Chemical synthesis, folding, and structural insights into O-fucosylated epidermal growth factor-like repeat 12 of mouse notch-1 receptor

Article abstract of DOI:10.1021/ja105216u

Notch receptors are cell surface glycoproteins that play key roles in a number of developmental cascades in metazoa. The extracellular domains of Notch-1 receptors are composed of 36 tandem epidermal growth factor (EGF)-like repeats, many of which are modified at highly conserved consensus sites by an unusual form of O-glycan, with O-fucose. The O-fucose residues on certain EGF repeats may be elongated. In mammalian cells this can be a tetrasaccharide, Sia±2,3Galβ1,4GlcNAcβ1,3Fuc±1→. This elongation process is initiated by the action of O-fucose-specific β1,3 N-acetylglucosaminyltransferases of the Fringe family. There is evidence that the addition of GlcNAc by Fringe serves as an essential modulator of the interaction of Notch with its ligands and the triggering of activation. Here we describe the efficient synthesis, folding, and structural characterization of EGF repeat 12 (EGF 12) of a mouse Notch-1 receptor bearing different O-fucose glycan chains. We demonstrate that the three disulfide bonds, Cys ⁴⁵⁶-Cys⁴⁶⁷ (C1-C3), Cys⁴⁶¹-Cys⁴⁷⁶ (C2-C4), and Cys⁴⁷⁸-Cys⁴⁸⁷ (C5-C6) were correctly formed in the nonglycosylated as well as the O-fucosylated forms of EGF 12. Three-dimensional structural studies by NMR reveal that the methyl group of fucose is in close contact with ILe⁴⁷⁵, Met⁴⁷⁷, Pro ⁴⁷⁸ residues and this stabilizes the conformation of the antiparallel β-sheet of EGF 12. The addition of the GlcNAc residue on O-fucosylated EGF 12 induces a significant conformational change in the adjacent tripeptide sequence, Gln⁴⁶²Asn⁴⁶³Asp⁴⁶⁴, which is a motif involved in the natural, enzymatic O-fucosylation at the conserved site (Cys⁴⁶¹X₄Ser/ThrCys⁴⁶⁷).

Full text of DOI:10.1021/ja105216u

Published on Web 09/30/2010
Chemical Synthesis, Folding, and Structural Insights into
O-Fucosylated Epidermal Growth Factor-like Repeat 12 of
Mouse Notch-1 Receptor
Kazumi Hiruma-Shimizu,†,‡,| Kensaku Hosoguchi,†,| Yan Liu,§ Naoki Fujitani,†
Takashi Ohta,‡ Hiroshi Hinou,† Takahiko Matsushita,†,‡ Hiroki Shimizu,‡ Ten Feizi,§
and Shin-Ichiro Nishimura*,†,‡
Graduate School of Life Science, Frontier Research Center for Post-Genomic Science and
Technology, Hokkaido UniVersity, N21, W11, Sapporo 001-0021, Japan, National Institute of
AdVanced Industrial Science and Technology (AIST), Sapporo 062-8517, Japan, and
Glycosciences Laboratory, Imperial College London, Northwick Park Campus, Harrow,
Middlesex, HA1 3UJ, U.K.
Received June 15, 2010; E-mail: shin@glyco.sci.hokudai.ac.jp
Abstract: Notch receptors are cell surface glycoproteins that play key roles in a number of developmental
cascades in metazoa. The extracellular domains of Notch-1 receptors are composed of 36 tandem epidermal
growth factor (EGF)-like repeats, many of which are modified at highly conserved consensus sites by an
unusual form of O-glycan, with O-fucose. The O-fucose residues on certain EGF repeats may be elongated.
In mammalian cells this can be a tetrasaccharide, SiaR2,3Galꢀ1,4GlcNAcꢀ1,3FucR1f. This elongation
process is initiated by the action of O-fucose-specific ꢀ1,3 N-acetylglucosaminyltransferases of the Fringe
family. There is evidence that the addition of GlcNAc by Fringe serves as an essential modulator of the
interaction of Notch with its ligands and the triggering of activation. Here we describe the efficient synthesis,
folding, and structural characterization of EGF repeat 12 (EGF 12) of a mouse Notch-1 receptor bearing
different O-fucose glycan chains. We demonstrate that the three disulfide bonds, Cys456-Cys467 (C1-C3),
Cys461-Cys476 (C2-C4), and Cys478-Cys487 (C5-C6) were correctly formed in the nonglycosylated as well
as the O-fucosylated forms of EGF 12. Three-dimensional structural studies by NMR reveal that the methyl
group of fucose is in close contact with ILe475, Met477, Pro478 residues and this stabilizes the conformation
of the antiparallel ꢀ-sheet of EGF 12. The addition of the GlcNAc residue on O-fucosylated EGF 12 induces
a significant conformational change in the adjacent tripeptide sequence, Gln462Asn463Asp464, which is a
motif involved in the natural, enzymatic O-fucosylation at the conserved site (Cys461X4Ser/ThrCys467).
occurs at the putative consensus sequences C2X4-5Ser/ThrC36
(where Ser/Thr is the modified amino acid, X can be any amino
Introduction
The Notch signaling pathway is a highly conserved cell
signaling system present in metazoans and involved in a wide
variety of developmental processes.1 Glycosylation of the Notch
receptor is essential for the regulation of the Notch signaling
pathway.2,3 The extracellular domain of a mouse Notch-1
receptor contains 36 tandem epidermal growth factor (EGF)-
like repeats, many of which are glycosylated with unusual forms
of O-glycans, namely O-fucose (Fuc) and O-glucose glycans,4
as well as the more recently described O-N-acetylglucosamine
(GlcNAc)5 glycans. The modification by O-fucose glycans
acid, and C2, C3 are the second and third conserved cysteines
in the EGF repeat) and is believed to be an important modulator
of Notch signaling.7,8 One of the most evolutionarily conserved
O-fucose sites is in EGF 12, which is considered to be a portion
of the Delta-Serrate-Lag2 (DSL) ligand-binding site.9,10 Al-
though the role of O-fucose glycans on EGF 12 appears to be
crucial for the direct interaction with DSL ligands, the structural
and functional roles of the glycosylation in EGF 12 have not
been clarified in detail.8,10,11 This is in part due to the disulfide-
rich nature of the EGF repeats which precludes large-scale
expression of native material for biophysical and biochemical
studies. The O-fucose moieties on EGF domains of Notch are
† Hokkaido University.
‡ National Institute of Advanced Industrial Science and Technology.
§ Imperial College London.
| These authors contributed equally.
(6) Shao, L.; Moloney, D. J.; Haltiwanger, R. S. J. Biol. Chem. 2003,
278, 7775–7782.
(1) Bray, S. J. Nat. ReV. Mol. Cell Biol. 2006, 7, 678–689.
(2) Luther, K. B.; Haltiwanger, R. S. Int. Biochem. Cell Biol. 2009, 41,
1011–1024.
(7) Okajima, T.; Irvine, K. Cell 2002, 111, 893–904.
(8) Rampal, R.; Arboleda-Velasquez, J. F.; Nita-Lazer, A.; Kosik, K. S.;
Haltiwanger, R. S. J. Biol. Chem. 2005, 280, 32133–32140.
(9) Rebay, I.; Fleming, R. J.; Fehon, R. G.; Cherbas, L.; Cherbas, P.;
Artavanis-Tsakonas, S. Cell 1991, 67, 687–699.
(3) Stanley, P. Curr. Opin. Struct. Biol. 2007, 17, 530–535.
(4) Moloney, D. J.; Shair, L. H.; Lu, F. M.; Xia, J.; Locke, R.; Matta,
K. L.; Haltiwanger, R. S. J. Biol. Chem. 2000, 275, 9604–9611.
(5) Matsuura, A.; Ito, M.; Sakaidani, Y.; Kondo, T.; Murakami, K.;
Furukawa, K.; Nadano, D.; Matsuda, T.; Okajima, T. J. Biol. Chem.
2008, 283, 35486–35495.
(10) Xu, A.; Lei, L.; Irvine, K. D. J. Biol. Chem. 2005, 280, 30158–30165.
(11) Lei, L.; Xu, A.; Panin, V. M.; Irvine, K. D. DeVelopment 2003, 130,
6411–6421.
9
10.1021/ja105216u  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 14857–14865 14857

A R T I C L E S
Hiruma-Shimizu et al.
Figure 1. Glycosylated EGF 12 of mouse Notch-1 (amino acids 452-489) and key synthetic materials. O-Fucose site Thr466 is in cyan, and the six cysteine
residues are in yellow.
often further elongated by fucose-specific ꢀ-1,3-N-acetylglucos-
aminyltransferases12,13 of the Fringe family. It is well docu-
mented that by adding GlcNAc to O-fucose moieties Fringe
regulates ligand-mediated Notch signaling.15 In mammals the
action of Fringe is the cue to the further elongation and
formation of a mature tetrasaccharide structure, SiaR2,
3Galꢀ1,4GlcNAcꢀ1,3FucR1f.4,12,15
Scheme 1. It is based on the microwave-assisted solid-phase
protocol established for the synthesis of tumor-associated mucin
glycopeptides.16-21 In previous studies we have demonstrated
that Fmoc-Thr/Ser derivatives having O-acetylated sugars such
as Fmoc-Thr/Ser(Ac3GalNAcR1f)-OH (Tn antigen), Fmoc-
Thr/Ser(Ac3Galꢀ1,3Ac2GalNAcf)-OH (core 1), Fmoc-Thr/
Ser(Ac3GlcNAcꢀ1,6Ac2GalNAcf)-OH (core 6), and Fmoc-
Thr/Ser[Ac3Galꢀ1,3(Ac3GlcNAc)ꢀ1,6GalNAcf]-OH (core 2)
are versatile building blocks for the construction of a highly
complex MUC1-related compound library.20,21 Indeed, this
strategy using a fully O-acetylated Fmoc-Thr(Ac3Fucf)-OH
(6) permits rapid and efficient synthesis of O-fucosylated EGF
12 (2) which is composed of 38 amino acid residues (Swiss-
Prot Q01705). Deprotection of the fucose residue proceeded
smoothly under general conditions [1 N NaOH (pH 12.5)/
MeOH, DTT, 25 °C, 2 h] and gave the desired linear glyco-
peptide of compound 2. For the synthesis of compound 3, we
As part of a program to investigate the roles of O-fucose
glycans in Notch-ligand interaction, we have prepared the
fucosylated amino acid building blocks FucR1-Thr(Fmoc) and
GlcNAcꢀ1,3FucR1-Thr(Fmoc). We considered that chemical
and enzymatic synthesis of EGF 12 peptides (1, 2, 3, 4, and 5)
listed in Figure 1 would be of interest in studies of the
significance of the glycosylation in Notch signaling as well as
the folding mechanism of EGF repeats bearing an O-fucosylated
motif. The present study has been mostly focused on the
chemical synthesis, in Vitro folding, and structural characteriza-
tion of EGF 12 having the disaccharide GlcNAcꢀ1,3FucR1f
moiety (3), this being the glycopeptide structure that is formed
in nature, which appears to be critical in the regulation of
Notch-ligand interaction. We show that the addition of GlcNAc
by Fringe to FucR1fEGF 12 induces a conformational change
in EGF 12, and we hypothesize that this tunes the binding
strength with Delta-like and Jagged Notch ligands. We suggest
that the key intermediate 3 is a useful general synthetic tool for
the construction of variously glycosylated Notch analogs by
further enzymatic sugar elongation with Gal and Sia residues.
used partially protected building blocks
7
[Fmoc-
Thr(Ac3GlcNAclꢀ1,3Fucf)-OH], as O-acetyl protection of the
axial hydroxyl group at the C-4 position of the fucose residue
was found to be difficult to remove under the above conditions
when the adjacent equatorial hydroxyl group at the C-3 position
is substituted by the bulky GlcNAc residue (see Supporting
(16) Matsushita, T.; Hinou, H.; Kurogochi, M.; Shimizu, H.; Nishimura,
S.-I. Org. Lett. 2005, 7, 877–880.
(17) Matsushita, T.; Hinou, H.; Fumoto, M.; Kurogochi, M.; Fujitani, N.;
Shimizu, H.; Nishimura, S.-I. J. Org. Chem. 2006, 71, 3051–3063.
(18) Fumoto, M.; Hinou, H.; Ohta, T.; Ito, T.; Yamada, K.; Takimoto, A.;
Kondo, H.; Shimizu, H.; Inazu, T.; Nakahara, Y.; Nishimura, S.-I.
J. Am. Chem. Soc. 2005, 127, 11804–11818.
Results and Discussion
Chemical Syntheses and Folding of EGF 12 (1-3). The
synthetic approach for glycosylated EGF 12 (3) is shown in
(19) Naruchi, K.; Hamamoto, T.; Kurogochi, M.; Hinou, H.; Shimizu, H.;
Matsushita, T.; Fujitani, N.; Kondo, H.; Nishimura, S.-I. J. Org. Chem.
2006, 71, 9609–9621.
(12) Moloney, D. J.; Panin, V. M.; Johnston, S. H.; Chen, J.; Shao, L.;
Wilson, R.; Wang, Y.; Stanley, P.; Irvine, K. D.; Haltiwanger, R. S.;
Vogt, T. F. Nature 2000, 406, 369–375.
(20) Ohyabu, N.; Hinou, H.; Matsushita, T.; Izumi, R.; Shimizu, H.;
Kawamoto, K.; Numata, Y.; Togame, H.; Takemoto, H.; Kondo, H.;
Nishimura, S.-I. J. Am. Chem. Soc. 2009, 131, 17102–17109.
(21) Matsushita, T.; Sadamoto, R.; Ohyabu, N.; Nakata, H.; Fumoto, H.;
Fujitani, N.; Takegawa, Y.; Sakamoto, T.; Kurogochi, M.; Hinou, H.;
Shimizu, H.; Ito, T.; Naruchi, K.; Togame, H.; Takemoto, H.; Kondo,
H.; Nishimura, S.-I. Biochemistry 2009, 48, 11117–11133.
(13) Jinek, M.; Chen, Y.-W.; Clausen, H.; Cohen, S. M.; Conti, E. Nat.
Struct. Mol. Biol. 2006, 13, 945–946.
(14) Haines, N.; Irvine, K. D. Nat. ReV. Mol. Cell Biol. 2003, 4, 786–797.
(15) Chen, J.; Moloney, D. J.; Stanley, P. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 13716–13721.
9
14858 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

O-Fucosylated EGF Repeat 12 of Mouse Notch-1 Receptor
A R T I C L E S
Scheme 1. Synthetic Route of Sugar-Modified EGF 12 of Mouse Notch-1 (Amino Acids 452-489)
Information, Figure S1 and Scheme S2). As anticipated, use of
the building block 7 allowed for the construction of full-length
EGF 12 with the disaccharide GlcNAcꢀ1,3Fucfmoiety (3) by
microwave-assisted solid-phase synthesis on PEGA resin. After
de-O-acetylation, the crude product was directly subjected to
oxidative folding in glutathione-based redox buffer at pH 8.22
The folding process was monitored with reversed-phase high
performance liquid chromatography (RPHPLC) analysis of
aliquots of the sample solution, and distinct profiles were
obtained before and after folding (Figure S2). The latter gave a
sharp peak at 24 min as the major component (Figure S2b), of
which the precursor ion is at m/z 4560.763 (Figure S2c). This
is consistent with the predicted molecular mass of the folded
glycopeptide 3 which has three disulfide bonds. Moreover, no
other peaks corresponding to the same molecular mass were
observed in this HPLC profile, and this indicates that it is
unlikely any glycopeptide isomer with a different disulfide
bonding pattern had been produced in the folding process. The
nonglycosylated (‘naked’) EGF 12 peptide 1 and its O-fucosyl
analog 2 were synthesized and subjected to the folding process
under similar conditions. The folding products of 1 and 2 were
analyzed by RP-HPLC and MS, and the results (data not shown)
are similar to those observed with peptide 3. Further studies
were performed with the peptide fragments generated and
isolated from HPLC purified fractions of folded naked EGF 12
peptide 1 and glycopeptides 2 and 3 after thermolysin digestion.
These peptide fragments were characterized by MALDI-TOFMS
and amino acid sequence analyses. The results of peptide 1
showed the three fragments, C1-C3 (m/z 870.2), C2-C4 (m/z
1284.9), and C5-C6 (m/z 1379.2), which demonstrated the
correct folding of the synthetic EGF 12 peptide 1 (Table S1).
In contrast, although the fragment of C5-C6 (m/z 1379.2) was
observed in glycopeptides 2 and 3, we did not detect any small
fragments containing C1-C3, C2-C4, or those corresponding
to peptides with mismatched disulfide bonds. This suggests that
glycosylation at the Thr466 residue may possibly afford steric
protection to susceptible sites against the enzymatic hydrolysis.
(22) Zamborelli, T. J.; Dodson, W. S.; Harding, B. J.; Zhang, J.; Bennett,
B. D.; Lenz, D. M.; Young, Y.; Haniu, M.; Liu, C.-F.; Jones, T.;
Jarosinski, M. A. J. Pept. Res. 2000, 55, 359–371.
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14859

A R T I C L E S
Hiruma-Shimizu et al.
Table 1. Structural Statistics of Average Conformational Energies
and RMSD for the 20 Accepted Structures of Glycosylated Mouse
Notch-1 EGF 12 (3)
a
Average potential energies (kcal mol-1
)
Etotal
18.446 ( 0.528
0.864 ( 0.043
8.155 ( 0.166
0.915 ( 0.049
3.008 ( 0.552
4.276 ( 0.384
0.0054 ( 0.004
Ebonds
Eangle
Eimpr
EVDW
b
b
ENOE
b
Ecdih
RMSD from idealized geometry
bonds (Å)
0.0012 ( 0
angles (deg)
impropers (deg)
0.346 ( 0.0031
0.1647 ( 0.0034
Pairwise rmsd of 20 structures from
Cys456 to Cys467, Gly472, vPhe474 and Cys476 to Ile489 (Å)
Backbone atoms (N, CR, C′)
All heavy atoms
0.62 ( 0.14
1.06 ( 0.15
a 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. b The
force constants for the calculations of EVDW, ENOE, and Ecdih were 4.0
Figure 2. NOESY spectrum (mixing time 400 ms) of EGF 12 (3). NOEs
between two ꢀ-protons of cross-linked cysteine residues were observed in
C1-C3 and C5-C6.
kcal mol-1 Å-4
, , ,
50 kcal mol-1 Å-1 and 200 kcal mol-1 rad-2
Enzymatic Sugar Elongation of EGF 12 (3). In order to obtain
a series of glycosylated EGF 12 peptides, we carried out
sequential enzymatic elongation of the disaccharide on folded
glycopeptide 3 by recombinant ꢀ-1,4-galactosyltransferase (ꢀ-
1,4-Gal-T) and R-2,3-sialyltransferase (R-2,3-Sia-T) (see Scheme
1). These enzymatic reactions efficiently allowed the production
of trisaccharide glycopeptide 4 and the tetrasaccharide glyco-
peptide 5. Both peptides were purified by RP-HPLC and were
corroborated by MS analyses. These compounds will be applied
to future structural and functional studies in Notch-ligand
interactions.
respectively.
tion spectroscopy (TOCSY) measurement, which was recorded
in D2O solvent, revealed that a total of 11 backbone amide
protons had slowly exchanging rates [Figure S4 (3)]. This
indicates that the formation of hydrogen bonding in these amino
acids may provide EGF 12 with a stable conformation. From
calculations using a NOE network, coupling constants, and
hydrogen bond information, we concluded that synthetic EGF-
like glycopeptide 3 contains a double-stranded antiparallel
ꢀ-sheet, which is typical for the EGF family [Table 1 and Figure
3A (3) and 3B (3)].
Structural Analysis of EGF 12 Peptides. In order to gain
insights into the structures of the folded products, we performed
NMR studies of purified naked EGF 12 (1) and glycosylated
EGF 12 peptides 2 and 3. The proton assignments and the
structure calculation were made for all three peptides. We
describe those for EGF 12 (3) as an example. The amide protons
of the 38 amino acid backbone were fully assigned, except for
those of the N-terminus Asp which were designated by
comparison with the published data on human Notch-1 EGF
10-13 (BMRB Entry 6031).23 NOESY spectra of peptide 3
showed two nuclear Overhauser enhancements (NOEs) between
ꢀ-protons of the cross-linked cysteine residues in C1-C3 and
C5-C6 (Figure 2). Although the NOE related to the C2-C4
cross-linkage could not be unambiguously assigned due to the
overlapping with other signals, the two NOEs observed are
sufficient to define the correct disulfide bonding pattern of
glycopeptide 3. The sequential connectivities between HR and
NH (dRN) were observed with strong intensity over the full length
of this peptide, whereas much weaker NOEs were observed with
the sequential connectivities between amide protons (dNN).
Furthermore, 21 amino acid residues of glycosylated EGF 12
(3) were shown to have a coupling constant between HR and
the NH proton (3JHNR) of more than 8.0 Hz. These data suggest
that glycosylated EGF 12 (3) is rich in ꢀ-sheet or extended
structures without helical conformation (Figure S3). The
hydrogen-deuterium exchange experiment using total correla-
All energies and rmsd values were calculated using the
CNS1.1 and MOLMOL programs, respectively.
To explore the effect of O-glycosylation on the structure of
the EGF-like domain, the solution structures of both the naked
EGF 12 (1) and the O-fucosylated EGF 12 (2) were calculated
in the same manner as that of EGF 12 (3) after full assignment
of proton signals. From the NOE patterns observed and results
of hydrogen-deuterium exchange experiments, we conclude that
peptides 1 and 2 also have the characteristic structures of the
EGF family [Supporting Information, Figure S4 (1 and 2)]. The
superposition of 20 final structures shown in Figure 3A and
the energy-minimized structures shown in Figure 3B indeed
supported the structural similarities of the three EGF 12
domains. For each peptide compound, the 20 final structures
(Figure 3A) were well converged except for the flexible region
at the N-terminus and the hinge region of the two antiparallel
ꢀ-sheet strands. These tendencies were in good agreement with
the solution structure of the EGF domain of blood coagulation
factors.24 The superposition of the backbone C, N, CR atoms
of the most energy-minimized structures of EGF 12 (1) and
the glycosylated analogs 2 and 3 are shown in Figure 4. These
structures were superimposed by minimizing the difference for
the backbone atoms of residues 456-468, 473-489 to give an
rmsd value of 1.33 ( 0.32 Å in backbone. The results suggest
that O-fucosylation does not cause a major conformational
change in the overall backbone structure of EGF 12 of the mouse
(23) Hambleton, S.; Valeyev, N. V.; Muranyi, A.; Knott, V.; Werner, J. M.;
McMichael, A.; Handford, P. A.; Downing, A. K. Structure 2004,
12, 2173–2183.
(24) Kao, Y.-H.; Lee, G. F.; Wang, Y.; Starovasnik, M. A.; Kelley, R. F.;
Spellman, M. W.; Lerner, L. Biochemistry 1999, 38, 7097–7110.
9
14860 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

O-Fucosylated EGF Repeat 12 of Mouse Notch-1 Receptor
A R T I C L E S
Figure 3. Three dimensional structures of naked and O-glycosylated EGF 12 analogs. These figures were generated with the program MOLMOL. (A) A
stereoview of the backbone C, N, CR atoms of the 20 final structures for EGF 12 analogs. These structures were superimposed by minimizing the difference
for the backbone atoms of residues 456, 459-467, 476-488 in naked EGF 12 (1); 456, 460-467, 476-489 in O-fucosylated EGF 12 (2); and 456-467,
472, 474, 476-489 in O-glycosylated EGF 12 (3). The structures are oriented with the N terminus at the top. The heavy atoms of Thr466 residues are shown
in cyan, and sugar moieties and Fuc and GlcNAc residues (all their atoms), in red and green, respectively. (B) A stereoview of the schematic representation
of the most energy-minimized structures. The ꢀ-sheet strands are predicted by arrows, and the sugar moieties, including sugar-linked threonine, are shown
in ball and stick. Oxygen is shown as a red ball, nitrogen as blue, hydrogen as white, and carbon as gray, respectively.
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14861

A R T I C L E S
Hiruma-Shimizu et al.
Figure 4. A comparison of the energy-minimized structures of the three
EGF 12 peptides (1, 2, and 3). Each EGF 12 was superimposed by
minimizing the difference for the backbone atoms of residues 456-468
and 473-488. The blue line is for the backbone of naked EGF 12 (1); the
orange line, for that of the O-fucosylated EGF 12 (2); and the green line,
for that of the GlcNAc-Fuc-modified EGF 12 (3). The rmsd values for the
backbone atoms are 1.33 ( 0.32 Å. This figure was generated with the
program PyMOL.
Figure 6. A schematic representation of O-fucosylated EGF 12 (2) focusing
on Fucose, Thr466, and 475Ile-Cys-Met-Pro478. This figure was generated with
the program PyMOL.
methyl group of the fucose residue could fit in the intersheet
spacing of the antiparallel ꢀ-sheet (data not shown). This further
supports that the fucose moiety might directly function as a
“bridge” in the formation of the antiparallel ꢀ-sheet to stabilize
the structure through the interaction between fucose and the
peptide backbone.
Next we compared all structures of this consensus region to
probe the structural effect of O-fucosylation. The 20 final
ensembles of energy-minimized structures of the O-fucose
consensus sequence, Cys461-Gln-Asn-Asp-Ala-Thr-Cys467, are
shown in Figure 7. Interestingly, while the conformation of this
region was well converged in 3, which has the disaccharide
GlcNAcꢀ1,3Fucf moiety, the structures of the same region in
nonglycosylated 1 and fucosylated 2 seemed to be somewhat
disordered. The addition of the fucose residue resulted in some
change in the conformation of glutamine and asparagine.
Moreover, the GlcNAc residue that followed also caused a
significant structural change in asparagine and aspartic acid as
deduced from the different strengths of the sequential NOE
connectivities between HR and NH (dRN) (Figure S4). This is
likely to be caused by the interaction of the carboxyl group of
the Asp464 residue with the GlcNAc residue. Some of the Asp464
carboxyl groups in the 20 refined structures of (3) in Figure 7
were shown to be located close to O-6 of the GlcNAc residue,
suggesting a potential hydrogen bonding between these moieties.
Thus, although O-fucosylation does not cause an overall
conformational change in EGF 12 of the mouse Notch-1
receptor, there are considerable differences around the fucosy-
lation site including the O-fucose consensus sequence region.
The O-fucose consensus site in EGF-like repeats is considered
to be C2X4-5Ser/ThrC3,6 where X can be any amino acid. In a
previous structural study of the EGF-like domain of the blood
coagulation factor24 on which the O-fucosylated region has the
sequence C2QNGGTC3, a conformational change was not
observed to be associated with the introduction of the fucose
residue. A possible explanation for this is the lack of amino
acid side chains in this region which can interact with the fucose
residue. C2QNGGTC3 also occurs as the most frequent
sequence among the 21 O-fucosylated EGF-like repeats on
mouse Notch-1. However, the sequence C2QNDATC3 is unique
Figure 5. Chemical shift differences (in ppm) of CHR protons of
O-glycosylated EGF 12 (2) and (3) from naked EGF 12 (1).
Notch-1 receptor. Several differences were observed, however,
in more detailed NMR studies on the structures of the region
close to the glycosylated residue Thr466: there were clear
differences in chemical shift and the strength of sequential NOEs
connectivities between HR and NH (dRN), and changes were
detected in slowly exchanging amide protons (Figures 5 and
S4). This indicates that the structures of glycosylated EGF were
changed in part. In both O-fucosylated EGF 12 (2) and (3),
NOEs were observed between the ring protons of the fucose
residues and Thr466 as expected. Interestingly, several NOEs
were also observed between the methyl protons (C6) of the
fucose residue and the methionine residue (Met477) which is
opposite to Thr466 on the other strand of the antiparallel ꢀ-sheet
(Figure 6). Some additional NOEs between the fucose methyl
group and the neighboring amino acids of Met477, namely ILe475
and Pro478, were also observed (Figure 6).
These findings, together with the change in the chemical shift
of the R proton of Met477 (amino acid position 28 in Figure 5)
in the glycosylated EGF peptides, demonstrated the close contact
of the fucose residue with these amino acids (Figures 5, 6). In
addition, the atomic radius molecular modeling showed that the
9
14862 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

O-Fucosylated EGF Repeat 12 of Mouse Notch-1 Receptor
A R T I C L E S
Figure 7. 20 final ensembles of structures of the O-fucose consensus region (461Cys-Gln-Asn-Asp-Ala-Thr-Cys466) for EGF 12’s. In all images corresponding
to each EGF 12 (1-3), hydrogen atoms are shown in gray; carbon atoms, black; oxygen atoms, red; sulfur atoms, yellow; and nitrogen atoms, blue. In the
figures of glycosylated EGF 12 (2, 3), the fucose residue is shown in magenta. These images were generated with the program MOLMOL.
to the EGF 12 motif, and as described above, the presence of
Fuc and GlcNAc has a clear influence on the structure of this
sequence. Thus, the rare C2QNDATC3 sequence in the O-
fucosylated site of EGF 12 might have a special role in Notch
function.1,2 This is yet to be elucidated, as it has been predicted
from NMR titration experiments that the ligand binding site is
on the face of EGF 12 that is opposite to (away from) the
O-fucose,25 with the inference being that the sugar chain on
threonine may not be directly involved in ligand binding.
Conclusion
In summary, five analogs of the EGF 12 repeat of a mouse
Notch-1 receptor (compounds 1-5) have been synthesized and
correctly folded. By comparing the structures of the two
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14863

A R T I C L E S
Hiruma-Shimizu et al.
TIS ) 95:2.5:2.5:1 for 1 h at room temperature. The solution was
filtered and concentrated by a flow of nitrogen gas, and the crude
peptide was precipitated using cold tert-butylmethylether. To
synthesize naked EGF 12 (1) this crude precipitate was involved
in a folding process. To deprotect the O-acetyl group of the sugar
moiety the solid material was dissolved in basic methanol (pH 12.5)
containing DTT (6.5 mM), and the solution was stirred. After 2 h
the solution was neutralized and concentrated by a flow of nitrogen
gas at 40 °C. The crude material underwent preparative RP-HPLC
to remove excess DTT.
Folding of Glycopeptides. The linear glycopeptides (1 mg) were
dissolved in a redox buffer (20 mL) containing 100 mM ammonium
acetate (pH 8.0), with 1 mM reduced and 0.1 mM oxidized
glutathione. The mixtures were routinely stirred for 3 days at room
temperature, although all oxidation reactions were complete within
48 h as evidenced by analytical RP-HPLC (data not shown). After
neutralization by treating with 30% aqueous acetic acid, the
solutions were lyophilized and subjected to purification by RP-
HPLC (see Supporting Information, Figure S2).
Sugar Elongation by Glycosyltransferases. Glycosylated EGF
12 (3) was treated with a mixture of 2.2 mU of ꢀ-1,4-GalT and
UDP-Gal (1.79 µmol) in a total volume of 890 µL of 50 mM
HEPES/NaOH buffer, pH 7.0, and 10 mM MnCl2. After incubation
for 2 h at 25 °C, the reaction mixture was directly subjected to
preparative RP-HPLC to purify compound 4, whose transfer ratio
of galactose residue was estimated to be 100% based on the HPLC
profile. Subsequently, compound 4 (0.95 µmol) was treated with a
mixture of 98 µU of R-2,3-SiaT and CMP-Sia (1.89 µmol) in a
total volume of 950 µL of 50 mM HEPES/NaOH buffer, pH 7.0,
and 10 mM MnCl2. After incubation overnight at 25 °C, the reaction
mixture was subjected to preparative RP-HPLC. The resulting
HPLC profile of the purified material clearly indicated that
sialylation of the trisaccharide moiety of 4 proceeded to completion
to give the EGF 12 derivative bearing tetrasaccharide 5.
Purification of EGF 12 Derivatives. Synthetic naked peptides
and glycopeptides were purified using a preparative C-18 reversed-
phase column (Inertsil ODS-3 10 mm × 250 mm) on a HITACHI
liquid chromatography system (HPLC) with an L7150 pump, at a
flow rate of 4 mL/min. The column temperature was 40 °C, and
UV monitoring was carried out at 220 nm. Buffer A was distilled
water containing 0.1% TFA, and buffer B was acetonitrile contain-
ing 0.1% TFA. A linear gradient of 10-50% of B over 60 min
was used unless otherwise stated. In the case for the purification
of the peptides containing sialic acid, another buffer was used: A
was 25 mM ammonium acetate buffer pH 5.8, and B was
acetonitrile containing 10% of 25 mM ammonium acetate buffer
pH 5.8. A linear gradient of 2-6% of B over 60 min was used.
Analytical runs were performed on a HITACHI liquid chromato-
graphy system (ELITE Lachrom) with an Inertsil ODS-3 (4.6 mm
× 250 mm) reversed-phase column. UV absorbance was measured
at 220 nm, and the column temperature was 25 °C. The flow rate
was 1 mL/min, and elution conditions were described as follows:
the elution buffer was the same as that described above. A linear
gradient of 20-45% of B over 45 min was used to elute peptides
1-4. Purification of sialylated compound 5 was performed using
10-55% of B over 45 min.
synthetic O-fucosylated forms of EGF 12 (2 and 3) with the
nonglycosylated form (1), we provide evidence for conforma-
tional changes induced by glycosylation. Although O-fucosy-
lation does not confer a perceptible change in the overall
backbone structure of EGF 12, our results suggest that the fucose
residue may serve to stabilize the conformation of the antipar-
allel ꢀ-sheet structure of EGF 12. More importantly, we have
shown that the addition of the GlcNAc residue on the fucose
induces a significant structural change of the O-fucosylated
region. This provides structural insights into possible mecha-
nisms of the regulation of Notch-ligand interaction by Fringe.
Our findings are analogous to the reported poly-L-proline type
II helix-like conformation of the tandem repeating polypeptide,
antifreeze glycoproteins [AFGPs, poly(Ala-Ala-Thr)n]26 that
result from the addition of multiple O-GalNAc moieties of
threonine residues. The report is to our knowledge the first to
describe the structure determination of an EGF-like module
having an extended disaccharide O-fucose modification. This
opens the way to investigating the contribution of this confor-
mational change to the Notch mediated signal transduction.
Experimental Section
Reagents and General Methods. All commercially available
solvents and reagents used for peptide synthesis were A grade and
used without purification. Recombinant human ꢀ-1,4-GalT (EC
2.4.1.22) was purchased from TOYOBO Co. Ltd. R-2,3-(N)-SiaT
(EC 2.4.99.6) was purchased from Calbiochem Co. Ltd. Thermol-
ysin was purchased from Sigma. All mixing was performed by using
a vortex mixer. MALDI-TOFMS spectra were recorded with a
Bruker AutoFLEX mass spectrometer in linear positive mode using
matrix (2,5-dihydroxybenzoic acid). Samples for MALDI-TOFMS
were desalted and concentrated using 10-µL C18 ZipTips (Millipore)
according to the manufacturer’s instructions. Typically, samples
were dissolved in 1 µL of 30% (v/v) aqueous acetonitrile and mixed
with the same volume of a saturated solution of 2,5-dihydroxy-
benzoic acid in water. The above mixtures (1 µL) were applied to
a stainless steel target MALDI plate and air-dried before analysis
in the mass spectrometer. All 1D and 2D 1H NMR spectra for the
identification of the synthetic peptides were collected with a Bruker
AV-600 spectrometer at 600.13 MHz for proton frequency equipped
with a Cryo-probe at 300 K. Two-dimensional homonuclear double-
quantum-filtered scalar-correlated spectroscopy (DQF-COSY), TOC-
SY with an MLEV-17 sequence, and nuclear Overhauser enhance-
ment spectroscopy (NOESY) spectra were recorded in the indirect
dimension using States-TPPI phase cycling.
Solid-Phase Synthesis of EGF 12 (1-3). The solid-phase
synthesis of the glycosylated EGF 12 analogs was carried out using
PEGA resin (0.03 mmol) functionalized with a Rinkamide linker
(0.24 mmol/g) according to our previous report.16 After removing
the Fmoc group on resin by 20% piperidine in DMF, each Fmoc-
amino acid (0.09 mmol) except for the cysteine residue was coupled
with resin in the presence of HBTU (0.09 mmol), HOBt (0.09
mmol), and DIEA (0.18 mmol) in DMF with microwave irradiation.
Fmoc-cysteine was introduced using FmocCys(Trt)COOPfp (0.09
mmol) in the presence of HOBt (0.09 mmol) to prevent undesired
epimerization. As for the Thr466 residue, the building block 6 or 7
was used instead of Fmoc-Thr(tBut) in each case. Removal of NR-
Fmoc protection and the coupling reaction were repeated until the
N-terminal amino acid residue. Upon completion of the synthesis,
each peptide resin was treated with a mixture of TFA/EDT/H2O/
EGF 12 (1): Retention time of analytical RP-HPLC was 28.2
min. MALDI-TOFMS (m/z) calcd for C175H262N45O62S7 [M + H]+
4209.678, found 4209.574.
EGF 12 (2): Retention time of analytical RP-HPLC was 28.9
min. MALDI-TOFMS (m/z) calcd for C181H271N45O66S7 [M + H]+
4358.813, found 4358.88.
EGF 12 (3): Retention time of analytical RP-HPLC was 24.5
min. MALDI-TOFMS (m/z) calcd for C189H284N46O71S7 [M + H]+
4562.006, found 4560.763.
EGF 12 (4): Retention time of analytical RP-HPLC was 23.5
min. MALDI-TOFMS (m/z) calcd for C195H294N46O76S7 [M + Na]+
4746.128, found 4747.35.
(25) Cordle, J.; Johnson, A.; Tay, J. Z. Y.; Roversi, P.; Wilkin, M. B.; de
Madrid, B. H.; Shimizu, H.; Jensen, S.; Whiteman, P.; Jin, B.; Redfield,
C.; Baron, M.; Lea, S. M.; Handford, P. A. Nat. Struct. Mol. Biol.
2008, 15, 849–857.
(26) Tachibana, Y.; Fletcher, G. L.; Fujitani, N.; Tsuda, S.; Monde, K.;
Nishimura, S.-I. Angew. Chem., Int. Ed. 2004, 43, 856–862.
9
14864 J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010

O-Fucosylated EGF Repeat 12 of Mouse Notch-1 Receptor
A R T I C L E S
EGF 12 (5): Retention time of analytical RP-HPLC was 25.3
min. MALDI-TOFMS (m/z) calcd for C206H311N47O84S7 [M + H]+
5015.401, found 5016.11.
and refinement. In order to calculate the family of structures, we
used a total of 377 distance restraints and 27 dihedral angle φ
restraints for EGF (1), a total of 374 distance restraints and 30
dihedral angle φ restraints for Fuc EGF12 (2), and a total of 517
distance restraints and 39 dihedral angle φ restraints used for
glycosylated EGF12 (3), respectively. Distance restraints for
calculations were estimated from the cross-peak intensities in
NOESY spectra with a mixing time of 400 ms; the estimated
restraints were then classified into five categories (very strong,
strong, medium, weak, and very weak) and assigned upper limits
of 2.6, 2.9, 3.5, 5.0, and 6.0 Å, respectively. In the first stage of
structure determination, synthesized peptides 1-3 were calculated
using only interproton distance information. After the validation
of fulfilling distance restraints for the obtained structure, the
restraints of the dihedral angle φ were adopted for structure
Thermolytic Digestion. Thermolysin was dissolved in 10 mM
phosphate buffer, and the concentration was adjusted to 0.03 mM
before use. After the purification, each folded EGF (100 µg) was
digested with diluted thermolysin (10 µL) in 25 mM ammonium
acetate buffer pH 5.8 (100 µL) at 37 °C for 24 h and separated by
RP-HPLC. Each fraction was analyzed by MALDI-TOFMS and
amino acid analysis. The fragments derived from compound 1 were
shown in the text. Although the fragment due to C5-C6 (477Ile-
Cys-Met-Pro-Gly-Tyr-Glu-Gly-Val-Tyr-Cys-Glu488, theoretical m/z
as [M + H]+ 1379.575) was identified at m/z 1379.19 for 2 and
m/z 1379.27 for 3, respectively, ion peaks due to the fragments
containing C1-C3 or C2-C4 could not be detected.
3
NMR Spectroscopy. Sequential assignments were achieved
according to the standard methods for small proteins established
by Wu¨thrich and co-workers.27 Naked EGF12 (1) was dissolved
at a final concentration of 0.75 mM, glycosylated EGF12 (2) was
0.67 mM, and disaccharide EGF12 (3) was 0.55 mM in 300 µL
containing 90% H2O/10% D2O at pH 5.3, respectively. The pH
was adjusted to 5.3 by addition of aq NaOH and aq HCl. TOCSY
experiments were applied for a spin-locking time of 100 ms, and
NOESY experiments were carried out with mixing times of 100,
200, and 400 ms. The suspension of water signal was performed
by presaturation during the relaxation delay (2 s) and by a 3-9-19
WATERGATE pulse sequence with a field gradient.28,29 TOCSY
and NOESY spectra which were acquired with 2048 × 2048 data
matrices were zero-field to yield 2048 × 2048 data matrices. DQF-
COSY spectra with 8192 × 512 frequency data points were also
recorded and zero-field to yield 8192 × 8192 matrices in order to
measure the coupling constants. The sweep width of 8192 Hz was
applied. Time domain data in both dimensions were multiplied by
a sine bell window function with a 90° phase shift prior to Fourier
transformation. All NMR data were processed by NMRPipe
software30 and analyzed using the Sparky program.31
calculation. When the coupling constant JHNR was more than 8.0
Hz and less than 6.0 Hz, the dihedral angle φ was constricted to
-120 ( 30° and -60 ( 30°, respectively.27 In the final step of
structure determination, hydrogen bond restraints were used as
distance constraints of 1.5-2.5 Å between amide protons and
carbonyl oxygen and 2.5-3.5 Å between amide nitrogen and amide
protons. Amide protons protected from exchange with solvent were
identified by amide hydrogen/deuterium exchange experiments
1
using a series of 1D H NMR and TOCSY spectra recorded after
reconstituting the EGF 12 in D2O. These protected amide protons
are presumably either involved in hydrogen bonds or buried in the
interior of the protein. When a hydrogen bond acceptor could be
clearly identified in the initial structures for a protected amide
proton, two distance constraints were then included in the refinement
stage of the structure calculation for each hydrogen bond. The
conformation of the sugar rings was held fixed to the chair
conformation. All analyses of rmsd values and the solution
structures of EGF 12 derivatives were performed with the
PROCHECK33 and MOLMOL34 programs.
Acknowledgment. This work was partly supported by grants
for “Innovation Project for Future Drug Discovery and Medical
Care” from the Ministry of Education, Culture, Science and
Technology and for “Development of Novel Diagnostic and Medical
Applications through Elucidation of Sugar Chain Functions” from
New Energy and Industrial Technology Development Organization
(NEDO), UK Research Councils’ Basic Technology Grant (GR/
S79268), and UK Engineering and Physical Research Councils
Translational Grant EP/G037604/1. We thank Mr. T. Hirose at the
Center for Instrumental Analysis, Hokkaido University, for amino
acid analysis.
The chemical shift assignments have been deposited in the
BioMagResBank database (http://www.bmrb.wisc.edu) under BMRB
accession number 11006 (3). The final constraints used in the
structure calculations have been deposited in the Protein Data Bank
with accession number 2RR0 (1), 2RR2 (2), 2RQZ (3).
Structure Calculation. Three-dimensional structures of synthe-
sized peptides 1-3 were calculated using the CNS 1.1 program32
with standard protocols for distance geometry-simulated annealing
(27) Wu¨thrich, K. NMR of Proteins and Nucleic Acids; John Wiley and
Sons: New York, 1986.
(28) Piotto, M.; Saudek, V.; Sklena´r, V. J. Biol. NMR 1992, 2, 661–665.
(29) Sklena´r, V.; Piotto, M.; Leppik, R.; Saudek, V. J. Magn. Reson., Ser.
A 1993, 102, 241–245.
Supporting Information Available: Experimental details for
the chemical synthesis of building blocks 6, 7 and the
supplemental data of structural analysis of peptides. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(30) Delaglio, F.; Grzesiek, S.; Vuister, G.; Zhu, G.; Pheifer, J.; Bax, A.
J. Biomol. NMR 1995, 6, 277–293.
(31) Sparky 3.111, a NMR assignment and integration software, was
downloaded on February 2, 2006 from http://www.cgl.ucsf.edu/home/
sparky/ and used. This site was updated on October 13, 2006: Goddard,
T. D. Kneller, D. G. SPARKY 3; University of California, San
Francisco, 1999.
JA105216U
(32) Bru¨nger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros,
P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.;
Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L.
Acta Crystallogr., Sect. D 1998, 54, 905–921.
(33) Laskowski, R. A.; Rullmann, J. A. C.; MacArthur, M. W.; Kaptein,
R.; Thornton, J. M. J. Biomol. NMR 1996, 8, 477–486.
(34) Koradi, R.; Billeter, M.; Wu¨thrich, K. J. Mol. Graph 1996, 14, 51–
55.
9
J. AM. CHEM. SOC. VOL. 132, NO. 42, 2010 14865