A PEGylated Photocleavable Auxiliary Mediates the Sequential Enzymatic Glycosylation and Native Chemical Ligation of Peptides

Abstract

Research aimed at understanding the specific role of glycosylation patterns in protein function would greatly benefit from additional approaches allowing direct access to homogeneous glycoproteins. Here the development and application of an efficient approach for the synthesis of complex homogenously glycosylated peptides based on a multifunctional photocleavable auxiliary is described. The presence of a PEG polymer within the auxiliary enables sequential enzymatic glycosylation and straightforward isolation in excellent yields. The auxiliary-modified peptides can be directly used in native chemical ligations with peptide thioesters easily obtained via direct hydrazinolysis of the respective glycosylated peptidyl resins and subsequent oxidation. The ligated glycopeptides can be smoothly deprotected via UV irradiation. Here we apply this approach to the preparation of variants of the epithelial tumor marker MUC1 carrying one or more Tn, T or sialyl-T antigens.

Posttranslational modifications (PTMs) are essential in determining the folding and activity of many proteins and their misregulation is often involved in the progression of severe diseases.^[1] Despite many advances in our understanding of protein modifications with carbohydrates,^[2] and the ability to generate glycosylated proteins by highly evolved chemical synthesis approaches^[3] as well as by combinations of enzymatic and synthetic means^[4] insights into the specific roles of this class of PTMs in protein function remain a challenge due to their complexity and heterogeneity. We have recently developed an efficient methodology for the preparation of site-specifically O-glycosylated Mucin 1 (MUC1) polypeptides,^[5] in which the presence of a PEG polymer at the N-terminus of the peptides allowed quantitative enzymatic glycosylation in solution and to recover the modified peptides by simple precipitation and spin column gel permeation chromatography (GPC), avoiding elaborate intermediate purifications. This method proved to be high-yielding and versatile, as it could be coupled with native chemical ligation (NCL), thereby giving access to longer site-specifically O-glycosylated peptides.

However, the proteolytic removal of the polymer required installation of a 6 amino acid protease recognition site and a non-native cysteine had to be introduced to carry out NCL reactions. To overcome these limitations we here combine both functionalities (PEG attachment^[6] and thiol group for NCL^[7]) in a photocleavable auxiliary, based on the 1-nitrophenyl-2-mercaptoethyl scaffold previously described by the groups of Aimoto^[8] and Dawson.^[9] Upon photolytic removal a native glycine residue remains at the ligation site. The mild deprotection conditions together with the high occurrence of glycine in proteins have made this approach suitable for several applications in protein (semi-) synthesis.^[10] Here we present the synthesis and application of a photocleavable auxiliary that allows attachment of the PEG polymer for efficient enzymatic glycosylation, provides the functional groups for NCL and is cleanly removed by UV irradiation (Scheme 1).

Scheme 1.

The auxiliary-mediated chemoenzymatic glycosylation approach. The solid phase peptide synthesis (SPPS) of a 20mer Mucin1 (MUC1) motif with a PEGylated auxiliary provides material for consecutive glycosylation and precipitation steps (right). The same peptide precursor can be converted into a peptide hydrazide and subsequently into a peptide thioester (left). Both peptides are used in ligation reactions to give extended glycosylated MUC1 repeats.

We based the first part of the auxiliary synthesis on the procedure developed by Dawson and coworkers.^[9] Starting from vanillin, the desired methyl butanoate, nitro and methylene groups were efficiently introduced (Scheme S1). The planned subsequent Sharpless aminohydroxylation^[11] was not efficient, even using different conditions. Therefore, a stepwise strategy was adopted, starting with the Sharpless dihydroxylation^[12] to obtain the diol, which was then converted into the corresponding amino alcohol in three steps and 63% overall yield.^[13]

The Mitsunobu reaction,^[14] followed by protecting group exchange (acetyl to tert-butylsulfanyl),^[15] gave the protected mercaptoaminoethyl group essential for the NCL reaction. Catalytic ester-amide exchange^[16] provided the attachment point for the PEG polymer and potentially for many other modifications. Final PEGylation of the free primary amine, removal of the Boc group and purification gave the desired auxiliary in 68% yield over the last two steps (Scheme S1).

The desired auxiliary in hand, we investigated the conditions for its incorporation into a MUC1 peptide replacing a native N-terminal glycine residue (Scheme 1). The MUC1 tandem repeat peptide was synthesized by Fmoc-based solid phase peptide synthesis (SPPS) and the N-terminus was modified with iodoacetic acid (Figure S3) via HBTU-mediated coupling and then submitted to an S_N2 reaction with auxiliary 7. Unfortunately, all tested conditions gave poor conversion even after several days. To improve this critical step, the synthesis of the MUC1 peptide was repeated on TentaGel resin, with lower peptide loading and better swelling properties when compared to the previously used Wang PS resin. To reduce any potential steric hindrance by the long PEG chain during the substitution reaction, auxiliary 8, missing the PEG moiety, was installed first and subsequently PEGylated on resin. The auxiliary synthesis was modified accordingly by introducing an Fmoc protecting group on the free primary amine (Scheme S1m).

The S_N2 reaction of auxiliary 8 with the iodoacetylated MUC1 peptidyl (Wang TentaGel) resin now proceeded with very high conversion (Figure S5) and PEGylation on resin was quantitative. After cleavage and purification, the pure MUC1-auxiliary conjugate Aux-MUC1 9 (Figure S7) was obtained in 34% yield.

This strategy was applied for the preparation of an auxiliary carrying MUC1 peptide including the Tn antigen (GalNacα1-O-Thr) at position 14 (Aux-MUC1(Tn) 10). The use of Fmoc-Thr(GalNAc-Ac₃)-OH ensures full control over the O-glycan attachment site. Alternatively Drosophila polypeptidyl GalNAc-transferase 1 (dGalNAcT1) can be employed for site-selective generation of the Tn antigen.^[5] Here iodoacetylation, attachment of the auxiliary, cleavage from the resin and subsequent hydrazinolysis of the acetyl protecting groups on the glycan gave Aux-MUC1(Tn) 10 (Figure S8) in 4% overall yield (based on synthesis scale).

In a first attempt to prepare 10 the GalNAc deprotection was performed on resin by overnight incubation with 5% (v/v) hydrazine monohydrate in methanol. Surprisingly, this did not only lead to acetyl removal but to complete release of the peptide from Wang TentaGel resin. Further investigation demonstrated that hydrazine quantitatively cleaved the protected peptide from resin by nucleophilic attack on the linker, forming a peptide hydrazide.^[17][18] This initially appeared as a nuisance on route to Aux-MUC1(Tn) 10, however since peptide hydrazides can be efficiently converted into the corresponding peptide thioesters,^[19] we decided to take advantage of this intermediate to generate larger MUC1 segments by sequential NCL reactions with auxiliary and hydrazide modified MUC1 peptides. To fully exploit the bi-functional MUC1 peptides we needed to prove that MUC1-NHNH₂ 16 can be easily converted into the corresponding thioester and that chemoenzymatic glycosylation works on Aux-MUC1-NHNH₂ 19. We knew from previous attempts that peptide thioesters hydrolyze under glycosylation conditions therefore a peptide hydrazide could be an ideal masked peptide thioester sufficiently stable to allow enzymatic glycosylation and subsequent conversion into the desired glycosylated peptide thioester. Moreover, the same peptidyl resin provided access both to the peptide acid and to the peptide hydrazide.

To test conversion of MUC1-NHNH₂ 16 into the corresponding MUC1-thioester MUC1-SR 17, MUC1-peptidyl resin was incubated with 5% (v/v) hydrazine monohydrate in methanol overnight. Protected MUC1-NHNH₂ 15 was obtained in 85% yield and deprotection of the side chains in solution by treatment with a mixture of TFA/TIS/H₂O (92.5:5:2.5) gave the desired MUC1-NHNH₂ 16 (Figure S9). Treatment with NaNO₂ followed by addition of MesNa and HPLC purification gave the MesNa thioester MUC1-SR 17 (Figure S10) in 14% overall yield.

Next we demonstrated that the PEGylated auxiliary was indeed able to facilitate the fast sequential enzymatic glycosylation of the MUC1 peptides as previously described for N-terminally PEGylated MUC1.^[5] Glycosylation of Aux-MUC1(Tn) 10 with human C1GalT1 gave Aux-MUC1(T) 11 with a single T antigen (Galβ1-3GalNAcα disaccharide attached to Thr14) in excellent conversion (Figure 1 & S12A, B). The incubation with a mixture of ethanol and diethyl ether at −80° C induced the precipitation of 11, which was then collected in 95% yield by centrifugation (Figure 1 & S12B) and used in the next glycosylation step without further purification. Incubation of Aux-MUC1(T) 11 with CMP-Neu5Ac in the presence of recombinant ST3Gal1 allowed efficient elongation of the disaccharide to the Neu5Acα2-3Galβ1-3GalNacα-trisaccharide and gave Aux-MUC1(sT) 12 with the sialyl-T antigen in 90% yield (Figure 1 & S12B). Using similar conditions for all glycosylation reactions but by increasing incubation times with ethanol and ether at −80°C (6-12 h instead of 4 h) we were able to simplify the sequential glycosylation procedure by omitting the GPC spin column step. As a control we have also used MUC1(Tn) without the auxiliary in this glycosylation-precipitation procedure and only 27% of MUC(T) were recovered, clearly demonstrating the advantages of PEGylation (Figure S12C).

Figure 1.

Sequential enzymatic glycosylation of Aux-MUC1(Tn) 10 gives Aux-MUC1(T) 11 and Aux-MUC1(sT) 12 for NCL reactions with MUC1-SR 17.MUC1: Tandem repeat sequence of Mucin 1 VTSAPDTRPAPGSTAPPAH, the O-glycans on the side chain of Thr14 are indicated in brackets. Tn: GalNacα, T: Galβ1-3GalNAcα, sT: Neu5Acα2-3Galβ1-3GalNacα. Left panel: From top to bottom: HPLC chromatograms of 10, 11 and 12 after precipitation and of the NCL reaction after 24h. Injection peak at 2 min. Peak at 12.14 min: Peptide thioester 17. Right panel: ESI-MS spectra from top to bottom: Aux-MUC1(Tn) 10, Aux-MUC1(T) 11, Aux-MUC1(ST) 12, MUC1-Aux-MUC1(sT) 13.

Next, the glycosylated peptide conjugates were linked to MUC1-SR 17 via NCL to demonstrate all advantages of the auxiliary. Non-glycosylated Aux-MUC1 9 was used to establish optimal ligation conditions. The thiol group of the auxiliary was deprotected via incubation with TCEP at 24°C for 6 h before addition of MUC1-SR 17. Optimized NCL conditions for MUC1 peptides (65% conversion after two days) are NaPi buffer (pH 7.5) at 30°C with Aux-MUC1 9 at 8 mM concentration and a 2.5-fold excess of 17 (Figure S15A). These conditions were applied for the synthesis of glycosylated MUC1-Aux-MUC1(Tn) 18, which was obtained from MUC1-SR 17 and Aux-MUC1(Tn) 10 in 78% conversion after 36 h incubation at 30°C (Figure S15B).

After demonstrating the individual functions of the auxiliary in MUC1 peptides we coupled glycosylation and NCL reaction by performing sequential enzymatic glycosylation of Aux-MUC1(Tn)10 to the sialylated core 1-containing conjugate Aux-MUC1(sT) 12, which was recovered by precipitation and directly used in a ligation reaction with 17 (Figure 1). In this case consecutive additions of TCEP were needed to efficiently remove the tertbutylsulfanyl group from the auxiliary (Figure S15C). Addition of MUC1-SR 17 and MesNa to the ligation mixture led to the desired product 13 in 1 d and with 70% conversion (Figure 1).

Finally, light induced removal of the PEGylated auxiliary was demonstrated for non-glycosylated MUC1-Aux-MUC1 as well as for 13 and 18. This was accomplished by UV irradiation of the crude ligation mixtures in water or in a water/acetonitrile mixture. In all cases no starting material was detectable after 30 minutes of irradiation with an UV-A lamp and simultaneously a new peak containing the desired product was formed and isolated by HPLC purification (Figures S16-18).

As described above, peptide hydrazides are useful thioester precursors that remain unaffected in glycosylation reactions in which peptide thioesters quickly hydrolyze. Hydrazine-induced cleavage of Aux-MUC1(Tn) peptidyl resin gave protected Aux-MUC1(Tn)-NHNH₂ 19 (Figure S11A) that, after acidic deprotection in solution, was successfully used in sequential glycosylation reactions giving the corresponding hydrazide Aux-MUC1(sT)-NHNH₂ 21 with similar yields as found for Aux-MUC1(sT)-OH 12 (Figure 2).

Figure 2.

Sequential enzymatic glycosylation of Aux-MUC1-NHNH₂ 19, a precursor for glycosylated peptide thioesters and the corresponding ESI-MS spectra (from bottom to top): Aux-MUC1(Tn)-NHNH₂ 19, Aux-MUC1(T)-NHNH₂ 20, Aux-MUC1(sT)-NHNH₂ 21.

Treatment with NaNO₂ followed by ascorbic acid (to suppress nitrosamine formation on the auxiliary) and MesNa gave access to the desired thioester Aux-MUC1(sT)-SR 22 in 63%.

To further explore the potential of this approach, the synthesis of fully unprotected conjugates comprising two glycosylated MUC1 peptides was accomplished (Figure 3). A MUC1 peptide carrying the Tn antigen at position Thr7 (23, instead of Thr14) was synthesized and converted into a peptide α-thioester. Hydrazine cleavage, deprotection, treatment with NaNO₂ and MesNa smoothly led to the desired MUC1(Tn⁷)-SR 24 in 43% yield (Figure S20). 24 was used in NCL with Aux-MUC1(Tn) 10 to give MUC1(Tn⁷)-Aux-MUC1(Tn) 25 with 35% yield (Figures S21).

Figure 3.

HPLC chromatograms and mass spectra of unprotected glycosylated peptides MUC1(T⁷)-G-MUC1(T) 27 and MUC1(sT⁷)-G-MUC1(sT) 29 obtained after native chemical ligation, glycosylation and UV-irradiation.

25 was efficiently used in the glycosylation-precipitation procedure leading to MUC1(T⁷)-Aux-MUC1(T) 26 (Figure S22), consisting of two MUC1 tandem repeats with a T antigen at different positions. This much longer peptide with only one auxiliary was efficiently recovered by precipitation (80% recovery) under similar conditions as described above. Subsequently, 26 was submitted to UV irradiation at 365 nm for 6 min and the desired MUC1(T⁷)-G-MUC1(T) 27 was obtained in 53% yield. MUC1(T⁷)-Aux-MUC1(T) 26 was also used in a further glycosylation-precipitation step leading to sialylated MUC1(sT⁷)-Aux-MUC1(sT) 28 (68% recovery, Figure S23). UV irradiation cleanly removed the auxiliary and pure MUC1(sT⁷)-GMUC1(sT) 29 was obtained in 12% yield after HPLC purification (Figure 3). These results demonstrate the power of this approach for the chemoenzymatic synthesis of glycosylated peptides and its combination with NCL to obtain larger polypeptides with different but specific glycosylation patterns.

In summary, we have developed a PEGylated ligation auxiliary that efficiently supports the sequential quantitative enzymatic glycosylation of peptides in solution without the need for chromatographic purification and can mediate NCL reactions. The auxiliary-modified (and glycosylated) peptides can be used in NCL reactions with other MUC1 tandem repeat peptides carrying a C-terminal thioester. All peptide α-thioesters used here are obtained via hydrazinolysis followed by oxidation of the corresponding hydrazide. Subsequent NCL reactions carried out in this study give conversions of > 65% and the auxiliary is cleanly removed from ligation products. Light-induced removal of the auxiliary leaves a glycine residue at the ligation site limiting the use of this approach to glycine residues, which are fortunately quite abundant in proteins. The new PEGylated auxiliary in combination with generating peptide α-thioesters via direct hydrazinolysis allows preparation of site-selectively O-glycosylated peptide α-thioesters, opening the way to the synthesis of polypeptides comprising two or more MUC1 tandem repeats with different glycosylation patterns. Controlling thioester generation and de-protection of the thiol group within the auxiliary also allows the controlled extension of each building block in C- and N-terminal direction. We will use this approach to create a library of site-selectively O-glycosylated MUC1 variants with different glycosylation pattern for a detailed study of these patterns in MUC1 function.

However, this approach is not limited to the synthesis of glycosylated MUC1 analogues but it is applicable to the synthesis of many larger, posttranslationally modified proteins, only limited by identifying a suitable glycine residue as well as the availability of chemistry or enzymes that introduce the desired PTMs.