To study the pathogenic potential of free-soluble material released by A. actinomycetemcomitans cells, we implemented our cell culture plate insert model for stimulation of human whole blood as described in Methods. Similarly to our previous studies using the insert model [17], the absence of A. actinomycetemcomitans cell lysis in the wells during the experiments was confirmed by immunoblot analysis of samples taken from the wells outside the inserts (filtrates), using an antiserum raised against the cytoplasmic protein cyclic AMP receptor protein (CRP) (data not shown). In addition, plating revealed no apparent death of A. actinomycetemcomitans cells, or contamination in the wells (data not shown).

According to our findings from the cytokine antibody array analysis, used to screen for the cytokine responses in human whole blood, the free-soluble material released by strain D7S (rough-colony wild type), grown in planktonic form, caused clearly enhanced production of the interleukins IL-6 and IL-8, and of macrophage inflammatory protein (MIP-1β), relative to no bacteria controls (Fig. 1A–B). The same was observed when whole human blood was stimulated with free-soluble material released from A. actinomycetemcomitans strain D7SS (smooth-colony variant) (Fig. 1C–D). In addition, very similar results were obtained when whole blood from another donor was used (data not shown), indicating no apparent inter-individual differences. Interestingly, production of IL-6, IL-8 and MIP-1β was also induced when D7S cells grown in biofilm were used (Fig. 1E–F). This suggests that A. actinomycetemcomitans cell populations growing in biofilms also release free-soluble material with proinflammatory activity to the surrounding environment. Taken together, we concluded that the free-soluble components released by live A. actinomycetemcomitans cells, grown either planktonic or in biofilm, could indeed induce proinflammatory effects in whole blood.

According to the above-described cytokine analyses (Fig. 1) the free-soluble material released from the test strains appeared to have no effect on the production of TNF-α and IL-1β. As a previous report indicated increased production of these cytokines in human whole blood, stimulated with A. actinomycetemcomitans LPS [19], the levels of TNF-α and IL-1β were also quantified using a highly sensitive approach (Dissociation enhanced lanthanide fluorescence immuno assay; DELFIA [see Methods]). As shown in Fig. 2, a significant increase in both TNF-α (10.3-fold) and IL-1β (4.2-fold) production was detected when whole human blood was stimulated with free-soluble material released from planktonic D7S cells in the cell culture model (see Methods), relative to no bacteria controls. Why this was not seen using the cytokine array (Fig. 1) is unclear. According to our findings (Fig. 2), the increase in TNF-α and IL-1β production was less pronounced in the presence of polymyxin B (PMB), which blocks LPS activity [20]. However, there was still a clear upregulation (4.1-fold and 1.7-fold, respectively), indicative of an LPS-independent effect on TNF-α and IL-1β production. Thus, we concluded that in addition to IL-6, IL-8 and MIP-1β, the production of TNF-α and IL-1β in whole blood was also enhanced by the free-soluble material released by D7S.

To confirm our observations of increased production of the above proinflammatory cytokines and chemokines in stimulated whole blood (Fig. 1 and Fig. 2, the amount of transcript of each gene was quantified using quantitative real-time PCR (qRT-PCR). As indicated in Fig. 3, there was up to 1000-fold increased transcription of each cytokine gene when human blood was stimulated with free-soluble material released from planktonic D7S cells relative to no bacteria controls. Although the increase in cytokine gene transcription was less intensive (up to 100-fold) when the whole blood stimulation was done in the presence of PMB (Fig. 3), each gene was still clearly upregulated, indicating that this effect was also LPS-independent. We could therefore conclude that transcription of IL-6, IL-8, MIP-1β, TNF-α and IL-1β in whole blood was enhanced by the free-soluble material released by D7S.

As the free-soluble material released by strain D7S provoked proinflammatory responses in human whole blood also independently of LPS, we investigated the possibility that this was due to release of leukotoxin A (LtxA) and/or cytolethal distending toxin (CDT). These immunomodulatory toxins were earlier shown to be extracellularly secreted by A. actinomycetemcomitans [21,22]. To test the contribution of LtxA and CDT we constructed a derivative of strain D7S having both the ltxA and cdtABC gene loci deleted (see Methods). The abolished production of LtxA was confirmed by immunoblotting, using specific antibodies (Fig. 4). According to our findings using the ex vivo insert model for stimulation of whole blood, inactivation of cdtABC and ltxA had no major effect. Instead, the free-soluble material released by the D7S double mutant, grown in planktonic form, induced a proinflammatory response in whole blood similar to that of the parental strain, D7S (Fig. 1A–B), i.e. production of IL-6, IL-8, and MIP-1β was clearly enhanced (Fig. 1G) relative to no bacteria controls (Fig. 1F). When this experiment was carried out in the presence of PMB (Fig. 1H) to inhibit LPS activity, the production of these cytokines was still induced relative to no bacterial controls (Fig. 1F), again indicating LPS-independent stimulation. However, the cytokine stimulation was less intensive, compared to when PMB was absent (Fig. 1G). Lack of contribution of LtxA and CDT to cytokine stimulation in human whole blood is consistent with the absence of these toxins in the filtrates released by strain D7S to PBS, which was also confirmed by immunoblotting (Fig. 4 and data not shown).

As PAL was identified in the free-soluble material released by both D7S and D7SS [17](Fig. 4), a pal mutant derivative (D7SS-p) was subsequently tested in the ex vivo model. According to our results, the free-soluble material released by planktonic D7SS-p induced a cytokine response in whole blood very similar to that of the parental strain, D7SS (data not shown). These findings together suggest that other factor(s) released in free-soluble form by A. actinomycetemcomitans may be involved in the induction of proinflammatory responses in whole blood.

To obtain more information about the free-soluble proteins secreted by A. actinomycetemcomitans, we used the cell culture insert model with PBS instead of serum/whole blood (see Methods). Plating revealed no apparent death of A. actinomycetemcomitans cells, or contamination in the wells (data not shown). As in our previous studies using the insert model [17], the absence of A. actinomycetemcomitans cell lysis in the wells during the experiments was confirmed by immunoblot analysis of filtrates, using an antiserum raised against the cytoplasmic protein, CRP (Fig. 4). SDS-PAGE analysis of filtrates from D7S and D7SS, grown both planktonic and in biofilm form (Fig. 5A), revealed several bands after Silver staining. Finding similar band profiles of material released through the 20 nm pores of the inserts from both D7S and D7SS is in accordance with the cytokine induction by these strains as determined by the present antibody array (Fig. 1).

Since our results hitherto obtained with D7S and D7SS were very similar with each other, we continued our analyses mainly using filtrates from D7S. As shown in Fig. 5B, two major protein bands could be detected in filtrates from D7S, by using Coomassie staining. These bands represented proteins of approximate molecular masses of 60 and 15 kDa, respectively. Bands of these sizes (Fig. 5C and data not shown) were also obtained with immunoblot analysis of the D7S filtrates, using an antiserum raised against whole-cell antigen of A. actinomycetemcomitans serotype a. Our previous results show that this antiserum reacts to multiple epitopes in A. actinomycetemcomitans [23]. Our current findings (Fig. 5C) therefore suggest that the 60 and 15 kDa-proteins are the two major protein antigens released in free-soluble form under the present conditions. As the 15 kDa-protein band was not detected in filtrates from the pal-deficient strain D7S-p (Fig. 5B), we concluded that this protein band represents PAL. This was consistent with immunoblotting (Fig. 4 and data not shown).

To determine the identity of the 60 kDa protein we used MALDI-TOF-MS, which identified A. actinomycetemcomitans GroEL-like protein (Hsp60; Cpn60) (Swiss-Prot entry P46398). The release of GroEL-like protein in free-soluble form to PBS by both planktonic and biofilm D7S was confirmed using immunoblotting and an antiserum raised against E. coli GroEL (Fig. 4). According to SDS-PAGE (Fig. 5), the pal-deficient strain D7S-p released a higher amount of GroEL-like protein than the parental strain. This observation is consistent with the role of the Tol-Pal complex in maintaining outer membrane integrity [24,25]. Taken together, we could conclude that GroEL-like protein is a major protein released in free-soluble form by A. actinomycetemcomitans D7S.

To assess the specificity of our present findings to A. actinomycetemcomitans, the ex vivo culture insert model (Methods), controlled as above (Fig. 6 and data not shown), was used to stimulate human whole blood with free-soluble material released from planktonic E. coli strain IHE3034 (serotype O18:K1:H7). As shown in Fig. 7A–C, there was a clear stimulation of IL-6, IL-8, and MIP-1β relative to no bacterial controls, indicative of a proinflammatory effect in whole blood caused by the E. coli free-soluble material. This observation is in accordance with previous studies using E. coli strains J5 and 789 (serotypes O18:K1:H7 and O78, respectively) [14,15], which demonstrated the release of Outer Membrane Protein A (OmpA), a prime target of the host immune system [26,27]. The release of OmpA in free-soluble form to PBS by planktonic IHE3034 was confirmed using immunoblotting and an antiserum against OmpA (Fig. 6). OMV-independent secretion of OmpA by E. coli is consistent with subcellular localization studies, finding OmpA in the fraction of soluble secreted proteins in addition to vesicles [14]. In contrast to OmpA, GroEL was absent from these filtrates (Fig. 6). This is in agreement with the cytoplasmic localization of GroEL and its presence in OMVs released by E. coli [28]. Our results therefore show that release of free-soluble surface material with proinflammatory effects in human whole blood is not restricted to A. actinomycetemcomitans.

A. actinomycetemcomitans serotype a strains D7S (rough colony type) and D7SS (smooth colony type), their corresponding PAL-deficient mutants, D7SS-p [23] and D7S-p [17], strain D7S Δltx/cdt (this work), and the leukotoxin highly producing strain, JP2 [56], were all cultured on blood agar plates (5% defibrinated horse blood, 5 mg hemin/l, 10 mg Vitamin K/l, Columbia agar base) incubated in air supplemented with 5% CO₂, at 37°C for 3 d as previously described [17]. Escherichia coli IHE3034 (serotype O18:K1:H7) isolated from meningitis [57] was cultivated 16–18 h on blood agar as described above. For biofilm growth, 2 × 10⁸ bacterial cells were inoculated in 2 ml tryptic soy broth (Difco) in 24-well cell culture plates (Nunc), which were incubated in static culture in air supplemented with 5% CO₂, at 37°C for 3 d. Biofilms were stained with crystal violet as previously described [58] and the amount of bound dye, which is proportional to the biofilm mass was quantitated by measuring its absorbance at 590 nm.

The D7S ltxA cdtABC markerless double-deletion mutant was constructed using the Cre/loxP recombination system, as described previously [59]. In brief, two DNA fragments flanking the target deletion site were generated by PCR and cloned into upstream and downstream loxP sites of the vector pLox2-Spe [59], which carries a Spe-resistance (Spe^R) cassette flanked by loxP sites. The recombinant plasmid was subsequently introduced into strain D7S, using natural transformation [60], selecting for spectinomycin-resistant transformants. The replacement of the target gene with loxP-Spe cassette-loxP was confirmed by PCR sequencing. The Spe cassette in the transformants was then removed by introducing the plasmid pAT-Cre [61], carrying the Cre recombinase. The removal of the Spe cassette was also confirmed by DNA sequence analysis. The resultant mutants had the cdtABC operon replaced by a loxP site spacer AGATCTGC, and the ltxA gene replaced by a loxP site with spacer ATGTATAC. This double-deletion strain construct was confirmed by PCR analysis, using the primer pairs CDT1 (5'-GGAGGCGATAACTCTACATCAGG-3') and CDT2 (5'-GTGTCACGTCGTCAAGCCGATG-3'), and LTX1 (5'-CTACTACGGGACCTGTCGCAGG-3') and LTX2 (5'-CCGGCTTTAGTAGCATTACGACCG-3'), respectively.

All procedures were conducted in accordance with the guidelines of the local ethics committee at the Medical Faculty of Umeå University, which are in compliance with the Declaration of Helsinki (59^th WMA General Assembly, Seoul, October 2008). Blood was drawn from a healthy volunteer, after informed consent, from antecubital vein into blood collection tubes containing heparin (BD Biosciences). A cell culture insert model, described previously [17], was then used to expose the whole blood to free-soluble components released by live A. actinomycetemcomitans and E. coli cells. For whole blood stimulation using planktonic bacteria, bacterial cells (final concentration 1 × 10⁸ CFU/ml) were suspended in serum (50% in PBS), separated from the same whole blood. The bacterial suspension (500 μl) was then added to cell culture inserts of pore size 20 nm (Nunc). The inserts were placed into the wells of a 24-well cell culture plate (Nunc), containing 500 μl whole blood. Serum (50% in PBS) containing no bacteria served as a negative control. The cell culture plate was subsequently incubated for 6 h at 37°C in air supplemented with 5% CO₂.

For whole blood stimulation using biofilm bacteria, A. actinomycetemcomitans biofilms (grown 3 d in 24-well cell culture plates; approximately 2 × 10⁸ CFU) were gently washed with PBS and then 500 μl serum (50% in PBS) was added to the wells. Human blood (500 μl) was then added to cell culture inserts of pore size 20 nm (Nunc), which were placed into the wells with biofilm. The cell culture plate was subsequently incubated as above. When indicated, experiments were performed with PMB (final concentration: 30 μg/ml), an inhibitor of LPS activity [20], to roughly estimate the extent of LPS-independent cytokine stimulation. Similarly to other authors [62], we did not use heat treatment of samples as this has been shown to reduce the cytokine-inducing activity of LPS [63-65].

For RNA isolation from stimulated whole blood, 100 μl blood was removed from the wells after 6 h of incubation and processed as described below. The remaining volume (400 μl) of stimulated whole blood was centrifuged 5 min at 5000 × g. Supernatants were stored at -80°C until cytokine analysis and profiling (see below).

The cell culture insert model was controlled for viability of the bacteria, absence of bacterial lysis, and the absence of bacterial contamination of the material released through the insert filters (filtrates) as described previously [17]. In brief, aliquots from within and outside the inserts were plated on blood agar plates, which were incubated as above for 3 d. Filtrates were also analyzed with immunoblotting, using an antiserum raised against the cytoplasmic protein cyclic AMP receptor protein (CRP). Biofilm integrity was confirmed by crystal violet staining as described previously [66] and also by plating aliquots taken from above the biofilms on blood agar.

For this, the cell culture insert model was employed and controlled as described above, however with the exception that PBS was used instead of serum or blood. After 6 h of incubation at 37°C in air supplemented with 5% CO₂, filtrates were collected and precipitated with acetone (80% final concentration) prior to SDS-PAGE analysis. Selected protein bands after Coomassie blue staining were subject to MALDI-TOF-MS analysis at the Umeå Protein Analysis Facility, Department of Chemistry, Umeå University.

The procedures employed for SDS-PAGE and immunoblot analysis have been described previously [23]. For immunoblots we used the following polyclonal antibodies raised in rabbits against A. actinomycetemcomitans LtxA [67] and PAL [68], E. coli GroEL (Sigma-Aldrich), E. coli OmpA [69], H. ducreyi Cdt [70], V. cholerae CRP [71], and against whole cells of A. actinomycetemcomitans serotype a [72]. Antisera were used at final dilutions of 1:1000, except for the antisera raised against GroEL (1:8000), PAL (1:10000), and OmpA (1:10000). Horseradish peroxidase-conjugated anti-rabbit secondary antibodies were used at a final dilution of 1:10000. Immunoreactive bands were visualized using SuperSignal^® (Pierce, Rockford, IL, USA).

To obtain a profile of the cytokines produced by human whole blood after stimulation with released free-soluble bacterial material, a cytokine antibody array (TranSignal™ Human Cytokine Antibody Array 1.0; Panomics, Redwood City, CA, USA), allowing simultaneous detection of the levels of 18 cytokines, was used according to instructions from the manufacturer. Dissociation enhanced lanthanide fluorescence immuno assay [73] was used for the quantification of the amounts of IL-1β and TNF-α produced by human whole blood. For this, 96-well microtiter plate wells were coated with capture-antibodies against human IL-1β and TNF-α (200 ng/well) and incubated overnight at 4°C with shaking. The wells were washed with DELFIA wash buffer twice and saturated with 1.5% BSA in water for 6 h at RT. The wells were then washed twice with 0.9 M NaCl and stabilized at RT for 15 h. Samples and the standards (recombinant IL-1β and TNF-α, e-Bioscience), each in triplicates, were added to the wells and incubated at RT for 1 h. After washing as above twice, detection antibodies against human IL-1β (Europium-labelled) and TNF-α (Samarium-labelled) were added to the wells (200 ng/well) and incubated at RT for 1 h. The wells were subsequently washed six times as above, and then 200 μl DELFIA enhancement solution was added to each well and the plates were incubated at RT on a shaker for 15 min. Europium and Samarium signals were measured using a multilabel reader (Victor³, Perkin Elmer).

Total RNA from whole blood was isolated from two separate stimulation experiments, using the RiboPure™-Blood kit (Ambion Inc.) according to instructions of the manufacturer. Quantitative real-time PCR was performed using the 7900HT Real-time PCR system (Applied Biosystems). Concentration of RNA was determined by measuring the absorbance at 260 nm. One μg of total RNA was converted to cDNA using Transcriptor First Strand cDNA Synthesis Kit (Roche) according to instructions from the manufacturer. Each qRT-PCR reaction (final volume 5 μl) was run in triplicates, containing 1× SYBR^® Green PCR Master Mix (Applied Biosystems), and with the cDNA diluted 100× (1000× in addition to assess PCR efficiency). The sequences and final concentrations of the forward and reverse oligonucleotide primers used in qRT-PCR are indicated in Table 1. The qRT-PCR data were normalized to an endogenous control (β-actin).

Unless otherwise stated all experiments were repeated three times. Mann-Whitney U test was used to assess the statistical significance of differences between experimental and control treatment of whole blood, observed using DELFIA and qRT-PCR. P-values of less than 0.05 were regarded as statistically significant.