Large-scale synthesis and structural analysis of a synthetic glycopeptide dendrimer as an anti-cancer vaccine candidate

Abstract

Herein, we report a new process that enables the gram-scale production of a fully synthetic anti-cancer vaccine for human use. This therapeutic vaccine candidate, named MAG-Tn3, is a high-molecular-weight tetrameric glycopeptide encompassing carbohydrate tumor-associated Tn antigen clusters and peptidic CD4⁺ T-cell epitopes. The synthetic process involves (i) the stepwise solid-phase assembly of protected amino acids, including the high value-added Tn building blocks with only 1.5 equivalents, (ii) a single isolated intermediate, and (iii) the simultaneous deprotection of 36 hindered protective groups. The resulting MAG-Tn3 was unambiguously characterized using a combination of techniques, including a structural analysis by nuclear magnetic resonance spectroscopy. The four peptidic chains are flexible in solution, with a more constrained but extended conformation at the Tn3 antigen motif. Finally, we demonstrate that, when injected into HLA-DR1-expressing transgenic mice, this vaccine induces Tn-specific antibodies that mediate the killing of human Tn-positive tumor cells. These studies led to a clinical batch of the MAG-Tn3, currently investigated in breast cancer patients (phase I clinical trial). The current study demonstrates the feasibility of the multigram-scale synthesis of a highly pure complex glycopeptide, and it opens new avenues for the use of synthetic glycopeptides as drugs in humans.

---

Introduction

The tumor-associated Tn antigen (α-D-GalNAc-O-Ser/Thr or S*/T*) is expressed in several carcinomas (e.g., breast, ovarian, colon, lung, and prostate cancers).¹ Therefore, the Tn and other carbohydrate antigens have attracted much interest as a target for the development of anti-tumor immunotherapy through the induction of a specific immune response against cancer cells.^2–11 The induction of carbohydrate-specific immunity has primarily relied on glycan conjugation to an immunogenic carrier protein. Although this approach is very successful, it has major limitations (i.e., uncertainty in the vaccine structure and composition, and carrier-induced epitopic suppression¹²). As an alternative, we designed a novel fully synthetic immunogen, named MAG-Tn3 (MAG: multiple antigenic glycopeptide) (Fig. 1), that associates Tn antigen clusters with peptidic CD4⁺ T-cell epitopes on a tetravalent lysine backbone.^13–15

Fig. 1 Synthesis of MAG-Tn3 (H-[S*T*T*-QYIKANSKFIGITEL]₄-K₂-K-βA-OH, 5, structural formula in the inset) with the unprotected (initial route I, R = H)¹⁵ or protected (new route II, R = Ac or R = Bn) carbohydrate building blocks. AA9–10 (NS) and AA15–16 (IT) were incorporated as pseudo-Pro dipeptides. See Table 1 for yield and purity.

MAG-Tn3 is based on a multiple antigenic peptide (MAP) construct which has been introduced by Tam¹⁶ and widely used to present peptidic epitopes to the immune system.¹⁷ Only a few examples of MAGs that feature a multivalent display of both peptidic and carbohydrate epitopes have been reported by us and others.^{9,13,14,18,19} Notably, the characterization of previously reported MAP or MAG immunogens is typically incomplete. The analytical methods mainly include amino acid analysis (AAA) and, at best, partial data of reverse-phase high-performance liquid chromatography (RP-HPLC), mass spectrometry (MS) or nuclear magnetic resonance (NMR).^19–23 In addition to allowing definitive assessment of the identity and purity of the MAG, NMR analysis has proven very useful for providing structural information.

We have previously demonstrated that MAG-Tn3 is an efficient strategy for inducing a high-level Tn-specific antibody response (both in mice and in non-human primates), and for increasing the survival of tumor-bearing mice when used as a therapeutic vaccine.^14,15,24 On the basis of these successful pre-clinical studies, MAG-Tn3 was identified as a promising therapeutic vaccine candidate for the treatment of carcinomas.

As part of our efforts to investigate the efficiency of MAG-Tn3 in a phase I clinical trial, we report the development of an improved solid-phase synthetic process that allows for a gram-scale production of high-quality MAG-Tn3 for human use. MAG-Tn3 was extensively characterized using a combination of techniques, including NMR. In addition, we demonstrated that the four peptidic chains are flexible in solution, with a more constrained and extended conformation at the Tn3 antigen motif. Finally, this vaccine candidate was shown to induce Tn-specific antibodies in HLA-DR1-expressing transgenic mice. These antibodies not only recognize human Tn-positive tumor cells, but also mediate their killing.

Results and discussion

Synthesis and physicochemical characterization of MAG-Tn3 (5)

The preparation of MAG-Tn3 is challenging because of (i) its high molecular weight (M_r 10 897), (ii) its branched structure (steric hindrance), and (iii) the presence of clustered glycosylated AAs (four Tn trimer antigens). The synthetic process involves stepwise solid-phase peptide synthesis (SPPS) using 9-fluorenylmethoxycarbonyl (Fmoc) chemistry, cleavage and a final deprotection step (Fig. 1). Because the intermediates could not be isolated and the purification of such large glycopeptides can be complex, the upstream process was precisely optimized for (i) peptide assembly and (ii) protection/deprotection steps.

Peptide assembly. The peptidic part of MAG-Tn3 was assembled using N,N′-diisopropylcarbodiimide (DIC)/N-hydroxybenzotriazole (HOBt) and benzotriazole-1-yl-oxy-tris-pyrrolidino-phosphonium hexafluorophosphate (PyBOP) as the coupling reagents. First, the lysine core was attached to the resin through a beta-alanine linker, providing four amino groups. This linker was then further elongated with amino acids (AAs) from the T-cell epitope sequence (QYIKANSKFIGITEL or TT peptide), with four copies of the same AA being sequentially added. To minimize aggregation during the assembly, Thr and Ser were introduced as pseudoproline dipeptides, (i.e., Fmoc-Ile-Thr(Ψ^Me,Mepro)-OH and Fmoc-Asn(Trt)-Ser(Ψ^Me,Mepro)-OH, respectively). MS analysis showed that the use of these precursors significantly decreased the amount of deletion products produced when Thr and Ser were introduced as single AAs (Fig. S1B versus S1A†).

In a typical MAG/MAP stepwise SPPS, interchain (and/or intrachain) clustering occurs as the peptide chains grow, which leads to the accumulation of multiple deletion products and complex mixtures.²⁵ A few strategies have been proposed to minimize these by-products.²⁶ Convergent syntheses, which involve ligation of purified peptide fragments to a preformed MAP core, have been developed.^22,27,28 However, they generate non-peptide bonds that may adversely affect the biological properties of the construct. In addition, when rigorously compared to identical peptide sequences, the stepwise approach was found to be more advantageous than the convergent approach.²¹ Other approaches have met with varying degrees of success. They include the introduction of flexible spacers, double couplings, capping and/or 1,8-diazabicyclo-[5.4.0] undecen-7-ene deprotection.^21,29,30 To reduce the aggregation effect, Boykins et al. used N-[2-hydroxy-4-methoxybenzyl] (Hmb) derivatives in the growing peptide chain of a divalent MAP.³¹ Pseudoproline dipeptides are also known to minimize aggregation and to limit deletion by-products in synthetic peptides.³² Accordingly, the use of two pseudoproline dipeptides significantly improved the quality of crude MAG-Tn3, thereby facilitating the downstream purification.

Glycosylated AA assembly and protection/deprotection strategy. When synthesized on a small-scale (1–10 mg), MAG-Tn3 (5) was obtained from unprotected Tn-AA synthons (route I, R = H, Fig. 1).¹⁵ This method was advantageous because it avoided the final deprotection step of 36 clustered protecting groups. However, the scale-up of the process failed because of incorporation of extra Tn residues which was difficult to control, thus affecting the overall yield (Table 1, Fig. 2A). To circumvent this problem, we developed a novel synthetic route involving protected Tn building blocks (route II, Fig. 1).

Fig. 2 MS analysis (deconvoluted spectra) of crude deprotected glycopeptide 5 obtained directly from peptide 2 (route I, calcd M_r 10 897.123) (A), crude protected glycopeptides 3 (route II, R = Ac, calcd M_r 12 410.465) (B), and 4 (route II, R = Bn, calcd M_r 14 141.610) (C). M ± S* and M ± T* were side-products due to incorporation defaults of the Tn building blocks. M-1Ac and M-1Bn, M-2Bn were partially deprotected compounds arising from resin cleavage under acidic conditions; they were not considered impurities because they all eventually led to 5.

Table 1 Different processes for the MAG-Tn3 5 synthesis (see Fig. 1 for details)

	Route I		Route II
Substrate (protecting group R)	2 (none, H)	2 (none, H)	3 (Ac)	4 (Bn)	4 (Bn)	4 (Bn)
a Isolated final product (net peptide content). b After RP-HPLC purification, calculated on the basis of the isolated net peptide content and includes all of the synthesis steps from 1. c Corresponding average yield per each coupling step > 96%. d As determined by RP-HPLC.
Deprotection method	—	—	NH2–NH2, H2O	TfOH	H2	H2
Scale^a (g)	0.001–0.01	> 0.01	∼0.005	∼0.005	0.225	3
Overall yield^b (%)	3	< 1	2.7	1.6	1.5	6^c
Purity^d (%)	94.5	< 95	96.4	95.9	96.3	96.6
Comment	Issues during scale-up		To be validated on a large scale	Partial degradation	Validated on a multigram scale

---

Two different hydroxyl protecting groups (R) were evaluated (i.e., acetyl (Ac) and benzyl ether (Bn)). Notably, the molar excess of the high value-added glycosylated building blocks was minimized to 1.5 equivalents. After resin cleavage and AA side-chain deprotection, MS analysis of the crude products indicated that, in both cases, N-acetylgalactosamine (GalNAc) protection prevented the side-reactions observed in route I (Fig. 2B, C versus A). Protected glycopeptides 3 and 4 were obtained as the major compounds, with a significant improvement in the crude purity. Advantageously, each glycopeptide constitutes the only intermediate isolated in the process.

Several deprotection methods were evaluated for both protecting groups. The best results are summarized in Table 1. Treatment of compound 3 with MeONa afforded MAG-Tn3 5 after RP-HPLC purification but with a moderate purity (91.4%) and a 1.4% overall yield (data not shown). In addition, the reaction progress was difficult to monitor, which may affect the process reproducibility. When 3 was treated with hydrazine hydrate, HPLC/MS monitoring indicated that MAG-Tn3 was formed as the major product after 1 h. Surprisingly, complete deprotection could not be achieved despite an increase in pH or an extension of the reaction time. The crude profiles in MeONa and hydrazine were similar, showing that 1–2 acetates (from among 36) were resistant to basic conditions, irrespective of the reagent used. This partial deacetylation may be due to a sterically hindered environment. The partially deprotected intermediates were easily separated by RP-HPLC, and this method afforded 5.8 mg of MAG-Tn3 in 96.4% purity, with an overall yield of 2.7%. Compound 4 was treated with a solution of trifluoromethanesulfonic acid (TfOH)/TFA/dimethylsulfide (DMS)/m-cresol³³ and HPLC/MS analysis indicated complete Bn removal after 1 h 15 min. RP-HPLC purification afforded 3.9 mg of MAG-Tn3 5 in 95.9% purity, with an overall yield of 1.6%. However, this method was not entirely satisfactory because a compromise is required between complete deprotection and degradation, which may adversely affect the reaction repeatability and result in scale-up issues. Catalytic hydrogenation is widely regarded as the method of choice for Bn deprotection because it typically proceeds under mild conditions and with limited degradation. The hydrogenolysis of 4 was performed using palladium catalysis under optimized deprotection conditions (10% Pd/C at 37 °C, 5 bar, 170 h, in NMP/H₂O (87.5/12.5)) enabling the simultaneous deprotection of the 36 Bn. The crude product (Fig. 3A) was purified by RP-HPLC to give 225 mg of 5 in 96.3% purity, with an overall yield of 1.5%. This process was further validated on a multigram scale with a final purity of 96.6% and an overall yield of 6% (Table 1). It is worth noting that the corresponding average yield is over 96% when related to each individual coupling step involved in the synthesis of this high molecular weight molecule.

Fig. 3 Analysis of the compound 5 obtained by hydrogenolysis before (A) and after (B, C) purification. RP-HPLC (A, B) and ESMS (deconvoluted spectrum, calcd M_r 10 897.123) (C) spectra. Magnification of the HPLC peak base is presented in the inset. The purity of 5 was calculated as the percentage of the total area under the curve at 220 nm (B). $ = fragmentation product.

To ensure both its purity and its identity, the MAG-Tn3 5 was analyzed by RP-HPLC (Fig. 3B), ESMS (Fig. 3C), quantitative AAA, and N-ter sequencing (data not shown). MAG-Tn3 was also thoroughly characterized by NMR (assignments of ¹H, ¹³C, and ¹⁵N nuclei in Table S1A, and in Fig. S2†). The four chains are equivalent even though small differences in the chemical shifts were observed in the C-ter part from Ile15 to Leu18. This result is explained by environmental differences due to the branching of Leu18 on either αNH (Leu18a and a′) or εNH (Leu18b and b′) of the lysine scaffold residues (Lys73 and Lys74) as well as by the steric hindrance close to the lysine core, as previously observed by Esposito and coworkers.²⁰ All of the NMR data (Tables S1A, S2, and S3, and Fig. S2 and S3†) confirmed the sequence and the presence of three GalNAc residues α-linked to Ser1, Thr2 and Thr3.

Despite recent progress in glycopeptide and glycoprotein synthesis,^11,34–37 access to such complex (macro)molecules remains a challenging task. Here, in a stepwise solid-phase process, we showed that the GalNAc protection was critical for minimizing the side reactions that occur during the Tn coupling, thereby allowing the scale-up. Two protocols emerged as convenient and repeatable strategies. The Ac/hydrazine method afforded the highest overall yield on a milligram scale. This process offers the advantages of short reaction time, and a lack of safety issues associated with the use of hydrogen. This method may be worthy of consideration in the future even though it remains to be validated on a large scale. The Bn/H₂ method successfully afforded MAG-Tn3 on a multigram scale. Notably, this optimized process was further improved and yielded 11.3 g of a “Good Manufacturing Practice” (GMP) batch.

Until the late 1990s, the crude MAPs and glycodendrimers were typically partially purified and the resulting products were poorly characterized. Inherent problems have been highlighted with their purity and chemical definition, especially by MS.^22,26,30 Improvements in both synthesis and analysis have allowed the more accurate characterization of MAPs/MAGs by MS, HPLC or NMR,^{18–23,26,28,31} even though only partial data are usually reported. Herein, we describe the complete and detailed characterization of the MAG-Tn3, with all of the results being consistent with a high quality product. As for the chemical characterization, literature reports have often lacked experimental data on the exact purity of the MAG/MAP, overall yield of the synthesis and the working scale. Nevertheless, by inference, most of the MAP/MAG syntheses have been performed on the 1–20 mg scale (M_r ranging from 4000 to 15 000). To the best of our knowledge, our results provide the first process that enables the gram-scale synthesis of a large pure dendrimer glycopeptide with comprehensive characterization using stringent quality controls, including NMR.

Immunological properties of MAG-Tn3 (5)

The MAG-Tn3 vaccine aims to induce specific anti-Tn antibody responses that are dependent on the activation of helper T-cells, specific for the inserted CD4⁺ T-cell peptidic epitope.¹⁴ The clinical-grade MAG-Tn3 designed for human use includes a pan-HLA-DR restricted TT peptide,³⁸ which we previously evaluated in non-human primates.¹⁵ To formally demonstrate the functionality of this TT epitope in an HLA-DR context, we tested the immunogenicity of MAG-Tn3 in HLA transgenic mice expressing human DR1 or DR1*A2. In contrast to the control groups, mice immunized with the vaccine developed high levels of anti-Tn IgG antibody titers, as detected by ELISA on a synthetic Tn3 cluster attached to a poly-glycine sequence (Fig. 4A and B). Also, the antibodies from MAG-vaccinated mice recognized physiological forms of Tn as shown by their binding to Tn-positive Jurkat tumor cells (Fig. 4C), which demonstrated the immunogenic properties of MAG-Tn3. Finally, sera from MAG-vaccinated mice mediated antibody-dependent cytotoxicity as assessed by the killing of Jurkat cells (Fig. 4D).

Fig. 4 MAG-Tn3 induced anti-Tn antibodies in HLA-transgenic mice. (A) DR1*A2 and (B) DR1 mice were immunized with 10 μg of MAG-Tn3 (n = 4) in adjuvant (Alum/CpG1826) or with only adjuvant (n = 4) or with 10 μg of the non-glycosylated control MAP-TT¹⁵ (n = 3). Sera from these mice were collected at days 20, 28 and 49 post-immunization (D20, D28 and D49, respectively), and they were analyzed for Tn recognition by ELISA (A and B) and by FACS on Tn-positive Jurkat cells (C). The antibody-mediated cytotoxicity was assessed on Jurkat cells with sera from panel B (D). The statistical significance of differences was determined by the Student t test (*P < 0.05).

Structural properties of MAG-Tn3 (5) by NMR

The NMR analyses were used not only as a quality control method but also for structural characterization. A detailed analysis of chemical shifts, coupling constants, and ¹H, ¹H NOEs is provided in the ESI.†

The chemical shift analysis indicated that the peptide chains behave similarly and are flexible with motions that are gradually restricted at N-ter from Ser1* to Ile6 or Lys7 (Table S1, Fig. S4†). The glycosylation of the N-ter residues (Ser1*, Thr2* and Thr3*) affects the chemical shift values of these modified residues as well as those of their neighbors. However, because these glycosylated AAs are not taken into account in the chemical shift databases, other NMR approaches were employed to confirm the presence of the favored extended conformation of the N-ter residues.

The coupling constant analysis (Fig. S3, Table S1A†) revealed: (i) the preferred extended conformation of the Tn3 antigenic motif, (ii) the restricted rotation around the peptide-sugar linkage upon glycosylation of threonine and, to a lesser extent, serine residues, as previously described (see ref. 39 and 40 and references therein), and (iii) the preferred orientation of the N-acetyl group of the GalNAc residues.

The backbone ¹H, ¹H NOE intensities highlighted the preferred extended conformation at N-ter up to Ile6. This was confirmed by the presence of a set of ¹H, ¹H NOEs between the GalNAc residues and the AA residues up to Tyr5 (Fig. S3, Tables S2 and S3†). A similar behavior was observed for a small glycosylated peptide (Ser*)₃ with strong dαN(i,i + 1) coupled to weak dNN(i,i + 1) NOEs, whereas medium dNN(i,i + 1) NOEs were obtained for the non-glycosylated peptide (Ser)₃.⁴¹

The ¹⁵N longitudinal and transverse relaxation time (i.e., T1 and T2, respectively) measurements for MAG-Tn3 (Fig. 5) suggested a restriction in the flexibility at the N-ter including the first three glycosylated AAs and several of the following residues. Indeed large T1 and T2 values (ranging from 0.51 s to 0.76 s and from 0.14 s to 0.34 s, respectively) were measured from Thr2* to Leu18, showing a high flexibility characteristic of an unfolded molecule. However, smaller T1 values were observed from Thr2* to Lys7, especially for Thr2* (T1 = 0.51 ± 0.03 s). Differences were also observed for T2 along the sequence with larger values from Thr2* to approximately Asn9, followed by decreasing values up to Leu18, as expected for residues adjacent to the lysine core. For an unstructured peptide, larger T2 values due to fast internal motions are typically observed as a consequence of fraying at the free N-ter. In the case of MAG-Tn3, the plateau values observed for T2 from Thr2* to Ala8/Asn9 may result from restricted flexibility induced by the glycosylation of the first three residues. This result is in agreement with the decrease in the T1 values at N-ter. Similarly, Coltart et al. concluded that conformational rigidity was persistent through residues 1 to 4 of a mucin-derived tri-glycosylated peptide (STTAV).⁴² Here, for the longer peptide chains of MAG-Tn3, the stiffening effect induced by glycosylation is perceptible up to approximately Ala8.

Fig. 5 ¹⁵N relaxation parameters for MAG-Tn3. The dotted lines highlight the different dynamic behavior of MAG-Tn3 along the sequence, notably the reduced N-ter end-fraying from Thr2* to approximately Ala8. For residues T16, E17 and L18, the mean value calculated from the differentiated chain values is reported.

For the GalNAc NH group, the ¹⁵N T1 and T2 values are larger for GalNAc 1 than for GalNAc 2 and 3 (T1 = 0.88, 0.72 and 0.71 ± 0.03 s and T2 = 0.49, 0.41 and 0.34 ± 0.03 s for GalNAc1, 2 and 3, respectively). This suggests greater freedom in the motions of the N-acetyl group in GalNAc-Ser1 than in GalNAc-Thr2 and -Thr3. In fact, it was shown that hydrogen bonding between the N-acetyl group of GalNAc residues and the carbonyl group of the glycosylated AA was always detected in the case of threonine but not in the case of serine.^40,43

Despite the large errors due to the low concentration of the ¹⁵N natural abundance sample, an increase in the NOEs from negative to positive values was observed from Thr2* to Leu18 (Fig. 5), which is in agreement with a progressive increase in flexibility when going from the lysine core bonded residues to the free N-ter residues.

In conclusion, all of the NMR results indicate that the four peptidic chains of the MAG-Tn3 molecule behave in an approximately similar fashion. Their flexibility increases starting from their lysine core attachment point up to the N-ter. However, the glycosylation of the three N-ter residues induces a preferred extended conformation in the Tn3 antigen motif and reduces the N-ter end-fraying.

The structural arrangement of the N-ter Tn3, and most likely the one of the overall MAG, plays a role in its immunological properties. A number of innovative studies have demonstrated that both the Tn display and environment within glycopeptides (e.g. antigen density, peptide backbone) are critical to modulate their molecular recognition by Tn-specific antibodies.^39,44–47 Moreover the impact of the proper design of Tn-based vaccines on their immunogenicity has been highlighted. In particular the carrier and the spacer arm attaching the Tn to the carrier have a profound effect on the level and on the diversity of antibody responses.^48,49

Because no conformational difference was observed between the MAG-Tn3 glycopeptide chains and the linear glycopeptide, the improved immunogenicity of the MAG-Tn3 tetramer compared to that of the monomer¹⁴ is not due to particular conformational glycopeptidic epitopes. It rather may account for the Tn density effect for its recognition by macrophage galactose-type lectin (MGL) on antigen-presenting cells.⁵⁰ MGL is expressed on dendritic cells and carries a single carbohydrate recognition domain for Gal/GalNAc residues.⁵¹ Thus, we have shown that antigen processing and presentation of the T-cell peptide are increased by the Tn density, most likely by clustering MGL on the dendritic cell surface that enhances receptor-mediated uptake of MAG-Tn3 in the dendritic cell endocytic processing compartment.

As reported earlier for the MAP construct by Esposito and co-workers,²⁰ the dendrimeric structure appears to be optimized for interactions, compared to the peptide monomer. The MAG-Tn3 allows the four-peptide chains to sweep the space with larger-amplitude motions at the N-ter, to present the antigenic motif at the surface as an outstretched hand ready for interaction with its partners. Thus, the MAG-Tn3 offers a multiplicity of interactions including with the galactose-specific lectins and the immunoglobulins which both contribute to its immunogenic properties. In addition, through its four flexible peptidic arms, MAG-Tn3 can display a large variety of Tn discrete conformational epitopes that may better mimic the diversity of cancer-associated Tn clusters.

Conclusions

Despite their enormous potential as medicines and/or immunotherapeutics, carbohydrate-based drugs remain underused, in part because of their challenging synthesis. Herein, we report a new process allowing multigram preparation of a highly pure glycopeptide as an anti-cancer therapeutic vaccine, and we demonstrate its potency in humanized mice. We also provide new insights into its molecular structure in solution, which may aid in the future design of such complex molecules. These results led to a phase I clinical trial, ongoing in breast cancer patients. In contrast to many others, this vaccine candidate is chemically-defined, which is a key advantage for consistent studies and safe clinical evaluation. The importance of carbohydrate- and peptide-based drugs is being increasingly recognized. The current study demonstrates the feasibility of the large-scale synthesis of a complex glycopeptide and it opens new avenues for the use of synthetic glycopeptides as drugs in humans.

Experimental section

General synthesis methods

The synthesis of 5 was performed by a stepwise solid-phase method using Fmoc chemistry. The AA side-chain protective groups used were Trt on Gln and Asn, Boc on Lys7 and Lys11, Fmoc on Lys19 and Lys20, tBu on Tyr, Ser, Thr, and Glu. The synthesis of 5via route I was performed according to a previously described protocol.¹⁵

The net peptide content was determined by nitrogen analysis or quantitative AAA using a Beckman 6300 analyzer after hydrolysis of the compounds with 6 N HCl at 110 °C for 20 h. The HPLC/MS analyses were performed on an Alliance model 2695 system coupled to a model 2487 UV detector (220 nm) and to a Q-Tofmicro™ spectrometer (Micromass) with an electrospray ionization (positive mode) source (Waters, France). The samples were cooled to 4 °C on the autosampler. A linear gradient was applied with acetonitrile + 0.025% formic acid (A)/water + 0.04% TFA + 0.05% formic acid (B) over a period of 20 min. Either a Zorbax 300SB C18 column (3.5 μm, 3 × 150 mm) (Agilent, France) (gradient 13–53% A) or an XBridge™ BEH130 C18 column (3.5 μm, 2.1 × 150 mm) (Waters, France) (gradient 15–40% A) was used. The source temperature was maintained at 120 °C and the desolvation temperature at 400 °C. The cone voltage was 40 V. The ESMS analyses were recorded in the positive mode by direct infusion in the same spectrometer with a source temperature and a desolvation temperature maintained at 80 °C and 250 °C, respectively. The samples were dissolved at a concentration of 5 μM in water/acetonitrile (1/1) containing 0.1% formic acid. MaxEnt 1 Software (Waters, France) was used for the deconvolution of mass spectra. The purity of 5 was analyzed by RP-HPLC using an Agilent 1200 pump system with a UV detector at 220 nm. A Zorbax 300SB C18 column (3.5 μm, 3 × 150 mm) (Agilent, France) was used, and a gradient of acetonitrile + 0.1% TFA (A)/water + 0.1% TFA (B) was applied over a period of 40 min, from 13 to 53% A (0.8 mL min⁻¹, retention time 20.5 min).

The molar equivalents of all of the reagents are indicated relative to the reactive amino groups. The molar amounts of crude protected intermediates 3 and 4 were calculated on the basis of the starting Fmoc-β-Ala-resin 1 substitution. The overall yields include all the synthetic steps from 1. These yields were calculated on the net peptide content of the isolated product 5 based on the first β-Ala residue loading on the resin.

Syntheses

[QYIKANSKFIGITEL]₄-K₂-K-β-Ala-Resin (protected peptide 2). Until the incorporation of Tyr5, the tetravalent peptide was synthesized from Fmoc-β-Ala-Tentagel R Trt resin 1 (36.9 g, 4.8 mmol) (Rapp Polymere, Germany) on a manual peptide synthesizer equipped with a Schmizo reactor. Prior to the elongation process, the resin was allowed to swell in DMF for 2 to 3 h and washed with DMF (240 mL) (3 times, 2 min per cycle). After each coupling, the Fmoc groups were removed with 20% piperidine in DMF (240 mL) (3 steps, 20 min each). For Glu17, Asn9 and Gln4, 2% HOBt was added to the deprotection solution. After each deprotection, the resin was successively washed with DMF (240 mL) (4 times, 2 min per cycle), 2% HOBt in DMF (240 mL) (twice, 5 min per cycle), and DMF (240 mL) (twice, 2 min per cycle).

The AA couplings (1.75 equiv. for Lys20 and 2 equiv. for the other AA) were performed in DMF (111 mL) at room temperature with DIC/HOBt (1.5 to 2 equiv. each). The AA in positions 15–16 and 9–10 were incorporated as Fmoc-Ile-Thr(Ψ^Me,Mepro)-OH and Fmoc-Asn(Trt)-Ser(Ψ^Me,Mepro)-OH (2 equiv.), respectively. After 30 min, a fresh portion of DIC (1.5 to 2 equiv.) was added to the reaction mixture. The coupling steps were monitored using the Kaiser test. From Leu18 to Ser1, after 1 h of coupling with DIC/HOBt, the PyBOP reagent was added (0.5 equiv. for Lys20 and dipeptides, 1 equiv. for the other AA) and the pH was adjusted to 7 by dropwise addition of diisopropylethylamine (DIPEA). After 30 min, the resin was washed with DMF (240 mL) (5 times, 2 min per cycle) and an acetylation step was carried out from Leu18 to Thr2. The acetylation was performed at room temperature with acetic anhydride (1 equiv.) and pyridine (1 equiv.) in DMF (111 mL). After 20 min, the resin was washed with DMF (240 mL) (6 times, 2 min per cycle). After the incorporation of Tyr5, the resin was extensively washed with DMF (240 mL) (8 times, 2 min per cycle) and CH₂Cl₂ (240 mL) (8 times, 2 min per cycle) prior to drying.

After the incorporation of Tyr5, the assembly was reiterated on a 0.15 and 4.65 mmol scale using the same protocol to afford the peptide-resin 2 for 3 and 4, respectively.

[S(α-D-GalNAc(OAc)₃)-T(α-D-GalNAc(OAc)₃)-T(α-D-GalNAc(OAc)₃)-QYIKANSKFIGITEL]₄-K₂-K-β-Ala (3). The synthesis was performed as previously described for 2. The coupling steps were performed with Fmoc-Thr(α-D-GalNAc(OAc)₃)-OH and Fmoc-Ser(α-D-GalNAc(OAc)₃)-OH¹³ (1.5 equiv.), DIC/HOBt (1.5 equiv.) and PyBOP (0.5 equiv.). Starting from 2 (0.15 mmol), the glycopeptide-resin was suspended in TFA/triisopropylsilane (TIS)/H₂O (95/2.5/2.5 v/v/v) (10 mL g⁻¹ of glycopeptide-resin) and stirred for 1 h at 20 °C. After filtration, the resin was washed twice with the same TFA mixture (2 mL g⁻¹ of glycopeptide-resin per wash). The combined filtrates were stirred for an additional 30 min at 20 °C. After concentration, the crude product was precipitated with diisopropyl ether (DIPE) (35 mL g⁻¹ of glycopeptide-resin). After filtration and washing with DIPE, the solid was dried to yield 750 mg of crude 3.

ESMS: 12 409.589 (C₅₅₃H₈₅₅N₁₀₇O₂₁₃ calcd 12 410.465).

[S(α-D-GalNAc(OBn)₃)-T(α-D-GalNAc(OBn)₃)-T(α-D-GalNAc(OBn)₃)-QYIKANSKFIGITEL]₄-K₂-K-β-Ala (4). The synthesis was performed as previously described for 2. The coupling steps were performed with Fmoc-Thr(α-D-GalNAc(OBn)₃)-OH and Fmoc-Ser(α-D-GalNAc(OBn)₃)-OH (Fischer Chemicals AG, Switzerland) (1.5 equiv.), DIC/HOBt (1.5 equiv.), and PyBOP (0.5 equiv.). Starting from 2 (4.65 mmol), 84.87 g of the glycopeptide-resin was obtained. The glycopeptide-resin (20 g, 1.096 mmol) was treated as previously described for 3 to afford 9.95 g of crude 4.

ESMS: 14 141.433 (C₇₃₃H₉₉₉N₁₀₇O₁₇₇ calcd 14 141.610).

[S(α-D-GalNAc(OH)₃)-T(α-D-GalNAc(OH)₃)-T(α-D-GalNAc(OH)₃)-QYIKANSKFIGITEL]₄-K₂-K-β-Ala or MAG-Tn3 (5). From 3: compound 3 (100 mg, 20 μmol) was dissolved in MeOH (3.2 mL). Hydrazine monohydrate (567 μL, 11.3 mmol) was added and the solution was stirred at room temperature. The reaction was monitored by HPLC/MS analysis. After 2.5 h, the solution was cooled to 0 °C and acetone (3.2 mL) was added. After an additional hour, the solution was concentrated and co-distilled five times with acetone. The crude glycopeptide was lyophilized to yield 117 mg. The product was purified by RP-HPLC using an Agilent 1200 pump system with UV detection at 230 nm. A Zorbax C18 column (5 μm, 300 Å, 9.4 × 250 mm) (Agilent, France) was used, and a gradient elution was performed with acetonitrile/water (0.1% TFA) over 20 min from 28/72 to 38/62. The purification yielded 5.8 mg (net peptide content) of 5 in 96.4% purity. The overall yield was 2.7%.

From 4, the TfOH method: compound 4 (200 mg, 0.014 mmol) was dissolved in TFA (2.96 mL), DMS (1.78 mL) (Sigma-Aldrich, France) and m-cresol (587 μL) (Sigma-Aldrich, France) at room temperature. The solution was cooled to −10 °C and TfOH (587 μL) (Fluka, France) was added. The mixture was stirred for 1 h 15 min at −10 °C (TfOH/TFA/DMS/m-cresol 1/5/3/1 v/v/v/v). The product was precipitated with diethyl ether and, after centrifugation, the pellet was dissolved in water and lyophilized to yield 372 mg of the crude glycopeptide. The product was dissolved in 0.05 M ammonium acetate buffer (7.7 mL) and the pH was adjusted to 7 with 1 M ammonia. After 1 h at room temperature, the solution was lyophilized to yield 412 mg of the crude product. The product was purified with a gradient from 27/73 to 40/60, as previously described. The purification afforded 3.9 mg (net peptide content) of 5 in 95.9% purity. The overall yield was 1.6%.

From 4, the H₂ method: compound 4 (10 g, 1.1 mmol) was dissolved in N-methylpyrrolidone (NMP)/H₂O (87.5/12.5, 800 mL v/v). After 10% Pd/C type 39 (4 g) (Johnson Matthey, UK) was added, the reaction was stirred at 37 °C under a pressure of 5 bars for 170 h. Two additional portions of the catalyst were added after 72 h (2 g) and 120 h (2 g). At the end of the reaction, the catalyst was filtered on Celite and washed with NMP/H₂O (87.5/12.5 v/v). The resulting filtrate (including that of similar reactions, 1.35 mmol in total) was diluted with H₂O (until NMP/H₂O 10/90 v/v) and purified by RP-HPLC in two steps. The primary purification was carried out on a Vydac C18 column (300 Å, 10–15 μm, 50 mL min⁻¹) with TFA/H₂O/CH₃CN (0.1/94.9/5.0 v/v/v) (A) and TFA/H₂O/CH₃CN (0.1/49.9/50.0 v/v/v) (B) as the eluents. The gradient was 0% B for 15 min, 0–40% B over 5 min, and 40–80% B over 60 min. A secondary purification with AcOH/H₂O/CH₃CN (0.5/94.5/5.0 v/v/v) (A) and AcOH/H₂O/CH₃CN (0.5/49.5/50.0 v/v/v) (B) was carried out on the same column. The gradient was 0% B for 15 min, 0–20% B over 5 min, and 20–60% B over 60 min. After concentration by RP-HPLC on a Daisogel SP-300-10-ODS-AP column (20 mL min⁻¹, isocratic TFA/H₂O/CH₃CN 0.1/49.9/50.0 v/v/v), the solution was evaporated on a rotary evaporator and lyophilized to afford 225 mg (net peptide content) of 5 in 96.3% purity. The overall yield was 1.5%.

AAA: Ala 4 (4), Asn 3.9 (4), Glu + Gln 7.8 (8), Gly 4.1 (4), Ile 11.0 (12), Leu 4.1 (4), Lys 11.0 (11), Phe 3.8 (4), Ser 7.0 (8), Thr 10.5 (12), Tyr 3.7 (4). ESMS: 10 897.387 (C₄₈₁H₇₈₃N₁₀₇O₁₇₇ calcd 10 897.123).

NMR experiments

NMR spectra were acquired at 303 K on a Varian NMR System 600 spectrometer equipped with a cryogenically-cooled triple resonance ¹H{¹³C/¹⁵N} PFG probe (Agilent Technologies). The samples were dissolved in a 15% v/v D₂O buffered solution (50 mM deuterated sodium acetate at pH 3.9 prepared with acetic acid D4 100% D, D₂O 99.97% D and sodium deuteroxide 40% w/w solution in D₂O 99.5% D, Euriso-top, Saint-Aubin, France) and transferred into a 4 mm NMR tube (Shigemi Inc., Alison Park, USA). The final sample concentration ranged from 0.3 to 0.8 mM, except for that used in the ¹⁵N relaxation study (1.2 mM). The ¹H chemical shifts were referenced to external DSS (0 ppm), and the ¹³C and ¹⁵N chemical shifts were referenced indirectly to DSS.⁵²

Resonance assignments were determined using the following 2D experiments: ¹H, ¹H COSY,⁵³¹H, ¹H TOCSY (mixing times of 40, 60 and 80 ms),⁵⁴¹H,¹H NOESY (mixing times of 50, 100 and 200 ms),⁵⁵¹H–¹³C edited HSQC and ¹H–¹³C HSQC-TOCSY (mixing time of 30 and 60 ms).⁵⁶ In addition, a ¹H–¹⁵N HSQC was employed to elucidate the overlapped HN, αH region in the NOESY experiments.⁵⁷

The conformational study of MAG-Tn3 5 was carried out using the following NMR measurements: (i) chemical shifts, (ii) ³J_NH,Hα and ³J_NH,H2 coupling constants extracted from 1D or 2D spectra with a ¹H resolution of 0.1 Hz and 0.8 Hz, respectively, (iii) ³J_Hα,Hβ and ³J_Hα,Hβ′ coupling constants measured in a 2D DQF-COSY experiment performed with a resolution of 0.6 Hz on a sample dissolved in 100% D₂O buffered solution, (iv) ¹H–¹H NOE cross-peak intensities measured for a mixing time of 200 ms (detailed in the ESI†) and (v) T1 and T2 ¹⁵N relaxation times and steady-state {¹H}–¹⁵N NOEs at natural abundance at 14.1 Tesla.^58,59

Analysis of immune responses

Mice were kept in the Institut Pasteur animal house with water and food supplied ad libitum. The study was performed in compliance with French guidelines (project CETEA no. 2013-126).

The immunogenicity of the MAG-Tn3 glycopeptide was assessed in DR1 or DR1*A2 mice which are transgenic for human HLA-DR1 or HLA-DR1 and HLA-A2.⁶⁰ For this purpose, the mice were i.p. immunized every three weeks with 10 μg of MAG-Tn3 in alum (1 mg) supplemented with 10 μg of CpG1826 oligonucleotide (Sigma). One week after each boost, sera were collected and tested for anti-Tn antibodies by ELISA using the Tn3-G7KG biotinylated glycopeptide.¹⁴ Serial dilutions of sera were performed and bound antibodies were revealed using the goat anti-mouse IgG peroxidase conjugate (Sigma) and an o-phenyldiamine/H₂O₂ substrate. The plates were read photometrically at 492 nm. The results are expressed as the mean antibody titer of individual sera ± SD.

Sera were also tested at 1/50 dilution by flow cytometry for recognition of Tn-expressing human Jurkat tumor cells.¹⁵ First the cells were incubated for 30 min with sera at 4 °C in PBS containing 5% fetal calf serum and 0.05% sodium azide. Then, the cells were incubated for 30 min with anti-mouse IgG conjugated to phycoerythrin (Caltag). The cells were analyzed on an FACS Calibur flow cytometer (BD), and the analysis was performed using Flowjo software (Tree Star Inc.). The data are shown as histograms corresponding to the fluorescence of cells incubated with the secondary reagent alone or with sera and reported as the mean of the fluorescence intensities.

The biological activity of the sera was evaluated for their ability to induce death of Tn-positive Jurkat cells. Jurkat cells were incubated with sera diluted 50-fold for 15 min at 4 °C and then with goat anti-IgG (Southern Biotech) for 45 min at RT. Cell death was measured by FACS using propidium iodide (PI). The cytotoxicity was calculated as the percent of PI-positive cells. The results are expressed as the mean cytotoxicity of individual test sera ± SD.

Author contributions

C. A., M. D., C. L., R. L-M., and S. B. designed and coordinated research; C. G., C. S., E. E., T. C., Y.-M. C., E. D., and F. B. performed research; C. S., M. D., C. L., R. L-M., and S. B. analyzed data; and C. S., R. L-M., and S. B. wrote the paper.

Conflict of interest

The authors declare no conflict of interest.

Acknowledgements

We are grateful to F. Groh for the AAA, to Dr Y. C. Lone for providing DR1 and DR1*A2 transgenic mice, and to Dr L. Mulard for critical review of the manuscript. We also wish to thank Dr J.-M. Poudrel, Dr J.-M. Cauvin, Dr C. Coindet, and Dr D. Monnaie (Lonza Braine SA) for their contribution to the process development. We gratefully acknowledge financial support from the Institut Pasteur, and we wish to thank the donators of the MAG-Tn3 program. This work was also supported by grants from the Ligue Nationale Contre le Cancer (Equipe Labellisée 2014) and Banque Privée Européenne to CL.

References

1. G. F. Springer, Science, 1984, 224, 1198–1206.
2. T. Buskas, P. Thompson and G.-J. Boons, Chem. Commun., 2009, 5335–5349.
3. D. Feng, A. S. Shaikh and F. Wang, ACS Chem. Biol., 2016, 11, 850–863.
4. M. Glaffig, B. Palitzsch, N. Stergiou, C. Schull, D. Strassburger, E. Schmitt, H. Frey and H. Kunz, Org. Biomol. Chem., 2015, 13, 10150–10154.
5. T. Ju, V. I. Otto and R. D. Cummings, Angew. Chem., Int. Ed., 2011, 50, 1770–1791.
6. V. Lakshminarayanan, P. Thompson, M. A. Wolfert, T. Buskas, J. M. Bradley, L. B. Pathangey, C. S. Madsen, P. A. Cohen, S. J. Gendler and G.-J. Boons, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 261–266.
7. P. Niederhafner, M. Reinis, J. Sebestik and J. Jezek, J. Pept. Sci., 2008, 14, 556–587.
8. O. Renaudet, G. Dasgupta, I. Bettahi, A. Shi, A. B. Nesburn, P. Dumy and L. BenMohamed, PLoS One, 2010, 5, e11216.
9. T. C. Shiao and R. Roy, New J. Chem., 2012, 36, 324–339.
10. S. F. Slovin, S. J. Keding and G. Ragupathi, Immunol. Cell Biol., 2005, 83, 418–428.
11. R. M. Wilson and S. J. Danishefsky, J. Am. Chem. Soc., 2013, 135, 14462–14472.
12. R. Roy, Drug Discovery Today, 2004, 1, 327–336.
13. S. Bay, R. Lo-Man, E. Osinaga, H. Nakada, C. Leclerc and D. Cantacuzene, J. Pept. Res., 1997, 49, 620–625.
14. R. Lo-Man, S. Vichier-Guerre, S. Bay, E. Dériaud, D. Cantacuzène and C. Leclerc, J. Immunol., 2001, 166, 2849–2854.
15. R. Lo-Man, S. Vichier-Guerre, R. Perraut, E. Dériaud, V. Huteau, L. BenMohamed, O. M. Diop, P. O. Livingston, S. Bay and C. Leclerc, Cancer Res., 2004, 64, 4987–4994.
16. J. P. Tam, Proc. Natl. Acad. Sci. U. S. A., 1988, 85, 5409–5413.
17. P. Niederhafner, J. Sebestik and J. Jezek, J. Pept. Sci., 2005, 11, 757–788.
18. H. Cai, Z.-Y. Sun, M.-S. Chen, Y.-F. Zhao, H. Kunz and Y.-M. Li, Angew. Chem., Int. Ed., 2014, 53, 1699–1703.
19. S. Keil, A. Kaiser, F. Syed and H. Kunz, Synthesis, 2009, 1355–1369.
20. G. Esposito, F. Fogolari, P. Viglino, S. Cattarinussi, M. T. Demagistris, L. Chiappinelli and A. Pessi, Eur. J. Biochem., 1993, 217, 171–187.
21. W. Kowalczyk, M. Monso, B. G. de la Torre and D. Andreu, J. Pept. Sci., 2011, 17, 247–251.
22. E. H. Nardin, J. M. Calvo-Calle, G. A. Oliveira, P. Clavijo, R. Nussenzweig, R. Simon, W. Zeng and K. Rose, Vaccine, 1998, 16, 590–600.
23. P. Niederhafner, L. Bednarova, M. Budesinsky, M. Safarik, S. Ehala, J. Jezek, L. Borovickova, V. Fucik, V. Cerovsky and J. Slaninova, Amino Acids, 2010, 39, 1553–1561.
24. D. Laubreton, S. Bay, C. Sedlik, C. Artaud, C. Ganneau, E. Dériaud, S. Viel, A. L. Puaux, S. Amigorena, C. Gérard, R. Lo-Man and C. Leclerc, Cancer Immunol. Immunother., 2016, 65, 315–325.
25. J. P. Tam and Y. A. Lu, J. Am. Chem. Soc., 1995, 117, 12058–12063.
26. J. Sebestik, P. Niederhafner and J. Jezek, Amino Acids, 2011, 40, 301–370.
27. M. Monso, W. Kowalczyk, D. Andreu and B. G. de la Torre, Org. Biomol. Chem., 2012, 10, 3116–3121.
28. J. C. Spetzler and J. P. Tam, Int. J. Pept. Protein Res., 1995, 45, 78–85.
29. V. Cavallaro, P. Thompson and M. Hearn, J. Pept. Sci., 2001, 7, 262–269.
30. M. Monso, J. Tarradas, B. G. de la Torre, F. Sobrino, L. Ganges and D. Andreu, J. Pept. Sci., 2011, 17, 24–31.
31. R. A. Boykins, M. Joshi, C. Syin, S. Dhawan and H. Nakhasi, Peptides, 2000, 21, 9–17.
32. T. Wohr, F. Wahl, A. Nefzi, B. Rohwedder, T. Sato, X. C. Sun and M. Mutter, J. Am. Chem. Soc., 1996, 118, 9218–9227.
33. J. P. Tam, W. F. Heath and R. B. Merrifield, J. Am. Chem. Soc., 1986, 108, 5242–5251.
34. N. Gaidzik, U. Westerlind and H. Kunz, Chem. Soc. Rev., 2013, 42, 4421–4442.
35. M. C. Galan, D. Benito-Alifonso and G. M. Watt, Org. Biomol. Chem., 2011, 9, 3598–3610.
36. D. P. Gamblin, E. M. Scanlan and B. G. Davis, Chem. Rev., 2009, 109, 131–163.
37. C. Nativi and O. Renaudet, ACS Med. Chem. Lett., 2014, 5, 1176–1178.
38. D. Valmori, A. Sabbatini, A. Lanzavecchia, G. Corradin and P. M. Matricardi, J. Immunol., 1994, 152, 2921–2929.
39. A. Borgert, J. Heimburg-Molinaro, X. Song, Y. Lasanajak, T. Ju, M. Liu, P. Thompson, G. Ragupathi, G. Barany, D. F. Smith, R. D. Cummings and D. Live, ACS Chem. Biol., 2012, 7, 1031–1039.
40. F. Corzana, J. H. Busto, G. Jimenez-Oses, M. Garcia de Luis, J. L. Asensio, J. Jimenez-Barbero, J. M. Peregrina and A. Avenoza, J. Am. Chem. Soc., 2007, 129, 9458–9467.
41. J. Schuman, D. Qiu, R. R. Koganty, B. M. Longenecker and A. P. Campbell, Glycoconjugate J., 2000, 17, 835–848.
42. D. M. Coltart, A. K. Royyuru, L. J. Williams, P. W. Glunz, D. Sames, S. D. Kuduk, J. B. Schwarz, X. T. Chen, S. J. Danishefsky and D. H. Live, J. Am. Chem. Soc., 2002, 124, 9833–9844.
43. F. Corzana, J. H. Busto, M. Garcia de Luis, J. Jimenez-Barbero, A. Avenoza and J. M. Peregrina, Chemistry, 2009, 15, 3863–3874.
44. O. Blixt, E. Clo, A. S. Nudelman, K. K. Sorensen, T. Clausen, H. H. Wandall, P. O. Livingston, H. Clausen and K. J. Jensen, J. Proteome Res., 2010, 9, 5250–5261.
45. H. Coelho, T. Matsushita, G. Artigas, H. Hinou, F. J. Canada, R. Lo-Man, C. Leclerc, E. J. Cabrita, J. Jimenez-Barbero, S. Nishimura, F. Garcia-Martin and F. Marcelo, J. Am. Chem. Soc., 2015, 137, 12438–12441.
46. N. Martinez-Saez, J. Castro-Lopez, J. Valero-Gonzalez, D. Madariaga, I. Companon, V. J. Somovilla, M. Salvado, J. L. Asensio, J. Jimenez-Barbero, A. Avenoza, J. H. Busto, G. J. Bernardes, J. M. Peregrina, R. Hurtado-Guerrero and F. Corzana, Angew. Chem., Int. Ed., 2015, 54, 9830–9834.
47. D. Mazal, R. Lo-Man, S. Bay, O. Pritsch, E. Deriaud, C. Ganneau, A. Medeiros, L. Ubillos, G. Obal, N. Berois, M. Bollati-Fogolin, C. Leclerc and E. Osinaga, Cancer Immunol. Immunother., 2013, 62, 1107–1122.
48. E. Kagan, G. Ragupathi, S. S. Yi, C. A. Reis, J. Gildersleeve, D. Kahne, H. Clausen, S. J. Danishefsky and P. O. Livingston, Cancer Immunol. Immunother., 2005, 54, 424–430.
49. Z. Yin, S. Chowdhury, C. McKay, C. Baniel, W. S. Wright, P. Bentley, K. Kaczanowska, J. C. Gildersleeve, M. G. Finn, L. BenMohamed and X. Huang, ACS Chem. Biol., 2015, 10, 2364–2372.
50. T. Freire, X. Zhang, E. Deriaud, C. Ganneau, S. Vichier-Guerre, E. Azria, O. Launay, R. Lo-Man, S. Bay and C. Leclerc, Blood, 2010, 116, 3526–3536.
51. M. Tsuiji, M. Fujimori, Y. Ohashi, N. Higashi, T. M. Onami, S. M. Hedrick and T. Irimura, J. Biol. Chem., 2002, 277, 28892–28901.
52. D. S. Wishart, C. G. Bigam, J. Yao, F. Abildgaard, H. J. Dyson, E. Oldfield, J. L. Markley and B. D. Sykes, J. Biomol. NMR, 1995, 6, 135–140.
53. M. Rance, O. W. Sorensen, G. Bodenhausen, G. Wagner, R. R. Ernst and K. Wuthrich, Biochem. Biophys. Res. Commun., 1983, 117, 479–485.
54. C. Griesinger, G. Otting, K. Wuthrich and R. R. Ernst, J. Am. Chem. Soc., 1988, 110, 7870–7872.
55. S. Macura, Y. Huang, D. Suter and R. R. Ernst, J. Magn. Reson., 1981, 43, 259–281.
56. W. Willker, D. Leibfritz, R. Kerssebaum and W. Bermel, Magn. Reson. Chem., 1993, 31, 287–292.
57. L. E. Kay, P. Keifer and T. Saarinen, J. Am. Chem. Soc., 1992, 114, 10663–10665.
58. N. A. Farrow, R. Muhandiram, A. U. Singer, S. M. Pascal, C. M. Kay, G. Gish, S. E. Shoelson, T. Pawson, J. D. Formankay and L. E. Kay, Biochemistry, 1994, 33, 5984–6003.
59. L. E. Kay, L. K. Nicholson, F. Delaglio, A. Bax and D. A. Torchia, J. Magn. Reson., 1992, 97, 359–375.
60. A. Pajot, M. L. Michel, N. Fazilleau, V. Pancre, C. Auriault, D. M. Ojcius, F. A. Lemonnier and Y. C. Lone, Eur. J. Immunol., 2004, 34, 3060–3069.

---

Footnotes

† Electronic supplementary information (ESI) available: Four figures, three tables, detailed NMR analyses (chemical shift, coupling constants, ¹H, ¹H NOEs). See DOI: 10.1039/c6ob01931e

‡ These authors contributed equally to this work.

§ Present address: Institut Pasteur, Unité d'Histopathologie Humaine et Modèles Animaux, Paris, France.

---