Combined TLR2 and TLR4 ligation in the context of bacterial or helminth extracts in human monocyte derived dendritic cells: molecular correlates for Th1/Th2 polarization

To study the molecular characteristics of DCs exposed to compounds that engage TLR2 and 4, yet lead to differential skewing of immune responses in terms of Th1 and Th2 induction, different pathogen derived products from bacterial or helminth origin with a known Th1 and Th2 inducing capacity were chosen and combined with LPS, as a reference maturation stimulus [6]. For this study Gram-negative E. coli and Gram-positive heat killed L. monocytogenes (HKLM) were used as bacterial stimuli that induce Th1 responses. A schistosome (a trematode) derived phosphatidylserine containing lipid preparation (schPS) and a similar preparation from the nematode worm A. lumbricoides (ascPS), both containing mainly phosphatidylserine species with two attached acyl chains and some lysophosphatidylserine species (with only a single attached acyl chain) (fig 1A and 1B, respectively), were chosen as Th2 inducing compounds [6]. Stimulation of HEK cells transfected with TLR showed that all stimuli could activate TLR2, with additional weak and potent TLR4 stimulation by the helminth lipids and E. coli, respectively (fig. 1C). IFN-γ and schistosome derived soluble egg antigen (SEA), stimuli that do not show TLR2 activating capacity in our experiments (fig. 1C), and induce Th1 and Th2 responses, respectively, were used as controls [6]. To assess the T cell polarizing capacity of DCs exposed to these compounds, stimulated human monocyte derived DCs were co-cultured for two weeks with allogeneic naïve CD4⁺ T cells and IL-4 as well as IFN-γ production was determined by intracellular staining upon T cell restimulation (fig. 1D). DCs were stimulated with the different compounds in the presence of LPS, to ensure equal maturation and to rule out potential effects on polarization due to differences in maturation status of the DCs. We found that in all conditions expression of maturation markers was significantly higher than levels measured on immature DCs and more similar to the levels induced by LPS alone (data not shown). As expected, E. coli induced a strong Th1 response comparable to DCs stimulated with IFN-γ, while HKLM induced a moderately polarized Th1 response. Conversely, the helminth derived compounds, as shown before for schPS [6], and SEA [4, 6], but also the A. lumbricoides derived phospholipids instructed DCs to drive Th2 skewed responses with the strongest polarization induced by SEA (fig. 1D). Based on intracellular IL-17 staining there was no sign of Th17 induction by the differently conditioned DCs, which is in agreement with other studies [12, 13].

To obtain a better understanding of the molecular processes in DCs that could underlie the observed differences in T cell polarizing capacity of these helminth- and bacteria-derived compounds, we set out to investigate in more detail the molecular characteristics of DCs exposed to the different stimuli. To study the intracellular signalling routes activated upon exposure to the helminth and bacterial derived products, we analysed the activation of the MAPK. ERK (ERK1/2) and p38 are two effector kinases of the MAPK family and are known to play an important role in shaping of immune responses [14]. p38 has been shown to regulate DC maturation and pro-inflammatory responses, while activation of ERK has been related to anti-inflammatory and Th2 responses [15]. As has been described before [16], exposure of DCs to LPS alone led to preferential phosphorylation of p38 (fig. 2A, B). The Th1 promoting stimuli IFN-γ and E. coli even further increased the activation of this MAPK resulting in a reduced p-ERK/p-p38 ratio, 20 minutes after stimulation (fig 2B, C), whereas for HKLM this ratio did not change. In contrast, the Th2 inducing compounds PS and SEA increased this ratio. Interestingly, the high p-ERK/p-p38 ratio induced by these Th2 polarizing stimuli was the result of different activation profiles for SEA versus the lipid preparations: SEA significantly induced phosphorylation of ERK, whereas the helminth derived lipids impaired p38 activation, but showed no effect on ERK activation (fig 2D and 2E). The p-ERK/p-p38 ratio showed a positive correlation with Th2, and negative correlation with Th1 polarization (R² = 0.36 and -0.47, respectively, figure 2F and 2G). In conclusion, for all components tested, the p-ERK/p-p38 ratio only 20 minutes after DC stimulation can be used to predict the outcome of the T cell response with regard to Th1 and Th2 polarization. This shows that very early events in DC activation already determine the fate of the DCs in terms of their T cell polarizing capacity.

To further characterize the molecular profile of the differentially stimulated DCs we performed mRNA expression analysis, using real-time PCR, on a selected set of genes involved in TLR signalling and T cell polarization (table 1[4, 15, 17‐48], Figure 3A). Upon maturation with LPS, the expression of most genes was increased (data not shown). All data shown are relative to what is seen in mature DCs without any polarizing agents added, i.e. DCs stimulated with LPS. Stimulation of DCs from different individuals with the same stimulus showed very consistent profiles (data not shown). Clustering analysis revealed that the gene expression profiles of Th1 and Th2 polarizing agents clustered in separate groups (top of figure 3A). Within the Th1 stimuli, DCs exposed to bacterial products derived of L. monocytogenes and E. coli had a remarkably similar profile that was different from the profile induced by IFN-γ. For the Th2 stimuli, both helminth derived lipid preparations showed a very comparable profile, which resembled the expression profile induced by SEA for most of the genes (fig 3A). However, expression levels in PS pulsed DCs were generally lower than in SEA stimulated DCs which is in accordance with the less pronounced effects on activation of the MAPK by the PS preparations.

Next, we related expression levels of individual genes to the T cell polarizing capacities of the DCs, to identify potential mechanisms through which different pathogen derived compounds induce differential T cell polarization. Members of the IL-12 cytokine family are well known for driving Th1 polarization [1]. Indeed the expression of both IL-12 p40 and p35, but also IL-23 p19 were shown to be upregulated in DCs stimulated with Th1 inducing stimuli and reduced in DCs stimulated with helminth derived compounds (fig. 3A). This was confirmed at the protein level when IL-12 and IL-23 production by DCs were measured by ELISA (Fig 3B, C).

With respect to T cell polarization, other genes of interest are the notch ligand family members delta-1, delta-4 and jagged-2, since expression of delta and jagged on DCs has been associated with induction of Th1 and Th2 responses, respectively [21]. For jagged-2 and delta-1 no significant differences were found between the stimuli (figure 3A). However, delta-4 was upregulated by the bacterial Th1 inducing stimuli, and downregulated by the Th2 inducing lipids. Moreover, expression levels of delta-4 correlated with the IL-4/IFN-γ cytokine ratio produced by the T cells of the stimuli that activate TLR2 (R² = -0.87, figure 3D). Yet, in SEA and IFN-γ stimulated DCs delta-4 expression was not altered. Therefore, Delta-4 seems to associate with T helper cell polarization only when TLR2 is also engaged.

Conversely, we found higher c-fos mRNA levels in PS and SEA pulsed DCs compared to HKLM and E. coli stimulated DCs. c-fos has been shown before to mediate SEA induced repression of IL-12 secretion by DCs [15]. Indeed, correlation analysis revealed that in DCs stimulated with bacterial products or helminth-derived lipids, mRNA levels of c-fos were positively correlated with Th2 induction (R² = 0.667, fig 3E).

Phosphatidylserine containing preparations (PS) were extracted from 4 gram of A. lumbricoides worms (expelled from infected humans) or from schistosomal worms, collected from golden hamsters 45–48 days after infection with S. mansoni, as described before [6]. Mass spectrometry was used to confirm the presence and composition of PS species in both lipid preparations, as described before [64]. Schistosomal egg antigen (SEA) was prepared from schistosomal eggs, collected from trypsin treated liver homogenate of the S. mansoni infected hamsters. E. coli (ATCC 11775) and L. monocytogenes (kind gift of J. van Dissel, LUMC, Leiden, The Netherlands) were grown at 37°C for 18 h in Brain Heart Infusion (BHI) bouillon (Biomerieux). Cultures were washed with PBS, quantified, and frozen in aliquots. In addition L. monocytogenes was heat inactivated for 2 hours and 45 minutes at 80°C before storage.

HEK-293-CD14, HEK-293-CD14/TLR2 and HEK-293-CD14/TLR4 cells (a gift from Dr. E. Latz, University of Massachusetts) were maintained in DMEM culture medium, supplemented with 10% FCS, 10 μg/ml ciprofloxacin and 5 μg/ml puromycin. For stimulation experiments, cells were seeded at 3.5 × 10⁴ cells/well in 96-well flatbottom plates and were stimulated the next day. For stimulation of HEK-293-CD14/TLR4 cells, 12.5% supernatant of MD-2 transfected cells was added. IL-8 production was measured in supernatants after 22 hours using a commercial kit (Sanquin, Amsterdam, The Netherlands), by following the manufacturer's recommendations.

Monocytes were isolated and immature DCs were cultured as described before [6]. At day 6 or 7 immature DCs were matured with LPS (ultrapure, E. coli 0111 B4 strain, Invivogen) (100 ng/ml) in the presence of IFN-γ (1000 U/ml), heat killed L. monocytogenes (HKLM; 10⁸/ml), E. coli (10⁷/ml), SEA (50 μg/ml), PS lipid extract derived from ascaris worms (an equivalent of 120 μg of worm per ml) or PS lipid extract derived from schistosomal worms (an equivalent of 20 worm-pairs per ml). For RNA isolation, DCs were harvested 16 hours after stimulation, as pilot experiments in our lab indicated that the expression levels of most genes had changed at this time point. DCs were snap-frozen in liquid nitrogen and kept at -80°C until RNA isolation. For measuring cytokine production by DCs and for co-culture with naïve T cells, DCs were matured for 48 hours after stimulation, after which secreted cytokines were measured in the harvested supernatant. Levels of IL12p70 were determined by ELISA using monoclonal antibodies 20C2 and biotinylated mouse-anti-human IL-12 C8.6 (both Becton Dickinson) as coating and detection antibodies, respectively. Levels of IL-23 were determined by ELISA using monoclonal antibodies ebio473p19 and biotinylated mouse-anti-human IL-12 C8.6 (both Becton Dickinson) as coating and detection antibodies, respectively. To determine T cell polarization, 5 × 10³ mature DCs were cocultured with 2 × 10⁴ naïve T cells that were purified using a human CD4+/CD45RO-column kit (R&D, Minneapolis, MN) in the presence of SEB (100 pg/ml; Sigma) in 96-well flat-bottom plates (Costar). On day 5, rhuIL-2 (10 U/ml, Cetus Corp., Emeryville, CA) was added and the cultures were expanded for another 5–9 days. To measure the frequency of IL-4- and IFN-γ-producing T cells, Th cells were restimulated with PMA and ionomycin in the presence of brefeldinA (all Sigma) during 6 hours and stained with anti-hu-IL-4-PE and anti-hu-IFN-γ-FITC (both BD Biosciences).

RNA isolation was performed using Trizol reagent (Invitrogen, Breda, The Netherlands) according to the manufacturers' instructions, with a minor modification: 3 μl of glycogen (Invitrogen) was added to all samples after they were homogenized in Trizol for a few minutes at room temperature. DNAse treatment and cDNA synthesis were performed following standard procedures.

Primers and Taqman probes were provided as a Taqman gene expression kit (Applied Biosystems, Foster City, California) or designed using Primer Express (Applied Biosystems) and synthesized by Biolegio (Malden, The Netherlands) and Eurogentec (Seraing, Belgium), respectively (sequences available upon request). Real time qPCR was performed using Eurogentec PCR reagents, in a volume of 25 μl on an ABI PRISM 7700 Sequence Detection System (SDS, Applied Biosystems), using the following program: 10 minutes at 95°C, 40 cycles of 15 seconds denaturation at 95°C and 60 seconds annealing and amplification at 60°C. Results were monitored and analysed with SDS software (Applied Biosystems).

Gene expression was normalized to the housekeeping gene TAF-1 and calculations were performed as described using the 2^-ΔΔC_T method [65]. Analysis of the expression of 6 different housekeeping genes in a subset of the samples indicated that TAF-1 was the most stable housekeeping gene in our samples upon stimulation. Spotfire software http://​spotfire.​tibco.​com was used to generate a heatmap and perform hierarchical clustering of the genes.

20 and 60 minutes after stimulation of immature DCs (day 6), cells were fixed for 10 minutes with 4% ultrapure formaldehyde (Polysciences) directly in the plate. Cells were harvested and washed twice in PBS/0.5% BSA. Subsequently, the DCs were permeabilized in 700 μl ice-cold 90% methanol in PBS in and left on ice for 30 minutes. Following two wash steps in PBS/0.5%BSA intracellular staining was performed for 2 hours at room temperature in the dark with anti-phospho-p44/42 MAPK AF-488 (T202/Y204) and anti-phospho-p38 MAPK AF-647 (T180/Y182), (Cell Signalling Technology). After one wash in PBS/0.5%BSA MAPK activation was determined by flow cytometry using a Becton Dickinson FACSCalibur flowcytometer (BD Biosciences) and analysed using FlowJo analysis software (Tree Star).

Data were analysed using SPSS (v14.0) and GraphPad Prism4. Differences among stimuli were analysed by a Mann-Whitney test. Differences relative to LPS stimulation were determined using a one sample t-test. Correlations between expression of genes and/or T-cell responses were calculated by a two-tailed Spearman's-rho test. Differences were considered significant when P-values were below 0.05.