Synthesis and Thermodynamic Evaluation of Sialyl-Tn MUC1 Glycopeptides Binding to Macrophage Galactose-Type Lectin

Abstract

Interactions between the tumor-associated carbohydrate antigens of Mucin 1 (MUC1) and the carbohydrate-binding proteins, lectins, often lead to the creation of a pro-tumor microenvironment favoring tumor initiation, progression, metastasis, and immune evasion. Macrophage galactose binding lectin (MGL) is a C-type lectin receptor found on antigen-presenting cells that facilitates the uptake of carbohydrate antigens for antigen presentation, modulating the immune response homeostasis, autoimmunity, and cancer. Considering the crucial role of tumor-associated forms of MUC1 and MGL in tumor immunology, a thorough understanding of their binding interaction is essential for it to be exploited for cancer vaccine strategies. The synthesis of MUC1 glycopeptide models carrying a single or multiple Tn and/or sialyl-Tn antigen(s) is described. A novel approach for the sialyl-Tn threonine building block suitable for the solid phase peptide synthesis was developed. The thermodynamic profile of the binding interaction between the human MGL and MUC1 glycopeptide models was analyzed using isothermal titration calorimetry. The measured dissociation constants for the sialyl-Tn-bearing peptide epitopes were consistently lower compared to the Tn antigen and ranged from 10 μM for mono- to 100 nM for triglycosylated MUC1 peptide, respectively. All studied interactions, regardless of the glycan’s site of attachment or density, exhibited enthalpy-driven thermodynamics.

Graphical Abstract

We have synthesized cancer-associated MUC1 glycopeptide models to explore the interactions between Tn and sialyl-Tn antigens and the human MGL lectin at the molecular level. The thermodynamic profiles of the binding interactions were analyzed by ITC. Findings from this study will shed more light on human MGL’s ability to regulate immune responses.

Introduction

Mucin 1 (MUC1) O-glycosylation is commonly truncated in malignant and premalignant epithelia.^[1–3] The human C-type lectin receptor, macrophage galactose-type lectin (hMGL), expressed by both dendritic cells and macrophages, has been identified as a receptor of Thomsen-nouvelle (Tn, CD175) and sialylated Tn (sialyl-Tn, CD175s).^[4,5] The interaction between Tn and sialyl-Tn antigens of tumor-associated MUC1 and hMGL likely plays a critical role in establishing the immunological tolerance to tumor cells.^[6] Yet, MGL targeting is an attractive approach to enhance cancer vaccine efficacy.^[7] The ability of hMGL to regulate immune response by engaging tumor-associated MUC1 to either augment the anti-tumorigenic effects or minimize the pro-tumorigenic effects is poorly understood. Zizzari et al.^[8] implied that the nature of the hMGL ligand may play a crucial role in shaping the immune responses.

We have recently studied the recognition of synthetic Tn antigen presenting MUC1 glycopeptides by hMGL by Isothermal Titration Calorimetry (ITC).^[9] An affinity enhancement of one order of magnitude with increasing glycan valency was observed. A shift in enthalpy-entropy compensations and contact position-specific effects, with the likely involvement of the peptide surroundings, were detected.^[9] KinITC analysis revealed a prolonged lifetime of the lectin-glycan complex with increasing glycan valency.^[9]

In this paper, we expanded the study to the interaction of the sialyl-Tn antigen presenting MUC1 glycopeptides by hMGL. This required the synthesis of the sialyl-Tn-bearing glycosylamino acid building block suitable for Fmoc SPPS of MUC1 glycopeptides. Alternatively, the synthesis of sialyl-Tn-MUC1 glycopeptides could be achieved by chemoenzymatic approaches that requires use of highly efficient α2–6-sialyltransferases.^[10–13] The all-chemical approach, undeniably presents several challenges too. The sialic acid’s unique structure, such as the hindered tertiary anomeric center at C-2 carbon, the presence of an electron-withdrawing carboxyl group at C-1 carbon, and the lack of a neighboring participating group at C-3 carbon lead to low yields and poor stereoselectivity of the glycosylation reaction.^[14–17] The early approaches towards the synthesis of α-glycosides suitable for SPPS of sialylglycopeptides involved the use of sialyl donors such as xanthate in the presence of methylsulfenyl triflate as the activator^[18–20] or diethyl phosphite donor.^[21,22] Alkyl (aryl) thioglycosides have emerged as versatile sialyl donors that can act as an anomeric protecting group and, at the same time, can be activated under mild conditions using a variety of promoters.^[16,23] A thioglycoside approach has been recently applied to synthesizing suitably protected 2,6-sialylated building blocks for preparation of tumor-associated MUC1 glycopeptide fragments.^[24] Using nitrile-based solvent^[25,26] and 5-N,4-O-oxazolidinone directing auxiliary^[27] successfully promoted α-selectivity and minimized competing 2,3-elimination reactions on the sialyl donor during activation. The glycosylated serine-and threonine-bearing sialyl-Tn and sialyl-T antigens were obtained in high yields (>80%) and significantly improved stereoselectivity, receiving almost exclusively α-sialosides.^[24] These building blocks were successfully coupled to the resin with minimally observed racemization during coupling conditions. However, Fmoc deprotection conditions were not orthogonal to the 5-N,4-O-oxazolidinone group, and on-resin deprotection was required before further elongation of the glycopeptide chain. Similarly, C-1 allyl ester of 1-adamantylthio sialoside that contained an N-acetyl-5-N-4-O-oxazolidinone gave excellent yields and α-selectivity in glycosylation reaction with the galactosyl receptor.^[28] Nevertheless, this approach required a two-step sidechain deprotection protocol upon the completion of the targeted glycolipopeptides, the on-resin Pd (PPh₃)₄ catalyzed allyl ester deprotection step, followed by deprotection of acetyl groups in 60% hydrazine in methanol. Thus, the alternate orthogonal protecting strategy is still of paramount importance.

Herein, we present a novel, fully orthogonal approach for synthesizing the α-2,6-sialyl-Tn Thr building block, resulting in a good overall yield and α-anomeric stereoselectivity. This building block was required to synthesize MUC1 glycopeptides models carrying one or two sialyl-Tn antigens at position Thr⁴, Thr⁹, and/or Thr¹⁶. To examine epitope heterogeneity, the glycoside cluster effect, and the steric hindrance effect of neighboring glycans on binding to hMGL, a mono- and multiple-glycosylated MUC1 peptide models carrying the sialyl-Tn and/or Tn O-glycans were prepared. The binding affinities and the thermodynamic profile of MUC1 glycopeptides binding to hMGL were determined by ITC.

Results and Discussion

Synthesis of the α-2,6-sialyl-Tn Thr Building Block

Our approach to synthesizing the α-2,6-sialyl-Tn Thr building block, including the installment of temporary and permanent protecting groups for the carbohydrate and the amino acid functional groups, is shown in Scheme 1. In this approach, we used phenylthiosialoside 12 as sialyl donor and the C6-OH Thr-Tn building block 8 as glycosyl acceptor. We performed glycosylation under N-iodosuccinimide/Trimethylsilyl trifluoromethanesulfonate (NIS/TMSOTf) activation conditions. In addition, this strategy relies on the efficiency of the pentafluorophenyl esters in coupling the glycosylated amino acid building blocks and the use of common cleavage/deprotection steps, acid followed by the base treatment for the side chain peptide and sugar moiety deprotection.

Scheme 1.

Synthesis of glycosylated amino acid building block 13. Full experimental procedures and compound characterization is provided in Supporting Information (Page S1-S19, Figure S1–S25).

Zemplen deacetylation^[29] of the phenyl thioglycoside 1^[30] in basic conditions (pH 8.5–9.5) using a fresh sodium methanolate solution resulted in 2 (97% yield).^[30] The ¹H NMR and ¹³C NMR spectra of compound 2 are provided in the Supporting Information (Figures S1−S2). Subsequently, regioselective 6-silyl protection via treatment with tert-butyldiphenylsilyl chloride (TBDPS-Cl) and imidazole gave thioglycoside 3 in 83% yield. The NMR characterization of the purified compound is given in the Supporting Information (Figures S3–S4). The acid labile isopropylidene group was chosen to protect the C3- and C4-OH side chains. Compound 4 was obtained in good yield after purification by silica gel column chromatography (87%). Figures S5−S6 (Supporting Information) show the respective ¹H NMR and ¹³C NMR spectra. The glycosylation of Fmoc-Thr-OPfp with the donor 4 in the presence of NIS/trifluoromethanesulfonic acid (TfOH) as a promoter system resulted in the orthogonally protected thiophenyl glycoside 6 in excellent yield (80%, α-anomer: 75%; β-anomer: 5.0%) and stereoselectivity (15:1). This can be attributed to the bulky silyl groups at the C6, affecting the conformation of the sugar ring, and concomitantly favoring attack from the alpha side.^[31] The desired α-anomer was effectively separated from the unwanted β-anomer by column chromatography on silica gel. The formation of the α-O-glycosidic linkage was confirmed by measuring the characteristic anomeric doublet at 5.02 ppm (β-proton should be at 4.42 ppm) and 99.0 ppm for the anomeric C-1α (Figures S7 and S8, Supporting Information). The buffered fluoridolysis was employed for the selective deprotection of the TBDPS group at C6.^[24] Lowering the basicity of TBAF with the AcOH to pH~8 ensured the stability of the Fmoc group, but not the OPfp ester. Thus, without further purification, the carboxyl group of 7 was converted back to the OPfp ester yielding the glycosyl acceptor 8 in 74% yield over two steps. Figures S9−S10 (Supporting Information) depict the respective ¹H NMR and ¹³C NMR spectra.

A phenylthiosialoside donor 12 was synthesized in four linear steps using commercially available N-acetylneuraminic acid (Neu5Ac, 9). The initial step of the synthesis was transforming the carboxylic acid group at position C-1 into a methyl ester 10.^[32,33] Amberlite IR-120 (H⁺) resin beads were used as an acid catalyst. Acetylation with acetic anhydride-pyridine mixture resulted in peracetylated compound 11 (65% yield, α-anomer: 15%; β-anomer 50%).^[32,34] Without the anomeric mixture separation, the addition of a thiophenol group in the presence of boron trifluoride led to the replacement of the acetoxy group located on C-1 by the thiophenol, forming the final product, phenyl-β-1-thiogalactoside 12 in 64% yield.^[33–35] The overall yield over four steps was 51%. Separation of anomers of peracetylated compound 11 by column chromatography on silica gel, followed by the thiophenol reaction with the β-anomer, resulted in a higher yield of thiogalactoside 12 reaction step (83% yield) and the total yield of 41%. ¹H NMR and ¹³C NMR spectra are provided in the Supporting Information (Figures S11–S16).

Accessing a single anomer and competing 2,3-elimination processes on the sialyl donor during activation are two major challenges in the preparation of α-sialoside.^[16,36] Our proposed approach relied on the high efficiency of the thioglycoside donor in glycosylation reaction and the stereodirecting effect of a nitrile solvent in promoting α-selectivity.^[18,37–43] We explored the impact of activator (TMSOTf/NIS or TfOH/NIS), temperature (−45 to 0 °C), solvent, and the excess equivalent of sialic acid thioglycoside donor, 12. Reaction conditions were most favorable at −20 °C with the solvent system dichloromethane:acetonitrile (DCM:ACN;1:1, v/v), 2 equivalents excess of sialic acid thioglycoside donor, and 0.5 equivalents of TMSOTf as an activator. The glycosylation of acceptor 8 with a phenylthiosialoside donor 12 under these conditions gave the sialylated glycosyl amino acid 13 (Scheme 1) in relatively high yield (72%, α-anomer: 62%; β-anomer: 10%) and good stereoselectivity (α:β=6:1). Performing the reaction in DCM, results in equal amounts of the α- and β-anomer, further confirming the critical role of nitrile effect in the stereoselectivity of the reaction (Figure S17, Supporting Information). Anomers were separated by column chromatography on silica gel and their purity was confirmed by ¹H NMR and ¹³C NMR spectra (Figure S18–S23, Supporting Information). The anomeric configuration of sialoside 13 was determined by analysis of chemical shift and coupling constant (H9a-H9b, H7 to H8, H4 and H3) and by comparison with the NMR data of known compounds.^[44–48] The distinctive δ α H-3 eq was observed at 2.62 ppm; correspondingly, the δ β H-3 eq was observed at 2.52 ppm, more upfield than the α H-3 eq.^[44–46] The α-anomer was further confirmed by the detected value of ~8.0 Hz for J7′,8′ coupling constant.^[47,48] According to ¹³C NMR analysis, the methyl ester C-1′ carbon for the α- and β- anomers was found to be at 168.0 ppm (which verifies the α-anomer) and 167.0 ppm (which verifies the β- anomer), respectively.^[44] Furthermore, the α H-4′ signal was detected at 4.89 ppm, indicating that the expected range for α-anomer is 4.89 to 4.93 ppm.^[47] α-Anomer is further confirmed by the difference between H9a and H9b [Δδ (H9a-HH9b)], which is less than 0.5 ppm.^[48] The NMR spectra (¹H and ¹³C), HPLC chromatograms, and MALDI-TOF MS are shown in Supporting Information (Figures S18–S25).

Synthesis and purification of the MUC1 (glyco)peptides

A panel of nine MUC1 (glyco)peptides 14-23, ranging from the single-site to triple glycosylated scaffolds, were synthesized using Fmoc SPPS (Table 1, and Supporting Information Figure S26, and Figures S27–S46), together with the non-glycosylated 20-mer MUC1 tandem repeat control peptide 14. In an iterative procedure on an automated peptide synthesizer, Fmoc-protected amino acids with HCTU/HOBt activation were employed in equimolar quantities (4-fold excess) (Figures S26, Supporting Information). The coupling of the orthogonally protected α-linked GalNAc building block^[9,30] and α-linked sialylated GalNAc building block 13 in a 1.5 equiv excess led to three monoglycosylated peptides 15–17, three double glycosylated peptides carrying either two sialyl-Tn antigens or the combination of the sialyl-Tn and Tn antigen 18–22, and one triple glycosylated peptide 23 (Table 1). After completion of the (glyco)peptides chain synthesis, the resin was treated with thioacetic acid for direct one-pot reductive acetylation of the azido group on the sugar moiety.^[49] The glycopeptides cleavage from the resin was achieved under acidic conditions (water/thioanisole/TFA). These conditions efficiently removed amino acids side chain protecting groups and the isopropylidene group from the sugar moiety. The two-step protocol was used for the deprotection of the O-acetyl and O-methyl protecting groups to circumvent problems during base hydrolysis, such as β-elimination.^[50,51] The O-acetyl groups were deprotected first with methanolic sodium methoxide (0.1 M NaOMe) followed by the 0.1 M aqueous NaOH respectively, conditions that ensured the methyl ester deprotection (Supporting Information, Figures S47–S50). The prepared (glyco)peptides were processed by RP-HPLC, and their purity was determined by analytical RP-HPLC and MALDI-TOF MS (Table 1 and Supporting Information, Figures S27–S46).

Table 1.

SPPS of MUC1 (glyco)peptides and their characterization by RP-HPLC and MALDI-TOF MS.

Peptide #	Amino acid sequence[a]	No. of sugars	HPLC	MALDI-TOFMS [M+H]+[c]
			tR[b] (min)	Expected (Da)	Observed (Da)
14	HGVTSAPDTRPAPGSTAPPA	0	21.09	1884.93	1883.68
15	HGVT**SAPDTRPAPGSTAPPA	1	19.59	2379.41	2379.19
16	HGVTSAPDT**RPAPGSTAPPA	1	19.96	2379.41	2379.24
17	HGVTSAPDTRPAPGST**APPA	1	20.50	2379.41	2379.72
18	HGVT**S*APDTRPAPGSTAPPA	2	18.62	2582.51	2582.61
19	HGVTSAPDTRPAPGS*T**APPA	2	18.45	2582.51	2582.13
20	HGVT**SAPDTRPAPGST**APPA	2	17.82	2873.31	2873.30
21	HGVT*S*APDTRPAPGSTAPPA	2	19.05	2291.49	2291.31
22	HGVTSAPDTRPAPGS*T*APPA	2	18.90	2291.53	2291.31
23	HGVT*SAPDT**RPAPGST*APPA	3	17.94	2785.32	2785.46

^[a]T** = Thr O-linked sialyl-Tn and T*/S* = Thr/Ser O-linked Tn.

^[b]HPLC analysis conditions can be found in Figures S27–S46.

^[c]The matrix used was α-cyano-4-hydroxycinnamic acid (Figures S27–S46).

As expected, the control peptide 14 displayed the highest retention time (t_R). Adding a single (hydrophilic) glycan moiety at either Thr⁴, Thr⁹, or Thr¹⁶ decreased t_R by 0.5 to 1.0 min (on average). Adding the second or third sugar further increased the peptides’ hydrophilicity (Table 1). For example, the triglycosylated peptide 23 eluted 3.2 min earlier from the column than the non-glycosylated peptide 14 (Table 1). Notably, diglycosylated peptide carrying two sialyl-Tn antigens 20 has a lower t_R compared to the MUC1 diglycopeptides that carry Tn/sialyl-Tn antigens 18 and 19 or two Tn antigens 21 and 22, respectively (Table 1). With the (glyco)peptides 14-23 in hand, their secondary structure could be investigated by circular dichroism (CD) spectroscopy.

CD analysis of the MUC1 (glyco)peptides

The CD analysis revealed no significant differences in the secondary structure amongst glycopeptides (Supporting Information, Figure S51). The presence of a random coil with polyproline II (PPII) helical components revealed the spectral molar ellipticity ([θ]) minima at 198 nm and the increased maximum ([θ]max) intensity at 222 nm (Supporting Information, Figure S51).^[52] Notably, glycan’s density and nature positively correlate with the degree of structured element presence. For example, compared to the diglycosylated MUC1 carrying Tn antigen 21 or 22, the diglycosylated MUC1 with sialyl-Tn antigen 20 had greater PPII content, followed by the triglycosylated peptide 23 that had the highest intensity at 222 nm (Figure S51). These outcomes agree with the corresponding decrease in the retention time on RP-HPLC for glycopetides 20 and 23 (Table 1). The next step would be to use this panel of glycopeptides as ligands for the endogenous lectin.

ITC analysis of the MUC1-Based (Glyco)peptides

We have previously evaluated the panel of Tn antigen-presenting MUC1 glycopeptides, covering the range from single site to fully glycosylated scaffolds, for binding to MGL to assess the role of the site of attachment, peptide epitope and multivalences on glycan-lectin interactions.^[9] In the current study, we have expanded the panel with the sialyl-Tn presenting MUC1 glycopeptides to evaluate the potential differences between Tn and sialyl-Tn antigen and the role of neighboring glycosylation in binding to MGL.

In order to determine the enthalpic and entropic contributions to binding for the interaction between hMGL and MUC1 glycopeptides, the ITC analysis of the binding affinities was performed. N-acetylgalactosamine (GalNAc, Tn antigen) served as an internal reference for the hMGL activity, which recognizes Tn antigen due to its exposed hydroxyl groups at C3 and C4.^[9,53,54] The obtained thermodynamic profile for the GalNAc is very similar to the one we previously reported.^[9] As expected, the non-glycosylated MUC1 peptide shows no activity.

The thermodynamic profiles and the measured K_D values for the interaction of hMGL with a series of MUC1 glycopeptides carrying the sialyl-Tn and/or Tn antigen are summarized in Table 2. The acquired stoichiometries (n-value, Table 2) correspond to titrations that reached complete saturation of the hMGL receptor, thus revealing the MUC1 glycopeptides functional valency (1/n). Similar to our previous data,^[9] under saturation conditions, a maximum of three glycans can bind to the trimeric hMGL protein.

Table 2.

ITC data for association of hMGL with GalNAc and MUC1 Glycopeptides in 10 mM HEPES sodium salt, 50 mM NaCl, and 2 mM CaCl2 (pH 7.4).^[b]

Peptide #	Ligand	KA (×104 M−1)	ΔG (kcal mol−1)	ΔH (kcal mol−1)	−TΔS (kcal mol−1)	n	KD (μM)
	GalNAc	6.80	−6.59	−11.60	4.97	0.99	14.70
15	HGVT**SAPDTRPAPGSTAPPA	8.77	−6.75	−10.60	3.84	0.98	11.40
16	HGVTSAPDT**RPAPGSTAPPA	9.80	−6.81	−10.90	4.13	0.98	10.20
17	HGVTSAPDTRPAPGST**APPA	10.42	−6.85	−10.50	3.68	0.99	9.59
18	HGVT**S*APDTRPAPGSTAPPA	33.78	−7.56	−20.80	13.20	0.55	2.96
19	HGVTSAPDTRPAPGS*T**APPA	34.10	−7.55	−20.20	12.60	0.55	2.93
20	HGVT**SAPDTRPAPGST**APPA	20.45	−7.19	−19.20	12.00	0.54	4.89
21	HGVT*S*APDTRPAPGSTAPPA	49.01	−7.77	−19.80	12.00	0.54	2.04
22	HGVTSAPDTRPAPGS*T*APPA	52.08	−7.80	−18.00	10.20	0.53	1.92
23	HGVT*SAPDT**RPAPGST*APPA	85.47	−8.09	−29.90	21.80	0.33	1.17

^[a]T** = Thr O-linked sialyl-Tn and T*/S* = Thr/Ser O-linked Tn.

^[b]Errors in ΔH ranged between ±0.13 and 0.52 kcal mol^–1 and between ±0.04 and 0.59 μM/experiment for K_D. The error values and concentrations of the ligand and lectin for each experiment are provided in the Supporting Information along with the corresponding thermograms (Figures S52–S55).

According to ITC analysis, all three mono-glycosylated MUC1 glycopeptides carrying sialyl-Tn antigen at Thr⁴, Thr⁹, or Thr¹⁶ (15, 16, or 17, respectively) had a lower K_D value for hMGL compared to GalNAc (Table 2, Figure 1A GalNAc: 14.70 μM and 1B 16: 10.20 μM, respectively). Although the affinity of sialyl-Tn for hMGL was better when presented on the MUC1 backbone, it was slightly worse compared to Tn antigen (Table 2, Thr⁴(sialyl-Tn) 15: 11.40 μM and Thr⁴-Tn: 6.82 μM^[9]). In addition, it did not seem that the site of attachment of the sialyl-Tn antigen (Thr⁴, Thr⁹, or Thr¹⁶) had an impact on the affinity of the glycopeptide for hMGL (Table 2). These findings imply that the nature of the microenvironment of each carbohydrate residue (at Thr⁴, Thr⁹, or Thr¹⁶) has only a minor influence under these conditions.

Figure 1.

ITC titration profiles for the interaction of (A) GalNAc (1.50 mM) with hMGL (25.0 μM), (B) MUC1-Thr⁹(sialyl-Tn) 16 (0.50 mM) with hMGL (15 μM), (C) MUC1-Thr^4,16(sialyl-Tn) 20 (0.50 mM) with hMGL (10 μM), and (D) MUC1-Thr⁹(sialyl-Tn), Thr^4,16(Tn) 23 (0.50 mM) with hMGL (16 μM) in buffer containing 10 mM HEPES sodium salt, 50 mM NaCl, and 2 mM CaCl₂ (pH 7.4). Injections of ligands were performed every 160 s at 298 K. The top panels show the experimental ITC data, and the bottom panels fit to a one-site model of the binding data using MicroCal PEAQ-ITC analysis software. Resulting values for the stoichiometry (n), binding affinity (K_a), dissociation constant (K_d), enthalpy (ΔH), and change in entropy with respect to temperature (TΔS) are shown in Table 2 and Supporting Information (Figure S52–S55).

To further investigate the role of the multivalency and neighboring glycosylation effect on hMGL binding, we have determined the ITC binding profile for di- and triglycosylated MUC1 glycopeptides bearing only sialyl-Tn antigen or mixture of sialyl-Tn and Tn antigens (Table 2).

In comparison to monoglycosylated sialyl-Tn peptides, the K_D of the diglycosylated MUC1 peptides carrying two sialyl-Tn antigens was approximately two times (20: 4.89 μM, Figure 1C) and three times lower for the diglycosylated peptides carrying sialyl-Tn and Tn antigens (18: 2.96 μM; 19: 2.93 μM, Figure 2A and 2C), respectively. Based on the observed binding affinities, we can conclude that the presence of the sialyl-Tn antigens, at Thr⁴ or at Thr¹⁶, to the neighboring Tn antigen, at Ser⁵ or Ser¹⁵, did not significantly affect affinity for hMGL (Table 2). Likewise, binding of diglycosylated MUC1 peptides 18 and 19 was similar in affinity to 21 and 22 (Table 2, Figure 2B and 2D). Similar to our findings, the neighboring glycosylation on Ser/Thr within the GVTS and GSTA epitopes did not disturb the binding to human MGL.^[55] This may suggest that the adjacent Tn or sialyl-Tn antigen is suitably accommodated by MGL.^[56] The triglycosylated peptide, carrying the Tn antigens at Thr⁴ and Thr¹⁶ and the sialyl-Tn antigen at Thr⁹ positions (23: 1.17 μM, Figure 1D), obtained the lowest K_D value, around ten times lower than that of mono-glycosylated sialyl-Tn peptides.

Figure 2.

Selected ITC binding curves, including thermographic profiles for the interaction of (A) MUC1-Thr⁴(sialyl-Tn), Ser⁵(Tn) 18 (0.50 mM) with hMGL (10 μM), (B) MUC1-Thr⁴(Tn), Ser⁵(Tn) 21 (0.50 mM) with hMGL (15 μm) (C) MUC1-Ser¹⁵(Tn), Thr¹⁶(sialyl-Tn) 19 (0.50 mM) with hMGL (10 μm) (D) MUC1-Ser¹⁵(Tn), Thr¹⁶(Tn) 22 (0.50 mM) with hMGL (15 μm) in buffer containing 10 mM HEPES sodium salt, 50 mM NaCl, and 2 mM CaCl₂ (pH 7.4). Injections of ligand were performed every 160 s at 298 K. The top panels show the experimental ITC data and the bottom panels fit to a one-site model of the binding data using MicroCal PEAQ-ITC analysis software. Resulting values for the stoichiometry (n), binding affinity (K_a), dissociation constant (K_d), enthalpy (ΔH), and change in entropy with respect to temperature (TΔS) are shown in Table 2 and Supporting Information (Figure S52–S55).

As to the binding of hMGL to MUC1-Tn glycopeptide models with increasing valency,^[9] the characteristic thermodynamic profiles of hMGL binding also result in a proportional increase in ΔH and enhanced entropic penalties when applying corresponding MUC1-sialyl-Tn glycopeptide models (Table 2). The evaluation of multivalent binding revealed negative cooperativity with cooperativity coefficient α values < 1. The constant, α, was calculated using the equation ΔG^mul,n = αnΔG^mono (Table 3).^[57] However, the affinity enhancement factor β^[58] (K_A^mul,n/K_A^mono) confirmed the involvement of multivalent avidity effects in enhancing the binding affinity as usually observed for the multivalent interactions in biological systems (Table 3). Further look into the enthalpic and entropic contributions of the monomeric interaction to the total enthalpy (ΔH^mul,n = αnΔH^mono) and entropy (ΔS^mul,n = αnΔS^mono) change implies the entropically enhanced multivalent interaction (Table 3).^[57] The calculated α factor for enthalpy contribution was equal or close to 1 (Table 3). Thus, the enthalpy is scaling proportionally to the number of binding epitopes. However, in the case of entropy, α was found to be greater than one (α > 1) (Table 3).

Table 3.

Enhancement factors (β) and coefficients (α) for multivalent MUC1 glycopeptides^[b]

Ligand	Peptide #	β	ΔG	α ΔH	−TΔS
HGVT**S*APDTRPAPGSTAPPA	18	2.88	0.60	1.03	1.78
HGVTSAPDTRPAPGS*T**APPA	19	2.74	0.60	1.02	1.74
HGVT**SAPDTRPAPGST**APPA	20	2.13	0.57	1.00	1.76
HGVT*S*APDTRPAPGSTAPPA	21	3.34	0.59	0.93	1.46
HGVTSAPDTRPAPGS*T*APPA	22	3.60	0.60	0.98	1.89
HGVT*SAPDT**RPAPGST*APPA	23	6.77	0.38	0.92	1.88

^[a]T** = Thr O-linked sialyl-Tn and T*/S* = Thr/Ser O-linked Tn.

^[b]β enhancement factor values are calculated using the equation β = K_A^multi/K_A^mono. β measures the increase in the affinity of each ligating unit obtained by its multivalent presentation. Factor α values are calculated using the equation ΔG^mul,n = αnΔG^mono, ΔH^mul,n = αnΔH^mono and ΔS^mul,n = αnΔS^mono.[57] ΔG^mono, ΔH^mono and ΔS^mono values were calculated by averaging the respective ΔG values of the glycan regions being evaluated.

The TΔS values for hMGL binding to multiple glycosylated peptides are not proportional to the number of single Tn and single sialyl-Tn. For example, TΔS is −21.80 kcal/mol for hMGL binding to trivalent MUC1 glycopeptide 23 as compared to an average of −4.0 kcal/mol for monoglycosylated peptides 15-17 (Table 2). This mode of affinity enhancement stems from multiple interactions between the lectin and the carbohydrate ligands presented on peptide backbone. It is a dynamic process in which lectin molecules bind to carbohydrate ligands and jump from ligand to ligand along the glycopeptides’ backbone. This mechanism was first observed and validated for MUC1-bearing Tn antigen ligands.^[59,60] Furthermore, simultaneous binding of two or more glycans in MUC1 glycopeptides (18-23), and in particular neighboring glycans in MUC1 glycopeptides (18, 19, 21 and 22), is difficult to imagine in light of recently published 3D models of trimeric MGL.^[61–63] The extended conformation, with the CRDs oriented perpendicular to the coiled-coil domain and 82 Å (8.2 nm) apart, is predicted.^[63] Another unique feature of this trimer model is the presence of the C162–C180 disulfide bond within the coiled-coil CRD linker, that strongly constrains the extension capabilities of CRDs from the neck.^[62] Certain plasticity of the MGL-CRD was proposed in relation to the accommodation of various GalNac-containing ligands.^[64] However, the observed plasticity can’t accommodate the binding of the glycans on the adjacent amino acids.

Enthalpy-entropy compensation (EEC)^[65] for hMGL interactions with the mono- and multivalent MUC1 glycopeptides carrying sialyl-Tn and/or Tn antigens was demonstrated by plotting the ΔH vs TΔS (Figure 3). The linear relationship indicates a flexible binding mode that optimizes the trade-off between enthalpic and entropic contributions.

Figure 3.

Enthalpy–Entropy compensation for glycopeptides vs hMGL interaction in H₂O. The enthalpy and entropy from each individual experiment for the glycopeptides binding with hMGL are plotted. Numbered points indicate individual measurements (see Table 2 for point identification). A linear fit to all the data points yielded a line with a slope of −1.077.

Conclusion

In summary, we developed a concise scheme to synthesize the sialyl-Tn antigen building blocks used in the solid-phase synthesis of MUC1 glycopeptide models. We employed orthogonally protected phenyl thioglycoside glycosyl donor in the presence of TfOH/NIS or TMSOTf/NIS as promoter system. The α-2,6-sialyl-Tn Thr building block was synthesized in a relatively high yield, 60%, but with moderate stereoselectivity, α/β ratio 6:1. The designed protocol was fully compatible with the Fmoc SPPS and allowed the relatively large-scale synthesis (0.1 mmol peptide) of mono- and multivalent MUC1 glycopeptides, also including mixed Tn and sialyl-Tn antigens. Mechanistic studies by ITC on the binding of synthesized MUC1 glycopeptides to trimeric hMGL revealed that the type of carbohydrate (Tn or sialyl-Tn) and the density of glycosylation affected the thermodynamics of the interaction. The presentation of the trimeric MGL CRDs is not flexible enough to allow adaptation to the shorter distances (<8.2 nm) between glycans, and the proposed “bind and jump” mechanism seems suitable model. The thermodynamic analysis pointed out the critical role of entropy in driving the affinity enhancement of multivalent interactions. Tighter contacts lead to increased molecular constraints, and thus, a more negative enthalpy compensates for conformational entropy reduction. However, the entropy contribution to binding can, to a considerable degree, originate from variations in the amount of water immobilized or released during binding interaction.