Structure Activity Relationships of N-linked and Diglycosylated Glucosamine-Based Antitumor Glycerolipids

1-O-Hexadecyl-2-O-methyl-3-O-(2'-amino-2'-deoxy-β-d-glucopyranosyl)-sn-glycerol (1) was previously reported to show potent in vitro antitumor activity on a range of cancer cell lines derived from breast, pancreas and prostate cancer. This compound was not toxic to mice and was inactive against breast tumor xenografts in mice. This inactivity was attributed to hydrolysis of the glycosidic linkage by glycosidases. Here three N-linked (glycosylamide) analogs 2–4, one triazole-linked analog 5 of 1 as well as two diglycosylated analogs 6 and 7 with different stereochemistry at the C2-position of the glycerol moiety were synthesized and their antitumor activity against breast (JIMT-1, BT-474, MDA-MB-231), pancreas (MiaPaCa2) and prostrate (DU145, PC3) cancer cell lines was determined. The diglycosylated analogs 1-O-hexadecyl-2(R)-, 3-O-di-(2'-amino-2'-deoxy-β-d-glucopyranosyl)-sn-glycerol (7) and the 1:1 diastereomeric mixture of 1-O-hexadecyl-2(R/S), 3-O-di-(2'-amino-2'-deoxy-β-d-glucopyranosyl)-sn-glycerol (6) showed the most potent cytotoxic activity at CC50 values of 17.5 µM against PC3 cell lines. The replacement of the O-glycosidic linkage by a glycosylamide or a glycosyltriazole linkage showed little or no activity at highest concentration tested (30 µM), whereas the replacement of the glycerol moiety by triazole resulted in CC50 values in the range of 20 to 30 µM. In conclusion, the replacement of the O-glycosidic linkage by an N-glycosidic linkage or triazole-linkage resulted in about a two to three fold loss in activity, whereas the replacement of the methoxy group on the glycerol backbone by a second glucosamine moiety did not improve the activity. The stereochemistry at the C2-position of the glycero backbone has minimal effect on the anticancer activities of these diglycosylated analogs.

. Structures of the synthesized glycolipids used in the study. Compound 1 is the reference GAEL while compounds 2-5 are glycolipid analogs differing in the nature of the glycosidic linkage and glycerol moiety. Compounds 6 and 7 are glycolipid analogs where the glycerolipid moiety contains two glycosidic linkages. Edelfosine is shown as a prototypic AEL analog.

Chemistry
In order to prepare easily accessible analogs with improved metabolic stability towards glycosidases we initially were interested in analogs that contain a glycosylamide linkage. Glycosylamides 2-4 were synthesized to evaluate the effect of amide linkage and nature of the lipid moiety on the antitumor properties. Compound 3 which lacks the methoxy substituent was prepared to explore how the methoxy group affects the antitumor properties. Compound 4 that contains a lipophobic polyfluorinated lipid tail instead of a lipophilic carbon chain was prepared to explore how modifications in the lipid tail affect the antitumor activity. Compound 5 was synthesized to evaluate the effect of a triazole linkage at the anomeric position. Both linkages, the glycosylamide and glycosyltriazole linkages are expected to be metabolically stable to hydrolysis by glycosidases in in vivo studies [11][12][13]. In addition, the glycosyltriazole linkage will be inert towards peptidases and proteases which may provide additional benefits for future in vivo studies [16].
GAEL mimetics 2-4 were prepared by coupling of glycosylamine 11 to carboxylic acids 13, 15 and 16 (Scheme 1). Glycosylamine 11 was synthesized in three steps from glucosamine hydrochloride (8) in 49% yield [17]. The amino substituent at C 2 position of glucosamine hydrochloride group was protected with phthalic anhydride followed by protection of the hydroxyl groups as acetate esters by reaction with acetic anhydride in pyridine to afford compound 9. The anomeric amino group in 11 was installed through conversion of 9 into the corresponding anomeric azide 10. Originally, we planned to introduce the anomeric azido group by nucleophilic displacement of the α-anomeric chloride. However conversion of the anomeric acetate into the chloride by reaction with PCl 5 did not afford the corresponding glucosylhalides [18]. However, the anomeric azide 10 was prepared by Fe(III) chloridepromoted reaction of 9 with trimethylsilyl azide to afford azide 10 in 75% yield. The anomeric azide was reduced to the amine by catalytic hydrogenation to provide glucosylamine 11 in 96% yield. Synthesis of the fatty acid compounds 13 and 15 was achieved by Jones oxidation of the corresponding commercially available alcohol 12 or known alcohol 14 [19]. The polyfluorinated fatty acid 16 was purchased from commercial source. Coupling of fatty acids 13, 15 and 16 to the glucosylamine 11 was achieved by using TBTU [20] as coupling agent to afford the protected glycolipids 17, 18 and 19, respectively. The acetate and phthalimido protecting groups were removed using an ethylenediaminebutanol mixture (1:1) at 90 °C for 3 h to afford desired target compounds 2, 3 and 4 respectively. The triazole analog 5 was synthesized by Cu(I)-promoted click chemistry [21] using azide 10 and 3-hexadecyloxyprop-1-yne (22) to produce glycosyltriazole 23. Deblocking using the same method as described above gave compound 5. Compounds 6 and 7 were synthesized to study the effects of diglycosylation and stereochemistry of the glycerol moiety on the antitumor properties. Glycolipids 6 and 7 were prepared from known thioglycoside donor 24 [22] by glycosylation with commercially lipid alcohol 25 or racemic alcohol 26 using silver triflate as catalyst and N-iodosuccinimde as promoter to afford protected glycolipids 27 or 28, respectively (Scheme 2). The acetate and phthalimido protective groups were removed using ethylenediamine:butanol mixture (1:1) at 90 °C for 3 h to provide compounds 6 and 7, respectively.

Cytotoxicity
The cytotoxicity of compounds 1-7 was evaluated against a number of epithelial cancer cell lines using the MTS assay [22]. The cells were derived from cancers of the breast (JIMT-1, BT-474, MDA-MB-231), pancreas (MiaPaCa2), and prostrate (DU145, PC3). The cytotoxic effect of compounds 2-6 were compared with that of 1 the most studied glycosylated antitumor ether lipid [5][6][7]9] (see Table 1). Exponentially growing cells were treated with test compound and then incubated for 48 h. The result for compounds 2-4 are shown in Figure 2, while that of compounds 5-7 are shown in Figure 3. The CC 50 values that lead to a reduction of cell viability by 50% relative to untreated control are reported in Table 1. At the highest concentration tested none of the N-linked glycolipids 2-4 was able to achieve 50% reduction in viability against all the four cell lines tested (see Figure 2). It is noteworthy that compound 2 with similar glycerolipid moiety as reference compound 1, is consistently more active than compounds 3 and 4, leading to a 19%-29% reduction in viability relative to untreated control at the highest dose tested (30 µM). Table 1. Cytotoxicity of compounds 1-7 on a panel of human epithelial cancer cell lines: breast (BT474, JIMT1, MDA-MB-231), prostrate (DU145, PC3), pancreas (MiaPaCa2). The CC 50 value is defined as the concentration required to decrease cell viability by 50% relative to the untreated control. Values were determined by MTS assay. The CC 50 values were obtained by estimating the drug concentration at 50% viability on the y axis using line plots (graph not shown); NT = not tested. Please note that CC 50 values for compound 1 were obtained from our previously published work [21].   At this dose and lower concentration, compound 4 with polyfluorinated lipid moiety is consistently the least active analogue among these N-linked (amide) glycolipids. This relative difference in activity is an indication that the nature of the lipids of this class of compounds plays an important role on their cytotoxicity against human cancer cell lines. The triazole-linked glycolipid compound 5 at concentration of 30 µM or below was unable to achieve CC 50 values against prostrate and pancreas cell lines used in this experiment, indicating that both the nature of the glycosidic linkage and the glycero moiety influences the antitumor properties. It is noteworthy that compound 5 was able to achieve more than 80% reduction in viability of JIMT1, a trastuzumab resistant cell line [4] at the dose of 30 µM, (CC 50 : 23µM) and more than 60% reduction in viability of BT474, a trastuzumab sensitive cell lines (see Figure 3). In vivo, triazole-linked glycolipid 5 is expected to be metabolically stable relative to lead compound 1, however the loss of activity does not make it a worthwhile compound for further development.
The diglycosylated analogues, compounds 6 and 7 with different stereochemistry at position 2 of the glycerol backbone have very similar activity but are significantly less active than GAEL-based reference compound 1 ( Figure 2 and Table 1) indicating that additional glucosamine moieties on the glycerol backbone reduce the cytotoxic effect of the compounds. This reduction in activity may be due to increased polarity of the compound which in turn can affect the absorption of the drug across cell membrane, and subsequent decreased drug concentration in the cell. The fact that both compounds despite the difference in stereochemistry at the C-2 position of the glycerol backbone have similar activity especially in the more sensitive prostrate cell lines ( Figure 2) is an indication that stereochemistry at this position has minimal to no effect on the anticancer activity.

General Methods
All fine chemicals such as glucosamine hydrochloride, phthalic anhydride, pyridine, 1-hexadecanol, 4-toluenesulfonyl chloride, 60% sodium hydride, dimethylformamide (DMF), 1,3-dihydroxypropane, sodium methoxide, concentrated sulfuric acid, palladium on carbon, trimethylsilyl azide, toluene, phosphorus pentachloride, boron trifluoride, N,N-diisopropyl-ethylamine, ethylenediamine and iron (III) chloride. All solvents such as dichloromethane (DCM), hexane, acetone, methanol, ethyl acetate, butanol were purchased from Sigma-Aldrich (Oakville, ON, Canada). Carboxylic acid 16 (2H,2H,3H,3H-perfluoroundecanoic acid) were purchased from Fluorous Technologies now available through Sigma-Aldrich. 1-O-hexadecyl-2-O-methyl-sn-glycerol was purchased from Chem-Implex (Wood Dale, IL, USA). 5 ¾ and 9 inch pipets used in lab were obtained from Fisher Scientific (Ottawa, ON, Canada). TLC plates (CCM Gel silica 60 F254) were visualized by ultraviolet light or by charring (9:1 methanol and sulfuric acid). All chromatography solvents were prepared by mixing hexane and ethyl acetate in various ratios based on the polarity of the compound. Column chromatography was performed by using silica gel p60 (40-63 µm) or reverse-phase C18 silica gel. Solutions were concentrated on reduced pressure rotary evaporator (IKA RV 10 Basic, that was connected to building vacuum and high vacuum pump (Welch 8907). 1 H-NMR and 13 C-NMR spectra were recorded on a 300 MHz (Bruker AMX-300 spectrometer). Low-resolution mass spectrometry (MS) data were obtained on a Varian 500-MS IT Mass spectrometer using electrospray ionization (ESI). (9). Glucosamine hydrochloride 8 (3.016 g, 14 mmol) and NaOH (28 mmol) were dissolved in water (50 mL). The resulting mixture was stirred at room temperature for 30 min. Phthalic anhydride (2.34 g, 0.0157 mol) was added to the solution. The mixture was stirred vigorously at room temperature for 16 h. The mixture was concentrated and dried using rotary evaporator. The residue was dissolved in pyridine (30 mL), and then Ac 2 O (19.8 mL) was added to the solution. The resulting solution was allowed to stir vigorously overnight. The reaction was checked by the TLC. Methanol (6 mL) was used to quench the excess of Ac 2 O, and then excess pyridine was removed under high vacuum. The remaining solid was dissolved in CH 2 Cl 2 (40 mL), and then the solution was washed with 10% HCl (40 mL×1), Saturated NaHCO 3 solution (40 mL×3), H 2 O (40 mL×1) and brine (40 mL×1) and dried over anhydrous MgSO 4 . The final solution was concentrated under reduced pressure, and the obtained product 8 (3.3g, 49.4%) was dried overnight. NMR data were consistent with data in the literature [23]. (10). Compound 9 (1.8 g, 8 mmol) and trimethylsilyl azide (1.7 g, 148 mmol) were both dissolved in CH 2 Cl 2 (20 mL) in a 100 mL-round bottom flask with vigorous stirring. Then FeCl 3 (1.77 g) was added to the reacting mixture. The reaction was allowed to stir for 24 h and then progress was checked by TLC. The dispersion solution was made by mixing hexane and ethyl acetate in 1:1 ratio. The solution was concentrated under reduced pressure by rotary evaporator. The product 10 was isolated and purified by column chromatography (1:2 ethyl acetate/hexane). The obtained product 10 (1.3 g, 76.53%) was a light yellow solid. NMR data were consistent with previously published data [16]. (11). Compound 10 (0.11 g, 0.24 mmol) was dissolved in methanol (2 mL) in a 25 mL-round bottom flask with vigorous stirring, and then Pd/C (0.29 g) was added. After that, round bottom flask was connected to a hydrogen balloon. The reaction was allowed to take place for half hour, and checked by TLC. The reaction was stopped when all starting material has disappeared, and the solution was concentrated to provide compound 11 (100 mg, 96%). This compound was not characterised by 1 H-NMR, because it is chemically unstable. Therefore it was directly used for the next step, i.e., coupling of carboxylic acids using TBTU as coupling reagent.  (15). A 2.7 M solution of Jones reagent (2.0 mL) was added dropwise into a stirred solution of compound 14 (0.124 g, 0.4 mmol) in acetone (20 mL) at 0 °C. The reaction was monitored by TLC and was completed after 1 h. Isopropyl alcohol was added dropwise until a stable green color appeared, to remove the excess of CrO 3 . The organic solvent was removed under reduced pressure. The solid was dissolved in water, and then compound 15 was extracted using ethyl acetate. The combined organic layers were washed by brine, dried using MgSO 4

General Procedure for the Synthesis of Diglycosylated Compounds 6 and 7
Compounds 6 and 7 were synthesized by diglycosylation of commercially available lipids 25 and racemic alcohol 26, respectively. The previously reported thioglycoside donor 24 [19] was synthesized from compound 9 and thiophenol using BF 3 . Et 2 O as promoter in DCM at 60 °C for 16 h. The glycosylation reaction was carried out using silver triflate and N-Iodosuccinimide in anhydrous DCM under argon atmosphere for 5 h. The reaction was stopped by addition of saturated sodium thiosulphate solution followed by washing with saturated NaHCO 3 solution (×3). The organic layer was concentrated under vacuum to give a brownish residue which was purified with flash chromatography using hexane/ethyl acetate mixture (6:4) to give protected diglycosylated glycolipids 27 and 28 as a white foam (yield 45%). Compounds 27 and 28 were subsequently deprotected using a 1:1 mixture of ethylenediamine and butanol for 4 h, followed by removal of solvent under vacuum and purified with Ethyl acetate/methanol mixture (7:3) to give compounds 6 and 7, respectively (yield 70%).

Cytotoxicity Assay
The cytotoxicity assay was carried out using a previously reported method [22]. Cell viability was determined with the cell Titre 96 AQueous One Solution (MTS assay, Promega, Madison, WI, USA. Equal numbers of cancer cells (7500-9500) in media (100 µL) were dispersed into 96-well plates. As blanks, media without cells (100 µL) were also placed in some wells and treated similarly to the cell containing wells. After an incubation period of 24 h, a solution of test compound ((100 µL) in medium at twice the desired concentration was added to each well. The treated cells were incubated for a further 48 h, after which time methanethiosulfonate (MTS) reagent (20% V / V ) was added to each well. The plates were incubated for 1-4 h on a Nutating mixer in a 5% CO 2 incubator, and then the optical density (OD) was read at 490 nm by using a SpectraMax M2 plate reader (Molecular Devices Corp., Sunnyvale, CA, USA). The blank values were substracted from each value, and the viability values of the treated samples relative to controls with vehicle were calculated. The values for the plots are the means ± standard deviation of six different wells.

Conclusions
Six novel cationic GAEL analogs were synthesized to explore novel SAR in this class of compounds. We were especially interested to explore how changes in the nature of the anomeric linkage, the nature of the hydrophobic lipid tail and the stereochemistry of the glycerol moiety affects the cytotoxic properties against breast, pancreas and prostate cancer cell lines. During our study, we discovered that replacement of the O-glycosidic linkage of lead compound 1 with an N-glycosidic linkage or a triazole linkage resulted in significant loss of anticancer activity against all six cancer cell lines tested. Moreover, replacement of the hydrophobic lipid tail by a fluorinated tail as well as replacement of the methoxy substituent by a second glucosamine moiety resulted in decreased cytotoxic activity.