Machine-Driven Chemoenzymatic Synthesis of Glycopeptide

Abstract

Historically, researchers have put considerable effort into developing automation systems to prepare natural biopolymers such as peptides and oligonucleotides. The availability of such mature systems has significantly advanced the development of natural science. Over the past twenty years, breakthroughs in automated synthesis of oligosaccharides have also been achieved. A machine-driven platform for glycopeptide synthesis by a reconstructed peptide synthesizer is described. The designed platform is based on the use of an amine-functionalized silica resin to facilitate the chemical synthesis of peptides in organic solvent as well as the enzymatic synthesis of glycan epitopes in the aqueous phase in a single reaction vessel. Both syntheses were performed by a peptide synthesizer in a semiautomated manner.

Glycans often express their functions by attaching to proteins (called glycoproteins) or lipids (called glycolipids).^[1] Of the more than 300 known posttranslational modifications, glycosylation (O-linked, N-linked, and C-linked glycosylation) is one of the most complex and abundant protein post modifications.^[2] It is predicted that more than 50% of mammalian proteins are glycosylated.^[3] Protein glycosylation mediates a diversity of physiological and pathological processes; this includes but is not limited to immune responses, angiogenesis and tumor cell metastasis, protein folding and degradation, cell-cell communications, and cell-pathogen interactions. Structurally defined glycopeptides have significant potential for application in enzymatic activity tests,^[4] clinical diagnostics,^[5] and in the development of glycopeptide antibiotics and carbohydrate-based vaccines.^[6] There are two main approaches for glycopeptide preparation. One strategy incorporates glycosylated amino acid building blocks in solid phase peptide synthesis.^[7] Alternatively, a short glycan epitope is grafted onto the peptide backbone and after completion of peptide synthesis. The glycan is further extended by chemical or enzymatic synthesis.^[8] However, both methods require skilled researchers to perform the synthesis, which is labor-intensive and time-consuming.

Historically, researchers have put considerable effort toward developing an automation system to produce natural biopolymers, aiming to allow non-specialists to access these biopolymers for biological studies easily. In 1963, Merrifield developed solid-phase synthesis to produce peptides.^[9] Since then, many automated synthesizers that are based on solid-phase synthesis have been successfully developed for the preparation of peptides and oligonucleotides by non-specialists.^[10] This has significantly advanced the development of natural science. Over the past few decades, breakthroughs have also been made in development of an automated synthesizer to produce oligosaccharides based on chemical glycosylation and enzymatic glycosylation.^[11] However, few works have been done about automation glycopeptide synthesis. Thurin and co-workers successfully achieved automation synthesis of N-glycopeptide based on the combination of solid phase peptide synthesis and a peptide synthesizer, by which a short peptide containing a GlcNAc or chitobiose residue was obtained.^[12] Using a special oligosaccharide synthesizer, Seeberger and co-corkers also synthesized several O-glycopeptides bearing monosaccharide or disaccharide in an automation manner.^[13] However, the synthesis of more complicated glycopeptides has not been reported due to the lack of a general platform.

Herein, we describe a machine-driven platform able to synthesize glycopeptides by a chemoenzymatic synthesis strategy. The principle is the chemical coupling of amino acids or glycoamino acids onto growing peptide chains followed by enzymatic extension of glycan epitopes to produce target glycopeptides. Chemoenzymatic synthesis of glycopeptides takes advantages of chemical synthesis of peptides and enzymatic extension of glycan epitopes. This could reduce the number of synthetic steps. The described platform is based on the use of an amine-functionalized silica resin to facilitate chemical synthesis of peptide in organic solvent and as well as enzymatic synthesis of glycan epitopes in aqueous phase in a single reaction vessel. Both stages were performed on a peptide synthesizer and controlled by a computer program (Scheme 1).

Scheme 1.

The principle of machine-driven chemoenzymatic synthesis of glycopeptide.

To achieve automated synthesis of glycopeptides, there are three prerequisites, including the cleavable linker, the solid support, and the automation instrument that used for performing chemical or enzymatic reactions. Chemical coupling of amino acids to a growing peptide chain needs to be performed in the organic phase, while enzymatic extension of glycan epitopes must be done in an aqueous solution. The popular solid resins that are used in peptide synthesis have high loading efficiency, but they are not well-accepted by glycosyltransferases. One possible reason is that most of starting substrates are located inside holes of the resin beads (this could increase the loading efficiency), while enzymes (big molecules) cannot enter inside of resins to catalyze glycan synthesis (Figure 1A). Meanwhile, the water-soluble resins that have been used in enzymatic solid-phase synthesis are not suitable for organic phase peptide synthesis.

Figure 1.

A) Comparison of traditional solid support and aminopropyl silica resin. B) Equations for loading efficiency and release efficiency. C) Conjugation of oligosaccharide primer (trisaccharide) with aminopropyl silica resin through a cleavable linker.

The popular cleavable linkers that used in solid-phase synthesis require either strong acid or base condition to release products. However, glycans are not stable in strongly acidic conditions and peptides will undergo racemization in strongly basic conditions. Moreover, the side chain of cysteine can be damaged under oxidation conditions. Therefore, we choose 3,4-diethoxy-3-cyclobuten-1,2-dione (squaric acid diethyl ester) as a cleavable linker in our design.^[14] This linker was first reported by Glu“senkamp and co-workers. It could be cleaved to generate a free amine group in the presence of 5% hydrazine in aqueous solution (Figure 1C). Importantly, both peptide and glycans are stable in this reaction condition. In addition to cleavable linker, a solid support is also necessary to perform automation synthesis on a peptide synthesizer. In our previous work about automation enzymatic synthesis of glycans by peptide synthesizer, we used a temperature-sensitive material (PNIPAM) as a support.^[15] PNIPAM is soluble in water at room temperature and precipitates out of aqueous solution when the temperature exceeds the lower critical solution temperature.^[16] Therefore, it is an ideal carrier support able to make use of the properties of commercially available peptide synthesizers to perform an automated process. However, it doesn’t work in organic solution such as DMF. To find a good resin for glycopeptide synthesis in this work, we tested many solid resins, and finally found an aminopropyl silica resin which could be used in our design (Figure 1A). Aminopropyl silica resin have been used in peptide synthesis.^[17] It doesn’t become popular because of its lower loading efficiency. Nevertheless, this silica resin could be used in enzymatic synthesis in aqueous solution.^[18] To test the loading efficiency of the selected solid resin and cleavable linker, 1 g of 2 (aminopropyl silica resin) was reacted with squaric acid diethyl ester (850 mg, 5 mmol) under triethyl amine (1 g, 10 mmol) in DMF (10 mL) to generate 3 (1 mL acetic anhydride was used to capping the amino group). 3 was shaken with 1 (625 mg, 1 mmol) and triethyl amine (1 g, 10 mmol) to produce 4 (1.1 g), giving a loading efficiency of 0.1 to 0.2 mmolg⁻¹ (calculated using the equation shown in Figure 1B), comparing to the loading efficiency of traditional resins used in peptide synthesis is 0.6 to 1.3 mmolg⁻¹. To release the conjugated oligosaccharide on 4, 4 (200 mg) was treated with 5% hydrazine aqueous solution overnight. The product was collected and concentrated. After purification by HPLC, 16 mg of product was obtained, indicating more than 90% of the oligosaccharide was released (calculated using the equation shown in Figure 1B). HPLC and NMR analysis indicated the product structure is 1. More importantly, the glycan is stable in the described reaction condition.

To test whether this aminopropyl silica resin can be used in our design, a reaction catalyzed by a β−1,4-galactosyltransferase from Neisseria meningitidis (LgtB)^[19] was chosen as an example (Figure 2). 500 mg (50 μmol) of 4 (prepared as described above) was incubated with LgtB (5 mg) and UDP-Gal (3 equiv) in a reaction buffer (50 mm of Tris-HCl, 5 mm of Mg²⁺, pH 8.0) for 24 h at 37 °C. Once reaction finished, the reaction was filtrated, and the silica resin was washed. Then, 5% aqueous hydrazine was added to release the oligosaccharide. After 12 h, the reaction mixture was filtrated, and the filtrate was collected. HPLC analysis indicated that more than 80% trisaccharide was converted to tetrasaccharide (Figure 2). This yield is similar with the yield of LgtB-catalyzed reaction in vitro. Similarly, we also tested the enzymatic activities of α1,2 fucosyltransferase from Helicobacter pylori (FucT), and α 1,3 N-acetylgalactosaminyltransferase from human (GTA), α 1,3 galactosyltransferase from human (GTB; Supporting Information, Figure SI2).^[20] All these enzymes could efficiently catalyze reactions when their substrates were linked with aminopropyl silica resin. These results indicated that this aminopropyl silica resin can be well accepted by glycosyltransferases and used in our design for automation synthesis of glycopeptides.

Figure 2.

HPLC analysis of an LgtB-catalyzed reaction using 4 as substrate.

Having solved the problems that mentioned above, a CEM Liberty Blue peptide synthesizer was employed to perform the designed automation process. The synthesizer could be used to synthesize peptide without any modification. Meanwhile, it can also be used to do enzymatic synthesis with small modifications (Supporting Information, Table S2).

Glycopeptide 13 was chosen as a target to test our design (Figure 3C). The peptide part of 13 is a repeating unit of a Mucin-type glycoprotein, which is overexpressed on the surface of many cancers.^[21] Many studies indicated that Mucin 1 glycopeptides may be a potential vaccine against cancers. The first amino acid was installed on the resin (Figure 3A) through a cleavable linker (squaric acid diethyl ester). Briefly, 3 (1 g) was reacted with diethylamine (120 mg, 2 mmol) in DMF (10 mL) to form 7. Then, 7 was shaken with Fmoc-Asp-OH (356 mg, 1 mmol), HATU (380 mg, 1 mmol), and DIPEA (129 mg, 1 mmol) in DMF (10 mL) to generate 8. The automated synthesis of the peptide backbone was started from 500 mg of 8 in chemical solution using peptide synthesizer by following the manufacturer instruction. The synthesizer was programmed to add deprotection solution (20% piperidine of DMF solution, 3 mL). After 3 min, the solution was removed by filtration. Deprotection steps were performed twice and the resin was washed three times with DMF (10 mL × 3). Then Fmoc-Pro-OH (2.5 mL, 0.2m), HATU (1 mL, 0.5m) and DIEA (1 mL, 0.5m) were injected into the reaction vessel to couple the second amino acid. The solution was removed by filtration after 10 min. The coupling reaction was performed one more time and the resin was washed three times with DMF (10 mL × 3) before performing for the next reaction cycle. The program was set up to perform nine cycles to produce 9 (Supporting Information, Scheme S12). In the position where glycan epitopes located, artificially synthesized glycol amino acid was used to replace natural Fmoc-Ser(tBu)-OH during reaction cycle (Figure 3C). Tn AA, esterification test of sugar free hydroxy groups was done to demonstrate the amino acid with free sugar could be used for SPPS (Supporting Information, Scheme S11 and Figure S3). 9 was treated with 5% aqueous hydrazine resulting a glycopeptide 10, which was confirmed by HPLC, MS, MS/MS and NMR analysis (Figures S4 and S5).

Figure 3.

A) Conjugation of the first amino acid with aminopropyl silica resin through a cleavable linker. B) Chemoenzymatic synthesis of glycopeptides in a two-phase solution. C) Chemoenzymatic synthesis of compound 13: i) 20% piperidine, 6 min, 50 °C; ii) AA (5 equiv), HATU(5 equiv), DIPEA(5 equiv), 20 min, 50°C; iii) glycotransferase (5 mg LgtB, 6.5 mg FucT, 5 mg GtA), glycan donor (2 equiv UDP-Gal, 2 equiv GDP-Fuc, 2 equiv UDP-GalNAc), 24 h, 37°C

Next, the reaction system was changed to water solution to perform automated extension of glycans by enzyme-catalyzed reaction. Enzymatic synthesis of 13 from 9 requires three glycosyltransferases (LgtB, FucT, and GTA, Figure 3C). Glycosyltransferases, reaction buffer, and sugar nucleotides were stored in the tubes where used for amino acids storage. Automation enzymatic synthesis was started by injecting LgtB (5 mg), UDP-Gal (2 equiv), and reaction buffer (10 mL, pH 8.0). The reaction was performed for 12 h at 37°C. Once the reaction finished, unreacted materials and by-products were removed by filtration. The enzymatic glycosylation reaction was performed twice, and the resin washed three times with water (10 mL × 3) before moving to next reaction cycle. All the steps were performed automatically under the control by a computer program, which was designed for enzymatic reactions. Similarly, reactions catalyzed by FucT and GTA were performed automatically to produce 11. 11 was treated with 5% hydrazine for 12 h and the resulting solution was collected after filtration. After HPLC purification, 10 mg of 13 was obtained in 8.3% yield with regarding to 8. The product was confirmed by HPLC, MS, MS/MS, and NMR analysis (Supporting Information, Figures S8 and S9). In a similar semiautomatic synthetic manner, glycopeptides 12, 14, 15, 16, 17, 18, 19, 20, 38, 39, 40, 38A, 39A and 40A were also successfully prepared but with different glycosyltransferases (Figure 4; Supporting Information, Schemes S12–27). Glycopeptide 10 to 15 containing O-GalNAc glycan Core 3 structure, glycopeptide 16 to 20 contain O-GalNAC glycan core 2 structure glycopeptide 38 (38A) to 40 (40A) contain O-GalNAC glycan core 3 structure with HIV peptide.^[22] All the products were confirmed by HPLC, and MS (Table 1), MS/MS and NMR analysis (Supporting Information, Figures S4–S41). All the products were verified by HRMS within 5 ppm.

Figure 4.

Glycopeptides employed to test the automation system reported herein (compounds 10–20 and 38–40 were obtained using HATU as a coupling reagent, while 38A–40A were synthesized using HATU and HOAT as a coupling reagent).

Table 1:

Analysis of glycopeptide by HRMS.

Entry	Glycopeptide	Formula	Theoretical mass	Charge	Observed m/z	Error [ppm]
1	10	C63H102N16O29	1546.6999	3	516.5758	2.65
2	12	C69H112N18O34	1708.7527	3	570.5934	2.40
3	13	C97H158N18O52	2407.0273	3	803.3528	3.20
4	14	C93H152N16O52	2324.9742	3	776.0017	3.23
5	15	C83H135N17O44	2073.8849	3	692.3038	1.49
6	16	C57H92N16O24	1384.6470	3	462.5574	1.30
7	17	C63H102N16O29	1546.6999	3	516.5756	2.26
8	18	C77H125N17O38	1895.8371	3	632.9555	3.64
9	19	C75H122N16O38	1854.8106	3	619.2802	3.56
10	20	C77H125N17O39	1911.8321	3	638.2873	3.35
11	38	C53H90N12O18	1182.6496	2	592.3336	1.69
12	39	C59H100N12O23	1344.7024	2	673.3601	1.64
13	40	C73H123N13O33	1709.8346	2	855.9276	2.92
14	38A	C53H90N12O18	1182.6496	2	592.3337	1.86
15	39A	C59H100N12O23	1344.7024	2	673.3609	2.83
16	40A	C73H123N13O33	1709.8346	2	855.9260	1.05

In summary, a machine-driven synthesis of glycopeptides based on the use of a peptide synthesizer has been successfully developed. The standardized process includes five steps: automated synthesis of a peptide backbone (with glycan epitope), reaction solution change, automated synthesis of the oligosaccharide, product release, and product purification. By this platform, a Mucin 1 peptide and a HIV peptide decorated with many important oligosaccharide epitopes were successfully prepared in a semiautomated manner. The use of a functionalized silica gel resin overcomes the limitations of traditional solid supports for enzymatic synthesis. It allows both chemical synthesis of peptides and enzymatic synthesis of oligosaccharides to be performed on a single solid support.

This study represents a proof-of-concept that a two-solution system for the synthesis of glycopeptides can be achieved in a semiautomated manner using a commercially available peptide synthesizer. This work will enable an improved instrument for total automation synthesis of glycopeptide in future. Such an improved instrument should include two parts: organic channel for peptide synthesis and water channel for enzymatic synthesis. This will avoid the change of reaction system during the synthetic process and facilitate hands-free automation glycopeptide synthesis.