Based on the reported chitinase-based MAMP-generating system in plants (5), we first sought to investigate whether circulating enzymes that generate MAMPs for specific PRRs may play a role in chitin sensing in humans, e.g., for TLR2 activation (9). Since acidic mammalian chitinase displays exochitinase activity (41), it would be expected to release only di-GlcNAc subunits, which are too small to stimulate TLR2 (9). However, in contrast to acidic mammalian chitinase, CHIT1 is an endochitinase, cleaving inside chitin chains (17, 42). Thus, CHIT1 could enable the generation of longer oligomeric chitin fragments (31) that may directly stimulate TLR2. To test this hypothesis, purified and LPS-free commercially available macroscopic chitin flakes from shrimp (sieved size approximately 0.5 × 1–2 mm; Supplementary Figure S1A and Materials and Methods) were digested with purified recombinant 39-kDa (lacking the CBD) or 50-kDa (containing the CBD) recombinant CHIT1 [using expression constructs and purification procedures validated before (43); see Materials and Methods] and the product-containing reaction transferred to TLR2-expressing NF-κB reporter HEK293T cells (HEK293T-TLR2). Interestingly, while undigested macroscopic shrimp chitin was unable to activate TLR2, incubation with 39-kDa CHIT1 rendered it TLR2-active ( Figure 1A ). A similar emergence of TLR2 activity was observed ( Figure 1B ) when the experiment was repeated using a transwell setup ( Supplementary Figure S1B , 8-µm pore size), separating the millimeter-scale chitin flakes and any fragments > 8 µm in the reaction from the cells. By doing so, we excluded the possibility that the observed CHIT1-induced TLR2-stimulating activity may be mediated by direct contact of cells (and hence TLR2) with the non-oligomeric macroscopic chitin. The acquisition of TLR2 activity of TLR2-transfected HEK293T cells ( Figure 1B ) was confirmed using Pam3 as a control stimulus ( Supplementary Figure S1C ). Conversely, lower-molecular-weight chitin oligomers (termed C10-15 throughout; estimated molecular weight 2,000–3,000, DP10-15, size ~5 nm; 9) were able to pass the filter pores and activate the cells ( Supplementary Figure S1C ). Although recombinant CHIT1 was clearly not able to activate TLR2 by itself, we further ruled out a possible role of confounding contaminants in recombinant CHIT1 protein preparations by ectopically expressing CHIT1 from a plasmid in HEK293T. In CHIT1 plasmid-transfected cells, both CHIT1 isoform proteins were secreted as expected (see Supplementary Figure S1D for immunoblots of cell supernatants). Importantly, incubation of CHIT1-containing cell culture media from these cells with macroscopic chitin flakes also rendered the otherwise inactive macroscopic chitin TLR2-active upon transfer to TLR2–HEK293T cells ( Figure 1C ). To demonstrate further that activation by CHIT1 was truly due to enzymatic degradation of macroscopic chitin, we mutated the catalytic site of CHIT1 (D138A E140L, cf. 44; Supplementary Figure S1E ). Medium from cells transfected with a mutant CHIT1 (mCHIT1) plasmid did contain secreted mCHIT1 ( Supplementary Figure S1D ) but, as expected, lacked chitinase activity ( Supplementary Figure S1F ). More importantly, the mutant was unable to generate TLR2 activity from macroscopic chitin flakes, compared to media from cells transfected with wild-type (WT) CHIT1 plasmid ( Figure 1D ). We conclude that especially the 39-kDa form of CHIT1 is able to generate diffusible TLR2 activating ligands from macroscopic shrimp chitin. When analyzing digestion reactions from CHIT1 incubation with chitin flakes, we were able to detect highly acetylated products of DP2-4 ( Supplementary Figure S1G ). Such oligomers are too short to activate TLR2 (9), but the analysis verifies CHIT1 endochitinase activity. Generally, it is highly challenging to detect intermediate enzyme products of DP6 or greater from such a cell culture system as their half-life, concentration, solubility, and ionization efficiency during mass spectrometry are exceedingly low. Provided that Kimura et al. (45), albeit in a cell-free system (45), detected fully acetylated chitin oligomers of DP6 to be released from a similar macroscopic substrate, colloidal chitin, and that at least DP4 oligomers were confirmed, we consider it highly plausible that CHIT1 degradation releases intermediate products that are sufficiently long and abundant for TLR2 engagement, which occurs with nanomolar K_D (9).

We then went on to investigate whether this effect applied to polymeric chitin from other sources, such as the disease-relevant HDM, Dermatophagoides pteronyssinus, and the pathogenic fungus, C. albicans. Using the same digestion assay, we observed a significant increase of reporter activity in HEK293T-TLR2 cells, especially upon pre-incubation of D. pteronyssinus whole cultures and mite bodies with recombinant 39-kDa CHIT1 in both the normal and transwell setups ( Figure 1E ). The same effect was observed for chitin derived from heat-killed C. albicans yeast and hyphae ( Figure 1F ). In this experiment, zymolyase, an enzyme cleaving β-glucan chains, only had a minor effect that may be attributable to degrading the scaffolding with which chitin is integrated into the fungal cell wall (46). The experiment was also performed specifically with macromolecular, highly purified C. albicans chitin (47), which triggered drastically increased TLR2-mediated activity upon 39-kDa CHIT1 digestion ( Figures 1G, H ). Collectively, these data show that CHIT1 is able to generate potent and diffusible TLR2-activating MAMPs from polymeric, poorly activating macroscopic chitin, reminiscent of observations from Drosophila (12) and plants (5).

To verify the effect of CHIT1 on fungal cell wall chitin, confocal fluorescence microscopy analysis of C. albicans yeast cells was performed upon incubation with CHIT1. For comparison, yeast cells were also treated with caspofungin, an inhibitor of β-1,3-d-glucan synthesis, which is known to increase chitin exposure on the fungal cell wall (48). Untreated and caspofungin-treated cells were heat-inactivated and pre-incubated with 39-kDa or 50-kDa CHIT1. Finally, yeast cells were labeled with Alexa647-conjugated, chitin-binding wheat germ agglutinin (WGA) lectin and Alexa488-conjugated concanavalin (Con) A, which binds to α-mannans and hence served as a control that should be unaffected by CHIT1. Confocal imaging revealed that WGA bound more prominently to C. albicans yeast cells after CHIT1 digestion ( Figure 2A , quantified in Figure 2B ), whereas the binding of ConA remained largely unchanged. Additionally, we stained for CHIT1 itself using anti-His primary and matching Alexa594-conjugated secondary antibodies. At the fungal cell wall, 50-kDa CHIT1 was visualized ( Figure 2C ), indicating a stable binding consistent with the presence of a CBD. No detectable binding of 39-kDa CHIT1 was observed, probably due to the lack of a CBD and hence only transient interaction with chitin (43). The lower retention time of the 39-kDa form of CHIT1 at the cell wall would be consistent with the generation of longer soluble oligomers, which are more favorable TLR2 ligands (9) due to lower processivity. In summary, recombinant CHIT1 can directly engage with the fungal cell wall, consistent with the release of oligomeric chitin fragments.

The TLR2–HEK293T system lacks many PRRs except TLR2 and TLR2 co-receptors (49). To extend our insights to a more physiologically relevant system, primary bone marrow-derived macrophages (BMDMs) were stimulated with native and CHIT1-digested D. pteronyssinus HDMs (whole cultures and mite bodies) or C. albicans hyphae and yeast. BMDMs produced IL-6 and TNF in response to undigested D. pteronyssinus and C. albicans. However, the stimulatory capacity of HDM bodies ( Figures 3A, B ) and C. albicans hyphae ( Figures 3C, D ) increased following CHIT1 39-kDa digestion. Parallel analysis of Tlr2 knockout (KO) BMDMs and use of the transwell setup showed that HDM preparations were not fully TLR2-dependent and thus probably contained other MAMPs that were not detectable in HEK293T cells but were increased by CHIT1 incubation and dominated over chitin in BMDMs. Moreover, BMDM levels of cytokine release were more similar when comparing transwell and direct incubation ( Figures 3A, B ). For C. albicans hyphae, however, the transwell setting clearly was associated with lower cytokine release but stronger TLR2 dependence than for direct contact ( Figures 3C, D ). The expected inter-individual differences in maximal release between primary cells from different animals precluded confirming these differences statistically. Nevertheless, observed trends suggest that in BMDMs and for C. albicans, TLR2 is able to elicit cytokine release to chitin-containing pathogens upon direct binding but, more importantly, can also sense diffusible ligands generated by CHIT1 for IL-6 and TNF secretion. Thus, CHIT1 digestion enables PRR-expressing cells to overcome the requirement of direct physical contact with chitin-bearing pathogens using oligomer- and diffusion-mediated distal chitin sensing.

A prominent feature of certain innate immune signaling pathways is the involvement of soluble and membrane-bound transfer proteins such as LBP and CD14, which both promote LPS sensing via TLR4, and Pam₂CSK₄ (Pam2) and Pam3₃CSK₄ (Pam3) lipopeptide sensing via TLR2/TLR6 and TLR1/TLR2, respectively (35, 36, 50). To test whether LBP and CD14 may play a similar role in shuttling oligomeric DP10-15 chitin (C10-15, see 9), we added recombinant LBP exogenously to and/or co-expressed CD14 in TLR2–HEK293T cells before measuring NF-κB reporter activity. Evidently, CD14 overexpression strongly increased TLR2-mediated NF-κB activity, and this was further increased by the addition of LBP similar to Pam3 ( Figure 4A ; Supplementary Figure S2 ), indicating that oligomeric chitin sensing employs components of other TLR ligand sensing systems in this recognition pathway. We next tested whether oligomeric chitin TLR2 sensing also required co-receptors such as TLR1 and TLR6 (51) using anti-TLR1 and anti-TLR2 blocking antibodies in WT BMDMs ( Supplementary Figure S3A ). Evidently, the response to all TLR2 ligands (including C10-15), but not LPS, was blocked by anti-TLR2 blockade. There was a subtle effect of anti-TLR1 blocking. To verify this further, we moved to a genetic system and compared WT and Tlr1 knockout (−/−, KO), Tlr2 KO, or Tlr6 KO BMDMs. Interestingly, in response to the oligomeric C10-15 chitin and Pam3, Tlr6 KO BMDMs responded like WT BMDMs by producing IL-6 and TNF, whereas both Tlr1 KO and Tlr2 KO BMDMs were reduced in response to C10-15 and Pam3 ( Figures 4B, C ; Supplementary Figures S3B, C ). This genetic analysis implicated both TLR1 and TLR2 in chitin sensing of immune cells, consistent with the earlier use of blocking antibodies directed against TLR1 or TLR6 in the HEK293T system (9). To validate TLR1 involvement further, TLR1 antibody-blocking experiments were also extended to BMDMs ( Supplementary Figure S3C ). To show that oligomeric chitin also triggered receptor dimerization as shown for TLR4 and TLR2, we conducted fluorescence complementation assays of TLR2 with the co-receptors, TLR1 or TLR6. In brief, we used expression constructs in which the TLR cytoplasmic domain is C-terminally fused to either the N- or C-terminal portion of the mLumin protein, a derivative of the far-red fluorescent protein variant of mKate (52). In this system, stable dimer formation is evidenced by the detection of mLumin fluorescence (52). In confirmation of KO and blocking results, C10-15 and Pam3 indeed induced the highest mLumin fluorescence in TLR1- and TLR2-expressing cells ( Figure 4D ), and Pam2 did so in TLR2- and TLR6-expressing cells ( Figure 4E ). Quantification of these data showed that TLR1/TLR2 co-expression induced significant mLumin fluorescence only upon stimulation with C10-15 oligomeric chitin or Pam3 ( Figure 4F ), whereas TLR2/TLR6 co-expression did so only upon stimulation with Pam2 ( Figure 4G ). Similar results were obtained for NF-κB induction in cells expressing TLR2 and co-receptors ( Figure 4H ). Additionally, we found that certain cell types, for example, N/TERT-1 keratinocyte-like cells (53), naturally responded to Pam2, whereas they failed to respond to both C10-15 and Pam3 ( Figure 4I ), indicating that the latter agonists behave similarly and that both prefer TLR1/TLR2 heterodimers. This is consistent with in silico analyses, which showed that a chitin decamer can be fitted into the structure of a TLR1/TLR2 heterodimer (PDB: 2Z7X (38), Figure 4J ), whereas the absence of a hydrophobic pocket in TLR6 prevented this in a TLR2/TLR6 heterodimer structure (PDB: 3A79; 39). Collectively, our results suggest that oligomeric chitin sensing in mammals relies not only on the generation of oligomeric MAMPs by CHIT1 but also on soluble and membrane-bound co-receptors, such as CD14 and TLR1, for optimal TLR2 signal induction.

Our data thus far suggest that CHIT1 can generate diffusible, TLR2-activating oligomers from chitin-containing organisms. We therefore became interested in how CHIT1 may be regulated. In murine BMDMs, RT-qPCR analyses revealed a relatively low expression of Chit1 compared to the housekeeping gene Tbp (encoding the TATA box binding protein). Exposure to oligomeric chitin (C10-15) or Pam3 did not increase Chit1 transcription ( Figure 5A ); neither did HDM bodies ( Figure 5B ). However, C. albicans hyphae induced significantly more than twofold Chit1 mRNA levels after 2-h stimulation ( Figure 5C ). The effect was transient and returned to baseline within 6 h. To test whether this induction was TLR2-dependent, WT and Tlr2 KO BMDMs were compared. Whereas Pam3, C10-15, and C. albicans hyphae elicited a strong, Tlr2-dependent Il6 upregulation, C. albicans hyphae induced Chit1 and Il6 in a Tlr2-independent manner ( Figures 5D, E ). Thus, in murine macrophages, rapid induction of Chit1 was independent of TLR2 and oligomeric chitin. These results are consistent with the fact that, in fungal cell walls, chitin is initially buried under β-glucan and mannoproteins (1, 54). At first encounter with a fungal pathogen, there would thus be little accessible chitin, and an induction of chitinase activity via another MAMP such as β-glucan would, hence, be more plausible. Therefore, we speculated that Chit1 mRNA induction by C. albicans—which is consistent with earlier analyses using macromolecular chitin in primary macrophages (9)—could be mediated by β-glucan sensing through Dectin-1 (encoded by Clec7A; 55). We compared the expression of Chit1 and Il6 using RT-qPCR analysis in mRNA extracted from WT and Clec7a KO immortalized macrophages (9, 56) stimulated with C. albicans, Pam3, C10-15, and zymosan, a yeast cell wall extract sensed through TLR2 and Dectin-1 (9). Interestingly, soluble TLR2 agonists (Pam3 and C10-15) did not show differences between cell types for Chit1 and Il6 transcription ( Figures 5F, G ). However, C. albicans and zymosan induced drastically less Chit1 and Il6 by Clec7a KO macrophages ( Figures 5F, G ). These data indicate that Dectin-1 may be responsible for inducing the expression of Chit1 in response to C. albicans. This would be consistent with β-glucan being a major and more readily exposed fungal MAMP (55) that may prime CHIT1-mediated oligomeric chitin MAMP generation.

Based on these results, increasing CHIT1 activity and the subsequent emergence of diffusible TLR2 ligands during infection over time would mean an increasing contribution of TLR2-dependent immunity to antifungal responses, which have been demonstrated in infection models (18, 33). This would make CHIT1 a potential target for fungal counterstrategies. Indeed, in plants, chitinases have been shown to represent proteolytic targets of fungal proteases (5). Since C. albicans can express multiple secreted aspartyl proteinases (Saps) (57), we tested whether C. albicans culture supernatants contained proteolytic activity to which CHIT1 may be sensitive. We incubated (His-tagged) recombinant CHIT1 preparations with fungal supernatants from C. albicans grown under conditions known to induce Sap secretion (in “induction media”, 58). This led to a decrease of anti-His signal, i.e., CHIT1 protein levels ( Figure 6A ), when compared to samples without C. albicans supernatant (lanes 9 and 10 in Figure 6A ). Degradation was not observed when C. albicans supernatant was harvested from fungal cultures that had been grown in standard yeast extract–peptone–dextrose (YPD) media ( Figure 6B ), where Saps are only expressed at low levels (58). This was confirmed by immunoblot quantification analysis from several experiments ( Figures 6C, D ). These results indicate that under these given conditions, C. albicans cells may secrete Saps to target CHIT1. To test this hypothesis further, CHIT1 was also exposed to purified recombinant C. albicans Sap2 and Sap6, representing Saps expressed by yeast and hyphae, respectively. Both 50-kDa and 39-kDa forms of CHIT1 were dose-dependent on the proteolytic activity of Sap2 and Sap6 ( Figure 6E ), as the effect was blocked by the aspartic protease inhibitor pepstatin A ( Figure 6F ). Thus, Chit1 is sensitive to the fungal virulence factors and proteases Sap2 and Sap6 and may therefore be targeted by the pathogenic fungus, C. albicans.

All chemicals used in the lab were from Sigma-Aldrich unless otherwise stated. The source and origin of all the ligands, recombinant proteins, and inhibitors are listed in Supplementary Table S1 . The preparation of C10-15 chitin oligomers has been described before (9). In brief, C10-15 chitin oligomers were generated from C10-15 chitosan (2,000–3,000 MW, equivalent to 10–15 subunits, Carbosynth), which was derived from crab shells, chemically hydrolyzed and High performance liquid chromatography (HPLC)-fractionated to >95% purity as confirmed by HPLC and mass spectrometry analysis; see Fuchs et al. (9) for further details. Using sodium bicarbonate and acetic anhydride acetylation (90), chitosan was acetylated to chitin. The resulting degree of acetylation was assessed upon trifluoroacetic acid hydrolysis (for C10-15, 2 h at 100°C) by Electrospray ionization (ESI) and Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry; see Fuchs et al. (9) for further details. Only batches with up to 90% of acetylation were used here. Prior to using sterile stimulation, acetylated chitin oligomers were suspended in endotoxin-free water and tested for endotoxin level using the limulus amebocyte lysate (LAL) assay (Lonza, CH). Levels below 0.25 EU/mL (<25 pg/mL LPS) in final dilutions were considered acceptable. For levels >0.25 EU/mL, the chitin preparation was incubated for 3 h with 10 µg/mL of the LPS binding agent polymyxin B (Thermo Fisher), washed by centrifugation, and re-assessed.

The plasmids used in this study are listed in Supplementary Table S2 . Gateway-compatible split-mLumin backbone plasmids for bimolecular complementary assay were described in Weiler et al. (91). Full-length TLR1 and TLR2 Open reading frame (ORF)-containing Gateway Entry plasmids were obtained from the PlasmID Repository at Harvard Medical School. With respect to TLR6, a full-length ORF flanked with Gateway attB sites was synthesized by GENEWIZ and transferred by BP Clonase reaction into an attL site-containing Gateway Entry plasmid. To transfer the hTLR1, hTLR2, and hTLR6 ORFs into the split-mLumin N-terminal and C-terminal Gateway Destination vectors, LR reactions using LR Clonase II Enzyme mix kit (Thermo Fisher) were used. The destination vectors were validated by restriction enzyme BsgI (New England Biolabs) (200–300 ng plasmid and 5 units of enzyme in 1X Tango Buffer, 1 h at 37°C water bath) and Sanger sequencing (GATC Biotech). After sequencing, correct protein expression was verified by Western blotting (see details below). Plasmids encoding the 50- and 39-kDa forms of CHIT1 as C-terminally His-tagged constructs have been described before (43) and used exactly as described. To simultaneously introduce the two point mutations D138A and E140L in the catalytic site of the chitinase, the QuikChange II XL site-directed mutagenesis kit (Agilent) was used together with mutagenesis primers ( Supplementary Table S3 ) designed by following the QuikChange manufacturer’s instruction and verified in silico using the Geneious R6 software (6.1.8 version). The presence of the desired mutation and the absence of unwanted mutations were confirmed by Sanger sequencing (GATC Biotech). Secretion of the proteins was confirmed by immunoblot, and the catalytic inactivity of the mutant CHIT1 proteins was checked by measuring chitinase activity.

C. albicans strain SC5314 was used in this study, in which the stock was a kind gift from Dr. Anurag Singh (Universitätsklinikum Tübingen), and stored frozen at −80°C in RPMI medium only (without Fetal calf serum (FCS)) containing 20% glycerol. When needed, cells were taken up from the frozen stock and grown at 30°C in YPD agar medium [1% (w/v) BactoYeast extract, 2% (w/v) BactoPeptone, 2% (w/v) dextrose, and 2% (w/v) agar] on 10-cm dishes. After overnight incubation, cells were stored at 4°C for up to 1 month. Before any experiment or treatment, cells were freshly prepared by sub-culturing from a 10-cm dish to a glass slant tube at 30°C. Cells were harvested by picking up a smear and re-suspending it in RPMI-1640 medium. To expose the chitin content on the surface of C. albicans, yeast cells were incubated with 0.032 µg/mL caspofungin (Sigma-Aldrich) for 6 h at 30°C in RPMI-1640 medium. After incubation, yeast cells were washed twice with Dulbecco's Phosphate Buffered Saline (dPBS) and tested once for viability. For hyphal induction, 1 × 10⁶ live yeast cells were re-suspended in YPD medium with 20% FCS and incubated at 37°C for at least 3.5 h until 90%–95% filamentation was observed. The hyphae were collected via a cell scraper and then washed twice with dPBS. For heat killing, both C. albicans yeast and hyphae were prepared by incubation at 65°C in a water bath for 1 h, with killing confirmed by plating in a YPD agar slant tube. For Saps, C. albicans was cultured in an induction medium (2 mM MgSO₄, 7.3 mM KH₂PO₄, 1% glucose, 0.5% bovine serum albumin, and 1% 100× HEPES, pH 4.0) for 48 h at 30°C as described previously (58). The culture supernatants were used to incubate with recombinant CHIT1 and to test the degradation effect by immunoblot.

Recombinant 50-kDa and 39-kDa CHIT1 proteins were expressed as C-terminally 8xHis-tagged constructs in HEK293-6E cells and purified from the cell culture supernatant by His-Trap (GE Healthcare) affinity as described in Stockinger et al. (43). C. albicans Sap2 and Sap6 were expressed in Pichia pastoris and purified via ion exchange chromatography and size exclusion as described in Schild et al. (92).

House dust mite D. pteronyssinus whole cultures and mite bodies were obtained from CITEQ Biologics. The powder of whole cultures and mite bodies was weighed and re-suspended in dPBS. Polymyxin B (Thermo Fisher) was then applied to a final working concentration of 100 units/mL to remove endotoxin. After incubation for at least 1 h at Room temperature (RT), whole cultures and mite bodies were re-suspended in dPBS to 10 mg/mL (stock concentration) and stored at −20°C. Whole cultures and mite bodies at a working concentration of 100 µg/mL were used. Chitin flakes from shrimp shells were bought from Sigma-Aldrich (Catalog # C9213-500G). The flakes were first sieved using 2-mm or 1-mm pore size steel sieves (Amazon) to sort small pieces of flakes ( Supplementary Figure S1A ). The obtained flakes were picked up by clean forceps into sterile 1.5-mL Eppendorf tubes and treated with polymyxin B (10 µg/mL) for 3 h at RT. The flakes were washed three times with dPBS, finally re-suspended into 1 mL dPBS, and stored at 4°C.

Chitin flakes, C. albicans preparations, house dust mite whole cultures, and mite bodies were incubated with human recombinant CHIT1 for digestion. Both 50-kDa and 39-kDa recombinant CHIT1 (stock solution 1 µM) were used at a working concentration of 4 nM (1:250 dilution). For expression of WT or mutant CHIT1-encoding plasmids in HEK293T cells, 250 ng of CHIT1 plasmids was transfected into HEK293T cells supplemented with 250 µL Dulbecco's Modified Eagle Medium (DMEM) complete medium. After 48-h transfection, the collected cultured medium was used for chitinase digestion. The secretion of both CHIT1 isoforms was confirmed by immunoblot. To digest chitin flakes, C. albicans preparations, house dust mite whole cultures, and mite bodies were incubated with recombinant or secreted CHIT1-containing media in 1.5 mL Eppendorf tubes on a rotation wheel for 18 h at RT. After CHIT1 digestion, the culture supernatants were used for stimulation or applied to transwells.

HEK-Dual™ hTLR2 cells were bought from InvivoGen. The cells were derived from human embryonic kidney 293 (HEK293) and stably transfected with the human TLR2 gene, an NF-κB/AP-1 inducible secreted embryonic alkaline phosphatase (SEAP) reporter construct, and a Lucia luciferase (a secreted luciferase), inserted under the control of the endogenous human IL8 promoter. Cells were kept under the antibiotic selection of Hygromycin B and Zeocin.

Macrophage progenitors derived from the bone marrow of C57BL/6 wild-type and Clec7a ^−/− mice were immortalized as described (56) and a kind gift of Philip Taylor, Cardiff University. Progenitor cells were kept and cultured in the RPMI-1640 complete medium supplemented with 1 µM (β-estradiol) and 10% murine granulocyte-macrophage colony-stimulating factor (mGM-CSF). Progenitor cells were split 1:10 depending on demand and maintained for less than 10 passages. Before differentiation, progenitor cells were washed and centrifuged twice with dPBS to remove β-estradiol. A density of 5 × 10⁵ progenitor cells was then seeded in a 12-well plate with 1.5 mL RPMI-1640 complete medium supplemented with 10% mGM-CSF without β-estradiol. An additional 1 mL of fresh culture medium was added on day 2. On day 5 of differentiation, adherent cells as differentiated macrophages were harvested and plated in 12-well plates (2 × 10⁶ cells/well) in RPMI-1640 complete medium without mGM-CSF. After 1 day, the cells were ready for stimulation.

N/TERT-1 cells were a gift from Prof. James Rheinwald (53) and acquired from local collaborator Birgit Schittek, Dermatology Department, University Hospital Tübingen. Cells were passaged and maintained in CnT-07 medium (Epithelial Proliferation Medium, from CELLnTEC, Catalog # CnT-07) for less than 10 passages. Before stimulation, N/TERT-1 cells were seeded at a 96-well plate (2 × 10⁴ cells/well) in CnT-07 medium. After 2 days of resting, cells were ready for stimulation. After 24 h of stimulation, supernatants were subjected to ELISA according to the manufacturer’s instructions (BioLegend).

C57BL/6 wild-type and Tlr2 KO mice on a C57BL/6 background (originally a gift from H. Wagner, Ludwigs-Maximilian University, Munich) were maintained in the animal facility of the Department of Immunology, University of Tübingen, used between 8 and 20 weeks of age, and sacrificed using CO₂. Animal breeding, handling, and sacrificing between 8 and 20 weeks of age were performed according to the local institutional guidelines and institutionally approved protocols. WT, Tlr1 KO, Tlr2 KO, and Tlr6 KO mice on a C57BL/6J background have been described before (93) and maintained at the animal facility of Lausanne University Hospital, Lausanne, Switzerland. Animal experiments performed in Lausanne were approved by the Service des Affaires Vétérinaires, Direction Générale de l’Agriculture, de la Viticulture et des Affaires Vétérinaires, état de Vaud (Epalinges, Switzerland) under authorizations 876.9 and 3587 and performed according to Swiss and ARRIVE guidelines. Bone marrow cells were isolated from femurs and tibias, which were cut and flushed out using a 24-gauge syringe, and re-suspended in RPMI-1640 medium supplemented with 10% FCS. A density of 1.2 × 10⁷ cells was seeded in 10-cm Petri dishes in RPMI-1640 complete medium containing 10% culture supernatant from mGM-CSF-producing cells (94). An additional 5 mL of fresh culture medium was added on day 3. On day 7 after differentiation, adherent cells were harvested and plated in 96-well plates (1 × 10⁵ cells/well) or 12-well plates (2 × 10⁶ cells/well) in RPMI-1640 complete medium without mGM-CSF. After 1 day of resting, cells were ready for stimulation.

Transwell inserts for 24-well plates (8-µm pore size, Nunc) were used, and 250 µL of media containing chitinase-digested chitin flakes, C. albicans, or house dust mite bodies was filled into the upper reservoir, which was placed onto the cell culture plate containing 250 µL media and TLR2-transfected HEK cells or BMDMs at its bottom. After 18-h incubation, almost two-thirds of the culture medium from the layer of transwell eventually passed through to the cell culture plate. The transwell was then removed and checked under the microscope. Large particles like chitin flakes and house dust mites were always still trapped in the upper part of the transwell insert. For a schematic impression, see Supplementary Figure S1B . For C. albicans, minor amounts of yeast cells or hyphae were sometimes found in the lower compartment after incubation. Cells or culture supernatants were processed as described for the respective assays.

Experiments using blocking antibodies were conducted as described previously (9). In brief, 5 × 10⁴ BMDM cells per well were seeded in RPMI media supplemented with 10% Fetal bovine serum (FBS) in 96-well plates. After 1 h of seeding, TLR1 or TLR2 blocking antibodies (InvivoGen) were added to a final concentration of 4 µg/mL, and cells were incubated for 1 h at 37°C, 5% CO₂—Pam2 (40 nM), Pam3 (40 nM), and LPS EB (40 nM; all from InvivoGen) or C10-15 (5 µM)—and incubated for 24 h. Isotypes were not included, as these commercially available reagents have been validated before, and LPS was included as a specificity control. Supernatants were then collected and measured for ELISA.

An inoculum of 5 × 10⁴ HEK293T cells was seeded in a 24-well plate with 500 µL DMEM complete medium and incubated overnight. The next day, cells were transfected with the following amounts of plasmid DNA per well: 100 ng NF-κB firefly reporter luciferase and 10 ng of Renilla luciferase under a constitutive promoter. To measure the TLR2 response, cells were transfected with 100 ng of human TLR2 or with 100 ng backbone as an empty vector control as described (9). Transfection was performed using 1 mL of X-tremeGENE™ HP DNA Transfection Reagent (Sigma-Aldrich) mixed into a total of 50 mL Opti-MEM together with the above-indicated plasmids. After 15 min of incubation at RT, 50 mL Opti-MEM-X-treme/plasmid mixture was added dropwise onto the cells. After 48 h of incubation, the medium was replaced by fresh DMEM complete medium with or without TLR agonists. Cells were stimulated for 18 h, and cell lysates were harvested. The concentrations of all stimuli and ligands are listed in Supplementary Table S1 . To analyze luciferase activity, the culture supernatants were removed, and the cells were washed with 350 mL dPBS. Cells were lysed in 60 mL passive lysis buffer (Promega). After shaking for 5 min, cells were frozen at −80°C for at least 15 min. Thawed lysates were cleared by centrifugation, and 10 mL of the cleared lysate was used to measure the luciferase activity on a FluoStar plate reader (BMG Labtech). In the plate reader, the substrates for firefly and Renilla luciferase were automatically injected. The analysis settings were used as recommended in the luciferase reporter system by Promega. The results were calculated as the ratio between firefly luminescence and Renilla luminescence ( Figures 1A–F , 4A, H ; Supplementary Figure S2 ).

Initially, 5 × 10⁴ HEK-Dual™ hTLR2 cells (InvivoGen) cells were seeded in a 96-well plate. After overnight incubation, the culture medium was exchanged with fresh DMEM complete medium with or without the stimuli. After 18 h, the cell culture supernatant was analyzed for NF-κB/AP-1-induced SEAP production using QUANTI-Blue reagents and IL-8-dependent expression of Lucia luciferase using QUANTI-Luc reagents (both InvivoGen). For QUANTI-Blue measurement, 20 µL of cell culture supernatant was added to 180 µL QUANTI-Blue solution in a 96-well flat plate. The plate was incubated for 30 min at 37°C. The SEAP levels were determined by a SpectraMax^® plate reader at 650 nm. For QUANTI-Luc measurement, 10 µL of cell culture supernatant was pipetted in a 96-well white plate. Lucia luciferase activity was measured using a FluoStar plate reader (BMG Labtech), which automatically added 50 µL of QUANTI-Luc solution. Data in Figures 1G, H were generated using this reporter assay.

Protein expression in HEK cells transfected with chitotriosidase- or TLR-encoding plasmids was verified by immunoblotting. In order to detect secreted chitotriosidase, after 24-h transfection and a further 24-h incubation in fresh Opti-MEM without any FCS, both whole cell lysates (WCL) and culture supernatants were harvested. For obtaining WCL, cells were washed once with dPBS and then lysed with 60 mL RIPA buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM Ethylenediaminetetraacetic acid (EDTA), 10% glycerol, 0.1% Sodium dodecyl-sulfate (SDS), 1% Triton X-100, and 0.5% deoxycholate) supplemented with EDTA-free protease inhibitor (Roche) and 0.1 mM Phenylmethylsulfonyl fluoride (PMSF). After maximum speed centrifugation, WCL were mixed with reducing reagent and sample loading buffer (Novex, Thermo Fisher) and then boiled for 5 min at 95°C to denature the proteins. For obtaining the supernatants, 60 mL of supernatants was treated as described for WCL. Samples (10 mL of WCL and 20 mL of supernatant) were subjected to electrophoresis on 10% Tris-glycine gels with SDS buffer (25 mM Tris-base, 250 mM glycine, and 0.1% SDS) and transferred to a 0.45-mm nitrocellulose membrane (GE Healthcare). The membrane was blocked with 5% bovine serum albumin (BSA) (w/v) in Tris-buffered saline solution with 0.1% (v/v) Tween-20 (TBS-T) for 1 h at RT and then left in 5 mL buffer containing primary antibody (see Supplementary Table S4 ) in 5% BSA in TBS-T overnight at 4°C with rotation. The next day, the membranes were washed three times with TBS-T for at least 5 min and then incubated with 10 mL TBS-T buffer containing secondary antibody in 5% non-fat milk for 1 h at RT with rotation. After incubation with secondary antibodies, membranes were washed three times with TBS-T for at least 5 min per wash. ECL reagent (Peqlab) was used to detect chemiluminescence, and the development was performed using a Licor camera Odyssey Imaging system in the chemiluminescence channel. Pictures were analyzed and edited in the Image Studio Lite software. The relative intensities of CHIT1 and His expression were also quantified by Image Studio. For the verification of TLR1, TLR2, and TLR6 constructs, TLR protein expression was verified using anti-TLR1, anti-TLR2, and anti-TLR6 as primary antibodies ( Supplementary Table S4 ). For whole protein staining, after transfer, nitrocellulose membranes were first rinsed with ddH₂O and then incubated with 5 mL Revert 700 Total Protein Stain Solution (Licor) for 5 min. After washing twice with 5 mL Wash Solution, the membranes were subjected to the Licor Odyssey Imaging system in the 700-nm channel. After imaging, the membranes were destained with 5 mL Revert Destaining Solution for 5–10 min until the stain was no longer visible to the eyes. The destained membranes were washed twice with ddH₂O, and the 5% BSA blocking was applied immediately.

Chitinase activity in the culture supernatant of CHIT1 plasmid-transfected HEK293T cells was measured using a fluorescence assay kit (Sigma-Aldrich, Catalog # CS1030). A 10-mL supernatant sample was mixed with 90 mL substrate solution containing 0.5 mg/mL of 4-methylumbelliferyl N,N′-diacetyl-β-d-chitobioside in an assay buffer provided as part of the kit and in a 96-well plate (black/clear bottom plate; Thermo Fisher). The reaction was incubated at 37°C for 30 min and then stopped by adding 100 mL stop solution. The fluorescence was measured at excitation of 360 nm and emission of 450 nm on a FluoStar plate reader (BMG Labtech).

Appropriate ELISA kits were used to quantify human or murine IL-6 and/or TNF concentrations according to the manufacturer’s recommendations (BioLegend). The 450-nm absorbance was measured using a SpectraMax^® plate reader.

Suspensions of 2 × 10⁴ HEK293T cells were seeded in 8-well 1.5H glass chamber slides (Ibidi). After 1 day of resting, cells were co-transfected with split-mLumin plasmids in the following two listed combinations: 1) hTLR2-LC151 (200 ng; mLumin C-terminal fragment) + hTLR2-LN151 (200 ng; mLumin N-terminal fragment) and 2) hTLR2-LN151 (100 ng; mLumin N-terminal fragment) + hTLR6-LC151 (300 ng; mLumin C-terminal fragment). After 48-h transfection, cells were stimulated with or without C10-15, Pam3, or Pam2 by replacing the culture medium. After 18 h of stimulation, HEK cells were gently washed once with 200 mL dPBS and then fixed with 4% paraformaldehyde in 150 mL dPBS for 10 min at RT. Cells were washed a second time and then stained with Hoechst 33342 (Invitrogen) at 1:10,000 dilution in dPBS for 8 min at RT. After the last wash, cells were mounted with Mounting Medium (Ibidi) and imaged using the Zeiss LSM 800 AiryScan Inverted Confocal Microscope at ×400 magnification.

Cultures of 2 × 10⁶ BMDMs were seeded into 12-well plates after 7 days of differentiation. After 1 day of resting, cells were stimulated with HDM, C. albicans, and TLR ligands for the indicated times. Cells were washed with dPBS and lysed in 350 mL RLT buffer (Qiagen) containing 1% (v/v) β-mercaptoethanol, and total RNA was isolated using the RNeasy Mini kit (Qiagen) following DNA digestion (RNase-free DNase set, Qiagen) and using a QIAcube instrument. The concentration of total RNA was measured using Nanodrop. mRNA reverse transcription to cDNA was performed using the High Capacity RNA-to-cDNA Kit (Thermo Fisher). Quantitative PCR was performed in a 10-mL mix containing 1 to 10 diluted cDNA, TaqMan Universal MasterMix II, and 0.3 mM TaqMan primer and RNA-free water. Each sample was analyzed in triplicates in a real-time cycler (Thermo QuantStudio 7 Flex, Thermo Fisher). The following cycle was used: 10 min/95°C; 15 s/95°C and 1 min/60°C for 40 cycles; and cooled and stored. The primers used are listed in Supplementary Table S5 .

To visualize a chitin decamer bound to a TLR1/TLR2 heterodimer in silico ( Figure 4J ), docking and an energy minimization simulation were performed as described previously (95). Briefly, 3D structures of different GlcNAc oligomers created with the Glycam Carbohydrate Builder (GLYCAM Web, at legacy.glycam.org) were docked into the monomers of the TLR1/TLR2 heterodimer (PDB: 2Z7X; 38) using AutoDock Vina-Carb 1.1.2 (96). One docking state each of (GlcNAc)₈ in TLR1 and (GlcNAc)₁₀ in TLR2 were selected and combined via PyMOL (Schrödinger) by aligning and pair fitting to create a TLR1/TLR2 heterodimer with a bound chitin decamer. This new complex is based on the mentioned crystal structure but features a slightly higher distance between the receptor monomers to allow optimal accommodation of the decamer. The outcome was finally optimized using the GROMACS 2019 (97) package with the force fields Amber ff14SB (98) and GLYCAM06 (99) by running an energy minimization simulation with the preparation and setup described previously (95).

Experimental data were analyzed in GraphPad Prism 8.3.1 (GraphPad Software, Inc.). Normal distribution was not formally tested but considered to apply judging from the data distribution. Hence, two-tailed Student’s t-tests and one-way or two-way ANOVA tests were used and adjusted for multiple testing as suggested by the analysis software and as indicated. Microscopy data were obtained using the ZEN blue (Zeiss) software and analyzed using ImageJ and Fiji. p-Values < 0.05 were generally considered statistically significant and were denoted by an asterisk throughout the figure legends, even if the actual p-values were considerably lower. Schematic images were created with BioRender.com.