HO OH Scheme 28
wanted reaction pathways and the need for tedious protecting group schemes or stereocontrolling auxiliaries. In addition, sialyltransferases have been shown to tolerate several acceptor modifications , although less is known about their donor specificity because the requisite CMP-NeuAc sugar donors are so difficult to synthesize.
Kinetic isotope effect studies with a particular transferase, namely, rat liver a-2,6-sialyltransferase, have shown that there is significant charge buildup at the an-omeric center in the transition state . This suggests that the glycosylation proceeds through an SN1-like mechanism. CMP-NeuAc appears to be the common donor for all the sialyltransferases. However, each enzyme varies in regioselectivity and to some degree in acceptor specificity. Under the premise that an enzyme exists for the formation of each natural sialoside, there should be at least nine sialyltransferases responsible for the synthesis of the following disaccharides: NeuAc-a-2,6-Gal, NeuAc-a-2,3-Gal, NeuAc-a-2,6-GalNac, NeuAc-a-2,4-Gal, NeuAc-a-2,4-GlcNac, NeuAc-a-2,6-GlcNAc, NeuAc-a-2,6-Man, NeuAc-a-2,8-NeuAc, and NeuAc-a-2,9-NeuAc. However, only a small subset of the known and postulated sialyltransferases has been isolated and used for synthesis purposes.
The sialyltransferases are membrane-bound proteins located in the endoplasmic reticulum (ER) and in the Golgi apparatus. Information about their sequence ho-mology is limited, but they do appear to share a common topography . A catalytic domain resides at the C-terminus followed by an N-terminal segment that anchors the enzyme into the ER or Golgi membrane. Soluble, catalytically active sialyltrans-ferases that lack the anchor segment have been isolated from milk, serum, and other body fluids, suggesting that this N-terminal anchor is not necessary for the enzyme to retain catalytic activity. However, the ability to obtain from natural sources quantities of most sialyltransferases that would be needed for synthesis applications is hampered by low tissue concentrations and difficult purifications.
The genes for members of some of the most common classes of sialyltrans-ferases have been cloned and expressed [33a]. Expressing and isolating membrane-bound enzymes in high catalytic activity can be difficult; however, Paulson and coworkers replaced the anchor segment with a cleavable peptide in the expression of Gal-^-1,4-GlcNAc a-2,6-sialyltransferase [35a,b]. This allowed the enzymes to be secreted from the cell into the medium, thus simplifying the process of isolation. This technology has been used to express several sialyltransferases in quantities
HO OH CMP-NeuAc H0 OH
HO OH CMP-NeuAc H0 OH
suitable for synthesis. Gaining sialyltransferase accessibility is becoming a less serious problem as these enzymes become commercially available.
The primary method for the synthesis of CMP-NeuAc 88 is through the use of CMP-NeuAc synthetase, an enzyme that catalyzes the condensation of cytidine triphos-phate (CTP) with sialic acid to produce CMP-NeuAc (Scheme 29) . CMP-NeuAc synthetase has been cloned from microbial sources and has been isolated from mammalian tissues [36,37]. The substrate specificity of each synthetase has been studied to some degree [36,37]. The mammalian version accepts C9 and some C8 modifications of NeuAc, as well as variations at the C5 position, such as replacement of the acetamide with OH (KDN) or hydroxylation of the acetamide (NeuAc). The microbial CMP-NeuAc synthetase has a high activity for C9-modified sialic acids, but does not tolerate alterations at the C5 position. Unfortunately, the microbial version is the more readily available enzyme, thus limiting the variety of analogs that can be prepared in this manner. Scheme 30 gives a representative sample of some important C9-derivatized CMP-NeuAc analogs . Modifications at the C5 position are somewhat limited by the specificity of CMP-NeuAc synthetase for an amide at this position. Nevertheless, several analogs were prepared that incorporate sterically and electronically diverse substituents at C5 (Scheme 31) .
Wong and coworkers first attempted the nonenzymatic synthesis of CMP-NeuAc by employing phosphoramidite chemistry in a key step involving the ligation of a sialyl phosphoramidite and a selectively protected cytidine analog (Scheme 32) . Treatment of NeuAc derivative 99 with 2-cyanoethyl chlorophosphoramidite resulted in the formation of the ^-sialyl phosphoramidite 100 in 89% yield. The sialyl donor was then coupled to the protected cytidine 101 under promotion by 1-H tetrazole to afford the intermediate phosphite, which was immediately oxidized with tert-butyl hydroperoxide to provide the protected CMP-NeuAc 102. Deprotec-tion of this intermediate was not reported; however, the Wong group demonstrated that the CMP-NeuAc core structure could be successfully synthesized utilizing phos-phoramidite chemistry.
The Schmidt group utilized a sialyl phosphite in a very different synthesis strategy (Scheme 33) . Upon treatment of sialyl donor 54 with cytidine phosphoric acid 103, a phosphite-phosphate exchange reaction occurred to give compound 104 exclusively as the /3 isomer. Deacylation by treatment with sodium meth-oxide followed by ester saponification through the addition of water provided CMP-NeuAc 88. This method circumvented the need for an oxidation step or phosphorus deprotection. This method was also applied to the synthesis of another naturally occurring CMP-NeuAc derivative 105 .
Hata and coworkers took a different approach to the ligation by utilizing cytidine phosphoramidite 107 and the tertiary anomeric alcohol of 106 as coupling partners (Scheme 34) . The coupling reaction successfully provided a phosphite intermediate, which was subsequently oxidized with tert-butyl hydroperoxide to provide the trialkyl phosphate 108. Attempts to purify the phosphotriester failed, but the production of 108 was supported by 31P NMR spectroscopy of the crude reaction mixture. The protecting groups were then removed under mild conditions by treatment with tetrakis(triphenylphosphine)palladium(0) in the presence of the allyl scavenger ra-butylammonium bicarbonate. Following purification by size exclusion chromatography, CMP-NeuAc 88 was obtained in acceptable overall yield (25% for three steps).
In a related synthesis of CMP-NeuAc, Kajihara and coworkers also used a cytidine phosphoramidite in the coupling reaction, although with different protecting group patterns on the coupling partners (Scheme 35) . The reaction of 99 with phosphoramidite 109 provided the phosphite 110 as a mixture of phosphorus dia-stereomers. Phosphite oxidation and subsequent treatment with methoxide led to decomposition. As an alternative, the cyanoethyl protecting group was first removed with DBU; then complete deprotection by treatment with sodium methoxide and sodium hydroxide afforded CMP-NeuAc 88 in good yield. This stragety was also applied to the synthesis of CMP-NeuAc-a-2,8-NeuAc 111 .
Halcomb and Chappell developed a route to CMP-NeuAc 88 that promises to be general for the synthesis of virtually any derivative thereof [44,45]. The route (Scheme 36) utilizes a condensation of sialic acid derivative 99 with the phosphoramidite 112 to afford the phosphite 113 in 62% yield. Oxidation of the phosphite provided the phosphotriester 114 , which was taken directly to the next transformation without purification (owing to its instability to chromatography). Deal-lylation of the phosphate gave compound 115 (61% for two steps), which was stable to silica gel chromatography. Compound 115 was deacylated with methoxide, and its methyl ester was subsequently saponified with NaOH to provide CMP-NeuAc 88. The derivatives shown in Scheme 37 were synthesized according to this protocol and were investigated as substrates for sialyltransferases (see below).
The synthesis of a CMP-NeuAc derivative that was bound to a solid support through the 9-position of the sialic acid has been reported by the Kajihara group . This derivative is quite useful in that it can be utilized to immobilize glyco-proteins onto a solid support by transferring the sialic acid to the terminus of the carbohydrate chain of the glycoprotein.
The Brossmer and Paulson groups have studied the sialyltransferase donor specificity with a series of C5 and C9 conjugates, all of which were prepared through the CMP-
Was this article helpful?