Abstract
Streptococcus pyogenes is a major cause of necrotizing fasciitis, a life-threatening subcutaneous soft-tissue infection. At the host infection site, the local environment and interaction between host and bacteria affect bacterial gene-expression profiles, but the S. pyogenes gene-expression pattern in necrotizing fasciitis remains unknown. In this study, we used a mouse model of necrotizing fasciitis and performed RNA-sequencing (RNA-seq) analysis of S. pyogenes M1T1 strain 5448 by using infected hindlimbs obtained at 24, 48, and 96 h post-infection. The RNA-seq analysis identified 483 bacterial genes whose expression was consistently altered in the infected hindlimbs as compared to their expression under in vitro conditions. The consistently enriched genes during infection included 306 genes encoding molecules involved in virulence, carbohydrate utilization, amino acid metabolism, trace-metal transport and vacuolar ATPase transport system. Surprisingly, drastic upregulation of 3 genes, encoding streptolysin S precursor (sagA), cysteine protease (speB), and secreted DNase (spd), was noted in the mouse model of necrotizing fasciitis (log2 fold-change values: >6.0, >9.4, and >7.1, respectively). Conversely, the consistently downregulated genes included 177 genes, containing genes associated with oxidative-stress response and cell division. These results suggest that S. pyogenes in necrotizing fasciitis changes its metabolism, decreases cell proliferation, and upregulates the expression of major toxins. Our findings could provide critical information for developing novel treatment strategies and vaccines for necrotizing fasciitis.
Author summary Necrotizing fasciitis, a life-threatening subcutaneous soft-tissue infection, principally caused by a Streptococcus pyogenes. At infection sites in hosts, bacterial pathogens are exposed to drastically changing environmental conditions and alter global gene expression patterns for survival and pathogenesis. However, there is no previous report about transcriptomic profiling of S. pyogenes in the necrotizing fasciitis. Here, we conducted comprehensive gene-expression analyses of S. pyogenes in the mouse model of necrotizing fasciitis at three distinct time points during infection. Our results indicated that S. pyogenes drastically upregulates the expression of virulence-associated genes and shifts metabolic-pathway usage during infection. The high-level expressions in particular of toxins, such as cytolysins, proteases, and nucleases, were observed at infection sites. In addition, the consistently enriched genes identified here included genes for metabolism of arginine and histidine, and carbohydrate uptake and utilization. Conversely, the genes associated with oxidative-stress response and cell division were consistently downregulated in the mouse model of necrotizing fasciitis. These data will provide useful information necessary for establishing novel treatment strategies (166 words).
Introduction
Streptococcus pyogenes causes diverse human diseases, ranging from mild throat and skin infections to life-threatening invasive diseases such as sepsis, necrotizing fasciitis, and streptococcal toxic-shock syndrome. Streptococcal necrotizing fasciitis cases are clinically characterized by fulminant tissue destruction and rapid disease progression [1]. In a majority of the cases, surgical treatment is required, including amputations, in addition to intensive care. Although this infection has been attracting increasing research and clinical interest, the mortality rate remains high [2, 3]. Investigation of the molecular pathogenesis of S. pyogenes in necrotizing fasciitis is expected to lead to the development of novel therapeutic strategies or effective treatments.
S. pyogenes typing has been historically conducted on the basis of antigenicity of M protein and T antigen (pilus major subunit). Currently, the sequence typing of the region encoding hypervariable region of M protein has been widely applied to classify this organism into at least 240 emm sequence types [4–6] and ~20 T serotypes [7, 8] have been identified. In industrialized societies, S. pyogenes serotype M1 (emm 1) isolates are considerably more common than other serotypes among invasive cases [6, 9–11], and the M1T1 clone in particular is the most frequently isolated serotype from severe invasive human infections worldwide [12, 13].
At infection sites in hosts, bacterial pathogens are exposed to drastically changing environmental conditions, which include the host cells, tissues, and immune response, as compared to laboratory growth conditions. However, no previous report has compared in vivo and in vitro transcriptome of S. pyogenes
Several comprehensive in vitro analyses of S. pyogenes gene expression performed using microarray or RNA-sequencing (RNA-seq) approaches and revealed the roles of S. pyogenes virulence-related regulators [14], such as CovRS [15, 16] and CcpA [17–19]. The transcriptome profile of S. pyogenes from a mouse soft-tissue infection, which was obtained using microarray analysis, indicated that S. pyogenes MGAS5005 (serotype M1) upregulated genes that are involved in oxidative-stress protection and stress adaptation [20]. The results of another microarray analysis on S. pyogenes MGAS5005 (serotype M1) demonstrated downregulation of glycolysis genes and induction of genes involved in amino acid catabolism and of several types of virulence genes in human blood [21]. These reports suggest that S. pyogenes changes its expression levels of virulence factor and metabolic pathways to adapt to the host environment.
Comprehensive understanding of bacterial transcriptomes in vivo will facilitate research aimed at developing therapeutic strategies or effective vaccine antigens. Although transposon-directed insertion-site sequencing using cynomolgus macaque model of necrotizing fasciitis was conducted recently [22], this method cannot assess the expression level of genes. In addition, transcriptome analysis of S. pyogenes in necrotizing fasciitis have not been performed to date. Here, we investigated the transcriptome profiling of S. pyogenes M1T1 strain 5448 by using a mouse model of necrotizing fasciitis, from the acute phase to the elimination phase, and we identified genes whose expression is consistently altered throughout the infection period. This new information may shed light on the development of novel therapeutic strategies for the infection.
Results
Establishing a technique for high-yield purification of bacterial RNA from mouse tissue
We used a mouse necrotizing fasciitis model as described previously, with minor modifications [23]. At 24 and 48 h after infection, the mouse model of necrotizing fasciitis was histologically similar to human necrotizing fasciitis in terms of tissue necrosis, the infection spread along the fascial planes, inflammatory-cell infiltration, hemorrhage and ulceration [24, 25] (Fig 1A and 1B). Extensive scab formation was detected at 48 h post-infection and elimination of pus from infected hindlimbs was observed at 96 h post-infection. At 96 h after infection, the weight of the mice tended to recover, and the bacterial burden at the infected site also decreased (S1 Fig). Therefore, we collected infected hindlimb samples from the 24, 48, and 96 h groups. To obtain bacterial RNA from infected tissues, we established a suitable protocol by using two types of beads (Fig 1C); this method allows us to remove most mouse RNA from samples and obtain high-yield purification of bacterial RNA (Fig 1D).
Similar gene-expression patterns of S. pyogenes at three distinct time points in infected hindlimbs
We performed RNA-seq analysis on S. pyogenes isolated from infected hindlimbs at 24, 48, and 96 h post-infection. RNA-seq data from S. pyogenes during the exponential growth phase in THY medium (Todd-Hewitt broth plus yeast extract) were defined as the control. To assess the global gene-expression profiles of the samples, we performed principal component analysis (PCA) (Fig 2A), hierarchical clustering analysis (S2 Fig), and k-means clustering (Fig 2B) by using the RNA-seq data. In PCA and hierarchical clustering analysis, bacterial RNA expression patterns of THY culture samples formed a cluster and the samples from the infected tissues were well separated. The 24 h_1 sample showed a global gene-expression profile that was distant from the profiles of other samples, whereas the heatmap of k-means clustering showed that the gene-expression profile of 24 h_1 was at least partially similar to that of samples from the infected tissues as indicated in Clusters A and B. The k-means clustering further suggested that most samples from infected hindlimbs show changes in global mRNA transcript patterns in opposite directions as compared to the THY group.
Consistently altered 483 bacterial genes at three time points in mouse necrotizing fasciitis model
Differentially expressed genes (DEGs; absolute log2 fold-change > 1 and adjusted P < 0.1) were detected between S. pyogenes in infected tissues and in THY broth (Fig 3A). In S1 Dataset, we provide the DEG details and list the following information for all the genes: gene ID, gene name, the gene-associated function, log2 fold-change, adjusted P value, and reads per kilobase per million mapped reads (RPKM) value. In comparisons of 48 h vs 24 h groups and 96 h vs 24 h groups, no DEGs were detected (Fig 3A), and only 4 DEGs were detected between 96 h and 48 h groups (S1 Dataset). These results indicate that S. pyogenes expresses similar genes at the three time points in the mouse model of necrotizing fasciitis. To identify the genes that are consistently enriched or downregulated in the mouse necrotizing fasciitis model, we drew Venn diagrams by using the DEGs from the comparisons of 24 h and THY groups, 48 h and THY groups, and 96 h and THY groups (Fig 3B); 28.0% of all 1,723 genes, 483 genes, were identified as consistently altered bacterial genes in infected hindlimbs (S2 Dataset). Among the 483 genes, 306 and 177 genes were upregulated and downregulated, respectively, at all three time points.
Marked upregulation of genes encoding virulence factors
The consistently enriched genes featured a high proportion of genes encoding virulence factors, such as genes for cytolysins (sagA-I, slo), nucleases (spd, spd3, sdaD2), cysteine protease (speB), factors involved in immune evasion (endoS, spyCEP, scpA, sic), superantigens (speA, smeZ), and adhesins (fbaA, lbp, emm) (Table 1; S2 Dataset). Surprisingly, the RPKM values of genes encoding streptolysin S precursor (sagA), speB, SpeB–inhibitor-encoding gene (spi), and spd were extremely high and were consistently ranked within the top four (S1 Dataset; Fig 4). As compared with their expression under the THY condition, sagA, speB, spi, and spd were expressed in mouse necrotizing fasciitis at the following levels (respectively): log2 fold-change = >6.0, >9.3, >9.4, and >7.1. By contrast, the gene encoding macroglobulin-binding protein (grab) was markedly downregulated at all three time points. Lastly, hyaluronic acid synthesis operon (hasABC) was significantly upregulated in the 48 h and 96 h groups, but no significant difference was detected in the 24 h group.
Upregulation of carbohydrate uptake and utilization genes
The consistently enriched genes also included most genes encoding ATP-binding cassette (ABC) transporters or phosphoenolpyruvate-phosphotransferase system (PTS) molecules responsible for carbohydrates transport (Fig 5; Table 1; S2 Dataset). In the glycolysis pathway, the expression of pgk (encoding phosphoglycerate kinase) and eno (encoding enolase) showed a slight decrease, but the RPKM values of these genes consistently remained at >1,500. Despite the sufficient expression of glycolysis-system molecules in the infected hindlimbs, the carbohydrate transport systems exhibited an overall increase. Shelburne et al. reported that the carbon catabolite protein CcpA upregulates the expression of most operons encoding transporters of carbohydrates, such as glucose, lactose, maltodextrin, mannose, fructose, cellobiose, lactose, galactose, and sialic acid, under glucose-limiting conditions [18]. Moreover, our results indicated that the genes encoding phosphocarrier protein (ptsH) and its kinase (ptsK) were consistently downregulated (Fig 5). When Gram-positive bacteria are in the presence of glucose, phosphocarrier protein (HPr) is phosphorylated at Ser46 by its kinase, HprK, which allows phosphorylated HPr to dimerize with CcpA; the dimerized proteins then bind to catabolite-response elements present in promoter sequences and elicit carbon catabolite repression [26]. These findings raise the possibility that S. pyogenes is relieved from carbon catabolite repression in the mouse model of necrotizing fasciitis.
Drastically enhanced arginine and histidine metabolism in infected hindlimbs
S. pyogenes is auxotrophic for at least 15 amino acids [27]. The consistently enriched genes identified here included operons for metabolism of arginine (arcABCD), histidine (hutDGHIU, ftcD, fchA, fhs.2), and serine (salB) (Table 1; Fig 4; Fig 6; S2 Dataset). Conversely, the bacterial genes encoding proteins for isoleucine metabolism (bcaT, acoC) were consistently downregulated in the infected hindlimbs (Fig 6). The operon for the dipeptide transporter dppABCDF, which is involved in the uptake of essential amino acids [28], was also upregulated in the infected hindlimbs (Table 1), and the expression of dppA, which encodes a dipeptide-binding protein, was remarkably enhanced (log2 fold-change > 2.25) (Fig 4; S2 Dataset).
The mean fold-changes in the transcript levels (i.e., the mean log2 fold-change values) for all genes in the operons for arginine and histidine metabolism were >5.67 and >6.38, respectively. In S. pyogenes, the arginine deiminase pathway (arcABCD) is reported to supplement energy production, help protect against acid stress, and compete with arginine-dependent NO production by host cells in the subcutaneous layer [29]. Another critical role of arginine metabolism is to serve as the source of uridine monophosphate (Fig 6A), whereas histidine metabolism is connected to the synthesis of inosine monophosphate (Fig 6B). These functions cooperate with pyrimidine and purine metabolism for the synthesis of DNA and RNA. The consistently enriched genes also included genes for pyrimidine and purine metabolism (S2 Dataset). These results suggest the possibility that bacterial synthesis of nucleic acids is active in infected hindlimbs, although we also observed the repression of certain genes related to cell division, such as genes encoding cell-division proteins (ftsA, ftsZ, ftsH), amino acid ligases (murD, murG), phospho-N-acetylmuramoyl-pentapeptide transferase (mraY), and ribonuclease III (rnc) (Fig 4; S1 and S2 Datasets).
Host-induced bacterial stress responses
Genes encoding superoxide dismutase (sodA) and glutathione peroxidase (gpoA) were consistently downregulated in the infected hindlimbs (Table 1; S2 Dataset). SodA and GpoA act to neutralize endogenous and exogenous peroxides, which contributes to detoxification of reactive oxygen species in vitro [30, 31]. Our results suggest that S. pyogenes is not exposed to substantial oxidative stress in the infected hindlimbs as compared to the stress encountered during the aerobic growth.
Transition metals are involved in several crucial biological processes in pathogens that are necessary for the pathogens to survive, proliferate, and cause diseases in their environmental niche. In S. pyogenes, contributions to virulence are made by the homeostasis of metals, including iron, manganese [32], and zinc [33], whereas hosts exploit this phenomenon and combat invading pathogens by restricting the availability of essential metals by using transferrin (iron), lactoferrin (iron), and calprotectin (manganese and zinc) [34]. Here, S. pyogenes was found to upregulate genes involved in iron and manganese transport (shr/shp/siaABCD, fhuGBDA) and zinc transport (adcRCB, htpA, lbp) in the mouse model of necrotizing fasciitis (Table 1; S2 Dataset).
Altered expression of virulence-related transcriptional-regulator genes
Expression of most virulence genes in S. pyogenes is under the control of two-component signal transduction systems (TCSs) and transcriptional activators/repressors [14]. Although phosphorylation is recognized as a key modification by which regulators exert regional transcriptional control [26, 35, 36], the alternation of regulator gene-expression levels could also influence the degree of regulation.
S. pyogenes showed altered expression of several genes encoding virulence-related regulators (Fig 7): The consistently enriched genes included TCS trxSR operon and genes encoding carbohydrate-sensitive regulators (lacD.1, ccpA), a member of the RofA-like protein type family of stand-alone virulence-related regulators (rivR/ralp4), and maltose repressor (malR). Conversely, the only consistently downregulated regulator gene was the gene encoding streptococcal regulator of virulence (srv), although certain other regulators also tended to show downregulation, including the genes for CovRS (covRS), the metabolic-control regulator VicRK (vicRK), metalloregulator (mtsR/scaR), and RofA regulator (rofA).
Discussion
This is the first report of comprehensive gene-expression analyses of S. pyogenes in a mouse model of necrotizing fasciitis. For RNA-seq analysis of bacteria in host tissues, deep sequencing has been previously used to obtain a sufficient number of reads [37]. However, our protocol is simple and inexpensive and appears to effectively enable in vivo RNA-seq analysis of Gram-positive bacteria without deep sequencing. Here, we also analyzed the transcriptome profiles of S. pyogenes at three distinct time points during infection. Our results indicated that S. pyogenes drastically upregulates the expression of virulence-associated genes and shifts metabolic-pathway usage in the mouse model of necrotizing fasciitis consistently, and the results showed high-level expression in particular of sagA, speB, and spd. By contrast, S. pyogenes downregulated genes associated with oxidative-stress response and cell division in infected hind limbs relative to that in THY culture at the mid-logarithmic phase.
Our RNA-seq analysis revealed that sagA, spi, speB and spd were extremely upregulated in mouse necrotizing fasciitis as compared to in bacterial culture medium. Streptolysin S (SLS; encoded by sagA-I) and SpeB (encoded by speB) are widely recognized virulence factors of S. pyogenes [38]. SLS is involved in cellular injury, phagocytic resistance, and virulence in murine subcutaneous infection [39, 40], and SLS and SpeB promote S. pyogenes translocation via a paracellular route by degrading epithelial junctions [41, 42]. SpeB is a secreted cysteine protease degrading a wide variety of host proteins including complement components and cytokines, and functions in escape of S. pyogenes from host immune response [43–47]. Moreover, SpeB has been shown to contribute its virulence substantially in mouse models of necrotizing myositis [23, 48]. The spi and speB genes are co-transcribed [49]. The spi gene encodes a specific SpeB inhibitor, Spi, to protect bacterial cell from the activity of residual unsecreted SpeB. We found that DNases encoded by sda1, spd3, and spd were also upregulated markedly. Sda1 allows S. pyogenes to escape killing in neutrophil extracellular traps and contributes to virulence in murine subcutaneous infection [50, 51]. The expression of spd, which encodes streptodornase B or mitogenic factor 1, was ranked 4th here, and a previous study has also reported its contribution to the virulence of S. pyogenes (serotype M89) [52]. Although S. pyogenes contains various virulence factors [12, 38, 53], these four genes showed outstanding up-regulation in our infection model. Our findings would help searching therapeutic targets for necrotizing facilities.
In this study, we also detected drastic upregulation of virulence genes encoding histidine triad protein (HtpA) [54] and laminin-binding protein (Lbp) [55]; our results could provide valuable insights regarding the utility of these molecules as therapeutic targets. Although we used a mouse intraperitoneal-infection model, we found that HtpA functions as an effective vaccine antigen against S. pyogenes [56]. Furthermore, analyses of sera from patients with uncomplicated S. pyogenes infection or rheumatic fever indicated the detectable humoral response against recombinant S. pyogenes Lbp [57].
The 483 genes that were consistently altered in this study overlap with the 150 low-glucose-induced genes of strain HSC5 (serotype 14) [17]. The overlapping genes include the upregulated genes encoding molecules involved in carbohydrate uptake and metabolism, arginine metabolism, V-Type ATP synthase, and lactate oxidase, whereas the overlapping downregulated genes contain molecules related to oxidative-stress response and cell division. In terms of the expression of genes encoding virulence factors, we observed the overlapping of upregulation of the genes for SLS, streptolysin O, and Spd and downregulation of GRAB gene. These findings suggest that S. pyogenes in the infected hindlimbs encounters a glucose-poor environment and relieves carbon catabolite repression [26].
Mutations in covRS of S. pyogenes serotype M1 (strains 5448) have been reported to enhance virulence during subcutaneous infection in mouse and might be responsible for loss of SpeB expression [51]. Graham et al. also reported that serotype M1 S. pyogenes (MGAS5005) showed reduced levels of the speB transcript during growth in human blood [21].
However, in this study, the gene encoding SpeB was drastically upregulated in the mouse model of necrotizing fasciitis (log2 fold-change > 9.38). The environment that S. pyogenes encounters in necrotizing fasciitis is considered to be distinct from that in blood and in subcutaneous tissue. Although blood pH is maintained in a narrow range around pH 7.4 in living organisms, inflammatory loci are typically associated with an acidic environment [58]. Moreover, our results suggested that S. pyogenes encounters glucose deprivation in necrotizing fasciitis. In S. pyogenes, speB expression at the early stationary phase can be substantially suppressed by glucose and buffered pH [59]. Generally, the stationary phase of bacterial growth is evidenced by glucose depletion and medium acidification. Thus, an environment similar to the bacterial stationary phase might have induced the strong expression of speB.
Graham et al. also characterized the MGAS5005 (serotype M1) transcript profile in a mouse soft-tissue infection model (subcutaneous infection) by using a wild-type strain and ΔcovR strain [20]; intriguingly, relative to the wild-type strain, ΔcovR strain exhibited drastic upregulation of sagA (18-fold), speB (2,053-fold), and spd (6-fold) in this model, and normalized expression levels of these 3 genes ranked 8th, 2nd, and 5th, respectively, in ΔcovR strain. In our study, S. pyogenes in the mouse model of necrotizing fasciitis also showed extremely high normalized expression levels of sagA (ranked 1st), speB (3rd), and spd (4th) among 1723 genes. One of the classic signs of acute inflammation is heat, and muscle temperature is considered to be higher than skin temperature [60]. S. pyogenes appears to encounter higher temperatures during myositis than during subcutaneous infection, which might lead to distinct transcriptome profiles of S. pyogenes.
Arginine and histidine are present in human muscle at high concentrations, ~1,000 and 500 μM, respectively [61]. Because a supply of amino acids is essential for protein and nucleic acid synthesis, the arginine and histidine metabolic pathways are likely to be enhanced, as was observed here, for pathogenicity to be exerted in necrotizing fasciitis. Moreover, for the uptake of essential amino acids, the operon encoding dipeptide transporter (DppABCDF) was consistently upregulated in the infected hindlimbs. Deletion of S. pyogenes dppA results in speB expression decreasing to one-eighth of its original level (serotype M49, strain CS101) [28]. Thus, dppA upregulation might contribute to the drastically increased expression of speB.
Recently, Zhu et al. identified the genes required for a cynomolgus macaque model of necrotizing fasciitis by using transposon-directed insertion-site sequencing [22]. The serotype M1 (MAGS2221) genes necessary for infection that were identified by Zhu et al. overlap with certain upregulated genes in our study, such as genes for carbohydrate metabolism (glgP, malM), arginine metabolism (arcABCD), and putative or known transporters (valine, braB; zinc, adcBC; SLS, sagGHI). However, in transposon-directed insertion-site sequencing, insertion sites are detected after DNA-sequencing, implying that gene-expression levels are not considered. In RNA-seq analysis, relative expression levels among all genes can be evaluated. For the investigation of therapeutic targets, it is critical to select highly expressed molecules, which suggests the importance of our study for this purpose.
No transcriptome profiling of S. pyogenes in necrotizing fasciitis have been previously reported. This study revealed that S. pyogenes in the mouse model of necrotizing fasciitis exhibited substantially altered global transcription as compared to that under in vitro conditions. S. pyogenes might have attempted to acquire nutrients by destroying tissues by markedly upregulating the expression of toxins such as SLS, SpeB, and Spd. Furthermore, genes encoding molecules involved in carbohydrate and amino acid utilization as well as metal-transporter genes were upregulated in the infected mouse hindlimbs. We also believe that our protocol for isolating bacterial RNA from infected tissues at high concentrations will facilitate studies involving global gene-expression analyses of bacteria in the in vivo host environment. Future studies could explore new therapies based on bacterial kinetics in vivo by exploiting our data or our methods. The accumulation of in vivo gene-expression profiles will provide useful information necessary for establishing novel treatment strategies or identifying effective vaccine antigens.
Materials and methods
Ethic statement
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 (28-002-0). Animals were cared for according to Guidelines for Proper Conduct of Animal Experiments (Science Council of Japan) and the policy laid down by the Animal Care and Use Committee of Osaka University Graduate School of Dentistry.
Bacterial strains and culture conditions
S. pyogenes M1T1 strain 5448 (accession: CP008776) was isolated from a patient with toxic-shock syndrome and necrotizing fasciitis; the strain is genetically representative of a globally disseminated clone associated with invasive S. pyogenes infections [62]. S. pyogenes strain 5448 was cultured in Todd-Hewitt broth (BD Biosciences, San Jose, CA) supplemented with 0.2% yeast extract (BD Biosciences) (THY) at 37°C. For growth measurements, overnight cultures of S. pyogenes strain 5448 were back-diluted 1:50 into fresh THY and grown at 37°C; growth was monitored by measuring the optical density at 600 nm (OD600).
Necrotizing fasciitis studies
We used 10-week-old male C57BL/6J mice (Charles River Japan Inc., Kanagawa, Japan) for the necrotizing fasciitis studies, as described previously [23]. After growing S. pyogenes cultures until the mid-exponential phase (OD600 = ~0.5), THY was replaced with PBS and the bacterial suspensions were stored in a refrigerator (−80°C). Viable cell counts of the suspensions were determined by plating diluted samples on THY blood agar. Mice were shaved and hair was removed through chemical depilation (Veet, Oxy Reckit Benckiser, Chartes, France), and then the mice were inoculated intramuscularly in both sides of hindlimbs with 2 × 107 CFU suspended in 100 μL of PBS, prepared immediately before infection by diluting frozen stocks.
Mice were euthanized at 24, 48, or 96 h after infection by means of lethal intraperitoneal injection of sodium pentobarbital, and then the infected hindlimbs were collected. The left hindlimbs were immediately placed in RNAlater (Qiagen, Valencia, CA) and stored at −80°C until use in RNA isolation, whereas the right hindlimbs were fixed with formalin, embedded in paraffin and sectioned, and stained with hematoxylin and eosin, as described previously [63].
RNA isolation
Thawed tissues were placed in lysing Matrix D microtubes containing 1.4-mm silica spheres (Qbiogene, Carlsbad, CA) with RLT lysis buffer (RNeasy Fibrous Tissue Mini Kit, Qiagen, Hilden, Germany) and homogenized at 6,500 rpm for 45 s by using a MagNA Lyser (Roche, Mannheim, Germany). The lysate was centrifuged, and the obtained pellet was resuspended in lysing Matrix B microtubes containing 0.1-mm silica spheres (Qbiogene) with the RLT lysis buffer and homogenized at 6,500 rpm for 60 s by using the MagNA Lyser. The final lysate was centrifuged, and bacterial RNA was isolated from the collected supernatant by using the RNeasy Fibrous Tissue Mini Kit according to the manufacturer’s guidelines and stored at −80°C (Fig 1B).
RNA-seq and data analysis
RNA integrity was assessed using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) (Fig 1C). For RNA-seq, 5 μg of bacterial RNA was treated for ribosomal RNA (rRNA) removal by using a Ribo-Zero rRNA Removal Kit (Mouse and Bacteria) (Illumina Inc., San Diego, CA). Directional RNA-seq libraries were created using TruSeq RNA Sample Prep Kit v2 (Illumina Inc.), according to the manufacturer’s recommendations. Libraries were sequenced using Illumina NovaSeq 6000 and HiSeq 2500 systems, with 100-bp paired-end reads being obtained (Macrogen, Daejeon, Korea). Data were generated in the standard Sanger FastQ format and raw reads were deposited into the DDBJ sequence read archive (DRA, accession number: DRA008246). Phred-type quality scores Q30 were used for quality trimming. RNA-seq reads were mapped against the S. pyogenes strain 5448 genome (accession CP008776) by using the commercially available CLC Genomics workbench (version 9.5.2, CLC Bio, Aarhus, Denmark). Differential expression analyses and global analysis of the RNA-seq expression data were performed using iDEP (http://ge-lab.org/idep/) [64], with the RPKM value of each sample being determined. Results were visualized using volcano plots (iDEP) and Venn diagrams (http://bioinformatics.psb.ugent.be/webtools/Venn/). EdgeR log-transformation was used for clustering and PCA (iDEP). The hierarchical clustering was illustrated by using the average-linkage method with correlation distance (iDEP). The data were also clustered by using k-means with 1,723 genes (k = 4) (iDEP). We classified the DEGs into functional categories based on the bacterial bioinformatics database and analysis resource PATRIC (www.patricbrc.org) [65], which is integrated with information from VFDB (http://www.mgc.ac.cn/VFs/) [66], Victors [67], subsystems technology toolkit (RASTtk) [68, 69], and KEGG map [70]. Genes were also classified into pathways based on BioCyc database [71]. The transcriptomic (RNA-seq) data are summarized in S1 Dataset.
Acknowledgment
We acknowledge the NGS core facility of the Genome Information Research Center at the Research Institute for Microbial Diseases of Osaka University for the support in RNA sequencing and data analysis.