C-type lectin-like receptor
fibronectin type II
Mannose receptor (MR) is the best characterised member of a family of four endocytic molecules that share a common domain structure; a cysteine-rich (CR) domain, a fibronectin-type II (FNII) domain and tandemly arranged C-type lectin-like domains (CTLD, eight in the case of MR). Two distinct lectin activities have been described for MR. The CR domain recognises sulphated carbohydrates while the CTLD mediate binding to mannose, fucose or N-acetylglucosamine. FNII domains are known to be important for collagen binding and this has been studied in the context of two members of the MR family, Endo180 and the phospholipase A2 receptor. Here, we have investigated whether the broad and effective lectin activity mediated by the CR domain and CTLD of MR is favoured to the detriment of FNII-mediated interaction(s). We show that MR is able to bind and internalise collagen in a carbohydrate-independent manner and that MR deficient macrophages have a marked defect in collagen IV and gelatin internalisation. These data have major implications at the molecular level as there are now three distinct ligand-binding sites described for MR. Furthermore our findings extend the range of endogenous ligands recognised by MR, a molecule firmly placed at the interface between homeostasis and immunity.
Resolution of inflammation involves the coordinate effort of cellular and soluble components of the immune system in cooperation with fibroblasts and endothelial cells, leading to the clearance of the initial insult, the elimination of cellular debris and tissue remodelling. Macrophages (MΦ) are important players in the regulation of this process. During the acute phase of the inflammatory response MΦ favour elimination of foreign material through microbial killing and phagocytosis. During the resolution stage of inflammation MΦ mediate clearance through phagocytic and enzymatic activity and influence extracellular matrix production and angiogenesis, through the production of growth factors. Furthermore, MΦ affect the deposition and organisation of collagen through the secretion of collagenase and other neutral proteinases. Under some conditions MΦ facilitate fibrosis and scarring through the release of fibroblast growth factors and promote collagen deposition through the up-regulation of arginase activity 1–3. These properties are particularly associated with alternatively activated MΦ generated in response to the Th2 T cell-derived cytokines IL-4 and IL-13 1. Another Th2 T cell-derived cytokine, IL-10, can also be produced following MΦ stimulation 4, 5. This cytokine induces a deactivated anti-inflammatory state in the Mϕ and is a key player in the control of inflammation in vivo6
One of the hallmarks of alternatively activated and deactivated MΦ is increased expression of the mannose receptor (MR, CD206) 1, 7. The MR was originally described as an endocytic receptor for lysosomal enzymes 8 and studies in MR-deficient mice support the idea that MR is primarily a homeostatic clearance system with an additional role as a pathogen receptor 9–12.
MR is a multifunctional receptor with two distinct lectin activities mediated by its extracellular region (Fig. 1A). The cysteine-rich (CR) domain recognises sugars terminated in SO4-3-galactose (Gal) or SO4–3/4-N-acetylgalactosamine (GalNAc) 13, 14, while mannose (Man), together with fucose (Fuc) and N-acetylglucosamine (GlcNAc), recognition is mediated by C-type lectin-like domain (CTLD) 4 with co-operation from CTLD5 15. A single fibronectin type II (FNII) domain is located between the CR domain and the CTLD 16–18. The FNII domain structure is conserved among all members of the MR family, which includes Endo180 (CD280, a receptor that associates with urokinase-plasminogen activator receptor complex 19), the DC receptor DEC205 (CD205) and the membrane-associated phospholipase A2 receptor (PLA2R) 20. The only member of the family with lectin activity in addition to MR is Endo180 that has recently been shown to bind simple mannose type ligands via its functional CTLD2 21.
We wanted to investigate if collagen should be included in the list of endogenous molecules recognised by MR. Collagen binding by the FNII domains of fibronectin, matrix metalloproteinases 2 and 9 and the members of the MR family Endo180 and phospholipase A2 receptor has been demonstrated previously 20, 22, 23, but we reasoned that this property could have been lost in MR in order to favour the unique lectin activity of the CR domain not seen in other members of the MR family. In this report, we demonstrate that MR can indeed recognise and internalise collagen, that the FNII domain is required for this interaction and that MR binding is not restricted to extracted collagens. Furthermore, we show that the MR is solely responsible for collagen IV and gelatin internalisation by murine MΦ in vitro. Comparison of the collagen-binding properties of Endo180 and MR revealed that both molecules display overlapping specificities consistent with a redundant role in vivo.
MR mediates internalisation of collagen IV and gelatin
To determine if the presence of MR in a cell could influence its capacity to internalise solubilized native collagen IV and denatured collagen (gelatin), WT and MR expressing cells (CHOwt and CHOMR, respectively 24) were incubated with OregonGreen-gelatin (gelatin-OG) and OregonGreen-collagen IV (collagen IV-OG) and processed for flow cytometry as described in the Materials and methods. Strikingly, collagen IV and gelatin behaved as the MR-specific ligands Man-BSA-FITC and anti-MR mAb MR5D3-Alexa 488 in that they were endocytosed by CHOMR but not by CHOwt (Fig. 1B). Similar results were obtained with NIH-3T3wt and NIH-3T3MR cells (data not shown). The method used to harvest the cells for analysis (trypsin-EDTA) ensured that no material associated with MR at the plasma membrane was being analysed as MR is a trypsin sensitive molecule. Confocal microscopy analysis confirmed the selectivity of the uptake and demonstrated that collagen was internalized into intracellular compartments in the CHOMR cells (data not shown).
Structural requirements for MR-collagen interaction
To define the domain(s) involved in collagen uptake, we prepared the Fc chimeric proteins CR-Fc 25, CR-FNII-CTLD1-Fc, CR-FNII-CTLD1-3-Fc and CTLD4-7-Fc 26 (Fig. 2A). All four constructs were readily secreted by transfected cells and appeared as a single entity by SDS-PAGE (Fig. 2B). Furthermore, gel filtration demonstrated that the vast majority of each protein in the preparations was, as expected, dimeric (Fig. 2C). CR-FNII-CTLD1-Fc and CR-FNII-CTLD1-3-Fc, together with CR-Fc and CTLD4-7-Fc, were used as inhibitors in endocytosis assays. As expected, Man-BSA uptake could be specifically blocked with CRD4-7-Fc and mannan but not by CR-Fc, CR-FNII-CTLD1-Fc (data not shown), or CR-FNII-CTLD1-3-Fc (Fig. 3A). No inhibition of collagen internalisation could be observed in the presence of CR-FNII-CTLD1-Fc (data not shown) or CR-FNII-CTLD1-3-Fc, indicating that multimerisation could play an important role in MR FNII-mediated recognition. To assess the requirement for protein cross-linking in the case of FNII domain-mediated recognition we analysed the binding of the chimeric proteins CR-Fc, CR-FNII-CTLD1-Fc, CR-FNII-CTLD1-3-Fc or CTLD4-7-Fc to collagen IV-coated plates using pre-complexed or non-complexed proteins. Pre-complexed proteins were prepared by preincubating the Fc proteins with anti-human Fc Ab. As shown in Fig. 3B, no binding of CR-Fc or CTLD4-7-Fc to collagen IV-coated plates was observed under any conditions and binding of CR-FNII-CTLD1-Fc and CR-FNII-CTLD1-3-Fc was only detected when proteins were pre-incubated with an anti-human Fc reagent. Data shown in Fig. 3B also demonstrate that collagen binding is calcium independent. The CR-FNII-Fc protein 25 was not included in these studies because questions regarding the conformation of FNII domain in this construct were raised by its poor production and the aggregated material observed after SDS-PAGE analysis. Our results indicate that the presence of CTLD1 could facilitate the correct folding of FNII. No differences between the binding properties of CR-FNII-CTLD1-Fc and CR-FNII-CTLD1-3-Fc were observed indicating that CTLD2 and CTLD3 do not contribute substantially to FNII domain-mediated binding.
To ascertain if MR could preferentially recognise a particular type of collagen, we investigated the binding of cross-linked Fc chimeric proteins to native collagens I, II, III and V and denatured collagen (gelatin). For comparison and as a positive control, an Fc chimeric protein containing the CR, FNII and CTLD1–4 domains of MR family member, Endo180, were included in these studies 23. Data shown in Fig. 4 demonstrate that MR displays a broad specificity as it is able to interact with all types of collagen tested, including denatured collagen (gelatin) and that there are no major differences concerning collagen recognition between MR and Endo180.
MR FNII ligands are broadly distributed in situ
As all the binding studies described earlier were performed using purified material extracted from biological samples, we decided to investigate the binding properties of CR-FNII-CTLD1-3-Fc in situ to determine if organised native collagen embedded in extracellular matrix would be a suitable ligand for MR. For this purpose we developed a binding assay based on that previously employed to detect CR domain ligands in lymphoid organs using CR-Fc 25, 27. Data shown in Fig. 5 demonstrate that CR-FNII-CTLD1-3-Fc binds to ligands in mouse tissues in situ and that these ligands are widely distributed. In agreement with previous binding assays, FNII-mediated binding could only be observed upon cross-linking. In lymphoid organs the presence of the FNII domain in the Fc constructs enabled detection of ligands in addition to those detected by CR-Fc alone; these were distributed as predicted for components of extracellular matrix. Furthermore, CR-FNII-CTLD1-3-Fc FNII also recognised structures in non-lymphoid organs such as pancreas, adrenal glands (Fig. 5), heart, testis and skin dermis (data not shown) that correlate with recognition of basement membranes. Interestingly, binding of the CR-Fc construct to lymphoid tissues 25, 27 was substantially enhanced when cross-linked. In particular, the white pulp of spleen, in addition to the marginal zone, was now readily recognised by this probe (Fig. 5). Binding of the CR-FNII-CTLD1-Fc construct was identical to that observed with the CR-FNII-CTLD1-3-Fc construct (data not shown).
MR is solely responsible for collagen and gelatin internalisation by bone marrow-derived MΦ
To investigate the contribution of endogenous MR to collagen internalisation, BM-derived MΦ from WT and MR–/– animals were assessed for their ability to internalise collagen IV and gelatin. As shown in Fig. 6, MΦ from MR–/– mice were not only unable to internalise the MR-specific ligands tested (Man-BSA and anti-MR mAb MR5D3) but also could not internalise collagen IV or gelatin. The same results were obtained when internalisation assays were performed for 20 min (data not shown) and 1 h, indicating that the differences observed were not due to altered kinetics of uptake or degradation in the MR–/– cells. To demonstrate that MR–/– MΦ did not have a general defect in receptor-mediated endocytosis we used the endocytic tracer specific for scavenger receptors (acetylated low-density lipoprotein) and determined that its internalisation was normal in MR-deficient BM-MΦ (Fig. 6, insets).
To investigate the relationship between MR and collagen in vivo we determined how MR expression related to collagen IV localisation in situ. A highly suggestive pattern was obtained in mouse skin (Fig. 7); in this anatomical location MR-positive cells (mostly MΦ, E. McKenzie et al, manuscript in preparation) were present in dermis and were arranged as rows lining collagen IV-containing basement membranes. These results are indicative of a contribution of MR to basal membrane biology.
MR contains three functional binding sites
MR is member of a family of proteins comprising three additional endocytic receptors, Endo180, PLA2R and DEC-205 20. A distinct ligand-binding profile has been described for each of these molecules illustrating how similar domain arrangements can be exploited to fulfil diverse physiological roles. While MR and Endo180 share the ability to bind similar monosaccharides in vitro through selected CTLD (CTLD4 in the case of MR and CTLD2 in the case of Endo180), no natural glycosylated ligands have been characterised for Endo180. MR is the only family member in which the CR domain contains the residues required for the formation of the neutral binding pocket mediating recognition of sulphated sugars. Therefore, recognition of glycoproteins bearing these acidic sugars is likely to be mediated exclusively by MR 20, 28.
The ability of MR to bind collagens adds a new dimension to the study of MR biology. The requirement for extensive cross-linking to visualise this interaction is indicative of a low affinity interaction requiring domain multimerisation to increase avidity. This property has been observed, to a lesser extent, in the case of CR recognition of sulphated sugars; we have observed that (i) targeting of CR containing Fc proteins to cells bearing CR ligands in vivo is enhanced by cross-linking 29, (ii) presence of CR domain containing Fc proteins does not block uptake of SO4-3-Gal by MR transductants (data not shown) and (iii) monomeric soluble MR displays very poor binding to sulphated sugars 30. Our observations predict that monomeric soluble MR would bind poorly to surfaces rich in sulphated sugars and would be unable to interact with collagen-rich surfaces. Multimerisation, either through interactions with polymeric mannosylated sugars 31 or through lack of terminal sialic acid in its N-linked carbohydrates 30 would favour both interactions (Fig. 8). On the other hand cell-associated MR, likely to be clustered at the plasma membrane, would be endowed with the ability to interact with three distinct ligands. This raises questions regarding the effect of the extracellular matrix on the sugar-recognition capacities of MR as steric hindrance could hamper recognition of sulphated sugars by collagen-engaged MR.
MR and self recognition
Our results increase the range of endogenous ligands identified for MR. Some of MR endogenous ligands are targeted by the immune system in the context of autoimmune diseases such as thyroglobulin in the case of thyroditis 26, 32 and myeloperoxidase in the case of glomerulonephritis 33. We previously hypothesised that inappropriate presentation of these ligands to the acquired immune system could be facilitated through their interaction with MR expressed by subpopulations of DC 12. It is intriguing that collagens II and IV can also be targeted by the acquired system since collagen II, a major component of cartilage, is an important autoantigen in rheumatoid arthritis patients 34, and collagen IV contains the antigen recognised by autoantibodies from Goodpasture syndrome patients 35. The involvement of MR to Ag presentation is still unclear since no major contribution of MR to immunity against pathogens has been evident in experimental models of infection but rather a non-redundant role in homeostatic clearance of lysosomal enzymes has been demonstrated 9–11. We have identified a novel MR-positive DC population in selected lymphoid organs that can bind MR ligands in vivo (E. McKenzie et al. manuscript in preparation). These cells are controlled by innate stimulation, leading us to propose that MR-mediated antigen presentation is favoured upon microbial infection. We hypothesise that in this environment enhanced presentation of endogenous ligands will also take place that could trigger activation of self-reactive T cells that escape central and peripheral mechanisms of tolerance.
MΦ and collagen
Even though MΦ produce a wide range of molecules relevant to tissue remodelling, the existence of compensatory mechanisms is illustrated by the normal foetal development and wound healing observed in mice lacking the transcription factor PU.1 which are deficient in MΦ 36. With this in mind it is not surprising that MR deficiency does not seem to have a major effect on organogenesis under steady-state conditions. Indeed, our preliminary results indicate that the distribution of ligands for the CR and FNII domains and of several MΦ populations is not altered in tissues from MR–/– animals (data not shown). Endo180 expressed by fibroblasts, among others, could compensate for lack of MR. Nevertheless, we have observed that MR plays a non-redundant role in collagen internalisation by MΦ. Based on this observation and together with the close relationship between MR+ cells and collagen IV observed in mouse skin, we predict that under MΦ-dominated pathologies such as granulomata, lack of MR would have a substantial effect.
In summary, we have demonstrated that murine MR recognises collagens, that this function is mediated by FNII domain and requires protein cross-linking. Our results are in agreement with a recent report by Napper et al.37 demonstrating collagen recognition by the MR FNII domain. We have also shown that MR is responsible for collagen internalisation by MΦ and that there is a close association between collagen fibres and MR+ cells in skin. Our studies extend the list of endogenous ligands recognised by MR and raise further questions regarding MR involvement in induction of autoimmunity and tissue homeostasis.
Materials and methods
WT BALB/c and WT and MR–/– C57BL/6 mice (kindly provided by Dr. M. Nussenzweig, Rockefeller University) were maintained under specific pathogen-free conditions and used at 8–12 weeks.
Fc chimeric proteins
cDNA encoding CR-FNII-CTLD-1-Fc and CR-FNII-CTLD1-3-Fc were amplified using HF-2 polymerase (Clontech, CA, USA) from a plasmid containing the full length mannose receptor cDNA 24 using reverse primer 5′-GGA GAT CTA CTT ACC TGT CAC ATG CTT GCT GAG GGA ATG ATA AAT G (CR-FNII-CTLD-1-Fc) and 5′-GGA GAT CTA CTT ACC TGT TGG ACA TTT GGG TTC AGG AGT TGT TGT GG (CR-FNII-CTLD1-3-Fc) and a common forward primer 5′-GGG ATA TCG ACC TTG GAC TGA GCA AAG GGG CAA CCT GG. EcoRV/BglII digested PCR products were introduced in-frame into the EcoRV and BamHI restriction sites of the expression vector pIg generating vectors encoding MR-Fc chimeric proteins under the control of the CMV promoter. CR-Fc, CRD4-7-Fc and E-CR-FNII-CTLD1-4-Fc have been previously described 23, 25, 26.
Plasmids encoding the Fc-chimeric proteins were introduced into the 293T cell line by transient transfection with Gene-Juice reagent (Novagen) following manufacturer's guidelines. After transfection cell media were replaced with media containing 4% IgG-depleted foetal bovine serum and media were conditioned for 7 days. After this time, media were harvested and Fc-chimeric proteins were recovered by affinity purification on Protein A-Sepharose (Amersham Biosciences) and elution with 0.1 M glycine, pH2.9. After neutralisation with 0.1 volumes of 1 M Tris, pH 9.5, proteins were dialysed against PBS. Of each reduced chimeric Fc protein, 4-10 μg was concentrated using the Strataclean method and run on 4–15% Bris-Tris gel (Invitrogen), alongside 7 µL of MultiMark protein ladder (Invitrogen).
Gel filtration analysis
Gel filtration was performed on a Superose 12 column (Pharmacia) using 10 mM Tris, 140 mM NaCl, 10 mM CaCl2, pH7.4 as running buffer; 20 μL of serum was run as a standard.
Solid phase collagen-binding assay
The 96-well high-binding plates (Costar) were coated with appropriate substratum diluted at 0.1 mg/mL, (100 µL/well) and incubated overnight at 37°C. Plates were washed three times with 150 µL PBS/0.3% BSA and blocked with 0.3% BSA in PBS for 1 h at room temperature (RT). Fc chimeric proteins (diluted at 2–10 µg/mL) were preincubated or not preincubated with anti-human Fc-alkaline-phosphatase (AP) conjugate (1:100–1:200, Sigma) for 1 h at RT in PBS/0.3% BSA. These were added to the plate and incubated for 1 h at RT. After three washes in PBS/0.3% BSA, plates were incubated with anti-human Fc-AP (1:500–1:1000) for 1 h at RT in PBS/0.3% BSA. After washing three times with PBS/0.3% BSA, wells were washed twice with AP substrate buffer (100 mM Tris-HCl, 100 mM NaCl, 1 mM MgCl2, pH 9.5) and 50 µL of substrate solution were added to each well (1 mg/mL p-nitrophenyl phosphate in substrate buffer) and incubated for 10–30 min at RT. Reactions were stopped by adding 50 µL of stop solution (100 mM EDTA). Absorbance was read at 405 nm. In some instances 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 mM CaCl2 was used as buffer in the assay.
CHOwt, CHOMR, and BM-MΦ from WT C57BL/6 and MR–/– mice were washed in serum-free medium and maintained in this medium for 1 h at 37oC. Different tracers (diluted at 5 μg/mL in serum-free media) were then added. The different tracers used were: Man-BSA-FITC (EY laboratories), Collagen IV-Oregon Green (OG) (Molecular Probes), gelatin-OG (Molecular Probes), anti-MR mAb MR5D3 and isotype control IgG2a both prepared in house and labelled with Alexa 488 (Molecular Probes) following the manufacturer's instructions. In some experiments CR-Fc, CR-FNII-CTLD-1-Fc, CR-FNII-CTLD1-3-Fc or CTLD 4-7-Fc diluted at 100 μg/mL or mannan (2 mg/mL) were present during the incubation. After incubation cells were washed in PBS, harvested using trypsin/EDTA (CHO cells) or lidocaine/EDTA (BM-MΦ) and fixed in 2% formaldehyde solution in PBS. Uptake was analysed using a BD FACScalibur and CellQuest software.
In situ detection of FNII ligands
Mouse tissues were collected, embedded in OCT compound and frozen in dry ice-cooled isopentane. The 5-μm sections were cut, placed on charged slides and kept frozen until use. To detect ligands for CR and FNII domains in situ, sections were fixed in 2% paraformaldehyde (10 min, 4oC), permeabilised using 1% Triton X-100 in PBS (1 h, RT) and blocked with 5% normal goat serum (NGS) in PBS (30 min, RT). Preparations were incubated with CR-Fc, CR-FNII-CTLD-1-Fc, CR-FNII-CTLD1-3-Fc pre-complexed for 1 h using goat-anti-human AP conjugate (Chemicon), diluted 1:100 in PBS (overnight, 4oC) in the presence of 5% NGS. Protein binding was detected using goat-anti-human AP conjugate diluted 1:100 in 5% NGS. The slides were developed using the BCIP/NBT alkaline phosphatase substrate kit IV (Vector Laboratories).
Detection of MR and collagen IV in tissue sections
Mouse ears were collected, embedded in OCT compound and frozen in dry ice-cooled isopentane. The 5-μm sections were cut and kept frozen until use. Sections were thawed at room temperature, fixed in 2% paraformaldehyde (10 min, 4oC), permeabilised using 0.1% Triton X100 in PBS (1 h), blocked with 5% NGS in PBS (30 min, RT) followed by blocking of endogenous biotin using a avidin/biotin kit (Vector Laboratories). Sections were incubated with rabbit anti-mouse Collagen IV (Immunologicals Direct) in 5% NGS in PBS (1 h) followed by Alexa488-conjugated goat anti-rabbit IgG (Molecular Probes) in PBS, to detect collagen, and then with biotinylated MR5D3 (10 μg/mL) in 5% NGS in PBS (1 h) and streptavidin Cy5 (Jackson Immunoresearch Laboratories) in PBS (30 min). Sections were counterstained with DAPI (Sigma) at 400 ng/mL prepared in H2O for 5 min, mounted using DAKO fluorescent mounting media (Dakocytomation) and analysed using a Zeiss Axioplan 2e microscope fitted with a 16-bit 1024 × 1024 pixels CCD camera. Images were processed using Metamorph software.
Funding sources: Medical Research Council, The Edward P. Abraham Research Fund and Breakthrough Breast Cancer.