New Pathogenesis Mechanisms and Translational Leads Identified by Multidimensional Analysis of Necrotizing Myositis in Primates

Abstract

A fundamental goal of contemporary biomedical research is to understand the molecular basis of disease pathogenesis and exploit this information to develop targeted and more-effective therapies. Necrotizing myositis caused by the bacterial pathogen Streptococcus pyogenes is a devastating human infection with a high mortality rate and few successful therapeutic options. We used dual transcriptome sequencing (RNA-seq) to analyze the transcriptomes of S. pyogenes and host skeletal muscle recovered contemporaneously from infected nonhuman primates. The in vivo bacterial transcriptome was strikingly remodeled compared to organisms grown in vitro, with significant upregulation of genes contributing to virulence and altered regulation of metabolic genes. The transcriptome of muscle tissue from infected nonhuman primates (NHPs) differed significantly from that of mock-infected animals, due in part to substantial changes in genes contributing to inflammation and host defense processes. We discovered significant positive correlations between group A streptococcus (GAS) virulence factor transcripts and genes involved in the host immune response and inflammation. We also discovered significant correlations between the magnitude of bacterial virulence gene expression in vivo and pathogen fitness, as assessed by previously conducted genome-wide transposon-directed insertion site sequencing (TraDIS). By integrating the bacterial RNA-seq data with the fitness data generated by TraDIS, we discovered five new pathogen genes, namely, S. pyogenes 0281 (Spy0281 [dahA]), ihk-irr, slr, isp, and ciaH, that contribute to necrotizing myositis and confirmed these findings using isogenic deletion-mutant strains. Taken together, our study results provide rich new information about the molecular events occurring in severe invasive infection of primate skeletal muscle that has extensive translational research implications.IMPORTANCE Necrotizing myositis caused by Streptococcus pyogenes has high morbidity and mortality rates and relatively few successful therapeutic options. In addition, there is no licensed human S. pyogenes vaccine. To gain enhanced understanding of the molecular basis of this infection, we employed a multidimensional analysis strategy that included dual RNA-seq and other data derived from experimental infection of nonhuman primates. The data were used to target five streptococcal genes for pathogenesis research, resulting in the unambiguous demonstration that these genes contribute to pathogen-host molecular interactions in necrotizing infections. We exploited fitness data derived from a recently conducted genome-wide transposon mutagenesis study to discover significant correlation between the magnitude of bacterial virulence gene expression in vivo and pathogen fitness. Collectively, our findings have significant implications for translational research, potentially including vaccine efforts.

Keywords: Streptococcus pyogenes; bacterial pathogenesis; bacterial virulence; dual RNA-seq; necrotizing fasciitis; pathogen-host interaction.

FIG 1

Mapping of transcript reads from in vivo samples to GAS and NHP genomes. Numbers of transcript reads per sample are shown. (A) Total number of sequence reads corresponding to each of the three NHPs (hatched columns). Reads simultaneously mapping to both NHP and GAS genomes were excluded (inset). Solid columns represent filtered reads, calculated by subtracting excluded reads from the total number of transcript reads. (B) Number of transcript reads mapping to the reference GAS genome (strain MGAS2221) corresponding to each of the three NHPs. (C) Number of transcript reads corresponding to each of the three NHPs mapping to the NHP genome (http://useast.ensembl.org/Macaca_fascicularis/Info/Annotation). The number of reads mapping to either the GAS genome (panel B) or the genome of mock-infected or infected NHPs (panel C), respectively, was 7.0 × 10⁵ (GAS) or 5.8 × 10⁸/1.9 × 10⁹ for NHP-1, 1.0 × 10⁶ (GAS) or 6.2 × 10⁸/5.8 × 10⁸ for NHP-2, and 2.5 × 10⁶ (GAS) or 6.3 × 10⁸/6.1 × 10⁸ for NHP-3.

FIG 2

Transcriptome signatures corresponding to GAS strains grown in vitro and in vivo. The heat map presents log-transformed normalized transcript counts obtained under the following two experimental conditions: (i) GAS grown in rich medium (in vitro) and harvested at two phases of growth (ME, mid-exponential; ES, early stationary) and (ii) GAS harvested from infected nonhuman primate (NHP) tissue (in vivo). The fold change value cutoff and adjusted P value cutoff were 1.5 and ≤0.05, respectively. (A) Heat map showing expression patterns of genes (n = 472) found to be significantly differentially expressed between in vitro and in vivo samples. Color coding is based on log-transformed normalized count values; blue indicates lower transcript expression, and red indicates higher transcript expression. Also indicated is hierarchical clustering of genes (dendrogram along the y axis) and of in vitro and in vivo GAS samples (dendrogram along the x axis). (B) Heat map highlighting expression profile of genes (n = 110) involved in GAS pathogenesis. Genes are assigned to three broad categories: genes involved in virulence factors, genes encoding transcriptional regulators, and genes involved in responses to stress. Gene names shown in red designate genes subsequently studied in more detail by constructing isogenic deletion-mutant strains.

FIG 3

GAS virulence genes highly transcribed during in vivo infection and upregulated in vivo compared to in vitro. (A) Virulence factor genes that are highly transcribed during in vivo infection are shown. Log-transformed mean expression (normalized) counts are plotted. Genes are considered highly transcribed if their normalized expression values are in the top 75th percentile. (B) The y-axis values represent the mean fold change values determined from combined data from the ME and ES in vitro growth phases. Both genes and operons are represented. The horizontal red line corresponds to 5-fold change, an arbitrarily chosen level designed to highlight the magnitude of change. Virulence genes upregulated ≥5-fold are considered to be substantially upregulated. (C) Transcriptional regulators highly transcribed during in vivo infection. Log-transformed mean expression (normalized) counts are plotted. (D) Transcriptional factors upregulated during in vivo infection compared to in vitro growth. Fold change data from in vivo versus in vitro ME and ES comparisons were averaged, and log-transformed mean log-fold changes are shown (y axis). Gene names shown in red designate genes subsequently studied in more detail by constructing isogenic deletion-mutant strains. The fold change value cutoff and adjusted P value cutoff were 1.5 and ≤0.05, respectively.

FIG 4

Spatial analysis of GAS and NHP sections. (A) Hierarchical clustering of GAS transcript profiles from individual tissue sections based on Euclidean distance. Data obtained from tissue sections 1 and 2 (closest to the inoculation site) are closely related and cluster into one group, and data from sections 3, 4, and 5 cluster into a second distinct group. (B) NHP transcripts from infected tissue. Data from sections 1, 2, 3, and 4 are closely related and cluster into one group. (C) NHP transcripts from infected and mock-infected tissue. Data from all five sections from the mock-infected tissue, denoted by asterisks, are closely related and cluster into one group. In each case, S1 through S5 correspond to data from analogous sections for the three NHPs pooled and analyzed together as individual units. Each cluster is demarcated by a red perimeter. (D) Differentially expressed GAS virulence genes comparing pooled sections 1 and 2 to sections 3, 4, and 5. Twenty virulence genes were differentially expressed. The fold change value cutoff and adjusted P value cutoff were 1.5 and ≤0.05, respectively. (E) Selected significantly upregulated immune response genes are depicted on the left, and selected downregulated extracellular matrix organization and skeletal muscle development genes are shown on the right. (Top panel) Comparison of pooled sections 1 through 4 to section 5. Bottom, comparison of pooled sections 1 and 2 to sections 3 through 5. The fold change value cutoff and adjusted P value cutoff were 1.5 and ≤0.05, respectively.

FIG 5

Upregulated GAS genes that contribute to fitness during infection. The y axis represents the product of the log-fold change (log-fold Δ) calculated as the product of in vivo fold change/in vitro fold change, multiplied by the log-fold change calculated in a previous study using the same NHP experimental methods and infecting GAS strain (46). The x axis is a linear representation of the GAS chromosome. The size of the circles is proportionate to the product of the 2-fold changes. Red circles represent genes chosen for further study. Gene names shown in red designate genes subsequently studied in more detail by constructing isogenic deletion-mutant strains. The fold change value cutoff and adjusted P value cutoff were 1.5 and ≤0.05, respectively. (A) Genes upregulated in vivo compared to in vitro ME growth phase. (B) Genes upregulated in vivo compared to in vitro in ES growth phase. (C) Genes upregulated in vivo compared to in vitro in both ME and ES growth phases.

FIG 6

Assessment of virulence of GAS isogenic mutant strains. (A) Schematic showing the Spy0281 (dahA) chromosomal region. dahA encodes a putative stress-related protein, and covRS encode a two-component system (TCS) involved in virulence. The red arrow marks the transcriptional start site for covRS (97). (B) Strains were grown at 37°C in THY medium. (C) Virulence of wild-type MGAS2221 and isogenic mutant MGAS2221 ΔdahA in a mouse model of necrotizing myositis (n = 20 mice per strain). (D) CFU recovered from the inoculation site of mice. Replicate data (n = 35) are expressed as means ± SEM. P < 0.001, Mann-Whitney U test. (E) Representative gross images of mouse hindlimbs infected with the strain MGAS2221 (left) and isogenic MGAS2221 ΔdahA strain (right). Lesion areas are boxed in white. Scale bar, 1 cm. (F) MGAS2221 and isogenic mutant strain MGAS2221 ΔdahA exposed to purified human PMNs. Percent bacterial survival was assessed at 1 and 3 h as indicated. Means ± SEM of the data from 7 separate experiments are shown. (G) Total number of differentially expressed genes comparing wild-type parental strain MGAS2221 and isogenic mutant strain MGAS2221 ΔdahA calculated using a fold cutoff value of ≥1.5 and a P value of ≤0.05. (H) Virulence of wild-type MGAS2221 and isogenic mutant MGAS2221 Δihk-irr in a mouse model of necrotizing myositis (n = 20 mice per strain). (I) Virulence of wild-type MGAS2221 and isogenic mutant MGAS2221 Δslr, MGAS2221 Δisp, and MGAS2221 ΔciaH in a mouse model of necrotizing myositis (n = 20 mice per strain). (J) Virulence of wild-type and isogenic mutant strains (all derived from parental strain MGAS2221) in an NHP model of necrotizing myositis (n = 4 animals per strain). Lesion volume data are expressed as means ± SEM. P < 0.05, Kruskal-Wallis test. (K) Counts of CFU recovered from the inoculation site of NHPs are expressed as means ± SEM. P < 0.05, Kruskal-Wallis test. All isogenic mutant strains were derived from parental strain MGAS2221.

FIG 7

Model of interaction of pathogen- and host-encoded factors to drive GAS-mediated necrotizing myositis pathology. GAS is well adapted to evade host innate immune responses in NHPs and humans. The model is based on new data presented here and includes selected other GAS and host factors crucial for various stages of invasive infection described in the literature, as detailed below. Note that not all pertinent genes and proteins could be included due to space constraints. (A) GAS initiates an acute inflammatory response caused by host recognition of streptococcal components such as lipoteichoic acid and peptidoglycan via extracellular (TLR1, TLR2, TLR4, TLR5, CR, TREM1, and FcR) and intracellular (TLR8) pattern recognition receptors. GAS subverts PMN activation and recruitment by expressing many factors such as hyaluronic acid capsule and M protein and degrading host inflammatory mediators involved in PMN recruitment. (B) GAS evades phagocytosis by producing many antiphagocytic factors (e.g., capsule, M protein). Phagocytosed GAS evade killing by PMNs by secreting cytotoxins (SLO and SPN) and destroying PMNs migrating to the infection site (SLS). GAS-encoded secreted DNases attenuate antimicrobial activity by degrading neutrophil extracellular traps (NETs). Other secreted factors such as streptococcal inhibitor of complement (SIC) and SpeB provide resistance to host antimicrobial peptides (AMPs). In addition, GAS resists host-induced trace metal starvation by upregulating genes involved in acquisition of zinc, iron, copper, and manganese. (C) Host cell destruction and tissue necrosis. GAS factors such as SLO, SLS, and SpeB directly damage cells and induce apoptosis and necrosis via accelerating cell death by triggering activation of proapoptotic host genes. PMN lysis and GAS virulence factors such as SpeB cause skeletal muscle destruction and necrosis as seen by the downregulation of genes involved in skeletal muscle maintenance and integrity. Proteins encoded by upregulated NHP genes are shown in red and downregulated genes in green. Proteins encoded by significantly upregulated GAS genes are shown in blue. AMP, antimicrobial peptide; PRR, pattern recognition receptor; GAS, group A streptococcus; NLRP, NOD-like receptor; CR, complement receptor; FcR, immunoglobulin receptor; TREM-1, triggering receptor expressed on myeloid cells; PGLYRP, peptidoglycan recognition protein; MPO, myeloperoxidase; BPI, bactericidal/permeability-increasing protein; MYLK3, myosin light chain kinase; MYH6, myosin heavy chain; ACTN2, actinin, alpha; ACTA1, actin, alpha; TCAP, titin cap; GSN, gelsolin; DMD, dystrophin; SGCB, sarcoglycan beta; EGF, epidermal growth factor; CTNNA3, catenin alpha; LAMA2, laminin subunit alpha; ADCY6, adenylate cyclase.