Skip to main content
Download PDF

Analysis of host-pathogen gene association networks reveals patient-specific response to streptococcal and polymicrobial necrotising soft tissue infections

BMC Medicine volume 20, Article number: 173 (2022) Cite this article

Abstract

Background

Necrotising soft tissue infections (NSTIs) are rapidly progressing bacterial infections usually caused by either several pathogens in unison (polymicrobial infections) or Streptococcus pyogenes (mono-microbial infection). These infections are rare and are associated with high mortality rates. However, the underlying pathogenic mechanisms in this heterogeneous group remain elusive.

Methods

In this study, we built interactomes at both the population and individual levels consisting of host-pathogen interactions inferred from dual RNA-Seq gene transcriptomic profiles of the biopsies from NSTI patients.

Results

NSTI type-specific responses in the host were uncovered. The S. pyogenes mono-microbial subnetwork was enriched with host genes annotated with involved in cytokine production and regulation of response to stress. The polymicrobial network consisted of several significant associations between different species (S. pyogenes, Porphyromonas asaccharolytica and Escherichia coli) and host genes. The host genes associated with S. pyogenes in this subnetwork were characterised by cellular response to cytokines. We further found several virulence factors including hyaluronan synthase, Sic1, Isp, SagF, SagG, ScfAB-operon, Fba and genes upstream and downstream of EndoS along with bacterial housekeeping genes interacting with the human stress and immune response in various subnetworks between host and pathogen.

Conclusions

At the population level, we found aetiology-dependent responses showing the potential modes of entry and immune evasion strategies employed by S. pyogenes, congruent with general cellular processes such as differentiation and proliferation. After stratifying the patients based on the subject-specific networks to study the patient-specific response, we observed different patient groups with different collagens, cytoskeleton and actin monomers in association with virulence factors, immunogenic proteins and housekeeping genes which we utilised to postulate differing modes of entry and immune evasion for different bacteria in relationship to the patients’ phenotype.

Peer Review reports

Background

Necrotising soft tissue infections (NSTI) are devastating bacterial infections characterised by impairment and injury in any layer of the soft tissue compartment, extending from the epidermis to the deep musculature [1, 2]. These infections are relatively rare (0.2 to 15.5 per 100,000 people/year) but their aggressive nature poses severe threats due to the high risk of mortality and long-term disability which often results from the extensive tissue loss and amputations often prescribed to control the infections [3] due to the fact that progression is rapid, and early diagnosis is vital for improving the prognosis of affected patients [4,5,6,7].

NSTIs can be caused by either a single bacterial species (monomicrobial NSTI, or Type 2 NSTI) or by multiple species (polymicrobial NSTI, or Type 1 NSTI), and the relative occurrence of the two types of NSTI differs significantly based on the geography and patient characteristics [3, 8].

Streptococcus pyogenes is the most common pathogen in monomicrobial NSTIs [3], but other streptococcal species [9] and Staphylococcus aureus are also known to cause monomicrobial NSTIs [10]. In this study, we only focus on type 2 NSTI caused by S. pyogenes, as well as polymicrobial NSTIs that are associated with a mixture of obligate anaerobic and facultative anaerobic bacteria [11], such as Enterobacteriaceae, Bacteroides spp., Porphyromonas spp., Prevotella spp., Peptostreptococcus spp. and Clostridium spp. [1, 12]

Monomicrobial NSTIs caused by S. pyogenes have been studied extensively and many of the virulence factors and toxins expressed by the bacterium to colonise the host tissue and bypass the host immune defences have been characterised [13]. In contrast, the pathogenic strategies and the complex dynamics of bacterial communities underlying polymicrobial NSTIs are poorly understood. One of the major limitations in our understanding of NSTI at the molecular level (and of bacterial infections in general) is the insufficient information about the web of molecular interaction, also known as interactome, between pathogens and the human host. In contrast with Mendelian diseases, where one or few genes can be directly linked to the disease, bacterial infections arise from the complex interactions between bacteria, the host immune system, predisposition, risk and environmental factors [14].

Interactomics focuses on the representation and the analysis of the interactions between biological features on a global scale [15] using network approaches to simplify a complex system like, in this case, a bacteria-host system, and to summarise it as components (nodes) and interactions (edges) between them [16]. Both nodes and edges can be different in nature, depending on the type of interactome considered. In this study, nodes are human and bacterial genes and edges represent the existence of a correlation between the expression profiles of these genes, thus representing the mutual response of host and pathogen and providing a global view of the observed interactions at the molecular level [17].

The present study builds and expands (on) the data obtained in the largest cohort study of NSTI patients in the world to date, the INFECT study. Thänert et al. [18] analysed dual RNA sequencing of NSTI patient biopsies together with microbial community profiling using 16S rRNA sequencing data [18] collected within the INFECT study [19], and showed that gene expression profiles of tissues from NSTI patients differed significantly between monomicrobial streptococcal and polymicrobial infections, identifying the core inflammatory signatures in both instances.

Here, we used network analysis to explore relationships between co-expressed host and bacterial gene pairs, complementing and expanding the results of Thänert et al. [18] by considering a larger number of samples with the aim to provide insight into the interaction and dynamics between the pathogens and the host and illuminate some of the underlying molecular mechanisms in the pathophysiology of NSTI using a systems biology approach [15, 17].

Methods

Study design

The study is founded in the INFECT study on clinical and pathogenesis in NSTI, where patients were included by prospective enrolment through 4.5 years in five Scandinavian referral hospitals (see Table 1). Study design and presentation of clinical results are detailed elsewhere [6, 19, 20]. The INFECT study is registered at ClinicalTrials.gov (NCT01790698).

Tissue biopsies and plasma samples on the day of hospital admission (day 0) were obtained from patients diagnosed with NSTI and admitted to Karolinska University Hospital in Stockholm, Copenhagen University hospital Rigshospitalet, Blekinge County Council Hospital in Karlskrona, Sahlgrenska University Hospital in Gothenburg and Haukeland University Hospital in Bergen in the framework of the EU project INFECT (https://permedinfect.com/projects/peraid/).

Diagnosis of NSTI was based on the presence of necrotic or deliquescent soft tissue with widespread undermining of the surrounding tissue. Patients were excluded in absence of reports of necrotic or deliquescent tissue. More details on patient characteristics and study design can be found in [19].

Ethical considerations

The INFECT study was conducted in accordance with the Declaration of Helsinki and was approved by the regional Ethical Review Board at the Karolinska Institutet in Stockholm, Sweden (Ethics Permits: 2012/2110-31/2), the National Committee on Health Research Ethics in Copenhagen, Denmark (Ethics permits: 1151739), the regional Ethical Review Board in Gothenburg, Sweden (Ethics permits: 930-12) and Bergen, Norway (2012/2227/REC West). All experiments were performed in accordance with the approved ethics applications specified above. All patients provided written informed consent. The INFECT study is registered at ClinicalTrials.gov (NCT01790698).

Experimental methods

RNA-seq sample preparation and sequencing

This study makes use of RNA-seq samples used in [18, 19] plus additional data not available at the time. All samples were handled and processed as described in [18, 19].

Sample selection

In line with the study by Thänert et al., we retained all those subjects/samples for which dual RNA-seq data (i.e., transcriptomics data for both host and pathogen) was available together with 16S bacterial rRNA gene sequencing data collected on the day of hospital admission: this results in 81 samples available for analysis (Fig. 1).

Fig. 1
figure 1

Flowchart for sample selection, mapping and filtering of dual RNA-seq for the generation of gene expression matrices for S. pyogenes monomicrobial infections (Hstrep -Human, Bstrep -Bacteria) and polymicrobial infections (Hpoly – Human, Bpoly -Bacteria). Bacterial and Human gene expression are measured on the same tissue biopsies from NSTI patients. Clinical information associated with patients and samples\biopsies is given in Table 1

Full size image

We followed the classification established in [18] that assigned the samples to 5 different types of infection based on their associated bacterial composition according to 16S rRNA gene sequencing. Classification was based on average-linkage hierarchical agglomerative clustering using the relative abundance of the identified bacterial communities. The optimal number of clusters in the resulting sample dendrogram was determined using the J-index, and distinct specimen clusters were defined to represent different types of NSTIs. This information was obtained from the Supplementary Data 3 from [18].

The sample size for Staphylococcus, Escherichia/Shigella/Bacteroides and other was not large enough to build robust correlation networks, thus we focused on the two larger groups, namely samples from patients with S. pyogenes and polymicrobial NSTI. We indicate with Hstrep and Hpoly the matrix of human gene expression measured on biopsies obtained from patients with diagnosis S. pyogenes NSTI and polymicrobial NSTI, respectively; we indicate with Bstrep and Bpoly the matrix of bacterial gene expression for S. pyogenes NSTI and polymicrobial NSTI, respectively. Relevant clinical information associated with the samples\biopsies can be found in Table 1

Table 1 Clinical parameters associated with the patients whose biopsies were analysed with dual RNA-seq and used to build the host-pathogen interactome in NSTI
Full size table

Bioinformatics analysis

RNA sequences were mapped against both human and bacterial genomes to obtain gene expression of both the host and resident pathogens. Quality control was performed with the tool FASTQC [21]. The mapping tool Kallisto [22] was used to map the sequences against the human genome (GRCh38 release 91). The same sequences were also mapped against several bacterial genomes using the published pipeline HUMAnN2 [23] from the UniRef database [24]. The final data sets for the Streptococcus monomicrobial classified samples contained 680 human genes and 721 bacterial Uniref90 sequences, and the polymicrobial classified samples contained 680 human genes and 703 bacterial Uniref90 sequences. Details of the methods used can be found in Additional file 1: Section S1.

Gene-gene association network inference

Gene-Gene Association networks were built with an algorithm developed to be robust against variation in sample size and noise called the Probabilistic Context Likelihood of Relatedness on Correlation (PCLRC) algorithm [25, 26]. We used partial correlations obtained from a Gaussian Graphical Model (GMM, GeneNet R package implementation) as a measure of association\covariation between human and bacterial genes [27, 28]. A significant correlation rij between the ith host and jth pathogen gene was established if the corresponding Benjamini-Hochberg adjusted P-value < 0.05 [29].

Functional analysis

TopGO R-package v2.42.0 was used for functional category enrichment analysis [30] using the human genes from each bacterial subnetwork as the target sets of interesting genes and the list of all genes from human genome build GRCh38.p12 downloaded from Ensembl Biomart [31] as the background set. The Biological Process ontology from Gene Ontology was used [32], and the Fisher’s exact test was selected to calculate statistical significance of enrichment for the genes of interest.

Inference of host-pathogen gene association networks at the patient level

We used the Linear Interpolation to Obtain Network Estimates for Single Samples (LIONESS) method to infer the single-sample networks [33]. Estimation was done separately for streptococcal and polymicrobial NSTI as described in [34]. For sample q (containing gene expression profiles for host and pathogen from the same biopsy) out of n samples, the corresponding LIONESS single-sample network is obtained as described in Section 1.10 in Additional file 1. The networks have been estimated using the same approach described in Network Inference section using partial correlations.

Clustering of patients based on single-sample networks

Each m × m single-sample network can be reduced to a ½m(m-1) × 1 vector containing the perturbations (edges) of the host-pathogen gene correlation. We collapsed these vectors in two matrices of size 2556 × 42 and 2691 × 24. For each matrix, pairwise distances (Euclidean) among samples were calculated, and hierarchical clustering was applied using the Ward linkage method [35].

Single-sample network edges were ranked for each group using the edge relevance as defined in Eq. 4 in Additional file 1: Section 3.12. For each group we retained the 10 most relevant associations.

Results

A total of 66 samples were included, comprising 42 monomicrobial Streptococcus pyogenes NSTIs and 24 polymicrobial cases. The most abundant genera detected across all samples are Streptococcus, Fusobacterium, Peptostreptococcus, Parvimonas, Peptoniphilus, Porphyromonas, Anaerococcus, Bacteroides and Escherichia. We detected sequences unique to the following species: Streptococcus pyogenes, Streptococcus dysgalactiae, Escherichia coli, Porphyromonas asaccharolytica, Parvimonas micra and Prevotella oris. The monomicrobial streptococcal NSTI samples were restricted to patient cases with Streptococcus pyogenes infection. A principal component analysis plot of S. pyogenes and polymicrobial NSTI gene expressions and a Random Forest classifier discriminating between them can be found in Additional file 1: Fig. S1.

Host-pathogen gene interaction networks

The interaction networks between human and bacterial genes are shown in Fig. 2. The network specific to monomicrobial S. pyogenes NSTI comprises the interaction of 20 human and 24 Streptococcus pyogenes genes, while the network for polymicrobial NSTI consists of 69 human and 79 bacterial genes.

Fig. 2
figure 2

Interactome network of the host-pathogen gene expression profiles derived from Dual RNA-seq of tissue biopsies of NSTI patients. The central column contains human genes, the left column contains S. pyogenes genes (i.e. bacterial genes found to be associated with human genes in monomicrobial NSTI), the right column contains genes from several bacterial species (i.e. bacterial genes found to be associated with human genes in polymicrobial NSTI). Nodes are colour coded by bacterial species; the node size is proportional to the node degree (connectivity, i.e. the number of associated genes; see Additional File 1, Equation S3). Red edges indicate positive partial correlation; blue edges indicate negative partial correlation; the colour intensity and the edge width are proportional to the magnitude of the partial correlations

Full size image

We observed NSTI type-specific responses in the host, with different sets of human genes highly correlated with bacterial gene expression depending on whether the infection is caused by S. pyogenes or by multiple bacteria. We found the polymicrobial correlation network to be divided into subnetworks with genes from three bacterial species (S. pyogenes, E.coli and P. asaccharolytica) that have a high relative abundance over the samples. While the input gene expression matrix also contained several gene sequences for Parvimonas micra and Prevotella oris, only a single gene interaction with a host gene for each of these species was observed in the resulting networks.

The genes from these association networks were isolated based on species for enrichment analysis. The set of human genes in the S. pyogenes monomicrobial network (consisting of S. pyogenes genes and associated human genes) were significantly enriched (adjusted P-value <0.05) in GO terms for cytokine production (GO:0080134) and regulation of response to stress (GO:0001816), which include genes coding for the interleukin receptors (IL1R2, IL18R1), CD55 and the heat shock proteins, HSPA5 and HSP90B1 (Fig. 3). The set of genes in the S. pyogenes subnetwork from the polymicrobial samples were significantly enriched in the GO term, cellular response to cytokine (GO:0034097), involving the genes for STAT1, IL18R1, POSTN, CXCL9, CXCL5, demonstrating different responses to mono- versus polymicrobial S. pyogenes NSTI.

Fig. 3
figure 3

GO (Gene Ontology) enrichment analysis of the human and bacterial genes found to be associated and visualised in Fig. 2. Results are given for monomicrobial streptococcal (S. pyogenes) and for the polymicrobial NSTI. Gene names listed in blue indicate genes annotated by the GO term also indicated in blue, and gene names in red are annotated by the GO term in red. Genes given in purple colour are annotated by both GO terms. A different type of infection elicits different response patterns in the host

Full size image

We observed a strong negative association between the human gene Zinc-Finger Protein 354B (ZFN354B) and three streptococcal genes, sic1 (Q1J9L2), SpyM3_0968 (Q8K763) and MGAS9429_Spy1542 (Q1JK93) in monomicrobial S. pyogenes. We also observed strong associations of S. pyogenes gene SpyM3_0408 (Q7CFC6) with TATA Box-binding protein-associated factor1D (TAF1D) in polymicrobial infections and with mitochondrially encoded NADH dehydrogenase 4L (MT-ND4L) in monomicrobial infections. In polymicrobial infections, we found correlations between the human pseudogene Ferritin (FTH1P2) and two S. pyogenes genes sagF (Q1JHQ0) and sagG (A2RFD6). S. pyogenes genes of known significance are given in Table 2 and the human genes in Table 3. For all human and bacterial gene interactions and their known functions from Uniprot, we refer you to the Additional file 1: Tables S1, S2, S3 and S4.

Table 2 Overview of the most relevant S. pyogenes genes obtained from the analysis of host-pathogen gene-gene association networks. S. pyogenes could be identified in both NSTI types, mono- and polymicrobial
Full size table
Table 3 Overview of the most relevant human genes obtained from the analysis of host-pathogen gene-gene association networks
Full size table

Host-pathogen gene association networks at the patient level

We used the patient-specific gene-gene correlations to characterise host-pathogen response at a patient level and to stratify patients based on such responses. Patient clusters based on single-sample network edges are shown in Fig. 4 for streptococcal (monomicrobial) (A) and polymicrobial (B) NSTI.

Fig. 4
figure 4

Hierarchical clustering of single sample networks, i.e. network derived at the patient level as perturbation networks (see Additional File 1 Section S1.10). A) Clustering of single sample networks from Monomicrobial (S. pyogenes) NSTI samples (see Discussion in Section 4.4 B) Clustering of single sample networks from Polymicrobial NSTI samples (see Discussion in Section 4.5)

Full size image

In the case of S. pyogenes NSTI, we found 6 distinct groups with 4 to 10 patients each, while for the polymicrobial NSTI, we individuated 4 distinct groups with 5 to 7 patients. For each one of these groups, we retrieved the top 10 most relevant host-pathogen gene associations, characterising the particular response to the infection of each patient group. These are shown in Table 4 and Table 5 for S. pyogenes (monomicrobial) and polymicrobial NSTI, respectively. For the polymicrobial case, the top relevant associations involve genes that were mapped to five bacteria species namely S. pyogenes, E. coli, P. asaccharolytica, P. micra and P. oris. We were unable to ascertain associations between the groups and clinical outcomes of patients with significant statistical power due to the lack of sufficient samples per group.

Table 4 List of the top associated gene pairs for the four patient groups obtained by hierarchical clustering of single sample networks in the case of monomicrobial (S. pyogenes) NSTI
Full size table
Table 5 List of the top associated gene pairs for the four patient groups obtained by hierarchical clustering of single sample networks in the case of polymicrobial NSTI
Full size table

Discussion

We explored the differences in the host-pathogen transcriptional responses at both the population and individual levels. At the global level, we investigated the differences between the interactomes associated with S. pyogenes and polymicrobial NSTI, across the entire cohort of NSTI patient tissue samples. The functions associated with the corresponding proteins in the interactome are shown in Fig. 5. To model the phenotypic heterogeneity observed in the host-pathogen interactions and dynamics at the patient level, we constructed patient-specific interactome networks. Our results provide further insights into the molecular mechanisms underlying the pathophysiology at the tissue site of infection in mono- and polymicrobial NSTIs albeit with some limitations.

Fig. 5
figure 5

Figure of protein functions associated to the host-pathogen gene expression profiles derived from Dual RNA-seq of tissue biopsies of NSTI patients represented by the interactome network given in Fig. 2. The central column contains human genes, the left column contains S. pyogenes genes (i.e. bacterial genes found to be associated with human genes in monomicrobial NSTI), the right column contains genes from several bacterial species (i.e. bacterial genes found to be associated with human genes in polymicrobial NSTI). Nodes are colour coded by bacterial species. The node size is proportional to the node degree (connectivity, i.e. the number of associated genes; see Additional File 1, Equation S3). Red edges indicate positive partial correlation; blue edges indicate negative partial correlation; the colour intensity and the edge width are proportional to the value of the partial correlations

Full size image

Host-pathogen interactome for Streptococcal NSTI

In the S. pyogenes monomicrobial interaction network, the most enriched GO categories for the human genes in the network were cytokine production (GO:0080134) and regulation of response to stress (GO:0001816) suggesting immune system defensive mechanisms in the host response to changes in expression of specific streptococcal genes. The host heat shock protein (HSPA5) is associated with streptococcal acyl carrier protein (P63443, acpP) which is involved in fatty acid biosynthesis in lipid metabolism. Eraso et al. have demonstrated the selection of mutations in the fabT gene, another gene involved in fatty acid biosynthesis during necrotising myositis infections in a non-human primate model [56]

Q1J9L2 is 90% identical to Sic1 [36]. Sic has several different mechanisms of actions, interference with complement and other host defences, and has been proposed to play a significant role in streptococcal infections [57]. It was recently shown that the Sic protein from M1 S. pyogenes, a type over-represented among severe invasive cases of NSTI [20], interacts with TLR2 resulting in release of pro-inflammatory cytokines [58]. A study by Kachroo et al. [59] revealed a positive correlation between sic and genes involved in the host immune response and inflammation when they examined the dual RNA-seq transcriptomes of S. pyogenes and host skeletal muscle from infected non-human primates. In addition, vaccination-induced anti-sic antibodies were effective in bacterial clearance in rabbit, mice, and in an ex vivo whole body assay [60].

Sic has been shown to bind to extracellular histones, a group of danger signals released during necrotising tissue damage. The aggregates formed from this interaction have been shown both in vitro and in co-localised biopsies from NSTIs resulting in the neutralisation of host antimicrobial activity [57]. The study by Frick et al. showed that Sic enhances bacterial survival in an animal model of subcutaneous infection [61]. .The increase in the expression of Sic (Q1J9L2) and its role in the inhibition of complement, accompanied by a downregulation of ZFN354B may be a Streptococcal strategy to evade the host innate immune response.

Different bacteria are known to target steps of host gene expression during pathogenic invasion, potentially as a mechanism to modify the expression of inflammatory genes [62]. The ZFN354B gene is also negatively correlated with the gene MGAS9429_Spy1542 (Q1JK93). Although, the sequence of this gene is unannotated, it is located immediately upstream to the gene encoding EndoS. The gene immediately downstream from the gene encoding EndoS, MGAS10270_Spy1608 (H8HD54), is also found in our analysis correlated with the human gene Neurensin (NRSN1). Even though the exact function of these two gene sequences is unclear, it should be noted that the protein EndoS displays endoglycosidase activity on immunoglobulin G (IgG) by hydrolysing the chitobiose core of the asparagine-linked glycan. EndoS modification of IgG antibodies results in impaired Fc-dependent effector function involved in phagocytic killing and elimination of antibody-antigen complexes from circulation [37, 63]. The strong connectivity of Sic, the EndoS-region, and others in the monomicrobial NSTI networks underlines the importance of the immune evasive strategies in S. pyogenes infections. Interestingly, we find that the transcription of both these virulence factors is abated by clindamycin, lending support to contemporary guidelines advocating adjunctive clindamycin treatment in streptococcal NSTIs [64].

LRRFIP1 gene was found associated with the S. pyogenes gene M1GAS476_1767 (J7MBD1) encoding Fba. Fba is a cell-wall-anchoring, surface-located protein that is found in M-type 1, 2, 4, 22, 28 and 49. These M-types constitute 55 out of the 95 sequenced isolates in the INFECT study [18]. On studying the effects of Fba in relation to the bacterial invasion of and adhesion to HEp-2 cells, Terao et al. inferred that the presence of both Fba and M-protein are required for the most efficient bacterial adhesion and invasion. In addition, the report showed that a Fba mutant displayed lower mortality in a murine skin infection model [38].

Host-pathogen interactome for polymicrobial NSTI

In the S. pyogenes subnetwork from the polymicrobial interaction network, the most enriched GO category for the human genes was GO:0034097 (Response to Cytokine) with 6 genes (STAT1, IL18R1, KYNU, POSTN, CXCL9, CXCL5) out of 20 annotated with this GO term. IL18R1, the IL18 receptor complex and the chemokine CXCL5 are associated with bacterial Spx which has an important role in growth, general stress protection and biofilm formation in S. aureus [65]. Therefore, the transcriptional response to cytokines in the host is associated with a change in the regulation of transcription in S. pyogenes that may impact its ability to form biofilm and modulate its response to stress.

The human gene Ferritin heavy chain 1 Pseudogene 2 (FTH1P2) was found associated with three S. pyogenes genes lbp (Q9ZHG8), sagF (Q1JHQ0) and sagG (A2RFD6). Although fth1p2 is a pseudogene, studies have recognised it to compete with fth1, the main intracellular iron-storage protein in the cytoplasm [52]. The study by Terao et al. on adhesion of S. pyogenes to the HEp-2 cells showed that the absence of Lbp significantly lowered the efficiency of adhesion to epithelial cells [45]. More studies have indicated that the primary function of Lbp is as a zinc-scavenger and referred to the gene as adcA [66, 67]. The genes sagF and sagG are two out of the nine genes that form the Streptolysin S-operon. The toxin Streptolysin S has been shown to be responsible for the Beta-haemolysis observed by S. pyogenes and has also been implicated in NSTI pathogenesis [46, 68,69,70]. The association of these S. pyogenes genes to fth1p2 is unclear.

Enrichment of immune-related host genes was also found in 6 out of the 12 host genes in the P. asaccharolytica subnetwork which are annotated with the GO category - Immune system process (GO:0002376). E. coli gene interactions with host genes did not reveal any enrichment of genes with immune-related functions. In this subnetwork, enrichment of host genes in cellular localisation and intracellular protein transport was observed and several of the host genes have roles in transcriptional regulation. The majority of the co-expressed bacterial genes in the E. coli subnetwork are uncharacterised proteins.

In the P.asaccharolytica subnetwork from the polymicrobial interaction network, the most highly connected human gene is KYNU which is correlated with eight P. asaccharolytica genes, five of which code for ribosomal proteins (rpmF, rplU, rpsR, rplF, rpsG), one is the small heat shock protein Hsp20 (encoded by Poras_0808) and two are proteins of unknown function [71]. Heat shock proteins are chaperones that can interfere with the uncontrolled protein unfolding that occurs under stress, such as the immune response of the host [72]. An increased kynurenine pathway activity has been linked to inflammation and immune activation and has been implicated in diverse diseases such as depression and cancer [73,74,75].

Expression of the host gene, TGFBI, an important cytokine with broad regulatory role in the immune system was also positively correlated with five ribosomal genes (rplU, rpsR, rplF, rpsG, rpsO) in P. asaccharolytica. Two other host genes, SIAE and HINT3, are associated with the ribosomal genes (rplU, rplF) and Hsp20 in this species. SIAE has been functionally associated with autoimmune diseases and preeclampsia [76], and although it has a role in the immune system, it has not previously been studied in the context of an infection. The host gene, SLC11A1, is also associated with the ribosomal gene rplF. SLC11A1 controls natural resistance to infection with intracellular parasites. Pathogen resistance involves sequestration of Fe(2+) and Mn(2+), cofactors of both prokaryotic and eukaryotic catalases and superoxide dismutase, not only to protect the macrophage against its own generation of reactive oxygen species, but also to deny the cations to the pathogen for the synthesis of its protective enzymes.

Differences between host-pathogen interactions in streptococcal infections in monomicrobial and polymicrobial NSTI

The expression of S. pyogenes gene SpyM3_0408 (Q7CFC6) was found to be correlated with the expression of human genes in both monomicrobial and polymicrobial infections. In monomicrobial infections, it was positively correlated with the gene mitochondrially encoded NADH dehydrogenase 4L (MT-ND4L). In polymicrobial infections, it was associated with the gene TATA Box-binding protein-associated factor 1D (TAF1D). The gene SpyM3_0408 (Q7CFC6) itself is transcribed as part of the three-gene scfAB-operon. It was hypothesised by Breton et al. that the scfAB-operon plays an integral role in enhancing adaptation and fitness of S. pyogenes during localised skin infection and potentially in the propagation to other deeper tissue in a genome-wide Tn-seq analysis to identify S. pyogenes genetic determinants necessary for in vivo fitness using a murine model of skin and soft tissue infection. The scfAB-operon is part of S. pyogenes core genome, and the gene encodes for a putative transmembrane protein and was found important at the subepithelial site of infection and for the dissemination of into the bloodstream. Homologues of the scfAB-operon are also found in other pathogenic streptococci and closely related gram-positive pathogens [44].

Although cytokine gene expression was associated with S. pyogenes gene expression in both polymicrobial and monomicrobial infections, the only common host gene involved was IL18R1. This occurrence of IL18R1 in both types of infections stands to reason as the Random Forest (RF) models built from cytokine concentrations in the study by Palma Medina et al. found IL18 to be the least important cytokine to differentiate between monomicrobial and polymicrobial NSTI [77].

The genes representing chemokines CXCL5 and CXCL9 were only found in the S. pyogenes subnetwork in polymicrobial infections and not in monomicrobial infections. Thänert et al. found CXCL9, CXCL10 and CXCL11 to be overexpressed in streptococcal NSTI [18] and a recent study analysing cytokines and chemokines in plasma samples from NSTI patients by Palma Medina et al. found CXCL10 to be a robust biomarker for differentiating between monomicrobial and polymicrobial NSTI infections [77]. In accordance with these studies, our analysis shows a difference in the involvement of these chemokines between monomicrobial and polymicrobial S. pyogenes response. The nature of this network analysis renders any information regarding over-expression or under-expression unascertainable.

Host-pathogen gene associations at the patient level in streptococcal NSTI

Overall, we observed different associations of genes coding for collagen proteins, consistent with the study by Singh et al. which showed that various human pathogens utilise the proteins found in the extracellular matrix (ECM) such as collagen proteins for the invasion of the host. Invasive pathogens infract the basal lamina and degrade the ECM proteins employing various proteases drafted from the host. Pathogens use these abilities to adhere to and invade the host tissue [47].

Group 2 differs from other groups only for perturbation of the association between COL3A1 and A2RGM6. In group 3, we find the association between collagen type V and the gene (F5U6Q2) encoding for a hypothetical protein with high similarity (91.4%) to an immunogenic secreted protein (Isp). This isp gene is located immediately downstream of the ihk-irr TCS and the gene is highly conserved among S. pyogenes strains [41]. The function of Isp remains elusive but Kachroo et al. showed that the isp gene contributed to virulence in a necrotising myositis model in non-human primates. Kachroo et al. also postulated the potential role of isp in cell-wall metabolism based on the CHAP domain located at the carboxy terminus [59].

The P0C0H1 (HasA) is required for hyaluronic acid capsule. The capsule represents an important virulence factor of S. pyogenes [43]. We find this gene associated with collagen VI in groups 2, 3, 5 and 6. Despite Collagen VI having antimicrobial properties [78], it has been shown to be a target of adherence by S. pyogenes for persistent infections and inducing invasions [79]. Immunodetection, in vitro binding assays and electron microscopy have shown S. pyogenes to have a strong affinity to Collagen VI and evolved adhesins with the ability to mediate interactions between S. pyogenes and the host [79]. Studies have reported the possibility of hyaluronic capsule to be a major virulence determinant and is also listed in the PATRIC database as a known virulence factor [80]. Dinkla et al. [81] demonstrated that in rheumatic fever, S. pyogenes had the unique capability to bind and aggregate membrane collagen type IV via the M3 protein or hyaluronic acid capsule in M18 serotype. It has also been shown that the upregulation of hyaluronic acid by S. pyogenes in blood is used as a mechanism to mask surface immunogenic determinants and evade antigen-specific antibodies to avoid bacterial death in blood. Dinkla et al. [82] managed to demonstrate that only the S. pyogenes abundant in hyaluronic acid capsule were capable of surviving in human blood containing high levels of antibodies directed against highly conserved bacterial surface proteins. This association may point to a mode of entry and immune evasion by S. pyogenes.

Host-pathogen gene associations at the patient level in polymicrobial NSTI

Group 4 differs from other groups only for perturbation of the association between the host gene TMSB4X and the P. micra gene A8SND5. The TMSB4X gene plays an important role in the organisation of the cytoskeleton and binds to and sequesters actin monomers (G actin) inhibiting actin polymerisation. A common target of bacterial pathogens is the host cell actin cytoskeleton, a dynamic system of filaments that is central to shape determination, movement, phagocytosis and intracellular trafficking [83]. After invasion, some pathogens remain within a membrane-bound compartment and target actin to subvert membrane trafficking [83] by polymerising actin on their surface and use filament assembly to power intracellular actin-based motility, generating actin comet tails that trail the moving bacteria [83,84,85]. The associated P. micra gene A8SND5 (rplC) highlights the importance of the expression of ribosomal proteins to increase bacterial protein synthesis indicating a possible remodulation of this interplay between host and pathogen.

The association of the E. coli gene encoding for protein E1IVY0 and several human genes are present in all 4 groups. The E1IVY0 is an uncharacterised protein with high similarity (86%) to a serine acetyl transferase from Pantoea ananatis, a plant pathogenic gram-negative, facultatively anaerobic gamma proteobacterium which has been shown to help the bacterium survive oxidative stress conditions [86, 87].

Limitations of the study

Limitations of this study include the (relatively) small number of samples, heterogeneity among patients with respect to their comorbidities, the time of infection and treatments, along with the differences in biopsy sampling, type of tissue, depth of infection and lack of longitudinal samples. Many of these limitations are unavoidable as the biopsies are obtained at a timing when surgery is clinically indicated and in areas where the tissue pathology warrants surgical removal. However, using important quality aspects such as the collection of biopsies by dedicated teams of clinicians using standardised SOPs, a prospective observational study design similar in all participating clinical centres and the employment of highly stringent statistical approaches strengthen this study.

Conclusions

Using a systems biology approach to explore host-pathogen interactions in NSTIs, we postulated several data-driven hypotheses which could be further evaluated experimentally. In attempting to elucidate the mechanisms underlying the progression and proliferation of NSTI infections, this study highlights the heterogeneity in the host-pathogen interactomes and strengthens the rationale for a personalised approach in the clinical management of NSTI patients. Furthermore, the identification of pivotal pathogenetic mechanisms is the first step towards identifying novel targets for intervention and expanding our therapeutic armamentarium.

Availability of data and materials

All the data used in this available at https://zenodo.org/record/5744186#.YlUsVdNBxPZ (DOI: 10.5281/zenodo.5744186).

Abbreviations

AMPs:

Antimicrobial peptides

CTGF:

Connective tissue growth factor

ECM:

Extracellular matrix

GMM:

Gaussian Graphical Model

GO:

Gene Ontology

HSP:

Heat shock protein

IgG:

Immunoglobulin G

isp:

Immunogenic secreted protein

LIONESS:

Linear Interpolation to Obtain Network Estimates for Single Samples

lm:

Laminin

NSTI:

Necrotising soft tissue infection

PCLRC:

Probabilistic Context Likelihood of Relatedness on Correlation

RF:

Random forest

References

  1. Goldstein EJC, Anaya DA, Dellinger EP. Necrotizing soft-tissue infection: diagnosis and management. Clin Infect Dis. 2007;44:705–10.

    Article  Google Scholar 

  2. Bonne SL, Kadri SS. Evaluation and management of necrotizing soft tissue infections. Infect Dis Clin N Am. 2017;31:497–511.

    Article  Google Scholar 

  3. Stevens DL, Bryant AE. Necrotizing soft-tissue infections. N Engl J Med. 2017;377:2253–65.

    PubMed  Article  Google Scholar 

  4. Pham TN, Moore ML, Costa BA, Cuschieri J, Klein MB. Assessment of functional limitation after necrotizing soft tissue infection. J Burn Care Res. 2009;30:301–6.

    PubMed  Article  Google Scholar 

  5. May AK. Skin and soft tissue infections: The New Surgical Infection Society Guidelines. Surg Infect. 2011;12:179–84.

    Article  Google Scholar 

  6. Madsen MB, et al. Patient’s characteristics and outcomes in necrotising soft-tissue infections: results from a Scandinavian, multicentre, prospective cohort study. Intensive Care Med. 2019;45:1241–51.

    CAS  PubMed  Article  Google Scholar 

  7. Chan T, Yaghoubian A, Rosing D, Kaji A, de Virgilio C. Low sensitivity of physical examination findings in necrotizing soft tissue infection is improved with laboratory values: a prospective study. Am J Surg. 2008;196:926–30.

    PubMed  Article  Google Scholar 

  8. Skrede S, Bruun T, Rath E, Oppegaard O. Microbiological etiology of necrotizing soft tissue infections; 2020. p. 53–71. https://doi.org/10.1007/978-3-030-57616-5_5.

    Book  Google Scholar 

  9. Bruun T, et al. Necrotizing soft tissue infections caused by Streptococcus pyogenes and Streptococcus dysgalactiae subsp. equisimilis of groups C and G in western Norway. Clin Microbiol Infect. 2013;19:E545–50.

    CAS  PubMed  Article  Google Scholar 

  10. Miller LG, et al. Necrotizing Fasciitis caused by community-associated methicillin-resistant Staphylococcus aureus in Los Angeles. N Engl J Med. 2005;352:1445–53.

    CAS  PubMed  Article  Google Scholar 

  11. Cocanour CS, et al. Management and novel adjuncts of necrotizing soft tissue infections. Surg Infect. 2017;18:250–72.

    Article  Google Scholar 

  12. Elliott D, Kufera JA, Myers RA. The microbiology of necrotizing soft tissue infections. Am J Surg. 2000;179:361–6.

    CAS  PubMed  Article  Google Scholar 

  13. Johansson L, Thulin P, Low DE, Norrby-Teglund A. Getting under the skin: the immunopathogenesis of Streptococcus pyogenes deep tissue infections. Clin Infect Dis. 2010;51:58–65.

    PubMed  Article  Google Scholar 

  14. Doron S, Gorbach SL. Bacterial infections: overview. In: International Encyclopedia of Public Health: Elsevier; 2008. p. 273–82. https://doi.org/10.1016/B978-012373960-5.00596-7.

    Chapter  Google Scholar 

  15. Ito T, Chiba T, Yoshida M. Exploring the protein interactome using comprehensive two-hybrid projects. Trends Biotechnol. 2001;19:S23–7.

    CAS  PubMed  Article  Google Scholar 

  16. Vidal M, Cusick ME, Barabási A-L. Interactome networks and human disease. Cell. 2011;144:986–98.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  17. Alonso-López D, et al. APID interactomes: providing proteome-based interactomes with controlled quality for multiple species and derived networks. Nucleic Acids Res. 2016;44:W529–35.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  18. Thänert R, et al. Molecular profiling of tissue biopsies reveals unique signatures associated with streptococcal necrotizing soft tissue infections. Nat Commun. 2019;10:3846.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  19. Madsen MB, et al. Necrotizing soft tissue infections - a multicentre, prospective observational study (INFECT): protocol and statistical analysis plan. Acta Anaesthesiol Scand. 2018;62:272–9.

    CAS  PubMed  Article  Google Scholar 

  20. Bruun T, et al. Risk factors and predictors of mortality in streptococcal necrotizing soft-tissue infections: a multicenter prospective study. Clin Infect Dis. 2021;72:293–300.

    CAS  PubMed  Article  Google Scholar 

  21. Simon A, Felix K, Anne S-P, Laura B, Krueger Christel WS. FastQC; 2010.

    Google Scholar 

  22. Bray NL, Pimentel H, Melsted P, Pachter L. Near-optimal probabilistic RNA-seq quantification. Nat Biotechnol. 2016;34:525–7.

    CAS  PubMed  Article  Google Scholar 

  23. Franzosa EA, et al. Species-level functional profiling of metagenomes and metatranscriptomes. Nat Methods. 2018;15:962–8.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  24. Suzek BE, Huang H, McGarvey P, Mazumder R, Wu CH. UniRef: comprehensive and non-redundant UniProt reference clusters. Bioinformatics. 2007;23:1282–8.

    CAS  PubMed  Article  Google Scholar 

  25. Saccenti E, Suarez-Diez M, Luchinat C, Santucci C, Tenori L. Probabilistic networks of blood metabolites in healthy subjects as indicators of latent cardiovascular risk. J Proteome Res. 2015;14:1101–11.

    CAS  PubMed  Article  Google Scholar 

  26. Jahagirdar S, Suarez-Diez M, Saccenti E. Simulation and reconstruction of metabolite–metabolite association networks using a metabolic dynamic model and correlation based algorithms. J Proteome Res. 2019;18:1099–113.

    CAS  PubMed  Article  Google Scholar 

  27. Opgen-Rhein R, Strimmer K. From correlation to causation networks: a simple approximate learning algorithm and its application to high-dimensional plant gene expression data. BMC Syst Biol. 2007;1:37.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  28. Schäfer J, Opgen-Rhein R, Strimmer K. Reverse engineering genetic networks using the GeneNet package. Newsl R Proj. 2006;6/5:50.

    Google Scholar 

  29. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B. 1995;57:289–300.

    Google Scholar 

  30. Alexa A, Rahnenfuhrer J. topGO: Enrichment analysis for gene ontology. R package version 2.42; 2020.

    Google Scholar 

  31. Kinsella RJ, et al. Ensembl BioMarts: a hub for data retrieval across taxonomic space. Database 2011; 2011. p. bar030.

    Google Scholar 

  32. Consortium, G. O. The gene ontology resource: 20 years and still GOing strong. Nucleic Acids Res. 2019;47:D330–8.

    Article  CAS  Google Scholar 

  33. Kuijjer ML, Tung MG, Yuan G, Quackenbush J, Glass K. Estimating sample-specific regulatory networks. iScience. 2019;14:226–40.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  34. Jahagirdar S, Saccenti E. Evaluation of single sample network inference methods for metabolomics-based systems medicine. J Proteome Res. 2021;20:932–49.

    CAS  PubMed  Article  Google Scholar 

  35. Rokach L, Maimon O. Clustering methods. In: Data Mining and Knowledge Discovery Handbook: Springer-Verlag. p. 321–52. https://doi.org/10.1007/0-387-25465-X_15.

  36. Merle NS, Church SE, Fremeaux-Bacchi V, Roumenina LT. Complement system part I–molecular mechanisms of activation and regulation. Front Immunol. 2015;6:262. https://doi.org/10.3389/fimmu.2015.00262.

  37. Collin M. EndoS, a novel secreted protein from Streptococcus pyogenes with endoglycosidase activity on human IgG. EMBO J. 2001;20:3046–55.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  38. Terao Y, et al. Fba, a novel fibronectin-binding protein from Streptococcus pyogenes, promotes bacterial entry into epithelial cells, and the fba gene is positively transcribed under the Mga regulator. Mol Microbiol. 2008;42:75–86.

    Article  Google Scholar 

  39. Engel H, et al. Heteroresistance to fosfomycin is predominant in Streptococcus pneumoniae and depends on the murA1 gene. Antimicrob Agents Chemother. 2013;57:2801–8.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  40. Bernish B, van de Rijn I. Characterization of a two-component system in Streptococcus pyogenes which is involved in regulation of hyaluronic acid production. J Biol Chem. 1999;274:4786–93.

    CAS  PubMed  Article  Google Scholar 

  41. McIver KS, Subbarao S, Kellner EM, Heath AS, Scott JR. Identification of isp, a locus encoding an immunogenic secreted protein conserved among group A streptococci. Infect Immun. 1996;64:2548–55.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  42. UniProt Consortium. UniProt: a worldwide hub of protein knowledge. Nucleic acids research. 2019;47( D1):D506–D515. https://doi.org/10.1093/nar/gky1049.

  43. Deangelis PL, Yang N, Weigel PH. The Streptococcus pyogenes hyaluronan synthase: sequence comparison and conservation among various group A strains. Biochem Biophys Res Commun. 1994;199:1–10.

    CAS  PubMed  Article  Google Scholar 

  44. Le Breton Y, et al. Genome-wide discovery of novel M1T1 group A streptococcal determinants important for fitness and virulence during soft-tissue infection. PLoS Pathog. 2017;13:e1006584.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  45. Terao Y, Kawabata S, Kunitomo E, Nakagawa I, Hamada S. Novel laminin-binding protein of Streptococcus pyogenes , Lbp, is involved in adhesion to epithelial cells. Infect Immun. 2002;70:993–7.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  46. Shumba P, Mairpady Shambat S, Siemens N. The role of Streptococcal and Staphylococcal exotoxins and proteases in human necrotizing soft tissue infections. Toxins (Basel). 2019;11:332.

    CAS  Article  Google Scholar 

  47. Singh B, Fleury C, Jalalvand F, Riesbeck K. Human pathogens utilize host extracellular matrix proteins laminin and collagen for adhesion and invasion of the host. FEMS Microbiol Rev. 2012;36:1122–80.

    CAS  PubMed  Article  Google Scholar 

  48. Kuivaniemi H, Tromp G. Type III collagen (COL3A1): gene and protein structure, tissue distribution, and associated diseases. Gene. 2019;707:151–71.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  49. Greenspan DS, et al. Human collagen gene COL5A1 maps to the q34.2→q34.3 region of chromosome 9, near the locus for nail-patella syndrome. Genomics. 1992;12:836–7.

    CAS  PubMed  Article  Google Scholar 

  50. Fitzgerald J, Rich C, Zhou FH, Hansen U. Three novel collagen VI chains, α4(VI), α5(VI), and α6(VI). J Biol Chem. 2008;283:20170–80.

    CAS  PubMed  Article  Google Scholar 

  51. Glavey SV, et al. Proteomic characterization of human multiple myeloma bone marrow extracellular matrix. Leukemia. 2017;31:2426–34.

    CAS  PubMed  Article  Google Scholar 

  52. Di Sanzo M, Quaresima B, Biamonte F, Palmieri C, Faniello MC. FTH1 pseudogenes in cancer and cell metabolism. Cells. 2020;9:2554.

    PubMed Central  Article  CAS  Google Scholar 

  53. Chang MS, McNinch J, Basu R, Simonet S. Cloning and characterization of the human neutrophil-activating peptide (ENA-78) gene. J Biol Chem. 1994;269:25277–82.

    CAS  PubMed  Article  Google Scholar 

  54. Erdel M, Laich A, Utermann G, Werner ER, Werner-Felmayer G. The human gene encoding SCYB9B, a putative novel CXC chemokine, maps to human chromosome 4q21 like the closely related genes for MIG (SCYB9) and INP10 (SCYB10). Cytogenet Genome Res. 1998;81:271–2.

    CAS  Article  Google Scholar 

  55. Tokunaga R, et al. CXCL9, CXCL10, CXCL11/CXCR3 axis for immune activation – A target for novel cancer therapy. Cancer Treat Rev. 2018;63:40–7.

    CAS  PubMed  Article  Google Scholar 

  56. Eraso JM, et al. Genomic Landscape of intrahost variation in group A Streptococcus: repeated and abundant mutational inactivation of the fabT gene encoding a regulator of fatty acid synthesis. Infect Immun. 2016;84:3268–81.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  57. Westman J, Chakrakodi B, Snäll J, Mörgelin M, Bruun Madsen M, Hyldegaard O, Neumann A, et al. Protein SIC secreted from Streptococcus pyogenes forms complexes with extracellular histones that boost cytokine production. Front Immunol. 2018;9:236. https://doi.org/10.3389/fimmu.2018.00236.

  58. Neumann A, et al. Streptococcal protein SIC activates monocytes and induces inflammation. iScience. 2021;24:102339.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  59. Kachroo P, Eraso JM, Olsen RJ, Zhu L, Kubiak SL, Pruitt L, Yerramilli P, et al. New pathogenesis mechanisms and translational leads identified by multidimensional analysis of necrotizing myositis in primates. MBio. 2020;11(1):e03363–19. https://doi.org/10.1128/mBio.03363-19.

  60. Tan LKK, et al. Vaccine-induced, but not natural immunity, against the Streptococcal inhibitor of complement protects against invasive disease. npj Vaccines. 2021;6:62.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  61. Frick I-M, et al. Streptococcal inhibitor of complement (SIC) modulates fibrinolysis and enhances bacterial survival within fibrin clots. J Biol Chem. 2018;293:13578–91.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  62. Denzer L, Schroten H, Schwerk C. From gene to protein—How bacterial virulence factors manipulate host gene expression during infection. Int J Mol Sci. 2020;21(10):3730. https://doi.org/10.3390/ijms21103730.

  63. Naegeli A, et al. Streptococcus pyogenes evades adaptive immunity through specific IgG glycan hydrolysis. J Exp Med. 2019;216:1615–29.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  64. Sartelli M, et al. 2018 WSES/SIS-E consensus conference: recommendations for the management of skin and soft-tissue infections. World J Emerg Surg. 2018;13:58.

    PubMed  PubMed Central  Article  Google Scholar 

  65. Pamp SJ, Frees D, Engelmann S, Hecker M, Ingmer H. Spx is a global effector impacting stress tolerance and biofilm formation in Staphylococcus aureus. J Bacteriol. 2006;188:4861–70.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  66. Tedde V, Rosini R, Galeotti CL. Zn2+ Uptake in Streptococcus pyogenes: characterization of adcA and lmb null mutants. PLoS One. 2016;11:e0152835.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  67. Bayle L, et al. Zinc uptake by Streptococcus pneumoniae depends on both AdcA and AdcAII and is essential for normal bacterial morphology and virulence. Mol Microbiol. 2011;82:904–16.

    CAS  PubMed  Article  Google Scholar 

  68. Siemens N, Norrby-Teglund A. Shocking superantigens promote establishment of bacterial infection. Proc Natl Acad Sci. 2017;114:10000–2.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  69. Arad G, et al. Binding of superantigen toxins into the CD28 homodimer interface is essential for induction of cytokine genes that mediate lethal shock. PLoS Biol. 2011;9:e1001149.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  70. Chatila T, Geha RS. Signal transduction by microbial superantigens via MHC class II molecules. Immunol Rev. 1993;131:43–59.

    CAS  PubMed  Article  Google Scholar 

  71. Santos F, et al. Maintenance of intracellular hypoxia and adequate heat shock response are essential requirements for pathogenicity and virulence of E ntamoeba histolytica. Cell Microbiol. 2015;17:1037–51.

    CAS  PubMed  Article  Google Scholar 

  72. Colaco CA, Bailey CR, Walker KB, Keeble J. Heat shock proteins: stimulators of innate and acquired immunity. Biomed Res Int. 2013;2013:461230.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  73. Wilson DN. Ribosome-targeting antibiotics and mechanisms of bacterial resistance. Nat Rev Microbiol. 2014;12:35–48.

    CAS  PubMed  Article  Google Scholar 

  74. Dyer MD, et al. The human-bacterial pathogen protein interaction networks of Bacillus anthracis, Francisella tularensis, and Yersinia pestis. PLoS One. 2010;5:e12089.

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  75. Sforzini L, Nettis MA, Mondelli V, Pariante CM. Inflammation in cancer and depression: a starring role for the kynurenine pathway. Psychopharmacology. 2019;236:2997–3011.

    CAS  PubMed  PubMed Central  Google Scholar 

  76. Tsai S, et al. Transcriptional profiling of human placentas from pregnancies complicated by preeclampsia reveals disregulation of sialic acid acetylesterase and immune signalling pathways. Placenta. 2011;32:175–82.

    CAS  PubMed  Article  Google Scholar 

  77. Palma Medina LM, Rath E, Jahagirdar S, Bruun T, Madsen MB, Strålin K, Unge C, et al. Discriminatory plasma biomarkers predict specific clinical phenotypes of necrotizing soft-tissue infections. J Clin Invest. 2021;131(14):e149523. https://doi.org/10.1172/JCI149523.

  78. Abdillahi SM, et al. Collagen VI contains multiple host defense peptides with potent in vivo activity. J Immunol. 2018;201:1007–20.

    CAS  PubMed  Article  Google Scholar 

  79. Bober M, Enochsson C, Collin M, Mörgelin M. Collagen VI is a subepithelial adhesive target for human respiratory tract pathogens. J Innate Immun. 2010;2:160–6.

    CAS  PubMed  Article  Google Scholar 

  80. Davis JJ, et al. The PATRIC Bioinformatics Resource Center: expanding data and analysis capabilities. Nucleic Acids Res. 2019. https://doi.org/10.1093/nar/gkz943.

  81. Dinkla K, et al. Rheumatic fever–associated Streptococcus pyogenes isolates aggregate collagen. J Clin Invest. 2003;111:1905–12.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  82. Dinkla K, et al. Upregulation of capsule enables Streptococcus pyogenes to evade immune recognition by antigen-specific antibodies directed to the G-related α2-macroglobulin-binding protein GRAB located on the bacterial surface. Microbes Infect. 2007;9:922–31.

    CAS  PubMed  Article  Google Scholar 

  83. Haglund CM, Welch MD. Pathogens and polymers: microbe–host interactions illuminate the cytoskeleton. J Cell Biol. 2011;195:7–17.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  84. Welch MD, Way M. Arp2/3-mediated actin-based motility: a tail of pathogen abuse. Cell Host Microbe. 2013;14:242–55.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  85. Truong D, Copeland JW, Brumell JH. Bacterial subversion of host cytoskeletal machinery: hijacking formins and the Arp2/3 complex. BioEssays. 2014;36:687–96.

    CAS  PubMed  Article  Google Scholar 

  86. Coutinho TA, Venter SN. Pantoea ananatis: an unconventional plant pathogen. Mol Plant Pathol. 2009;10:325–35.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  87. Weller-Stuart T, De Maayer P, Coutinho T. Pantoea ananatis : genomic insights into a versatile pathogen. Mol Plant Pathol. 2017;18:1191–8.

    CAS  PubMed  PubMed Central  Article  Google Scholar 

Download references

Acknowledgements

We acknowledge the contribution of Jasper Koehorst in the assistance with the preliminary bioinformatics. We thank clinical and research personnel for providing routine diagnostics, in particular respective Departments of Microbiology at the Clinical Centers, as well as the patients and their relatives.

#INFECT study group (Trond Bruun, Eivind Rath, Torbjørn Nedrebø, Per Arnell Anders Rosen, Morten Hedetoft, Martin B. Madsen, Mattias Svensson, Johanna Snäll, Ylva Karlsson, & Michael Nekludov)

Funding

This work was supported by the Center for Innovative Medicine (CIMED) and Region Stockholm (no. 20180058); the Swedish Research Council (2018-02475); the European Union Seventh Framework Programme (FP7/2007-2013) under the grant agreement 305340 (INFECT project); the Swedish Governmental Agency for Innovation Systems (VINNOVA), Innovation Fund Denmark (8114-000005B) and the Research Council of Norway under the frame of NordForsk (project no. 90456, PerAID); the Swedish Research Council, Innovation Fund Denmark (8113-000009B), the Research Council of Norway, the Netherlands Organisation for Health Research and Development (ZonMW) and DLR Federal Ministry of Education and Research, under the frame of ERA PerMed (project 2018-151, PerMIT); and the Swedish Children’s Cancer Foundation (TJ2018-0128).

Author information

Affiliations

  1. Laboratory of Systems and Synthetic Biology, Wageningen University & Research, Stippeneng 4, 6708, WE, Wageningen, the Netherlands

    Sanjeevan Jahagirdar, Nirupama Benis, Vitor A. P. Martins dos Santos & Edoardo Saccenti

  2. Lifeglimmer GmbH, Markelstraße 38, 12163, Berlin, Germany

    Lorna Morris & Vitor A. P. Martins dos Santos

  3. Present affiliation: Department of Medical Informatics, Amsterdam Public Health Research Institute, Amsterdam UMC, University of Amsterdam, Meibergdreef 9, Amsterdam, Netherlands

    Nirupama Benis

  4. Department of Medicine, Division for infectious diseases, Haukeland University Hospital, Bergen, Norway

    Oddvar Oppegaard & Steinar Skrede

  5. Center for Infectious Medicine, Department of Medicine, Karolinska Institutet, Karolinska University Hospital, Huddinge, Sweden

    Mattias Svenson & Anna Norrby-Teglund

  6. Department of Anesthesia, Centre of Head and Orthopaedics, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark

    Ole Hyldegaard

  7. Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark

    Ole Hyldegaard

  8. Department of Clinical Science, University of Bergen, Bergen, Norway

    Steinar Skrede

Authors
  1. Sanjeevan Jahagirdar
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Lorna Morris
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Nirupama Benis
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Oddvar Oppegaard
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Mattias Svenson
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Ole Hyldegaard
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Steinar Skrede
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Anna Norrby-Teglund
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Vitor A. P. Martins dos Santos
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Edoardo Saccenti
    View author publications

    You can also search for this author in PubMed Google Scholar

Consortia

INFECT Study group

  • Trond Bruun
  • , Eivind Rath
  • , Torbjørn Nedrebø
  • , Per Arnell
  • , Anders Rosen
  • , Morten Hedetoft
  • , Martin B. Madsen
  • , Mattias Svensson
  • , Johanna Snäll
  • , Ylva Karlsson
  •  & Michael Nekludov

Contributions

SJ, LM and ES designed the study; NB performed preliminary bioinformatics analysis; SJ and LM performed the analysis; SJ, LM, OO, MS, OH, SS, ANT, VAdP and ES provided advice and interpretation of results; OO, MS, OH, SS, ANT, VAdP and ES provided Material support; SJ and LM visualised the results; SJ, LM and ES wrote the manuscript; OO, MS, SS, VAdP and ES critical revision. All authors reviewed the manuscript and approved the final version.

Authors’ information

Nirupama Benis: Work was performed while affiliated at Laboratory of Systems and Synthetic Biology, Wageningen University & Research

Corresponding author

Correspondence to Edoardo Saccenti.

Ethics declarations

Ethics approval and consent to participate

The INFECT study was conducted in accordance with the Declaration of Helsinki and with the approval of the regional Ethical Review Board at the Karolinska Institutet in Stockholm. Sweden (Ethics permits: 2012/2110-31/2), the National Committee on Health Research Ethics in Copenhagen, Denmark (Ethics permits: 1151739), the regional Ethical Review Board in Gothenburg, Sweden (Ethics permits: 930-12) and Bergen, Norway (2012/2227/REC West). Specimens from patients admitted to five Scandinavian hospitals, Righospitalet (Copenhagen, Denmark), Karolinska University Hospital (Stockholm, Sweden), Blekingesjukhuset (Karlskrona, Sweden), Sahlgrenska University Hospital (Gothenburg, Sweden) and Haukeland University Hospital (Bergen, Norway), were included in this study.

Consent for publication

Written informed consent was obtained from all patients or their surrogate. The INFECT study is registered at ClinicalTrials.gov (NCT01790698).

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Section S1.

Extended Methods. Figure S1. (A): PCA of gene expression profiles. (B): Random Forest classification using gene expression profiles. Table S1. Corresponding protein functions of all interactions between S.pyogenes and human genes in monomicrobial infections. Table S2. Corresponding protein functions of all interactions between S.pyogenes and human genes in polymicrobial infections. Table S3. Corresponding protein functions of all interactions between P.asaccharolytica and human genes in polymicrobial infections. Table S4. Corresponding protein functions of all interactions between E.coli and human genes in polymicrobial infections.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Jahagirdar, S., Morris, L., Benis, N. et al. Analysis of host-pathogen gene association networks reveals patient-specific response to streptococcal and polymicrobial necrotising soft tissue infections. BMC Med 20, 173 (2022). https://doi.org/10.1186/s12916-022-02355-8

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12916-022-02355-8

Keywords

  • Bacterial infection
  • Dual RNA-seq
  • Polymicrobial infection
  • Single-sample networks
  • Streptococcus pyogenes
  • Transcriptomics

BMC Medicine

ISSN: 1741-7015

Contact us