Substrate Recognition of Glycoprotein Folding Sensor UGGT

Jul 25, 2017 - UDP-glucose:glycoprotein glucosyltransferase (UGGT) distinguishes glycoproteins in non-native conformations from those in native confor...
0 downloads 16 Views 2MB Size
Download PDF
Article pubs.acs.org/JACS

Substrate Recognition of Glycoprotein Folding Sensor UGGT Analyzed by Site-Specifically 15N‑Labeled Glycopeptide and Small Glycopeptide Library Prepared by Parallel Native Chemical Ligation Masayuki Izumi,†,⊥ Rie Kuruma,† Ryo Okamoto,† Akira Seko,‡ Yukishige Ito,*,‡,§ and Yasuhiro Kajihara*,†,‡ †

Department of Chemistry, Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan ERATO Ito glycotrilogy project, Japan Science and Technology Agency (JST), 2-1 Hirosawa, Wako, Saitama 351-0198, Japan § Synthetic Cellular Chemistry Laboratory, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan ‡

S Supporting Information *

ABSTRACT: UDP-glucose:glycoprotein glucosyltransferase (UGGT) distinguishes glycoproteins in non-native conformations from those in native conformations and glucosylates from only non-native glycoproteins. To analyze how UGGT recognizes non-native glycoproteins, we chemically synthesized site-specifically 15N-labeled interleukin 8 (IL-8) C-terminal (34−72) glycopeptides bearing a Man9GlcNAc2 (M9) oligosaccharide. Chemical shift perturbation mapping NMR experiments suggested that Phe65 of the glycopeptide specifically interacts with UGGT. To analyze this interaction, we constructed a glycopeptide library by varying Phe65 with 10 other natural amino acids, via parallel native chemical ligation between a glycopeptide-α-thioester and a peptide library consisting of 11 peptides. UGGT assay against the glycopeptide library revealed that, although less hydrophobic glycopeptides could be used as substrates for UGGT, hydrophobic glycopeptides are preferred.

■

INTRODUCTION

Biosynthesis of glycoproteins starts in the endoplasmic reticulum (ER).1,2 In the ER, a significant fraction of newly synthesized glycoproteins fail to fold into their native structures.3 To prevent the accumulation of deleterious misfolded glycoproteins and maximize the efficiency of glycoprotein biosynthesis, the glycoprotein quality control (QC) system in the ER captures non-native glycoproteins and converts them into their native structures.4−6 UDP-glucose:glycoprotein glucosyltransferase (UGGT) is a key enzyme in the glycoprotein QC system. UGGT distinguishes non-native glycoproteins bearing a Man9GlcNAc2 (M9) oligosaccharide from native glycoproteins and transfers a glucose residue from UDP-glucose onto the M9 oligosaccharide, converting it to Glc1Man9GlcNAc2 (G1M9) (Figure 1).7,8 UGGT is thus referred to as a folding sensor due to its ability to sense the folding status of the glycoprotein. A non-native glycoprotein bearing a G1M9 oligosaccharide that formed as a result of this process can interact with lectin chaperones calnexin/ calreticulin, which refolds the glycoprotein into its native conformation. The ability of UGGT to distinguish glycoproteins in nonnative conformations from those folded in the native conformation is well-known; however, it is still unclear as to how UGGT is able to recognize the folding status of glycoproteins. Proteins in non-native conformations tend to expose hydrophobic amino acid residues; therefore, UGGT is thought to recognize hydrophobic patches exposed on the © 2017 American Chemical Society

Figure 1. Discrimination of folding status of glycoprotein by folding sensor UDP-glucose:glycoprotein glucosyltransferase (UGGT) in the glycoprotein quality control system. UGGT transfers a glucose residue only to the M9 oligosaccharide on a glycoprotein having non-native conformation.

surface of the non-native glycoprotein. Parodi and co-workers reported that a glycoprotein mimic exposing hydrophobic patches on the surface, which was characterized by 1anilinonaphthalene-8-sulfonic acid (ANS) binding assay, is the more preferred substrate for UGGT.9 Ito and co-workers found that high-mannose-type oligosaccharides having a Received: April 1, 2017 Published: July 25, 2017 11421

DOI: 10.1021/jacs.7b03277 J. Am. Chem. Soc. 2017, 139, 11421−11426

Article

Journal of the American Chemical Society hydrophobic fluorescence group, such as methotrexate or 4,4difluoro-4-bora-3a,4a-diaza-s-indacene (BODIPY), at their reducing ends are good substrates for UGGT.10,11 Recently, Kato and co-workers reported a 3D structure of the third thioredoxin-like domain in the N-terminal folding sensor region of Chaetomium thermophilium UGGT and found an extensive hydrophobic patch which may function as a substrate binding site.12 We have been studying the substrate recognition properties of UGGT using chemically synthesized homogeneous glycoprotein probes. We synthesized interleukin 8 (IL-8) bearing an M9 oligosaccharide both in native form and in misfolded forms, which we produced by forming non-native disulfide bonds through intentional misfolding processes. We were able to find that the hydrophobic misfolded M9-IL-8 dimer is a better substrate for UGGT than less hydrophobic misfolded M9-IL-8.13 We also found that folding intermediates could be substrates for UGGT, as well.14 These increasing reports support the hypothesis that UGGT recognizes hydrophobic patches on the surface of a non-native glycoprotein; however, to the best of our knowledge, no direct evidence of the interaction between UGGT and this hydrophobic surface has been reported. In this article, we describe chemical shift perturbation mapping NMR experiments to analyze the interaction between site-specifically 15N-labeled M9-glycopeptide and UGGT. We also discuss our findings of the involvement of a specific hydrophobic amino acid residue in UGGT recognition of nonnative glycopeptides. We also prepared a glycopeptide library consisting of 11 glycopeptides by varying this specific amino acid residue with 10 other proteinogenic amino acids, using parallel native chemical ligation of a single glycopeptide-αthioester and a peptide library consisting of 11 peptides. UGGT assay against the glycopeptide library revealed that glycopeptides having hydrophobic amino acid residues are the preferable substrates for UGGT, although UGGT is able to glucosylate all 11 glycopeptides.

oligosaccharide that can be isolated from egg yolk limit the availability of this small glycopeptide containing naturally abundant 15N atoms. In order to overcome these problems, we designed a protocol to introduce 15N labels to nine hydrophobic amino acid residues (Ala35, Val41, Leu43, Gly46, Leu49, Val58, Val62, Phe65, Ala69) out of the 39 amino acid residues in the M9-IL-8(34−72) glycopeptide 1a, as shown in Figure 2a. We chose to label the hydrophobic amino acids

■

RESULTS AND DISCUSSION Design and Synthesis of Site-Specifically 15N-Labeled M9-IL-8(34−72) Glycopeptide and Chemical Shift Perturbation Mapping NMR Experiment with UGGT. In the course of our study on the substrate specificity of UGGT using M9-IL-8 derivatives, we found that derivatives having a non-native disulfide bond between Cys34 and Cys50 are preferable substrates for UGGT.15 Among these derivatives, we decided to use the M9-IL-8(34−72) glycopeptide as a model substrate of UGGT in this study, due to its suitable affinity with recombinant Aspergillus oryzae UGGT (Km = 3.4 ± 0.3 μM) and its convenient size (39 amino acid residues with one M9 oligosaccharide), for the chemical synthesis and NMR analysis. To analyze which part of the glycopeptide is recognized by UGGT as an indicator of a non-native glycoprotein, we employed a chemical shift perturbation mapping NMR experiment, which is a widely used method to map protein− protein interactions. M9-IL-8(34−72) glycopeptide bearing homogeneous M9 oligosaccharide can only be obtained via chemical synthesis. Ideally, all amino acid residues would be labeled with 15N for efficient assignment of the NMR signals and complete analysis of the interaction. Unfortunately, not all 15 N-labeled amino acids with suitable protected forms for solidphase peptide synthesis (SPPS) are commercially available. Conducting 1H−15N NMR experiments is difficult because the minute amounts of UGGT that can be expressed and M9

Figure 2. (a) Schematic representation of site-specifically 15N-labeled M9-IL-8(34−72) glycopeptide 1a. Amino acid residues indicated in black are positions labeled with 15N. Black bar between Cys34 and Cys50 indicates a disulfide bond. (b) RP-HPLC profile of purified 1a. (c) ESI-MS spectrum of purified 1a. [M + H]+ calcd 6396.8, found 6396.6. (d) Overlap of 1H−15N HSQC spectra of 1a in the presence of recombinant Penicillium chrysogenum UGGT. The ratios between 1a and UGGT were 1:0 (black), 50:1 (blue), 30:1 (green), and 20:1 (red). Red arrow indicates the peak shift observed depending on the ratio between 1a and UGGT.

because UGGT is reported to recognize hydrophobic patches on the surface of a non-native glycoprotein and because Bocprotected 15N-labeled hydrophobic amino acids required for SPPS are commercially available. We did not label all hydrophobic residues because we hypothesized that UGGT recognizes a hydrophobic patch that is formed with several amino acid residues. 15N labeling of one amino acid within the hydrophobic patch is expected to show a clear interaction between UGGT and the hydrophobic patch, using NMR chemical shift perturbation. Therefore, we chose to introduce 15 N-labeled hydrophobic amino acids far from each other, allowing us to cover the entire surface of the glycopeptide. We also limited the number of the labeled residues for each amino acid type, as similar signals would be detected if multiple 11422

DOI: 10.1021/jacs.7b03277 J. Am. Chem. Soc. 2017, 139, 11421−11426

Article

Journal of the American Chemical Society

methoxy-2,3,6-trimethylbenzenesulfonyl) as opposed to Arg(diZ) (Z = carbobenzoxy) because we frequently encountered incomplete deprotection of the latter during deprotection of glycopeptide side-chain protecting groups using a low-acidity trifluoromethanesulfonic acid (TfOH) cocktail. Hydrolysis of mannose residues in M9 oligosaccharide occurred concomitantly when a high-acidity TfOH cocktail was employed for the deprotection step. Two-step deprotection by treatment with trifluoroacetic acid (TFA)/dimethylsulfide (DMS)/TfOH/mcresol/1,2-ethanedithiol (EDT) = 50:30:10:8:2 at 0 °C for 1 h followed by TFA/DMS/TfOH/m-cresol = 50:30:10:10 at 0 °C for 2 h was required to remove all Mtr groups. Thiolysis of the resin-bound glycopeptide-α-thioester with sodium 2-mercaptoethanesulfonate yielded the desired glycopeptide-α-thioester 2a (Figure S1). For the preparation of C-terminal peptide 3a having four 15N-labeled amino acid residues, we employed standard Boc SPPS on a Boc-Ser(Bzl)-OCH2-Pam resin, which was utilized because we had to use 15N-labeled Boc-amino acids (Figure S2). Native chemical ligation of glycopeptide-αthioester 2a and peptide 3a was carried out under the standard conditions, 19,20 and one-pot conversion of N-terminal thiazolidine to cysteine21 was carried out to obtain glycopeptide 4a (Figure S3). The desired M9-IL-8(34−72) glycopeptide 1a was obtained after formation of a disulfide bond between Cys34 and Cys50 under air oxidation conditions (Figures 2b,c and S4). We then employed chemical shift perturbation mapping NMR experiments using glycopeptide 1a, containing nine 15Nlabeled amino acid residues, and Penicillium chrysogenum UGGT. Since recombinant human UGGT1 is difficult to produce in mammalian cells in sufficient amounts for NMR experiments, we used P. chrysogenum UGGT that is easy to produce in Escherichia coli (Supporting Information). 1H−15N HSQC spectra were recorded by varying the ratio between 1a and UGGT from 1:0 to 50:1, 30:1, and 20:1. As shown in Figure 2d and Figure S5, most of the peaks were well overlapped, but one peak (δH 8.0 ppm, δN 117 ppm) gradually shifted depending on the ratio between 1a and UGGT. We speculated that this chemical shift perturbation was caused by the specific interaction between glycopeptide 1a and UGGT. We noticed that another peak (δH 8.2 ppm, δN 120 ppm) showed two signals at the ratio of 1a and UGGT to 20:1 (red). This may be due to the existence of two different conformations caused by tighter binding of the glycopeptide to UGGT at higher ratios. However, as the perturbation was not dose-dependent, we thus focused our attention on the peak at δH 8.0 ppm, δN 117 ppm. As a control experiment, titration NMR experiments using 1a and hydrophobic bovine serum albumin (BSA) were performed to examine if there were nonspecific hydrophobic interactions. No significant peak shift was observed even when using the ratio of 1a and BSA to 10:1 (Figure S6), suggesting that the chemical shift perturbation observed with 1a and UGGT is due to a specific interaction. To assign the peak that showed the chemical shift perturbation, we synthesized glycopeptide 1b containing four 15 N-labeled amino acid residues at the C-terminal region, using nonlabeled glycopeptide-α-thioester 2b and site-specifically 15 N-labeled C-terminal peptide 3a (Scheme 1b and Figure S7a,b). The desired peak was observed in the 1H−15N HSQC spectrum of 1b. We then carefully inspected the NOESY spectrum of 1a and speculated that the residue might be Phe65. We thus synthesized glycopeptide 1c containing the 15N label only at Phe65 with nonlabeled glycopeptide-α-thioester 2b and

quantities of the same amino acid are labeled, making peak assignment difficult. Site-specifically 15N-labeled M9-IL-8(34−72) glycopeptide 1a was synthesized chemically, as depicted in Scheme 1a (see Scheme 1. Synthesis of Site-Specifically 15N-Labeled M9-IL8(34−72) Glycopeptides (a) 1a and (b) 1b and 1ca

a

Letters indicate positions of 15N-labeled amino acid residues (1a: A35, V41, L43, G46, L49, V58, V62, F65, A69; 1b: V58, V62, F65, A69; 1c: F65). Reagents and conditions: (1) 6 M Gn·HCl/0.2 M Na2HPO4, pH 6.7/0.1 M MPAA/20 mM TCEP, overnight; (2) 0.2 M MeONH2·HCl, pH 4.5; (3) 1 M Gn·HCl/0.1 M Tris-acetate, pH 8.5, freshly bubbled with air for 5 min.

Supporting Information for full details of the synthesis). The full-length glycopeptide, consisting of 39 amino acids, was divided into two segments: M9-IL-8(34−49)-α-thioester 2a and IL-8(50−72) peptide 3a. Glycopeptide-α-thioester 2a was prepared on an amino-PEGA resin bearing a mercaptopropionamide linker,16 and in situ neutralization Boc protocol was employed for the peptide elongation step.17 Two percent of commercially available 15N-labeled Boc-amino acid was mixed with nonlabeled Boc-amino acid for the coupling step of the designated 15N-labeling position. The sensitivity of NMR spectroscopy is sufficient enough to detect such a small amount of 15N-labeled amino acid, which is beneficial as it reduces the amount of the expensive 15N-labeled Boc-amino acids needed. Boc-Asn(M9-oligosaccharide)−OH was prepared from hen egg yolk as previously reported and used in the coupling at Asn36.18 We chose to use Boc-Arg(Mtr)−OH (Mtr = 411423

DOI: 10.1021/jacs.7b03277 J. Am. Chem. Soc. 2017, 139, 11421−11426

Article

Journal of the American Chemical Society 15

N-labeled peptide 3b, which had the 15N label only at Phe65 (Scheme 1b and Figure S7c,d). Only one peak at δH 8.0 ppm, δN 117 ppm was observed in the 1H−15N HSQC spectrum of 1c (Figure S8b), confirming that the peak showing chemical shift perturbation is Phe65. Synthesis of M9-IL-8(34−72) Glycopeptide Library by Parallel Native Chemical Ligation and UGGT Assay. Based on the results of the chemical shift perturbation mapping NMR experiments, we hypothesized that UGGT recognizes a hydrophobic patch formed with several amino acid residues including Phe65. To examine this hypothesis, we designed a small glycopeptide library by varying the amino acid residues at position 65. To vary the hydrophobicity and analyze the effects of other characteristics of amino acids on the substrate specificity of UGGT, we selected 11 amino acids that are acidic, basic, hydrophilic, and hydrophobic: Glu, Lys, Gln, Ser, Tyr, Gly, Ala, Pro, Phe, Leu, and Val. For efficient synthesis of the glycopeptide library, we introduced a parallel native chemical ligation (pNCL) strategy. As shown in Figure 3a, we planned on constructing M9-IL-8(34−72) glycopeptide library 7 varying in amino acid type at position 65 by pNCL of M9-IL-8(34−49) glycopeptide-α-thioester 2b and IL-8(50−72)

peptide library 5, which consists of 11 peptides. We envisaged that all peptides in library 5 could be ligated with glycopeptideα-thioester 2b by a single NCL reaction in a parallel manner to obtain the desired glycopeptide library 7. Not only does this library approach allow for the rapid analysis of UGGT substrate specificity, it also allows us to avoid the need to purify each glycopeptide, making the most of the precious M9 oligosaccharide. We decided to use 11 amino acids out of the 20 natural amino acids in this study for the clear LC-MS analysis of the new pNCL reaction. As mentioned earlier, we think that these 11 amino acids will cover the major characteristics of amino acids, such as hydrophilic, hydrophobic, acidic, and basic amino acids. Peptide library 5 was prepared by the split and mix method22 using standard Fmoc SPPS using 4-hydroxymethylphenoxyacetamide-PEGA resin. After coupling of Leu66, the resin was divided into 11 portions, and each portion of the resin was coupled with one of the above-mentioned 11 amino acids. All of the resin was then combined, and peptide elongation was continued to Cys50. The crude peptide library was purified by RP-HPLC, and fractions of the desired peptides were combined to yield peptide library 5 (Figure 3b and Figure S9). Parallel NCL of glycopeptide-α-thioester 2b and peptide library 5 consisting of 11 peptides was then examined under the standard conditions,19 employing about 3 mM of the peptide library 5 and 1.9 mM of 2b. Formation of all the desired ligated glycopeptides was confirmed by LC-MS analysis (Figure S10). Subsequent one-pot conversion of N-terminal thiazolidine to cysteine afforded the library of deprotected glycopeptide 6. Finally, glycopeptide library 6 was subjected to air oxidation to form a disulfide bond between Cys34 and Cys50, yielding the desired glycopeptide library 7 after RP-HPLC purification (Figures 3c and S11). LC-MS analysis confirmed that all 11 glycopeptides were included in the library 7 (Figure 4), although the relative amounts of each glycopeptide estimated from the peak area of the UV chromatogram varied from 0.71 to 1.45 (Table S1). The ratio between each glycopeptide is similar to that of peptide library 5, indicating that ligation of each peptide proceeded equally. The glucose transfer rate of recombinant human UGGT was examined using glycopeptide library 7. As depicted in Figure 5a, UGGT reaction was carried out using the assay solution containing all 11 glycopeptides (totaling approximately 5 μM, an estimation based on average molecular weight). The glucose transfer yield of each glycopeptide was estimated by the signal intensities of the LC-MS spectra of each library component, under the assumption that glycopeptides bearing M9 oligosaccharide and G1M9 oligosaccharide have similar ionization efficiencies (Figures S13−S16). Although estimated yields based on MS spectra could not be interpreted quantitatively, the LC-MS-based assay was simple and efficient enough to analyze the tendency of the substrate preference of UGGT. As shown in Figure 5b, after 1 h, it is apparent that glucosylation yields of glycopeptides having hydrophobic amino acids (Val, Phe, Leu) were higher than those of other glycopeptides. As the solution contained all 11 glycopeptides, it can be said that this assay mimicked the environment of the biosynthesis of glycoproteins in the ER. Regardless of these competitive conditions, UGGT transferred glucose preferentially to glycopeptides having hydrophobic residues at position 65. These results are consistent with the result of the chemical shift perturbation mapping NMR experiment, which suggested that there was a hydrophobic interaction between UGGT and

Figure 3. (a) Synthesis of M9-IL-8(34−72) glycopeptide library 7 by parallel native chemical ligation (pNCL). Letters indicate amino acid residues varied at position 65. Reagents and conditions: (1) 6 M Gn· HCl/0.2 M Na2HPO4/0.1 M MPAA/20 mM TCEP/pH 6.7, overnight; (2) 0.2 M MeONH2·HCl/pH 4.5; (3) 1 M Gn·HCl/0.1 M Tris-acetate/pH 8.5. (b) RP-HPLC profile of purified peptide library 5. (c) RP-HPLC profile of purified glycopeptide library 7. 11424

DOI: 10.1021/jacs.7b03277 J. Am. Chem. Soc. 2017, 139, 11421−11426

Article

Journal of the American Chemical Society

Figure 5. (a) Glucosylation of M9-IL-8(34−72) glycopeptide library 7 by recombinant human UGGT. (b) Glucosylation yield of each component of library 7 after 1 h estimated by LC-MS. (c) Time course of the glucosylation yield of library 7.

hydrophobic residues, especially those with aromatic residues in this case, were glucosylated faster (Figure S18b). This suggests the importance of the hydrophobic residue in the formation of a hydrophobic patch which plays a vital role in substrate recognition of UGGT. However, eventually all 11 glycopeptides in the library 7 were glucosylated after 24 h (Figure 5c). The same result was obtained for library 10 (Figure S18c). Hydrophilic glycopeptides were glucosylated after most of the hydrophobic substrates were consumed. These results may be due to the formation of a hydrophobic patch by hydrophobic residues at other positions or due to the possible multiple binding sites of UGGT. The fact that UGGT preferentially glucosylates hydrophobic substrates is important because hydrophobic glycopeptides have higher tendencies to form aggregates that can be toxic to the cells. The ability to glucosylate relatively hydrophilic substrates is also an important property of UGGT as a broad range of misfolded species formed during the biosynthesis of glycoprotein in the ER must be glucosylated. Recently, electron microscopy structures of UGGT were reported, and the authors proposed that hydrophobic surfaces at the central cavity of UGGT recognize hydrophobic residues exposed on the surface of misfolded glycoproteins.23 This large cavity may accommodate the hydrophobic interaction with a hydrophobic patch but not with a single hydrophobic residue.

Figure 4. LC-MS spectra of each component of glycopeptide library 7. Amino acid residue at the position 65 is (a) Gln: m/z calcd 6377.8, found 6337.9; (b) Pro: m/z calcd 6346.8, found 6347.0; (c) Gly: m/z calcd 6306.7, found 6306.1; (d) Lys: m/z calcd 6377.8, found 6377.1; (e) Ser: m/z calcd 6336.7, found 6337.1; (f) Glu: m/z calcd 6378.8, found 6379.1; (g) Tyr: m/z calcd 6412.8, found 6413.3; (h) Ala: m/z calcd 6320.7, found 6320.9; (i) Val: m/z calcd 6348.8, found 6348.0; (j) Phe: m/z calcd 6396.8, found 6396.9; (k) Leu: m/z calcd 6362.8, found 6362.1.

the amino acid residue at position 65 of the M9-IL-8 glycopeptide. It is also noteworthy that rough estimations of the initial velocities of the glucosylation of glycopeptides having hydrophobic amino acids are about 2-fold higher than those of other glycopeptides (Figure 5b). These differences resulted in about 4−5-fold difference of the half-life estimated from the time course curve shown in Figure 5c even under the competitive assay conditions. This efficient glucosylation activity of UGGT toward more hazardous hydrophobic substrates, shown by our experiments, may contribute to the fast calnexin/calreticulinrefolding cycle in the ER. We also prepared another glycopeptide library 10, varying hydrophilic residue Lys54 with hydrophobic amino acids using a pNCL strategy (Scheme S1 and Figure S17). The librarybased UGGT assay also revealed that the glycopeptides having

■

CONCLUSION In conclusion, we have shown that UGGT recognizes M9-IL8(34−72) glycopeptide 1a as a misfolded glycoprotein through hydrophobic interactions of the area containing Phe65. The specific interaction between UGGT and glycopeptide 1a was found by chemical shift perturbation mapping NMR experiments using chemically synthesized glycopeptides having a 15N label at the designated hydrophobic amino acid residues. This site-specific 15N labeling is an alternative approach to analyze protein−protein interaction or protein conformation. This method may be used to accelerate such studies, as it does not require assignments of all NMR peaks, a drawback admittedly 11425

DOI: 10.1021/jacs.7b03277 J. Am. Chem. Soc. 2017, 139, 11421−11426

Article

Journal of the American Chemical Society

(7) D’Alessio, C.; Caramelo, J. J.; Parodi, A. J. Semin. Cell Dev. Biol. 2010, 21, 491. (8) Trombetta, E. S.; Parodi, A. J. Methods 2005, 35, 328. (9) Caramelo, J. J.; Castro, O. A.; Alonso, L. G.; de Prat-Gay, G.; Parodi, A. J. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 86. (10) Totani, K.; Ihara, Y.; Tsujimoto, T.; Matsuo, I.; Ito, Y. Biochemistry 2009, 48, 2933. (11) Totani, K.; Ihara, Y.; Matsuo, I.; Koshino, H.; Ito, Y. Angew. Chem., Int. Ed. 2005, 44, 7950. (12) Zhu, T.; Satoh, T.; Kato, K. Sci. Rep. 2015, 4, 7322. (13) Izumi, M.; Makimura, Y.; Dedola, S.; Seko, A.; Kanamori, A.; Sakono, M.; Ito, Y.; Kajihara, Y. J. Am. Chem. Soc. 2012, 134, 7238. (14) Dedola, S.; Izumi, M.; Makimura, Y.; Seko, A.; Kanamori, A.; Sakono, M.; Ito, Y.; Kajihara, Y. Angew. Chem., Int. Ed. 2014, 53, 2883. (15) Izumi, M.; Komaki, S.; Okamoto, R.; Seko, A.; Takeda, Y.; Ito, Y.; Kajihara, Y. Org. Biomol. Chem. 2016, 14, 6088. (16) Izumi, M.; Murakami, M.; Okamoto, R.; Kajihara, Y. J. Pept. Sci. 2014, 20, 98. (17) Schnölzer, M.; Alewood, P.; Jones, A.; Alewood, D.; Kent, S. B. H. Int. J. Pept. Res. Ther. 2007, 13, 31. (18) Makimura, Y.; Kiuchi, T.; Izumi, M.; Dedola, S.; Ito, Y.; Kajihara, Y. Carbohydr. Res. 2012, 364, 41. (19) Johnson, E. C. B.; Kent, S. B. H. J. Am. Chem. Soc. 2006, 128, 6640. (20) Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. H. Science 1994, 266, 776. (21) Bang, D.; Kent, S. B. H. Angew. Chem., Int. Ed. 2004, 43, 2534. (22) Lam, K. S.; Salmon, S. E.; Hersh, E. M.; Hruby, V. J.; Kazmierski, W. M.; Knapp, R. J. Nature 1991, 354, 82. (23) Calles-Garcia, D.; Yang, M.; Soya, N.; Melero, R.; Menade, M.; Ito, Y.; Vargas, J.; Lukacs, G. L.; Kollman, J. M.; Kozlov, G.; Gehring, K. J. Biol. Chem. 2017, 292, 11499. (24) Mijalis, A. J.; Thomas, D. A.; Simon, M. D.; Adamo, A.; Beaumont, R.; Jensen, K. F.; Pentelute, B. L. Nat. Chem. Biol. 2017, 13, 464. (25) Simon, M.; Maki, Y.; Vinogradov, A. A.; Zhang, C.; Yu, H.; Lin, Y.-S.; Kajihara, Y.; Pentelute, B. L. J. Am. Chem. Soc. 2016, 138, 12099.

being, however, that complete interaction or conformation cannot be analyzed. We then constructed a glycopeptide library by parallel native chemical ligation of a glycopeptide-α-thioester to a peptide library consisting of 11 peptides. This synthetic strategy proved to be effective for the preparation of longer peptides or a protein library. Rapid preparation of proteins differing at a single residue is important for the detailed analysis of protein function. One solution has been the use of accelerated SPPS.24,25 Our pNCL approach is another option for the rapid preparation of a precisely designed protein library. UGGT assay against the glycopeptide library revealed that UGGT preferentially glucosylates hydrophobic glycopeptides; however, relatively hydrophilic glycopeptides can also be glucosylated at slower rates. We demonstrated that the precisely designed glycopeptide probes are valuable tools to address the substrate recognition mechanism of UGGT. The homogeneous glycopeptide probe that has a 15N label at a specific position and the glycopeptide library bearing homogeneous oligosaccharides constructed by parallel native chemical ligation are accessible only by chemical synthesis.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b03277. Experimental procedures, characterization, NMR spectra, and MS spectra of library-based UGGT assay (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] ORCID

Masayuki Izumi: 0000-0001-6486-9678 Ryo Okamoto: 0000-0001-9529-2525 Yukishige Ito: 0000-0001-6251-7249 Yasuhiro Kajihara: 0000-0002-6656-2394 Present Address ⊥

Department of Chemistry and Biotechnology, Faculty of Science and Technology, Kochi University, 2-5-1 Akebono-cho, Kochi 780-8520, Japan. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was partially supported by Financial support from the Japan Society for the Promotion of Science (KAKENHI Grant Number 16H06290 to Y.I.). We thank all referees for their useful and kind advice.

■

REFERENCES

(1) Parodi, A. J. Annu. Rev. Biochem. 2000, 69, 69. (2) Helenius, A.; Aebi, M. Science 2001, 291, 2364. (3) Hartl, F. U.; Hayer-Hartl, M. Nat. Struct. Mol. Biol. 2009, 16, 574. (4) Mäaẗ tänen, P.; Gehring, K.; Bergeron, J. J. M.; Thomas, D. Y. Semin. Cell Dev. Biol. 2010, 21, 500. (5) Lederkremer, G. Z. Curr. Opin. Struct. Biol. 2009, 19, 515. (6) Aebi, M.; Bernasconi, R.; Clerc, S.; Molinari, M. Trends Biochem. Sci. 2010, 35, 74. 11426

DOI: 10.1021/jacs.7b03277 J. Am. Chem. Soc. 2017, 139, 11421−11426