Sugar-Assisted Glycopeptide Ligation with Complex Oligosaccharides: Scope and Limitations

Synthesis of Glycosyl Amino Acid Building Blocks

The initial phase of this study involved the construction of the auxiliary-containing glycosyl amino building blocks for solid-phase peptide synthesis (SPPS) of the model glycopeptides (Figure 1). We envisioned that glycopeptides 1–3 could be constructed through enzymatic elaboration of a monosaccharide-containing glycopeptide following SPPS. As such most of the targets could be synthesized from the previously reported amino acid 5 (Figure 2).³⁴ For the construction of glycopeptide 4 it was necessary to synthesize fucose-containing amino acid 6 (Figure 2), as there is no readily available enzyme for modification at the C-6 position of the bridgehead GlcNAc.

Synthesis of 6 began with transformation of known glucosamine derivative 7⁴⁰ into triol 8 through a three-step sequence (Scheme 2). Triol 8 was coupled to fucosyl bromide 12⁴¹ under halide ion conditions to afford disaccharide 9 in modest yield.^42,43 At this point it was necessary to replace the benzyl protecting groups with acetates in order to stabilize the glycosidic linkage during SPPS. Hydrogenolysis of the benzyl groups occurred with partial removal of the 2,2,2-trichloroethyl carbamate (Troc), necessitating its reinstallation prior to global acetylation. The resulting peracetylated disaccharide was reacted with benzylamine in THF to effect selective cleavage of the anomeric acetate, followed by treatment with 2,2,2-trichloroacetonitrile in the presence of potassium carbonate to afford glycosyl trichloroacetimidate 10.⁴⁴ This was coupled to protected serine 13⁴⁵ in the presence of TMSOTf and activated sieves to afford glycosyl amino acid 12 in excellent yield. Removal of the Troc protecting group was followed by HBTU/DIEA-mediated coupling of the free amine to S-trityl-2-mercaptoacetic acid⁴⁶ and removal of the allyl protecting group⁴⁷ to provide glycosyl amino acid 6 in high yield.

Solid-Phase Peptide and Glycopeptide Synthesis

Synthesis of glycopeptides 14a–d (Scheme 3), which served as precursors for enzymatic elaboration, was achieved using SPPS on Rink-amide resin.⁴⁸ Due to the complexity of 5, only one equivalent of amino acid was used and the coupling time was extended to 6 h. As reported previously,^32,34 these conditions were sufficient for efficient coupling as determined by the UV absorption of the Fmoc/piperidine adduct at 302 nm. Following coupling of 5, the resin was split into four equal parts for the synthesis of Gly, Asp, His and Gly-Gly glycopeptides 14a–d. These targets were chosen because the N-terminal Gly, Asp, and His glycopeptides perform particularly well in SAL; while the Gly-Gly peptide provides a simple extended glycopeptide model. Following the final coupling, the acetate protecting groups on the sugar were removed by hydrazinolysis prior to release from the resin using trifluoroacetic acid/triisopropylsilane/thioanisol/water (17:1:1:1). The resulting products were purified by HPLC to afford the desired glycopeptides in 45–82% yield.

Synthesis of glycopeptide 4 was also achieved using Fmoc-based SPPS (Scheme 4). Due to the challenges associated with the construction of amino acid 6, we chose to only synthesize an N-terminal Asp glycopeptide as glycopeptides possessing this residue at their N-terminus provided some of the best yields for SAL. Similar to the synthesis of 14a–d, we found that acceptable yields could be achieved using 1 equivalent of 6 and a coupling time of 6 h. Because we were concerned about the stability of the sensitive α-1,6-fucose linkage we chose to release the glycopeptide from the resin prior to removal of the carbohydrate protecting groups. Following removal from the resin under standard conditions and HPLC purification the glycopeptide, Zemplén deacetylation at pH 9 (to prevent β-elimination of the glycan)⁴⁹ afforded glycopeptide 4.

Peptide thioesters bearing amino acids Gly, Tyr, His and Ala at the C-terminus were synthesized using the Boc-strategy, as reported previously.^32,34 These peptide thioesters were selected so that direct comparisons could be made with SAL reactions bearing monosaccharides which were reported previously. The N-termini of the peptide thioesters were capped with acetates prior to removal from the resin in order to prevent any unwanted aminolysis during the ligation reaction.^50–52

Enzymatic Elaboration of Glycopeptides

Having synthesized glycopeptides 14a–d, we next turned our attention to the enzymatic synthesis of glycopeptides 1–3a–d. To achieve this, we chose to focus on glycosyltransferase-mediated elaboration of 14a–d as many of these enzymes are know to have a broad substrate tolerance with regard the glycosyl acceptor.⁵³ Importantly, from previous studies we already knew that the presence of the sulfhydryl group at C-2 of the bridgehead sugar did not interfere with at least one glycosyltransferase, β-1,4-galactosyltransferase (GalT).³² As such, treatment of glycopeptides 14a–d with GalT and UDP-galactose (UDP-Gal) in the presence of alkaline phosphatase and MnCl₂ afforded the desired disaccharide products 1–d in 81–86% isolated yields (Scheme 5).^54–56 These products were readily converted to sialic acid containing trisaccharides 2a–d upon treatment with CMP-sialic acid (CMP-sial) and α-2,3-(N)-sialyltransferase (SialT), under conditions similar to those described for the GalT reaction.⁵⁷ Conversion of 2a–d to glycopeptides 3a–d bearing the sialyl Lewis X motif using α-1,3-fucosyltransferase (FucT)⁵⁸ proved to be much more challenging, initially providing the desired target glycopeptides in poor yields. This was not entirely surprising given the proximity of the unnatural sulfhydryl group to the C-3 glycosylation site on the bridgehead sugar. Pleasingly, however, we found that it was possible to obtain target glycopeptides 3a–d through the use of an excess of GDP-fucose, FucT, and prolonged reaction times (see supporting information).

Scope of Sugar-Assisted Ligation with Complex Glycans

With the target glycopeptides in hand, we turned our attention to investigating the efficiency of SAL in the presence of more complex glycans. Our initial investigations focused on the ligation between glycopeptides 1a–d with peptide thioester 15a, bearing a C-terminal glycine residue, under mixed solvent system conditions (4:1 v/v N-methyl pyrrolidinone/6M guanidine hydrochloride, 100 M HEPES, pH 8.5, 3% PhSH, 37 °C),³³ in order to minimize peptide thioester hydrolysis. Under these conditions glycopeptides 1a and 1b, possessing glycine and aspartic acid at their N-terminus, underwent ligation with 15a in 63% and 70% yield, respectively (Table 1, entries 1 and 2). The fact that these yields were lower than those observed in previous studies with the C-2 auxiliary-containing monosaccharide (91% for glycine)³⁴ suggests that the increased steric bulk of the galactose moiety may effect the efficiency of the reaction. This effect is even more pronounced with glycopeptide 1c, possessing an N-terminal histidine, which undergoes ligation with 15a in 34% yield (entry 3). This result was surprising given that in previous studies glycopeptides bearing an N-terminal His gave good yields in SAL reactions. The lower yield observed with the more complex glycan is presumably due to unfavorable interactions between the larger His side chain and the auxiliary-containing disaccharide during the S→N acyl shift. Finally, extension of the N-terminus of the glycopeptide by an additional glycine residue (1d) results in a drop in yield, presumably due to the increased size of the proposed ring transition state (entry 4).

Table 1.

Scope of SAL with glycopeptides 1a–d.^a

Entry	Glycopeptide(AA1)	Thioester (AA2)	Ligation Junction–AA2-AA1-	Isolated Ligation Yieldb
1	1a (Gly)	15a (Gly)	Gly-Gly	63%
2	1b (Asp)	15a	Gly-Asp	70%
3	1c (His)	15a	Gly-His	36%
4	1d (Gly-Gly)	15a	Gly-Gly	49%
5	1b	15b (Tyr)	Tyr-Asp	70%
6	1b	15c (His)	His-Asp	70%
7	1b	15d (Ala)	Ala-Asp	55%

Open in a new tab

^a

Conditions: 4:1 v/v NMP:6M Gn·HCl, 1M HEPES, pH8.5, 2% PhSH, 37 °C, 96 h.

^b

Isolated yield after HPLC purification.

We next turned our attention to the effect of extended glycosylation on the reaction with more sterically demanding peptide thioesters. To this end, we examined the reaction of 1b with peptide thioesters 15b–d, possessing Tyr, His and Ala at their C-terminus, respectively (table 1, entries 5–7). These peptide thioesters had been used in our previous studies and would therefore allow a better assessment of the effects of the larger glycan on the reaction. Pleasingly, we found that the presence of the sterically demanding Tyr residue in peptide thioester 15b did not interfere with the reaction, providing the ligation product in yields comparable to those observed with peptide thioester 15a (entry 5). Similarly, glycopeptide thioester 15c, possessing a C-terminal His residue, also underwent ligation in good yield (entry 6). Consistent with our previous studies,^32–34 the ligation between 1b and peptide thioester 15d, possessing an N-terminal Ala residue, gave a slightly lower yield (55%), however this is still deemed to be synthetically useful (entry 7). This data suggests that the C-terminal residue of the peptide thioester is less sensitive to the presence of larger glycans than the N-terminus of the glycopeptide, in line with trends observed in previous SAL studies.^31–34

In order to further probe the effects of glycosylation at C-4, we examined the efficiency of SAL with trisaccharide glycopeptides 2a–d. Pleasingly, we found that the addition of a sialic acid residue to C-3 of the terminal galactose had little effect on the reaction between peptide thioester 15a and glycopeptides 2a and 2b (Table 2, entries 1 and 2). By comparison, the reaction between glycopeptide 2c, possessing an N-terminal His, and 15a proceeded in poor (14%) yield (entry 3), with no observable increase in yield with prolonged reaction times. The large decrease in yield upon moving to the larger glycan, relative to that obtained with disaccharide containing glycopeptide 1c (Table 1, entry 3), further underscores our observation that larger amino acids experience unfavorable steric interactions with the extended glycans in the reaction. The increased bulk of the glycan did not affect the reaction between 15a and extended glycopeptide 2c, which gave the desired product in yields comparable to those obtained with the disaccharide containing glycopeptide 1d. From these results, we conclude that for sterically unencumbered amino acids extending glycosylation at C-4 beyond a disaccharide does not have a major impact on the reaction. Ligation products could be desulfurized in an efficient manner by treatment with 10% palladium on alumina in the presence of tris(2-carboxyethyl)phosphine hydrochloride (see supporting information).³²

Table 2.

Scope of SAL with trisaccharide-containing glycopeptides 2a–d.^a

Entry	Glycopeptide(AA1)	Ligation Junction–AA2-AA1-	Isolated Ligation Yieldb
1	2a (Gly)	Gly-Gly	57%
2	2b (Asp)	Gly-Asp	70%
3	2c (His)	Gly-His	14%
4	2d (Gly-Gly)	Gly-Gly	44%

Open in a new tab

^a

Conditions: 4:1 v/v NMP:6M Gn·HCl, 1M HEPES, pH8.5, 2% PhSH, 37 °C, 96 h.

^b

Isolated yield after HPLC purification.

We next turned our attention to the more sterically congested sialyl Lewis X glycopeptides 3a–d. Given the effects of steric hindrance on the reaction observed with glycosylation at C-4, we were concerned that the introduction of the fucose residue onto the C-3 position of the bridgehead sugar would be deleterious to the reaction. This indeed proved to be the case, and we were unable to isolate any product from ligations between 3a–d and peptide thioester 15a, even after prolonged reaction times (Scheme 6). These results indicate that the SAL reaction is not compatible with extended glycosylation at positions adjacent to the thiol-auxiliary. This observation provides valuable information about which glycan structures can be utilized for SAL reactions and should prove useful for planning syntheses of complex glycopeptides and glycoproteins in future studies.

We next examined the effect of glycosylation at C-6 of the bridgehead sugar using glycopeptide 4. Given the distance from the auxiliary, we were hopeful that this modification would not have a significant negative effect on the reaction. To this end, we examined the reaction between glycopeptide 4 and peptide thioesters 15a and 15d (Scheme 7). In both cases the SAL reaction proceeded in moderate (44%) yield to afford glycopeptides 16a and 16b. While these yields are synthetically useful, they are lower than those observed with auxiliary-containing monosaccharides or C-4 substituted auxiliary-containing glycans. This suggests that the glycan at C-6 must somehow be interacting with the peptide backbone. Importantly, given the use of the generally more reactive glycopeptide possessing an N-terminal Asp in the SAL reaction, these results indicate that care must be used in selecting N-terminal residues in SAL reactions when the auxiliary-containing glycan is substituted at C-6.

N-Linked Ligation Study

Having examined the effects of complex O-glycosylation on the efficiency of the SAL reaction, we sought to determine if complex N-linked auxiliary-containing glycopeptides were compatible with the reaction. To this end, we envisioned enzymatic elaboration of glycopeptide 17³² using endo-β-N-acetylglucosaminidase A or M (Endo A or M) mediated transglycosylation.^59,60 Unfortunately, all attempts to transfer a complex oligosaccharide to the auxiliary-containing glycan in 17 using Endo M failed. Rationalizing that the C-2 sulfhydryl group might be interfering with the reaction we next examined Endo A-mediated transfer of sugar oxazoline 18,⁶¹ as this has been reported to be a much more efficient substrate for Endo mediated transfer.^62–64 Pleasingly, this strategy proved to be more effective, delivering the target glycopeptide 19 in 38% yield (Scheme 8). This glycopeptide underwent SAL with peptide thioester 15a to afford glycopeptide 20 in 40% yield after HPLC purification.

Conclusion

The presence of extended glycosylation on the auxiliary-containing glycan in SAL can have a significant impact on the outcome of the reaction. Glycosylation at C-4 has the smallest effect on the reaction, where in many cases the reaction proceeds with high levels of efficiency. The effect of extended glycosylation at C-6 on the SAL reaction is more pronounced, resulting in lower, but still synthetically useful, ligation yields. In both of these cases, the selection of the N-terminal amino acid in the auxiliary-containing glycopeptide is critical, and sterically unencumbered amino acids are necessary for optimal yields. Importantly, it appears that the peptide thioester is much less sensitive to extended glycosylation at C-4 and C-6 on the auxiliary-containing glycan, as even peptide thioesters possessing bulky C-terminal residues smoothly undergo the reaction with glycopeptides bearing appropriate N-terminal residues. SAL is not tolerant of glycosylation at positions adjacent to the auxiliary, however, as glycosylation at C-3 prevents the reaction from taking place. Finally, both O-linked and N-linked auxiliary-containing glycopeptides can be extended, although it appears that the latter is more sensitive to larger glycans. Based on these results, we urge caution in using SAL when a novel glycosylation pattern is encountered in a glycoprotein, as structural changes in more complex glycans can have major and unexpected effects on the reaction, as evidenced by the results obtained with the C-6 substituted glycopeptide 4. Nevertheless, we feel that when appropriate SAL is a useful addition to the glycopeptide, and potentially glycoprotein, synthesis toolbox, as it permits access to ligation junctions that cannot be constructed using other ligation strategies.