Lack of the purinergic receptor P2X₇ results in resistance to contact hypersensitivity

To study the role of extracellular ATP in CHS, we tested C57BL/6 WT and P2X₇-deficient (P2X₇^−/−) mice for their ear swelling reaction to epicutaneous sensitization with the contact allergen TNCB. Mice were treated on the abdominal skin followed by elicitation of CHS on day 5 by application of TNCB to the ear skin. The increase in ear thickness was measured 24 h after elicitation. Although WT mice developed a normal ear-swelling response, P2X₇^−/− mice did not develop TNCB-induced CHS (Fig. 1 A). Similar results were obtained for the contact allergen oxazolone (Fig. 1 B). In contrast, P2X₇-deficient mice developed normal ear swelling in response to the irritant croton oil (Fig. 1 C). These results identify a crucial role for the ATP-gated cation channel P2X₇ in CHS induced by contact allergens. The results from ear thickness measurements are supported by histology of the ear skin. In contrast to P2X₇^−/− mice, the ear skin of WT mice treated with TNCB shows a significant swelling and a more pronounced inflammatory infiltrate (Fig. 1 D).

DCs are the main interface between innate and adaptive immunity in CHS (Martin and Jakob, 2008; Freudenberg et al., 2009). Thus, the role of the P2X₇ receptor on DCs in the induction and/or elicitation of CHS was assessed. BM-derived DCs (BMDCs) of WT and P2X₇^−/− mice were modified with TNBS and subsequently injected i.c. (intracutaneously) into WT recipients as described previously (Martin et al., 2008). TNBS-modified WT BMDCs effectively sensitized WT recipients (Fig. 2 A). In contrast, TNBS-modified P2X₇^−/− BMDCs failed to sensitize WT mice under identical conditions. Together with the finding that TNBS-modified WT BMDCs were able to readily sensitize P2X₇^−/− recipients (Fig. 2 B), these data indicate that expression of P2X₇ on DCs is essential during the sensitization phase but not during the effector phase of CHS.

The failure of P2X₇^−/− BMDCs to induce sensitization may be caused by their failure to produce mature IL-1β via the NLRP3 inflammasome (Ferrari et al., 2006; Mariathasan et al., 2006; Di Virgilio, 2007; Surprenant and North, 2009). Therefore, we bypassed the need for P2X₇ triggering with alum, a potent P2X₇-independent activator of the NLRP3 inflammasome (Eisenbarth et al., 2008, Franchi and Núñez, 2008; Kool et al., 2008; Li et al., 2008). BMDCs were pretreated with alum before adoptive transfer. TNBS-modified and alum-pretreated P2X₇^−/− BMDCs regained their sensitizing potential upon transfer into both WT and P2X₇^−/− recipients (Fig. 2 C). Alum pretreatment of WT BMDCs did not significantly alter their sensitizing potential (Fig. 2 D). These data further support the assumption that P2X₇ on DCs is mandatory for efficient induction of CHS and acts via processing and release of inflammasome-dependent cytokines such as IL-1β (Ferrari et al., 2006; Mariathasan et al., 2006; Di Virgilio, 2007; Martinon et al., 2009; Surprenant and North, 2009).

To analyze the role of the NLRP3 inflammasome in DCs and the mechanism of the P2X₇ bypass with alum, WT mice were sensitized with NLRP3^−/− (Martinon et al., 2006) BMDCs in comparison with WT and P2X₇^−/− BMDCs with and without alum pretreatment (Fig. 3 A). No CHS was elicited by TNCB challenge in mice sensitized with unmodified control DCs or alum pretreated control DCs of all three mouse strains. Moreover, we demonstrate that the lack of NLRP3 in BMDCs modified with TNBS abrogates their sensitization capacity. This defect was not prevented by alum pretreatment, indicating that the alum bypass of P2X₇ is mediated by the NLRP3 inflammasome. In addition, we used BMDCs from ASC^−/− mice (Sutterwala et al., 2006) for sensitization (Fig. 3 B). As shown for NLRP3^−/− BMDCs, TNBS-modified ASC^−/− BMDCs also failed to sensitize WT mice for CHS, a defect which again could not be bypassed by alum pretreatment. These data support the notion that the NLRP3 inflammasome is important in CHS (Sutterwala et al., 2006; Watanabe et al., 2007) and demonstrate that functional ASC and NLRP3 are required in DCs for successful sensitization. To analyze the role of NLRP3 in P2X₇-mediated IL-1β processing, WT or NLRP3^−/− BMDCs were stimulated in vitro with LPS and alum or with LPS and ATP. Neither alum nor ATP stimulation increased the LPS-induced IL-1β secretion from NLRP3^−/− BMDC mice in contrast to BMDCs from WT mice (Fig. 3 C). These data demonstrate that the P2X₇^−/−-dependent IL-1β release is mediated via the NLRP3 inflammasome and confirm the important role of NLRP3 in alum-mediated IL-1β processing (Eisenbarth et al., 2008; Kool et al., 2008; Li et al., 2008). IL-6 release was triggered efficiently in both WT and NLRP3^−/− BMDCs, showing that there is no general functional defect of NLRP3^−/− BMDCs (unpublished data).

To further address the functional consequence of P2X₇ deficiency in DCs, WT and P2X₇^−/− BMDCs were analyzed for the production and secretion of pro– and mature IL-1β by Western blotting (Fig. 4 A) and for secreted IL-1β by ELISA (Fig. 4 B) in response to LPS and ATP in vitro. This cytokine is essential for the sensitization process in CHS (Shornick et al., 1996; Antonopoulos et al., 2001, 2008; Cumberbatch et al., 2001). Both WT and P2X₇^−/− BMDCs produced similar amounts of pro–IL-1β in response to LPS (Fig. 4 A), but only WT BMDCs secreted mature IL-1β when challenged with LPS and ATP, whereas P2X₇^−/− BMDCs did not (Fig. 4 B). These data show that induction of IL-1β release by TNCB is crucially dependent on ATP-mediated triggering of P2X₇. It is worth noting that in both WT and P2X₇^−/− BMDCs, ATP alone was able to efficiently induce pro–IL-1β (Fig. 4 A). This points to a P2X₇^−/−-independent effect on pro–IL-1β production. When BMDCs from both mouse strains were treated with graded concentrations of alum alone in vitro, no IL-1β release was observed (Fig. 4 C). However, the addition of 1 µg/ml LPS resulted in a dose-dependent IL-1β release from BMDCs of WT and P2X₇^−/− mice (Fig. 4 D). These results and the results shown in Fig. 3 show a link between LPS-induced pro–IL-1β production and ATP- and P2X₇-dependent IL-1β processing and release. An important finding is that ATP triggering of P2X₇ can be replaced by alum treatment of the BMDCs, which results in LPS-induced IL-1β production and release of processed IL-1β not only in WT but also in P2X₇^−/− BMDCs (Fig. 4, C and D). Importantly, the alum effect is no longer visible when BMDCs from NLRP3^−/− or ASC^−/− mice are used for sensitization in CHS (Fig. 3, A and B). Moreover, alum also fails to restore defective LPS-induced IL-1β release in NLRP3^−/− BMDCs in vitro (Fig. 3 C). These data support the notion that defective IL-1β release of P2X₇^−/− DCs is responsible for their failure to sensitize mice for CHS. They also strongly suggest that contact allergen–induced ATP triggering of P2X₇ leads to IL-1β processing via the NLRP3 inflammasome, as demonstrated in other systems (Ferrari et al., 2006; Mariathasan et al., 2006; Di Virgilio, 2007; Idzko et al., 2007; Ghiringhelli et al., 2009; Surprenant and North, 2009).

We next analyzed whether ATP is released in the skin after application of the contact allergen TNCB. Bioluminescence imaging (Edinger et al., 2003) revealed that TNCB rapidly triggers ATP release, as detected by ATP-dependent luciferin-induced bioluminescence from HEK293-pmeLUC cells (Pellegatti et al., 2008) after painting of the ear skin of mice (Fig. 5). These data reveal that P2X₇ is triggered by the endogenous danger signal ATP that is released from cells in the skin after TNCB treatment.

To further investigate the role of the P2X₇ receptor in CHS in vivo, experiments with the P2X₇ receptor antagonist KN-62 (Baraldi et al., 2004) were performed in WT mice. Recipients were injected with KN-62 into the left ear pinna, followed by sensitization with TNCB on the same ear 4 h later. CHS was elicited on day 5 with TNCB on the right ear, and the increase in ear thickness was measured 24 h after elicitation. As depicted in Fig. 6 A, pretreatment with KN-62 before TNCB sensitization led to abrogation of CHS in WT mice. This effect was dose and time dependent (Fig. S1). Similar results were obtained with the purinergic receptor antagonist suramin (Fig. S2; Baraldi et al., 2004). These data again indicate an essential role for P2X₇ receptor signaling in CHS. To assess the role of endogenously released ATP in CHS, WT mice were injected with apyrase, an ATP-degrading enzyme, to remove ATP from the extracellular space (Idzko et al., 2007) using the same approach as described for KN-62 (Fig. 6 A). Injection of apyrase into the ear pinna before sensitization prevented CHS in WT mice (Fig. 6 B). A dose- and time-dependent reduction of the ear swelling response was observed (Fig. S3). These results underline the crucial role for the release of extracellular ATP in the sensitization phase of CHS.

Finally, to assess the role of IL-1β in CHS, we neutralized the effects of this cytokine by i.v. injection of the recombinant IL-1Ra anakinra before sensitization of WT mice with TNCB. After ear challenge, we observed that anakinra treatment efficiently prevented the CHS response (Fig. 7 A). When we tried to induce CHS in IL-1R^−/− mice, we observed a similar effect. In contrast to IL-18^−/− mice, CHS was significantly reduced (Fig. 7 B). Because IL-1R signaling is not confined to IL-1β, we addressed the role of this cytokine by pretreatment of mice by IL-1β injection before sensitization. Although this treatment did not affect the CHS response in WT mice, it fully restored it in the P2X₇^−/− mice (Fig. 7 C). These data underline the role of IL-1R signaling and the crucial pathogenic role of IL-1β for the sensitization to contact allergens (Shornick et al., 1996; Antonopoulos et al., 2001, 2008; Cumberbatch et al., 2001).

The mechanisms by which contact allergens activate innate immune and stress responses in skin are only poorly understood. We have recently demonstrated a crucial role for the innate immune receptors TLR2 and TLR4 and the cytokine receptor IL-12Rβ2 in the sensitization phase of CHS (Martin et al., 2008). These receptors are indirectly triggered by contact allergens via induction of endogenous ligands such as fragments of the extracellular matrix component hyaluronic acid. So far, only Ni²⁺ has been identified as a contact allergen that directly triggers human TLR4 by binding to conserved histidine residues that are not present in mouse TLR4 (Schmidt et al., 2010). This explains why CHS to Ni²⁺ in mice requires coapplication of adjuvants.

Besides the TLRs, a role for the NLRP3 inflammasome in CHS has been demonstrated (Sutterwala et al., 2006; Watanabe et al., 2007). In the present study, we show that P2X₇ signaling in DCs is a crucial step in the sensitization phase of CHS. Interestingly, irritant contact dermatitis induced by croton oil is not affected in P2X₇^−/− mice. However, these mice are resistant to contact allergen–induced CHS. P2X₇^−/− DCs are incapable of sensitizing WT mice for CHS, whereas WT DCs efficiently sensitize P2X₇^−/− mice. The reconstitution of the sensitization capacity of P2X₇^−/− DCs by pretreatment with alum, a P2X₇-independent activator of the NLRP3 inflammasome (Eisenbarth et al., 2008, Franchi and Núñez, 2008; Kool et al., 2008; Li et al., 2008), supports the functional role of the inflammasome in CHS (Sutterwala et al., 2006; Watanabe et al., 2007) and suggests a P2X₇-dependent pathway in the processing and secretion of the key cytokines IL-1β and IL-18 in CHS (Shornick et al., 1996; Antonopoulos et al., 2001, 2008; Cumberbatch et al., 2001). P2X₇-mediated K⁺ efflux from the cell can activate the NLRP3 inflammasome and caspase-1, which is a requirement for IL-1β processing and secretion (Solle et al., 2001, Mariathasan et al., 2006; Dinarello, 2009). Similar effects are observed upon depletion of intracellular K⁺ by nigericin and maitotoxin (Ferrari et al., 2006; Mariathasan et al., 2006; Di Virgilio, 2007; Surprenant and North, 2009). Our findings suggest that P2X₇ is involved in contact allergen–induced and NLRP3 inflammasome–dependent IL-1β release.

The previously reported link between P2X₇ and posttranslational IL-1β processing, which involves the NLRP3 inflammasome (Ferrari et al., 2006; Mariathasan et al., 2006; Di Virgilio, 2007; Idzko et al., 2007; Ghiringhelli et al., 2009; Surprenant and North, 2009), is further supported by our in vitro findings. Stimulation of WT BMDCs with LPS and ATP results in production and secretion of mature IL-1β, whereas IL-1β processing and secretion is abrogated in P2X₇^−/− BMDCs. The fact that we can restore this defect with the inflammasome activator alum, which bypasses P2X₇ by direct activation of the NLRP3 inflammasome (Eisenbarth et al., 2008, Franchi and Núñez, 2008; Kool et al., 2008; Li et al., 2008), and the lack of IL-1β release from NLRP3^−/− DCs after stimulation with LPS and ATP reveal a functional link between ATP, P2X₇, and the NLRP3 inflammasome in IL-1β processing and secretion in the sensitization phase of CHS. Similar results were reported for LPS-primed macrophages from P2X₇^−/− mice, which secrete IL-1β when pretreated with the inflammasome activator nigericin (Solle et al., 2001). This is further supported by our results demonstrating a failure of TNBS-modified BMDCs from ASC^−/− and NLRP3^−/− mice (Martinon et al., 2006; Sutterwala et al., 2006) to sensitize WT mice. In contrast to the results with P2X₇ BMDCs, alum pretreatment of these DCs fails to restore their sensitizing capacity.

Interestingly, our data support the notion that IL-1β as a rapidly induced pathogenic cytokine in the sensitization phase of CHS (Enk and Katz, 1992) plays a crucial role. Keratinocytes and Langerhans cells are sources for IL-1β in the skin. A recent in vivo imaging study identified CD45⁺CD11b⁺ myeloid leukocytes as primary sources for IL-1β by monitoring messenger RNA induction in vivo (Matsushima et al., 2010). In contrast to the failure to sensitize WT mice treated with the IL-1Ra anakinra or the absence of CHS in IL-1R^−/− mice, CHS to TNCB was normal in IL-18^−/− mice. Moreover, pretreatment of P2X₇^−/− mice with IL-1β fully restored CHS. These results show the crucial role of IL-1β in the sensitization to TNCB. The extent of IL-1β messenger RNA induction correlates with the dose as well as with strength of the contact allergen (Lass et al., 2010). Interestingly, it was shown that CHS to DNFB (2,4-dinitrofluorobenzene) is also dependent on IL-18 (Klekotka et al., 2010). In this study, a role for IL-1R on radiosensitive bone marrow–derived cells and for IL-18 in radioresistant host-derived cells was demonstrated in bone marrow chimeras. Our findings on the role of IL-1β are in line with the study by Antonopoulos et al. (2008), who showed that deficient CHS in caspase-1^−/− mice can be restored by pretreatment with IL-1β but not with IL-18. The authors concluded that IL-18 acts upstream of IL-1β.

In the light of recent data showing that ATP can activate regulatory T cells and control CHS (Ring et al., 2010), a dual role for ATP in inflammation and immunoregulation becomes evident. It remains to be determined how the balance between proinflammatory and regulatory immune responses is regulated. The kinetics of these responses, the target cells, and the purinergic receptors involved may all play a role.

Our study identifies ATP triggering of P2X₇ on DCs as a crucial step in the induction of skin inflammation by contact allergens but not skin irritants such as croton oil. Similar to our findings with indirect TLR2 and TLR4 triggering by contact allergens such as TNCB, oxazolone, and FITC, via induction of endogenous TLR ligands (Martin et al., 2008), we provide evidence that P2X₇ activation by contact allergens is also indirect. We show that TNCB rapidly induces the release of the P2X₇ agonist ATP in the skin. We demonstrate that TNCB-induced CHS can be readily prevented by injection of the P2X₇ antagonist KN-62 and the purinergic receptor antagonist suramin or by the ATP-degrading enzyme apyrase. Further studies will show whether P2X₇-specific antagonists are also effective in the control of established CHS. In this context, our finding that treatment of mice with the IL-1Ra anakinra efficiently prevents CHS opens the possibility to test its efficiency in patients with ACD.

The fact that TLR2 and TLR4, P2X₇, or NLPR3 and ASC must be functional on DCs to enable them to sensitize mice for CHS and that the lack of function of one of these pathways is sufficient to abrogate their sensitization potential supports our concept of an essential collaborative action of these pathways in CHS (Freudenberg et al., 2009). The essential role of the DC is explained by its unique function as the cell type that migrates from the inflamed skin to the lymph nodes, communicates the activation of the innate immune system to contact allergen–specific T cells, and directs effector T cells to the skin (Dudda and Martin, 2004; Edele et al., 2007; Sigmundsdottir and Butcher, 2008).

Our study may open new perspectives for the treatment of ACD with available antagonists of the P2X₇ receptor, as recently reported for allergic airway inflammation (Idzko et al., 2007). So far, our data suggest that P2X₇ antagonists may prevent sensitization to contact allergens. It remains to be studied whether P2X₇ signaling also plays a role in the elicitation phase or in chronic contact dermatitis.

C57BL/6 WT, C57BL/6 P2X₇^−/− (Solle et al., 2001), and ASC^−/− mice (Sutterwala et al., 2006) were bred at the animal facility of the University Medical Center Freiburg under specific pathogen-free conditions. P2X₇^−/− mice were purchased from The Jackson Laboratory. NLRP3^−/− mice (Martinon et al., 2006) were received from M. Kopf (Swiss Federal Institute of Technology Zurich, Zurich, Switzerland) with permission from J. Tschopp (University of Lausanne, Lausanne, Switzerland). IL-1R^−/− (Labow et al., 1997) and IL-18^−/− mice (Wei et al., 1999) on C57BL/6 background were a gift from M. Freudenberg (Max-Planck-Institute of Immunobiology, Freiburg, Germany). All of the experimental procedures were in accordance with institutional, state, and federal guidelines on animal welfare. The animal experiments were approved by the Regierungspräsidium Freiburg and supervised by the Animal Protection Representatives of the University Medical Center Freiburg.

TNBS, FITC, oxazolone, and croton oil were purchased from Sigma-Aldrich. TNCB was purchased from VeZerf Laborsynthesen GmbH. LPS of Salmonella abortus equi was purchased from Enzo Life Sciences, Inc. IL-1β ELISA was purchased from BD (OptEIA ELISA set). Recombinant IL-1β was purchased from R&D Systems. Flow cytometry was performed as described previously (Martin et al., 2008) on a FACScan instrument with CellQuest Pro software (BD).

BMDCs were prepared as previously described (Martin et al., 2008). Overall DC viability was not significantly influenced by the lack of P2X₇ as controlled by growth rate and cell yield in DC cultures from P2X₇^−/− compared with WT mice.

Mice were sensitized by painting the abdominal skin with 100 µl of 3% TNCB/acetone or acetone alone as vehicle control on day 0. Epicutaneous application of 20 µl of 1% TNCB on the dorsum of both ears on day 5 was used for elicitation. Alternatively, mice were sensitized by i.c. injection of 3 × 10⁵ unmodified or TNBS-modified BMDCs in 2 × 50 µl PBS into two sites of the shaved abdomen (Martin et al., 2008) followed by elicitation on day 5 by painting the dorsum of both ears with 1% TNCB. CHS to oxazolone was induced by sensitization with 150 µl of 3% oxazolone in ethanol, followed by ear challenge with 20 µl of 1% oxazolone/ethanol 6 d later. Ear measurement was performed before and 24 h after challenge using an engineer’s thickness gauge (Mitutoyo).

The P2X₇ receptor antagonist KN-62 (Sigma-Aldrich) was dissolved in DMSO at 7.14 mg/ml. Mice were injected with KN-62 diluted in 71.4 µg/ml PBS, and 50 µl was injected into the pinna of the left ear 4 h before sensitization unless indicated otherwise. Apyrase (2 U potato apyrase grade 7 [Sigma-Aldrich] in 50 µl PBS) was injected into the ear pinna immediately before sensitization unless indicated otherwise. Mice were sensitized with 3% TNCB on the same ear and challenged 5 d later on the right ear with 1% TNCB. 24-h ear-swelling responses were measured. Anakinra Kineret (Amgen) was injected at a dose of 200 µg/mouse i.p. in 200 µl PBS on days 5, 3, and 1 before sensitization. The optimal doses and the timing for KN-62 and apyrase treatment were determined as shown in Figs. S1 and S3. IL-1β (50 µg in 30 µl PBS/0.1% BSA) or PBS/BSA as vehicle control was injected into WT and P2X₇^−/− mice 30 min before sensitization in the dorsum of the left ear as described previously (Antonopoulos et al., 2008). Mice were sensitized by topical application of 20 µl of 3% TNCB on the dorsum of the left ear on day 0. Challenge was performed on day 5 with 20 µl of 1% TNCB on the dorsum of the right ear.

Ears from vehicle or allergen-sensitized mice were removed 24 h after allergen ear challenge and fixed in 3.7% buffered formaldehyde. Organ slices of 5 µm were prepared and stained with hematoxylin and eosin (H&E).

Mice were painted on both ears with 20 µl of 1% croton oil (Sigma-Aldrich) in acetone/olive oil (4:1). The ear thickness was measured before and 24 h after application of croton oil.

BMDCs were harvested on day 7 and modified with TNBS as described previously (Martin et al., 2008). For alum stimulation, BMDCs were incubated for 2 h at 37°C in RP-10 containing 240 µg/ml aluminum hydroxide (diluted 1:166.7; Imject Alum; Thermo Fisher Scientific) unless indicated otherwise. For NLRP3^−/− and ASC^−/− BMDC transfers into WT mice, 80 µg/ml alum was used to avoid stimulation of recipient WT cells caused by alum carryover. Cells were washed three times in PBS before adoptive transfer by i.c. injection.

In vivo bioluminescence imaging was performed as previously described (Edinger et al., 2003). In brief, 5 × 10⁶ HEK293-pmeLUC cells (Pellegatti et al., 2008) were injected into the ear skin in 50 µl PBS. Mice were treated 24 h after cell transfer with the contact allergen TNCB (20 µl of 3% TNCB/acetone), the vehicle control acetone, or by injection of 50 µl of 1.25 mM ATP/PBS as a positive control, followed by i.p. injection of luciferin. 10 min later, the mice were imaged with an exposure time of 5 min with an IVIS 100 charge-coupled device imaging system (Xenogen) and Igor Pro Carbon (WaveMetrics).

Stimulation of 10⁶ BMDCs was performed in 500 µl RP-10 for 12 h with 1 µg/ml LPS in the presence or absence of 5 mM ATP or the corresponding controls. No significant differences in the viability of BMDCs from WT and P2X₇^−/− DCs were observed by propidium iodide/annexin-V staining. Supernatants were used for Western blotting and ELISA to measure IL-1β. ELISA (OptEIA; BD) was performed according to the manufacturer’s recommendations. Peritoneal macrophages were prepared and stimulated with LPS and ATP as described previously (Metkar et al., 2008), and the supernatant was used as positive control for Western blots. For alum stimulation, BMDCs were incubated for 12 h at 37°C in RP-10 containing graded concentrations of aluminum hydroxide in the presence or absence of 0.1 µg LPS.

Statistical analysis was conducted using one-way analysis of variance with Tukey’s multiple comparison post test and Prism version 5.00 (GraphPad Software, Inc.). Data are shown as means ± SD. Differences between groups as marked by asterisks were statistically significant at P < 0.05 (*), P < 0.005 (**), or P < 0.001 (***).

Fig. S1 shows the dose titration and time kinetics for the P2X₇ antagonist KN-62. Fig. S2 shows suppression of the CHS response in WT mice treated with the purinergic receptor antagonist suramin before sensitization. Fig. S3 shows the dose titration and time kinetics for the ATP-degrading enzyme apyrase.

We are grateful to Anton Grubisic, Thomas Stehle, and Bettina Hartmann (Max-Planck-Institute of Immunobiology, Freiburg, Germany), Eva Bachtanian (Allergy Research Group, Department of Dermatology, University Medical Center Freiburg, Freiburg, Germany), and Melanie Grimm (Chronic Obstructive Pulmonary Disease and Asthma Research Group, Department of Pneumology, University Medical Center Freiburg) for expert technical assistance.

F.C. Weber was supported by a fellowship from the Research Commission of the Medical Faculty, University Medical Center Freiburg. The work was supported in part by a grant of the European Commission as part of the project “Novel Testing Strategies for In Vitro Assessment of Allergens (Sens-it-iv)” to S.F. Martin (LSHB-CT-2005-018681) and an Emmy Noether grant of the Deutsche Forschungsgemeinschaft to M. Idzko (ID7/4-1).

The authors have no conflicting financial interests.