Chemical and biological characterization of a polysaccharide biological response modifier from Aloe vera L. var. chinensis (Haw.) Berg.

Three purified polysaccharide fractions designated as PAC-I, PAC-II, and PAC-III were prepared from Aloe vera L. var. chinensis (Haw.) Berg. by membrane fractionation and gel filtration HPLC. The polysaccharide fractions had molecular weights of 10,000 kDa, 1300 kDa, and 470 kDa, respectively. The major sugar residue in the polysaccharide fractions is mannose, which was found to be 91.5% in PAC-I, 87.9% in PAC-II, and 53.7% in PAC-III. The protein contents in the polysaccharide fractions was undetectable. NMR study of PAC-I and PAC-II demonstrated the polysaccharides shared the same structure. The main skeletons of PAC-I and PAC-II are beta-(1-->4)-D linked mannose with acetylation at C-6 of manopyranosyl. The polysaccharide fractions stimulated peritoneal macrophages, splenic T and B cell proliferation, and activated these cells to secrete TNF-alpha, IL-1 beta, INF-gamma, IL-2, and IL-6. The polysaccharides were nontoxic and exhibited potent indirect antitumor response in murine model. PAC-I, which had the highest mannose content and molecular weight, was found to be the most potent biological response modifier of the three fractions. Our results suggested that the potency of aloe polysaccharide fraction increases as mannose content and molecular weight of the polysaccharide fraction increase.

The polysaccharides may have phosphate, acetyl group, carboxylic group, peptide, and lipid attached. The sugar composition of the polysaccharides might contain variable amounts of glucose, mannose, galactose, arabinose, and xylose. The linkages of the polysaccharides can be a-, b-, (132), (133), (134), or (136)-linkages. The polysaccharides are thus very diverse in their components and linkages except those derived from fungi and aloe. In fungi, the majority polysaccharide BRMs are b-D-glucans and those from yeast are a-D-mannan and glucomannan. Polysaccharide BRMs derived from aloe are mainly mannose polymer with b-(134)-D linkage.
Although a lot of polysaccharide BRMs have been reported, there is little information regarding to the structure and potency relationship of the polysaccharides. Such relationship study is hindered by the very complex sugar compositions and linkages of the polysaccharides. However, aloe polysaccharides having simple sugar composition and linkage are ideal models for the studies of the structure and potency relationship. The simplicity of the polysaccharides makes the structure determination feasible, and then the studies between structure and potency can be performed. In this study, three purified polysaccharide fractions were prepared from Aloe vera L. var. chinensis (Haw.) Berg. and were characterized chemically and biologically to delineate the effect of sugar content and molecular weight on the potency of the polysaccharide fractions.

Purification of polysaccharides in PAC50
Polysaccharides in PAC50 (see Materials and methods) were further separated by gel filtration. PAC50 was resolved into five distinct peaks by gel filtration high-performance liquid chromatography (HPLC) ( Figure 1A). The yields of the fractions, calculated according to their polysaccharides, were 60%, 10%, 5%, 16%, and 9%. According to dextran standards, the molecular weight (MW) of PAC-I, PAC-II, PAC-III, PAC-IV, and PAC-V were 10,000 kDa, 1300 kDa, 470 kDa, 20 kDa, and 10 kDa. The purity of HPLC-separated polysaccharides were monitored by capillary electrophoresis (CE). There existed only one major peak in each capillary electropherogram of PAC-I, PAC-II, and PAC-III ( Figure 1B), and each peak was close to symmetry.

Composition analysis
The purified polysaccharide fractions contained nearly exclusively carbohydrate. The percentage of carbohydrate in PAC-I, 98.07,and 95.63. The protein content was undetectable in these polysaccharides. PAC-I and PAC-II shared similar monosaccharide composition and had about 90% mannose, 5±6% galactose, some glucose, and some arabinose. PAC-III had about 54% mannose, 33% galactose, 13% arabinose, and trace amounts of glucose ( Figure 2).

NMR spectroscopy
The 13 C nuclear magnetic resonance (NMR) and 1 H NMR spectra of PAC-I and PAC-II were basically indistinguishable ( Figure 3 and Figure 4). The 13 C NMR of PAC-I and PAC-II were found to be very similar to that of galactomannan. The NMR spectra of methyl glycosides and legume-seed galactomannans were used as references (Grasdalen and Painter, 1980). The carbon resonance of monosaccharide unit in polysaccharide is almost same as the corresponding free monosaccharide with the same configuration and substitution pattern. Hydroxyl group of C-4 of mannopyranosyl in PAC-I is glycosylated as there is no resonance of corresponding unglycosylated C-4 (66±68 ppm). In fact, the chemical shifts of C-4 and C-5 in PAC-I accorded with that of the galactomannan NMR references. C-6 of mannopyranosyl in PAC-I resonated at 64.5 ppm and 62.5 ppm. The hydroxyl group of C-6 in PAC-I was not glycosylated as glycosylated C-6 has chemical shift of 69.6 ppm. However, C-6 in PAC-I might be acetylated as acetylation caused the corresponding carbon resonance downfield by 1.8 ppm. The presence of acetyl group was confirmed by proton NMR (Figure 4) and infrared spectrum (unpublished data). Chemical shift in the anomeric region indicated that anomeric configuration was beta for mannopyranosyl. C-1 of mannopyranosyl in PAC-I had chemical shift of 102.2 ppm, which was similar to that of galactomannan. The 13 C NMR and 1 H NMR spectra of PAC-III were too complicated for structural interpretation ( Figure 3C and Figure 4C).
In vitro and in vivo splenic lymphocytes proliferation All the polysaccharide fractions stimulated the proliferation of both splenic T and B cells in vitro. The stimulation demonstrated a dose-dependent pattern ( Figure 5A and 5B). The polysaccharide fractions were potent B cell stimulators and were less potent in T cell proliferation stimulation. PAC-I was the strongest stimulator and was followed by PAC-II and then PAC-III. The polysaccharide fractions when injected IP stimulated splenic lymphocyte proliferation 3 days after injection. PAC-I was the strongest stimulator and then PAC-II. PAC-III was the weakest ( Figure 5C). The results accorded with in vitro observations. Endotoxin is a cell surface component of Gram-negative bacteria and is a potent B-lymphocyte stimulator. The stimulatory effect of PAC polysaccharides that was due to the contamination of LPS (endotoxin) was excluded by Limulus amebocyte lysate (LAL) test. The coagulation result of LAL test was negative and indicated that there was undetectable amount of endotoxin in PAC-I, PAC-II, and PAC-III preparation. PAC-I was the largest peak and represented 60% of PAC50. PAC-II represented 10% and PAC-III represented 5%. The molecular weights of PAC-I, PAC-II, and PAC-III were 10,000 kDa, 1300 kDa, and 470 kDa according to dextran standards. (B) Capillary electropherograms of PAC-I, PAC-II, and PAC-III. Sample of 20 ml at 200 mg/ml eluant (H 3 BO 3 -KOH, pH 10) was electrophoresed at 200 V. Polysaccharide was detected by monitoring at optical density 254. Only one major peak was observed in each sample, and the peak was close to symmetry. M.Y.K. Leung et al.

Elicitation of cytokines secretion
The polysaccharide fractions stimulated peritoneal macrophage to secrete interleukin (IL)-1b and tumor necrosis factor (TNF)-a, and stimulated splenic lymphocytes to secrete IL-2, IL-6, and interferon (INF)-g. The stimulation followed a dose-dependent pattern and stimulation saturated at 500 mg/ml polysaccharide concentration. Cytokine level in culture medium of lymphocyte treated with polysaccharide at 500 mg/ml were tabulated in Table I. The polysaccharides when injected IV into mice triggered substantial secretion of serum TNF-a (Table I).
In vivo antitumor activity Only PAC-I and PAC50 mediated significant regression of Sc-180 tumor in vivo. PAC-I caused a 58% reduction in tumor size and PAC50 caused a 50% reduction. PAC-II and PAC-III were unable to cause significant tumor regression ( Figure 6). However, all the polysaccharide caused no significant regression of Hep-G2 tumor in nude mice (unpublished data).

Discussion
The molecular weight of PAC-IV and PAC-V determined by HPLC were lower than the specified molecular weight cutoff (MWCO) of the membrane (50 kDa) used for the preparation of PAC50. The specified MWCO of the membrane was for globular protein but not thread-like polysaccharide. The MWCO of the same membrane for polysaccharide was lower. Furthermore, the column used had very large fractionation range (10±1,000 kDa), so that the estimation of molecular weight of small molecules might not be accurate.
CE results of PAC-I, PAC-II, and PAC-III suggested the purified polysaccharide fractions might still contain some impurities. However, the purity was good enough for chemical and biological study. Protein or peptide, which can be immunostimulatory, was undetectable in PAC-I, PAC-II, and PAC-III. The result accorded with elemental analysis. Elemental analysis demonstrated that there existed no nitrogen in PAC-I, PAC-II, and PAC-III (unpublished data). Methylation analysis of the polysaccharides had been attempted. However the sizes of the polysaccharide fractions were too large for all the free hydroxyl groups to be methylated completely. The 13 C NMR spectrum of PAC-I was found to be very similar to that of legume-seed galactomannan (Grasdalen and Painter, 1980). Legume-seed galactomannan has manopyranosides in the main skeleton linked together by b-(1-4)-D-linkage and galactopyranoside attached at C-6 of the skeletal manopyranosides.
PAC-II, which had nearly the same composition and NMR profiles as PAC-I, had the same main skeletal structure as PAC-I but smaller in size. PAC-I had molecular weight 7.8 times that of PAC-II. The deduced structure of the main linkage of PAC-I and PAC-II is illustrated in Figure 7. A. vera polysaccharide characterization complicated than PAC-I and PAC-II. The possibility of impurities in PAC-III that contributed to the complex NMR spectra of PAC-III was rather low. CE results indicated that the purity of PAC-III was comparable to that of PAC-I and PAC-II.
All the PAC polysaccharide fractions tested were potent immunostimulators or BRMs. Immunomodulatory activities of the polysaccharide fractions, which were due to the contamination of endotoxin, was excluded by the result of the LAL test. The polysaccharide fractions stimulated peritoneal macrophage and splenic T and B cell proliferation, and activated these cells to secrete TNF-a, IL-1b, INF-g, IL-2, and IL-6. We focused on these cytokines because they are commonly used in tumor immunotherapy (Iwashita et al., 2003;Kuninaka et al., 2000;Rhines et al., 2003;Terlikowski, 2002).
Both PAC-I and PAC50 mediated a significant regression of Sc-180 in vivo ( Figure 6). The in vivo tumor regression properties of the aloe BRMs might be due to direct toxicity of the polysaccharide fractions, tumor-sensitive toxic metabolic derivatives of the polysaccharide fractions, or immunostimulatory properties of the polysaccharide fractions. The direct toxicity of the polysaccharide fractions was excluded because antitumor effect was only observed in vivo but not in vitro (unpublished data). The polysaccharide fractions mediated no tumor regression in nude mice, Fig. 3. 13 C NMR spectra of PAC-I (A), PAC-II (B), and PAC-III (C). Polysaccharide was dissolved in deuteriated water at 50 mg/ml. The NMR spectra were recorded by INOVA-600. The chemical shifts (d) of C-1, C-2, C-3, C-4, C-5, and C-6 were 102.2, 72.1, 73.5, 78.5, 77.7, 62.5, and 64.5 ppm. The spectra of PAC-I and PAC-II were nearly the same. The NMR spectrum of PAC-III was too complicated for structural interpretation.
M.Y.K. Leung et al. which suggested that aloe polysaccharides were not metabolized to tumor-sensitive toxic substances. Therefore, aloe BRMs must mediate tumor regression in vivo by the activation of host immune effector mechanisms.
Similar to PAC-I and PAC-II, b-(133)-D-glucans have simple monosaccharide composition and linkage. The in vivo antitumor properties of fungal b-(133)-D-glucans are initiated by the binding of the glucans to b-glucan receptor, such as dectin-1 (Brown and Gordon, 2001;Taylor et al., 2002), on immune cells. The target receptor for aloe BRMs might be mannose receptor (MR) on macrophages and dendritic cells. Aloe polysaccharides might bind to MR of dendritic cells and macrophages and lead to activation of immunity. The stimulation of B cells by PAC polysaccharides might through another mechanism, because the presence of MR on B cells was not reported. The potent B cell stimulatory effect of PAC polysaccharides in vitro might be related to its highly repeating structures, which cross-link surface IgM of polysaccharide-responding B cells in a multivalent fashion. The cross-linking activates the B cells to proliferate (Anderson and Blomgren, 1971;Dorries et al., 1974).
The mechanism of activation of immunity by b-glucan is well established. Lesson from b-glucan study may provide clues to the immune activation by mannose-rich polysaccharide at a molecular level. Dectin-1 is a pattern , and PAC-III (C). Polysaccharide was dissolved in deuteriated water at 50 mg/ml. The NMR spectra were recorded by INOVA-600. Resonance at 2.21 ppm was very strong and indicated the presence of an acetyl group. The spectra of PAC-I and PAC-II were nearly the same. The NMR spectrum of PAC-III was too complicated for structural interpretation.
A. vera polysaccharide characterization recognition receptor. Dectin-1 cooperates with Toll-like receptors (TLRs), and many other surface receptors for the recognition of different microbial products such as fungal cell wall, lipopolysaccaharide, lipoprotein, flagellin, and bacterial DNA (Gantner et al., 2003;Underhill, 2003). The repertoire of receptors that recognize microbial products determines the immune effector mechanism to be activated. Zymosan particle, which is composed mainly of b-glucan, mannan, mannoproteins, and chitin, is recognized by dectin-1, TLRs, and CD14. The receptors work together to enhance immune response to the particles. Dectin-1 can trigger phagocytosis and reactive oxygen production directly. TLRs induce signaling through NF-kB results in the secretion of inflammatory cytokines. The TLR induction is enhanced by dectin-1. Mannose-rich polysaccharide may also collaborate with TLRs for the activation of immunity.
Altogether, mannan from aloe has a simple composition, and the linkage is an excellent model for the study of structure and activity relationship. PAC-I and PAC-II prepared from A. vera L. var. chinensis (Haw.) Berg. has potent BRMs. The polysaccharide fractions mediated indirect tumor regression in vivo through the activation of antitumor effector cells and the secretion of antitumor cytokines. The skeletons of the polysaccharide fractions were b-(134)-Dlinked mannose with O-acetylation at C-6. High molecular weight and high mannose content may be essential for high antitumor potency of aloe polysaccharide.

Materials and methods
Fractionation of A. vera L. var. chinensis (Haw.) Berg. A. vera L. var. chinensis (Haw.) Berg., a variant of A. vera (barbadensis Miller), was supplied by South China Institute of Botany, Chinese Academy of Sciences, China. Fleshy layers of aloe leaves were collected and blended to viscous watery gel. Insoluble fibers were removed by centrifugation (865 Â g for 10 min). The viscous supernatant was allowed to incubate at 4 C overnight and then centrifuged at 5000 Â g for 20 min to remove precipitate. The supernatant was loaded onto a column of XAD-4 resin (Rohm and Haas, Philadelphia, PA) with dimension of 5 cm Â 15 cm. Collected eluate, which contained the crude polysaccharide, was fractionated by membrane with MWCO of 50 kDa. Membrane-retained fraction was Fig. 5. The stimulatory effect of PAC polysaccharides on the proliferation of murine splenic lymphocytes. PAC polysaccharides stimulated the proliferation of murine splenic T lymphocytes (A) and B lymphocytes (B) in a dose-dependent manner in vitro. PAC-I was the most potent stimulator for both T and B cell stimulation. Stimulation due to PAC-I was significantly higher than that due to PAC-II and PAC-III. PAC-II and PAC-III had similar potency on T cell stimulation. However PAC-II was more potent than PAC-III on B cell stimulation. LPS of 5 mg/ml was used as positive control in B lymphocyte proliferation and Conacanavalin A of 1 mg/ml was used as positive control in T lymphocyte proliferation. The simulation index of LPS was 133 AE 8, and the simulation index of Conacanavalin A was 283 AE 45. PAC polysaccharides exhibited mitogenic activity on murine splenic lymphocytes in vivo (C). Female albino BALB/c mice were injected IP with PBS, PAC50, PAC-I, PAC-II, or PAC-III. Splenic lymphocytes were prepared from treated mice 3 days after injection. PAC-I was a stronger stimulator than PAC-II, followed by PAC-III. An asterisk indicates lymphocyte proliferation was significantly higher than that of PBS-treated control. SI ratio is calculated by dividing the radioactive count of sample-treated lymphocytes with radioactive count of mediumtreated lymphocytes. Values are means AE SD (n 5 for each group). M.Y.K. Leung et al. concentrated with a thin film evaporator LZ-005 at 70 C (Xia Mei Yi Liao Jixie Chang, China) and then lyophilized. The fraction was designated PAC50.

Protein content determination
Bradford micro-protein assay (Bradford, 1976) was used to determine protein content. Bovine serum albumin (Sigma, St. Louis, MO) was used as protein standard. Filtered Bradford reagent (BioRad, Hercules, CA) (0.2 ml) was reacted with 0.8 ml sample at 1 mg/ml at room temperature for 15 min. Absorbance of the reaction mixture was measured at optical density 595.

Carbohydrate content determination
Dubois method (Dubois et al., 1956) was used to measure carbohydrate content. D-mannose was used as standard. A sample of 400 ml at 500 mg/ml was mixed with 200 ml 6% phenol solution and 1 ml concentrated sulfuric acid. The mixture was reacted for 20 min at room temperature. Absorbance of the reaction mixture was measured at optical density 490.

HPLC of PAC50
PAC50 was dissolved in 0.1 M NaCl at 5 mg/ml. Five milliliters of PAC50 were loaded onto a Gel filtration column (Jordi Aqueous GPC Glucose Bound 10,000 s 5 U SS column; Alltech, Nicholasville, KY) with dimensions of 22 mm Â 500 mm. NaCl (0.1 M) was the mobile phase, and the flow rate was maintained at 3 ml/min. Dextrans of different molecular weights were used as standard. The eluate was collected as 3-ml fractions. The carbohydrate content in each fraction was determined by Dubois's method.

CE
Honda's method (Honda et al., 1989) was used to determine the purity of column separated polysaccharides. Sample of 20 ml at 200 mg/ml eluant (H 3 BO 3 -KOH, pH 10) was applied to Quanta 400E (Waters, Milford, CT) CE system All the PAC polysaccharide fractions induced significant secretion of cytokines when compared with medium control. TNF-a is the serum TNF-a level determined at 1.5 h after IV injection of PBS or PAC polysaccharides. Concanavalin A (ConA) and LPS were used as positive controls. NA, undetectable. Sarcoma 180 cells (10 7 cells per ml PBS of 50 ml) were injected SC into the right groin of male ICR albino mice. Mice were then injected IP with PAC polysaccharides (1 mg/ml, 0.2 ml) or PBS for 10 consecutive days. PBS was used as a negative control. On day 21, tumor weight was determined. An asterisk indicates that tumor growth was suppressed significantly when compared with PBS-treated control. Values are means AE SD (n 10 for each group). A. vera polysaccharide characterization and electrophoresed at 20 kV. Polysaccharide was detected by measuring at optical density 254.

Monosaccharide composition analysis
Monosaccharide composition analysis followed the method of Wozniewski et al. (1990). A polysaccharide fraction of 3 mg was hydrolyzed with 10 ml 2 M aqueous trifluoroacetic acid in a sealed tube at 120 C for 1 h. The hydrolysate was concentrated with rotary evaporator under reduced pressure at 60 C, and remaining traces of trifluoracetic acid was removed by coevaporating with methanol (3 Â 10 ml). The concentrated hydrolysate was dried over P 2 O 5 in vacuum overnight. The dried hydrolysate was dissolved in 5 ml dH 2 O, and 10 mg sodium borohydride (NaBH 4 ) was then added. The mixture was stirred at room temperature for 18 h. Excess NaBH 4 was removed by acetic acid. The mixture was then concentrated and dried under reduced pressure. Dried residue was dissolved in 6 ml anhydrous pyridine. Acetic anhydride (6 ml) was added, and the mixture was stirred overnight at room temperature. After incubation, the mixture was coevaporated with toluene (3 Â 10 ml) at 55 C under reduced pressure and finally dried in vacuum. Dried residue was dissolved in minimal amount of chloroform for gas chromatography±mass spectrometry (GC-MS) analysis with a GC analyzer (Varian VISTA 6000, Palo Alto, CA). The GC column was OV-17 with dimension of 0.2 mm Â 300 mm (Agilent Technologies, Wilmington, DE). The injector and column temperature was 250 C.

C NMR and 1 H NMR spectroscopy
Samples were dissolved in deuteriated water at 50 mg/ml for both 13 C NMR and 1 H NMR (Agrawal, 1992). 1 H NMR spectra were recorded with an INOVA-600 spectroscopy at 30 C in D 2 O, and chemical shifts were referred to HOD. 13 C NMR spectra were recorded with an INOVA-600 spectroscopy at 30 C in D 2 O, and chemical shifts were referred to Me 4 Si.

LAL coagulation test
The LAL test is commonly employed to determine the presence of LPS or endotoxin (lipopolysaccharide) in herbal preparation. The Pyrotell (Associates of Cape Cod) testing reagent was used according to instructions. Pyrogen-free water and LPS (0.05 ng/ml) from Escherichia coli of serotype 0127:B8 (Sigma) were used as controls.
In vitro lymphocyte blast transformation assay Splenic lymphocytes were prepared from female albino BALB/c mice by Ficoll-Plaque centrifugation. Three spleens were ground with a sterilized sieve and a 2.5-ml syringe plunger in 10 ml RPMI-1640 medium (Invitrogen, Carlsbad, CA). The ground cell suspension was transferred to a 50-ml centrifuge tube, and volume was adjusted to 20 ml with medium. Equal volume of Ficoll was then added to the bottom of the cell suspension. The tube was centrifuged at 2200 Â g for 20 min. Cells at interface were splenic lymphocytes and were collected. T cells were prepared by B cells depleting column (Cedarlane Labs, Hornby, Ontario, Canada), and B cells were prepared by T cells depleting column (Cedarlane Labs, Carlsbad, CA). Prepared T and B cells were resuspended in medium supplemented with 10% fetal calf serum (Invitrogen, Carlsbad, CA) at 5 Â 10 6 cells/ml. Cell suspension of 100 ml was distributed per well of a 96-well plate. Cells were incubated with filter-sterilized polysaccharide at 37 C for 48 h. Cell proliferation was monitored by thymidine incorporation.
In vivo lymphocyte blast transformation assay Polysaccharide fraction of 0.2 ml at 5 mg/ml phosphate buffered saline (PBS) was injected IP into female albino BALB/c mice. Spleens were obtained from mice on 3 days after sample administration. Splenic lymphocytes were prepared as described. Cell suspension of 100 ml at 5 Â 10 6 cells/ml was distributed per well of a 96-well plate. Cell proliferation was monitored by thymidine incorporation at 48 h after incubation.
In vitro IL-2, IL-6, and INF-g secretion Splenic lymphocytes were prepared from female albino BALB/c mice. Cell suspension of 100 ml at 5 Â 10 6 cells/ml concentration was plated per well of a 96-well plate. Cells were incubated with filter-sterilized polysaccharide at 37 C for 48 h. After the incubation, 100 ml medium was removed from each well, and the level of IL-2, IL-6, and INF-g in removed medium was assayed by enzyme-linked immunoasorbent assay (ELISA) kit (Pharmingen, San Diego, CA).
In vitro IL-1b and TNF-a secretion Thioglycollate broth (Difco Laboratories, Detroit, MI) of 1.5 ml at 0.3% concentration was injected IP per male BALB/c mouse. Three days after injection, the peritoneal cavity was washed with 20 ml PBS. Cells in the peritoneal wash were washed twice with RPMI-1640 medium. Cell concentration was adjusted to 10 8 cells/ml medium. Cell suspension of 100 ml was added per well of a 96-well plate. The cells were incubated at 37 C for 3 h. After incubation, nonadherent cells were removed by washing. The adherent cells, which were peritoneal macrophages, were incubated with filter-sterilized polysaccharide at 37 C for 48 h. After the incubation, 100 ml medium was removed from each well. The level of IL-1b and the level of TNF-a in removed media were assayed by murine IL-1b and TNF-a ELISA kit (Pharmingen), respectively.
In vivo TNF-a secretion Polysaccharide of 0.2 ml at 500 mg/ml was injected IV per male BALB/c mouse. LPS at 10 mg/ml was used as positive control. Blood was collected 1.5 h after polysaccharide administration by heart puncture. Collected blood was clotted at room temperature and then centrifuged at 4000 Â g for 30 min. Serum was collected for TNF-a measurement by murine TNF-a ELISA kit (Pharmingen).
In vivo antitumor assay Sarcoma 180 (Sc-180) cells (ATCC) of 50 ml at a cell concentration of 10 7 cells per ml PBS were injected SC into the right groin of male ICR albino mice (10 mice/group) on day 1. Mice were then injected IP with 0.2 ml polysaccharide M.Y.K. Leung et al. fraction at 1 mg/ml PBS or 0.2 ml PBS (PBS as negative control) for 10 consecutive days from day 1. On day 21, mice were euthanized, and solid tumors were excised for weight determination.

Statistical analysis
Student's t-test was used for the statistical analysis of results at a level of significance of p 5 0.05.