Synthesis of O‑Sulfated Human Syndecan-1-like Glyco-polypeptides by Incorporating Peptide Ligation and O‑Sulfated Glycopeptide Cassette Strategies

Abstract

A successful synthesis of O-sulfated syndecan-1-like (Q23–E120) glyco-polypeptide was accomplished. The synthesis features the integration of an O-sulfated carbohydrate-bearing glycopeptide cassette with efficient protein ligation strategies, overcoming the acid lability of carbohydrate sulfates as a major hurdle in solid-phase peptide synthesis. Crucial to the synthesis is the microwave-assisted Ag(I) ligation, which afforded the ligation product in improved overall yield. This O-sulfated syndecan-1 (Q23–E120) is the longest O-sulfated glyco-polypeptide prepared to date.

Graphical Abstract

Proteoglycans are a class of complex glycoproteins commonly present on the surface mammalian cells and in the extracellular matrix.¹ They play roles in many important biological events, including growth factor binding, anti-inflammation, and tumor metastasis. Proteoglycans contain one or more sulfated glycosaminoglycan chains covalently attached to the core protein backbone through serine hydroxyl groups in serine–glycine dipeptides.² It has been increasingly appreciated that both the core protein and the glycosaminoglycan can significantly impact the biological activities of proteoglycans.^3,4 Because of the high structural heterogeneity of the glycan chains attached to proteoglycans isolated from nature, it is important to develop synthetic methodologies toward homogeneous proteoglycans.

Over the past decade, tremendous progress has been made in glycoprotein synthesis, especially for N-linked glycoproteins, with multiple full-sized glycoproteins having been synthesized.⁵ In comparison, proteoglycan synthesis has been under-developed, with major hurdles in incorporating the acid-sensitive sulfated glycans into proteins. To date, several synthetic proteoglycan glycopeptides have been reported, with the longest one bearing a decapeptide backbone.^6,7 Therefore, strategies need to be developed to synthesize proteoglycan glycopeptides with a more extended peptide chain.

Herein we report the synthesis of O-sulfated syndecan-1-like glyco-polypeptide 1 with 98 amino acid residues in the peptide backbone corresponding to Q23–E120 of human syndecan-1. Our target sequence 1 spans a functionally active region of the ectodomain of syndecan-1, from the N-terminus Q23 (after the proteolytic cleavage of the signal peptide) to the end of the integrin binding site (E120).⁸ This is the longest O-sulfated glyco-polypeptide prepared to date. Syndecan-1 is a proto typical proteoglycan, which is expressed in high levels by epithelial tumors.⁹ Syndecan-1 affects mesenchymal tumor cell proliferation, adhesion, migration, and motility and modulates genes involved in cell growth regulation. To overcome the challenges in the synthesis of syndecan-1-like glyco-polypep-tide 1, multiple strategies have been investigated. A successful method has finally been established by devising a tetrapeptide carrying the O-sulfated glycan as a cassette for peptide synthesis. The backbone of the glycopeptide was extended with several powerful ligation strategies including native chemical ligation (NCL),¹⁰ serine/threonine ligation (STL),¹¹ and Ag(I)-mediated ligation^12,13 under microwave assistance, leading to glyco-polypeptide 1 with an acid-labile O-sulfate. Our study lays the groundwork toward the chemical synthesis of highly demanding O-sulfated glycoprotein molecules.

The synthetic design of syndecan-1 glyco-polypeptide 1 was based on several considerations. To improve the convergence of the synthesis, peptide ligation strategies were explored. We selected STL because the syndecan-1 sequence is rich in serine and threonine residues. In addition, the NCL method was chosen because it is orthogonal to STL, enabling further peptidyl chain elongation. Thus we designed the strategic disconnections at the G50–A51 for NCL followed by desulfurization and at the S89–T90 for STL, leading to three fragments 2–4 with similar lengths (Figure 1). Glycopeptide 2 contains the O-sulfated xylose moiety. Peptide 3 bears the thiozolidine (Thz)-protected cysteine at the N-terminus for future NCL¹⁴ and a salicylaldehyde ester at the C-terminus for STL with peptide 4.

Figure 1.

Retrosynthetic plan of human syndecan-1 ectodomain (Q23–E120) glyco-polypeptide 1.

Our synthesis started from the preparation of glycopeptide 2. The acid sensitivity of O-sulfate prevents the glycopeptide assembly via standard Fmoc-based solid-phase peptide synthesis (SPPS) due to the need for strong acids for deprotection. Thus we first explored the installation of the O-sulfate post glyco-peptide assembly (Scheme 1A).^15,16 Whereas many sulfation reactions of oligosaccharides have been successfully performed, the direct sulfation of syndecan-1 glycopeptide fragment (Q23–G50) S2 encountered significant difficulties (Table S1). The yields of the desired product were negligible under either basic or acidic conditions, even with heating or microwave irradiation.

Scheme 1.

Synthetic Strategies toward Glycopeptides Bearing a Sulfated Glycan

An alternative strategy to O-sulfate is to use protected sulfate esters such as trichloroethyl (TCE) ester or dichlorovinyl (DCV) ester (Scheme 1B),^17,18 which are known to have enhanced acid stabilities. Such sulfate esters have been employed to install aryl sulfates to synthesize tyrosine sulfated peptides.¹⁹ Recently, we utilized DCV ester to prepare syndecan-4 glyco-decapeptide.⁴ Unfortunately, although the glycopeptide S5 with DCV-protected sulfate ester was obtained, the glycopeptide unexpectedly decomposed during the deprotection of the DCV sulfate (Scheme S1).

The difficulties encountered prompted us to explore an alternative cassette approach (Scheme 1C). A short glycotetrapeptide (S47–G50) 5 was prepared as a cassette (Scheme 2), on which glycan manipulation and sulfation would be carried out. Subsequently, the sulfated glycopeptide cassette would be ligated to yield the glyco-polypeptide.

Scheme 2.

Preparation of Glycopeptide Cassette 5

The preparation of tetrapeptide 5 started by the reaction of t-butyldimethylsilyl (TBS)-protected xylosyl donor 6 with the Fmoc serine 7 in 82% yield (Scheme 2). The simultaneous removal of the TBS and the p-methoxybenzyl (PMB) groups from 8 by trifluoroacetic acid (TFA) gave a free carboxylic acid intermediate, which was coupled to tripeptide 9, generating glycopeptide 10 in 91% yield for the two steps. The sulfation of the free hydroxyl group in 10 with the Et₃N·SO₃ complex proceeded smoothly, leading to the sulfated glycopeptide 11 in 82% yield, the Fmoc group of which was then removed by piperidine, producing cassette 5. The successful preparation of 11 suggested that the aforementioned difficulty encountered in the sulfation of glycopeptide fragment (Q23–G50) S2 was likely due to the long peptide at the reducing end of the xylose, which potentially shielded the free hydroxyl group of the xylose, reducing its nucleophilicity.

To extend the glycopeptide, we first attempted to prepare the peptide fragment (Q23–G46) bearing a free carboxylic acid at the C-terminus for direct coupling to the cassette 5 (Scheme 3A). Aspartimide was found to form during the peptide synthesis of fragment (Q23–G46). Thus pseudopro-line^20,21 dipeptide building blocks were utilized to minimize this side reaction, leading to the peptide in good yield. (See the Supporting Information for details.) With the peptide (Q23–G46) in hand, its coupling to 5 was tested. Unfortunately, little desired glycopeptide product was obtained with a variety of peptide coupling agents, including 1-[bis(dimethylamino)-methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexa-fluorophosphate (HATU) and 1-ethyl-3-(3-dimethylamino propyl)carbodiimide (EDC) (Table S2). As an alternative, we explored the Ag(I)-mediated ligation¹³ to facilitate this peptide bond formation, where Ag(I) selectively activated the C-terminal thioester reacting with the N-terminal free amine of the glycopeptide cassette (Scheme 3B). The mild condition of Ag(I)-mediated ligation can potentially reduce the side reactions such as the loss of the O-sulfate or the elimination of the glycosyl chain often observed in sulfated glycopeptide synthesis.^6,22

Scheme 3.

Glycopeptide Cassette Approach

The Ag(I)-mediated ligation was investigated between Q23–G46 thioester 12 and the glycopeptide cassette 5. The DMF/THF mixed-solvent system was used in our initial trial.²³ With AgNO₃ as the activator in the presence of 3-hydroxy-1,2,3-benzotriazin-4-one (HOOBt) as an additive and ⁱPr₂NEt as a base, the formation of product 13 was not detected by electrospray ionization mass spectrometry (ESI-MS), whereas the thioester 12 gradually hydrolyzed (Table 1, entry 1). Elongation of the reaction time led to an increase in the side product aspartimide formation from 12. To enhance the nucleophilicity of the N-terminal amine to favor peptide coupling, we removed the electron-withdrawing Bz protection from the xyloside to give tetrapeptide 5a. However, that did not improve the yield of the desired product (entry 2). When the reaction was performed in an aqueous NaHCO₃ solution with N-hydroxysuccinimide (NHS) as the additive (entry 3),¹² the corresponding NHS ester was observed as a major product within 0.5 h, which gradually underwent hydrolysis, indicating the poor reactivity of the amine. Switching the reaction solvent to DMSO gave rise to the desired product, albeit with a low yield (<10%) (entries 4–6), which was not further improved with the varied stoichiometry of either 5 or 5a.

Table 1.

Exploration of Ag(I)-Mediated Ligation^a

entry	solvent (v/v)	cassette	Ag (I)	additive	product (%)
room temperature, 24 h
1	DMF/THF (1/2)	5	AgNO3	HOOBt	n.d.
2	DMF/THF (1/2)	5a	AgNO3	HOOBt	n.d.
3	aq. NaHCO3, pH 8	5a	AgNO3	NHS	n.d.
4	DMSO	5a	AgNO3	HOOBt	<10
5b	DMSO	5a	AgNO3	HOOBt	<10
6	DMSO	5	AgNO3	HOOBt	<10
microwave conditions: 50 °C, 20 min
7	DMF/THF (1/2)	5	AgNO3	HOOBt	30
8	DMSO/dioxane (10/1)	5	AgNO3	HOOBt	16
9c	DMSO/acetate buffer (pH 5.5) (1/1)	5	AgNO3	HOOBt	21
10	DMSO	5	AgSbF6	HOOBt	33
11	DMSO	5	AgSCN	HOOBt	16
12	DMSO	5	AgPF6	HOOBt	53
13	DMSO	5	AgBF4	HOOBt	32
14	DMSO	5	AgNO3	HOOBt	75

^aReaction was performed with cassette 5 or 5a in excess (3 equiv). The concentration of peptide thioester 12 was 3–5 mM at room temperature and 1 mM under microwave conditions. The formation of 13 was calculated based on the HPLC integration of UV absorbance at 220 nm. n.d. = not detected.

^bGlycopeptide 5a was added in 0.67 equiv.

^cNo iPr₂NEt was added.

To promote this peptide coupling, we decided to test microwave irradiation to the Ag(I) ligation reaction. Under microwave conditions at 50 °C for 20 min in THF/DMF, 30% of the desired product 13 was observed based on HPLC analysis (entry 7). Other solvent systems, including the acidic aqueous buffered condition,²⁴ did not improve the reaction (entries 8 and 9). DMSO appeared to be the best solvent (entries 10–14). The effect of various Ag(I) salts was investigated next, among which the best result was obtained with AgNO₃ in DMSO, leading to 75% yield of the desired glycopeptide 13 (entry 7). It is worth mentioning that prolonged treatment should be avoided, which led to aspartimide product hardly separable from the desired product (Figure S15).

In parallel, STL was carried out with peptide fragment 3 salicylaldehyde ester and fragment 4 bearing an N-terminal threonine. The ligation yielded the desired product, the cysteine of which was deprotected, yielding peptide 14 (Scheme 4).

Scheme 4.

Preparation of Fragment 14 via STL

For the union of peptide 14 and glycopeptide 13, we resorted to NCL chemistry. To activate glycopeptide 13 for NCL, hydrazinolysis of the C-terminus methyl ester of 13 was the next key step.²⁵ Surprisingly, the treatment of 13 with a hydrazine/MeOH mixture⁶ resulted in a slow reaction and a major side product (around 30%) with an increase of 15 Da in molecular weight (Table 2, entry 1). This indicated that the hydrazinolysis likely occurred on the side-chain amide residues as well, such as Asn23, Asn34, Gln28, or Gln43, to form a side-chain acyl hydrazide (Scheme S2). The addition of 20% water to the reaction significantly mitigated the side reaction, which, on the contrary, led to hydrolysis of the glycopeptide methyl ester 13 (entry 2). After multiple trials, it was discovered that the amount of water added was critical to providing an expedited reaction rate while suppressing the formation of the side-chain hydrazide side product (entries 3 and 4). With a ratio of 80:1 MeOH/water (entry 3), an optimal outcome was obtained (complete within 0.5 h, 77% isolation yield after HPLC purification), which not only formed the desired hydrazide product 15 but also removed the benzoyl protective groups from the xylose ring.

Table 2.

Optimization of Hydrazinolysis^a

entry	MeOH/H2O	time (h)	major side product formed
1	MeOH only	12	side-chain acyl hydrazide
2	80/20	12	hydrolysis of methyl ester
3	80/1	0.5	none
4	80/1	4.5	side-chain acyl hydrazide

^aConcentrations: peptide 13 (10 mM), N₂H₄·H₂O (4 M).

The acyl hydrazide 15 was subjected to nitrous acid treatment for hydrazide activation.²⁵ Caution was taken to perform the reaction at −15 °C, pH 3 to prevent the loss of O-sulfate from 15 due to acid treatment (Figure 2A). The resulting acyl azide intermediate was incubated in situ with 4-mercaptophenylacetic acid (MPAA) to form the thioester followed by NCL with peptide fragment 14 (C51–E120) in one pot. The desired ligation product 16 was obtained with high conversion, as judged by HPLC and ESI-MS (Figure 2A). The reaction was highly reproducible, with two batches giving 35 to 36% yields for the three steps from acyl hydrazide 15. With 16 in hand, P–B desulfurization²⁶ was performed, which successfully produced the syndecan-1-like glyco-polypeptide 1. The structure and the purity of product 1 were supported by ¹H NMR, HPLC, and MS analysis (Figure 2 and Figures S9–S14).

Figure 2.

(A) Final assembly of syndecan-1-like glyco-polypeptide 1. (B) HPLC and (C) ESI-MS analysis of 1. (See Figure S9 for details.) (D) ¹H NMR analysis of glyco-polypeptide 1 was performed at 298 K in D₂O. See Figures S10–S14 for 2D NMR spectra.

In conclusion, we successfully synthesized the syndecan-1-like glyco-polypeptide corresponding to Q23–E120 bearing an O-sulfated carbohydrate. This sulfated glycopeptide has the longest polypeptide chain prepared to date. Our approach features an effective integration of the glycopeptide cassette bearing a short tetrapeptide with high-efficiency protein chemical ligation technologies (Ag(I)-mediated ligation, NCL, and STL). The rationally designed glycopeptide cassette enabled the facile glycan manipulation and incorporation of acid-labile O-sulfates, whereas the microwave-assisted Ag(I)-mediated ligation was critical in overcoming the low nucleophilicity of the free N-terminus of the sulfated glycopeptide cassette in peptide bond formation. Both Ag(I)-mediated ligation and NCL were compatible with the O-sulfated glycan. Whereas glyco-polypeptide 1 only bears a xylose unit, it can provide valuable knowledge toward the synthesis of O-sulfated glycopeptides approaching the complexities of naturally existing proteoglycan.