A pivotal role for the Streptococcus iniae extracellular polysaccharide in triggering proinflammatory cytokines transcription and inducing death in rainbow trout

Authors


  • Editor: Jeff Cole

Correspondence: Avi Eldar, Department of Poultry and Fish Diseases, The Kimron Veterinary Institute, PO Box 12, Bet Dagan 50250, Israel. Tel.: +972 3 968 1760; fax: +972 3 968 1739; e-mail: eldar@agri.huji.ac.il

Abstract

Streptococcus iniae is a major pathogen of fish, causing considerable economic losses in Israel, the United States and the Far East. Containment of mortalities through vaccination was recently compromised due to the emergence of novel vaccine-escape strains that are distinguished from previous strains by their ability to produce large amounts of extracellular polysaccharide (EPS) that is released to the medium. In vitro and in vivo data now indicate that the EPS is a major virulence factor, capable of triggering the proinflammatory cytokine machinery and inducing mortality of fish. Streptococcus iniae EPS might therefore be considered to be responsible for sepsis and death just as lipopolysaccharide is for Gram-negative pathogens.

Introduction

Current opinion perceives sepsis as the consequence of the excessive activation of the innate immune system through Toll-like receptors, ensuing in an uncontrolled release of multiple proinflammatory and anti-inflammatory cytokines that are largely responsible for the experimental and clinical symptoms of sepsis and septic shock (Bhakdi et al., 1991; Anderson et al., 1992; Bone, 1993; Cavaillon, 1995; Wenzel et al., 1996; Medzhitov & Janeway, 1997a, b; Gao et al., 1999; Opal & Cohen, 1999; Sriskandan & Cohen, 1999; Ashare et al., 2005; Bozza et al., 2007). Although heterogeneous bacterial components [including bacterial wall components, peptidoglycan, lipoteichoic acid (LTA) and bacterial DNA (Heumann et al., 1994; Mattsson et al., 1994; de Kimpe et al., 1995; Timmerman et al., 1995; Vallejo et al., 1996; Sparwasser et al., 1997; Kengatharan et al., 1998; Gao et al., 1999; Opal & Cross, 1999)], commonly termed ‘pathogen-associated molecular pattern’ molecules (Medzhitov & Janeway, 1997a, b) have been implicated as initiating these responses, it is widely accepted that, in Gram-negative bacterial sepsis, the pathophysiology basically involves an early and excessive release of lipopolysaccharide (LPS)-induced cytokines (Suffredini et al., 1989; Danner et al., 1991). It is also believed that, among the various cytokines, tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β) and IL-6 are the pivotal factors, mediating reactions associated with clinical deterioration, multiorgan system failure and death (Waage et al., 1991; Anderson et al., 1992; Beutler & Grau, 1993; Bone, 1993, 1994; Casey et al., 1993; Muller-Alouf et al., 1994; Wenzel et al., 1996; Silverstein et al., 1997; Okusawa et al., 1998; Cohen & Abraham, 1999).

Unlike the pathophysiology of shock caused by Gram-negative bacteria, which has been extensively investigated, comparatively little is known about the pathogenesis of the sepsis and shock induced by Gram-positive pathogens and, despite the fact that several Gram-positive bacterial components have been shown to trigger cytokine release by monocytes (Bone, 1993, 1994; Heumann et al., 1994; Mattsson et al., 1994; Timmerman et al., 1995; Vallejo et al., 1996; Kengatharan et al., 1998), a common pattern of bioactive molecules has not been defined. The conviction that LTA is the unequivocal counterpart of LPS in terms of pathogenesis of Gram-positive bacteria (Ginsburg, 2002) is also wavering (Nealon & Mattingly, 1985; Bhakdi et al., 1991; Vallejo et al., 1996; Han et al., 2003). In some instances (i.e. GAS), a number of different active substances that can initiate sepsis have been identified (Bone, 1993; Curtis, 1996). It therefore appears that the triggering molecules of Gram-positive bacteria are heterogeneous, and that these pathogens lack a common single LPS comparable mediator capable of initiating the entire cascade of inflammatory cytokines (Vallejo et al., 1996). Likewise, the cytokine network that accompanies Gram-positive sepsis is uncertain, with relatively few studies suggesting the equivalent involvement of cytokines in Gram-positive and Gram-negative sepsis (Wakabayashi et al., 1991; Timmerman et al., 1995), while other evidence substantiates the possibility of a kinetically and qualitatively different machinery (Anderson et al., 1992; Muller-Alouf et al., 1994; Silverstein et al., 1997; Cohen & Abraham, 1999).

Recent studies from our own laboratory point out the emergence of novel virulent strains of the Gram-positive fish pathogen Streptococcus iniae that are producers of large amounts of free extracellular polysaccharide (EPS). So far, production of EPS has been described exclusively for food-grade lactic acid bacteria (LAB) of industrial interest (Cerning, 1990, 1994; de Vuyst & Degeest, 1999; Broadbent et al., 2003). For these bacteria, it was speculated (Stingele et al., 1996) that EPS synthesis by LAB might be a trait that was carried over in evolution from organisms for which the polysaccharides provided a selective advantage (Rubens et al., 1987). For S. iniae, secretion of large quantities of EPS is advantageous, as it enables the pathogen to evade host humoral immune response that is directed primarily against saccharidic moieties located on the exterior capsular polysaccharidic layer (Eyngor et al., 2008). We also noticed that infection with S. iniae EPS-producing strains results in a stormy and generalized septic disease with high bacterial counts disseminated throughout the organism, suggesting the possibility that EPS is also a virulence factor (Eyngor et al., 2008). This has never been investigated thoroughly.

In light of these unresolved issues, we set out to further analyze the function of the EPS produced by novel strains to obtain a more comprehensive understanding of its role in relationship to the host innate immune response against S. iniae bacterial sepsis of fish. The present study has been predicated on the concept that S. iniae EPS is likely to play a major role in the pathophysiology of the disease, functioning as a crucial inducer of proinflammatory cytokines that are released during sepsis. To pursue this goal, a series of in vitro studies using purified EPS and viable S. iniae EPS-producing strains in coculture with trout macrophages were carried out in an effort to reproduce as closely as possible the in vivo host inflammatory response. We demonstrate here that the introduction of purified EPS and viable S. iniae EPS-producing strains into a trout macrophage culture results in release of TNF-α, IL-1β and IL-6 mRNA transcripts at a magnitude that is higher than that of LPS. In vivo assays demonstrate that administration of EPS results in death of fish in a dose-dependent fashion, associated with significant increase in the transcription levels of the pivotal proinflammatory cytokines network. We therefore conclude that S. iniae EPS is a critical virulence factor and a potent cytokine inducer that is able to initiate the entire cascade of proinflammation, comparable to LPS of Gram-negative bacteria.

Materials and methods

Fish

Rainbow trout, weighting 50 g each, were obtained from a S. iniae-specific pathogen-free facility and maintained in a UV-treated pathogen-free environment at a constant temperature of 16 °C.

Bacterial strains and culture conditions

Streptococcus iniae KFP404 (ADH-positive type II strain; nonproducer of EPS) and KFP 477 (ADH-positive type II strain; EPS producer) are both clinical isolates, recovered in 2000 and 2005 (respectively) from the kidneys of diseased rainbow trout. Staphylococcus caseolyticus KFP 776 is a commensal strain recovered in 2007 from a healthy rainbow trout by striking a skin sample on Baird–Parker agar base (Becton Dickinson, Sparks, MD) supplemented with 0.01% sodium azide. Aeromonas salmonicida ITP 20598 (kindly donated by Dr C. Ghittino, IZS Umbria, Italy) is a virulent strain collected in 2003 from the kidney of a rainbow trout with clinical furunculosis. All bacterial isolates were stored at −70 °C in brain–heart infusion (BHI) broth (Oxoid, Basingstoke, UK) with 15% glycerol. Cultures were routinely grown on Columbia blood agar (Oxoid) at 18 °C.

For infection assays, bacteria were grown for 8 h in BHI broth at 18 °C; OD640 nm was measured with a spectrophotometer (Shimadzu Corporation, Kyoto, Japan), and viable CFU counts were determined. Mid-log-phase cultures (108 CFU) were found to correspond to an OD of 0.30–0.35. Bacterial suspensions were washed twice (with fresh L-15 medium) and concentrated so that, for experiments, approximately 5 × 108 CFU in a 20-μL volume were added to each tissue-culture well [multiplicity of infection (MOI) of 100].

Isolation, purification and structure confirmation of S. iniae EPS

EPS was purified from the supernatant of S. iniae KFP 477 fermented in BHI (Oxoid) supplemented with 3% glucose. Fermentations were carried out in a 20-L fermentor (Novaferm, Sweden) with constant stirring (40 r.p.m.) for 24 h at 27 °C; pH 6.8 was regulated with 2 N NaOH. Bacterial cells were discarded and EPS was obtained as described elsewhere (Eyngor et al., 2008). Briefly, bacterial cells and proteins were removed from the culture by adding an equal volume of trichloroacetic acid (40%) followed by centrifugation (10 000 g for 15 min). Two volumes of ice-cold acetone were then added to the supernatant, and the precipitated EPS was recovered by centrifugation, dissolved in distilled water, and the solution was adjusted to pH 7.0 before dialysis against distilled water for 24 h. Insoluble material was removed by ultracentrifugation, and the supernatant containing the EPS was freeze dried (Christ). Finally, the EPS composition was determined by GC, which revealed a monosaccharide composition of rhamnose–fucose–ribose–arabinose–xylose– mannose–galactose–glucose at a weight ratio of 1.2 : 0.7 : 6.6 : 1.8 : 0.5 : 68.6 : 4.7 : 15.9, as described previously (Eyngor et al., 2008).

The purity of the EPS was determined by measuring the protein and endotoxin contents by conventional silver staining after polyacrylamide gel electrophoresis and by Limulus amebocyte lysate assay (BioWhittaker, Walkersville, MD), respectively. DNA or RNA contaminations were excluded by measuring UV adsorption at 260 and 280 nm.

Cell line, cell culture and infection

The salmonid RTS11, a functional macrophage cell line (Ganassin & Bols, 1998; Brubacher et al., 2000), was a gift from Dr N. Bols (Waterloo, Canada). RTS11 cells were cultured at 18 °C in Leibovitz (L-15) medium (Gibco Laboratories, Grand Island, NY) supplemented with 10% fetal calf serum (Gibco Laboratories), l-glutamine (300 mg L−1), HEPES (1%), penicillin (100 μg mL−1), streptomycin (100 μg mL−1) and amphotericin B (0.25 μg mL−1). The cell line was subcultured every 3 weeks by dividing cells and conditioned medium evenly between two flasks, and adding an equal volume of fresh medium. Cells used in this study had been passaged between 15 and 25 times.

For stimulation of RTS11 macrophages, cells were seeded at 5 × 106 cells per well in a six-well tissue culture-treated plate (Costar), in serum-free and antibiotic-free L-15 medium. Cells were left undisturbed at 18 °C for 48 h to allow for any manipulation-induced gene expression to return to constitutive levels. For infection assays with viable bacteria cells, RTS11 cells were infected with 20 μL of bacterial suspension (MOI of 100) for different time intervals. LPS (50 mg mL−1 of LPS 0127:B8 purchased from Sigma) stimulated cells were used as positive controls. Phosphate-buffered saline (PBS)-stimulated macrophages were used as negative controls. Macrophages with medium alone served as controls for spontaneous cytokine release. At different time intervals (0, 3, 6, 9, 12 and 24 h), cells were harvested from individual wells, aliquoted and kept frozen in liquid nitrogen until RNAs were extracted. All experiments were performed three times (in triplicates).

For EPS stimulation assays, 20 μL of fresh medium containing EPS (50 mg mL−1) was added to each well. Positive and negative controls are the same as listed above. All cytokine induction mixtures were incubated at 18 °C and assessed at the intervals specified above. Experiments comparing cytokine production in response to the viable bacteria EPS were always run concurrently.

Determination of cell viability throughout the experiments

Given that our in vitro experiments may lead to potential variations in cell viability (ongoing macrophage death or damage), we have considered the question of whether possible differences (in terms of cytokine induction) following stimulation with the various bacteria or bacterial components are unrelated to cell viability, or whether a substantial variation in cell number – urging for a recalculation of data on the basis of live cells – is required.

To assess this possibility, macrophage viability during the time frame set was assessed by determining ATP levels (Crouch et al., 1993; Dexter et al., 2003) of stimulated cells (6- and 24-h postinfection/stimulation), using a 3-[4,5-dimethyl thiazol-2-yl]-2,5-diphenyltetrazoliumbromide spectrophotometric assay (Mossmann, 1983). ATP levels of stimulated cells were compared with those on nonsimulated control cells as determined with the (bioluminescent) Vialight Plus assay (Cambrex Bio Science, Rockland, ME).

Primer and probe design

For TNF-α, which possesses two biologically active isoforms (TNF-α1 and TNF-α2; Zou et al., 2002) whose functional roles are poorly understood (Bridle et al., 2006), two sets of primers designed by Purcell et al. (2004) were used. IL-1β primers are also from the same source. IL-6 primers [IL6AZ-1 (TTTGCTCCGCCTCCAACAAG) and IL6AZ-2 (GGTCTTTGACCAGCCCTATCAG)] were designed using the primer express software v2.0 (Applied Biosystems) from sequences deposited in GenBank (accession number DQ866150).

Relative cytokine mRNA levels were determined by normalization of the signal with that for β-actin (Zou et al., 2004). The suitability of β-actin gene as a normalizing gene was compared with that of several other known housekeeping genes, namely: elongation factor-1α (EF-1α), rainbow trout histone H2A and rainbow trout 18S rRNA gene (see Real-time PCR data analysis).

PCR (RT-PCR) for relative measurement of cytokine-specific mRNAs

As a detection system that quantitates fish cytokines is not available, and as most cytokines are transcriptionally regulated (Brorson et al., 1991), cytokine induction and quantification were assessed through cytokine mRNA transcript levels (Livak & Schmittgen, 2001). Unrelated studies, comparing the rise of mRNA transcripts with the (ELISA) quantification of cytokines have shown that these correlate (Cui et al., 2000), and that the assay is reproducible (Stordeur et al., 2002; O'Dwyer et al., 2006).

Total RNA was extracted from isolated RTS-11 cells and pronephros using the peqGOLD TriLFast (Peqlab), following the manufacturer's instructions. RNA was then eluted in 200 μL of RNAse-free water, quantified by Nanodrop (ND 1000) and stored at −80 °C until use. The synthesis of cDNA was initiated by incubating 500 ng of RNA with 5 mM dithiothreitol (ABgene), 1 U of RNasin (Promega) and 0.25 U of RNAse-free Cloned DNAse I (Takara) for 30 min at 37 °C followed by 10 min at 65 °C. Next, 500 ng of oligo (dT) primer and 400 ng of random primers (ABgene) were added and annealed at 70 °C for 5 min, and for 5 min on ice. Finally, 5 mM dNTPs (ABgene), reverse transcriptase buffer and 50 U of Reverse-iT RTase blend (ABgene) were added, and the mixture was incubated for 50 min at 47 °C; the mixture was then incubated at 75 °C for 10 additional minutes. To minimize variation, all samples representing a single time point were run from the same bulk cocktail of cDNA synthesis reagents. The 20-μL cDNA synthesis reactions were diluted to a final volume of 100 μL and stored at −20 °C until use.

Real-time PCR

Quantification of cytokine gene expression was accomplished by real-time PCR using the ABsolute Blue SYBR Green Rox Mixes (Thermo scientific) according the manufacturer's instructions. The PCR mixture was composed of 10 μL SYBR Green Mix, 5 μL cDNA (25 ng of total RNA) and specific primers (final concentration: 300 nM); PCR-grade water was then added to obtain a final volume of 20 μL. The mixtures were run with the following thermal cycling parameters: enzyme activation at 95 °C for 15 min, 40 cycles of denaturation at 95 °C for 15 s, annealing and extension at 60 °C for 1 min. The PCR assay was followed by a melt curve step with a heating rate of 0.5 °C s−1 (for 10 s) and continuous fluorescence measurement. All PCR products were of the predicted molecular weight, indicating that specific amplification had occurred.

Real-time PCR data analysis

The amplification efficiency of each cytokine to β-actin mRNA expression (internal control) was determined by evaluating and analyzing the ΔCt variation (final amount of cDNA template=25 ng per well). Relative quantification (RQ) was obtained using the 2−ΔΔCt method, by adjusting the mRNA cytokine expression to the expression of β-actin mRNA and considering the adjusted expression in the control group as reference (RQ=1) (Livak & Schmittgen, 2001).

The stability of the (housekeeping) β-actin gene throughout the time course of the various assays was assessed by comparing results with those obtained using other known housekeeping (normalizing) genes, namely: EF-1α (forward 5′-CATGTCGACTCCGGCAAGTC-3′; reverse 5′-TGCCTCCGCACTTGTAGATCA-3′; GenBank accession number AF498320; Ooia et al., 2008); rainbow trout histone H2A (forward TCCCCAAGAAGACTGAGAAGG; reverse TTTGTTGAGCTAGGTGGTTGG; TC85036 in TIGR database; Qiu et al., 2008); and rainbow trout 18S rRNA gene (forward TGTGCCGCTAGAGGTGAAATT, reverse CGAACCTCCGACTTTCGTTCT; GenBank accession number AF308735; Løvoll et al., 2007). Results point out that the expression of β-actin as well as that of EF-1α remained constant along the time axis (P>0.05), whereas that of the 18S rRNA gene was lower (P<0.05), thus indicating the suitability of the chosen β-actin as normalizing gene.

Data were analyzed by the Applied Biosystems stepone software v2.0 and expressed as RQ. Descriptive statistics (mean±standard deviation of mean) was carried out to describe RQ in both in vivo and in vitro experiments.

In vivo effects of EPS and LPS: lethality and cytokine mRNA transcription

The in vivo production of proinflammatory cytokines following EPS administration to fish was assessed through a model based on dose–response and time-course parameters. Different dosages of S. iniae EPS (0.55, 1.1 and 2.2 mg per fish, dissolved in 50 μL of PBS) were administered to groups of 30 fish by (slow) injection of 50 μL into the caudal vein; mortalities were monitored for 14 days and dead fish were subjected to complete necroscopic examination. Thereafter, doses of 1.1 mg of EPS (see Results) were injected in each fish (30 fish per group); at fixed intervals (1, 3, 6, 9, 12 and 24 h) three fish were sacrificed and their entire spleens were removed aseptically and frozen in liquid nitrogen until RNA extraction.

Control fish were injected with PBS or LPS (1.1 mg of LPS 0127:B8 per fish). Experimental procedures with live fish were performed in accordance with National Institutes of Health guidelines and according to the principles of the Animal Care Committee of the Kimron Veterinary Institute (Ministry of Agriculture), Israel.

Statistical analysis

Results of all experiments are presented in Figs 1–5 as means±SDs of the dependent variables RQ (Figs 1, 2, 4 and 5) and mortality rate (Fig. 3). Data were obtained from three independent experiments. Data were analyzed by two-way anova for both time and treatment, followed by Duncan's multiple range test (GLM procedures, sas software, version 5). Differences with P-values of 0.05 or lower were considered significant.

Figure 1.

 Kinetics of in vitro mRNA transcription levels of TNF-α, IL-1β and IL-6 by RTS11 macrophages after stimulation with live Streptococcus iniae KFP 477(•), killed S. iniae KFP 477 (×), live Aeromonas salmonicida ITP 20598 (⋄), killed A. salmonicida ITP 20598 (▵), live Staphylococcus caseolyticus KFP 776 (*) and killed S. caseolyticus KFP 776 (○). RTS11 macrophages were lysed in peqGOLD TriLFastTM reagent before RNA extraction and reverse transcriptase-PCR analysis. Data are expressed as the average RQ of cytokine over (β-actin) housekeeping gene expression pooled from three independent experiments.

Figure 2.

 Kinetics of in vitro mRNA transcription levels of TNF-α1, TNF-α2, IL-1β and IL-6 by RTS11 macrophages after stimulation with (50 mg mL−1) of EPS or LPS.

Figure 3.

 Kinetics of EPS and LPS-induced accumulated mortalities in rainbow trout. Different dosages of purified Streptococcus iniae EPS (0.55, 1.1 and 2.2 mg) and LPS (fixed dose of 1.1 mg) were injected in fish (in groups of 30 fish) by (slow) injection of 50 μL into the caudal vein. Mortalities were monitored for 14 days. Data are presented as the accumulated mortality means from three experiments. The differences in mortality rates between the two high EPS groups and the LPS group were statistically nonsignificant (P>0.05; χ2 test for comparison of proportions). However, the mortality induced by the low EPS dose was statistically significant in comparison with the other groups (P<0.001). The rate of the S. iniae EPS- vs. LPS-induced mortalities was statistically significant (P<0.001 using the above χ2 test).

Figure 4.

 Kinetics of in vivo mRNA expression of TNF-α, IL-1β and IL-6 by rainbow trout after stimulation with EPS (1.1 mg per fish). Pronephros was lysed in peqGOLD TriLFast reagent before RNA extraction and reverse transcriptase-PCR analysis. Data are expressed as the RQ means±SD of cytokine over (β-actin) housekeeping gene expression pooled from three independent experiments.

Figure 5.

 Kinetics of in vivo mRNA expression of TNF-α, IL-1β and IL-6 by rainbow trout after stimulation with LPS (1.1 mg per fish). Pronephros was lysed in peqGOLD TriLFastTM reagent before RNA extraction and reverse transcriptase-PCR analysis. Data are expressed as the RQ means±SD of cytokine over (β-actin) housekeeping gene expression pooled from three independent experiments.

A rank test for the RQ values was performed to overcome the uncertainty that they were not distributed normally. In all experiments, significance levels of the rank test (P-values) ranged between 0.05 and 0.001, indicating normal distribution of the data. Also, differences between rank scores resembled those of absolute levels.

Results

Streptococcus iniae and A. salmonicida evoke comparable in vitro patterns of proinflammatory cytokine release from trout RTS11 macrophages

The primary goals in this study were to appraise whether the interaction between pathogenic S. iniae bacteria and rainbow trout macrophages would lead to an increased proinflammatory cytokines response, and to assess whether the ensuing cytokine kinetic patterns approximate those observed after stimulation by a Gram-negative rod that is a LPS producer (the fish pathogen A. salmonicida; positive control). To pursue this, cultures of RTS11 macrophages were cocultured with viable or killed S. iniae and A. salmonicida bacteria and the production of three pivotal proinflammatory cytokines (TNF-α, IL-1β and IL-6) was assessed by quantifying specific RNA transcripts collected at fixed time intervals throughout a 24-h incubation period.

On the whole, the magnitude and the kinetics in the rise of proinflammatory mRNA cytokine transcript levels in the present study resemble those reported previously in comparable (but unrelated) systems (Cui et al., 2000; Khan et al., 2002; Sigh et al., 2004; Segura et al., 2006), and can be summarized as follows:

TNF-α transcription levels

As shown in Fig. 1, infection with both live and killed S. iniae or A. salmonicida induced an early and considerable increase in TNF-α transcription levels. It also appears that, with the exception of live A. salmonicida, an essentially comparable kinetic pattern in the rise of TNF-α1 and TNF-α2 transcription levels was observed after stimulation with the various pathogens, and that transcript levels peak 6–9-h postinfection (live S. iniae or killed S. iniae/killed A. salmonicida, respectively). Instead, whereas during the first 9 h of stimulation with live A. salmonicida, only a relatively moderate (but significant; P<0.001) increase in TNF-α transcription levels (1.7–3.2±0.4-fold increase) was recorded, at later times live A. salmonicida stimulated the production of high amounts of TNF-α transcripts (16±1.8-fold increase at 24-h postinfection). This phenomenon is coupled with decreased cell survival (16% survival in A. salmonicida infection vs. 54% of survival in S. iniae cocultured cells at 24-h postinfection).

However, meticulous analysis of TNF-α mRNA transcription patterns reveals that, depending on (1) bacterial type and (2) bacterial viability, two substantial quantitative differences in TNF-α transcription levels can be perceived. First, live bacteria constantly induced higher levels of TNF-α1 and TNF-α2 mRNA expression compared with heat-killed bacteria (16±1.8- vs. 4.1±0.5- or 10.4±1.6-fold increase for A. salmonicida, P<0.01, at 24 h; 3.7±0.2- or 6.6±0.8- vs. 2.5±0.4- or 5.2±0.6-fold increase for S. iniae, P<0.01, at 6 h). Secondly, infection with A. salmonicida, whether live or dead, induced higher TNF-α transcription levels than infection with S. iniae (16±1.8- or 4.1±0.5- to 10.4±1.6- vs. 3.7±0.2- to 6.6±0.8- or 2.5±0.4- to 5.2±0.6-fold increase in TNF-α1 and TNF-α2 transcription levels for live or dead A. salmonicida or S. iniae, respectively; P<0.05 for live bacteria throughout the experiment and P<0.01 for dead bacteria at 9 h).

LPS (positive control) stimulation of RTS11 macrophages gave rise to a time-dependent increase of TNF-α transcription levels (5.2±0.8- to 5.7±0.6-fold increase for TNF-α1 and TNF-α2, peaking at 9 h; P<0.001) that resembles bacterial stimulation (Fig. 2). No differences in cytokine expression levels were recorded following PBS stimulation. The overall similarity (both from the kinetic and the quantitative aspects) in the increase of TNF-α transcription patterns following LPS stimulation and the coculture of RTS11 trout macrophages with specific pathogens strengthens the reliability of the experimental model. This is further demonstrated by an additional control, consisting of coculture of RTS11 macrophages with live or killed S. caseolyticus KFP 776, a commensal Gram-positive strain recovered from the skin of a healthy rainbow trout. Staphylococcus caseolyticus induced only a minimal increase in TNF-α1 transcription levels (1.4±0.3- or 1.7±0.2-fold increase after coculture with dead or live bacteria, respectively); induction of TNF-α2 transcription (3.6±0.5- or 4.5±0.6-fold increase after coculture with dead or live bacteria, respectively) was also lower than that of A. salmonicida or S. iniae (P<0.01 for both).

IL-1 transcription levels

The amplitude of IL-1 mRNA transcription levels in RTS11 macrophages stimulated by killed S. iniae cells closely resembled that of the same cells cocultured with LPS or A. salmonicida-positive controls (4.5±0.6, 5.4±0.7 SD and 5.3±0.3-fold increase, respectively; all peaking at 9-h postinfection) (Fig. 1). Interestingly, live S. iniae were found to be poor stimulants of IL-1 mRNA transcription, and the (apparent biwave) rise in IL-1 mRNA transcription levels is notably lower than what was observed with other stimulators (P<0.01 or P<0.05). Although relatively low, coculture of RTS11 macrophages with live or killed cells of the commensal S. caseolyticus KFP 776 still induced an increase in transcription levels (2.6±0.5–3.0±0.6-fold increase). But, in contrast to the early rise of proinflammatory cytokine transcriptional levels that was typical of true pathogens, the commensal strain induced only a late and minor increase.

IL-6 transcription levels

For IL-6, LPS and live S. iniae-induced augmented mRNA transcription levels that peaked at an earlier time (6-h poststimulation) compared with heat-killed (A. salmonicida or S. iniae) or live A. salmonicida bacteria (9-h poststimulation). As was the case for TNF-α and IL-1, stimulation with commensal S. caseolyticus induced only a small (and late, for killed cells) increase in levels of mRNA transcription levels (Fig. 1).

Streptococcus iniae EPS induces, in trout RTS11 macrophages, TNF-α, IL-1 and IL-6 mRNA transcripts at a magnitude that exceeds that of LPS

To further delineate the role of S. iniae EPS in the induction of cytokine transcriptions, RTS-11 macrophages were stimulated by purified EPS. The experimental set was analogous to other in vitro experiments and was performed contemporaneously with those where whole bacterial cells were used.

Figure 2, illustrating the magnitude and the kinetics of TNF-α, IL-1 and IL-6 transcriptional levels induced by equivalent concentrations of EPS or LPS during 24 h of incubation, indicates that although both substances are highly effective in inducing an early response (6–9-h poststimulation), EPS is more efficient than LPS in all parameters tested. In terms of TNF-α mRNA transcript induction, S. iniae EPS induced significant activation of macrophage, which resulted in mRNA transcription levels that are practically double those observed after LPS stimulation [5.2±0.8- vs. 10.4±1.4-fold for TNF-α1 (P<0.005), and 5.7±0.6- vs. 10±1.5-fold increase for TNF-α2 (P<0.01)].

Similar results were detected with IL-1 transcription levels [5.3±1.7-fold increase with LPS; 10.3±1.3-fold increase with EPS (P<0.05)]. The highest degree of transcription levels was that of IL-6 (43±7.9-fold increase with EPS; 8.8±2.3-fold increase with LPS).

Differences between cytokine inductions are unrelated to cell viability

The plausibility that the observed differences in levels of cytokine transcription are related to macrophage viability was assessed by determining cellular ATP levels (Mossmann, 1983). When macrophages were stimulated (by LPS or EPS), infected by live S. iniae or A. salmonicida or cocultured with killed S. iniae or A. salmonicida, the percentages of living macrophages at 6-h poststimulation, as determined in quadruplicate experiments, were as follows: 97.3±1.5% (80.7±1.5% at 24 h) – EPS-stimulated cells; 97.3±1.5% (70±2.6% at 24 h) – LPS-stimulated cells; 90±2% (75±2% at 24 h) – coculture with killed A. salmonicida; 75±2% (19.7±3.5% at 24 h) – infection with live A. salmonicida; 98.3±1.5% (74±2.6% at 24 h) – coculture with killed S. iniae; 97.3±1.5% (50.7±3.1% at 24 h) – infection with live S. iniae; 97.7±1.5% (91.7±3.5% at 24 h) – coculture with killed S. caseolyticus commensal strain; 90.7±2.1% (48.7±4.2% at 24 h) – infection with live S. caseolyticus and 99±1% (97.3±1.5% at 24 h) – untreated cells. Thus, there are no apparent or systematic differences in macrophage viability during the initial 6-h incubation period of the experiment, corresponding to the period of cytokine peak.

EPS administration to rainbow trout results in specific death and augmented proinflammatory cytokine transcription

In light of the in vitro proinflammatory cytokine induction of S. iniae EPS, we were next interested in determining whether similar events also occur in vivo, and in revealing the clinical outcomes following EPS inoculation. To accomplish this we first constructed a dose-effect (lethal) model. Mortality rates were affected by both time and group (EPS/LPS dosages). As shown in Fig. 3, EPS induced death of fish in a dose-dependent fashion: low doses (0.55 mg per fish) resulted in 10% mortality, while higher doses resulted in increased mortality rates (P<0.001). Administration of 2.2 mg of EPS per fish resulted in 60% mortality within the first 24 h, while 1.1 mg of EPS per fish yielded 40% mortality during the same period (P<0.01 between these doses). Mortality in fish injected with the higher doses continued for several more days, cumulating in 90% at 144-h postinoculation, resembling that of the LPS-induced septic shock in a mouse model (An et al., 2008) and the (24-h delayed) LPS-induced mortality (80%) of trout observed in the present work (Fig. 3). None of the PBS-injected fish succumbed. Gross pathological findings in dead and moribund fish consisted in discoloration of skin (mainly around the tail), presence of ascitic fluids in the celomic cavity and inflammation with ecchymotic hemorrhages in the gut and peritoneum. Thus, 1.1 mg of EPS per fish was used as the effective dosage in subsequent experiments where cytokine-specific mRNA transcripts levels were assessed.

Relative cytokine mRNA levels analysis revealed that augmentation of specific transcripts was significantly superior to that of the in vitro system. Following inoculation of EPS, TNF-α2 transcription levels peaked at 12-h postinjection (1320-fold increase) and remained elevated for a considerable time (71-fold increase at 24 h), whereas TNF-α1 transcription levels, peaking at 12-h postinjection, were relatively lower (18.1-fold increase) and decreased to a 2.8-fold increase at 24 h (P<0.01 for the difference between the two cytokines) (Fig. 4). LPS injection (Fig. 5) resulted in 115.4-fold increase of TNF-α2 transcripts (remaining elevated throughout the experiment) and 25.9-fold increase of TNF-α1 transcripts (at 9 h). Differences between the two cytokines were nonsignificant. Injection of PBS (negative control) did not affect cytokine transcription levels.

IL-1 transcript level among the EPS-injected fish was increased by 209-fold; IL-6 transcript level of the same fish was increased 560.9-fold. LPS-injected fish showed a 252.1-fold increase of IL-1 transcripts and a 536.7-fold increase of IL-6 transcripts (P<0.001). All of the IL transcripts peaked at 6–9-h postinjection.

The finding that, in contrast to the in vitro results where the two isomers of TNF-α were equally transcripted, the in vivo EPS-induced upregulation of TNF-α2 was significantly higher than that of TNF-α1, is not unexpected. Similar to what was observed in our present study, differential expression of TNF-α isoforms was demonstrated after stimulation with LPS or stimulation of the hemoparasite Trypanoplasma borreli, with a predominant rise in TNF-α2 (Zou et al., 2002; Bridle et al., 2006). Rainbow trout infected with the protozoan parasites Tetracapsuloides brysalmonae (the causative agent of proliferative kidney disease) and Neoparamoeba sp. (causative agent of amoebic gill disease) also displayed an increased expression of TNF-α2 relative to TNF-α1. In contrast, stimulation by IHN virus (causative agent of infectious hematopoietic necrosis) by the protozoan Ichthyophthirius multifiliis (‘white spot’ disease) or by the monogenean parasite Gyrodactylus derjavini (skin fluke) induced an increase in the expression of the TNF-α1 isoform at a higher magnitude than that of the TNF-α2 isoform. Thus, the differential expression of TNF-α isoforms is apparently dependent on the species of pathogen or stimulus, the tissue sampled and the species of fish studied (Purcell et al., 2004; Bridle et al., 2006), and the results obtained here probably reflect the interaction of S. iniae EPS with different cell types, including granulocytes and nongranulocytes present in the blood and organs. Indeed, the use of an in vivo system may help to preserve the integrity of cellular interactions, as well as the effect of lymphocyte-derived factors on proinflammatory cytokine production and, similarly to other studies, ensues in elevated cytokine levels (O'Dwyer et al., 2006; Bozza et al., 2007).

Discussion

The role of EPS in S. iniae pathogenesis is poorly understood. There is evidence, however, that the interaction between the immune system and the EPS produced by this pathogen play an important role in both the development of the disease and protection against the pathogen (Eyngor et al., 2008). Not surprisingly, it is now revealed that EPS is also a key molecule in S. iniae pathogenesis; the failure to control the inflammatory cascade following EPS administration is accompanied by a considerable increase in the secretion of proinflammatory cytokines that are likely to be at the origin of clinical manifestations and poor outcome, both of which are typical of septic shock. Indeed, several inflammatory and infectious diseases are associated with the overproduction of proinflammatory cytokines and chemokines, and the recruitment and activation of different leukocyte populations are a hallmark of acute inflammation (Saukkonen et al., 1990; Welbourn & Young, 1992). These cytokines are believed to mediate responses associated with clinical deterioration, multiorgan system failure and death from septic shock (Waage et al., 1991; Anderson et al., 1992; Bone et al., 1992; Beutler & Grau, 1993; Bone, 1993; Casey et al., 1993; Muller-Alouf et al., 1994; Wenzel et al., 1996; Silverstein et al., 1997; Okusawa et al., 1998; Cohen & Abraham, 1999).

The knowledge that stimulation by various Gram-positive pathogens, for example Group B streptococci (Gibson et al., 1991; Teti et al., 1992, 1993), viridans streptococci (Hanage & Cohen, 2002), Streptococcus pneumoniae (Benton et al., 1998), Streptococcus suis (Segura et al., 2006) and Staphylococcus aureus (Cui et al., 2000), generates a signal for elevated release of proinflammatory cytokines that are correlated with disease severity and mortality (Metz & Murray, 1990; Wakabayashi et al., 1991; Casey et al., 1993) has highlighted some probable similarities in septic shock pathophysiology, leading to an increased research interest aiming to identify the counterpart of LPS, the pivotal molecule in Gram-negative sepsis. Despite great efforts, results are often inconclusive or contradictory. For example, while some works clearly suggest that purified type and/or group-specific GBS polysaccharides induce considerable TNF-α secretion, (Vallejo et al., 1996; Cuzzola et al., 2000), in vivo data often do not support these results (Williams et al., 1993; Ling et al., 1995). Similar findings were described in the case of S. pneumoniae (Tuomanen et al., 1985). In view of the essential role of EPS in S. iniae pathogenesis, the belief that LTA is the unequivocal counterpart of LPS in terms of pathogenesis of Gram-positive bacteria (Ginsburg, 2002) should be reassessed, especially as other studies have reported that staphylococcal and GBS LTA is a weak TNF-α inducer (Nealon & Mattingly, 1985; Vallejo et al., 1996; Han et al., 2003) and that pneumococcal LTA is completely unable to induce cytokine production (Bhakdi et al., 1991). Taken together, these data indicate that despite the fact that the possible disparity in results may be due to technical differences in the assay systems (cell types and culture conditions, variations in the chemical structures, for example CPS from different pathogens, or even minimal biochemical changes between compounds considered similar, for example microheterogeneity among pneumococcal LTAs), the mechanisms underlying the septic shock induced by Gram-positive cocci are very likely heterogeneous. It appears that several of the cell wall components may act together or with other extracellular molecules, perhaps synergistically (Vallejo et al., 1996), to induce TNF-α production.

While an in vitro cell-line system cannot completely mimic the complexity of the natural milieu, and therefore can hardly stand alone in witnessing the role of EPS, the addendum of in vivo data, also essentially supporting the concept of resemblance in the cytokine network triggered after stimulation by Gram-positive and Gram-negative microorganisms (but sustaining the theory of the absence of a common LPS-like denominator among Gram-positive pathogens), now indicates that, in the case of the disease induced by S. iniae EPS-producing strains, EPS is a central molecule that is capable of triggering a process that resembles septic shock. This molecule, which has never been previously associated with proinflammation, is capable of causing dose-dependent death associated with TNF-α (and the whole main orchestra of proinflammatory cytokines) transcription levels that are practically double of those induced by LPS.

These results are appealing when viewed from an evolutionary perspective, suggesting that the collective immune response of the host population shapes the antigenic diversity of S. iniae to produce EPS that is responsible for sepsis just as LPS is for Gram-negative sepsis.

Ancillary