The C-type lectin member 5A (CLEC5A) is a pattern recognition receptor for members of the Flavivirus family and has critical functions in response to dengue virus and Japanese encephalitis virus. Here we show that CLEC5A is involved in neutrophil extracellular trap formation and the production of reactive oxygen species and proinflammatory cytokines in response to Listeria monocytogenes. Inoculation of Clec5a−/− mice with L. monocytogenes causes rapid bacterial spreading, increased bacterial loads in the blood and liver, and severe liver necrosis. In these mice, IL-1β, IL-17A, and TNF expression is inhibited, CCL2 is induced, and large numbers of CD11b+Ly6ChiCCR2hiCX3CR1low inflammatory monocytes infiltrate the liver. By day 5 of infection, these mice also have fewer IL-17A+ γδ T cells, severe liver necrosis and a higher chance of fatality. Thus, CLEC5A has a pivotal function in the activation of multiple aspects of innate immunity against bacterial invasion.
The myeloid C-type lectin 5A (CLEC5A), also known as myeloid DAP12-associated lectin-11, is a spleen tyrosine kinase (Syk)-coupled receptor abundantly expressed by monocytes, macrophages and neutrophils. Whereas CLEC7A detects beta-glucans2 and has critical functions in host defense against Candida albicans invasion3, CLEC5A has been shown to recognize mannose and fucose expressed on envelope proteins of Flavivirus family members, and has critical functions in the pathologies of dengue virus-induced lethal diseases4 and Japanese encephalitis virus-induced neuroinflammation5. Activation of CLEC5A by dengue virus and Japanese encephalitis virus can result in secretion of proinflammatory cytokines such as TNF, IL-6, CXCL-8, and IP-10. Moreover, dengue virus activates the NLRP3 inflammasome via CLEC5A to induce the production of IL-1β and IL-186, 7. We have previously shown that dengue virus activates CLEC5A to upregulate osteolytic activity and dysregulate bone homeostasis8. Blockade of CLEC5A by anti-CLEC5A monoclonal antibody (mAb) not only attenuates dengue virus-induced and Japanese encephalitis virus-induced lethality, but also suppresses dengue virus-mediated activation of the NLRP3 inflammasome and osteolytic activity4, 8. These findings suggest that CLEC5A has critical functions in flavivirus-induced inflammatory reactions. However, whether CLEC5A is involved in the pathogenesis of other microbial infections has not been investigated.
Via phagocytosis and production of reactive oxygen species (ROS), macrophages, neutrophils and CCR2+ inflammatory monocytes are the first line of host defense against bacterial invasion9, 10, and depletion of these cells increases susceptibility to systemic Listeriosis11. Listera monocytogenes is an intracellular Gram-positive bacterium that can cause disease in immunocompromised individuals. In animal models, L. monocytogenes can induce systemic sepsis and liver abscess12. In addition to phagocytosis and ROS production, neutrophils produced web-like structures known as neutrophil extracellular traps (NETs), which comprise decondensed chromatin, histones, and subsets of granules and cytoplasmic proteins to ensnare variety of microbes13. Moreover, activated macrophages secrete proinflammatory cytokines (such as IL-1β, TNF and IL-6) and limit growth and spreading of bacterial populations12. IL-17A-producing γδ T cells have been shown to have a critical protective function against Listeria infection14: however, the pattern recognition receptors (PRRs) responsible are unclear. MyD88-deficient mice are highly susceptible to L. monocytogenes infection, suggesting that Toll-like receptors (TLRs) may be involved in host recognition of L. monocytogenes. Although TLR2 can recognize components of Gram-positive bacteria, TLR2 does not contribute significantly to host survival after Listeria infection12, 15. Thus, the key PRR responsible for resistance to L. monocytogenes infection is unknown.
Here we show that CLEC5A is a promising candidate PRR in macrophage-mediated and neutrophil-mediated defense against L. monocytogenes. CLEC5A not only has important functions in the production of free radicals and cytokines, but also drives NET formation (NETosis). Compared to wild-type (WT) littermates, the expression of IL-1β, TNF, and IL-17A is downregulated in Clec5a−/− mice, an effect associated with rapid bacterial spreading and increased bacterial loads in liver and blood. Moreover, the livers of Clec5a−/− mice develop severe necrosis associated with fewer IL-17A+ γδ T cells and massive infiltration of CCR2+ inflammatory monocytes. Together, these data indicate a critical role for CLEC5A in L. monocytogenes-induced innate immunity.
CLEC5A is involved in L. monocytogenes-induced IL-1β production
Macrophages are principle effector cells in the clearance of bacterial infections by exerting several bactericidal actions12 and coordinating innate immune responses16. Among the proinflammatory cytokines secreted from macrophages after bacterial infection, IL-1β is crucial for induction of a febrile response and differentiation of TH17 and IL-17A-producing γδ T cells17, 18. It has been reported that bacteria promote IL-1β production via TLR-dependent NF-κB19 activation and Syk-mediated signaling20. Thus, we asked whether CLEC5A is involved in bacteria-induced inflammasome activation. To address this question, L. monocytogenes were incubated with WT, Clec5a−/− and Tlr2−/− macrophages, followed by examination of Syk phosphorylation and NF-κB activation. We found that L. monocytogenes induced the phosphorylation of Syk (Fig. 1a) and p65 (Fig. 1b) in WT macrophages; these responses were significantly impaired in Clec5a−/− and Tlr2−/− macrophages, respectively (Fig. 1a, b). Compared to WT macrophages, Clec5a−/− and Tlr2−/− macrophages produced less IL-1β after L. monocytogenes stimulation (Fig. 1c). In addition, IL-1β mRNA expression was downregulated in Clec5a−/− and Tlr2−/− macrophages incubated with live or UV-inactivated L. monocytogenes (Fig. 1d). These data indicate that both TLR2 and CLEC5A are involved in IL-1β production upon challenge with L. monocytogenes and that is independent of intracellular replication of the bacteria. It is interesting to note that caspase I activity is only downregulated in Clec5a−/− macrophages (Fig. 1e, f), indicating that CLEC5A, but not TLR2, is responsible for L. monocytogenes-induced caspase I activation. As CLEC5A is a DAP12-associated receptor1, we asked whether DAP12 is required for L. monocytogenes–induced IL-1β production. As shown in Supplementary Fig. 1a, b, IL-1β production was downregulated dramatically in Dap12−/− macrophages, suggesting that CLEC5A-mediated IL-1β production during L. monocytogenes infection is mediated via DAP12.
It has been demonstrated that L. monocytogenes is sensed by multiple inflammasomes (including NLRP3, NLRC4, and AIM2 inflammasomes) that collectively orchestrate the production of IL-1β21,22,23. Compared to WT macrophages, nlrc4 transcript was downregulated in Clec5a−/− macrophages while aim2 transcript was downregulated in both Clec5a−/− and Tlr2−/− macrophages following L. monocytogenes infection (Fig. 1g). In contrast, the expression of Nlrp3 transcript was similar in WT, Clec5a−/−, and Tlr2−/− macrophages.
Together, these observations suggest that CLEC5A and TLR2 are involved in regulating the transcription of IL-1β and inflammasome components, and that the CLEC5A-DAP12 signaling cascade regulates canonical caspase I-dependent inflammasome activation, in response to L. monocytogenes.
CLEC5A is critical in neutrophil-mediated defense mechanisms
We went on to test whether CLEC5A is involved in neutrophil-mediated immune responses to L. monocytogenes, i.e., phagocytosis, cytotoxicity, and ROS production, using Ly6Ghi neutrophils, from Clec5a−/−, Tlr2−/−, and WT mice. While the phagocytic activity of Clec5a−/− Ly6Ghi neutrophils was similar to WT (Supplementary Fig. 2a), Clec5a−/− and Tlr2−/− neutrophils had impairments in both cytotoxic activity (Fig. 2a) and production of ROS (Fig. 2b, c; Supplementary Fig. 2b). ROS production is not only an essential bactericidal action, but also contributes to bacteria-induced NET formation24, 25. Therefore, we further investigated the role of CLEC5A in NET formation by incubating human neutrophils with L. monocytogenes in the presence of an antagonistic anti-CLEC5A mAb4, an isotype control or an anti-TLR2 blocking mAb26. Examination and quantitation of NET formation revealed that L. monocytogenes induced NET formation 3 h post infection, and that this was almost completely abolished by blockade of CLEC5A (Fig. 2d, e). In contrast, blockade of TLR2 had a much weaker inhibitory effect on NET formation (Fig. 2d, e). We further validated these observations using Clec5a−/− and Tlr2−/− neutrophils. NET formation was detected in murine Ly6Ghi neutrophils and peaked at 90 min post-infection with L. monocytogenes (Supplementary Fig. 3). In contrast, NET formation was impaired in both Clec5a−/− (Fig. 2f) and Dap12−/− (Supplementary Fig. 1c) neutrophils; mild impairment of NET formation was noted in Tlr2−/− neutrophils under the same conditions (Fig. 2f). To further explore the mechanism of CLEC5A-mediated NET formation, we examined peptidylarginine 4 (PAD4)-mediated histone citrullination27, which causes histone decondensation and enhanced NET formation28. We found that histone 3 citrullination was severely impaired in Clec5a−/− neutrophils, but not in Tlr2−/− or WT neutrophils (Supplementary Fig. 4). Thus, CLEC5A contributes significantly to NET formation via ROS production and histone citrullination during L. monocytogenes infection.
Increased bacterial spreading in Clec5a-deficient mice
Since NET formation is known to limit bacterial spreading in vivo, we asked whether CLEC5A deficiency impairs this aspect of host immunity. Mice were inoculated intravenously (i.v.) with live luminescent L. monocytogenes (Xen32) and livers were collected to determine the extent of NET formation 4 h later. In WT livers, fibrous DNA structures associated with citrullinated histone 3 (mean NET histone area: 23972 μm2) and the neutrophil-specific marker myeloperoxidase (MPO) were detected (Supplementary Fig. 5a, b); digestion of the fibrous DNA with DNase I did not significantly reduce citrullinated histone 3 (mean NET histone area: 20,153 μm2) (Supplementary Fig. 5a, b). While only mild impairment of NET formation was observed in Tlr2−/− livers (mean NET histone area: 16580 μm2), citrullinated histone 3 (mean NET histone area: 6748 μm2) and fibrous DNA were reduced significantly in Clec5a−/− livers (Supplementary Fig. 5a, b).
Inactivation of L. monocytogenes in vivo is known to be mediated by neutrophils at 2–6 h after i.v. inoculation29,30,31,32. Here we investigated (using a minimally invasive imaging system) whether impaired NET formation in Clec5a−/− mice resulted in enhanced bacterial growth and spreading. In WT and Tlr2−/− mice, luminescence intensity in the liver peaked at 2 h post-infection with Xen32, before decreasing steadily (Fig. 3a, b). In contrast, luminescence intensity was still increasing at 4 h post-infection in the livers of WT/DNase I and Clec5a−/− mice (Fig. 3a, b) and this correlated with bacterial loads in the liver (Fig. 3c) and blood (Fig. 3d) at the 4 h time point. These observations suggest that CLEC5A-mediated NET formation contributes to clearance of L. monocytogenes infection.
CLEC5A is critical for induction of IL-17A+ TCR γδ T cells
We further investigated the impact of CLEC5A deficiency on systemic Listeriosis by administering a sublethal dose of L. monocytogenes to mice (1 × 105 CFU per mouse, i.v.) and monitoring bacterial load in the liver. At day 1 post-infection, the highest bacterial load was observed in WT/DNase I mice (Fig. 4a). While bacterial load increased gradually in Tlr2−/− mice, it was significantly elevated in WT/DNase I, and Clec5a−/− mice at day 3 post-infection (Fig. 4a). By day 5, Clec5a−/− mice had the highest load in the liver and significant elevation was also seen in Tlr2−/− mice. This observation suggests that TLR2 contributes to limiting of L. monocytogenes proliferation in the later stages of infection.
Since L. monocytogenes-mediated activation of both CLEC5A and TLR2 induces phosphorylation of Syk and p65 (Fig. 1a, b), we investigated whether CLEC5A and TLR2 collaborate in host defense against systemic Listeriosis. Clec5a−/− mice were crossed with Tlr2−/− mice to generate Clec5a−/−Tlr2−/− mice (Supplementary Fig. 6). We found that bacterial load in the liver was dramatically elevated in Clec5a−/−Tlr2−/− mice at days 3 and 5 post-infection with L. monocytogenes (Fig. 4a). Similarly, bacterial load in peripheral blood was elevated in CLEC5A- and TLR2-deficient mice (Clec5a−/−Tlr2−/− > Clec5a−/− > Tlr2−/−) (Fig. 4b). These observations suggest that co-activation of CLEC5A- and TLR2-mediated pathways are required for optimal host immunity against systemic L. monocytogenes infection. To further test this hypothesis, we examined the cytokine profiles in livers from Clec5a−/−, Tlr2−/−, and Clec5a−/−Tlr2−/− mice post L. monocytogenes infection. We found that the expression of tnf, Il-1β, and Il-17a was downregulated in Clec5a−/− and Tlr2−/− mice (ANOVA, P < 0.05), while these cytokines were further downregulated in Clec5a−/−Tlr2−/− mice (ANOVA, P < 0.01) (Fig. 4c). Interestingly, ccl2 transcript was upregulated in Clec5a−/−, Tlr2−/−, and Clec5a−/−Tlr2−/− mice (Fig. 4c).
IL-17A induction in the liver occurs very early post L. monocytogenes infection and this cytokine plays a critical role in host defense14. Here we determined the population of IL-17A-producing cells in liver at 5 days post L. monocytogenes infection. We found that in WT mice IL-17A was expressed by CD3+CD4−TCRγδ+ T cells (1.65% of CD3+ cells; Fig. 4d, e), which have been shown to play a critical role in innate immunity against L. monocytogenes infection in liver14. In contrast, IL-17A-proudcing CD3+CD4−TCRγδ+ T cells were almost undetectable in Clec5a−/− mice (0.36% of CD3+ cells) and Clec5a−/−Tlr2−/− mice (0.17% of CD3+ cells) (Fig. 4d, e). Even though the populations of IL-17A-producing γδ T cells were reduced in mice deficient in CLEC5A and TLR2, the total populations of CD3+CD4+ T cells and CD3+CD4− TCR γδ T cells were similar in all groups of mice tested (Supplementary Fig. 7). This suggests that CLEC5A and TLR2 are required specifically for the induction of IL-17A-producing γδ T cells, and not for the development of CD3+CD4−TCR γδ T cells, in the liver.
CLEC5A is not detectable in γδ T cells, even though it is highly expressed by myeloid cells4 (Supplementary Fig. 8). Previous studies have shown that dengue virus can activate the NLRP3 inflammasome and induce the secretion of IL-1β7, which is critical for the induction of an IL-17A-secreting CD3+CD4−TCRγδ+ population33. Because IL-1β production is suppressed in L. monocytogenes-infected Clec5a−/− macrophages (Fig. 1c, d), we concluded that the induction of IL-17A-secreting CD3+CD4−TCRγδ+ population during Listeria infection occurs via CLEC5A-dependent IL-1β production from myeloid cells, rather than direct activation of CD3+CD4−TCRγδ+ cells.
CLEC5A deficiency results in liver inflammation and mortality
Because CCL2 is critical for the recruitment of CCR2+CD11b+Ly6c+ inflammatory monocytes34, and is upregulated in Clec5a−/− mice after L. monocytogenes infection (Fig. 4c), we asked whether CLEC5A deficiency affected the recruitment of CCR2+CD11b+Ly6c+ cells to the liver. Compared to WT littermates, extensive distribution of inflammatory foci with Ly6B.2+ cells were noted in the livers of knockout mice (Clec5a−/−Tlr2−/− > Clec5a−/− > Tlr2−/− mice) (Fig. 5a). Flow cytometry further demonstrated that the majority of the infiltrating cells were CD11b+Ly6c+ cells (monocytes), rather than CD11b+Ly6g+ cells (neutrophils) (Fig. 5b). In addition, the expression of CCR2 was increased in the livers of Clec5a−/− and Clec5a−/−Tlr2−/− mice (as determined by RT-PCR; Supplementary Fig. 9a), and multicolor flow cytometry revealed elevated numbers of CD11b+Ly6chiCCR2hiCX3CR1low inflammatory monocytes among the infiltrating CD11b+Ly6c+ cells in these animals (WT: 18%, Clec5a−/−: 28.7%, Tlr2−/−: 23.2%, and Clec5a−/−Tlr2−/−: 33% of CD11b+CX3CR1low cells) (Supplementary Fig. 9b). Thus, CLEC5A deficiency enhanced, rather than suppressed, the recruitment of inflammatory monocytes to the liver post L. monocytogenes infection.
Since CCR2+ inflammatory monocytes mediate host defense against bacterial infection, we investigated whether this cell population was functionally impaired in Clec5a−/− and Clec5a−/−Tlr2−/− mice. Compared to WT macrophages, cytotoxic activity and ROS production were reduced dramatically in Clec5a−/− and Clec5a−/−Tlr2−/− macrophages following incubation with L. monocytogenes in vitro (Supplementary Fig. 10). The production of IL-1β and TNF (Supplementary Figs. 11, 12), as well as the expression of inducible nitric oxidase synthase (iNOS), were also downregulated (Supplementary Fig. 13) post-infection in vivo. Furthermore, the extent of cell infiltration correlated with liver inflammation, necrosis (Fig. 5c and Supplementary Fig. 14), and elevation of serum aminotransferase (ALT) (Fig. 5d) in knockout mice (Clec5a−/−Tlr2−/− > Clec5a−/− > Tlr2−/−). These observations suggest that Clec5a−/− inflammatory monocytes are functionally impaired and unable to limit L. monocytogenes proliferation in vivo; CCL2 production in the liver is, therefore, elevated in order to attract more inflammatory monocytes and thereby compensate for the deficiency in bactericidal activity. The high mortality rate amongst the knockout mice post Listeria infection (Fig. 5e) was attributed to increased bacterial load and septic shock resulting from severe hepatocyte damage.
We went on to further examine the effect of attenuated NET formation by comparing the infiltrating cell populations in the livers of WT and WT/DNase I mice. We found that the majority of infiltrating cells were CD11b+Ly6c+ cells rather than CD11b+Ly6g+ cells (Supplementary Fig. 15a, b). Compared to WT mice, CD11b+Ly6chiCCR2hiCX3CR1low inflammatory monocyte numbers were elevated in WT/DNase I mice (WT: 20.4%, WT/DNase I: 31.5% of CD11b+CX3CR1low cells; Supplementary Fig. 15c), suggesting that impaired NET formation also resulted in a compensatory recruitment of inflammatory monocytes to the liver. Moreover, elevated serum ALT levels and an increased mortality rate were observed in WT/DNase I mice compared to WT following L. monocytogenes infection (Fig. 5d, e). Interestingly, adoptive transfer of WT neutrophils into Clec5a−/− mice partially reduced L. monocytogenes loads in blood and liver (Supplementary Fig. 16), further supporting the concept that NET formation contributes significantly to host defense against L. monocytogenes infection. Higher mortality rates in knockout mice (Clec5a−/−Tlr2−/− and Clec5a−/−) compared to WT/DNase I mice indicates that the mechanism of CLEC5A-mediated protection against Listeria infection involves both NET formation and cytotoxic effects of inflammatory monocytes and neutrophils.
Dissection of CLEC5A/TLR2 co-activated signalling pathways
Microorganisms express multiple pathogen-associated molecular patterns and it is likely that these activate multiple PRRs simultaneously to enhance host immunity. Consistent with this, we observed that both CLEC5A and TLR2 were upregulated and co-localized in neutrophils after incubation with L. monocytogenes (Fig. 6a). Moreover, co-activation of CLEC5A and TLR2 on neutrophils by agonistic anti-CLEC5A and TLR2 ligands not only enhanced chemokine production (Fig. 6b), but also further upregulated ROS production (Fig. 6c) and NET formation (Fig. 6d and Supplementary Fig. 17). Similar responses were observed in macrophages co-activated with agonistic anti-CLEC5A mAb and TLR2 ligand (Pam3CSK4) (Supplementary Fig. 18). It is interesting to note that NETosis was suppressed efficiently by inhibitors of p38 (SB203508), PI3K (LY294002), and Syk (Fig. 6e), where L. monocytogenes activates both AKT and p38 (Fig. 6f). Moreover, activation of CLEC5A induced phosphorylation of AKT (MFI = 14.2), but not p38 (Fig. 6f), while TLR2 ligands mainly induced p38 phosphorylation (MFI = 14.1) (Fig. 6f). In contrast, co-activation of both CLEC5A and TLR2 enhanced the phosphorylation of both AKT (MFI = 23.6) and p38 (MFI = 18.6) (Fig. 6f). These observations suggest that co-activation of CLEC5A and TLR2 by L. monocytogenes will stimulate p38- and PI3K-Akt- pathways and thereby enhance host immunity against infection (Fig. 7).
In this study, we investigated the role of CLEC5A in the innate immune response to bacterial infection using L. monocytogenes (a Gram-positive, intracellular bacterium) as a model system. Our data reveal that CLEC5A plays a critical role not only in macrophage activation and cytokine production, but also in neutrophil-mediated cytotoxicity and NET formation. Elimination of NETs, via DNase I treatment, enhanced bacterial spreading in vivo (Figs. 3, 4a), suggesting that NET formation, which is suppressed in Clec5a−/− neutrophils, contributes significantly to host defense against L. monocytogenes. In addition, CLEC5A deficiency led to impaired production of TNF-α and IL-1β as well as reduced numbers of IL-17A-producing γδ T cells, which are critical for immunity to L. monocytogenes14. Moreover, Clec5a−/− mice were more sensitive than Tlr2−/− animals to infection with a sub-lethal dose of L. monocytogenes. All these observations provide compelling evidence that that CLEC5A is a critical PRR in Listeria infection.
NET formation is dependent on the proxduction of ROS, which enables the release of neutrophil elastase (NE) and MPO from azurophilic granules35. In addition, PAD4 is responsible for histone citrullination, which promotes decondensation of nuclear DNA28. ROS production was similar in Clec5a−/− and Tlr2−/− neutrophils, but PAD4 activity was impaired in Clec5a−/− neutrophils post L. monocytogenes infection. It has been shown that histone citrullination by PAD4 is facilitated by elevated ROS36 and cytosolic calcium concentration37, 38. Since activation of CLEC5A induces both calcium mobilization1 and ROS production (Fig. 6c), it is likely that the dominant role of CLEC5A, compared to TLR2, in promoting NET formation is associated with a PAD4−mediated mechanism.
Studies in humans and mice have shown that neutralization of IL-1 by IL-1R antagonist (IL-1RA; Kineret, which is used to treat rheumatoid arthritis) is associated with increased susceptibility to bacterial infections39, 40. This is consistent with the impaired immune response to Listeria in Clec5a−/− mice, where IL-1β production is suppressed. TLR2-deficient mice have reduced Il-1β expression compared to WT at 6 h post-infection with Staphylococcus aureus, but not at 24 h41, suggesting that the role of TLR2 in IL-1β production is most significant in the early stages of infection. Although we observed less IL-1β production in Tlr2−/− than Clec5a−/− macrophages incubated with L. monocytogenes (Fig. 1c, d), Tlr2−/− mice were more resistant than Clec5a−/− mice to Listeria infection (Fig. 5e). Furthermore, the higher mortality rate and severe liver inflammation and necrosis in Clec5a−/−Tlr2−/− mice (compared to Clec5a−/− and Tlr2−/− animals) suggests that, while the DAP12-assoicated CLEC5A is a critical PRR for L. monocytogenes (Fig. 5e and Supplementary Fig. 19), co-activation of multiple CLEC5A- and TLR2-mediated signaling pathways is required to initiate a robust immune response to systemic Listeria infection.
Since IL-1β is critical for host defense against bacterial infections in general, we investigated whether CLEC5A is required for IL-1β production in response to other bacteria. We found that, compared to WT cells, Clec5a−/− macrophages produced less IL-1β after incubation with S. aureus (Gram-positive, extracellular bacterium) and K. pneumoniae (Gram-negative, extracellular bacterium), but not S. typhimurium (Gram-negative, intracellular bacterium) (Supplementary Fig. 20), suggesting that CLEC5A is an important PRR for S. aureus and K. pneumoniae. This is further supported by our observations of impaired NET formation (Supplementary Fig. 21) and ROS production (Supplementary Fig. 1b) in Clec5a−/− neutrophils after incubation with S. aureus or K. pneumonia. Moreover, Clec5a−/− mice showed increased susceptibility to S. aureus infection, with rapid bacterial spreading to regional lymph nodes after subcutaneous inoculation (Supplementary Fig. 22).
Using a CLEC5A.Fc fusion in a microbial glycan microarray analysis42, we found that CLEC5A also binds to other bacteria, including Pseudomonas aeruginosa, Streptococcus pneumonia, Escherichia coli, and Shigella flexneri (Supplementary Table 1). Since L. monocytogenes was not represented in the glycan array used here, cell wall components were extracted from L. monocytogenes and the CLEC5A.Fc fusion protein was shown to bind to this extract in a dose-dependent manner (Supplementary Fig. 23a). Moreover, fluorochrome-conjugated L. monocytogenes, S. aureus, and K. pneumonia (but not S. typhimurium) bound to 293T cells overexpressing CLEC5A and TLR2 (Supplementary Fig. 23b). These observations demonstrated the direct interactions of bacteria with CLEC5A and TLR2, further supporting the role of CLEC5A as a PRR in defense against a range of bacterial infections.
To determine whether CLEC5A binds to terminal sugars of L. monocytogenes, we extracted cell wall polysaccharides from various L. monocytogenes strains (see Supplementary Table 2). In all cases, we observed similar binding to CLEC5A.Fc by ELISA and similar ability to induce NET formation and cytokine production (data not shown). This observation suggests that terminal sugars (rhamnose, galactose, glucose) on wall teichoic acids do not play significant roles in the interaction of L. monocytogenes with CLEC5A and subsequent activation of the host immune response. In contrast, CLEC5A binding to Listeria cell wall extracts was inhibited by N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) disaccharides; this observation suggests that CLEC5A binds GlcNAc-MurNAc disaccharide backbone, rather than terminal sugar, of L. monocytogenes.
Here we have shown that CLEC5A is an important PRR in immune responses to bacteria, including L. monocytogenes and S. aureus via activation of macrophage-, neutrophil- and γδT cell-mediated effector functions. Previously, CLEC5A and other DAP12-associated receptors have been implicated in responses to mycobacterial infection (Mycobacterium bovis)43, Our observation that CLEC5A can promote differentiation of IL-17A-producing CD3+CD4- TCRγδ T-cells, which are necessary for granuloma formation during mycobacterial infection44, might provide a mechanism for this activity.
Culture media/supplements and HBSS buffer were from Invitrogen GIBCO. Chemical reagents and chemical inhibitors were purchased from Sigma and Calbiocam, respectively. TLR ligands, including pam2csk4 (tlr-pm2s-1), pam3csk4 (tlr-pms), and LPS (tlr-eklps) were obtained from Invivogen. The monoclonal antibodies used in this study was listed as following: mouse IL-1β ELISA kit (DY 401), human TNF ELISA kit (DY210) and blocking antibody to human TLR2 (MAB2616) were purchased from R&D Systems, antibodies to histone H3 (citrulline R2 + R8 + R17; ab5103, 1:100), myeloperoxidase (MPO, sc-52707 or sc-34159, 1:100), phospho-syk (#2701, 1:1000), phosphor-p65 (#3033, 1:1000), histone 1–4 (Mab 3422, 1:200). Alexa Fluor® 488 Conjugated antibody against phosphor-p38 (#690, 1:100) and phosphor-AKT (Ser473, #2336, 1:100) were from Cell Signaling. CLEC5A antagonistic mAb has been characterized before4.
Bacteria, including L. monocytogenes (10403S, Serotype 1/2a), S. aureus (25923), S. typhimurium (13311) were obtained from ATCC. K. pneumoniae was from a clinical isolation (NTUH K2044, serotype K1). L. monocytogenes Xen32 possessing a stable copy of the modified Photorhabdus luminescens lux operon at a single integration site on the bacterial chromosome was derived from the parental strain L. monocytogenes 10403S (Serotype 1/2a wild-type strain), was purchased from Caliper Life Science and used for the Xenogen IVIS optical imaging technology in this study. For serogroups of L. monocytogenes with different glycan substitutents on the cell wall teichoic acid structure, including EGDe (1/2a), EGDe (lack of rhamnoase), EGDe (lack of N-acetylglucosamine), EGDe (lack of rhamnoase, and N-acetylglucosamine; GlcNAc)45, 4MT (serotype 4b, ATCC10527), XL7(lack of glucose on TA)46, M44 (lack of galactose on TA)47. ATCC19111 (serotype 1), ATCC19112 (serotype 2), ATCC19113 (serotype 3), ATCC19114 (serotype 4a), ATCC19115 (serotype 4b) are from Dr Stowell Sean.
All experiments were conducted with mice housed under SPF conditions that were age matched (8–10 weeks old) and backcrossed for at least twelve generations onto the C57B/6 background. Clec5a−/− mice5 were backcrossed for twelve generations into C57/B6 background; and Tlr2−/− mice (C57/B6) were from the Jackson Laboratory (Bar Harbor, ME). Clec5a−/−Tlr2−/− mice (C57/B6) were generated by crossing Clec5a−/− and Tlr2−/− mice, and the F1 offspring were further interbred to generate F2 offspring (Supplementary Fig. 6). Dap12−/− mice (C57/B6) were from Dr NJ Chen (a gift from Dr Toshiyuki Takai, Department of Experimental Immunology, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan). All gene-deficient mice and littermate control were bred in the animal facility within campus, and all animal experiments were in compliance with the Guide for the Care and Use of Laboratory Animals of the Taiwanese Council of Agriculture in a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of Academia Sinica (IACUC #1510872).
For human neutrophil isolation, peripheral venous blood from the healthy donors was taken according to a protocol approved by the Academia Sinica Ethics Committee (AS-IRB-BM-14006). Blood samples were incubated with acid citrate dextrosefollowed by centrifugation to remove plasma and platelets before subjected to Ficoll-Paque (Amersham Biosciences) density-gradient centrifugation to isolate granulocytes. The erythrocytes were removed by two rounds of hypotonic lysis with RBC lysis buffer, while the viability and purity of neutrophils were determined by CD66 and 7-AAD double staining using FACS Caliber (Becton-Dickinson). Human neutrophil were obtained from healthy donors under a protocol (AS-IRB02-103202) approved by the IRB of the Clinical Center of the Department of Health, Taiwan. Written informed consent was obtained from all donors. Isolation of mouse neutrophils from the bone marrow was modified as described previously48. Briefly, bone marrow was flushed out of the tibia and the femur using RPMI medium supplemented with penicillin/streptomycin, and passed through a 70 μM cell strainer to obtain a single-cell solution. Cells were pelleted, and the remaining erythrocytes were lysed by RBC lysis buffer, followed by washing with Hank’s balanced-salt solution (HBSS) without CaCl2/MgCl2. Bone marrow cells were further suspended in 45% (v/v) Percoll/HBSS and overlaid onto a discontinuous Percoll gradient comprising 52% (v/v), 62% (v/v) and 81% (v/v) Percoll in HBSS, followed by centrifugation at 1,000 × g for 30 min at room temperature. Mouse Neutrophils were recovered from the interphase between layers of 62 and 81% Percoll, and washed with HBSS. The purity (>90%) of isolated neutrophils was confirmed by FACS analysis.
Detection of NET structure by immunofluorescence staining
Human and mouse neutrophils were suspended in RPMI with 10% of autologous serum and seeded on the coverslip, then allowed to settle for 2 h. Cells were either infected with bacteria or treated with TLR ligands followed by fixing cells with 4% paraformaldehyde at indicated time points. Fixed cells were subjected to permeabilization of cell membranes with 0.5% Triton X-100 in phosphate-buffer saline (PBS), then blocked with 3% bovine serum albumin in PBS for 1 h at room temperature, and incubated with primary antibodies against anti-histone and MPO at 4 °C overnight. Secondary antibodies conjugated with Alex488 or TRITC were applied to detect primary Ab, and Hoechst 33342 were used as a counter stain. NET images were observed under confocal microscope (Leica: TCS-SP5-MP-SMD), and images were analyzed by Leica LAS AF software. For the liver histology, livers were dissected and mounted in OCT embedding compound, and then frozen at -80 °C to generate 10-microm sections for immunofluorescence staining. Slides were fixed with cold acetone at 4 °C for 10 min, followed by the air dry for 30 min, and then incubated with Hoechst 33342 and primary antibodies: antibody to histone and MPO. Stained slides were detected by fluorochrome-conjugated secondary antibodies as mentioned above. Images were obtained by using Leica confocal microscope with white light laser system (TCS-SP5-MP-SMD). Image was analyzed by Leica LAS AF software.
Quantification of NET formation
NET formation was quantified by histone area. All images were obtained by using Leica confocal microscope (TCS-SP5-MP-SMD) and processed under Leica LAS AF software. Histone image of neutrophil were analyzed using MetaMorph® software. Briefly, NETs were visualized in at least five random fields (40 × magnification), and signal intensity of histone per fields was individually measured; and the pixels of each image were converted into area (μm2) using a calibration unit (0.8333). Mean of histone area was from the five independent fields (Supplementary Fig. 24) were denoted as NET histone area (μm2).
Preparation of macrophages
For human macrophage preparation, peripheral blood mononuclear cells (PBMCs) were isolated from the whole blood of healthy human donors by standard density-gradient centrifugation with Ficoll-Paque (Amersham Biosciences). After centrifugation, the buffy coat containing leukocytes (PBMC) and platelets was further washed with PBS, and CD14+ cells were purified using the VarioMACS technique with anti-CD14 microbeads (Miltenyi Biotec GmbH). Cells were then cultured in complete RPMI 1640 medium supplemented with 10 ng/ml human M-CSF (R&D Systems) for 6 days4. For preparation of murine bone marrow-derived macrophages, bone marrow cells were isolated from femurs and tibias and cultured in RPMI 1640 complete medium supplemented with 10% (v/v) fetal calf serum (FCS) and 10 ng/ml of recombinant mouse M-CSF (R&D Systems) for 6–8 days. Human monocytes were obtained from healthy donors at the Taipei Blood Center of the Taiwan Blood Services Foundation, under a protocol (AS-IRB02-103202) approved by the IRB of the Clinical Center of the Department of Health, Taiwan. Written informed consent was obtained from all donors.
The caspase-1 activity was measured after 6 h post bacterial stimulation, with a Caspase-1 Colorimetric Assay kit (BioVision, K111), whereas caspase-1 processing was detected by mouse caspase 1 ELISA kit to measure the amount of processed 20-kilodalton (p20) and 10-kilodalton (p10) caspase-1 in the culture supernatants (AdipoGen; AG-45B-0002-KI01).
CLEC5A agonistic Ab synergizes with TLR ligands on NETosis
Human neutrophils seeded on the coverslip were pre-incubated with agonistic CLEC5A mAb (clone: 2D8G9; 2.5 μg for 1 × 106 neutrophils) (Supplementary Fig. 25) for 10 min at room temperature, followed by cross-linking with a secondary antibody to mouse IgG (Jackson ImmunoResearch, 115-005-146) for 20 min, and then subjected to addition of TLR ligands at the amount of 30 μg per reaction for another 2 h NET components were detected by immunofluorescence staining using anti-MPO, anti-citrullinated histone H3 (cit-H3) antibodies and Hoechst 33342, followed by observation with a confocal microscope (Leica).
Measurement of ROS production and NBT reduction assay
Measurement of ROS was modified as described previously49. Briefly, human or mouse neutrophils (1 × 105 cells per well) were suspend in HBSS plus Ca2+ and Mg2++, and incubated with live bacteria (m.o.i. = 3) for 30 min, followed by addition of pre-warmed (37 °C) mixture containing luminol (50 μM) and HRP (2 U per well) in HBSS to immediately prior to quantification. Luminescence was measured at an interval of 10 min till 120 min incubation at 37 °C in a luminometer (Titertek–Berthold). For the nitro blue tetrazolium (NBT) detection, neutrophils were seeded on the coverslip (1 × 105 cells per well) for 1 h, followed by incubating the live bacteria (m.o.i. = 3) and NBT solution to a final concentration of 4 mM simultaneously. Cells were observed every 10 min till the appearance of indigo color (about 60-min post infection) and were fixed with 4% paraformaldehyde for 2 h at 4 °C. Cells were observed under a light microscopy and numbers of NBT positive were calculated for the average of five individual fields in each group.
Measurement of bacterial killing and phagocytosis
For phagocytosis analysis, mouse neutrophils (1 × 105 cells per well) were seeded and infected with bacteria (m.o.i. = 0.1) for 60 min; supernatants were collected to determine the bacteria counts as the un-engulfed bacterial number. Percentage of phagocytosis was calculated as following formula: (1-unengulfed bacterial numbers/original bacterial numbers) × 100% engulfed bacterial numbers/original bacterial numbers) × 100%. For the killing assay, the formula is (1-engulfed bacterial numbers/original bacterial numbers- un-engulfed bacterial numbers) × 100%.
Mouse neutrophils (3 × 105) incubated with live L. monocytogenes for 1 h were collected and lysed with lysis buffer (50 mM Tris pH 6.6, 10% SDS, 10% 2-Mercaptoethanol, 50% glycerol and 0.05% bromophenol blue). Immunoblot was firstly performed with anti-citH3 Ab (citrulline R2 + R8 + R17, ab51031, 1:1000), and then the used PVDF member were stripped and re-blotted with anti-H3 Ab (Sigma H0164, 1:1000) as an internal control.
Monitoring bacterial growth in vivo
A Xenogen IVIS Imaging System 200 series (PerkinElemer Inc.) was used to quantify luminescent live L. monocytogenes (Xen32) over time. Live lag-phase L. monocytogenes (Xen32) was washed and suspended as an amount of 5 × 108 CFUs in 100 μl saline, then i.v. injected into mice. To assess the effect of NETs on the L. monocytogenes dissemination, mice received an intraperitoneal injection of DNase I (4 KU) 1h before i.v. inoculation of Xen32. Bacterial dissemination was quantified by collectingblood at 2 and 4 h after bacterial inoculation. Mice were anesthetized with isoflurane while in the imaging chamber and imaged at 30 min, 1, 2 and 4 h after bacterial inoculation. Photons were measured during 60-s exposure, and total photon emissions from uniform area of each mouse were quantified by using the Living Image software (Caliper Life Science). At 4 h, the mice were killedand liver and spleen were homogenized and plated on TSA agar plates containing kanamycin and grown overnight to quantify bacterial CFUs.
Systemic Listeriosis model was performed with L. monocytogenes 10403S strain in 10- to 12-weeks-old mice. All groups of mice were i.v. injected with 1 × 105 CFUs of L. monocytogenes and monitored for 15 days. For the DNase I treatment, mice were intraperitoneally with DNase I (2 KU) every 2 days. Blood and liver tissue was collected as indicated time points for bacterial load determination. Livers were also collected and lysed by TRIzol to extract total RNA for reverse-transcription into complementary DNA (cDNA) for analyzing gene expression. The primer sequences for real-time PCR analysis were listed in the Supplementary Table 3.
Detection of leukocyte subpopulations in liver by FACS
Mice were perfused with PBS buffer containing heparin (10 U/ml) prior tokilling. The livers were collected and minced into pieces followed by being digested with the buffer containing 1 mg/ml collagenase type IV (Sigma C5138) and 300 U/ml DNAase I (Sigma D5025) at 37 °C for 30 min. Liver homogenates were passed through cell strainer (40 μm), and lysed with RBC lysis buffer to remove the remaining red blood cells. The homogenates were further resuspended in 40% (w/v) Percoll-HBSS solution and overlaid onto 70% (w/v) Percoll-HBSS, followed by centrifuging at 2000 rpm for 30 min. After centrifugation, hepatic mononuclear cells (MNC) were collected from the layer of interphase and washed with PBS to remove residual Percoll. The isolated hepatic MNC were resuspended in FACS buffer for the following staining. To determine the IL-17A producing T cells, MNC (5 × 105/reaction) were incubated with a cocktail of surface marker antibodies including CD3-PE (clone 17A2), CD4-APC (clone GK1.5) and γδ TCR-Brilliant Violet 421(clone GL3). The stained cells were then were fixed and permeabilized with CytoFix/perm solution (BD Bioscience 554714), followed by washing with BD wash buffer and staining with anti-IL17A-PE.Cy7 (clone TC11-18H10) and IFN-γ-Alexa Flour 700 (clone XMG 1.2) for 30 min on ice in the dark. After washing with FACS buffer, cells were resuspended in 1×PBS and analysed with BD FACSVerse. To determine monocytes and neutrophils, MNC were incubated with a cocktail of surface markers, including CD11b-Alexa Fluor 647 (clone M1/70), Ly6G-FITC (clone 1A8), Ly6C-PE (clone AL-21), CCR2-Alexa Flour 700 (clone SA203G11) and CX3CR1-PerCP.Cy5.5 (clone SA011F11). The stained cells were then were fixed and permeabilized with CytoFix/perm solution, followed by washing with BD wash buffer and staining with anti- IL-1β-PerCP (clone B122), TNF-PE (clone MP6-XT22) and iNOS-FITC (clone 6/iNOS). Stained cells washed with FACS buffer and resuspend in the PBS for analysing the subpopulation with BD FACSVerse. All the data were processed using FlowJo software (version 10, Treestar).
Formalin-fixed paraffin embedded liver sections were deparaffinized in xylene, rehydrated in graded alcohols and washed in distilled water. Antigen retrieval for neutrophil binding epitopes was performed by boiling sections in citrate buffer (10 mM, pH 6.0) for 24 min, follow by incubating the slides with trypsin (1 mg/ml) at 37 °C for 15 min. To block endogenous peroxidase activity, slides were incubated with 3% H2O2.methanol. Liver sections were washed and incubated with 10% BSA in PBS containing biotin (25% v/v) and avidin (25% v/v) to minimize non-specific binding of primary antibody. Slides were then incubated with primary antibody, anti-Ly6B.2 (1:200, clone: 7/4, Biorad) or isotype control (RIgG2a, 1:200) overnight at 4 °C. After washing, the sections were incubated with diluted secondary antibody (rabbit anti-rat IgG, 1:200, Vector BA4001), followed by VECTASTAIN ABC reagent (Vector PK6100). Slides were developed with chromogen diaminobenzidine (DAB, DAKO K3468) and counterstained with Gills Haematoxylin. Stained sections were digitized using Aperio Digital Pathology System and analyzed with ImageScope V9 softeware.
Liver histological scores and assessment of liver injury
The liver lobes were collected from experimental animals and placed in neutral buffered formalin (Sigma-Aldrich). The formalin-fixed tissue was then embedded in paraffin, cut in 5-microm sections, and stained with hematoxylin and eosin. For the degree of liver, inflammation was determined by a blinded histopathology score as previous study50. In brief, a score of 1 indicates that the number of micro abscesses on each liver section was below 10 and no necrosis region was found. A score of 2 indicates that the number of micro-abscesses on each liver section was >10 but <20 and no necrosis region was found. A score of 3 indicates that the number of micro-abscesses on each liver section was >20 but <30 and no necrosis region was found. A score of 3 indicates that the number of micro-abscesses on each liver section was >20 but <30 and no necrosis region was found. A score of 4 indicates that the number of micro-abscesses on each liver section was over 30 and no necrosis region was found. A score of 5 indicates that the number of necrosis region was below 5. A score of 6 indicates that the number of necrosis region was >5 but <10. A score of 7 indicates that the number of necrosis region was >10 but <15. A score of 8 indicates that the number of necrosis region was >15. The average score in each group was generated by the examination of liver sections from 5 mice. Serum levels of alanine ALT were detected in blood obtained via cheek pouch bleeding and as an indicator of hepatocellular injury. Samples were analyzed with Fuji Dri-Chem 4000i and results are expressed as international units per liter of serum.
Microbial glycan microarray screening
The interaction of CLEC5A.Fc with the microbial glycan array was determined in Dr. Richard Cumming’s laboratory, as described previously42. The Fc-tagged CLEC5A sample and IgG control were diluted to 200 mg/ml in TSM binding buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM CaCl2, 2 mM MgCl2, 0.05% v/v Tween-20, and 1% w/v BSA), and 100 ml was added to the surface of the Microbial glycan microarray (MGM) slide. After placing a coverslip over the surface, samples were incubated for 1 h at room temperature in the dark. Slides were washed 4× in TSM wash buffer 1 (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM CaCl2, 2 mM MgCl2, 0.05% v/v Tween-20) and 4× in TSM wash buffer 2 (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM CaCl2, 2 mM MgCl2). Detection of binding was achieved with Alexa Fluor-488-labeled anti-human IgG (Invitrogen) diluted to 5 mg/ml in TSM binding buffer on the slide for 1 h at room temperature in the dark. Slides were washed 4× in TSM wash buffer 1, 4× in TSM wash buffer 2, and 4× in water. Binding was quantitated with a microarray scanner (Scan Array Express, PerkinElmer Lifer Sciences) and processed in terms of average relative fluorescence intensity (RFU) of 4 replicates using ImaGene software (BioDiscovery).
Extraction of cell wall components from L. monocytogenes
The preparation of cell wall and sugar component of L. monocytogenes was as described previously51, 52. Briefly, L. monocytogenes (0.5 g) was ground and subjected to alkaline lysis (0.5 N NaOH, 2.0 ml, 60 °C, 8 h), followed by neutralization with 1 M AcOH (1.0 ml) and removal of solvent by a rotavapor. The precipitate was resuspended in MeOH (5 ml × 2), followed by centrifugation at room temperature for 30 min (4000 rpm) to collectsupernatant (MeOH-soluble fraction). The undissolved precipitate was resuspended with H2O (5 ml × 2), followed by centrifugation at room temperature for 30 min (4000 rpm) to collect the supernatant, followed by dialysis in distilled water (H2O-soluble fraction). The collected fractions were lyophilized and used for TA sugar compositional analysis.
To determine CLEC5A binding to cell wall components, bacteria cell wall extracts (100 μg/ml fraction; 100 μl per well) were incubated on Maxi-Sorp ELISA plate (Nunc #469949) overnight at 4 °C, followed by addition of CLEC5A.Fc fusion protein (0–200 pM). The bound CLEC5A.Fc fusion protein were detected by HRP conjugated anti-human IgG antibody with 3,3′,5,5′-tetramethylbenzidine (TMB) (BD Pharmingen) substrate. In this ELISA assay, fusion protein and secondary antibody were diluted in a commercial protein-free blocking buffer (Thermo #37585). For cell- based binding assay, bacteria were labeled with Syto 60® (molecular probe S11342) at the finial concentration of 5 μM for 30 min at room temperature, followed by incubation with 293T cells (1 × 105) overexpressing human CLEC5A and TLR2, respectively, at 4 °C for 30 min. Samples were washed with PBS to remove unbound bacteria, followed by flow cytometry analysis to detect bacteria-cell interaction.
All the data were presented as mean ± SEM and analyzed using GraphPad Prism software (Version 5.0). An unpaired, two-tailed student’s t-test (for parametric data) or Mann–Whittney test (for nonparametric data) was used to determine the significance between two sets of data. When more than two groups were compared, a one-way ANOVA with the post hoc Bonferroni test (for parametric data) or a Kruskai–Wallis with post hoc Dunn’s test (for nonparametric data) was used for multiple comparisons. Kaplan–Meier survival curves with log rank were used for analyzing mouse survival.
All the data generated or analyzed during this study are available within the article and Supplementary Information files, or available from the authors upon request.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Drs John Kao, Caroline Milner, Fu-Tong Liu, Jen-Tsan Ashley Chi, and Judith Chou for critical comments. We are grateful for Drs Wei-Chieh, Cheng, and Tsui-Ling Hsu provides peptidoglycan and technical assistance. Dr Richard Cumming, Dr Chih-Ya Yang, Xing-Jei Lin, I-Shuen Tsai, Sung Pei-Shen, and Li-Wen Lo for technical assistance. We are indebted to Dr Daniel Portnoy, Dr Sean Stowell, Dr Martin Loessner, Dr Yang Shen, and Dr Fu-Tong Liu for sharing various serogroups of L. monocytogenes. We are also appreciative to the technical services provided by the “Transgenic Mouse Model Core Facility of the National Core Facility Program for Biotechnology, National Science Council” and technical services provided by Flow cytometry Core Facility of National Yang Ming University. CLEC5A-microbial glycan interaction was performed by the Consortium for Functional Glycomics (Richard D. Cummings). This work was supported by Academia Sinica and Ministry of Science and Technology (MOST 106-2321-B-001-037, MOST 103-2320-B-001-010-MY3, MOST 104-2320-B-010-0440-MY3, Taipei Medical University (TMU102-AE1-B26), Yen Tjing Ling Medical Foundation (CI-106-16) and Summit and Thematic Research Projects.
Electronic supplementary material
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.