Glycosylation of threonine of the repeating unit of RNA polymerase II with β-linked N-acetylglucosame leads to a turnlike structure

Article abstract of DOI:10.1021/ja982312w

Two models of the repeating C-terminal domain of RNA polymerase II (Ac- SYSPTSPSYS-NH₂; Ac-SYSPT(β-O-GlcNAc)SPSYS-NH₂) were prepared and their conformations in water studied using 1-D and 2-D ¹H NMR spectroscopies, CD spec

Full text of DOI:10.1021/ja982312w

J. Am. Chem. Soc. 1998, 120, 11567-11575
11567
Glycosylation of Threonine of the Repeating Unit of RNA
Polymerase II with â-Linked N-Acetylglucosame Leads to a Turnlike
Structure
Eric E. Simanek,† Dee-Hua Huang,† Laura Pasternack,† Timothy D. Machajewski,†
Oliver Seitz,† David S. Millar,‡ H. Jane Dyson,‡ and Chi-Huey Wong*,†
Contribution from the Departments of Chemistry and Molecular Biology, The Scripps Research Institute,
10550 North Torrey Pines Road, La Jolla, California 92037
ReceiVed July 2, 1998
Abstract: Two models of the repeating C-terminal domain of RNA polymerase II (Ac-SYSPTSPSYS-NH2;
Ac-SYSPT(â-O-GlcNAc)SPSYS-NH2) were prepared and their conformations in water studied using 1-D
1
and 2-D H NMR spectroscopies, CD spectrophotometry, fluorescence anisotropy, and molecular mechanics
and dynamics calculations. The data suggest that glycosylation of the native, randomly coiled peptide with a
single, biologically relevant sugar leads to the formation of a turn. This report represents the first structural
study of a new class of glycoproteins monoglycosylated with N-acetylglucosamine on threonine.
Introduction
by many groups: examples have been reported where glyco-
sylation leads to differences in structure between native and
glycosylated sequences.13-15 Herein we report that glycosyla-
tion with a single â-linked GlcNAc on the threonine residue of
a model of the polymeric domain of RNA polymerase II, 1,
The polymeric C-terminal domain (CTD) of the largest
subunit of the holoenzyme responsible for the synthesis of
mRNA, RNA polymerase II, consists of a highly conserved
repeating sequence, (YSPTSPS)n.1,2 The length of the sequence
varies with the evolutionary complexity of the organism;
mammals have 52 repeats,3,4 Drosophila have 45 repeats,4,5
yeasts have 27 repeats,6,7 and plasmodia have 18 repeats.8 While
it is generally agreed that hyperphosphorylation of the CTD
converts the polymerase from an inactive, promoter-bound state
to an actively transcribing assembly,9 the pattern of phospho-
rylation remains uncertain.10,11 Recently, Hart and colleagues
recognized that the CTD is also glycosylated with a single,
â-linked N-acetylglucosamine (GlcNAc) on threonine.12 While
the role that glycosylation plays is unclear, the modification of
the CTD through glycosylation or phosphorylation is mutually
exclusive.
Ac-Ser-Tyr-Ser-Pro-Thr(â-O-GlcNAc)-
Ser-Pro-Ser-Tyr-Ser-NH2
1
Ac-Ser-Tyr-Ser-Pro-Thr-Ser-Pro-Ser-Tyr-Ser-NH2
2
leads to increased structure over the corresponding sugar-free
peptide. We find evidence for a nonrandom, turnlike structure
about the site of glycosylation for 1 and random coil structure
for 2. The evidence for the turnlike structure comes primarily
from 1H NMR: nuclear Overhauser effect (NOE) connectivities,
variable temperature coefficients of the amide protons, chemical
shift differences of amide and HR protons of 1 and 2, and the
The search for general principles that describe the role of
glycosylation in peptides13-15 and proteins16 is being pursued
† Department of Chemistry.
1
rates of H/2H exchange are all consistent with a turn. CD
‡ Department of Molecular Biology.
spectrophotometry and fluorescence anisotropy measurements
corroborate this conclusion. Computation using NMR-derived
constraints produces a model with turnlike structure about the
site of glycosylation.
Model System. Molecules 1 and 2 contain the heptad repeat
unit and additional amino acids appended to the N- and
C-termini. Our choice to include these residues is based on
previous studies of models of the native peptide (similar to 2)
by Suzuki,17 Harding,18 and Corden.19
On the basis of data from sedimentation and fluorescence
quenching studies, Suzuki concluded that the uncapped se-
quence, YSPTSPSY, bound to DNA. Noting the similarity of
the sequence to the known bisintercalator, Triostin A, Suzuki
(1) Corden, J. L. TIBS 1990, 15, 383-387.
(2) Woychik, N. A.; Young, R. A. TIBS 1990, 347-351.
(3) Corden, J. L.; Cadena, D. L.; Ahearn, J. M., Jr.; Dahmus, M. E. Proc.
Natl. Acad. Sci. U.S.A. 1985, 82, 7934-7938.
(4) Allison, L. A.; Wong, J. K.-C.; Fitzpatrick, V. D.; Moyle, M.; Ingles,
C. J. Mol. Cell. Biol. 1988, 8, 321-329.
(5) Zehring, W. A.; Lee, J. M.; Weeks, J. R.; Jokerst, R. S.; Greenleaf,
A. L. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 3698-3702.
(6) Allison, L. A.; Moyle, M.; Shales, M.; Ingles, C. J. Cell 1985, 42,
599-610.
(7) Nonet, M.; Sweetser, D.; Young, R. A. Cell 1987, 50, 909-915.
(8) Li, W. B.; Bzik, D. J.; Gu, H.; Tanaka, M.; Fox, B. A.; Inselburg, J.
Nucleic Acids Res. 1989, 17, 9621-9636.
(9) Dahmus, M. E. Biochim. Biophys. Acta 1995, 1261, 171-182.
(10) Baskaran, R.; Dahmus, M. E.; Wang, J. Y. J. Proc. Natl. Acad. Sci.
U.S.A. 1993, 90, 11167-11171.
(11) Zhang, J.; Corden, J. L. J. Biol. Chem. 1991, 266, 2290-2296.
(12) Kelly, W. G.; Dahmus, M. E.; Hart, G. W. J. Biol. Chem. 1993,
268, 10416-10424.
(15) Andreotti, A. H.; Kahne, D. J. Am. Chem. Soc. 1993, 115, 3352.
(16) Varki, A. Glycobiology 1993, 3, 97.
(17) Suzuki, M. Nature 1990, 344, 562-565.
(18) Harding, M. M. J. Med. Chem. 1992, 35, 4658-4664.
(19) Cagas, P. M.; Corden, J. L. Proteins 1995, 149.
(13) Blanco, F. J.; Jimenez, M. A.; Herranz, J.; Rico, M.; Santoro, J.;
Nieto, J. L. J. Am. Chem. Soc. 1993, 115, 5887.
(14) O’Connor, S. E.; Imperiali, B. J. Am. Chem. Soc. 1997, 119, 2295.
10.1021/ja982312w CCC: $15.00 © 1998 American Chemical Society
Published on Web 11/03/1998

11568 J. Am. Chem. Soc., Vol. 120, No. 45, 1998
Scheme 1. Synthesis of the Glycosylated Amino Acid
Simanek et al.
Figure 1. Chemical shift differences from values expected for random
coils23 for the NH and HR of 1 and 2.
1
NMR Spectroscopy. One-dimensional H NMR spectros-
copy afforded data on coupling constants, chemical shift, VT
coefficients, and amide NH exchange rates. Two-dimensional
ROESY and NOESY experiments provided a consistent set of
NOEs. These data are discussed individually in the following
paragraphs.
(a) Coupling Constants. Table 1 summarizes relevant
coupling constants and chemical shift information collected for
1 and 2. For 1, we observed a 3JHN,HR(Thr5) ) 8.6 Hz, implying
a transoid relationship between these protons: we adopted a
dihedral constraint of -160° < φ < -80° in our computational
proposed that the tyrosine residues intercalated into the helix.
Harding’s examination of the conformation of the uncapped
sequence YSPTSPSY in water by 1H NMR found that a small
population exists in the conformation proposed by Suzuki. Two
â-turns were identified as important for placing the aromatic
rings of the tyrosine residues in the correct geometry. Contrary
to Harding’s results, Corden found that (YSPTSPS)8 exists as
a random-coil structure in water but adopted a â-turn structure
in trifluoroethanol.
To reconcile these results, we chose to add serine residues
to both the N- and C-termini of the Suzuki/Harding model to
increase the steric constraints. We also capped the N- and
C-termini with acetyl and carboxamide groups, respectively, to
reduce ion-pairing interactions.
3
models.22 The JHN,HR of the remaining amides of 1, and for
those of 2 all fall within a range (6.0-8.0 Hz) where structural
predictions are difficult. The very small coupling constant
observed between Thr5HR and Thr5Hâ (J ) 1.8 Hz) suggests
that there is little free rotation about the CR-Câ bond. We
adopted dihedral constraints for (HR-CR-Câ-Hâ) in which
55°< φ < 60° and 110°< φ < 120°.
(b) Chemical Shift Differences. Local structure can be
inferred by comparing the chemical shifts of 1 and 2 to values
tabulated for random coil sequences (Figure 1).23 While there
are not substantial differences in the shifts of HR for 1 or 2,
the amide protons of Thr5 and Ser6 of 1 are markedly different
than those of 2 and the values recorded for peptides with random
structure.
Results
Synthesis of 1 and 2. Peptides 1 and 2 were synthesized
using Fmoc chemistry20 in a peptide synthesis vessel using Rink
Amide AM resin with a loading density of 0.65 mmol/g as the
solid support. The amino acids were purchased (Fmoc-Ser(O-
t-Bu)COOH, Fmoc-Tyr(O-t-Bu), Fmoc-Pro; Novabiochem) or
prepared (Fmoc-Thr-(O-â-GlcNAc(OAc)3)COOH) by a known
protocol (Scheme 1).21 Details of the iterative synthesis
(Scheme 2) are provided in the Experimental Section.
Peptide Purification. Purification of 1 and 2 was effected
in three phases. The product oil obtained from the cleavage
cocktail was purified by size exclusion on an LH-20 column.
Fractions containing peptide were combined, evaporated, and
subjected to silica gel chromatography. The fractions containing
purified peptide were combined and submitted to HPLC
purification using preparative reverse-phase chromatography.
These materials gave single ions (M + Na+) by MALDI-MS.
For 1, Zemplen deprotection afforded a single product by HPLC
which gave satisfactory MALDI-MS (M + Na+) and 1H NMR.
Full details are available in the Experimental Section.
(c) NOES. Two-dimensional NOESY and ROESY spec-
troscopy (Figures 2-4) yielded a series of cross-peaks that are
represented with double headed arrows in Scheme 3. The
existence of proline residues led us to probe for â-turn
structures.24 The NOEs that are diagnostic for turns (where i
+ 1 ) Pro) include dNN(i+2,i+3), dRN(i,i+2), and either dRN
-
(i+1,i+2) for type II or dδN(i+1,i+2) for type I turns. While
we see cross-peaks at dNN(i+2,i+3) and dRN(i+1,i+2) for both
potential turn regions (i ) Ser3 and Ser6) in 1, we see no cross-
turn NOEs at dRN(i,i+2) for either region (shown as empty boxes
in Figure 5). The absence of these critical NOEs in water
prevents us from assigning type II â-turn structure to these
regions. We propose, however, that a turnlike structure exists
(Vida infra). Tables 2 and 3 give NOE and dihedral constraints
utilized for computational modeling.
(d) Variable Temperature Coefficients. The extent to
which an amide proton is exposed to solvent is reflected in the
dependence of its chemical shift on temperature. Exposed
(20) Sheppard, R. C.; Williams, B. J. J. Chem. Soc., Chem. Commun.
1982, 587.
(22) Botuyan, M. V.; Toy-Palmer, A.; Chung, J.; Blake, R. C.; Beroza,
P.; Case, D. A.; Dyson, H. J. J. Mol. Biol. 1996, 263, 752.
(21) Meinjohanns, E.; Meldal, M.; Bock, K. Tetrahedron Lett. 1995, 36,
9205.
(23) Wishart, D. S.; Sykes, B. D. Methods Enzymol. 1994, 239, 363.
(24) Nelsoney, C. L.; Kelly, J. W. Bioorg. Med. Chem. 1996, 4, 739.

Glycosylation of Threonine of RNA Polymerase II
J. Am. Chem. Soc., Vol. 120, No. 45, 1998 11569
Scheme 2. Iterative Synthesis of 1 and 2
amides such as those in a randomly coiled peptide shift more
than those in an ordered structure.25 The variable temperature
spectra for 1 (Figure 6) reveal that the VT coefficients of Thr5
and Ser6 of 1 are significantly lower than those recorded for
the other amide protons (Table 1), suggesting that these protons
are involved in structure either directly through hydrogen
bonding or indirectly by being buried within the protein. The
values for the remaining amide protons of 1, as well as those
recorded for 2, reveal that these protons are not effectively
protected from solvent.
(e) Rates of 1H/2H Exchange. Amide NH involved in
structure or engaged in hydrogen bonding show different rates
of exchange than those amides of a randomly coiled peptide.25
Adding 25% (v/v) D2O to a solution of 1 in H2O at 278 K
reveals that the all amide protons except those of Thr5 and Ser6
exchange rapidly. Exchange for these two residues is >20×
slower at 278 K, thus offering further evidence for their
involvement in a structured domain resulting from glycosylation.
All amide protons of 2 exchange at similar rates.
CD and Fluorescence Measurements. CD spectrophotom-
etry (Figure 7) and fluorescence anisotropy measurements
corroborate a structural change upon glycosylation. A minimum
around 198 nm in the CD spectra is consistent with a peptide
having a random-coil structure. While this feature is the only
feature in the spectrum of 2, the traces of 1 show local maxima
at 190 nmsa feature consistent with a turnlike structure.26
Fluorescence anisotropy measurements for 1 and 2 suggest that
tyrosines of 1 are held more rigidly in place than those of 2
(0.0065 vs 0.0045). While these values are significant, their
magnitude suggests that the ends of both peptides remain highly
flexible.
Computation. Independent computational analyses using
Monte Carlo conformation searches and simulated annealing
calculations reveal a common low-energy structure (Figure 8).
(a) Monte Carlo Calculations using Macromodel. Monte
Carlo conformational searches were performed on an SGI
Octane workstation using Macromodel v5.5. Our initial model-
ing was carried out on the complete model of 1. The NOE
distance and dihedral constraints (Tables 2 and 3) were applied
(25) Wutrich, K. NMR of Proteins and Nucleic Acids; Wiley: New York,
1986.
(26) Woody, R. W. The Peptides; Academic Press: New York, 1985;
Vol. 7.

11570 J. Am. Chem. Soc., Vol. 120, No. 45, 1998
Table 1. Assignment of Glycopeptide, VT Coefficientsa
Simanek et al.
residue of 1
NH
3JRN
VTa
VTb
HR
Hâ1,2
3.59
3.60, 2.81
3.60
2.10
4.15
3.73
2.10
3.60
3.63, 2.87
3.62
Hγ1,2
other
AcSer1
Tyr2
8.19
8.24
8.16
6.9
7.5
7.4
3.9
4.8
3.5
4.7
5.5
4.4
4.20
4.46
4.57
4.12
4.31
4.61
4.29
4.22
4.41
4.20
Me, 1.74
o, 6.92; m, 6.62
Ser3
Pro4
1.80, 1.77
1.84, 1.65
δ
1,2, 3.55, 3.45
Me, 0.91
Thr5
7.80
7.93
8.5
6.4
1.0
1.6
1.9
2.6
Ser6
Pro7
δ1,2, 3.68, 3.58
Ser8
8.25
8.11
8.12
7.3
6.5
7.0
3.2
3.3
3.2
3.3
4.4
4.0
Tyr9
o, 6.95; m, 6.64
NHa,b, 7.10, 6.85
SerNH210
GlcNAc: NH, 8.29; H1, 4.41; H2, 3.56; H3, 3.40; H4, 3.31; H5, 3.31; H6a, 3.65; H6b, 3.78
residue of 2
NH
3JRN
VTb
HR
Hâ1,2
Hγ1,2
other
AcSer1
Tyr2
8.22
8.21
8.25
8.1
7.8
7.7
3.5
4.2
3.9
4.29
4.53
4.67
4.25
4.26
4.73
4.39
4.34
4.57
4.33
3.76, 3.73
2.99
3.71
2.23
4.18
3.83
2.21
3.71
2.93
Me, 1.84
o, 6.90; m, 6.60
Ser3
Pro4
1.90-1.93
δ
1,2, 3.55, 3.64_
Me, 1.12
Thr5
8.12
8.27
7.0
7.0
3.0
3.9
Ser6
Pro7
1.96, 1.77
δ1,2, 3.79, 3.68
Ser8
8.36
8.35
8.30
7.0
6.7
7.1
4.4
5.0
4.4
Tyr9
o, 7.02; m, 6.70
NHa,b
SerNH210
3.71
1
a The assignment of the H NMR spectra of 1 and 2 acquired in 90%:10% H2O:D2O at pH 3.25. Chemical shifts are reported in parts per
3
million. JRN are reported in hertz. VT coefficients measured (a) between 278 and 298 K and (b) between 278 and 308 K are included (-∆ppb/
∆K).
1
Figure 2. NH-HR region of the 2-D H NMR spectrum of 1 recorded in 90%:10% H2O:D2O at pH 3.5.
with the default force constant of 100 kJ/(mol‚Å2). In those
cases where the distance constraint could be applied to more
than one hydrogen (methylene or methyl hydrogens), the
constraint was set to the corresponding carbon with an ap-
propriate increase in the acceptable deviation from the mean
distance. The Monte Carlo simulation was set to vary all
dihedral angles except the amide bonds, which were constrained
in the trans orientation. A total of 6200 Monte Carlo steps using
AMBER*27/PRCG minimization with the GB/SA solvent (H2O)
treatment28 to a gradient convergence (<0.05 kJ/mol) were
conducted. A low-energy structure was found which contained
a turn at the site of glycosylation.
(27) Senderowitz, H.; Parish, C.; Still, W. C. J. Am. Chem. Soc. 1996,
118, 2078.
(28) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am.
Chem. Soc. 1990, 112, 6127.

Glycosylation of Threonine of RNA Polymerase II
J. Am. Chem. Soc., Vol. 120, No. 45, 1998 11571
1
Figure 3. NH-NH region of the 2-D H NMR spectrum of 1 recorded in 90%:10% H2O:D2O at pH 3.5.
Scheme 3. Structural Model for 1a
Table 2. NOEs Recorded for 1 and the Assigned Distances
mean distance
(Å)
deviation
((Å)
atom 1
atom 2
GlcNAc-H1
GlcNAc-H1
GlcNAc-H1
GlcNAc-NH
GlcNAc-NH
GlcNAc-NH
GlcNAc-NH
GlcNAc-NH
Ser3-HR
Thr5-âMethyl
Thr5-Hâ
2.0
2.0
2.0
3.0
4.0
3.0
2.0
3.0
3.0
3.0
4.0
2.0
3.0
4.0
2.0
3.0
3.0
4.0
2.0
3.0
3.0
2.0
2.0
2.0
3.0
3.0
1.1
1.0
1.0
1.0
1.1
1.0
1.1
1.0
1.5
1.5
1.5
1.0
1.0
1.1
1.0
1.1
1.0
1.1
1.0
1.0
1.0
1.5
1.5
1.5
1.5
1.0
GlcNAc-H3
GlcNAc-H5
Thr5-âMethyl
GlcNAc-H1
GlcNAc-NAc
GlcNac-H3
Ser3-Câ
Ser3-HR
Pro4-Cδ
Thr5-NH
Thr5-NH
Thr5-NH
Thr5-NH
Thr5-HR
Thr5-Hâ
Ser6-NH
Ser6-NH
Ser6-NH
Ser6-NH
Ser6-NH
Ser6-HR
Ser6-NH
Ser6-HR
Ser3-NH
Ser3-NH
Pro4-Câ
Pro4-HR
Thr5-HR
Thr5-âMethyl
Thr5-Hâ
Thr5-âMethyl
Thr5-HR
a NOEs are shown as double headed arrows. The dashed line
indicates a potential hydrogen-bonding interaction. The cross-turn NOE
at dRN(Ser3,Thr5) is not observed. The actual position of the carbohydrate
is not shown: it lies above the plane of the turn. See Figure 8 for
more details.
Thr5-âMethyl
Thr5-NH
Thr5-Hâ
Ser6-HR
Ser6-Câ
To reduce the computational time, we examined the confor-
mational search of a truncated model of 1, Ac-Ser-Pro-Thr(â-
O-GlcNAc)-Ser-Pro-NH2. The nOe distance and dihedral
constraints were applied as before. A total of 15 000 Monte
Carlo steps using AMBER*/PRCG minimization without solvent
treatment were conducted. Of the 645 unique structures
obtained, the lowest energy conformer was found 18 times: an
additional 42 related conformers were found to be within 2.0
kcal/mol. All of these structures exhibited a turn at the site of
glycosylation. These conformations superimpose with an aver-
Ser6-Câ
Pro7-Cδ
Ser3-Câ
Ser3-HR
age backbone root mean square deviation of 0.560 Å from the
lowest energy conformer. The lowest 13 structures are shown
in Figure 9.
As a test of the validity of the computation on the truncated
model, the lowest energy conformations obtained from both

11572 J. Am. Chem. Soc., Vol. 120, No. 45, 1998
Simanek et al.
1
Figure 4. Detail of the aliphatic region of the 2-D H NMR spectrum of 1 recorded in 90%:10% H2O:D2O at pH 3.5.
Table 3. Coupling Constant Dihedral Angle Constraints for 1
using the Discover module of InsightII. Models for 1 and 2
were constructed using the biopolymer builder module. Initial
molecular dynamics (MD) calculations of the entire glycopeptide
indicated high variability of the end residues. Thus subsequent
calculations used the Ac-Ser-Pro-Thr-Ser-Pro-NH2 section and
carbohydrate.
mean angle deviation
atom 1
atom 2
atom 3
atom 4
(deg)
((deg)
Thr5-NH Thr5-N Thr5-CR Thr5-HR
Ser6-NH Ser6-N Ser6-CR Ser6-HR
-120.0
-120.0
40
40
Dynamics runs included minimization, randomization of
coordinates, equilibration at 1000 K, ramping of force constants
at 1000 K, cooling, and finally minimization using conjugate
gradients (to a maximum derivative of less than 0.01 kcal/Å).
A total of 26 experimentally determined constraints were applied
during the MD runs. Calculations were carried out on structures
in vacuo without the explicit inclusion of water molecules. A
distance-dependent dielectric constant of 4r was used. A total
of 60 final structures were obtained and categorized by visual
inspection (Figure 11). Of the 60 models 25 fell into a single
group whose relative energy was 2.6 kcal lower than any of
the other groups. This group was chosen for further analysis.
Superimposition of all 25 models gave an all atom root mean
square deviation of 1.18 Å with a maximum violation of NMR
constraints of 0.510 Å. A second structure of higher energy
was also found in which the carbohydrate was placed in the
plane of the turn (identified as the “open” structure of Figure
11).
Figure 5. Portions of the 2-D NOESY spectra of 1 recorded at 5 °C
in 90%:10% H2O:D2O at pH 3.5: (a) NH-NH region; (b) NH-HR
region. The boxes indicate the positions where dRN(i,i+2) cross-peaks
might have confirmed the existence of type II â-turns.
searches were superimposed (Figure 10). The turn is clearly
defined in both structures with only minor deviations in the
peripheral residues.
(b) Molecular Dynamics. Molecular dynamics calculations
utilizing the Amber force field with Homans’ carbohydrate
potentials29 were performed on an SGI Indigo 2 workstation
Discussion
What emerges from these studies is a model of the glyco-
sylated CTD of RNA polymerase II. Using computation guided
1
(29) Homans, S. W. Biochemistry 1990, 29, 9110.
by H NMR-derived distance constraints, we find that glyco-

Glycosylation of Threonine of RNA Polymerase II
J. Am. Chem. Soc., Vol. 120, No. 45, 1998 11573
Figure 7. CD spectrogram of 1 (bold lines) and 2 (dashed lines) at
278 and 298 K. The inset shows the traces expected for R-helix (R),
â-sheets (â), and random coils (RC).
Figure 8. (Top) Computational model of 1. Analysis of the computa-
tion-derived structures from Ac-Ser-Pro-Thr(GlcNAc)-Ser-Pro-NH2
revealed a stable conformation for the carbohydrate residue. The model
obtained from computation is consistent with experimental data.
Interestingly, the sites identified for phosphorylation, Ser3 and Ser6,
are close enough that a large conformational change might be expected
upon phosphorylation.11 (Bottom) CPK model of 1. The structure on
the left shows the carbohydrate over the analyzed domain, Ser-Pro-
Thr(GlcNAc)-Ser. The carbohydrate has been removed in the structure
on the right in order to show the buried amides of Thr5 and Ser6 (purple).
Figure 6. Variable temperature spectra for 1.
sylation leads to the formation of a turn. Both the VT
coefficients for the amides of Thr5 and Ser6 and their lower
rates of 1H/2H exchange in comparison to the other amides are
consistent with the structural model.20
Of greater interest perhaps is the comparison of two inde-
pendent calculations. Both Monte Carlo and molecular dynam-
ics calculations proVide the same low-energy structure in which
the carbohydrate lies over the plane of the turn.
While the biological relevance is unclear, we draw attention
to three aspects of the model that result from the close proximity
of the side chains of Ser3 and Ser6. In the lowest energy
structure they lie on the same face of the turn beneath the
carbohydrate group. First, in the lowest energy conformer these
two residues appear to make hydrogen bonding contacts with
hydroxyl groups and the glycosidic oxygen. The downfield shift
of the â-protons of Ser6 is consistent with such interactions.
Second, these two serine residues have been identified as sites
for phosphorylation.11 The existence of negative charges on
these two groups would presumably lead to severe ion-ion
repulsion, resulting in a dramatic conformational change. We
are currently exploring this hypothesis. Third, the model also
explains one aspect of the exclusivity of glycosylation and
phosphorylation: the carbohydrate shields the hydroxyl groups
of Ser3 and Ser6 from solvent and soluble factors (i.e. enzymes).
In summary, we have investigated, for the first time, the effect
of glycosylation on the local structure of an important class of
glycoproteins that are glycosylated with a single GlcNAc
residue. We have chosen water as a solvent, as it is likely to
best mimic the environment which is biologically significant.
Capping both N- and C-termini prevents ion-pairing interactions.
While a turn structure is revealed, the spectrum remains too
complex to interpret additional structural details: the region
around δ 3.6 ppm which contains signals for the â-protons of
five serine residues, some carbohydrate protons, and the
δ-protons of the prolines is too crowded for detailed analysis.
Additional model compounds are warranted. Work is in
progress to investigate further details of this monoglycosylation-

11574 J. Am. Chem. Soc., Vol. 120, No. 45, 1998
Simanek et al.
Rink Amide Resin were purchased from Novabiochem. Dimethylfor-
mamide (DMF) was obtained from Fluka. Other reagents for synthesis
were purchased from Aldrich Chemical Co. The peptide synthesis
vessel was obtained from Chem Glass.
Peptide Synthesis. (a) Deprotection. To 250 mg of resin, 10 mL
of a solution of 1:1 DMF:morpholine was added. The reaction vessel
was placed on an Innova 2000 platform shaker (160 rpm) for 50 min.
After removal of solvent by vacuum filtration, and repetitive rinsing
with DMF, the coupling sequence was followed.
(b) Coupling. First, protected amino acid (4 equiv, based on resin
loading), HOBT (6 equiv), and N-methylmorpholine (4 equiv) were
added to 10 mL of DMF. Upon dissolution, this solution was
transferred to a vial containing HBTU (4 equiv). Upon dissolution,
this solution was transferred to the reaction vessel. The vessel was
placed on the platform shaker for 4 h (160 rpm). After removal of
solvent by vacuum filtration, and repetitive rinsing with DMF, the
capping sequence was followed.
(c) Capping. Unreacted amine groups were capped by adding a
1:1 mixture of Ac2O and pyridine to the reaction vessel. The vessel
was placed on the platform shaker for 10 min. After removal of solvent
by vacuum filtration, and repetitive rinsing with DMF, the deprotection
sequence was followed.
Figure 9. Superimposition of the 13 lowest energy conformations of
the truncated model of 1 as determined from Monte Carlo simulations.
(e) Cleavage. Prior to cleavage, the N-terminus was capped
following the protocols established for deprotection and capping. To
cleave the peptide from the solid support, the resin was first washed
with copious amounts of CH2Cl2 and dried thoroughly under vacuum.
Next, a cleavage cocktail (10 mL) was prepared containing 90% (v:v)
trifluoroacetic acid (TFA), 5% H2O, 2.5% ethanedithiol, and 2.5%
anisole. The solution was added to the resin and the vessel placed on
the platform shaker for 3 h. The resulting solution was collected using
vacuum filtration, and the resin was washed with TFA.
Peptide Purification. A product oil was obtained upon removal of
cleavage cocktail with a rotary evaporator. This oil was loaded onto
a LH-20 column 50% methanol in chloroform. Fractions were screened
by thin layer chromatography (TLC, using 10% MeOH in CHCl3 as
an eluent) using ceric ammonium molybdate as a stain. Those fractions
containing peptide were combined, and the solvent was removed. Silica
gel chromatography using the same conditions provide a white solid
upon evaporation of the eluent. The purity of this material was assessed
using analytical HPLC.
Using a Hitachi HPLC system (L-6200A Intelligent Pump; L-4000
UV Detector; D-2500 Chromato-Integrator), the peptides were separated
using an acidic gradient of MeCN (98% MeCN; 2% H2O; 0.1% TFA)
and water (98% H2O; 2% MeCN; 0.1% TFA) on a protein and peptide
C-18 column from VYDAC. For analytical separation the gradient
started with 0% MeCN and grew to 10% MeCN over 30 min under a
constant 1 mL/min flow rate. Both 1 and 2 were judged by integration
to be >90% pure. Preparative scale HPLC was carried out using the
same acidic eluents at flow rates of 3 mL/min and a gradient of 0-20%
MeCN over 60 min using a VYDAC C-18 prep column. By collecting
only the “top” of the peaks, material that appeared >95% pure was
obtained. No visible impurities were observed in the analytical trace
Figure 10. Superimposition of the lowest energy conformations from
computation using the entire model of 1 (red; for clarity only Ser-Pro-
Thr(GlcNAc)-Ser-Pro is shown) with the truncated model of 1 (white).
1
or in the H NMR spectrum. These materials gave single ions (M +
Na+) by MALDI-MS.
The peracetylated glycopeptide was subjected to Zemplen depro-
tection. To a solution of dry methanol, Na° was added until the pH
was 8.5. This solution was then added to the peracetylated glycopep-
tide, and the reaction was monitored by analytical HPLC. In 3 h, a
distinct, single line was observed in the HPLC trace. Acidification of
the reaction mixture with AcOH and removal of solvent provided 1,
Figure 11. Visual inspection of the 60 lowest energy structures of 1
by molecular dynamics reveals two classes of structures: the low-energy
“closed” structure and a higher energy “open” structure. The low-energy
structure obtained is very similar to that obtained from Monte Carlo
simulations.
1
which gave satisfactory MALDI-MS (M + Na+) and H NMR.
1H NMR Spectroscopy. NMR experiments were carried out on
3-5 mM solutions of 1, using a Bruker AMX600. Data were recorded
in 100% D2O and 90%:10% H2O:D2O at pH ) 3.5 and pH ) 7.4. All
spectra were recorded at 5 °C. Adjusting the pH did not affect C-H
lines or NOE connectivities. No differences were observed when the
2-D spectra were recorded in the presence of 20 mM phosphate.
Spectra were recorded over spectral widths of 7200 Hz with quadrature
detection employed throughout. Two-dimensional spectra were ac-
quired in the phase-sensitive mode using time-proportional phase
induced conformational change, and how glycosylation affects
protein function.
Experimental Section
Fmoc-protected amino acids, hydroxybenztriazole (HOBT), o-
benzotriazolyltetramethyluronium hexafluorophosphate (HBTU), and

Glycosylation of Threonine of RNA Polymerase II
J. Am. Chem. Soc., Vol. 120, No. 45, 1998 11575
incrementation. Two-dimensional data sets typically containing 512
increments of t1 were acquired and zero filled to 1024 points of a 2048
point experiment. The number of scans varied from 32 to 64 for 2-D
experiments with a recycle delay of N ms. TOCSY spectra of 1 and
2 in 90%:10% H2O:D2O were acquired using the WATERGATE pulse
sequence for water suppression and a 65 ms mixing time. NOESY
and ROESY spectra of 1 and 2 in 90%:10% H2O:D2O were acquired
using the WATERGATE pulse sequence for water suppression and
three different mixing times (200, 400, and 600 ms). Data were
subjected to sine-bell weighting functions in f1 and f2, and a base-line
correction was applied.
was 2.5 mM. Traces recorded at pH ) 7 and pH ) 3.5 were
indistinguishable.
Acknowledgment. The authors wish to thank Dr. Linda
Tenant who assisted in the collection of CD spectra and Dr.
Jeff Kelly for providing a thoughtful reading of the manuscript.
This work was supported by NSF Grants CHE-9700391 (C.-
H.W.) and CA 27489 (H.J.D.).
Supporting Information Available: Text and figures
Fluorescence Anisotropy. Values for fluorescence anisotropy were
collected on a SimAmicon fluorimeter using an excitation wavelength
of 275 nm and monitoring at an emission wavelength of 310 nm. Data
were collected at pH ) 7 in a buffered phosphate solution (100 mM).
CD Spectrophotometry. CD traces were recorded on an Aviv 60I
spectrophotometer fitted with a Peltier block. Sample concentration
describing peptide synthesis and purification, computation,
Monte Carlo conformational search, and H NMR of 1 (17
pages, print/PDF). See any current masthead page for ordering
information and Web access instructions.
1
JA982312W