The mannose receptor on murine liver sinusoidal endothelial cells is the main denatured collagen clearance receptor


  • Potential conflict of interest: nothing to report.


The purpose of this study was to identify the receptor responsible for endocytosis of denatured collagen from blood. The major site of clearance of this material (at least 0.5 g/day in humans) is a receptor on liver sinusoidal endothelial cells (LSECs). We have now identified an 180-kDa endocytic receptor on LSECs, peptide mass fingerprinting of which revealed it to be the mannose receptor. Challenge of mannose-receptor knockout mice and their cultured LSECs revealed significantly reduced blood clearance and a complete absence of LSEC endocytosis of denatured collagen. Organ analysis of wild-type versus knockout mice after injection of denatured collagen revealed significantly reduced liver uptake in the knockout mice. Clearance/endocytosis of ligands for other receptors in these animals was as that for wild-type mice, and denatured collagen uptake in wild-type mice was not affected by other ligands of the mannose receptor, namely mannose and mannan. Furthermore, unlike that of mannose and mannan, endocytosis of denatured collagen by the mannose receptor is calcium independent. This suggests that the binding site for denatured collagen is distinct from that for mannose/mannan. Mannose receptors on LSECs appear to have less affinity for circulating triple helical type I collagen. Conclusion: The mannose receptor is the main candidate for being the endocytic denatured collagen receptor on LSECs. (HEPATOLOGY 2007.)

Collagen is the most abundant protein in vertebrates, making up about 25% of all protein in mammals. Collagens have numerous structural and cell-biological functions, and mutations/deletions of collagen genes can result in serious diseases, e.g., osteogenesis imperfecta or embryonic lethality. Inappropriate collagen deposition is a serious complication of liver fibrosis1 and other diseases.

Much is known about the degradation of collagens by matrix metalloproteinases.2 However, in normal tissue no extracellular matrix proteases are capable of breaking down collagen into its constituent amino acids. A latent vertebrate collagenase in tissue catalyses a clip in the collagen triple helix, generating 2 fragments.3 The α-chain fragments of the free collagen (which result from the physiological denaturation of the collagenolytic split products at 37°C) are too large to be excreted via the kidneys. Knowledge about how these collagen α-chains are cleared is insufficient. It has been suggested that local intra- or extracellular degradation may be responsible for some collagen removal. Some uptake and intralysosomal degradation in local cells may be mediated by Endo180, an endocytic receptor capable of binding native collagen type I (NatColl) and denatured collagen type I (DenColl).4 However, most collagen α-chains in rat interstitia are transported out to circulation and removed rapidly by receptor-mediated uptake into liver sinusoidal endothelial cells (LSECs),5 which do not express Endo180.6 Furthermore, collagen fragments liberated during bone turnover are drained directly into the bloodstream and not via the lymphatic system. Therefore, it is of paramount importance that clearance of this material and its subsequent lysosomal degradation7, 8 by LSECs occur rapidly; otherwise, excessive circulating denatured collagen could interfere with biophysical processes (e.g., plasma viscosity) or biochemical processes (e.g., inhibition of the matrix-degrading enzymes necessary to reverse fibrosis).

We were able to identify this clearance receptor on LSECs using gelatin-sepharose affinity chromatography of rat and pig LSECs to purify a 180-kDa protein. Peptide mass fingerprinting of this protein identified it as the mannose receptor (MR). The MR is a type I transmembrane protein with an extracellular portion consisting of 8 membrane-proximal C-type carbohydrate-recognition domains followed by a domain containing a fibronectin type II (FNII) repeat and a membrane-distal cysteine-rich domain.9 MRs are expressed on other cell types, including macrophages,10 but only LSECs, each of which has 20,000-25,000 MRs on its cell surface,11 have the ability to endocytose large amounts of DenColl.12, 13 This capacity is probably a result of the MR having a short surface half-life (10 seconds) and being very rapidly recycled back to the plasma membrane after endocytosis.11 We have already shown that the LSEC MR is the main clearance receptor for C-terminal procollagen propeptides,14 and our finding that the LSEC MR is also the main clearance receptor for denatured collagen reveals yet another clearance function for this protein, making it an interesting target for the study of hepatic fibrosis.


DenColl, denatured collagen type I; DTAF, 5-isomer of fluorescein dichlorotriazine (5-[4,6-dichlorotriazine-2-yl] amino fluorescein hydrochloride); FSA, formaldehyde-treated bovine serum albumin; LSECs, liver sinusoidal endothelial cells; MR, mannose receptor; NatColl, native collagen type I; PBS, phosphate-buffered saline; RT, room temperature.

Materials and Methods

Chemicals and Reagents.

Sephadex G-25 (PD-10 columns), gelatin-sepharose 4B, and Percoll came from GE-Health (Uppsala, Sweden). Bovine serum albumin, Triton X-100, mannose, EGTA, mannan, pepstatin-A, N-ethylmaleimide, Hepes, phosphotungstic acid, and DMSO came from Sigma Chemical Co. (St. Louis, MO). Complete Protease Inhibitor Cocktail Tablets came from Roche (Oslo, Norway). RPMI-1640 culture medium, supplemented with 20 mM sodium bicarbonate, 0.006% (wt/vol) penicillin, and 0.01% (wt/vol) streptomycin, came from Gibco BRL (Roskilde, Denmark); human serum albumin from Octapharma (Ziegelbrucke, Switzerland), and collagenase P from Worthington Biochemical Corp. (Lakewood, NJ). The carrier-free Na[125I] was from PerkinElmer Norge AS (Oslo, Norway), and the 1,3,4,6-tetrachloro-3α,6α-diphenylglycoluril (Iodogen) from Pierce Biotechnology Inc. (Rockford, IL). The 5-(4,6-dichlorotriazine-2-yl) amino fluorescein hydrochloride (DTAF) came from AnaSpec Inc. (San Jose, CA) and the bovine collagen type I (Vitrogen 100) from Cohesion Technologies (Palo Alto, CA). Formaldehyde-treated bovine serum albumin (FSA) was prepared as described.15 The ProtoBlue Safe colloidal Coomassie blue G-250 protein stain came from National Diagnostics (Hessle Hull, UK).


Castrated male piglets (Sus scrofa domesticus, Norwegian Landrace, 7-8 kg) came from a local farm and male Sprague-Dawley rats from Scanbur BK AB (Sollentuna, Sweden). Wild-type C57BL/6 mice came from Harlan (France), and the MR/ knockout C57BL/6 mice were kindly provided by Professor M. Nussenzweig (Rockefeller University, New York, NY). The MR knockout animals were backcrossed for 2-3 generations with wild-type C57BL/6 mice at the Animal Department of the University of Tromsø (Tromsø, Norway) before breeding the homozygous MR knockout C57BL/6 mice for the experiments. MR knockout status was tested by PCR.16 The animals were housed in the Animal Department under controlled conditions in rooms designed for each species. The rodents were fed a standard chow (Scanbur BK, Nittedal, Norway) ad libitum. All experimental protocols were approved by the Norwegian Animal Research Authority in accordance with the Norwegian Animal Experimental and Scientific Purposes Act of 1986.

Statistical and Pharmacokinetic Analyses.

Statistical calculations (independent-sample t tests) were performed with the SPSS statistical package for Windows version 13.0 (SPSS Inc., Chicago, IL). Clearance kinetics were analyzed as described previously.7

Isolation of LSECs from Pig, Rat, and Mouse Livers.

LSECs from pig17 and rat18 livers and mouse LSEC cultures19 were prepared as previously described.

Ligand Labeling Procedures.

Heat-denatured collagen type I, DenColl (3.2 mg/ml: 60°C for 60 minutes), was labeled with DTAF as follows. To 1 ml of DenColl was added 100 μl of 1M Hepes (pH 8) and 10 mg/ml DTAF in 45 μl of DMSO. This was incubated overnight at 4°C and dialyzed extensively against phosphate-buffered saline (PBS) to yield DTAF-DenColl with a maximum substitution of 1 DTAF per 40 amino acid residues of DenColl (or a maximum of 23 DTAFs per collagen alpha chain). This reagent has proven to be very stable and amenable to 125I labeling without becoming a ligand for scavenger receptors (data not shown). The DTAF-DenColl was stored in aliquots at −20°C, and heated to 60°C for 60 minutes before use. DTAF-DenColl, NatColl, mannan, and FSA were dissolved in PBS and labeled with 125I using Iodogen.20 Free 125I was removed with PD-10 columns equilibrated with PBS. The resulting specific radioactivities were: [125I]DTAF-DenColl, approximately 4 × 106 cpm/μg; [125I]NatColl, 1.2 × 106 cpm/μg; [125I]mannan, 2.5 × 106 cpm/μg; and [125I]FSA, 2.5 × 106 cpm/μg.

Isolation and Purification of the Receptor for Denatured Collagen.

To label the surfaces of rat and pig LSECs, suspensions of 100 million LSECs were labeled with 125I using the lactoperoxidase method21 with modifications.22 Briefly, suspended LSECs were washed 3 times with ice-cold PBS, then resuspended in 1 ml of PBS containing 5 mM glucose, 0.5 mCi 125I, 30 μg/ml lactoperoxidase, and 1.2 μg/ml glucose oxidase. Cells were incubated for 15 minutes at room temperature (RT) with gentle mixing, after which they were washed 3 more times with ice-cold PBS. They were then solubilized with PBS containing 1% Triton X-100 with the protease inhibitors NEM (2 mM), pepstatin (1 μg/ml), and Roche protease inhibitor (PI) cocktail end over end overnight at 4°C.

For small-scale purification of the DenColl receptor, the [125I]LSEC extracts were centrifuged at 10,000g for 10 minutes, after which the supernatant was diluted to 10 ml with PBS, divided into 2 aliquots, and applied to gelatin or control sepharose columns (6-ml bed volume, 1.5 cm in diameter) equilibrated with PBS/0.1% Triton X-100/PI at 5 cm/hour at RT. Both columns were then washed with 45 ml PBS/0.1% Triton X-100/PI at 5 cm/hour at RT. Bound proteins were eluted with 0.5M NaCl, 0.1% Triton X-100, PI, 0.1 M NaCH3COOH, pH 3.5 (18 ml), at 10 cm/hour at RT, as 1-ml fractions into tubes containing 100 μl of 2M Tris (pH 8.8). The most radioactive fractions were analyzed by SDS-PAGE (6% running and 4% stacking gels) and phosphorimaging (Fuji BAS5000).

For large-scale purification of the DenColl receptor, a suspension of 600 million pig LSECs was prepared17 and solubilized in 10 ml PBS/1% Triton X-100/PI end over end at 4°C overnight. The cell extract was then centrifuged at 2,500 rpm for 10 minutes, after which the supernatant was filtered through a 0.45-μm filter. The filtrate was diluted with PBS to 100 ml, and radioactive tracer containing the purified [125I]DenColl receptor was added. This was applied to a gelatin-sepharose column (30-ml bed volume, 1.8 cm in diameter) equilibrated with PBS/0.1% Triton X-100/PI at 7.5 cm/hour at RT. The column was washed with 120 ml PBS/8 mM CHAPS/PI at 7.5 cm/hour at RT. Bound proteins were eluted with 90 ml 0.5 M NaCl, 8 mM CHAPS, PI, 0.5 mM EDTA, 0.1 M NaCH3COOH (pH 3.5) and collected as 2-ml fractions in tubes containing 200 μl of 2M Tris (pH 8.8). The most radioactive fractions were concentrated (100,000 MWCO) to 250 μl and subjected to preparative SDS-PAGE, colloidal Coomassie blue G-250 protein staining, and phosphorimaging (Fuji BAS5000). After excision of the Coomassie-stained band, the dried gel was subjected to further phosphorimaging to confirm excision of the correct (radioactive) band.

Protein Digestion and Peptide Purification.

The excised proteins bands were reduced and alkylated,23 and prior to MS analysis the eluted trypsin-generated peptides were concentrated and desalted on OMIX C18 pipette tips (Varian Inc., Palo Alto, CA) according to the manufacturer's instructions.

Mass Spectrometry for Protein Identification.

Peptide mass spectra were obtained on a MALDI micro MX (Micromass, UK) spectrometer. The purified tryptic digest (1 μl) was mixed with α-cyano-4-hydroxy-trans-cinnamic acid (10 mg/ml in 1:1 [vol/vol] 0.1% trifluoroacetic acid:acetonitrile) directly on the target and dried at RT. Mass spectra were collected as a summation of up to 100 laser shots on an instrument externally calibrated using sodium iodide and PEG 200, PEG 600, PEG 1000, and PEG 2000 mixtures. Internal calibration using 200 Glu-fibrinopeptide B (Glu-Fib, MH+ = 1,570.6773; Sigma) was applied after acquisition. Peptide mass fingerprint spectra from tryptic fragments that had been used for identification were searched against the NCBInr databases using an in-house license of the MASCOT search engine (Matrix Science, London, UK). For peptide mass fingerprinting it was assumed that the peptides were monoisotopic, oxidized at methionine residues, and carbamidomethylated at cysteine residues. Up to 1 missed trypsin cleavage was allowed. Mass tolerance of 100 ppm was the window of error allowed for matching the peptide mass values.

Blood Clearance and In Vivo Distribution of Radiolabeled Ligands.

The tail veins of anesthetized wild-type and MR knockout mice were injected with [125I]DTAF-DenColl (approximately 0.04 mg/kg in PBS) or [125I]NatColl (approximately 0.20 mg/kg in PBS). Immediately after the injections, 5 μl blood samples were collected over short intervals from the tail tip and mixed with 0.3 ml 13 mM citric acid, after which 0.3 ml 2% bovine serum albumin in water and 0.6 mL of 20% (wt/vol) trichloroacetic acid/0.5% (wt/vol) phosphotungstic acid were added to the samples to precipitate nondegraded protein. Acid-soluble radioactivity represented degraded ligand. Thirty minutes after injection of [125I]DTAF-DenColl or 60 minutes after injection of [125I]NatColl, the animals were killed in a 100% CO2 atmosphere. The abdomens and thoraces were opened, and the animals were perfused intracardially with cold PBS to remove free tracer from the vasculature before the organs were excised and analyzed for radioactivity. The amount of tracer recovered was considered the sum of the radioactivity of individual organs, the carcass, and blood at sacrifice. Blood volume was calculated as described.24

In Vitro Endocytosis of Radiolabeled Ligands.

Primary cultures of mouse LSECs (3–5 × 105 cells per culture) were established in 1-cm2 wells (48-well culture plates; Falcon, Becton Dickinson, Plymouth, UK) precoated with 0.1% native collagen (Vitrogen 100) and maintained in serum-free RPMI-1640 medium at 37°C in a humidified atmosphere of 5% CO2. Endocytosis experiments were conducted in fresh cultures within 4 hours of plating. After washing, each culture was supplied with 100 μl of fresh RPMI-1640 medium containing 1% human serum albumin, inhibitors, and trace amounts (approximately 10 ng) of radiolabeled protein. The incubations, which were carried out for 2 hours to measure endocytosis, were terminated by transferring the media along with 1 wash (500 μl) with PBS to tubes containing 800 μl of 20% trichloroacetic acid/0.5% phosphotungstic acid. Ligand degradation was determined by measuring the amount of labeled acid soluble radioactivity after centrifugation. Because mammalian cells lack lysosomal enzymes that can degrade mannan, the polymannose conjugate will be trapped intralysosomally, and 125I attached to the protein core will not leave the cells. Consequently, endocytosis of [125I]mannan did not lead to release of 125I to the medium, and TCA precipitation was omitted. Cell-associated ligand was quantified by measuring the radioactivity in the washed cells solubilized in 1% (wt/vol) SDS, and total endocytosis was determined by adding cell-associated and acid-soluble radioactivity.

Results and Discussion

Purification of the Denatured Collagen Receptor.

Using gelatin (DenColl)-sepharose affinity chromatography of [125I]surface-labeled rat and pig LSECs, we identified a 180-kDa cell surface protein that binds DenColl (Fig. 1). In preparing pig LSECs, a weak band at approximately 400-kDa was also consistently noted. A 115-kDa band was eluted from control (i.e., nonsubstituted) sepharose when [125I]surface-labeled rat LSECs were used. This band was absent in pig LSEC preparations and was not eluted from the gelatin-sepharose columns, suggesting these columns are very heavily substituted with gelatin.

Figure 1.

Purification of DenColl receptors from LSECs. The [125I]surface-labeled rat and pig LSECs were solubilized in Triton X-100 and applied to gelatin (denatured collagen) sepharose (G) or control sepharose (C). Specifically bound material was eluted at pH 3.5 and subjected to SDS-PAGE autoradiography.

MALDI mass data of the Coomassie-stained bands of about 180 and 400 kDa from pig LSECs yielded 19 and 21 peptides, respectively, which matched pig MR type C, with respective Mascot protein scores of 100 and 124 (a score > 78 is statistically significant). Because the SDS-PAGE was nonreducing, it is possible that the 400-kDa species is a dimer of MRs. The MR has recently been reported to confer collagen-binding properties to fibroblasts transfected with same,25 but its role in the endocytosis of collagen fragments in vitro was unknown until recently.10

MR Knockout Mice Have a Drastically Reduced Ability to Clear DenColl In Vivo.

Challenge of MR knockout mice and their cultured LSECs revealed significantly reduced clearance from blood and the complete absence of endocytosis of DenColl by LSECs. Linear decay plots indicated that after intravenous injection [125I]DTAF-DenColl was very efficiently cleared from the blood of wild-type mice (Fig. 2A) and more slowly cleared from the circulation of MR knockout mice (Fig. 2B). Semilogarithmic decay plots revealed a biphasic pattern of elimination (not shown) in both mouse groups. In wild-type mice, 90.25% ± 2.23% of the radioactivity was eliminated during an initial rapid α-phase, with a blood t1/2 of 0.51 ± 0.11 minutes, whereas the remainder was eliminated with a t1/2 of 36.88 ±17.43 minutes during a terminal slow β-phase. In MR knockout mice, the t1/2 of the α phase was significantly slower, 1.55 ± 0.45 minutes (P < 0.05), and only 33.28% ± 2.13% of the radioactivity was eliminated during the initial α phase, whereas 66.72% ± 2.13% was eliminated with a t1/2 of 38.44 ± 10.73 minutes during the terminal β-phase. There was minimal release of free 125I, probably because of the entrapment of DTAF-labeled radioisotope in the lysosome. The clearance pattern of directly labeled [125I]NatColl was similar and very slow in both wild-type (Fig. 2C) and MR knockout (Fig. 2D) mice. This suggests that the LSEC MR is not involved in NatColl clearance, in contrast to the role suggested in a study of MR-transfected CHO cells by Martinez-Pomares et al.,10 who reported endocytosis of native type IV collagen, an ability that was also absent in cultured macrophages from MR knockout mice. However, Martinez-Pomares et al. used OregonGreen-labeled native type IV collagen at concentrations 50 times greater than those of radiolabeled NatColl and DenColl in the present study, which would suggest a weaker affinity of macrophage MRs for native/denatured collagen. It has been previously reported that LSECs have no affinity for native collagen,12, 26 which is perhaps not surprising, given that there are typically no native collagens in the blood that would otherwise stimulate platelet adhesion/aggregation.27

Figure 2.

Blood clearance of injected DenColl and NatColl in wild-type and MR knockout mice. Data from a single mouse are indicated by rings, triangles, or squares. (A) Clearance (30 minutes) of [125I]DTAF-DenColl (0.04 mg/kg) in 3 wild-type mice (filled shapes). (B) Clearance (30 minutes) of [125I]DTAF-DenColl (0.04 mg/kg) in 3 MR knockout mice (open shapes). (C) Clearance (60 minutes) of [125I]NatColl (0.20 mg/kg) in 3 wild-type mice (filled shapes). (D) Clearance (60 minutes) of [125I]NatColl (0.20 mg/kg) in 3 MR knockout (open shapes) mice. Degradation (not shown) was negligible. Radioactivity of a blood sample radioactivity was plotted against time after injection. Radioactivity of the 1-minute sample was set as 100%.

Whole-organ analysis of wild-type (n = 3) and MR knockout (n = 3) mice 30 minutes after injection of [125I]DTAF-DenColl revealed the following distribution of radioactivity: wild type—56% liver, 4% kidneys, 24% body, 8% blood; MR knockout—28% liver, 10% kidneys, 22% body, 32% blood. This indicates a significant (P < 0.01) reduction in liver uptake, a corresponding increase in blood level (P < 0.01), and increased kidney clearance/uptake (P < 0.05) in the knockout mice (Fig. 3A). The organ distribution of injected [125I]NatColl was identical in both wild-type and knockout mice (Fig. 3B), with most remaining in the blood after 60 minutes. The clearance/endocytosis of a ligand (FSA) for other LSECs receptors (stabilin-1 and stabilin-2)22, 28 by MR knockout animals was similar to that for wild-type mice (not shown).

Figure 3.

Anatomical distribution of radiolabeled (A) DenColl and (B) NatColl in wild-type and MR knockout mice. The animals used in the serum half-life studies (Fig. 2) were analyzed for anatomical distribution of [125I]DTAF-DenColl or [125I]NatColl 30 or 60 minutes, respectively, after intravenous injection into 3 wild-type (filled bars) and 3 MR knockout (open bars) mice. The sum of the radioactivity in the listed tissues and carcass was considered 100%. Results are presented as percentage of total recovered radioactivity ± SD (n = 3); *P < 0.01, **P < 0.05.

The livers of MR knockouts retained some [125I]DTAF-DenColl because of some unknown mechanism. Fluorescence microscopy of liver sections from DTAF-DenColl-injected knockout mice (100 μg/mouse) did not reveal any cell-specific uptake (not shown).

LSECs from MR Knockout Mice Lack Ability to Endocytose DenCollIn Vitro.

LSECs from wild-type and knockout mice were challenged with [125I]DTAF-DenColl. The knockout LSECs were unable to endocytose the [125I]DTAF-DenColl (Fig. 4A), unlike the wild-type LSECs (Fig. 4B). LSECs from both the MR knockout and the wild-type mice endocytosed FSA equally well (Fig. 4A, B). In the wild-type mice, uptake of [125I]DenColl by LSECs was not affected by mannan, EGTA, or mannose (Fig. 5A), nor was the endocytosis of [125I]mannan inhibited by DenColl or NatColl (Fig. 5B). This suggests that the binding site for denatured collagen is distinct from that for mannose and mannan and is probably the FNII domain, which can bind gelatin.25 NatColl inhibited LSEC uptake of DenColl (Fig. 5A), but not by more than 60%, suggesting that the affinity for the latter is greatest, which we have seen previously.12 However, trace amounts of radiolabeled NatColl (0.1 μg/ml) were neither bound nor taken up by cultured LSECs after a 2-hour incubation at 37°C (not shown), suggesting that the LSEC MR has a much reduced affinity for this type of collagen in its native form. This lack of NatColl binding is at odds with findings of others, who demonstrated that MR-derived recombinant constructs are able to bind immobilized native collagen types I-V10 and when expressed in fibroblasts, native collagen types I, III, and IV, but not type V.25 However, that we did not observe NatColl binding in LSECs could simply reflect the need to have the entire MR for specificity and/or the specificity of the LSEC MR, which will normally not be exposed to circulating native collagen. Indeed, the MR FNII domain has greater affinity for gelatin than does the entire MR extracellular domain, suggesting that the other domains modulate this affinity.25 That the MR has a reported affinity for native collagens (albeit in vitro, on cells other than LSECs, and either to immobilized collagens25 or collagen at concentrations 50-fold greater10 than those used in the present study) could indicate that in certain environments it has a broader binding spectrum and a more general (perhaps affinity-dependent) role in the scavenging of collagen and collagen fragments in other cell types, including macrophages. It would therefore be interesting to study if Kupffer cells can penetrate/extravasate through the sinusoidal layer to phagocytose collagen deposits via MRs, thereby preventing hepatic fibrosis.

Figure 4.

LSECs mediated endocytosis of DenColl and FSA in (A) MR knockout mice and (B) wild-type mice. Radiolabeled ligands (0.1 μg/ml) were given to confluent 1-cm2 LSEC cultures for 2 hours at 37°C. To check the specificity of [125I]DTAF-DenColl endocytosis, unlabeled DenColl (100 μg/ml) was added (middle bar). Results represent the mean endocytosis ± SEM for 3 independent experiments representing 3 different MR knockout or wild-type animals. The results of each experiment, presented as the percentage of added ligand, are means of duplicate measurements.

Figure 5.

Specificity of endocytosis of (A) DenColl and (B) mannan in LSECs from wild-type mice. (A) [125I]DTAF-DenColl and (B) [125I]mannan were given to 1-cm2 LSEC cultures for 2 hours at 37°C at 0.1 μg/mL in the absence and in the presence of the inhibitors listed on the x axis. Inhibitor concentrations: DenColl, NatColl, mannan, and FSA: 100 μg/mL; EGTA, 2.5 mM; mannose, 50 mM. Each figure shows the mean for 3 independent experiments using different animals. The results of each experiment (n = 3), presented as a percentage of the control value, are means of duplicate measurements. (A) Mean endocytosis of [125I]DenColl in control cultures as a percentage of the total added ligand was 34% ± 4% (mean ± SEM). (B) Mean endocytosis of [125I]mannan in control cultures was 14% ± 2% (mean ± SEM). Error bars represent the SEM.

There remains the issue of DenColl uptake in MR knockout livers. To address this, we cultured parenchymal and nonparenchymal liver cells and challenged them with DTAF-DenColl. We saw no uptake of this ligand in LSECs, Kupffer cells, or stellate cells from MR knockout mice (not shown). However, with large doses of DTAF-DenColl (10 μg/ml), we saw some vesicular uptake of this ligand in hepatocytes from both MR knockout and wild-type livers (not shown). When we challenged dense MR knockout hepatocyte cultures with [125I]DTAF-DenColl (0.1 μg/ml) for 2 hours, less than 2% of the label was taken up (compared to the 23% taken up by confluent wild-type LSECs), and this could not be inhibited by excess DenColl. It is therefore possible that hepatocytes possess a DenColl uptake mechanism unrelated to the MR that may be responsible for the slow clearance of this ligand in MR knockout mice.

That the LSEC MR is the main site of DenColl clearance from the blood has several implications for the study of hemostasis and fibrosis. DenColl is a major waste product cleared from the blood by LSECs. For example, collagen liberated during bone turnover is drained directly to the blood. Given that 10% of human bone mass is turned over yearly, this alone would result in the daily release of approximately 0.5 g of collagen fragments to the blood.29, 30 Furthermore, most collagen α-chains in the rat interstitium are transported out to circulation and removed rapidly by uptake into LSECs,5 adding to that released from bone. Given this daily load, it is likely that proper function of the LSEC receptor for denatured collagens is essential for maintaining normal plasma viscosity and protein balance. However, the MR knockout mice (at least up to young adulthood) appeared normal, so the fate of the collagen fragments in these animals needs to be addressed. Some of the injected DenColl was cleared via the kidneys in the MR knockout mice, but a portion remained in the liver, possibly taken up by other mechanisms. We are currently monitoring these animals as they age for any collagen that might be deposited (or other effects) in the liver and other organs.

In normal animals, aging and its effects on LSECs need also to be considered with respect to denatured collagen/collagen fragment clearance. The phenomenon of pseudocapillarization,31 that is, age-related changes in liver sinusoids whereby fenestration is lost and a (normally absent) basement membrane is established (and which shares many morphological characteristics with liver fibrosis32, 33), could result in the inhibition of LSEC-mediated removal of collagen fragments from the space of Disse because of the physical barrier of high-density collagen. A potential reduction in the endocytic function of LSECs during pseudocapillarization/fibrosis, resulting in increased levels of denatured collagen and other connective tissue waste products,22 also needs to be considered. Increased circulating levels of such material would competitively inhibit matrix-degrading enzymes that would otherwise degrade and thereby prevent the irreversible deposition of collagen. It could thus be contemplated that a vicious circle could develop, whereby reduced removal of collagen by LSECs could lead to increased pseudocapillarization/fibrosis that in turn could lead to even greater reductions of LSEC-mediated collagen removal.

The collagen-binding25 MR FNII domain is now an interesting target for identifying individuals with dysfunctional MRs in order to determine if polymorphisms in this region are a risk factor for liver fibrosis. Furthermore, liver fibrosis is generally reversible if the primary injury is removed,32 so it may be interesting to study the role of LSECs in this process, especially with respect to developing treatment regimes that enhance LSEC-mediated collagen removal during fibrosis or the recovery from same.


We thank Cristina Øie for providing excellent rat LSEC preparations.