Tissue targeting of multivalent Le(x)-terminated N-linked oligosaccharides in mice.

The target site for N-linked biantennary and triantennary oligosaccharides containing multiple terminal LeX determinants was analyzed in mice. N-Linked oligosaccharides containing a single tert-butoxycarbonyl-tyrosine attached to the reducing end were used as synthons for human milk α-3/4-fucosyltransferase to prepare multivalent LeX (Galβ1-4[Fucα1-3]GlcNAc) terminated tyrosinamide oligosaccharides. The oligosaccharides were radioiodinated and examined for their pharmacokinetics and biodistribution in mice. The liver was the major target site in mice at 30 min, which accumulated 18% of the dose for LeX biantennary compared with 6% for a nonfucosylated Gal biantennary. By comparison, LeX- and Gal-terminated triantennary accumulated in the liver with a targeting efficiency of 66 and 59%, respectively. The liver targeting of LeX biantennary was partially blocked by co-administration with either galactose or L-fucose whereas LeX triantennary targeting was only reduced by co-administration with galactose. In contrast to these results in mice, in vivo experiments performed in rats established that both LeX and Gal terminated biantennary target the liver with nearly identical efficiency (6-7%). It is concluded that the asialoglycoprotein receptor in mice preferentially recognize LeX biantennary over Gal biantennary, whereas little or no differentiation exists in rats. Thereby, the mouse asialoglycoprotein receptor apparently possesses additional binding pockets that accommodate a fucose residue when presented as LeX.

In mammals, carbohydrate/protein interactions often involve the binding of an oligosaccharide ligand to a cell surface receptor (1,2). The ligands are most frequently N-or O-linked oligosaccharides that are covalently attached to a glycoprotein. N-Linked oligosaccharides possess a common pentasaccharide core structure which contains a branch point resulting in two or more nonreducing end sugar residues (3). It is often these terminal sugar residues on N-linked oligosaccharides which bind to spatially resolved binding sites on a lectin (4).
The mammalian lectins discovered to date have been grouped into several subcategories (5). Several membrane spanning lectins are known to be C-type lectins which contain a carbohydrate recognition domain named for its calcium-dependent ligand binding. Of the C-type lectins, the asialoglycoprotein receptor (ASGP-R) 1 found on hepatocytes has been most thoroughly studied for its binding specificity and its intracellular routing of ligands (6 -8). N-Linked oligosaccharides containing multiple terminal Gal residues bind with high affinity, although GalNAc terminated N-linked oligosaccharides are much more potent ligands (9 -11). A biantennary oligosaccharide possessing only two terminal GalNAc residues is a superior ASGP-R ligand compared with a triantennary possessing three terminal Gal residues (12).
In addition to the ASGP-R, several other mammalian lectins have been isolated and characterized in liver. Kupffer cells possess a C-type lectin that binds avidly to Fuc-bovine serum albumin but also has shared affinity for Le X -bovine serum albumin (13)(14)(15)(16). A Gal-specific lectin that binds asialyl-tetraantennary N-linked oligosaccharides and a Fuc/GlcNAc binding lectin have been found on rat alveolar macrophages and Kupffer cells (17,18). Also, a lectin that binds GalNAc-4-sulfate on N-linked oligosaccharides has been shown to exist on endothelial cells (19).
The ligand specificity of most mammalian lectins is deduced from an in vitro measurement of their relative affinity for natural glycoproteins, neoglycoproteins, or simple glycosides (20). However, comparison of results from different binding assays using different receptor preparations contributes to the uncertainty of the identity of the natural ligand(s) for mammalian lectins (21).
Elucidating the binding specificity of liver receptors is further complicated by the unique specificity found for different animals. For example, chicken hepatic lectin fails to recognize Gal terminated oligosaccharides, but instead functions in binding GlcNAc terminated oligosaccharides (22). The most prevalent lectin in alligator liver binds Man-and Fuc-terminated ligands (23). Thereby, the preferred carbohydrate ligand determined using animal models may not necessarily result in comparable binding specificity for liver receptors in man.
In order to discover the tissue location and binding specificity of mammalian lectins in their native environment, we have initiated studies that use well defined N-linked oligosaccharides as probes to reveal lectin activity resulting from ligand uptake into the target organ. This approach offers certain advantages in that only high affinity (K d Ͼ micromole) binding is detected, all receptors with serum exposure are assayed simultaneously, and the exogenous ligand must compete with endogenous ligands, leading to physiologically relevant results that can be directly translated into the design of glycotargeted drug delivery systems (24).
In the present study, we have examined the target site for a series of Le X biantennary and triantennary oligosaccharides.
The results indicate that terminal Gal and L-Fuc residues on biantennary oligosaccharides are simultaneously recognized by the ASGP-R in mice but not in rats. These data further exem-plify subtle differences in carbohydrate receptor ligand specificity between closely related species.
Preparation and Characterization of Le X Biantennary and Triantennary-Triantennary and biantennary oligosaccharides containing a tert-butoxycarbonyl-tyrosine attached to the reducing end GlcNAc through a ␤-glycosylamide linkage were prepared from bovine fetuin (25) and porcine fibrinogen (26) as described previously. The oligosaccharides were used as substrates for human milk ␣3/4-fucosyltransferase. The enzyme was partially purified from frozen human milk using a modification of the method described by Palcic et al. (27). Human milk (300 ml) was defatted by centrifugation for 10 min at 5000 ϫ g. The protein was precipitated with 65% saturated ammonium sulfate and recovered following centrifugation for 25 min at 5000 ϫ g. The precipitate was dialyzed at 4°C for 24 h in M r 12,000 cut-off tubing against 2 ϫ 1 liter of cacodylate buffer (25 mM, pH 6.5, containing 5 mM manganese chloride), and the remaining insoluble protein was removed by centrifugation. The retentate was loaded onto a CM-Sephadex C-50 column (2.5 ϫ 37 cm) equilibrated at 4°C with 25 mM cacodylate buffer and eluted with a 300-ml step gradient of cacodylate buffer containing 0.1 M, 0.2 M, and 0.3 M sodium chloride. The protein eluting in 0.2 M sodium chloride was detected in a 50-ml fraction by A 280 nm and was pooled. The fraction was concentrated at 4°C to 0.5 ml in a Micro-Pro Dicon concentrator (Spectrapore, CA) using a M r 10,000 cut-off membrane while dialyzing against 2 liters of cacodylate buffer.
Analytical transferase reactions were performed by reacting 2 nmol of tyrosinamide biantennary prepared in 7.5 l of 25 mM cacodylate enzyme buffer, pH 6.5 (containing 8 mM manganese chloride and 1.6 mM ATP), with 2 l of enzyme, 1.5 l (42.4 nmol) of GDP-Fuc, and 1 l (1 unit) of alkaline phosphatase. The reaction was incubated at 37°C and after 120 min was diluted 10-fold with water, then analyzed (200 pmol) on analytical C18 reverse phase (RP)-HPLC. The column (50°C) was eluted isocratically at 1 ml/min with 0.1% acetic acid and 11% acetonitrile while monitoring tyrosine fluorescence at an excitation of 275 nm and an emission 305 nm. Fucosyltransferase activity was quantified from the initial rate (10 min) by integration of the biantennary oligosaccharide peak eluting at 22 min relative to the monofucosylated product eluting at 21 min. Each batch of human milk yielded approximately 5 milliunits of enzyme with specific activity of 150 microunits/mg of protein, where 1 unit is defined as the amount of enzyme to transfer 1 mol/min of L-Fuc to a biantennary oligosaccharide. Protein concentrations were determined by the method of Bradford (28).
Preparative fucosyltransferase reactions were performed by treating 500 nmol of biantennary with 4 milliunits of enzyme, 8 mol of GDP-Fuc, and 5 units of alkaline phosphatase in a total volume of 500 l of cacodylate enzyme buffer. Triantennary oligosaccharide (500 nmol) was treated with 10 milliunits of fucosyltransferase, 12 mol of GDP-Fuc, and 5 units of alkaline phosphatase prepared in 500 l of cacodylate enzyme buffer. The reactions were incubated at 37°C for 48 h, which resulted in the formation of a single earlier eluting product peak when monitoring by RP-HPLC as described above.
Le X oligosaccharides were purified from the enzyme reaction from a mixed bed ion exchange column (1 ϫ 37 cm; top: AG50WX2 acid form, bottom: AG1-X2 acetate form) eluted with water while detecting A 214 nm . The oligosaccharide peak eluting at the void of the column (14 ml) was collected and freeze-dried, then purified to homogeneity by RP-HPLC. Multiple 100-nmol injections were performed on an analytical PRP-1 column eluted at 1 ml/min with 0.1% acetic acid and 11% acetonitrile while detecting A 280 nm 0.05 absorbance units at full scale. The peak eluting at 20 -25 min was collected and freeze-dried, reconstituted in water, and a yield of 70% was determined by A 280 nm (⑀ ϭ 1330 M Ϫ1 ).
Monosaccharide analysis of Le X biantennary and triantennary oligosaccharides was performed following trifluoroacetic acid hydrolysis according to method of the Hardy et al. (29). Oligosaccharides were prepared for 500-MHz proton NMR spectroscopy by repeatedly freezedrying 0.4 -0.5 mol in D 2 O. The sample was prepared in 0.5 ml of 99.98% D 2 O containing 0.01% acetone as an internal standard and analyzed on a Bruker 500-MHz NMR spectrometer operating at 23°C. All spectra were processed utilizing resolution enhancement parameters supplied by the Felix software package (Hare Research, Eugene, OR).
Each oligosaccharide was analyzed by FAB-MS by preparing the sample (5 nmol) in 10 l of water containing 1 l of ␣-monothioglycerol. The water was removed by speed vacuum, and the 1-l sample was applied to the probe of the Finigan Matt 900 FAB-MS operated in the positive ion mode.
Radiolabeling and Pharmacokinetic Analysis of N-Linked Oligosaccharides-Iodinations were performed using a modification of the chloramine-T method as described previously (12). The purity of each iodinated oligosaccharide was analyzed by spotting 1 l of diluted sample (2 nCi) at the origin of a TLC plate developed with ethyl acetate:acetic acid:pyridine:water at a ratio of 3:3:2:2 then analyzed by quantitative autoradiography on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
The pharmacokinetics of Le X -terminated biantennary and triantennary oligosaccharides were analyzed as described for other tyrosinamide-oligosaccharides (12). Serial blood time points were analyzed by direct ␥ counting, after which the oligosaccharide was extracted and analyzed by TLC/autoradiography (12). Pharmacokinetic parameters were derived from direct blood counts versus time for three data sets for each oligosaccharide. Iterative nonlinear least squares fits for individual data sets were obtained with PCNONLIN (SCI Software, Lexington, KY) using a two-compartment open model as described previously (12).
Biodistribution of N-Linked Oligosaccharides-Qualitative whole body autoradiography and quantitative biodistribution were determined in mice as described previously (12). Biodistribution was also performed in triplicate in 150-g rats (Sprague-Dawley, Harlan Indianapolis, IN) by dosing 25 pmol (3.5 Ci) of oligosaccharide in 30 l of saline through a jugular vein cannula. At 30-min postadministration, rats were euthanized, and the amount of radioactivity was determined by direct ␥ counting of dissected tissues.
In vivo inhibition experiments were performed in mice by jugular vein dosing 10 pmol (1.8 Ci) of oligosaccharide containing 0.1, 0.2, or 0.3 mmol of Man, L-Fuc, or Gal prepared in 200 l of sterile water. The mice were sacrificed 30-min postadministration and the tissue distribution analyzed as described above.
In vivo saturation experiments were performed by increasing the oligosaccharide dose from 10 pmol to 100 nmol while keeping the radioactive dose (1.8 Ci) constant. Oligosaccharide biodistribution was analyzed at 30-min postadministration as described above.

RESULTS
Asialyl biantennary, with and without core Fuc, and a triantennary oligosaccharide were isolated from porcine fibrinogen and bovine fetuin as described previously (25,26). Prior to purification, the oligosaccharides were derivatized at the reducing end to form an oligosaccharide-glycosylamine which coupled to tert-butoxycarbonyl-tyrosine results in tyrosinamide-oligosaccharides (30). This aglycone was radioiodinated prior to biodistribution experiments, providing oligosaccharide probes that have greater metabolic stability compared with natural glycopeptides, which are rapidly metabolized and exocytosed from liver (12).
tert-Butoxycarbonyl-tyrosinamide-oligosaccharides were used as substrates to generate Le X -terminated oligosaccharides. Partially purified human milk fucosyltransferase catalyzed the transfer of L-Fuc from GDP-Fuc to the 3-hydroxyl group on the subterminal GlcNAc of each of the antennae of a biantennary or triantennary oligosaccharide. Optimal reaction conditions were determined by varying the concentration of GDP-Fuc, fucosyltransferase, and oligosaccharide substrate until transfer was complete (48 h). The transferase reactions were analyzed using RP-HPLC while monitoring the fluorescence of the tyrosinamide group in order to detect earlier eluting peaks resulting from successive transfer of L-Fuc to the substrate. Biantennary oligosaccharides produced one intermediate and a final product (Fig. 1), whereas triantennary resulted in two intermediates and a single product.
Le X oligosaccharides I, II, and III were prepared in 500-nmol quantities using biantennary and triantennary substrates (Fig.  2, structures IV and V) and purified using a polymeric RP-HPLC column. Each purified oligosaccharide demonstrated the presence of either two or three Fuc residues by monosaccharides analysis. The detailed structure of Le X oligosaccharides was elucidated using a combination of proton NMR and FAB-MS. Mass spectral analysis provided a dominant molecular ion at m/z of 2363.7, 2217.5, or 2729.8 corresponding to within 0.8 mass unit of the anticipated mass (M ϩ 23) for oligosaccharides I, II, and III, respectively.
Proton NMR analysis of oligosaccharides I, II, and III resulted in signal patterns consistent with multivalent Le X oligosaccharides ( Fig. 3) (29). Biantennary oligosaccharides I and II had characteristic anomeric protons for Man-4,4Ј (5.107-5.109 ppm) and GlcNAc-1 (4.913-4.917 ppm) but also demonstrated two new anomeric protons for Fuc A and B (5.123-5.133 ppm). The presence of two new L-Fuc residues was also observed from the C6 methyl protons at 1.17-1.18 ppm.
Triantennary oligosaccharide III represents a novel structure not reported previously. The NMR spectra for III contained anomeric (5.110 -5.131 ppm) and methyl resonances (1.173-1.176 ppm) assigned to L-Fuc residues A, B, and C. The chemical shifts for the structural reporter group signals for oligosaccharides I-III are included in Table I.
Le X oligosaccharides I-III were radioiodinated and analyzed for their pharmacokinetics and biodistribution in mice. Following jugular vein dosing, pharmacokinetic analysis verified that the oligosaccharides were rapidly cleared from the blood within 1 h with biphasic kinetics (Fig. 4). Extraction of oligosaccharides from blood followed by TLC separation and autoradiography demonstrated that the oligosaccharides were cleared without the formation of detectable metabolites (Fig. 4). Therefore, blood time points were directly ␥-counted and fit by nonlinear least squares analysis to derive pharmacokinetic constants as described previously (12). Oligosaccharides I and II had a total body clearance (Cl tb ) and steady state volume of distribution (Vd SS ) that was comparable with that reported previously for a Gal biantennary (Table II) (12). Also, the presence of a core Fuc residue on II did not influence the clearance rate of this oligosaccharide. Oligosaccharide III showed a much more rapid elimination from the blood as compared with I and II (Fig. 4), resulting in a further elevation of Cl tb and Vd SS (Table II). The mean resonance time, which approximates the serum half-life, was 14 -17 min for all three oligosaccharides.
Whole body autoradiography established that at 30-min postadministration, oligosaccharides I, II, and III each targeted the liver as the major site with only a minor percentage of the dose (1-4%) targeting kidney and small intestine (Fig. 5). Other tissues possessed only background counts, whereas the urine was the major route of elimination. ␥ counting of the major organs established that biantennary oligosaccharides I and II each targeted the liver with similar efficiency (17-18% of Oligosaccharides I and II represent Le X biantennary oligosaccharides, whereas III is a novel Le X triantennary. Oligosaccharide IV and V were used as synthons to prepare II and III. the dose at 30-min postadministration), whereas triantennary oligosaccharide III targeted liver with approximately 3-fold higher (66%) efficiency (Fig. 5). In contrast, studies in rats revealed the liver targeting for either Le X (II) or Gal (IV) biantennary was approximately 5-6% (Table II).
The elimination rate of biantennary I and II from mouse liver was analyzed to determine if core fucosylation would slow metabolism as was previously noted for GalNAc-terminated biantennary oligosaccharide (12). At times ranging from 30-to 180-min postadministration, both I and II were eliminated from liver with a similar half-life and appeared in the small intestine, suggesting that the ligands were internalized into the target cells and excreted into the bile (Fig. 6). However, the elimination rate for Le X oligosaccharides was comparatively slower (t1 ⁄2 ϭ2.5 h) to that determined previously for GalNAc biantennary (t1 ⁄2 ϭ 1 h) (12).
In vivo inhibition studies were performed by co-administering Man, L-Fuc, or Gal with iodinated II, III, or V in mice. Preliminary experiments established that 0.1-0.3 mmol was adequate to compete for receptor binding, whereas higher concentration resulted in increased mortality of the mice. Also, the viscosity of the dosing solution in these experiments resulted in slightly higher (3-5%) targeting efficiencies compared to dosing in saline.
The liver targeting of Le X biantennary II was not inhibited by co-administration with Man but was partially inhibited by L-Fuc or Gal (Fig. 7). The most potent inhibitor was Gal which reduced liver targeting in a dose-dependent fashion to 3% when 0.3 mmol was used. L-Fuc was a weaker inhibitor of II, only reducing the targeting to 12% at 0.3 mmol (Fig. 7). The liver targeting of Le X triantennary III was also partially inhibited by Gal, resulting in 43% targeting at a dose of 0.3 mmol of Gal (Fig. 7), but was not affected by co-administration with Man or L-Fuc. This is in contrast to triantennary V, which was substantially inhibited to 13% by co-administration with 0.3 mmol of Gal but not with Man or L-Fuc (Fig. 7).  The structural reporter group signal for I and II were assigned by comparison with a previous report (31). The assignments made to III are tentative. See Fig. 2 for structure and residue nomenclature and Table I for chemical shift data.
To further substantiate that Le X oligosaccharides underwent receptor-mediated internalization in mice, 0.3 mmol of L-Fuc or Gal was administered at 26-min postinjection of radioiodinated II or V. Pharmacokinetic analysis was used to monitor the reappearance of the radiolabeled ligand in the blood which established that neither oligosaccharide could be released from the cell surface receptor by monosaccharide ligand.
To confirm that the liver targeting efficiency was not a function of dose, the targeting efficiency was measured for oligosaccharides dosed in the range of 10 pmol to 100 nmol (Fig. 8). The results indicate that receptor saturation is reached in the range of 10 -30 nmol when dosing with either II, III, or V. This established that the targeting efficiency is insensitive to dose in the 10 -50-pmol range normally used. The results also suggest that the number of liver receptors that bind oligosaccharides II, III, and V are roughly equivalent. DISCUSSION The biodistribution pattern of complex oligosaccharides in mice can be used to search for the existence of new carbohydrate receptors and elucidate the carbohydrate binding specificity of known receptors (12,32). Utilizing structurally well defined N-linked oligosaccharides eliminates heterogeneity often encountered in glycoproteins or neoglycoproteins and allows the determination of pharmacokinetic and biodistribution parameters on oligosaccharides that resemble natural ligands found in animal glycoproteins. Thereby, comparison of the bioactivity of closely related oligosaccharides provides the opportunity to study structure activity relationships for N-linked oligosaccharides.
Enzymatic remodeling of the outer antenna of N-linked oligosaccharides is a powerful approach to improve on the structural diversity of oligosaccharides obtained from common gly- FIG. 4. Pharmacokinetic analysis of Le X biantennary and triantennary oligosaccharides in mice. The concentration of oligosaccharide determined from direct ␥ counting of blood time points is plotted versus time for oligosaccharides I, II, and III. The fitted line is the nonlinear least squares fit of the data determined by PCNONLIN resulting in parameters reported in Table II (12). The right panels illustrate the autoradiograph of TLC analysis of each time point, demonstrating the lack of significant metabolites. The rapid clearance of oligosaccharide III is evident from both the direct counts and the TLC analysis.

TABLE II Pharmacokinetic analysis and targeting efficiency of N-linked oligosaccharides
Data derived from fitting three data sets to a two compartment open model using PCNONLIN as described previously (12). Targeting efficiency is defined as the percent of dose in the target organ at 30-min postadministration. Oligosaccharide c ND, not determined. d Data derived from a previous study (12). coprotein sources. The synthesis of multivalent Le X N-linked oligosaccharides reported here, and GalNAc-terminated oligosaccharides reported previously (12), are examples of the utility of this approach to produce rare or completely novel N-linked oligosaccharides. Although Le X biantennary oligosaccharides have been previously identified in animal glycoproteins (33,34), these represent rare sources which would not allow the isolation of micromole quantities of oligosaccharide necessary to perform detailed in vivo analysis. Likewise, Le X triantennary represents a novel structure which may exist in nature but most probably in quantities which would make its detection difficult. Tyrosinamide derivatization of N-linked oligosaccharides was developed as a means to produce glycoconjugates that could be used to study the pharmacokinetics and biodistribution of N-linked oligosaccharides (12,32). However, in addition to radioiodination, the tyrosinamide group aids both in providing enhanced chromatographic resolution on RP-HPLC and in allowing sensitive (100 -200 pmol) monitoring via fluorescence (35). In the present study, the fluorescence of tyrosinamideoligosaccharides allowed the optimization of the fucosyltransferase reaction using partially purified enzyme obtained from human milk (Fig. 1). This enzymatic assay is able to resolve partially fucosylated intermediates which may find utility when analyzing the enzyme kinetics of fucosyltransferases acting on multivalent oligosaccharides. Additional attributes of tyrosinamide-oligosaccharides, such as their reversibility to form reducing oligosaccharides and their utility in glycopeptide synthesis, have been reported recently (35,36).
This study examined the influence of terminal fucosylation on N-linked oligosaccharides with respect to target site in mice.

FIG. 5. Biodistribution of Le X biantennary and triantennary in mice.
Whole body autoradiography was used to determine the major target sites for oligosaccharide I, II, and III (see insets). Quantitative biodistribution was performed by direct ␥ counting of the dissected tissues (bars). The targeting efficiency (percent of dose in the target organ at 30-min postadministration) was 17% for I, 18.5% for II, and 66% for III. The kidney and small intestine were the only other organs with radioactivity in excess of 1% of dose. Each oligosaccharide was analyzed in triplicate.
FIG. 6. Elimination rate of biantennary oligosaccharides I and II from liver. The influence of core fucosylation on elimination from the target site was examined by comparing the percent of dose in liver (solid lines) and small intestine (dashed lines) for biantennary I (E, q) and II (Ⅺ, f). The results established that core fucosylation did not influence clearance rate as had been previously detected for GalNAc biantennary (12). However, the clearance rate is much slower (t1 ⁄2 ϭ 2.5 h) for Le X biantennary versus GalNAc biantennary oligosaccharide. Each data point was determined from dosing three different mice. The pharmacokinetic and biodistribution data both indicate that Le X biantennary and triantennary oligosaccharides are rapidly cleared from the blood (t1 ⁄2 ϭ 14 -17 min) and primarily target the liver (Fig. 5). These experiments do not discern which of the liver receptors are involved in binding of these ligands, since the hepatocyte ASGP-R and Kupffer cell fucose receptor of mouse liver both reportedly bind Le X -terminated neoglycoproteins to varying degrees (13)(14)(15)(16). But given that the number of mouse liver receptors that bind Le X or Gal triantennary and Le X biantennary is roughly equivalent (Fig. 8), and that both types of ligand are endocytosed by the liver and eliminated via the bile to the small intestine, it is assumed in the following discussion that the ASGP-R is primarily responsible for the liver binding activity being measured in mice.
The primary specificity of Le X oligosaccharide targeting to liver is related to terminal Gal residues since the targeting of both II and III were partially inhibitable by co-administration with Gal but not Man (Fig. 7). However, several criteria indicate that L-Fuc on I, II, and III is also recognized by the liver receptor. Oligosaccharide I and II each possess two Le X determinants resulting in a liver targeting efficiency of 17-18%, whereas Gal biantennary IV targets liver with only 7% efficiency (12) (Table II). Therefore, either L-Fuc influences the conformation of oligosaccharide I and II to enhance terminal Gal recognition or contributes to the binding affinity directly by interacting with the liver receptor.
The finding that the liver targeting of Le X biantennary II is partially inhibited by co-administration with L-Fuc supports the latter hypothesis. High concentrations (0.1-0.3 mmol) of L-Fuc were necessary to inhibit the targeting of II, since monosaccharides are weak ligands for mammalian lectins compared with multivalent oligosaccharides (20). Thereby, this result suggests that the multivalency of Le X determinants on biantennary contributes to the tight binding to the mouse liver receptor.
However, the location of L-Fuc in the oligosaccharide must also be important, because comparison of I and II indicates that the presence or absence of core fucosylation on Le X biantennary oligosaccharide results in the same degree of liver targeting activity in mice. Thus, the simultaneous presence of a terminal Gal and L-Fuc as exists in Le X appears to be responsible for the enhanced targeting efficiency of biantennary oli-gosaccharides. Although L-Fuc is clearly involved in the binding of Le X biantennary to the liver receptor, its contribution to the overall affinity is minor relative to Gal which was found to be a more potent antagonist.
These experiments are contrasted with those performed in rats which established that both Le X (II) and Gal (IV) terminated biantennary oligosaccharides target liver equally (5-6%) ( Table II). The results for Le X biantennary are therefore markedly different for mice and rats, which strongly suggests this is related to a specificity difference for the ASGP-R between these two species as suggested previously (13).
One of the most potent Gal-terminated ligands for the ASGP-R is triantennary oligosaccharide V, which has a measured K d of 4 nM for binding to rat hepatocytes and a targeting efficiency of 59% for mouse liver (12,37). Le X triantennary III differs from V by the addition of three terminal L-Fuc residues. Surprisingly, the liver targeting of oligosaccharide III is only 7% higher than V despite this major structural difference. Therefore, compared with the liver targeting activity of a Gal-NAc-terminated triantennary (85%), III displays only a modest enhancement in receptor affinity over V. This may arise from the failure of all three Fuc residues to participate in binding to the receptor when presented as Le X triantennary III.
Le X (III) and Gal (V) terminated triantennary display a nearly indistinguishable (66 Ϯ 4% versus 59 Ϯ 4%) targeting efficiency in mice. However, when co-administered with 0.3 mmol of Gal, the liver targeting of III is inhibited to 43%, while V was inhibited to 13% (Fig. 8). While it is assumed that triantennary V is a selective ligand for the ASGP-R, the L-Fuc residues on Le X triantennary III cause this ligand to retain high affinity for the liver, even when administered with 0.3 mmol of Gal. Furthermore, even the highest dose (0.3 mmol) of L-Fuc was incapable of causing inhibition of the liver targeting of III. This result could be due to the inability of monovalent L-Fuc to compete for receptor binding with an oligosaccharides possessing multiple (three Fuc and Gal residues) determinants. Therefore, the partial inhibition observed for 0.3 mmol of Gal reflects an enhanced receptor affinity for III, which most likely occurs as the result of binding of some of the L-Fuc residues on Le X triantennary.
The presence or absence of a core Fuc did not affect the elimination rate for Le X biantennary I and II from the liver. This could relate to the finding that Le X oligosaccharides are refractory to digestion with bovine testes ␤-galactosidase. Thereby, metabolism may depend on ␣-fucosidase to act first before exoglycosidase trimming can proceed. If this enzyme is rate-limiting, it could explain why the liver elimination rate of biantennary I and II were identical and much slower than GalNAc biantennary oligosaccharides as reported previously (12). Nonetheless, the finding that Le X biantennary II appears in the bile and could not be displaced from the liver by postadministration at 26 min with either 0.3 mmol of Gal or L-Fuc suggests this ligand is internalized into its target cell, which is a property associated with the ASGP-R but not the fucose receptor (14 -16).
The dose is an important parameter that must be considered when designing experiments that attempt to reveal lectin activity in animals (12,36). Liver targeting efficiencies in mice were constant up to a dose of 5 nmol. At doses of 10 -100 nmol the percent of dose bound to the liver decreased. Apparently this occurs as the receptor becomes saturated, allowing the excess ligand to escape from the circulation by rapid renal filtration. This suggests that the number of receptors that bind oligosaccharides III and V are nearly identical and further implicates the involvement of the ASGP-R, since V is a potent and selective ligand for this receptor system. However, it is also FIG. 8. Influence of dose on targeting efficiency. Oligosaccharides II (q), III (Ç), and V (f) were dosed in mice via vial jugular vein injection in a constant volume of 50 l containing 10 pmol to 100 nmol of oligosaccharide (1.8 Ci). At 30 min the liver was removed and the amount of radioactivity determined by direct ␥ counting. Receptor saturation was only evident at doses of 10 nmol or greater. evident that while the binding affinity and targeting efficiency of biantennary II is less than either III or V, the dose that saturates the receptor is nearly equivalent (Fig. 8). This evidence supports the proposal that Le X biantennary I and II both target the mouse ASGP-R.
The present study has supplied new information on the biodistribution of Le X containing N-linked oligosaccharides. Differences in the targeting efficiency for biantennary ligands in mice and rat liver point to subtle differences in receptor specificity. It will be important to determine if the human ASGP-R also preferentially binds Le X -over Gal-terminated N-linked oligosaccharides and if so how L-Fuc orients in the binding pocket.