Subscriber access provided by UNIV OF NEWCASTLE
Article
Chemical synthesis demonstrates dynamic O-glycosylation regulates folding and functional conformation of a pivotal EGF12 domain of human NOTCH1 receptor Shun Hayakawa, Yasuhiro Yokoi, Hiroshi Hinou, and Shin-Ichiro Nishimura Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.7b00372 • Publication Date (Web): 26 Jul 2017 Downloaded from http://pubs.acs.org on July 27, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Biochemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Chemical synthesis demonstrates dynamic O-glycosylation regulates folding and functional conformation of a pivotal EGF12 domain of human NOTCH1 receptor Shun Hayakawa,† Yasuhiro Yokoi,† Hiroshi Hinou,†,‡ and Shin-Ichiro Nishimura*†,‡ †
Graduate School of Life Science and Faculty of Advanced Life Science, Hokkaido University, N21, W11, Kita-ku, Sapporo, 001-0021 Japan
‡
Medicinal Chemistry Pharmaceuticals, Co. Ltd., N9, W15, Chuo-ku, Sapporo, 060-0009 Japan
ABSTRACT: The interaction of human NOTCH1 receptor and its ligands is a crucial step to initiate the intracellular signal transductions, in which O-glycosylation of the extracellular EGF-like domain strongly affects multiple aspects of cell differentiation, development, and cancer biology. However, consequences of biosynthetic O-glycosylation processes in ER/Golgi on the folding of EGF domains remain unclear. Synthetic human NOTCH1 EGF12 modules allow for new insight into the crucial roles of Oglycosylation in the folding and conformation of this pivotal domain. Here, we show for the first time that predominant Oglucosylation at Ser458 facilitates proper folding of EGF12 domain in the presence of calcium ion, while non-glycosylated linear EGF12 peptide affords large amounts of misfolded products (>50%) during the in vitro oxidative folding. Strikingly, Ofucosylation at Thr466 prior to O-glucosylation at Ser458 totally impedes folding of EGF12 independent of calcium ion, whereas modification of the Fucα1→ moiety with β-linked GlcNAc enhances dramatically folding efficiency. In addition, we elicit that extension of the Glcβ1→ moiety with xyloses is a negative regulation mechanism in the folding of EGF12 when synthesis of a trisaccharide (Xylα1→3Xylα1→3Glcβ1→) dominates over the posttranslational modification at Thr466. Comprehensive NMR studies of correctly-folded EGF12 modules demonstrate that non-covalently bonded bridges between sugars and peptide moieties, namely sugar bridges, contribute independently to the stabilization of anti-parallel β-sheet in the ligand-binding region. The present results provide evidence that the dynamic O-glycosylation status of EGF12 domain elaborated in ER/Golgi strongly affects folding and trafficking of human NOTCH1 receptor.
INTRODUCTION The Notch signaling pathway is a highly conserved cell-cell communication system, which involves central roles in cellular development processes. In particular, Notch signaling strongly influences cancer biology including cancer stem cells, angiogenesis, cancer immunity, and embryonic differentiations.1 Notch signaling is mediated by five ligands [Jagged1 and Jagged2, and Delta-like ligand 1 (DLL-1), DLL-3 and DLL-4] on signalsending cells and Notch (NOTCH1~4) extracellular domain (ECD) on signal-receiving cells. The Notch signaling pathway is triggered by ligand-receptor binding which induces the subsequent proteolysis by ADAM10/17 metalloproteases and γsecretase to cleave Notch intercellular domain (ICD). Cleaved Notch ICD moves into nucleus and regulates the expression of Notch target genes.2,3 Posttranslational O-glycosylations in human NOTCH1 ECD that is composed of 36 epidermal growth factor (EGF) like repeats have been reported to regulate the Notch signaling pathway in terms of Notch receptor-ligand binding and intracellular signal activation.4 Each EGF domain has six cysteine residues forming three disulfide bonds (C1-C3, C2-C4, and C5-C6). Glycosylations such as O-glucosylation,5 O-fucosylation, and O-Nacetylglucosaminylation6 have been found in the consensus sequence between two cysteine residues, C1-X-S-X-(P/A)-C2,7 C2X-X-X-X-(T/S)-C3,8 and C5-X-X-G-X-(T/S)-G-X-X-C6,9 respectively (where X is any amino acid). In mammalian Notch receptors, α-O-fucose and β-O-glucose attached to Thr/Ser residues are further modified with specific sugars, forming unusual glycoforms, Neu5Acα2→3/6Galβ1→4GlcNAcβ1→3Fucα1→ and
Xylα1→3Xylα1→3Glcβ1→,10 respectively. O-Fucosylation of human NOTCH1 EGF12 is one of the most important modifications in Notch signaling activity because further glycosylations from the O-fucose residue at Thr466 involved in the ligand binding region11–13 influence strongly binding specificity and strength of the affinity of EGF12 domain.14–16 Especially, it seems likely that Fringe-mediated GlcNAc modification of the fucose residue may be an essential process for the regulation of the canonical NOTCH signaling pathway in cancer biology.15 Crystallography of the recombinant human and mouse EGF111316-19 indicated highly converged three-dimensional (3D) structure of the ligand-binding region of EGF12 domain including two unique O-glycosylations at Ser458 and Thr466. Our interest was focused on dynamic conformational impacts of the putative Oglycosylation states involved in the central EGF12 domain. To investigate the effects of various O-glycoforms on the structure and functions of EGF12 domain, we needed to develop an efficient and versatile method for the synthesis of multiply crosslinked EGF-like peptides bearing various glycoforms at potential glycosylation sites. We showed that a synthetic strategy for the cancer-related mucin domains20 is feasible for the construction of glycosylated mouse Notch1 EGF12 domain including three-pairs of intramolecular disulfide bonds.21 Recently, a synthetic human NOTCH1 EGF12 module having two unique O-glycans, Xylα1→3Xylα1→3Glcβ1→ at Ser458 and GlcNAcβ1→3Fucα1→ at Thr466, revealed an important role of calcium ion in the folding process of this domain.22 In addition, NMR-based 3D structure of this fully O-glycosylated EGF12 model allowed us to understand that posttranslational O-
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
glycosylations also contribute to the stabilization of β-hairpin involved in the functional antiparallel β-sheet structure of this pivotal domain16-19 through the interaction of sugars with neighboring amino acid residues, namely the intramolecular “sugar bridge” structure.22 Structural and functional roles of O-glucosylation at Ser458 in human NOTCH1 EGF12 domain remain unclear when compared with those of modification at Thr466 residue.23,24 It is likely that O-glucosylation-initiated modification at Ser458 residue may not directly participate in the NOTCH ligand binding.18,19 However, glucosyltransferase (POGLUT), which is responsible for the introduction of glucose to the Ser458 residue, may affect the processing of human NOTCH1 receptor and regulate signaling by ICDs.25,26 Overexpression of xylosyltransferases (GXYL1, GXYL2, and XXYLT1), which are responsible for addition of xylose residues to the proximal O-glucose residue, results in negative regulation of Notch signaling in Drosophila, suggesting that modification of O-glucose with xylose residue(s) might inhibit trafficking of Notch receptor.27 These observations may indicate that some possible glycoforms at Ser458, such as Glcβ1→ and Xylα1→3Xylα1→3Glcβ1→ generated by these enzymes, have different effects on the structure and functions of human NOTCH1 EGF12 domain from those of O-fucocosylationinitiated modification at Thr466 residue. Although we have accomplished the synthesis of human NOTCH1 EGF12 domain carrying two matured O-glycans, Xylα1→3Xylα1→3Glcβ1→ at Ser458 and Neu5Acα2→3Galβ1→4GlcNAcβ1→3Fucα1→ at Thr466 residues, concurrently (Figure 1A)22 based on the glycomics of Drosophila and mouse Notch1 EGF12 domain,27,28 we do not know yet the full portrait of O-glycosylations of the human NOTCH1 EGF12 domain. Importantly, it should be noted that posttranslational O-glycosylations of NOTCH1 receptor is tissue/cell type-specific and dynamic events that may afford microheterogeneity of total glycan structures in each extracellular EGF domains having consensus sequences to be glycosylated, namely “glycosylation status” at putative glycosylation sites. In case for the human NOTCH1 EGF12 domain, it is also clear that the variants carrying different glycoforms can be produced at Ser458 and/or Thr466 residue(s) during the dynamic biosynthetic processes in ER/Golgi. A systematic chemical synthesis of novel EGF12 analogs having the competent glycosylation states elicits for the first time the significance of posttranslational Oglycosylation in the folding process as well as conformational stabilization of this central domain of the human NOTCH1 receptor. EXPERIMENTAL PROCEDURES Synthesis of human NOTCH1 EGF12 modules. All new EGF12 modules were synthesized as linear (glyco)polypeptides using sugar amino acid derivatives 7~10 according to the general procedure for the synthesis of module 6 reported previously.22,29.30 The detail conditions and procedure of the solid-phase synthesis, and characterization data were shown in the Supporting Information (Figure S1). Disulfide bond formation of EGF12 modules. The linear EGF12 precursors 1~5 were subjected to the in vitro oxidative folding reaction in the redox buffer (pH8.0) containing 1 mM reduced glutathione and 0.2 mM oxidized glutatione at 25oC or 4oC in the presence or absence of calcium ion. The detail procedures for the characterization by using HPLC (Figure 2A, 3A, and 3B), purification of the folding products, and the condition of thermolysin digestion were described in the Supporting Information. NMR study and structural calculation of EGF12 modules. NMR experiments of EGF12 modules 1~5 were performed on a
Page 2 of 13
Bruker AVANCE 800 MHz spectrometer according to the methods reported previously.21,22 Three-dimensional structures of synthetic EGF12 modules 1~5 were calculated and constructed by the procedure established for the structural study of EGF12 module 622 and the detail experimental conditions and all procedures were described in the Supporting Information. Statistics of NMR structure and restraints for the structural calculation of EGF12 modules were summarized in Table 1 and 2. RESULTS Chemical synthesis of human NOTCH1 EGF12 modules having designated O-glycosylation states at Ser458 and Thr466. Figure 1B shows a general synthetic scheme of the human NOTCH1 EGF12 modules 1~6 used mainly in the present study. To investigate systematically the effects of the individual O-glycans on the folding and conformation of a pivotal EGF12 domain, it is considered that possible folding intermediates 2, 3, 4, and 5 having some key glycoforms that can be elaborated during biosynthesis should be newly synthesized in addition to the nonglycosylated 1 and the matured counterpart 6 bearing both Xylα1→3Xylα1→3Glcβ1→ at Ser458 and GlcNAcβ1→3Fucα1→ at Thr466 residues, concurrently.22 All EGF12 modules were synthesized efficiently by a general protocol based on the microwave-assisted solid phase glycopeptide synthesis29,30 using key building blocks, Fmoc-Thr(Ac3Fucα1→)OH (7),21 Fmoc-Thr[Ac3GlcNAc(β1→3)Fucα1→]-OH (8),21 Fmoc-Ser(Ac4Glcβ1→)-OH (9),22 and FmocSer[Ac3Xyl(α1→3)Ac2Xyl(α1→3)Ac3Glcβ1→]-OH (10)22 (Figure 1B). Linear EGF12 peptide/glycopeptides composed of 38 amino acid residues as precursors were released from resin and subjected directly to in vitro glutathione-mediated oxidative folding reaction for the construction of intramolecular three disulfide bonds.22 Misfolded structures generated from non-glycosylated EGF12 linear peptide. In the preceding communication,22 we succeeded in the synthesis and NMR analysis of properly folded human NOTCH1 EGF12 module 6 having matured O-glycan chains at two putative glycosylation sites, Ser458 and Thr466 residues. However, it is important to note that the results do not provide any information for the effects of dynamic Oglycosylation states on the folding process of this crucial domain in the human NOTCH1 receptor. Our previous study highlighted the importance of calcium ion in the folding and occurrence of the intramolecular sugar bridges involved in the ligand-binding region of the folded EGF12 model 6 having one of the matured glycoforms, Xylα1→3Xylα1→3Glcβ1→ at Ser458 and GlcNAcβ1→3Fucα1→ at Thr466, concurrently.22 Therefore, it is clear that careful assessment of in vitro folding process of nonglycosylated EGF12 peptide 1 is the first step in understanding how posttranslational O-glycosylation pathways contribute to the molecular mechanism in the construction of properly folded EGF12 domain.
ACS Paragon Plus Environment
Page 3 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Figure 1. Synthetic human NOTCH1 EGF12 modules. (A) Human NOTCH1 EGF12 domain having two matured O-glycans at Ser458 and Thr466 residues, concurrently.22 (B) A general synthetic protocol of human NOTCH1 EGF12 modules 1~6 designed and used in this study. Compounds 7~10 were sugar-modified amino acid derivatives used for the synthesis of modules 2~6.
As shown in Figure 2A, when in vitro folding of the naked EGF12 peptide (0.1 mg/mL) was conducted at 25oC in the absence of calcium ion in 50 mM Tris-HCl buffer (pH 8.0) containing 1.0 mM reduced-type glutathione and 0.2 mM oxidized-type glutathione for 24 h, the linear non-glycosylated EGF12 peptide afforded three major product peaks in HPLC eluted at 20, 22, and 23 min, respectively. The products isolated from fractions #1, #2, and #3 gave an expected mass value corresponding to the theoretical mass of the properly folded EGF12 (1) as shown in Figure 2B. However, further MALDI-TOFMS analysis of the enzymatic digests of these products revealed that the ideal ion peaks corresponding to the fragments containing C2-C4 (m/z 1270.768) and C5-C6 linkages (m/z 1353.896 and m/z 1375.863) are observed only in case of the fraction #3 while digests containing C1-C3 could not be detected (Figure 2C and 2D). It was also uncovered that the folding conducted at 25oC without assistance of calcium ion gives rise to properly folded non-glycosylated EGF12 (1) as a minor product and large amounts of misfolded products containing improperly crosslinked structures identified as ion peaks corresponding to C4-C6 (m/z 903.526) of the fraction #1 and C2-C4 (m/z 1270.768) of the fraction #2, respectively. When this folding reaction was performed in the presence of calcium ion, it was revealed that the ratio of the properly folded product (fraction #3) was enhanced significantly (~30%) while considerable amounts of misfolded products (over 50%) still formed (Figure 2A, green line).
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 2. Glutathione-mediated oxidative folding reactions of naked (nonglycosylated) linear EGF12. (A) HPLC profiles of the mixtures after folding reactions. Reactions were conducted by the conditions for 24 h as follows: 5 mM CaCl2 at 4ºC (magenta), 4ºC in the absence of calcium ion (blue), 5 mM CaCl2 at room temperature (green), and room temperature in the absence of calcium ion (black), respectively. (B) MALDI-TOFMS indicates that three major products in the fractions #1, #2, and #3 exhibit desired mass value corresponding to the correctly folded EGF12 (1). (C) MALDI-TOFMS showing positions of disulfide bonds in the products obtained from the fractions #1, #2, and #3. Proteolytic digests obtained by treating with thermolysin were directly subjected to MALDI-TOFMS analysis. (D) Plausible folded structures found in the fractions #1, #2, and #3, respectively. Regions colored in red indicate that positions of disulfide bonds are identified by crucial ion peaks detected in MALDI-TOFMS (Figure 2C). Arrows represent the possible cleavage sites by thermolysin from Geobacillus stearothermophilus.
A similar effect was observed in the folding at 4oC even in the absence of Ca2+ (Figure 2A, blue line), suggesting that reduced flexibility facilitates the formation of a thermodynamically stable structure using hydrogen bonds between anti-parallel β-sheets. Interestingly, calcium ion-associated in vitro folding at 4oC gave the highest yield (~50%) of the properly folded non-glycosylated EGF12 (1) (Figure 2A, magenta line). Glycosylation states strongly influence folding of human NOTCH1 EGF12 domain. Considering that both Ser458 and Thr466 reside within the two consensus sequences between two cysteine residues, C1-X-Ser458-X-(P/A)-C2 and C2-X-X-X-XThr466-C3, lying in close proximity to each other, mechanism of the formation of three disulfide bonds in the EGF12 domain may be influenced significantly by dynamic O-glycosylation biosynthetic pathways in ER/Golgi compartments. We hypothesized that the efficacy in the accurate folding of NOTCH1 EGF12 domain depends strongly on the interaction of each sugar moieties with the proximal peptide regions and/or short segments involving a cysteine residue forming the intramolecular disulfide bond(s) in an unfolded/disordered state. In addition, it seems likely that steric hindrance and flexibility of individual glycoforms linked to the Ser458 and Thr466 residues greatly affect the thermodynamic stability of the folding intermediates generated through the disulfide shuffling reactions during the folding process. To test this hypothesis, in vitro folding reactions of the unstructured EGF12 precursors 2~6 were assessed by monitoring HPLC profiles of the products generated under various environments (Figure 3). HPLC profiles of the products elaborated during in vitro folding from linear precursors 2~6 (Figure 3) were apparently different from those of non-glycosylated EGF12 (1) as shown in Figure 3. The results clearly indicate that posttranslational Oglycosylation states at Ser458 and Thr466 residues influence directly the folded state as well as folding efficacy of unstructured EGF12 domain even in the presence of calcium ion. It was clear that the folding potency of each linear EGF12 precursors strongly depends on both glycoforms and glycosylation sites.
ACS Paragon Plus Environment
Page 4 of 13
Page 5 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry 25ºC in the presence of calcium ion as shown in Figure 3A. These results imply that a trisaccharide structure (Xylα1→3Xylα1→3Glcβ1→) attached to the Ser458 residue of linear glycopeptides 5 and 6 perturbs folding of EGF12 domain under a non-natural low temperature condition. Properly folded structures of newly synthesized modules 1~5 were evidenced by NOE correlations between Hβ of two cysteine residues (Figure 3C and 3D, see also Figure S2), in which NOE signals indicating correctly formed disulfide bond pairs are detected in 2.4~3.4 ppm as those observed in case for the module 6.6.22
Figure 3. Glutathione-mediated oxidative folding reactions of linear EGF12 precursors 1~6 (0.1 mg/mL). HPLC profiles of the folding conducted at 25ºC (A) and 4ºC (B), respectively. The peaks indicated by asterisk show the fractions containing properly folded EGF modules, respectively. NOESY spectrum of the module 1 isolated from the fraction represented by asterisk (C), in which enlarged NOEs from 2.4~3.4 ppm show cross peaks between two β-protons of cross-linked cysteine residues (D). See also NOESY spectra of compounds 2~5 in Figure S2.
Dominant modification at Ser458 residue with a trisaccharide (Xylα α1→3Xylα α1→3Glcβ β1→) inhibits folding of human NOTCH1 EGF12 domain. Surprisingly, it was uncovered that O-glucosylation at Ser458 in the unstructured precursor 4, an initial modification step in ER/Golgi, gives a positive regulatory effect on the folding when compared with the result of folding of the linear EGF12 polypeptide 1 as shown in Figure 3. In contrast, O-fucosylation of Thr466 residue in the unstructured precursor 2, another first glycosylation step of EGF12 domain in the same intracellular compartment, appears to perturb folding process of this domain. Remarkably, subsequent modification of Glcβ1→ moiety at Ser458 residue with xyloses leading to a trisaccharide moiety of the module 5, Xylα1→3Xylα1→3Glcβ1→, was proved to be a negative regulation mechanism in the folding of EGF12 domain. Inversely, it was demonstrated that the extension with GlcNAc at O-fucosylated Thr466 facilitates greatly the folding of the precursor 3 in a similar manner to the folding of the precursor 6 having matured O-glycans both at Ser458 and Thr466 residues (Figure 4). Intriguingly, folding efficacy of precursor 6 at 25oC (Figure 3A) is reduced distinctly by conducting folding at 4oC (Figure 3B), while folding of the unstructured module 3 lacking modification at Ser458 residue proceeds more smoothly at 4oC when compared with the result performed at 25oC (Figure 3A).
O-Fucosylation at Thr466 impedes in vitro folding while Oglucosylation at Ser458 promotes proper folding of NOTCH1 EGF12 domain. As shown by the arrows in Figure 3A, the unfolded precursor 2 having Fucα1→ moiety at Thr466 residue as well as 5 bearing Xylα1→3Xylα1→3Glcβ1→ moiety at Ser458 residue were proved to afford considerable amounts of misfolded products during in vitro folding reaction at 25ºC even in the presence of calcium ion. Surprisingly, it was also uncovered that Ofucosylation of EGF12 disturbs totally the formation of the threepairs of disulfide bonds when folding is conducted at 25ºC in the absence of calcium ion, whereas all other compounds even precursor 5 provide products involving properly folded EGF12 domain. On the other hand, folding of precursors 3, 4, and 6 proceeds smoothly at 25ºC in the presence of calcium ion and gives properly folded EGF12 domain as major product in high efficiency, respectively (Figure 3A). It is of particular interest that precursors 3, 4, and 6 do not provide any misfolded structures and/or partly folded intermediates during in vitro folding reaction under the above general condition (25ºC and 5 mM CaCl2). From a view point of chemical synthesis of human NOTCH1 EGF domains as models for deciphering structural and functional roles, it should be noted that synergistic effects of calcium ion and low-temperature environment on the in vitro folding found in nonglycosylated EGF12 (1) (Figure 2) facilitate an efficient and large-scale synthesis of properly folded NOTCH1 EGF12 modules, notably compounds 2 (7.8 mg, 7.4%), 3 (9.3 mg, 8.5%), and 4 (6.5 mg, 6.4%), respectively (Figure 3B). Here, “%” represents isolated yields of the correctly folded products calculated from the solid-phase synthesis. In contrast, the folding efficacy of modules 5 and 6 is drastically reduced at 4ºC (Figure 3B) and the optimal synthetic yields including the folding step for compound 5 (2.8 mg, 2.4%) and 6 (4.9 mg, 4.1%)22 were found to be achieved at
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 4. Altered effect of O-fucosylation of Thr466 on the folding of EGF12 domain when Ser458 is occupied by Xylα1→3Xylα1→3Glcβ1→ moiety. (A) Synthetic scheme of a new EGF12 module 11 having Xylα1→3Xylα1→3Glcβ1→ moiety at Ser458 and Fucα1→ moiety at Thr466 residue and comparison of the folding profiles of synthetic unstructured glycopeptides 2, 5, 6 and 11. (B) MALDI-TOFMS of thermolytic digests of the isolated 11 showing the positions of disulfide bonds in the main product.
These observations provide evidence that (i) the trisaccharide moiety (Xylα1→3Xylα1→3Glcβ1→) modified at Ser458 residue inhibits formation of a thermodynamically stable disulfide bond between Cys456 and Cys467 (C1-C3) due to the steric effect while fucosylation of Thr466 residue locating between Cys461 and Cys467, another negative regulatory mechanism, appears to enhance the flexibility of this hydrophilic peptide region, and (ii) attachment of GlcNAcβ1→ moiety to the proximal O-fucose of Thr466 residue is a critical step to accelerate proper folding of EGF12 domain. Altogether, we postulated that dynamic glycosylation states at two potential O-glycosylation sites in the ER/Golgi biosynthetic pathway may be an essential molecular mechanism to control normal and impaired folding of EGF12 domain. To test this hypothesis and elicit underlying molecular mechanism in the O-glycosylation-mediated protein folding, we then synthesized and investigated folding profile of a novel EGF12 precursor 11 bearing Xylα1→3Xylα1→3Glcβ1→ moiety at Ser458 and Fucα1→ moiety at Thr466 residue (Figure 4A), because this important glycosylation status in the human NOTCH1 EGF12 domain is associated significantly with a regulation mechanism in the binding with putative NOTCH ligands.14-16 Glycosylation states define the dynamic folding pathway of human NOTCH1 EGF12 domain. Accumulating evidence has indicated that alteration from the glycosylation status at Thr466 of module 11 (Fucα1→) into that of 6 (GlcNAcβ1→3Fucα1→) seems to be a crucial step in the NOTCH signaling pathway to control interaction with Jagged1 and DLL1, whereas glycosylation of Thr466 may not affect the affinity of this domain with DLL4.17 However, it is noteworthy that the difference in the effects of glycosylation states of two important models 11 and 6 on the folding of EGF12 domain remains to be elucidated. Given the fact that both Fucα1→ moiety at Thr466 of the precursor 2 and Xylα1→3Xylα1→3Glcβ1→ moiety at Ser458 of 5 render proper folding of the EGF12 domain difficult (Figure 3), the glycosylation status integrating these two negative-regulation glycoforms appears to inhibit or abolish normal folding of the precursor 11. Surprisingly, folding of the synthetic unstructured EGF12 polypeptide 11 proceeds in a quite similar manner to that of 6 at 25ºC in the presence of calcium ion, whereas folding under other environments shows a marked propensity to accompany with a complicated mixture of misfolded and partially folded intermediates (Figure 4A). MALDI-TOFMS analysis gave evidence of the correctly folded state of the isolated major product as shown in Figure 4B, indicating that O-fucose of Thr466 residue may function as a positive regulation mechanism in the folding of this domain
Page 6 of 13
when the Ser458 residue is concurrently modified with a bulky trisaccharide (Xylα1→3Xylα1→3Glcβ1→), another negativeregulation glycan motif. Judging from the entirely different folding profile of module 11 from those of 2 and 5, it is considered that the effect of the negative-regulation glycan (Xylα1→3Xylα1→3Glcβ1→) of Ser458 residue on the folding of EGF12 domain alters significantly in accordance with the glycosylation status at Thr466 residue (Figure 4B). More importantly, attachment of GlcNAcβ1→ moiety to the proximal fucose of the module 11 restricts markedly generation of the misfolded products as shown in the folding of module 6, demonstrating that this glycosylation step universally facilitates proper folding of the EGF12 domain independent of the glycosylation status at Ser458 residue. It seems likely that the glycan biosynthetic pathways conduct largely folding of unstructured EGF12 polypeptides under a novel molecular mechanism. We now appreciate that flexibility and steric effects of two different O-glycans locating at Ser458 and Thr466 residues between Cys456, Cys461, and Cys467 residues contribute concertedly to the thermodynamic stability of the folding intermediates in the disulfide shuffling reactions resulting in a dynamic folding energy landscape composed of conformational ensembles in unstructured (unfolded) states, misfolded states, and properly folded state. O-Fucosylation of Thr466 facilitates formation of anti-parallel β-sheet in the ligand-binding region of correctly folded EGF12 domain. Our attention was next directed to the effect of individual O-glycoforms linked to the Ser458 or Thr466 residue in the properly folded modules 2~5, respectively on the conformational stabilization of the ligand-binding loop region in the EGF12 domain. As indicated in Figure 3, our results uncovered that Oglucosylation of Ser458 and attachment of GlcNAc to the fucose at Thr466 facilitate formation of normally folded state of EGF12 domain while the initial O-fucosylation of Thr466 residue makes proper folding of EGF12 domain difficult. However, effect of each glycoforms in the folded modules 2~5 on the local conformational ensembles such as anti-parallel β-sheet and β-hairpin involved in the ligand-binding region remains unclear. We considered that the glycosylation status of the EGF12 domain also entails alternative molecular mechanism in the modulation of the functional ligand-binding region of this central domain. In other words, functional roles of O-glycosylation states both in the protein folding and local conformational stabilization, and their interplays during dynamic
ACS Paragon Plus Environment
Page 7 of 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry tra of folded EGF12 modules 1, 2, and 4 in the absence (left panel) and presence (right panel) of calcium ions (25 mM CaCl2).
folding process have not yet been fully delineated. To decipher the conformational impact of the glycosylation states of EGF12 domain in the local conformational stabilization, purified synthetic modules 2~5 were subjected to the solution NMR experiments such as two-dimensional NMR spectra, TOCSY, NOESY, DQFCOSY, and 13C-edited HSQC spectra collected in the solution of 90% H2O-10% D2O or in 99.9% D2O. In 2D NMR, the sequential connectivity of peptide backbone between Hα (i) and HN (i+1) were detected in all amino acid residues except for proline in NOESY spectra and all side chain protons as well as sugar protons were assigned using TOCSY and NOESY connection (Table S1). As shown in the proton NMR spectra measured in the absence of calcium ion (Figure 5A), modules 2 and 3 exhibit characteristic signals due to Hα and HN protons of β-sheet contributing residues (Cys467, Leu468, Phe474, Gln475, Cys476, and Ile477) at 5.0 to 5.5 ppm and at 8.8 to 9.5 ppm, respectively. These lower field shifts of both Hα and HN protons are due to electronwithdrawing effects by hydrogen bonds formed between antiparallel β-sheet. Notably, these signals were observed only in cases of the modules 2 (Fucα1→) and 3 (GlcNAcβ1→3Fucα1→) containing O-fucose moiety at Thr466 but not in non-glycosylated 1, modules 4 (Glcβ1→), and 5 (Xylα1→3Xylα1→3Glcβ1→) having O-glucose moiety at Ser458 residue. Proton NMR spectra measured in the presence of high concentration of calcium ions (25 mM CaCl2) afforded new signals due to hydrogen bonds between anti-parallel β-sheet in modules 1, 4, and 5 while those of 2 and 3 did not show any alteration in the spectra (Figure 5B). TOCSY spectra also demonstrate evidence that module 2 having O-fucose specifically gives well-separated cross peaks without the assistance of calcium ion when compared with those of modules 1 and 4, whereas calcium ions may improve significantly the separation of the spectra (Figure 5C). These observations clearly indicate occurrence of the local conformational equilibrium between the normal anti-parallel β-sheet of O-fucosylated EGF12 modules 2 or 3 and other unnatural conformations of nonglycosylated 1 or O-glucosylated modules 4 and 5 independent of the intracellular and extracellular concentration of calcium ions. It is particularly important to note that O-fucosylation is required for the stabilization of the anti-parallel β-sheet structure of the functional EGF12 domain after accurate disulfide bond formation, and O-glucose modification has little effect on the stabilization of this region.
Figure 5. Functional role of O-fucosylation of Thr466 residue in the conformational stabilization of anti-parallel β-sheet in the ligand-binding region of folded EGF12 domain. (A) Proton NMR spectra of amide proton region (4.7~11.0 ppm) of folded EGF12 modules 1~5 in the absence and (B) presence of calcium ions (25 mM CaCl2). Arrows indicate HN and Hα protons contributing to β-sheet. NMR spectra were measured in 90%H2O/10%D2O to observe backbone amide protons. (C) TOCSY spec-
Proximal fucosylation at Thr466 and distal non-reducing xylosylation at Ser458 contribute independently to the conformational stabilization of EGF12 domain. Further NOESY experiments revealed that sugar residues of the synthetic EGF12 modules 2~5 have characteristic NOE interactions with peptide moiety (Figure 3C and Figure S2). Notably, O-fucose residue has NOE interactions with proximal Thr466 and Ala465, and with Ile477 and Met479 locating in the opposite β-strand of Thr466. These amino acid residues, notably Ala465, Ile477 and Met479, appear to interact with methyl protons at C6 position of O-fucose residue through hydrophobic interaction. Similarly, NOE interactions between O-fucose and these amino acids were also observed in case of module 3 while GlcNAc residue did not show any NOE connection with proximal peptide region. On the other hand, NOE interactions of O-glucose residue of the module 4 were found only with β-protons of the proximal Ser458 residue. However, it was shown that the module 5 exhibits specific NOE connections between anomeric proton of non-reducing distal xylose residue of Xylα1→3Xylα1→3Glcβ1→ moiety and α-proton of the Glu473
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
residue of the β-sheet region. It is interesting to note that module 6 having two mature O-glycans, Xylα1→3Xylα1→3Glcβ1→ and GlcNAcβ1→3Fucα1→, shows quite similar NOE contacts to those observed in the modules 2~5.22
Page 8 of 13
Hα proton at Ser458 residue (Figure 6B). It is interesting to note that module 5 having xyloses appears to influence the lower-field shift over 0.2 ppm (Figure 6B) while NOE experiment exhibited connection between non-reducing xylose and Glu473 residue involved in the β-sheet. Differences in the chemical shift of Hα at Ser458, Thr466, Phe474, and Met479 residues between modules 2~5 and 1 were also observed in the module 622 (Figure 6C), suggesting that the roles of the individual O-glycans in the conformational stabilization of anti-parallel β-sheet are independent of those of the NOTCH1 protein folding processes. Table 1. Statistics of NMR structure of synthetic NOTCH1 EGF12 modules 1
2
3
4
5
Average potential energies (kcal mol-1)[a] Etotal
23.45
±
2.06
47.08
±
0.85
73.96
±
3.24
59.73
±
2.06
66.86
±
4.47
Ebonds
1.35
±
0.15
2.43
±
0.08
3.84
±
0.29
3.35
±
0.26
3.61
±
0.40
Eangle
8.59
±
0.63
20.68
±
0.33
27.53
±
1.10
16.97
±
1.13
18.99
±
1.65
Eimpr
0.72
±
0.14
0.81
±
0.04
2.37
±
0.29
3.42
±
0.40
3.71
±
0.31
EVDW[b]
3.98
±
0.99
4.00
±
0.59
11.43
±
1.98
12.33
±
2.24
13.18
±
2.10
ENOE[b]
8.71
±
0.92
17.63
±
0.61
26.40
±
1.57
22.95
±
2.44
25.11
±
3.39
Ecdih [b]
0.10
±
0.09
0.51
±
0.03
0.66
±
0.07
0.20
±
0.09
0.20
±
0.12
Deviations from idealized geometry Bond lengths (Å)
0.0016
±
0.0001
0.0022
±
0.0001
0.0026
±
0.0001
0.0024
±
0.0001
0.0024
±
0.0001
Bond angles (°)
0.24
±
0.01
0.83
±
0.01
0.84
±
0.01
0.39
±
0.03
0.47
±
0.02
Improper (°)
0.13
±
0.01
0.16
±
0.00
0.25
±
0.01
0.29
±
0.02
0.31
±
0.01
Average pairwise r.m.s. deviation (Å) Cys5-Asp18, Glu22-Pro29 Backbone atoms
1.03
±
0.40
0.98
±
0.23
1.10
±
0.31
0.91
±
0.30
0.94
±
0.29
heavy atoms
1.75
±
0.49
1.63
±
0.37
1.80
±
0.41
1.58
±
0.45
1.62
±
0.43
[a]
All energies and root mean square values were calculated by using the programs CNS1.1 and MOLMOL, respectively. Eimpr, EVDW, ENOE and Ecdih are the energy of improper torsion angles, the van der Waals repulsion energy, the square-well NOE potential energy, and the dihedral potential energy, respectively. [c] The force constants for the calculations of Eimpr, EVDW, ENOE and Ecdih were 4.0 kcal mol-1Å-4, 50 kcal mol-1Å-1 and 200 kcal mol-1rad-2, respectively. [b]
Table 2. Restraints for structural calculation of synthetic NOTCH1 EGF12 summary of restraints
1
2
3
4
5
447
424
451
456
490
289 0
281 11
277 26
301 13
296 38
96 0
75 0
87 2
85 0
91 4
24 0
20 0
24 0
26 0
22 0
38 0
32 0
30 0
28 0
34 0
within the same glycosylated residue
0
2
2
3
3
glycans on other peptide residues
0
3
3
0
2
22 22 0
30 21 9
42 22 20
26 20 6
38 20 18
distance restraints total intra-residue peptide glycan sequential (|i-j|=1) peptide glycan medium-range (2
