEN ACCESS S. pyogenes arginine catabolism changes global gene expression We conducted RNA sequencing (RNA-seq) analysis of the S. pyogenes strains in CDM with arginine (Arg(+)) or without arginine (Arg(À)) to assesses transcriptional consequences of altering the ADI pathway. Principal-component analysis (PCA) showed DarcA Arg(À), DarcA Arg(+), and WT Arg(À) samples clustered together, whereas WT Arg(+) samples positioned clearly apart from the others (Figure 2A). Differentially expressed genes (DEGs) from comparisons of the Arg(+) and Arg(À) groups were detected only in the WT S. pyogenes background and not with the DarcA (Figure 2B; Data S1 and S2). The cumulative results indicated that arginine-dependent transcriptome changes in S. pyogenes occurred only in the presence of an intact ADI pathway. A heatmap using selected characteristic genes showed that upregulated genes in WT Arg(+) samples included those encoding virulence factors, such as an enzyme for maturation of the cytolysin streptolysin S (sagB), another cytolysin streptolysin O (slo), nucleases (spd3), and NADase (nga), but also genes comprising the pyrimidine biosynthesis pathway (pyrD, pyrE, and pyrF) ( Figure 2C). On the other hand, cell-division-associated genes (ftsH, ftsL, and ftsZ) and genes encoding F 0 F 1 -type ATP synthase (atpB-H) were downregulated in WT S. pyogenes in the presence of arginine.
The most well-studied S. pyogenes two-component signal transduction system is the cluster of virulence (Cov) intracellular responder (CovR)/extracellular sensor (CovS) CovRS, which influences 15%-20% of the S. pyogenes genome (Graham et al., 2002). CovS sensed environmental changes and transmitted to CovR by phosphorylation-dephosphorylation (Horstmann et al., 2015). Of particular interest, the covS gene was downregulated in WT S. pyogenes-sensing arginine ( Figure 2C; Data S2). To investigate whether arginine catabolism influences CovR phosphorylation, we monitored phosphorylation of the regulator using a 5 0 FLAG-CovR strain ( Figure 2D). Exponential phase growth of S. pyogenes in THY broth showed low levels of CovR phosphorylation. For S. pyogenes within CDM, the WT Arg(+) sample tended to show a low level of CovR phosphorylation, but no significant difference was found by the densitometric analysis ( Figure 2E).

ADI pathway contributes to the development of cutaneous lesions
The SC of skin is relatively deficient in glucose (Sylvestre et al., 2010), whereas arginine is abundant (Kubo et al., 2013). We hypothesized that S. pyogenes may use arginine to acquire energy on the skin surface and examined this possibility by using a mouse model of epicutaneous infection ( Figure 3A). By 3 days post-challenge, the skin surface of mice infected with WT or Wr S. pyogenes had peeled off, whereas it remained intact in DarcA-infected mice ( Figure 3B). Furthermore, CFU recovered from skin lesions 3 days post-infection were significantly reduced in DarcA-infected mice compared to the WT or Wr ( Figure 3C). An analysis of gene expression showed the WT S. pyogenes strain had higher in vivo expression of the genes encoding the streptolysin S precursor (sagA) and streptolysin O (slo) but reduced expression of the gene encoding cysteine protease SpeB (speB) compared to the DarcA mutant ( Figure 3D). These results indicate that arginine catabolism contributes to the pathogenesis of S. pyogenes on the skin surface while promoting the expression of cytolysins.
To assess the ADI-dependent effects on S. pyogenes systemic pathogenicity, we conducted intravenous challenge of mice along with an ex vivo bactericidal assay by using mouse blood. In both sets of experiments, there were no significant differences between WT and DarcA ( Figures 3E and 3F); this arginine catabolism may not be involved in the virulence or viability of S. pyogenes in blood. We speculated that arginine catabolism was suppressed in blood due to high concentrations of glucose. With that in mind, we evaluated arcA gene expression of WT S. pyogenes in blood and on the skin surface by using qPCR. Using an expression level of arcA in the exponential phase growth in THY medium set to 1, we found that arcA expression was 10 to 100 times lower in blood and 10 times greater on the skin than under the culture conditions ( Figure 3G). This result suggests that the expression on the skin surface was elevated 2-to 3log-fold compared to that in blood, where low-glucose concentrations are found. Expanding the repertoire of infection models, we found that in a subcutaneous mouse soft-tissue infection model, which causes necrosis of fascial tissue and adjoining muscle, the DarcA showed reduced virulence compared to WT and Wr ( Figure S2A), whereas there were no differences in virulence in another systemic challenge by intraperitoneal infection ( Figure S2B). It is speculated that these contrasting results reflect differences in available concentrations of glucose and arginine in localized versus systemic disease compartments.
Next, an epicutaneous infection study was conducted using streptozotocin-induced diabetic mice. As expected, 48 h after depilation of dorsal skin, glucose levels on the skin surface in non-diabetic (WT) mice were below the detection limit (<1 mM) in nearly all samples, whereas those in the diabetic mice were much higher (4 mM-70.4 mM; Figures S2C and S2D). In response to an epicutaneous infection in diabetic mice, DarcA showed comparable recovered CFU from the skin lesion to WT ( Figure S2E). Both WT and DarcA in CDM showed an upregulation of the slo gene in the presence of glucose (Figures S2F and S2G). These results suggested that the abundance of glucose on the skin surface in diabetes allows the pathogen to exhibit pathogenicity independently of arginine catabolism, consistent with the known increased risk of S. pyogenes skin and soft tissue infections in patients with diabetes (Lin et al., 2011).
In our RNA-seq data in vitro, 4 genes associated with phosphotransferase system (PTS) were upregulated in WT under arginine-rich conditions (ptsD, SP5448_05675, SP5448_08850, and SP5448_08855). Therefore, we evaluated the expression levels of these transporter genes at 3 h post-epicutaneous infection in WT mice ( Figure S3). The expression levels of each gene tended to be upregulated in WT compared to those of the DarcA mutant. This result also suggests that the skin surface of WT mice has a low glucose concentration and S. pyogenes upregulates the expression of transporter genes in a manner that depends on arginine catabolism.
ADI-pathway-dependent bacterial virulence was canceled in Flg À/À mice Arginine is abundant in filaggrin, an abundant protein within the SC of skin, and a filaggrin knockout (Flg À/À ) mouse has a markedly lower SC arginine level than a corresponding WT mouse (Kawasaki et al., 2012;Kubo et al., 2013). The degradation of filaggrin into amino acids occurs in the SC layers by host-derived enzymes, including caspase-14, calpain 1, and bleomycin hydrolase (Hoste et al., 2011). We hypothesized that S. pyogenes acquires arginine from filaggrin in the SC for nutritional requirements and its pathogenicity on the skin surface. At baseline, it is difficult to appreciate any phenotypic differences between WT mice and Flg À/À mice ( Figure 4A). However, following epicutaneous challenge assessed at 3 days post-infection, the previously observed difference of virulence between WT and DarcA disappeared in the skin of Flg À/À mice, as verified at three different challenge inocula ( Figure 4B). These results were corroborated with similar histopathology of the skin lesions at the 3-day post-infection time point (Figure 4C). The ArcA protein is associated with the S. pyogenes bacterial cell surface (Henningham et al., 2012). To confirm that S. pyogenes expresses ArcA during infection on the skin surface, immunofluorescence staining was performed at 24 h post-infection. Strong signals for ArcA expression were seen only in the case of S. pyogenes WT on the skin surface of WT mice (Figures 4D and 4E). In Flg À/À mice, S. pyogenes did not appear to upregulate ArcA for arginine catabolism to achieve infection.
S. pyogenes showed different expression levels of covS, arcA, and slo genes on the skin surface 3 h after epicutaneous infection in WT compared to Flg À/À mice ( Figure S4). Because Flg À/À mouse skin exhibits enhanced permeability of SC as compared to WT mouse skin (Kawasaki et al., 2012), these transcriptional differences might reflect differences of nutritional or stress environments.
In intestinal epithelial cells, it is thought that caspase-1 contributes to both pyroptosis and apoptosis during an inflammation (Lei-Leston et al., 2017). To investigate whether S. pyogenes caused programmed cell death of skin epithelial cells, immunofluorescence staining of caspase-1 was performed 24-h post-epicutaneous infection. We saw a marked decrease of caspase-1 expression in epithelial cells infected with the DarcA in only WT mice ( Figures 4F and 4G). S. pyogenes induces keratinocyte apoptosis by SLO-mediated membrane damage (Cywes Bentley et al., 2005). In S. pyogenes infection of HaCaT cells with CDM, an increase in released lactate dehydrogenase (LDH) in culture supernatants indicated that SLO contributed partially to arginine-catabolism-dependent cytotoxicity ( Figure S5A). Next, we determined whether pyroptosis or apoptosis was induced in infected epithelial cells by measuring interleukin-1b (IL-1b) release and DNA fragmentation by TUNEL staining. Because it was difficult to discriminate between pro-IL-1b and mature IL-1b by ELISA, we assessed signaling activity using HEK-Blue IL-1b reporter cells for quantifying pyroptosis. The secretion of mature IL-1b from Ha-CaT cells was enhanced in both WT and Dslo by arginine catabolism ( Figure S5B), with SLO partially contributing to the pyroptosis phenotype, paralleling the cytolytic effect. On the other hand, in a TUNEL assay designed to detect apoptotic cells, arginine-catabolism-dependent apoptosis was not observed ( Figure S5C). Taken together, S. pyogenes SLO contributes significantly to the pyroptosis of HaCaT cells induced by arginine catabolism, and there are other S. pyogenes factors that are likely involved in arginine-catabolism-independent pyroptosis.

DISCUSSION
The S. pyogenes ADI pathway is controlled by virulence-related metabolic regulators (Dmitriev et al., 2006;Shelburne et al., 2010) and is highly expressed in vivo (Graham et al., 2006;Hirose et al., 2019) and ex vivo (Graham et al., 2005;Shelburne et al., 2005). Here, we show that the S. pyogenes ADI pathway influences virulence factor expression and contributes to keratinocyte cytotoxicity under low-glucose conditions in vitro and subcutaneous infection in vivo. Our data support a model in which S. pyogenes uses arginine abundant in filaggrin of the SC, can secure nutrition, survive, and produce skin infection associated with local tissue destruction.
We found that S. pyogenes can survive for more than 40 days if it can maintain neutral pH by metabolizing arginine during the stationary phase. This result is consistent with a previous report that proved the long-term survival potential of S. pyogenes in neutral pH (Savic and McShan, 2012). The ability of the S. pyogenes ADI pathway to supplement energy production and provide protection against acid stress might greatly contribute to its viability in specialized environments such as skin.
Comparative transcriptome analysis reveals arginine-catabolism-dependent gene regulation under glucose-starvation conditions. Upregulated genes in the WT strain include genes in the pyrimidine biosynthesis pathway (pyrD, pyrE, and pyrF). In S. pyogenes, the ADI pathway cooperates with the pyrimidine biosynthesis pathway to acquire pyrimidine ribonucleotides and ATP (Hirose et al., 2019). Conversely, genes encoding F 0 F 1 -type ATP synthase were downregulated in the WT strain compared to those in the DarcA mutant. Although it has been reported that the generation of ATP by the ADI pathway and a functional F 0 F 1 -type ATP synthase work in concert to adapt to acid stress (Cusumano and Caparon, 2015), it is speculated that S. pyogenes in CDM at neutral pH decreases the concomitant hydrolysis of ATP to ADP to reduce ATP consumption. Genes related to cell division (ftsH, ftsL, and ftsZ) were also downregulated in WT S. pyogenes, which might prioritize the expression of cytolytic virulence factors when there are not enough local nutrition sources to proliferate.
The deletion of covR induces the upregulation of virulence genes contained within the streptolysin S operon (sagABCDEFGHI) and the nga-slo operon and also shows downregulation of a dipeptide permease operon (dppABCDE) (Shelburne et al., 2010). These (D) Expression levels of bacterial virulence genes sagA, slo, and speB on the skin surface. The sagA, slo, and speB gene expression levels in the DarcA were examined by qPCR and shown relative to that of the WT strain. The 16S rRNA was used as the internal control. Data from three independent qRT-PCR assays, each performed in triplicate, were pooled and normalized. Vertical lines represent the mean + SE. **p < 0.01. Report ll OPEN ACCESS findings were mirrored in our results from RNA-seq analysis. Although CovR phosphorylation enhances DNA binding of CovR (Graham et al., 2002) and CovR affects the expression of 15%-20% of S. pyogenes genes (Horstmann et al., 2015), S. pyogenes arginine catabolism did not significantly influence CovR phosphorylation in our assay. Taken together, these results suggest that the main mechanism of S. pyogenes pathogenicity under glucose-depleted conditions may be dependent on acquiring ATP by catabolism of arginine.
Although transcript levels of the ADI operon were reduced in serotype M1 S. pyogenes (MGAS5005) after human blood exposure, temporal and mild upregulation of ADI operon were observed within 90 min of blood exposure (Graham et al., 2005). These investigators also reported that the deletion of the S. pyogenes covR regulator led to upregulation of the ADI operon in blood. These results partly contrast with our data. However, human blood is relatively poor in arginine but rich in glucose (Canepa et al., 2002;Sylvestre et al., 2010). Therefore, although CovRS might mediate some environmental signals in the blood and upregulate ADI operon expression, we speculate that arginine catabolism changes are not sufficient to drive pathogenesis of S. pyogenes in blood. In contrast, the DarcA mutant showed lower virulence than the WT S. pyogenes strain in a mouse soft-tissue infection model that was associated with localized necrosis in adjacent muscle. This difference in pathogenicity might be explained by high concentrations (approximately 1,000 mM) of arginine in muscle tissue (Canepa et al., 2002).
In a murine model of epicutaneous infection and in our RNA-seq analysis, S. pyogenes WT upregulated the arcA, slo, and sagA genes. High expression levels of these genes were reported in S. pyogenes isolated directly from mouse soft tissue infection (Graham et al., 2006) and a mouse model of necrotizing fasciitis (Hirose et al., 2019). Streptolysin S is involved in cellular injury, phagocytic resistance, and virulence in murine subcutaneous infection models (Datta et al., 2005;Humar et al., 2002). The upregulation of SLO has been correlated with a highly virulent S. pyogenes phenotype (Zhu et al., 2015). Our finding that ADI contributes to the expression of the sagA and slo genes might be important information for mitigating the pathogenicity of S. pyogenes.
The SC is the outermost layer of the epidermis and acts as the first line of structural defense against pathogens and toxins. Filaggrin, a major structural protein in the SC (Sandilands et al., 2009), contributes to the mechanical strength and integrity of the SC in vivo (Kawasaki et al., 2012), and filaggrin breakdown products form natural moisturizing factors that are believed to play a major role in SC hydration (Rawlings and Harding, 2004). Arginine is a major component of filaggrin-derived natural moisturizing factors (Kubo et al., 2013). In our epicutaneous infection model with Flg À/À mice, S. pyogenes might more easily penetrate to reach viable epidermal cells below the SC whereupon cytotoxic factors can allow the pathogen to secure nutrition from the viable host epidermal cells. Thus, DarcA could exert pathogenesis by using ADI-independent mechanisms on the skin surface of Flg À/À mice.
The SC of skin is also rich in lipids, such as ceramides, cholesterol, and free fatty acids (Elias and Schmuth, 2009), and bacterial infection of the skin also activates the host immune response (Cogen et al., 2008) and promotes acidic pH (Proksch, 2018). In our results, there were certain differences between our in vitro and in vivo findings that could not be explained based solely on S. pyogenes ADI pathway activity. Further experiments will be required to confirm whether other factors are involved in the pathogenesis of S. pyogenes and reveal more details about interactions between S. pyogenes and host skin tissue.
We revealed that S. pyogenes induced increased pyroptosis of HaCaT cells in a manner dependent on arginine catabolism, whereas almost no apoptosis was observed both dependently and independently of arginine catabolism. However, Cywes Bentley et al. (2005) reported that S. pyogenes SLO contributed to keratinocyte apoptosis. This discrepancy may be due to the difference of the keratinocyte cell line used or nutritional conditions. Because SLO was not fully responsible for the observed pyroptosis, further exploration will be needed to fully clarify the arginine-catabolism-dependent pyroptosis-inducing factors of S. pyogenes.
In summary, our findings suggest that S. pyogenes uses arginine from SC-derived filaggrin to adapt to glucose starvation on the skin surface. Despite the fact that arginine is a molecule that contributes to natural moisturizing of the skin, it can be simultaneously exploited by S. pyogenes that may metabolize arginine to promote its pathogenesis.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

RESOURCE AVAILABILITY
Lead contact Further information and requests for reagents may be directed to, and will be fulfilled by the corresponding author Yujiro Hirose (yujirohirose@dent.osaka-u.ac.jp).

Materials availability
This study did not generate new unique reagents.

Data and code availability
The accession number for the bacterial RNA-seq reads reported in this paper is SRA: DRA009112 .

EXPERIMENTAL MODEL AND SUBJECT DETAILS
Bacterial strains and culture conditions Streptococcus pyogenes M1T1 strain 5448 (GenBank: CP008776.1) was isolated from a patient with toxic shock syndrome and necrotizing fasciitis that is genetically representative of a globally disseminated clone associated with invasive S. pyogenes infections (Kansal et al., 2000). S. pyogenes strains were grown at 37 C in a screw-cap glass tube (Pyrex; Iwaki Glass, Tokyo, Japan) filled with Todd-Hewitt broth (BD Biosciences, San Jose, CA, USA) supplemented with 0.2% yeast extract (BD Bioscience) (THY broth) in an ambient atmosphere and standing cultures. To obtain cultures for experiments and observe pH change, overnight cultures of S.pyogenes were back diluted 1:50 into fresh THY broth or phenol red broth (Sigma Aldrich, St Louis, MO, USA) supplemented with 30 mM arginine. CFUs were determined by plating diluted samples on THY blood agar. Escherichia coli strain XL-10 Gold (Agilent Technologies, Santa Clara, CA, USA) was used as a host for derivatives of plasmids pSET4s (Takamatsu et al., 2001) and pQE30 (QIAGEN, Hilden, Germany). E. coli strains were cultured in Luria-Bertani medium (Nacalai Tesque, Kyoto, Japan) at 37 C with agitation. For selection and maintenance of strains, antibiotics were added to the medium at the following concentrations: spectinomycin, 100 mg/mL for S. pyogenes and E. coli: carbenicillin, 100 mg/mL for E. coli.

Construction of mutant strains
An in-frame arcA deletion mutant (DarcA) and its revertant strain (Wr) with a background of strain 5448 (WT) were constructed using the pSET4s temperature-sensitive shuttle vector, as previously reported (Nakata et al., 2011). A pSET4-ArcAKO plasmid harboring the DNA fragment, in which upstream and downstream regions of arcA were linked by overlapping PCR, was electroporated into Report ll OPEN ACCESS strain 5448 and grown in the presence of spectinomycin. The plasmid was then integrated into the chromosome via first allelic replacement at 37 C, after which it was cultured at 28 C without antibiotics to induce the second allelic replacement. The deletion of arcA was confirmed by site-specific PCR using purified genomic DNA. Primers are listed in Table S2.

Cell culture and media
We used the immortal human keratinocyte line, HaCaT cells. HaCaT cells were cultured in Dulbecco's Modified Eagle Media (DMEM, Cat#: 10-013-CV, Corning, NY, USA), with 10% Fetal Bovine Serum (Cat# 97068-085, VWR International LLC, Radnor, USA). The cell culture was maintained in a humidified 5% CO 2 atmosphere at 37 C. The cells were cultured to around 70% confluence. To subculture cells, adherent cells were rinsed with PBS without calcium and magnesium, and detached by using trypsin/EDTA solution (0.05% trypsin, 0.53 m EDTA) for~10 min, added fresh culture medium, centrifuged, resuspended cells in fresh culture medium, and dispensed into new culture vessels. For all experiments, freshly trypsinized cells were seeded at a density of~1 3 10 5 cells/ cm 2 one day prior to bacterial infection.

Murine model of epicutaneous and intravenous infections
All mouse experiments were conducted in accordance with animal protocols approved by the Animal Care and Use Committee of Osaka University Graduate School of Dentistry (30-011-0) and University of California San Diego Institutional Animal Care and Use Committee (IACUC). The filaggrin null mouse strain (B6.Cg-Flg < tm1 > , RBRC05850) was provided by RIKEN BRC through the National Bio-Resource Project of the MEXT, Japan. Epicutaneous infections were performed using a previously reported with minor modifications (Nakamura et al., 2013;Sumitomo et al., 2018). Briefly, bacterial cultures during exponential phase were centrifuged, washed with and resuspended in PBS. Dorsal skin of C57BL/6 wild-type (wt) mice (6-to 7-week-old, both female and male; Japan SLC, Shizuoka, Japan) and the filaggrin null (Flg À/À ) mice (Kawasaki et al., 2012) (6-to 7-week-old, both female and male) was depilated 2 days before infection. A bacterial suspension (5 3 10 5 -10 7 CFU in 100 mL PBS) was placed on a 1 3 1 cm patch of sterile gauze, which is secured to the shaved skin with a transparent bio-occlusive dressing. At 3 hours post-infection, bacteria on the skin surface were collected by using a stainless dental scaler, then bacterial RNA was extracted and quantified by qPCR as described above. At 3 days post-infection, cutaneous tissue was excised for histopathologic analyses and assessment of bacterial burden. Cutaneous tissue samples were obtained and fixed with formalin, then embedded in paraffin, sectioned, and subjected to hematoxylin and eosin (HE) staining. Bacterial counts in cutaneous tissue homogenates were determined after plating serial dilutions, with those in the cutaneous tissue corrected for differences in tissue weight.
For intravenous infection, C57BL/6 wild-type mice (6-to 7-week-old, both female and male; Japan SLC) were intravenously infected with 2 3 10 6 CFU of S. pyogenes during exponential phase, and survival was monitored for 14 days.
Mouse model of S. pyogenes necrotizing skin infection Invasiveness of S. pyogenes in mouse skin was measured by modification of a previously described S. pyogenes infection model (Nizet et al., 2001). All mouse experiments were conducted in accordance with animal protocols approved by the Animal Care and Use Committee of Osaka University Graduate School of Dentistry (30-011-0). The CD-1 (Slc: ICR) mice (6 weeks old, female; Japan SLC, Shizuoka, Japan) were shaved and hair removed by chemical depilation (Veet, Oxy Reckit Benckiser, Chartes, France). S. pyogenes were cultured until the log phase (OD 600 = 0.5~0.6), and adjusted 1 3 10 7 CFU in 200 mL of PBS were injected subcutaneously. Areas with ulcer were defined as lesions and areas were measured daily for up to 3 days after infection.

Streptozotocin-induced diabetic mice
To induce diabetes mellitus, C57BL/6 male mice (4-week-old) were injected i.p. with streptozotocin (Adipogen, San Diego, CA, USA) at 80 mg/kg/dose in 200 mL of 0.1 M citrate buffer daily for 4 days (Patras et al., 2020). Control mice received 4 daily treatments of 200 mL of 0.1 M citrate buffer. Mice were weighed weekly thereafter. The concentration of blood glucose in 7-week-old mice was determined 24 h prior to infection. Sample glucose was determined using an AimStrip Plus blood glucose meter kit (Germaine Labs, Indianapolis, IN, USA).

METHOD DETAILS
Quantitative real-time PCR (qPCR) Bacterial cultures during exponential phase (OD 600 = 0.5-0.6), early stationary phase (OD 600 = 1.2), or decline phase (overnight culture) were centrifuged and immediately placed into RNAprotect Bacteria Reagent (QIAGEN) prior to RNA isolation. In the RNA isolation from S. pyogenes cultured within CDM, bacterial cultures during exponential phase (OD 600 = 0.5-0.6) were centrifuged, e3 Cell Reports 34, 108924, March 30, 2021 Report ll OPEN ACCESS resuspended into CDM, incubated in a screw cap glass tube (Pyrex; Iwaki Glass, Tokyo, Japan) for 1 h at 37 C, and immediately placed into RNAprotect Bacteria Reagent. S. pyogenes was resuspended into lysing Matrix B microtubes containing 0.1-mm silica spheres (Qbiogene, Carlsbad, CA, USA) with RLT lysis buffer (RNeasy Mini Kit; QIAGEN), and homogenized at 6,500 rpm for 60 s using the MagNA Lyser (Roche Molecular Diagnostic, Mannheim, Germany). RNA was isolated from the lysate with RNeasy Mini Kit according to the manufacturer's guidelines, and then cDNA was synthesized using a Superscript VILO cDNA synthesis kit (Thermo Fisher Scientific, Waltham, MA, USA). Real-time reverse transcription PCR analysis was performed using a StepOnePlus real-time PCR system (Applied Biosystems, Foster City, CA, USA) and Toyobo SYBR green RT-PCR master mix kit (Toyobo Life Science, Osaka, Japan). Data for 16S rRNA or rpoB were used as the internal control. Primers used for qPCR are listed in Table S2.

Measurement of glucose and ammonium ion concentrations
Culture supernatant at each growth phase obtained from S. pyogenes WT and DarcA cultured in THY broth was filtered through a 0.22 mm membrane, then and directly analyzed with BioProfileâ FLEX2 analyzer following manufacturer's instruction (Nova Biomedical, Inc., Waltham, MA, USA).

Measurement of extracellular pH
For phenol red broth (Sigma Aldrich, St Louis, MO, USA) supplemented with 30 mM arginine, culture supernatant at each point was measured the absorbance at 550 nm. A calibration curve was determined in phenol red broth which adjusted to pH values ranging from 4 to 10. The pH values were assessed up to 40 days in the sample which included surviving bacteria. For THY broth, culture supernatant at each point was supplemented with 5 mg/mL phenol red and the absorbance was measured at 550 nm. A calibration curve was determined in THY broth which was supplemented with 5 mg/mL phenol red and adjusted to pH values ranging from 4 to 10.

Measurement of intracellular pH
The cytosolic pH of S. pyogenes was determined based on the previously described fluorescent probe method (Do et al., 2019). Briefly, S. pyogenes grown to log phase of growth in THY broth were centrifugated, washed in 150 mM NaCl, and resuspended in 50 mM HEPES buffer (pH 8.0). The cells were then incubated for 20 min at 37 C in the presence of 10 mM carboxyfluorescein diacetate succinimidyl ester (cFDASE, Invitrogen, Grand Island, NY, USA). cFDASE is hydrolyzed to carboxyfluorescein succinimidyl ester (cFSE) in the cell and subsequently conjugated to aliphatic amines of the intracellular proteins. After incubation, cells were washed and suspended in 50 mM potassium phosphate buffer (pH 7.5). To eliminate nonconjugated cFSE, cells were incubated with 10 mM glucose for 30 min at 30 C. Subsequently, S. pyogenes were washed, and suspended, and incubated in CDM at 37 C within a screw-cap glass tube. At 1 h, 5 h, and 15 h post-incubation, fluorescence intensities were determined with an excitation spectrum of 400-500 nm wavelength range that includes excitation wavelengths 490 nm (pH-sensitive) and 435 nm (pH-insensitive) (Spark 10M; TEKAN, Mä nnedorf, Switzerland). Emission was determined at 520 nm. The ratio of the emission resulting from excitation at 490 and 435 nm obtained for both cell suspension (C) and filtrate (F) was calculated as R 490/435 = (C490 -F490)/ (C 435 -F435). A calibration curve was determined in CDM adjusted to pH values ranging from 5.5 to 8.0 and a cubic equation for the ratio value was determined. Intracellular pH values of S. pyogenes were calculated using the cubic equation from the calibration curve.

Infection of HaCaT cells with S. pyogenes
The composition of Chemically Defined Medium (CDM) is shown in Table S1. S. pyogenes in log phase growth in THY broth were centrifuged and resuspended into CDM supplemented with or without 1000 mM arginine. HaCaT cells were infected with S. pyogenes (MOI = 500). At 20 h post-incubation, supernatants were collected by centrifugation and analyzed for cytotoxicity assays, IL-1b signaling assay, and cells were used for the apoptosis determination.

Cytotoxicity assays
Cell viability was assessed using the LIVE/DEADâ Viability/Cytotoxicity Kit (Thermo Fisher Scientific). LDH activity in the culture supernatant was measured by using LDH Assay Kit-WST (Dojindo, Kumamoto, Japan). At 1 h, 5 h, and 15 h post-infection, the bacterial count in CDM with cultured HaCaT cells was evaluated by combining CFUs from supernatant and those from associated with HaCaT cells. To determine bacterial association, HaCaT cells were harvested with PBS containing 0.05% trypsin and 0.025% Triton X-100.

Measurement of intracellular ATP levels
At 1 h, 5 h, and 15 h post-incubation in CDM, the intracellular ATP levels of S. pyogenes were also evaluated by an ATP-bioluminescent assay using Kinshiro (TOYO B-Net, Tokyo, Japan) according to manufacturer's instructions. Briefly, ATP extractant solution was added to equal amount of the mixture of CDM and S. pyogenes. After the incubation for 10 s at room temperature, 100 mL samples were mixed with equal amount of bioluminescent reagent, and bioluminescence was measured with a luminometer (Infinite 200 Pro multiplate reader, TEKAN, Mä nnedorf, Switzerland), immediately. After establishing of the calibration curve, the ATP concentrations in the samples were determined, and the percentage of ATP levels for S. pyogenes at 0 h post-incubation was calculated.
1 Figure S1. Effects of arginine catabolism on extracellular or intracellular pH, and the cytotoxicity. Figures 1 and 2. (A) Arginine catabolism in S. pyogenes. Through the multienzyme pathway, arginine is transported into the cell via the antiporter ArcD and catabolized by ArcA, ArcB, and ArcC to produce one molecule of carbon dioxide, one molecule of ATP, and two molecules of ammonia. ArcA,  respectively. The percentage of TUNEL-positive cells was calculated as the ratio between TUNELpositive and DAPI-stained nuclei ×100. Representative data obtained from at least 3 independent experiments are shown. Vertical lines represent the mean + S.E. (n = 6). Statistical differences between groups were analyzed using a Mann-Whitney U test. **p < 0.01.**p < 0.01.