The Amadori Rearrangement for Carbohydrate Conjugation: Scope and Limitations

The Amadori rearrangement was investigated for the synthesis of C‐glycosyl‐type neoglycoconjugates. Various amines including diamines, amino‐functionalized glycosides, lysine derivatives, and peptides were conjugated with two different heptoses to generate non‐natural C‐glycosyl‐type glycoconjugates of the d‐gluco and d‐manno series. With these studies, the scope and limitations of the Amadori rearrangement as a conjugation method have been exemplified with respect to the carbohydrate substrate, as well as the amino components.

conjugates of the D-gluco and D-manno series. With these studies, the scope and limitations of the Amadori rearrangement as a conjugation method have been exemplified with respect to the carbohydrate substrate, as well as the amino components.
We have demonstrated earlier that the Amadori rearrangement can also be considered as a versatile synthetic method for the preparation of glycoconjugates. This reaction allows the conjugation of aldoses 1 to amino components achieving Cglycosyl-type neoglycoconjugates such as 2 (Scheme 1). [11] Scheme 1. Amadori rearrangement furnishes C-glycosyl-type neoglycoconjugates (1-amino-1-deoxy ketoses 2) from aldoses 1.
The conjugation through a C-glycosidic linkage is of particular interest for biological investigations, because this type of linkage is not sensitive towards enzymatic hydrolysis, in contrast to the more common O-and N-glycosidic bonds. In spite of this, the Amadori rearrangement has not been extensively employed in the past because of several challenges accompanying this reaction. For example, the reversibility of several steps including the introduction of the amine and the subsequent isomerization to the aminodeoxy ketose product often leads to low yields. Furthermore, the rearrangement product can enter the Maillard reaction cascade, [12] leading to a range of side and degradation products. In addition, the formation of anomeric mixtures of both furanoside and pyranoside Amadori products causes difficulties during the isolation of one desired isomer. Nevertheless, we have earlier shown the versatility of the Amadori rearrangement in various applications; [11b,11c,13] therefore, we commenced a program to further investigate the scope and limitations of this rearrangement as a new glycoconjugation method. Herein, we focus on the use of (i) diamines, (ii) more complex amino-functionalized glycosides, and tors of bacterial adhesion. [13c] Treatment of aldoheptose 9 with 1,6-diaminohexane, under the same reaction conditions exemplified with aldoheptose 3, gave, after 2 d at 70°C, the disubstituted product 10 in 60 % yield.
The analogous reaction with 2,2′-(ethylenedioxy)bis(ethylamine) gave rearrangement product 7 in only 23 % yield and the corresponding monosubstituted Amadori rearrangement product 6, which was concomitantly obtained in 56 % yield. In this case, the low yield of 7 might be explained by the limited availability of the second amino group, which remains free after the first Amadori rearrangement, leading to 6. Likewise, when sugar substrate 9 was employed with 2,2′-(ethylenedioxy)bis(ethylamine) under the same reaction conditions, only 29 % of the bis-compound 13 was isolated, whereas the corresponding mono-compound 12 was isolated in a yield of 55 %. On the other hand, when 4,7,10-trioxa-1,13-tridecanediamine was employed as the amino component and an excess of aldoheptose 3 (3 equiv.) was used, the Amadori rearrangement gave exclusively the disubstituted product 8 in a yield of 58 %. Aldoheptose 9 gave, under the same reaction conditions, exclusively product 14 in a yield of 54 %. Hence, the amount of monosubstitution vs. disubstitution of diamines can be controlled by selecting the amount of aldose in a quantifiable manner. Additionally, the availability of the second amino group increases with the chain length of the spacer of the respective diamines.
During our investigation applying various amines in the Amadori rearrangement, we observed an H/D exchange in the NMR spectra at position C-1. This isotopic exchange was previously detected by Heyns and co-workers [14] who noted that signals of protons at the position C-1 in Amadori rearrangement compounds decreased on prolonged storage of solutions in D 2 O because of H/D exchange, which significantly accelerated with increasing pH values [15] (see the Supporting Information).
Inspired by the success of the double Amadori rearrangement with p-xylylenediamine, we next investigated the use of 4-aminobenzylamine (15) as a bidirectional linker. Here, we aimed at a regioselective single Amadori rearrangement involving the benzylic amino group, whereas the arylamino group was intended to remain available for subsequent modifications. Indeed, the reaction of both aldoheptoses 3 and 9 with diamine 15 gave exclusively the singly modified rearrangement products 16 and 17 in 73 % in both cases (Scheme 3). Scheme 3. Diamine 15, having two differing pK a values, allows for regioselective Amadori rearrangement.
The regioselective ligation reaction with 15 can be rationalized on the basis of the pK a values of the amino group. Given that basicity is a crucial parameter for the nucleophilicity of amines, which is required for the success of the Amadori rearrangement, (4-aminobenzyl)amine (15), with pK a values of 9.3 for the benzylic amine and 4.6 for the aniline amino group, reacts selectively at the benzylic position. Products 16 and 17, respectively, can be further ligated in amide coupling reactions, thiourea bridging or other ligation reactions involving amines. In addition, the amino group can be converted into an azido function, which is amenable for click chemistry.

Amadori Rearrangement with Amino-Functionalized Carbohydrates
To utilize the Amadori rearrangement for the synthesis of more complex glycoconjugates, amino-functionalized glycosides were used as amino components. To this end, 3-O-(aminopropyl)-functionalized mannosides 20 and 23 were synthesized by applying a procedure for tin-mediated regioselective etherification of glycosides, which we have disclosed earlier (Scheme 4). [16] Thus, mannosides 18 [17] and 21, [18] respectively, were treated with dibutyltin oxide to achieve the corresponding stannylidene acetal intermediates. These were treated with N- (3-bromopropyl)phthalimide in the same reaction vessel after the solvent was exchanged from MeOH to N,N-dimethylformamide (DMF) to deliver the respective 3-O-functionalized mannosides 19 and 22 after regioselective opening of the tin acetal ring. Although the yields were moderate in this step, this is an advantageous direct approach to selectively 3-O-functionalize mannosides without the need for protecting-group chemistry. The free amines 20 and 23 were obtained after hydrazinolysis [16] of 19 and 22 in 80 and 89 % yields, respectively.
less, both compounds are valuable intermediates, because they offer orthogonal groups at the anomeric position for further modification: compound 27, bearing a masked amino function, and compound 28, a versatile propargyl group.

Synthesis of Amino Acid Glycoconjugates by Amadori Rearrangement
We then investigated the use of the Amadori rearrangement as a conjugation method for the synthesis of C-glycosyl-type amino acid glycoconjugates. Thus, we employed partly protected lysine derivatives 29, 31, and 33 as well as dipeptide 35 and tripeptide 38 as amino components (Scheme 6).

Conclusions
We demonstrated that the Amadori rearrangement is an attractive conjugation method for the synthesis of C-glycosyl-type glycoconjugate mimetics without the need for protectinggroup manipulations.
Symmetrical diamino components gave the respective disubstituted Amadori rearrangement products in acceptable yields. Interestingly, the ratio of mono-sugar conjugation vs. bissugar Amadori products can be controlled in a preparatively useful manner by controlling the amount of sugar substrate employed. The regioselectivity of the Amadori rearrangement employing unsymmetrical diamines can be controlled by taking the pK a value as a parameter for the nucleophilicity of the respective amino components into account. Amines with a pK a range between 8 and 12 are more efficient nucleophiles for this reaction than less basic amines. Amino-functionalized glycosides can also be employed as amino components in the Amadori reaction. By the choice of the aglycon, such as a masked amino function or a versatile propargyl group, different applications for the obtained building blocks can be envisaged. Lysine derivatives as well as lysine-containing di-and tripeptides in the Amadori rearrangement lead to C-glycosyl-type glycopeptide mimetics, which can be used as building blocks for glycopeptide and glycoprotein synthesis. In all cases the α-anomeric configuration of the respective Amadori products was obtained exclusively, as has been shown earlier in the D-gluco as well as D-manno series. [11b,11c] In the context of configuration of the sugar substrate, no significant differentiation in the obtained yields was observed.
Despite the fact that yields for this conjugation method may occasionally be found in a preparatively moderate range and product purification may be extensive in a few cases, conjugation through a C-glycosidic linkage leads to versatile building blocks for different applications, in particular for biological investigations, because this type of conjugation is not sensitive towards enzymatic or chemical hydrolysis in a physiological environment.

Experimental Section
Materials and General Methods: All chemicals were purchased from Sigma Aldrich and used without further purification. Moisturesensitive reactions were carried out under nitrogen in dry glassware. NMR spectra were recorded with Bruker Ultrashield spectrometers at 300.36 ( 1 H) and 75.53 ( 13 C) MHz, respectively. Higher resolution NMR spectra were recorded with VARIAN INOVA 500 MHz at 500.619 ( 1 H) and 125.894 ( 13 C) MHz, respectively. Chemical shifts are reported relative to internal tetramethylsilane (δ = 0.00 ppm), Thin-layer chromatography was performed on precoated silica gel plates on aluminum 60 F254 (E. Merck 5554). Detection was effected by UV and/or charring with 10 % sulfuric acid in EtOH as well as with ceric ammonium molybdate (100 g of ammonium molybdate/8 g of ceric sulfate in 1 L of 10 % H 2 SO 4 ) followed by heat treatment at ca. 180°C. Flash chromatography was performed on silica gel 60 (0.035-0.070 mm, 60 A, Acros Organics 24036) using distilled solvents. Ion exchange chromatography was performed with a strong cation exchanger (Amberlite® CG-120-II, Na + form, 200-400 mesh, Fluka 06449) using water and a water/NH 4 OH conc. mixture.

General Method A (Diamines):
To a solution of the respective aldoheptose (2 equiv.) in a mixture of EtOH and 1,4-dioxane, the amino compound (1 equiv.) and acetic acid (2 equiv.) were added, and the reaction mixture was stirred at 70°C until TLC showed satisfactory consumption of the starting material. The reaction mixture was concentrated under reduced pressure, and the crude product was purified by ion exchange chromatography as well as silica gel chromatography with the solvent system indicated.

General Method B (Amines):
To a solution of the respective aldoheptose (1 equiv.) in a mixture of EtOH and 1,4-dioxane, the amino compound and acetic acid (1 equiv.) were added, and the reaction mixture was stirred at 70°C until TLC showed satisfactory consumption of the starting material. The reaction mixture was concentrated under reduced pressure, and the crude product was purified by silica gel column chromatography with the solvent system indicated.