Global epithelial cell transcriptional responses reveal Streptococcus pyogenes Fas regulator activity association with bacterial aggressiveness

Authors


E-mail Bernd.Kreikemeyer@med.uni-rostock.de; Tel. (+49) 381 494 5912; Fax (+49) 381 494 5902.

Summary

The bacterial human pathogen Streptococcus pyogenes (group A streptococci, GAS) is able to adhere to, internalize into and cross-talk on multiple levels with its host cells. To gain insight into the Fas function in pathogenesis we used Affymetrix human genome DNA-arrays to measure temporal and global transcriptional responses of HEp-2 cells infected with M49 S. pyogenes wild-type bacteria and ΔfasX, an isogenic S. pyogenes two-component-signal-transduction system mutant. A modified stringent statistical analysis method identified a total of 86 HEp-2 cell genes as differentially transcribed upon infection over the investigated time course. Increased expression of genes encoding proteins involved in GAS host cell adherence and internalization (fibronectin, integrin-α5) was found as a common response. In contrast to earlier reports investigating other GAS serotype strains, Ras superfamily and RhoA pathways are exploited by M49 GAS, suggesting serotype specific interactions with the host cell cytoskeleton. Despite transcriptional induction, secreted IL-8 levels of ΔfasX mutant infected cells were below those of non-infected cells, indicating an absence of Fas expression could be important for GAS tissue colonization and long-term intracellular persistence. Oppositely, activity of the S. pyogenes Fas-system apparently promotes high adherence and internalization rates, massive cytokine gene transcription and cytokine release, host cell apoptosis via a caspase-2 activation pathway, and cytotoxicity. Thus, the S. pyogenes Fas two-component signal transduction system could be involved in local tissue destruction and general bacterial aggressiveness towards host cells.

Introduction

Streptococcus pyogenes (group A Streptococcus, GAS) is an important Gram-positive bacterial pathogen that exclusively infects humans. Diseases caused by GAS range from mild superficial infections of mucosal surfaces and skin (pharyngitis, impetigo) to occasional life-threatening severe invasive diseases such as streptococcal toxic shock syndrome and necrotizing fasciitis, both associated with high mortality rates (Cunningham, 2000).

Group A Streptococcus adherence to and internalization into host epithelial cells of mucosal surfaces and the skin are key initial steps in infection. In contrast to our detailed knowledge on the molecular mechanisms of these processes (LaPenta et al., 1994; Molinari and Chhatwal, 1998; Courtney et al., 2002; Kreikemeyer et al., 2004a) much less is known about the transcriptional and translational response of the infected host cell. The presence of GAS has been shown to induce inflammatory chemokine gene transcription, chemokine release and upregulation of chemokine ligands from several cell types upon exposure (Darmstadt et al., 1999; Raeder et al., 2001; Goldmann et al., 2003; Veckman et al., 2004). GAS-host cell contact during adherence and internalization leads to clustering of cytoskeleton-associated factors FAK and GTPases (Ozeri et al., 2001), triggers phosphatidylinositol 3-kinase cascades (PI 3-K; Purushothaman et al., 2003), and activates NF-κB and STAT pathways (Miettinen et al., 2000), as well as several other host cell kinases (Pancholi and Fischetti, 1997).

To date, only two systematic attempts have been made to investigate global transcriptional responses in GAS-infected host cells. Nakagawa et al. (2004) used SAGE (serial analysis of gene expression) technology to study interaction of a serotype M6 GAS strain and a paired adherence/internalization-negative mutant with non-phagocytic cells and found that numerous host cell genes were differentially transcribed 3 h post infection (p.i.). Kobayashi et al. (2003), using Affymetrix human genome arrays, investigated interaction of GAS with PMN (polymorphonuclear leukocytes). This study focused on effects on PMN apoptosis pathways and reported a unique pattern of gene expression following phagocytosis of GAS that was not observed with other pathogens (Kobayashi et al., 2003). GAS significantly accelerated PMN apoptosis resulting in pathogen release and survival of the oxidative burst (Kobayashi et al., 2003).

In a natural GAS infection, PMNs are usually not the first host cells encountered. Epithelial cell barriers of mucous membranes and the skin are more relevant for initial contact during bacterial adherence/internalization. Infection of epithelial cells can provide GAS with an environment for long-term persistence, an avoidance of innate immune responses and prevent exposure to antibiotics. Such intracellular sanctuaries could act as the main reservoir for GAS in asymptomatic humans and provide a site from which recurrent GAS infections could arise (Österlund and Engstrand, 1997; Podbielski and Kreikemeyer, 2001; Podbielski et al., 2003).

To persist within host cells, GAS would be expected to downregulate any bacterial gene that would encode a cytotoxic product. If this phenotype changed and cytotoxins were generated, the infected cell would die and release the bacteria promoting a disseminative infection. GAS posses several independent pathways which can lead to host cell damage. These include soluble cytolysins like streptolysin S (Carr et al., 2001; Nizet, 2002; Hynes, 2004), which acts even in larger distances from the initial GAS-host cell contact site, and the GAS secreted cystein protease SpeB as well as other bacterial products that can induce apoptosis (Tsai et al., 1999). Another host cell apoptosis-induction pathway has been described after GAS adherence and internalization and works via induction of mitochondrial dysfunction and subsequent caspase-9 activation (Nakagawa et al., 2001). The most sophisticated system associated with GAS is the cytolysin-mediated translocation (CMT) system. This resembles type III secretion machines from Gram-negative bacteria and acts through streptolysin O forming a pore into the host cell membrane which facilitates passage of the effector NAD-glycohydrolase. The effector molecules diffuse into the host cell cytoplasm where they exert damage by a so far unknown mechanism(s) (Bricker et al., 2002).

The transition from a persistently infected cell to an imbalance that leads to cell death and bacterial release is a complex and dynamic balance between the GAS and the host cell. The pattern of expressed bacterial genes will be critical as will be the response of the host cell to different patterns of GAS gene expression. Changes in bacterial gene expression are controlled by the action of single transcriptional regulators and 2-component-signal-transduction systems (TCSs), some of which act in a growth-phase dependent regulatory network (Kreikemeyer et al., 2003; 2004a). Among the 13 identified GAS TCSs, the FasBCAX regulator system is atypical, because it consists of two sensor kinases (FasB, FasC), and one response regulator (FasA). A putative regulatory RNA (fasX) acts as the main effector of this system. Transcriptional analyses of a ΔfasX-mutant indicated several adhesin and aggressin genes are downregulated while genes for secreted tissue-destructive factors are concomitantly upregulated (Kreikemeyer et al., 2001). Thus, Fas activity resembles the function of the Staphylococcus aureus Agr-quorum sensing system (Novick and Muir, 1999).

The preliminary profile of genes controlled by the Fas system suggests that this regulatory pathway could function as a molecular switch in GAS, which would control a phenotype that either favoured a locally adherent and persisting infection or promoted a tissue destructive phenotype. To test this hypothesis it is necessary to be able to analyse the host cell response to GAS infection with wild-type (wt) and Fas mutant bacteria. Previous studies have shown the value of the genome array technology for deciphering transcriptional changes in host cells after bacterial infection and to gain insight into the dynamic bacteria–host cell interaction (Belcher et al., 2000; Binnicker et al., 2003; Hess et al., 2003; Falkow, 2004). The system of bacterial persistence is multifactorial involving not only adherence and internalization of bacteria but also the interplay of bacteria with the host cell environment and the corresponding changes in gene expression by both cell types. The goal for this study was to elucidate the role of the GAS Fas two-component regulator in this complex process by combining bacterial mutation strategies with host cell transcriptomics and quantitative measurements of host cell factors in order to characterize the temporal and global transcriptional response of GAS-exposed epithelial cells. By comparing the host cell response to bacteria with or without a functional Fas regulator we sought to elucidate the role of this unique GAS regulatory pathway in the ability of GAS to persistently infect host cells.

Results

Adherence and internalization assays

For this study we used a serotype M49 GAS strain which belongs to a phylogenetic lineage that does not display tissue site preference for initial infection (McGregor et al., 2004). The preliminary functional characterization of the ΔfasX-mutation in this serotype M49 GAS strain indicated an effect of the Fas regulon on the interaction with epithelial host cells (Kreikemeyer et al., 2001). To potentially confirm this observation, we compared the adherence and internalization efficiency of both the wt-bacteria and the ΔfasX-mutant to HEp-2 cells (Table 1). Upon exposure of host cells to GAS, the number of adherent bacteria increases for both strains, reaching a maximum at 2 h p.i. Unexpected, at this time point the wt-bacteria revealed significantly higher adherence compared with the ΔfasX-mutant and internalization of the ΔfasX-mutant was found significantly decreased compared with wt-bacteria at all time points studied (Table 1). Using microscopic techniques, these differences could not be accounted for by differences in bacterial chain length or the average number of infected HEp-2 cells (data not shown). The observed decrease of internalization over time is potentially a function of cytotoxicity to the host cells, which would expose the intracellular bacteria to the antibiotics containing cell culture medium and a function of intracellular killing of the internalized bacteria. Our experimental set-up did not discriminate between these processes.

Table 1. Streptococcus pyogenes adherence to and internalization into HEp-2 cells.
TimeStrain% of inoculumStatistical significance
  • Number of adherent and internalized wt-bacteria and ΔfasX-mutant compared with inoculum for the indicated times points.

  • *

    P-value < 0.05.

Adherence and internalization (= 8–12)
0.5 hwt6.1 ± 1.9 
ΔfasX7.1 ± 0.8 
1 hwt15 ± 3.2 
ΔfasX9.4 ± 4.3 
2 hwt62 ± 14*
ΔfasX20 ± 16 
Only internalization (= 8–10)
4 hwt8.2 ± 5.5*
ΔfasX2.0 ± 1.6 
6 hwt4.1 ± 3.5*
ΔfasX0.4 ± 0.2 
8 hwt1.9 ± 2.3*
ΔfasX1.2 ± 0.6 × 10−1 
10 hwt1.4 ± 1.4*
ΔfasX0.9 ± 1.3 × 10−1 
12 hwt4.2 ± 3.9*
ΔfasX5.3 ± 6.6 × 10−2 
14 hwt1.7 ± 1.9*
ΔfasX4.1 ± 3.0 × 10−2 
16 hwt2.1 ± 2.2 × 10−1*
ΔfasX5.0 ± 6.0 × 10−3 
20 hwt2.0 ± 1.9 × 10−1*
ΔfasX3.0 ± 3.4 × 10−3 
24 hwt5.8 ± 6.4 × 10−1*
ΔfasX1.6 ± 1.1 × 10−3 
48 hwt1.1 ± 0.9 × 10−1*
ΔfasX1.5. ± 1.5 × 10−4 

The HEp-2 cell transcriptome and data analyses following GAS-infection

The epithelial cell response to the presence of adherent and intracellular GAS has not previously been determined on a global transcriptome level. In the next series of experiments, we used Affymetrix microarray analysis to monitor the transcriptional changes of HEp-2 cells following exposure to GAS. Two different statistical methods were used for data analyses and revealed quantitative differences. Analyses of data sets with standard methods published by other groups (Kobayashi et al., 2003) revealed 732 genes as differentially transcribed in the infected HEp-2 cells. However, despite the statistical significance of small transcriptional changes determined by the paired Students t-test, the biological significance of such changes is questionable (Murphy, 2002).

Data analyses with a more stringent error function based statistical evaluation method (Supplementary material) identified a total number of 86 differentially transcribed genes, of which the majority was found to be upregulated during infection (90%). This method, based on a trapezoidal error function, reduced the number of differentially transcribed genes and increased the confidence (Fig. S1, Supplementary material) in the results with respect to the biological variation and measurement error. Thus, further data analysis was based on this method. Detailed data of all differentially expressed genes not included in Table 2 are provided in Table S1 (Supplementary material).

Table 2.  Selected differentially transcribed HEp-2 cell genes over the time course of infection.
Signal ofnon-infectedcells (average)Fold change in HEp-2 cellsexposed to wt-bacteriaGene designationAccession numberFold change in HEp-2 cellsexposed to ΔfasX-mutant
2 h p.i.4 h p.i.6 h p.i.8 h p.i.2 h p.i.4 h p.i.6 h p.i.8 h p.i.
  1. Using a modified statistical method, 86 differentially expressed genes in GAS-exposed HEp-2 cells were found when compared with non-infected HEp-2 cells. Thirty-nine of the genes displaying transcriptional changes are contained in this Table and are grouped corresponding to the function of the encoded proteins. The other affected genes with metabolic and unknown function are listed in Table S1 (Supplementary material).

  2. –, no change in transcription.

 Transcription 
49.35.4ZNF297BNM_014007
2856.94.0ZAPBG533558
2463.93.0ATF3NM_001674
257.65.98.0FOSBNM_006732
49.66.66.2DUSP4NM_001394
72.818.2NOR-1U12767
259.65.9COPEBAB017493
24.46.4SNT-1NM_006654
278.27.17.4MAFFAL021977
53.311.4SCML1AI431345
128.56.8TRA2AAW157450
85.35.1GTF2IBC004472
521.65.9SUPT16HAK024072
148.8p66αBF0258915.0
 Inflammation and interferon induction
17.57.825.732.8IL-8AF043337
101.28.821.6IL-6NM_0006006.1
578.75.1GROβM57731.1
24.65.011.89.9GROγNM_002090
15.417.245.1HSP70BNM_002155
26.05.9HLA-DPB1AW402154
554.84.8PTGS2NM_000963
135.78.1ISG20U88964
701.55.08.8IFIT1NM_0015485.1
1297.76.314.1IFIT2AA1310417.0
2782.95.29.6IFIT4AI0754075.0
1998.53.9G1P2NM_005101
 Apoptosis 
403.24.87.8GADD34NM_014330
1174.03.34.6TNF-α iP3NM_006290
678.36.2OASLNM_003733
698.83.86.6PMAIP1AI857639
 Cytoskeleton and matrix proteins
41.49.810.2GEMNM_005261
488.7−7.9PXND86862.1
104.27.015.2PTGER4NM_0009585.3
302.311.9ITG-α 2N95414
925.75.6PLAURU08839.1
49.46.5Claudin4NM_001305
1362.63.7PMP22L03203.1
649.3−5.0GLP2RNM_004246
131.95.9cig5AW189843

For 46 of the identified genes database searches predicted a function for their encoded products [PubMed, Netaffx (Affymetrix)]. A predominant portion (14 genes, 16%) of differentially expressed genes clustered into the category encoding transcription factors.

A comparison of HEp-2 cells infected with wt-bacteria and ΔfasX-mutant revealed a different effect on transcription. Eighty-five per cent of the total 86 differentially regulated genes were exclusively found in HEp-2 cells exposed to wt-bacteria. Only two genes were solely found to be differentially transcribed in cells interacting with the regulatory GAS mutant that were not transcribed in response to wt infection. Thirteen per cent of the genes were equally up- or downregulated during infection with both strains and appeared to represent the common response in the GAS-HEp-2 cell interaction (Table 2).

Transcriptional responses of wt and ΔfasX-mutant infected HEp-2 cells

The initial infection phase of GAS-host cell adherence and internalization and most likely the complete intracellular GAS life cycle requires bacterial interaction with proteins of the host cell cytoskeleton. Microarray analysis revealed a differential transcription of several genes encoding proteins involved in building and regulation of the cytoskeleton (Table 2). In host cells which interact with wt-bacteria, transcription of genes encoding a small GTP-binding protein (GEM) and ITG-α 2 were upregulated during infection. Furthermore, the transcription levels of FN1 (fibronectin) and ITG-α 5 (integrin-α 5) encoding genes were upregulated from 10 h p.i. on in cells interacting with both strains (Table 3). Control studies were carried out using staurosporine-treated HEp-2 cells as well as cells grown to different confluences and expression of ITG-α 2, ITG-α 5 and FN1 measured by real-time polymerase chain reaction (PCR). Only HEp-2 cells exposed to GAS demonstrated changes in expression of these genes (data not shown).

Table 3.  Confirmation of differential gene transcription in GAS-infected HEp-2 cells by real-time PCR.
Gene designationDifferential HEp-2 cell gene expression upon exposure to GAS assessed by quantitative real-time PCR
strain2 h p.i.6 h p.i.10 h p.i.14 h p.i.24 h p.i.
  1. Altered transcript amounts of selected genes in HEp-2 cells exposed to GAS were confirmed by real-time PCR. In expansion to the DNA array analysis, we also investigated additional time points at 10 h p.i., 14 h p.i. and 24 h p.i. Several genes (FN1, ITG-α5, JUN) were only differentially expressed from 10 h p.i. on and others [p66α, expressed sequence tag 5 (EST5)] were found exclusively in HEp-2 cells interacting with the ΔfasX-mutant.

  2. –, no change in transcription.

Transcription factors
p66 αwt
ΔfasX6.2 ± 0.6
c-JUNwt 10.7 ± 0.79.5 ± 5.5
ΔfasX
ATF3wt6.6 ± 2.025.8 ± 5.920.8 ± 2.010.9 ± 4.8
ΔfasX5.0 ± 1.3
NOR-1wt5.7 ± 2.05.1 ± 0.2
ΔfasX
DUSP4wt7.4 ± 1.118.0 ± 0.810.5 ± 1.7
ΔfasX
FOS Bwt10.2 ± 0.97.3 ± 5.828.1 ± 6.67.5 ± 1.9
ΔfasX18.3 ± 1.95.0 ± 2.4
Inflammation
IL-8wt7.8 ± 0.759.8 ± 11.873.3 ± 10.39.9 ± 1.3
ΔfasX5.7 ± 0.46.4 ± 1.49.7 ± 0.47.2 ± 3.6
IL-6wt5.0 ± 0.446.0 ± 15.238.5 ± 0.854.8 ± 15.9
ΔfasX30.1 ± 4.216.5 ± 7.17.6 ± 0.4
ISG20wt48.4 ± 8.030.0 ± 3.924.6 ± 1.1
ΔfasX8.7 ± 1.59.0 ± 2.97.7 ± 1.0
Apoptosis
TNF α IP3wt16.7 ± 5.623.4 ± 6.88.1 ± 3.810.6 ± 7.9
ΔfasX7.4 ± 0.413.0 ± 0.923.1 ± 3.7
GADD34wt12.4 ± 3.27.9 ± 2.03.8 ± 1.6
ΔfasX
OASLwt27.0 ± 16.929.1 ± 5.648.0 ± 21.1
ΔfasX5.6 ± 1.36.1 ± 0.25.4 ± 0.8
PMAIP1wt3.6 ± 1.816.5 ± 10.311.8 ± 0.67.0 ± 3.0
ΔfasX4.7 ± 0.6
Cytoskeleton and matrix proteins
ITG α 5wt12.2 ± 2.410.0 ± 0.148.2 ± 10.9
ΔfasX5.5 ± 2.98.4 ± 1.97.3 ± 0.2
FN1wt23.9 ± 10.221.6 ± 12.836.9 ± 12.2
ΔfasX9.2 ± 3.27.9 ± 4.810.9 ± 6.4
GEMwt18.7 ± 3.1365 ± 956.5 ± 4.4
ΔfasX11.0 ± 5.05.1 ± 0.112.3 ± 1.1
ITG α 2wt6.1 ± 0.58.7 ± 4.86.3 ± 4.516.0 ± 4.8
ΔfasX5.8 ± 0.08.3 ± 3.1
Unknown function
EST5wt
ΔfasX6.1 ± 0.4

A previous study indicated that several host cell genes are differentially transcribed during GAS adherence and internalization (Nakagawa et al., 2004). The transcriptome analysis of GAS-infected HEp-2 cells performed in our study identified that 16% of all differentially transcribed genes fall into the category of transcription factors (Table 2). To validate the microarray results, and extend the analysis up to 24 h p.i., selected transcription factor genes were studied by real-time PCR (Table 3). We found c-JUN, NOR-1 (neuron-derived orphan receptor 1) and DUSP4 (dual specific phosphatase) (Table 3) among those differentially transcribed transcription factor genes that are exclusively changed in wt-infected HEp-2 cells. These genes obviously required an intact GAS Fas-system for activation during bacterial adherence and internalization.

Two transcription factor genes, FosB and ATF3, were found differentially transcribed in HEp-2 cells infected with wt and the ΔfasX-mutant, however, with less extensive changes induced in mutant-infected cells (Table 3). Upregulation of these genes appears to be part of the common cellular response to GAS infection. Two genes, the transcription factor p66α (DNA-methylation machinery) and an EST with unknown function (EST5), were only upregulated in ΔfasX-mutant-infected cells, suggesting an association with loss of fasX function (Table 2, Table S1 in Supplementary material). The latter results were confirmed by real-time PCR (Table 3).

The release of chemokines and cytokines is a measure of infected cells to alert the innate immune response and to recruit an acute inflammatory infiltrate at sites of infection. We found eight genes encoding proteins related to inflammatory responses exclusively upregulated in wt-infected cells, and four inflammation-related genes were changed as part of the common cellular response to GAS infection (Table 2). Genes encoding the pro-inflammatory cytokines IL-6 and IL-8 were significantly upregulated in both array and real-time PCR experiments. Although no induction of IFN-γ and TNF-α gene transcription was detected by the array experiments, we observed transcriptional induction of corresponding responsive genes ISG20 (IFN-stimulated gene factor 20) and TNF-αiP3.

In conjunction with innate immune responses the induction of the host cell death program is another host response that can be used to combat adherent and internalized GAS (Tsai et al., 1999; Nakagawa et al., 2001). Based on array analysis (Table 2) several genes involved in apoptosis were differentially transcribed. These results were confirmed by real-time PCR (Table 3). As part of the common response, transcription of the 2′-5′oligoadenylate synthetase like (OASL) and phorbol-12-myristate-13-acetate-induced protein 1 (PMAIP1) mRNA was upregulated in infected host cells, but at a higher level in HEp-2 cells exposed to wt-bacteria. In contrast, the growth arrest and DNA damage-inducible protein 34 (GADD34) is upregulated in host cells exposed to wt-bacteria and unchanged in cells exposed to the ΔfasX-mutant. Of note, no transcriptional changes of genes encoding caspases could be detected by array or real-time PCR experiments.

Functional assays investigating the correlation of transcriptome and phenotype of GAS-infected epithelial cells

The common upregulation of genes encoding cellular fibronectin and its respective integrin-α5 starting at 10 h p.i. was unexpected, because it would promote additional bacterial adherence and internalization. To investigate whether the enhanced message expression correlated with phenotype, surface fibronectin expression was measured using an ELISA format. The results shown in Fig. S2 (Supplementary material) demonstrate increased abundance of cellular fibronectin on HEp-2 cells exposed to GAS.

Quantitative measurement of cytokines demonstrated that IFN-γ, TNF-α, IL-6 and IL-8 were secreted into the supernatant during infection (Fig. 1A–D). The measured concentrations of IL-6 and IL-8 in the supernatant did not correlate with the transcriptional changes of these cytokines. HEp-2 cells exposed to both strains upregulate the transcription levels of genes encoding these cytokines (Table 3), but the concentration of IL-6 in the supernatant of infected cells is up to 100 times higher than that of IL-8 (Fig. 1C and D). Both strains are able to induce a significantly higher amount of secreted IL-6 compared with non-infected cells (Fig. 1C). Whereas the IL-8 gene is transcribed in higher amounts in cells coincubated with both strains starting 2 h p.i. (Table 3), differences in secreted IL-8 levels in cells exposed to wt-bacteria compared with non-infected cells could be detected only after 14 h p.i. (Fig. 1D). The ΔfasX-mutant induced a different profile, as IL-8 secretion was significantly decreased compared with control cells starting 2 h p.i. and lasting up to 24 h p.i. These results suggested additional translational and post-translational levels of regulation that contribute to the observed cytokine levels.

Figure 1.

Measurement of cytokine release from GAS-exposed HEp-2 cells. Upon exposure of HEp-2 cell cultures to GAS the cytokines IFN-γ (A), TNF-α (B), IL-6 (C) and IL-8 (D) were found to be secreted in the supernatants. The concentration of each cytokine was related to the HEp-2 cell number and calculated to the basis of 106 cells. Significant differences between the HEp-2 cells interacting with wt-bacteria (wt), with ΔfasX-mutant (fas) and non-infected HEp-2 cells (c) were indicated by an asterisk, and differences between host cells infected with wt-bacteria and such interacting with ΔfasX-mutant were marked by a number symbol. The significance was shown for the P-value of < 0.05.

In experiments to monitor host cell viability and apoptosis, the number of viable HEp-2 cells following infection with both GAS strains was determined by utilizing trypan blue staining and LDH-release assays (Figs. S3 and S4, respectively, Supplementary material). Trypan blue staining revealed that the number of non-infected control HEp-2 cells increased threefold during a 48 h incubation period, while the number of viable host cells exposed to wt-bacteria decreased by a factor 2. A significant decrease in cell number was detected by 2 h p.i. and persisted throughout the experiment. In contrast, only a 30% reduction in viable cells was observed in ΔfasX-mutant infected cells, reaching a significant level in comparison with the wt-infected cells at 10 h p.i. This general trend on cell viability was confirmed using an independent LDH-release method (Fig. S4, Supplementary material).

According to apoptosis measurements the two GAS strains and a positive control (staurosporine) induced the cell death program. Apoptosis in the infected cells was associated with induction and activation of caspases-2 and -3 in a time-dependent fashion (Fig. 2A and B). However, caspase-8 and, contrary to previous reports, caspase-9 (Nakagawa et al., 2001; 2004) was not activated by any treatment (data not shown). These results were further confirmed with a second, independent assay method. The appearance of annexin V at the outer leaflet of the membrane and the activation of caspase-3 was quantified by FACS analysis at 8 h and 30 h p.i. (Fig. S5, Supplementary material). After 8 h p.i. up to 30% of all infected cells activated their apoptosis program, reaching numbers as high as 75% at 30 h p.i.

Figure 2.

Quantification of activated caspases in GAS-exposed HEp-2 cells. The caspases-2 (A) and -3 (B) were found to be activated during exposure of HEp-2 cells to GAS. The specific caspase-2 inhibitor Z-VDVAD-FMK reduces the activity of caspases-2 and -3 significantly (C). Significant differences between the HEp-2 cells infected with the GAS and non-infected HEp-2 cells were indicated by an asterisk and differences between host cells interacting with wt-bacteria and those interacting with the ΔfasX-mutant were symbolized by a number symbol. HEp-2 cells treated with staurosporine were marked by a plus sign. The significance was shown for the P-value of < 0.05.

Further studies in which the specific caspase-2 inhibitor (Z-VDVAD-FMK) was added to HEp-2 cells prior to infection with wt-bacteria demonstrated that caspase-3 is exclusively activated via caspase-2 in GAS serotype M49 infected HEp-2 cells. This treatment had no effect on the GAS inherent protease activity (data not shown), however, reduced the activation of caspases-2 and -3 significantly compared with wt-infected host cells without inhibitor (Fig. 2C). Incorporation of a general caspase inhibitor (Z-VAD-FMK) in these experiments produced a similar result (data not shown). Together, these results demonstrate that the serotype M 49 GAS strain induced host cell apoptosis in HEp-2 cells via caspase-2 activation.

Discussion

Group A Streptococcus have the ability to adhere to and become internalized by host cells. The consequences of the infection can be the establishment of an active infection or an asymptomatic carrier state in which a persistent intracellular infection can be detected. Establishing a persistent intracellular state is a complex and dynamic process that will require a precise balance between expression of bacterial genes that are not toxic to the host cell and a pattern of host cell gene expression that does not elicit an effective host response that results in destruction of infected cells.

Previous studies from our laboratory had identified a unique regulatory system consisting of two sensor kinases (FasB, FasC), one response regulator (FasA), and a putative regulatory RNA (fasX) which acts as the main effector of this system. (Kreikemeyer et al., 2001). The profile of genes controlled by the Fas system suggest that this regulatory pathway could function as a molecular switch in GAS, which would control a phenotype that either favoured a locally adherent and persisting infection or promoted a tissue destructive phenotype. To investigate this potential function, we have used a comprehensive global transcriptional analysis of epithelial cells (HEp-2) infected with either wt or a paired Fas mutant of a GAS serotype 49 strain in this study. Because transcriptional analysis using array technology does not always correlate with phenotype the array data collected at a limited number of time points was confirmed using real-time PCR approaches and analysis of accompanying phenotypes. This extensive experimental approach is necessary to obtain a comprehensive understanding of the potential bacterial–host interaction that is occurring.

Our data revealed that GAS wt bacteria adhered to and internalized into HEp-2 cells in larger numbers compared with the ΔfasX-mutant throughout the investigated infection periods. This was unexpected because the preliminary Fas characterization suggested that a functional Fas regulator would lead to decreased adherence and internalization rates resulting from downregulation of bacterial adhesion genes (Kreikemeyer et al., 2001). This phenotypic discrepancy could not be explained in this study and has to be solved by uncovering the GAS genome wide Fas-regulon in future experiments. As a direct consequence of its obviously positive involvement in eukaryotic cell adherence/internalization, Fas promotes a strong cytotoxic and apoptotic phenotype, most probably via transcriptional induction of genes encoding secreted toxins like streptolysin S (Kreikemeyer et al., 2001).

In addition to determining these FasBCAX-driven GAS–host cell interaction parameters transcription analysis of GAS-wt and -mutant exposed epithelial cells revealed several new aspects of the GAS–host cell interaction. Our data suggested that 13% of all differentially transcribed eukaryotic genes appear to represent the common transcriptional changes induced by GAS, irrespective of the presence or absence of a functional FasBCAX operon. Consequently, these transcriptional changes appear to be independent of the number of bacteria attaching to and internalizing into host cells. If HEp-2 cells potentially express Toll-like receptors (TLRs), which could act in bacterial recognition and explain the common core transcriptional changes in the infected cells needs to be determined in future experiments. It has also to be noted that the current experimental set-up does not clearly distinguish between differences in the host cell transcriptome solely related to adherence and internalization differences from those dependent on Fas expression.

One novel aspect of the common host cell response is the transcriptional induction of genes encoding cellular fibronectin (FN1) and its cognate integrin (ITG-α 5). The adherence and internalization of several GAS strains was found to depend on these molecules (Molinari et al., 1997; Courtney et al., 2002; Hynes, 2004; Kreikemeyer et al., 2004a,b). Control experiments have shown that the observed differential transcription of ITG-α 5 and FN1 genes in the presence of GAS is independent of host cell apoptosis and cell-to-cell as well as cell-to-surface contact. Thus, the observed effects exclusively depend on the extra- and intracellular presence of GAS. The fact that transcriptional changes also led to increased amounts of these proteins on the surface of infected cells further supports the biological relevance of this observation. A bacterial infection-dependent upregulation of crucial host adherence factors beneficial for subsequently increased adhesion and/or internalization steps appears to emerge as a common theme, because also respiratory epithelial cells have been shown to enhance expression of an adherence-promoting substance (mucin) upon infection with Bordetella pertussis (Belcher et al., 2000).

Additional GAS induced responses, independent of the Fas-regulator mutation, included the transcriptional activation of membrane-associated signalling cascades which link integrin activity with cytoskeletal re-arrangements. A previous study reported that binding of the serotype M6 GAS to fibronectin is associated to activity of focal adhesion kinase, tyrosine phosphorylation, and two members of the Rho family of GTPases, Rac1 and CDC42, whereas RhoA has no function in this process (Ozeri et al., 2001). However, a global transcriptome analysis of GAS M6 serotype strain-infected HEp-2 cells did not support these findings on the transcription level (Nakagawa et al., 2004).

Distinct to results by Ozeri et al. (2001), HEp-2 cells infected with the serotype M49 GAS strains used in our study upregulate GEM, which belongs to the Ras superfamily (Aresta et al., 2002). GEM associates with GEM-interacting protein (Gmip), which in turn could stimulate the GTPase activity of RhoA, but not those of Rac1 and CDC42 (Aresta et al., 2002). This suggests that RhoA, activated via GEM and Gmip, could play a role in the uptake of the GAS serotype M49 strain, as opposed to the serotype M6 strain that was used by Ozeri et al. (2001). Apparently, different GAS serotypes exploit the host cytoskeleton in different ways. This concept is supported by the fact that serotype M6 and M49 strains adhere and internalize via protein F1/SfbI and protein F2, respectively, in a process dependent or independent of host cell-lamelipoda respectively (Ozeri et al., 2001; Kreikemeyer et al., 2004b; B. Kreikemeyer and M. Rohde, unpubl.). Similar phenotypes and their association with either Rac1/CDC42 or RhoA were also observed in FcγR-mediated (type I) and complement (CR3)-mediated (type II) phagocytosis respectively (Caron and Hall, 1998).

At the phenotypic level significant amounts of secreted TNF-α, Il-6 and IL-8 were detected in the supernatants of wt- and mutant-infected host cells. An unexpected picture emerged from the IL-6 and IL-8 measurements. Transcription of genes encoding these two cytokines as well as the level of released cytokines, were induced in cells infected with both GAS strains. However, the GAS wt induced a much stronger increase throughout the measured time points. Of note, levels of secreted IL-8 in mutant-infected cells were below those of uninfected cells, suggesting a major selective advantage for the extra- and intracellular survival of the GAS fas-mutant, because IL-8 acts in neutrophile recruitment to the site of infection. This effect is likely a result of specific degradation of IL-8 by a Fas-dependent trypsin-like GAS protease. This protease has been demonstrated to be active in GAS-associated necrotizing soft-tissue infections (Hidalgo-Grass et al., 2004), but could also play a role during GAS persistence in the tonsillar ultrastructure. In this scenario Fas would act as a negative regulator for the protease in wt bacteria. Again, such a function does not support the initial hypothesis that Fas-activity promotes secretion of virulence factors (Kreikemeyer et al., 2001). However, this initial hypothesis was only based on a limited number of genes tested for dependence on Fas regulation. Uncovering of the complete Fas-regulon will resolve these issues. Taken together, the cytokine profile suggests GAS wt-bacteria induced a massive pro-inflammatory response. In a tissue environment such a strong response would lead to both cell and tissue destruction at the site of infection, thereby breaching the barriers for the bacteria to reach deeper anatomical compartments.

In addition to host defence mechanisms, many bacterial pathogens either specifically induce or repress host cell apoptosis (Rosenberger and Finlay, 2003). After been successfully invaded by GAS, eukaryotic cells actively switch on their apoptosis program (Tsai et al., 1999; Nakagawa et al., 2001; 2004). Especially in PMNs, GAS alter the apoptosis differentiation program, thereby accelerating apoptosis rates. This mechanism promotes pathogen survival and disease complications (Kobayashi et al., 2003).

In the present study, GAS-exposed HEp-2 cells increased transcription of several genes involved in apoptosis. Effects on apoptosis-related genes are apparently dependent on the number of adherent and internalized bacteria, because the fasX mutant-induced transcriptional changes were less pronounced. Many of these factors (NOR-1, OASL, GADD34, TNF-α iP3, PMAIP1) have direct or indirect influence on the host cell apoptosis program (Cheng et al., 1997; Hollander et al., 1997; Zhou et al., 1997). TNF-α iP3 has also been described as upregulated in B. pertussis-infected BEAS-2b cells (Belcher et al., 2000) and in PMNs infected with S. pyogenes M1 (Kobayashi et al., 2003), suggesting common mechanisms in infections caused by Gram-positive and Gram-negative bacteria.

In spite of the upregulation of PMAIP1 observed in this study caspase-9 activation was not observed, when HEp-2 cells were infected with either the M49 wt or Fas-mutant. Similar findings were obtained in control experiments using HEp-2 cells treated with staurosporine, suggesting that PMAIP1 could be involved in additional HEp-2 cell specific pathways. From all investigated initiator caspases, only caspase-2, which can be activated through binding of TNF-α to its receptors (Paroni et al., 2001), was activated upon presence of our serotype M49 GAS strains suggesting an important role for this enzyme in GAS M49 induced apoptosis initiation. This novel function of caspase-2 for the GAS-associated apoptosis pathway in HEp-2 cells was confirmed by specific inhibition experiments. Apparently, caspase-2 is a key player in GAS M49 induced apoptosis initiation.

The results of our global transcriptional studies using a serotype M49 isolate and those of other investigators, using a serotype M6 isolate, have significant differences in transcriptome profiles (Nakagawa et al., 2004). These differences most probably reflect different global transcription analysis methods used in both studies, which apparently have different sensitivities and specificities, or could be explained by different post-infection time points used in both studies. Our global and temporal analysis revealed most pronounced changes in host cell transcription after 6 h p.i. The 3 h p.i. time point exclusively used by Nakagawa et al. (2004) could be too early and the investigation might have missed significant changes. However, it is also very likely that divergent results simply reflect serotype-specific transcriptional changes in the infected host cells. Apoptosis related genes belong to this category, as has been documented to great detail in the present study and that of Nakagawa et al. (2001; 2004).

In conclusion, our study elucidated for the first time the temporal genome-wide transcriptional impact of the presence of a GAS isolate with average virulence traits on a human epithelial cell line. Although this interaction may result in a clinically uniform picture, various GAS strains apparently have serotype-specific means to severely affect their host cells. The results for our M49 strain concerning the cellular uptake pathways (specific matrix and cytoskeletal proteins) and the final consequences for the host cell (caspase-2 activated apoptosis) clearly differ from previously published observations and therefore, support this concept. The reduced aggressiveness of GAS with a mutated Fas regulator in terms of the host cell response, i.e. decreased cytokine production, apoptosis induction and reduced cytotoxicity, would be consistent with a central function of Fas for the host persistence of the bacteria. Obviously, the bacteria have the choice between an uneventful local persistence (Fas turned off) at the price of low multiplication rates and short-lasting extreme invasiveness (Fas turned on) with high multiplication rates. The recently described low fas-operon expression of intracellular persisting serotype M6 isolates (Podbielski et al., 2003) shows the direction for future measurements of specific host cell and bacterial messages. With the present investigation, the potentially significant host cell target genes for such studies have been identified.

Experimental procedures

Bacterial strains and cell culture conditions

The S. pyogenes serotype M49 strain used in this study is a patient isolate kindly provided by R. Lüttiken (Aachen). This serotype is frequently associated with skin infections, however, it can also be found in intracellular locations of ex vivo material from patients with recurrent tonsillopharyngitis (Podbielski et al., 2003). The ΔfasX-mutant was previously described (Kreikemeyer et al., 2001). Bacteria were grown in Todd-Hewitt broth supplemented with 5% yeast extract and incubated under 5% CO2 atmosphere at 37°C. The human laryngeal epithelial cell line HEp-2 (ATCC, CCL23) was cultured in Dulbecco's modified Eagle's medium supplemented with 10% FCS (both obtained from Life Technologies, Karlsruhe, Germany) at 37°C in an atmosphere containing 5% CO2.

Adherence and invasion assay

A standard antibiotic protection assay was used for quantification of adherent GAS after 0.5 h, 1 h and 2 h (Molinari et al., 1997). Intracellular bacteria were quantified at time points 4 h, 6 h, 8 h, 10 h, 12 h, 14 h, 16 h, 20 h, 24 h and 48 h p.i.

Oligonucleotide microarray analysis

Hybridization of RNA to Affymetrix U133A and U133B chips (Affymetrix, Santa Clara, CA, USA) were performed according to published standard protocols (see Supplementary material).

Two biological replicate experiments were performed to estimate the bias of the gene expression pattern of infected and non-infected HEp-2 cells. Microarrays hybridized with RNA from 2 h of non-infected HEp-2 cells were used as reference chips for the comparison with microarrays hybridized with RNA from 2 h and 4 h of eukaryotic cells exposed to wt-bacteria and ΔfasX-mutant. As a reference for chips hybridized with RNA prepared from 6 h p.i. and 8 h p.i. of both GAS-infected HEp-2 cells we used chips that were hybridized with RNA isolated from non-infected cells 8 h p.i. We also compared the microarray data from 2 h of non-infected HEp-2 cells with those from 8 h of non-infected HEp-2 cells to determine the influence of the extended culture on the non-infected cells. Only such genes which were differentially regulated after infection with wt-bacteria and ΔfasX-mutant infected cells and not differentially present in unequal amounts between the 2 h and 8 h of controls were included in the subsequent statistical analysis.

Microarray data processing and analysis

For the statistical analysis of microarray data we calculated the ratio relating the expression level of a gene under non-infected and GAS-infected conditions. A detailed description of the statistical method used for data analysis is provided as supplementary information. The microarray data generated in this study have been deposited in the Arrayexpress database at the European Bioinformatics Institute (http://www.ebi.ac.uk/arrayexpress) under the Accession number E-MEXP-271.

Real-time PCR

RNA quantification by real-time PCR detection was performed with the ABI Prism 7000 Light Cycler (Applied Biosystems, Darmstadt, Germany). RNAs from at least three independent infection experiments were analysed. Reverse transcription of 5 µg total mRNA was carried out using the Ready To Go kit (Amersham Biosciences, Freiburg, Germany) according to the instructions of the manufacturer. The obtained cDNA was purified using NucleoSpin columns (Machery-Nagel, Düren, Germany). An optimal PCR reaction was established employing the SYBR Green PCR Master Mix (Applied Biosystems, Darmstadt, Germany). The full length coding sequence of the genes of interest were used to create the primer pairs (Primer Express 2.0, Applied Biosystems, Darmstadt, Germany). The nucleotide sequences of the primer pairs (Thermo Electron, Ulm, Germany) are available in Table S2 (Supplementary material). Signals (CT-values) for the genes of interest from each RNA sample were normalized to the signal for GAPDH in the same sample. The fold change of samples of infected HEp-2 cells was calculated relatively to the signal observed for non-infected HEp-2 cells.

We used 2 h p.i., 6 h p.i., 10 h p.i., 14 h p.i. and 24 h p.i. wt-bacteria and ΔfasX-mutant infected HEp-2 cells as samples. Furthermore, both the influence of staurosporine (700 nM, Sigma, Steinheim, Germany) on the eukaryotic cells and the influence of different confluences of non-infected HEp-2 cells on the expression levels of selected genes were investigated.

Determination of cytotoxicity

The cytotoxicity of wt-bacteria and ΔfasX-mutant bacteria during infection compared with non-infected HEp-2 cells was determined by the trypan blue staining method. Briefly, HEp-2 cells were trypsinized and diluted 1:2 in trypan blue (Invitrogen, Karlsruhe, Germany) to distinguish between dead (blue stained) and life (stainless) cells (Sandstroem, 1965). The cell number was subsequently counted in a Neubauer cell chamber (Merck Biosciences, Darmstadt, Germany). A second independent method, LDH-release from damaged cells, was employed according to the instructions of the manufacturer (Roche, Germany) and confirmed data revealed by trypan blue staining.

Measurement of activated caspases

In order to elucidate the time-dependent activation of caspases-2, -3, -8 and -9 in infected and non-infected HEp-2 cells, monolayers were trypsinized and cells subsequently lysed. For quantification of activated caspases the BD ApoAlert Caspase Assay Plate system was used (BDClontech, Heidelberg, Germany). Staurosporin (700 nM, Sigma, Steinheim, Germany) treated HEp-2 cells were employed as positive controls for apoptosis induction (Binnicker et al., 2003). Activated caspase-3 and annexin V were additionally quantified by FACS analysis (Fig. S5, Supplementary material).

Inhibition of selected caspases

The irreversible general caspase inhibitor Z-VAD-FMK (Bachem, Heidelberg, Germany) and the specific caspase-2 inhibitor Z-VDVAD-FMK (Calbiochem, Bad Soden, Germany) were both dissolved in dimethyl sulphoxide and stored at −20°C as stock solutions. Both inhibitors were added to cells 1 h prior to infection with wt-bacteria in a final concentration of 20 µM. After 2 h p.i., both inhibitors were also added to the medium containing 5 mg × l−1 penicillin-streptomycin (see above, Invitrogen, Karlsruhe, Germany). Cells were lysed after 14 h p.i. for quantification of activated caspases (see above).

Measurement of cellular fibronectin

HEp-2 cells were grown in serum-free DMEM and monolayers were infected with bacteria as described above. At 12 h p.i. infected monolayers were washed with PBS and subsequently incubated with a 1:2000 diluted antifibronectin antibody (Biomol, Germany) for 1 h. An anti-rabbit HRP-conjugated secondary antibody (1:5000 dilution) and HRP-development substrate (Bio-Rad, Germany) were used for detection. In parallel wells live cells were counted after the infection period and final values of the fibronectin detection were adjusted to equal cell numbers.

Cytokine measurements

The concentration of GM-CSF, IFN-γ, IL-2, IL-4, IL-6, IL-8, IL-10 and TNF-α was determined in each supernatant from three replicate experiments using the Bio-Plex Human Cytokine Analysis system with Panel A according to the instructions of the manufacturer (Bio-Rad, München, Germany). The concentration of the cytokines was calculated from standard curves prepared for each cytokine, and values were normalized to the number of HEp-2 cells counted in each sample prior to the measurement (pg × 106 cells). Specificity of the results was verified by proving non-existing cross-reactivity of the antibodies included from the kit.

Statistical analysis

The adherence and invasion assay (Table 1) and the cytotoxicity experiments (Fig. S3, Supplementary material) were each repeated 8–12 times and a Student t-test was used for statistical analysis.

All real-time PCRs were repeated three times with RNA from three independent infection experiments (Table 3). The detection of secreted cytokines (Fig. 1) and the quantification of activated caspases (Fig. 2) were repeated three times from individual infection experiments. For analysis of significant differences in these assays a Dunnets Multiple Comparison test was used.

For each test, three significance levels were calculated (0.05, 0.01 and 0.001). However, as a result of the high complexity of the figures only levels < 0.05 were shown.

Acknowledgements

This work was supported by a grant from the DFG (Deutsche Forschungsgemeinschaft) Priority Program 1047 (Ecology of Bacterial Pathogens: Molecular and Evolutionary Aspects) awarded to B.K. (KR 1765/2-1). The authors would like to acknowledge the expert technical assistance of Jana Normann, Yvonne Humboldt and Cordula Lembke. The authors acknowledge Burkhard Krüger for granting access to the Cytofluor 2350 (Millipore, Schwalbach, Germany) and Christian Zimmermann from Bio-Rad Laboratories for providing access to the Bio-Plex Suspension Array system. The authors also thank Michael Boyle (Department of Biology, Juniata College, Huntingdon, PA 16652-2196, USA) for proofreading the manuscript and many helpful comments.

Ancillary