INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE MANNOSE RECEPTOR IS A C-type lectin containing multiple carbohydrate recognition domains (CRDs) and is expressed on Kupffer cells (16), alveolar (42), peritoneal (40), and splenic macrophages (9), monocyte-derived dendritic cells (4), and subsets of vascular and lymphatic endothelial cells (44). It specifically recognizes and binds to ligands having terminal nonreducing D-mannose, N-acetylglucosamine, or L-fucose units in a Ca²⁺-dependent manner (2, 30). These carbohydrates are not normally displayed at the ends of carbohydrate chains of mammalian cells but are frequently found on the surfaces of microorganisms (1). Several in vivo functions have been proposed for the mannose receptor on macrophages: endocytosis of extracellular peroxidases and hydrolases during the resolution phase of inflammation (39), phagocytosis of unopsonized pathogens (41), and antigen capture for eventual presentation to T cells (37).

Besides the mannose receptor, mannan-binding proteins (MBPs) also bind to a variety of pathogens by recognizing D-mannose, N-acetylglucosamine, or L-fucose on their surface (52). MBP is a large oligomeric serum protein of hepatic origin, and it belongs to the family of Ca²⁺-dependent collagenous lectins, most of which are components of the innate immune system (12). MBP also activates the complement system following binding to its ligands in an antibody- or C1q-independent manner (15, 27). It has been isolated from serum of several species, including human (43, 48), rabbit (20), rat (36), bovine (18), and mouse (8, 13).

Both groups of mannose-specific lectins, i.e., mannose receptors and MBP, are believed to be involved in the in vivo recognition of glycoproteins terminating with mannose. Taylor et al. (46) carried out a pharmacokinetic analysis of the in vivo uptake and processing of mannose-terminating glycoproteins by rat hepatic mannose receptors. However, they did not examine the binding of the ligands to serum lectins, such as MBP, an event that might affect their biodistribution profiles. An in vitro experiment indicated that human serum lectins inhibit the uptake of glycoproteins by the hepatic mannose receptor (47). Although the recognition characteristics of both mannose-specific lectins have been extensively investigated in vitro (22, 29-30, 45, 46), it is not yet known whether serum mannose-specific lectins affect the in vivo hepatic uptake of mannosylated ligands via mannose receptor-mediated endocytosis.

In vitro observations suggest that the configuration, number, and density of mannose on ligands are critical determinants for their interaction with both types of lectins (22-23, 29, 45, 46). Furthermore, since the MBP is a relatively bulky and large lectin with several CRDs, the size of the ligands is an important factor in determining their recognition. Therefore, to elucidate the in vivo recognition characteristics of mannosylated ligands by hepatic mannose receptors and MBP, we designed the following experiments. Three proteins, recombinant human superoxide dismutase (SOD, 32 kDa), bovine serum albumin (BSA, 67 kDa), and bovine immunoglobulin G (IgG, 150 kDa) were selected as model ligands having different molecular masses. Since these proteins themselves have no mannose-terminating oligosaccharides, they were chemically modified with mannose to obtain mannosylated glycoproteins with varying numbers of mannose residues. Their in vivo disposition was examined in mice after intravenous bolus injection at different doses, and their biodistribution profiles were analyzed using a pharmacokinetic model to obtain parameters representing the receptor affinity of both lectins. The importance of the degree of mannosylation and the molecular mass of ligands is discussed in relation to the molecular characteristics of mannose receptors and MBP.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials. BSA (fraction V) and IgG were purchased from Sigma (St. Louis, MO). Recombinant human SOD was supplied by Asahi Kasei (Tokyo, Japan). D-Mannose was obtained from Nacalai Tesque (Kyoto, Japan). ¹¹¹InCl₃ was supplied by Nihon Medi-Physics (Takarazuka, Japan). All other chemicals were reagent-grade products obtained commercially.

Animals. Male ddY mice (25-28 g) were obtained from the Shizuoka Agricultural Cooperative Association for Laboratory Animals (Shizuoka, Japan). All procedures were examined by the Ethics Committee on Animal Experiment at the Kyoto University, and animal care was in accordance with the National Institutes of Health Guidelines for Animal Experiments and the law of the Japanese government.

Synthesis of glycosylated proteins. Coupling of mannose moieties to protein was carried out by the method of Lee et al. (24). Briefly, cyanomethyl 2,3,4,6-tetra-O-acetyl-1-thiomannoside was prepared from the respective pseudothiourea derivatives and chloroacetonitrile. The nitrile group in these cyanomethyl thioglycosides can be converted to a methyl imidate group by treatment with sodium methoxide in dry methanol to yield 2-imino-2-methoxyethyl 1-thioglycosides. Cyanomethyl 1-thiomannoside was treated with 0.01 M sodium methoxide at room temperature for 24 h, and a syrup of 2-imino-2-methoxyethyl-1-thiomannoside was obtained after evaporation of the solvent. A quantity of the resultant syrup was added to protein in borate buffer (pH 9.5). The number of mannose residues per protein molecule was controlled by the molar ratio of the starting reagents. After 24 h at room temperature, the reaction mixture was dialyzed to remove any unreacted compound and lyophilized. The physicochemical properties of the synthetic mannosylated proteins with different sugar densities are summarized in Table 1. The apparent molecular mass of each mannosylated protein was estimated by SDS-PAGE. The number of mannose residues was determined by calculating the mannose content of the mannosylated protein solution using the anthrone-sulfuric acid method (7). The protein content was calculated by subtracting the weight of mannose from that of mannosylated protein. The final number of mannose residues was obtained by dividing the molar amount of mannose by the molar amount of protein.

¹¹¹In labeling of mannosylated proteins. Each mannosylated protein was radiolabeled with ¹¹¹In using the bifunctional chelating agent diethylenetriaminepentaacetic acid (DTPA) anhydride, according to the method of Hnatowich et al. (10). Briefly, 10 µl DTPA anhydride in dimethyl sulfoxide was added to each mannosylated protein. The mixture was stirred for 30 min at room temperature and purified by using a Sephadex G-25 column to remove any unreacted DTPA. To ¹¹¹InCl₃ solution was added 1 M sodium acetate (pH 6.0), followed by DTPA-coupled mannosylated protein. After 30 min, the mixture was purified using a PD-10 column and the fractions containing derivatives were collected and concentrated by ultrafiltration. The protein concentration was determined by the method of Lowry et al. (25).

In vivo distribution experiment. Each ¹¹¹In-labeled mannosylated protein in saline was injected intravenously into mice via the lateral tail vein at a dose of 0.05, 0.1, 1, 10, or 20 mg/kg. At periods (1, 3, 5, 10, 30, and 60 min) after injection, blood was collected from the vena cava under ether anesthesia. The mice were then killed; the heart, lung, liver, spleen, kidney, and muscle were removed, rinsed with saline, and weighed; and then the radioactivity was assayed in a well-type NaI scintillation counter (ARC-500; Aloka, Tokyo, Japan).

Calculation of area under the curve and clearances. The distribution data of compounds after intravenous injection were analyzed in terms of the organ uptake clearance (CL_org) (33). Because of limited efflux of radioactivity from tissues due to the characteristics of ¹¹¹In-DTPA labeling, CL_org (ml/h) can be expressed (see Ref. 34 for details) as

(1)

where X_i (normalized to % of the dose) represents the amount of radioactivity in tissue i after the administration of the ¹¹¹In-labeled compound, C_p (% of dose/ml) is the concentration of radioactivity in the plasma, and AUC_t (% of dose · h · ml¹) represents the area under the plasma concentration-time curve from time 0 to t, calculated by fitting a monoexponential equation to the plasma concentration-time data of the ¹¹¹In radioactivity-time profile using the nonlinear least squares program MULTI (53). Then CL_org can be calculated simply from Eq. 1 at several time points after administration. In addition, the total body clearance (CL_total) was calculated by dividing the dose by the AUC up to infinity.

Pharmacokinetic analysis based on a physiological model. The time courses of the plasma concentration and liver accumulation of ¹¹¹In-mannosylated proteins were analyzed using the model shown in Fig. 1 (33). In this model, three compartments, i.e., the plasma pool (PP), the sinusoidal and Disse spaces in the liver (EC), and the intracellular space in the liver (IC), represent the body as a whole. The PP and EC compartments have apparent volumes of distribution (V_p and V_l, respectively). The PP compartment represents all plasma spaces within blood vessels of all tissues except the liver; it is connected to the EC by hepatic plasma flow (Q). The uptake of mannosylated proteins from the EC to the IC is expressed as a saturable process following Michaelis-Menten kinetics, with a maximum rate of uptake (V_max,l; nmol/h) and a Michaelis-Menten constant (K_m,l; nM). Extrahepatic elimination from the PP is assumed to be a saturable process represented by V_max,p (nmol/h) and K_m,p (nM). Since some derivatives showed capacity-limited plasma protein binding, this process was expressed as a maximum binding concentration (B_max; nM), and the dissociation constant (K_d; nM) was expressed as

(2)

where C_total is the total concentration of mannosylated protein in the PP and C_free is the concentration of free (unbound) mannosylated proteins in the PP. Only free mannosylated protein is assumed to be involved in the hepatic uptake and extrahepatic elimination. At time 0, the injected mannosylated protein is assumed to be distributed in the PP and EC compartments at the same concentration. The mass balance equation for the concentration of mannosylated proteins in PP, EC, and IC is expressed as

(3)

(4)

(5)

where C_p,free and C_l,free are the concentrations of free (unbound) mannosylated proteins in the PP and EC, respectively, X_l is the amount accumulated in the IC, and Q is the hepatic plasma flow rate. The values of V_p, V_l, and Q were assumed to be 1.5, 0.15, and 85 ml/h, respectively (6). To obtain the pharmacokinetic parameters, these equations were fitted to the data using the nonlinear least squares method MULTI associated with the Runge-Kutta-Gill method, [MULTI(RUNGE)] (damping Gauss Newton method) using an M-382 mainframe computer located in the Kyoto University data processing center.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Physiological pharmacokinetic model for analyzing the disposition of mannosylated proteins. Km,p and Km,l, Michaelis-Menten constants for plasma pool and liver, respectively; Vmax,p and Vmax,l, maximum rates of uptake for plasma pool and liver, respectively; Kd, dissociation constant for plasma protein binding; Bmax, maximum concentration of plasma protein binding; Q, hepatic plasma flow rate; Cbound, concentration of bound protein; Cfree, concentration of free protein.

Isolation of MBP from mouse serum. MBP was isolated from BALB/c mouse serum (Japan Bio-supply) using a Sepharose-mannan or Sepharose-Man-BSA column according to the procedure reported by Kawasaki et al. (19). Briefly, mannan or Man₄₆-BSA was coupled to N-hydroxysuccinimide (NHS)-activated Sepharose 4 Fast Flow (Pharmacia Biotech, Uppsala, Sweden) according to the instructions of the manufacturer. The mouse serum was diluted with an equal volume of a buffer consisting of 40 mM imidazol-HCl, pH 7.8, 40 mM CaCl₂, and 2.5 M NaCl. The mixture was applied to the Sepharose-mannan or Sepharose-Man-BSA column, which had been equilibrated with a loading buffer (20 mM imidazol-HCl, pH 7.8, 20 mM CaCl₂, and 1.25 M NaCl). The binding protein was eluted with an elution buffer (20 mM imidazol-HCl, pH 7.8, 1.25 M NaCl, and 2 mM EDTA). The eluate was applied to the second and third smaller affinity columns. The final column was washed with the loading buffer and eluted with more loading buffer containing 100 mM mannose. All of the procedures were carried out at 4°C.

In vitro assay of serum protein binding. To deplete MBP, mouse serum was applied to a Sepharose-Man-BSA column twice at a rate of 1 drop/15 s to obtain MBP-depleted serum. ¹¹¹In-Man₁₆-, ¹¹¹In-Man₂₅-, and ¹¹¹In-Man₄₆-BSA were incubated with mouse serum, MBP-depleted serum, or buffer containing purified MBP (Man-BSA:MBP; 2:1 molar ratio) for 30 min at 4°C, and then the mixture was applied to a Sephadex G-200 column (Pharmacia) and eluted with buffer (1.25 M NaCl, 20 mM CaCl₂ and 20 mM imidazol). In one case, 100 mM mannose or 2 mM EDTA was added to the buffer. The ¹¹¹In radioactivity in each fraction was determined by scintillation counting.

In situ liver perfusion. To evaluate the hepatic uptake of ¹¹¹In-Man-BSAs under serum-free conditions, an in situ liver perfusion experiment was carried out as reported previously (31). Briefly, mouse liver was perfused in a single-pass mode at a flow rate of 2 ml/min with Krebs-Ringer-bicarbonate buffer containing 10 mM glucose and 3% (wt/vol) BSA. The buffer was oxygenated with 95% O₂-5% CO₂, adjusted to pH 7.4, and incubated at 37°C. The liver was perfused for 2 min to allow it to stabilize, and then 100 µl of each ¹¹¹In-Man-BSA (10 µg/ml) was administered to the liver as a bolus through the portal cannula. Then, 3 min later, the liver was excised and its radioactivity was counted. Statistical analysis was performed by ANOVA. P < 0.001 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Tissue distribution of ¹¹¹In-mannosylated proteins after intravenous injection. After intravenous injection, all ¹¹¹In-mannosylated proteins were mainly recovered in the liver. However, their distribution patterns varied depending on the molecular mass of the protein, the number of mannose residues, and the dose administered (Figs. 2, 3, and 4). Figure 2 shows the plasma concentration and liver accumulation time courses of ¹¹¹In-Man-SODs after intravenous injection. At the lower doses of 0.05 and 0.1 mg/kg, >80% of the injected dose was recovered in the liver. The plasma elimination rate was approximately inversely related to the hepatic uptake. Increasing the dose reduced the amount and rate of liver accumulation of both derivatives because of saturation of the mannose receptor-mediated hepatic uptake. In such cases, radioactivity was also detected in kidneys and urine, because the molecular mass of the Man-SODs was less than the glomerular filtration threshold (28).

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Time courses of plasma concentration (top) and liver accumulation (bottom) of 111In-Man17-SOD (A) and 111In-Man21-SOD (B) after intravenous injection into mice at doses of 0.05 (), 0.1 (), 1 (), 10 (), and 20 () mg/kg. Results are means ± SD of 3 mice.

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Time courses of plasma concentration (top) and liver accumulation (bottom) of 111In-Man12-BSA (A),111In-Man16-BSA (B),111In-Man25-BSA (C),111In-Man35-BSA (D), and 111In-Man46-BSA (E) after intravenous injection into mice at doses of 0.05 (), 0.1 (), 1 (), 10 (), and 20 () mg/kg. Results are means ± SD of 3 mice.

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Time courses of plasma concentration (top) and liver accumulation (bottom) of 111In-Man32-IgG (A) and 111In-Man42-IgG (B) after intravenous injection into mice at doses of 0.05 (), 0.1 (), 1 (), 10 (), and 20 () mg/kg. Results are means ± SD of 3 mice.

In vitro interaction of ¹¹¹In-mannosylated proteins with serum-type MBP. Under reducing conditions, a single band (~30 kDa) could be detected on the SDS-PAGE of the protein(s) purified from mouse serum using an affinity column (Fig. 5). Several high-molecular-mass bands were seen under nonreducing conditions. These results agreed with those of previous papers (8, 13, 19-20), indicating that the purified protein is serum-type MBP. In the following experiments, MBP and MBP-depleted serum were prepared by this affinity column procedure as described in MATERIALS AND METHODS.

View larger version (58K): [in this window] [in a new window]   Fig. 5.   SDS-PAGE of isolated mannose binding protein. Purified binding protein was subjected to 10% polyacrylamide gel electrophoresis under reducing conditions (-mercaptoethanol). Lane 1, molecular mass marker; line 2, mannose binding protein using a Man-BSA-Sepharose affinity column; lane 3, mannose binding protein using a Mannan-Sepharose affinity column. The binding protein band has an apparent molecular mass of ~30 kDa.

View larger version (12K): [in this window] [in a new window]   Fig. 6.   Gel filtration (Sephadex G-200) patterns of 111In-Man-BSAs. A: 111In-Man46-BSA alone incubated with buffer (), incubated with serum (), eluted with buffer containing mannose 100 mM (), and eluted with buffer containing EDTA 2 mM (). B: 111In-Man46-BSA incubated with serum-depleted mannan-binding protein (MBP) () or incubated with purified MBP () (Man-BSA:MBP; 2:1 mol ratio). C: 111In-Man25-BSA incubated with buffer () or incubated with serum (). D: 111In-Man16-BSA incubated with buffer () or incubated with serum () for 30 min at 4°C and eluted with buffer (1.25 M NaCl, 20 mM imidazol and 20 mM CaCl2).

Calculation of AUC and clearances. Table 2 summarizes the AUC, CL_total, hepatic uptake clearance (CL_liver), and tissue uptake rate index (tissue uptake clearance per unit tissue weight) for representative tissues of ¹¹¹In-mannosylated proteins after intravenous injection at various doses. As for all ¹¹¹In-mannosylated proteins, the CL_liver made the major contribution to the CL_total. The CL_liver of ¹¹¹In-Man-SODs as well as ¹¹¹In-Man₁₂- and ¹¹¹In-Man₁₆-BSA decreased on increasing the dose, whereas the CL_liver of other derivatives peaked at a dose of 1 mg/kg.

View larger version (32K): [in this window] [in a new window]   Fig. 7.   AUC of mannosylated compounds at a dose of 0.05 mg/kg plotted against molecular mass and mannose content (wt/wt%)

Pharmacokinetic analysis based on a physiological model. On the basis of the pharmacokinetic analysis carried out so far, it is obvious that both the molecular mass and mannose content of the mannosylated proteins play crucial roles in determining the interaction with plasma MBP. To achieve a more precise and quantitative analysis of the processes of hepatic uptake and plasma protein binding of mannosylated proteins, the biodistribution profiles of ¹¹¹In-Man-BSAs were analyzed using the physiological pharmacokinetic model shown in Fig. 1. Differential Eqs. 2-4 were simultaneously fitted to the plasma concentration and liver accumulation data for each ¹¹¹In-mannosylated protein at five doses, and the pharmacokinetic parameters were estimated (Table 3) and plotted against the number of mannose units per BSA (Fig. 8). Of these parameters, only K_m,l and K_d, which correspond to the affinity of mannosylated proteins for the hepatic mannose receptors and MBP, respectively, varied depending on the number of mannose units. The K_m,l values were all fairly similar (34-68 nM) except for ¹¹¹In-Man₁₂-BSA (300 nM). On the other hand, the K_d of ¹¹¹In-mannosylated proteins increased exponentially on increasing the number of mannose residues from 3,000 nM for ¹¹¹In-Man₁₂-BSA to 0.27-0.3 nM for ¹¹¹In-Man₃₅-BSA and ¹¹¹In-Man₄₆-BSA.

View larger version (15K): [in this window] [in a new window]   Fig. 8.   Effect of the number of mannose residues on the pharmacokinetic parameters of Man-BSAs: Km,l (A), representing the affinity for the hepatic receptor, and Kd (B), representing the plasma protein binding, respectively. The parameters obtained by fitting Eqs. 3-5 to the experimental data were plotted as means ± SD against the number of mannose residues for each protein.

In situ liver perfusion. Fig. 9 shows the hepatic recovery (%dose/g liver) of ¹¹¹In-Man-BSAs in perfused mouse liver after bolus administration to the liver via the portal vein. There were no significant differences in recovery between ¹¹¹In-Man₁₆-, ¹¹¹In-Man₂₅-, ¹¹¹In-Man₃₅- and ¹¹¹In-Man₄₆-BSAs (11-14% of the dose/g tissue), but that of ¹¹¹In-Man₁₂-BSA was significantly lower than those of other ¹¹¹In-Man-BSAs (6.5% of dose/g tissue).

View larger version (10K): [in this window] [in a new window]   Fig. 9.   Percentage recovery of radioactivity in liver of 111In-Man-BSAs after bolus injection via the portal vein in the mouse liver perfusion experiment. Three minutes after injection, the liver was excised and radioactivity was measured. Results are expressed as the means ± SD of 5 mice. #P < 0.001 vs. other Man-BSAs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Although various types of cells have mannose receptors on their surface, ligands with mannose are largely taken up by liver nonparenchymal cells after intravenous injection. Therefore, it has been demonstrated that naturally or chemically mannosylated ligands are promising candidates for drug carriers targeting liver nonparenchymal cells via mannose receptors (3, 17, 38). In one case, the in vivo targeting efficiency of Man-BSA to the whole liver exceeded 80% of the injected dose (35, 46), and the endothelial and Kupffer cells contribute about 66% and 21% of the uptake, respectively, after intravenous injection (35). On the other hand, Kawasaki et al. (19) have reported that MBP, another mannose-specific lectin, circulates in serum and recognizes pathogens that have mannose units on their surface. Although the in vitro binding specificity and characteristics of hepatic mannose receptors and serum MBP have been studied in detail (22, 29-30, 45, 46), little information is available on their in vivo recognition characteristics (46). Furthermore, little attention has been paid to the in vivo interaction between serum MBP and mannosylated ligands. Therefore, in the present study, we tried to obtain information about the in vivo recognition characteristics of these mannose-specific lectins by analyzing the biodistribution of mannosylated proteins with different physicochemical properties.

Using a metabolizable ¹²⁵I labeling, Taylor et al. (46) studied the uptake and metabolic fate of mannosylated ligands after systemic administration. Instead, we applied an ¹¹¹In labeling combined with DTPA anhydride, since ¹¹¹In labeling prepared by this method is metabolically stable in the bloodstream and is locked within cells that take up the labeled compounds even after intracellular degradation (10). These characteristics of labeling enabled us to quantitatively trace the plasma profile and tissue uptake of mannosylated proteins by simply counting radioactivity.

Because of the limited number of mannose receptors, increasing the dose of ligands saturates the receptors followed by a decrease in hepatic uptake when normalized with the administration dose. This is the case for ¹¹¹In-Man-SODs as well as ¹¹¹In-Man₁₂- and ¹¹¹In-Man₁₆-BSA; their hepatic uptake simply decreased with increasing dose, as observed with galactosylated proteins (32). Although uptake clearances by other tissues decreased on the increasing the dose in some cases, the absolute values were very small compared with those by the liver (Table 2). Therefore, the mannose receptor-mediated hepatic uptake seems to be the only mannose-specific and capacity-limited process determining the biodistribution of these Man-SODs and Man-BSAs with fewer mannose units.

When mannosylated ligands interact with serum lectins like MBP, their biodistribution becomes more complicated. Incubated in mouse serum in vitro, ¹¹¹In-Man₂₅-BSA and ¹¹¹In-Man₄₆-BSA showed an interaction with a serum component (Fig. 6) that was identified as MBP by determining its molecular size and in vitro binding characteristics. Martinez-Pomares et al. (26) found in mouse serum the presence of soluble mannose receptors that interact with carbohydrate chains in a mannose- or fucose-dependent manner. However, we could not detect any other band on SDS-PAGE except for 30 kDa (which corresponds to MBP), indicating that the level of soluble mannose receptors was too low to be detected in normal serum or that they were so unstable that we could not separate them by this procedure.

MBP is known to activate the complement system on binding to its ligands, but the in vivo fate of MBP/mannosylated ligand complex is hardly understood. Several studies showed that MBP is recognized by collectin receptors expressed on macrophages, monocytes, and neutrophils (5, 12, 49-50). However, on the basis of the biodistribution of ¹¹¹In-Man-BSAs and ¹¹¹In-Man-IgGs, MBP-bound mannosylated ligands do not seem to be rapidly removed by hepatic mannose receptors or by collectin receptors. This is also supported by our preliminary observation that the uptake of ¹¹¹In-Man₄₆-BSA by cultured peritoneal macrophages is not assisted by complex formation with MBP. In addition, ¹¹¹In-Man₄₆-BSA preincubated with purified MBP (1:1 molar ratio) showed a slower hepatic uptake than ¹¹¹In-Man₄₆-BSA itself after intravenous injection in mice (data not shown). Therefore, it is reasonable to assume that MBP/mannosylated protein complex is not taken up by liver nonparenchymal cells via mannose receptor-mediated endocytosis. On the basis of this assumption, the AUC value at the lowest dose of 0.05 mg/kg is a good index for the semiquantitative comparison of the binding strength of each mannosylated protein to MBP in vivo: a large AUC indicates strong binding to MBP, whereas a small AUC means weak or no binding (Fig. 7). ¹¹¹In-Man-SODs had very small AUCs in spite of their high mannose content (wt/wt). Although the ¹¹¹In-Man-IgGs used in this study have a lower mannose content than ¹¹¹In-Man₄₆-BSA, their AUCs were larger than that of ¹¹¹In-Man₄₆-BSA. These results suggest that the molecular mass of mannosylated proteins is very important for their interaction with MBP. When the molecular mass is identical, the number of mannose moieties also has a significant effect on recognition by MBP.

MBP-mediated retardation of hepatic uptake was only obvious at the low doses of 0.05 and 0.1 mg/kg. This phenomenon could be explained by the limited amount of MBP in serum; its concentration in mouse serum was reported to be from 5 to 80 µg/ml (8). Increasing the dose will saturate the binding of ligands to MBP, and excess ligands, existing in the free form in the circulation, will be recognized by hepatic mannose receptors. At >1 mg/kg, the mannose receptors were also saturated with the ligands and the hepatic uptake was retarded.

On the basis of the same model as this study, we concluded that the in vivo affinity of galactosylated proteins with the asialoglycoprotein receptor is simply increased on increasing density of galactose on the ligand surface (32). In the present study, a capacity-limited binding in plasma was also assumed due to the presence of MBP in serum. We also assumed that the bound form of the ligand does not participate in the hepatic uptake on the basis of the biodistribution experiments.

Increasing the number of mannose units from 12 (Man₁₂-BSA) to 16 (Man₁₆-BSA) greatly reduced the K_m,l, indicating that the hepatic mannose receptors require at least 16 mannose residues on BSA for efficient uptake in vivo. The binding affinity of ligands for mannose receptors has been reported to depend on the clustering and geometric organization of the mannose residues on a branched sugar chain (30). Man-BSA, having a larger number of mannose residues, could take part in a multivalent interaction with mannose receptors, followed by rapid uptake. On the basis of an in vitro binding study using alveolar macrophages, Hoppe and Lee (14) reported that the binding affinity of Man-BSA increased on increasing the number of mannose units from 5 to 43. In the present study, however, the K_m,l of ¹¹¹In-Man-BSAs, except for ¹¹¹In-Man₁₂-BSA, remained relatively constant regardless of the number of mannose units. To confirm the results obtained by the analysis, the in situ hepatic uptake of ¹¹¹In-Man-BSAs was examined in perfused mouse livers where no blood components were involved. All ¹¹¹In-Man-BSAs, except for ¹¹¹In-Man₁₂-BSA, showed a similar level of hepatic uptake, supporting the parameters obtained by the pharmacokinetic analysis. This finding, that the affinity does not increase on increasing the number of mannose units above 16 per BSA, is not consistent with the previous report (14). This discrepancy could be due to differences in the experimental conditions: cultured cells vs. whole animals, and/or alveolar macrophages vs. hepatic nonparenchymal cells. The results obtained here were based on the in vivo disposition of mannosylated proteins, so they reflect the in vivo recognition characteristics of the mannose receptors more accurately than those obtained in vitro.

In contrast to the affinity with hepatic mannose receptors, the K_d on the protein binding in serum, which correlates with the affinity of ¹¹¹In-Man-BSAs to MBP, fell markedly on increasing the number of mannose residues from 3,000 to 0.27-0.3 nM. This discrepancy between the two types of mannose-specific lectins in the characteristics of ligand recognition can be explained by the difference in the molecular structures of the hepatic mannose receptors and MBP. The mannose receptor is a type I transmembrane protein with eight different C-type CRDs in a single polypeptide. On the other hand, the monomer of MBP contains only one CRD at the COOH-terminus, three monomers are held together by interaction with the -helical neck region, and two to six sets of the trimer form the bouquet-like structure of MBP (52). These structural characteristics of MBP could account for the strong binding of highly mannosylated BSAs (11). The relationship between the density of the sugar units and the K_d was similar to that between the density of galactose units on proteins and the K_m,l of their hepatic uptake (32). These findings suggest that MBP exhibits a significant cluster effect in its recognition of ligands larger than BSA, as is the case with the asialoglycoprotein receptors (51).

In summary, serum MBP only recognized ligands with a larger molecular mass, and its binding affinity significantly increases on increasing the molecular mass of the ligands and the density of mannose units. It was quantitatively shown to have a strong cluster effect. In contrast, recognition of the hepatic mannose receptors was hardly affected by the ligand size and depended only on the density of the mannose units on ligands. The differences in the recognition of mannosylated ligands by hepatic mannose receptors and serum MBP could be explained by the fact that the hepatic mannose receptors contain multiple different CRDs in the single polypeptide, whereas serum MBP is composed of six or more monomers with only a single CRD. These findings will prove useful not only for understanding the physiological roles of these lectins in host defense but also for designing drug carriers targeting liver nonparenchymal cells.