Hydrogen exchange study on the hydroxyl groups of serine and threonine residues in proteins and structure refinement using NOE restraints with polar side-chain groups

Article abstract of DOI:10.1021/ja206799v

We recently developed new NMR methods for monitoring the hydrogen exchange rates of tyrosine hydroxyl (Tyr-OH) and cysteine sulfhydryl (Cys-SH) groups in proteins. These methods facilitate the identification of slowly exchanging polar side-chain protons in proteins, which serve as sources of NOE restraints for protein structure refinement. Here, we have extended the methods for monitoring the hydrogen exchange rates of the OH groups of serine (Ser) and threonine (Thr) residues in an 18.2 kDa protein, EPPIb, and thus demonstrated the usefulness of NOE restraints with slowly exchanging OH protons for refining the protein structure. The slowly exchanging Ser/Thr-OH groups were readily identified by monitoring the ¹³C_β-NMR signals in an H ₂O/D₂O (1:1) mixture, for the protein containing Ser/Thr residues with ¹³C, ²H-double labels at their β carbons. Under these circumstances, the OH groups exist in equilibrium between the protonated and deuterated isotopomers, and the ¹³C _β peaks of the two species are resolved when their exchange rate is slower than the time scale of the isotope shift effect. In the case of EPPIb dissolved in 50 mM sodium phosphate buffer (pH 7.5) at 40 °C, one Ser and four Thr residues were found to have slowly exchanging hydroxyl groups (k _ex < ～40 s^-1). With the information for the slowly exchanging Ser/Thr-OH groups in hand, we could collect additional NOE restraints for EPPIb, thereby making a unique and important contribution toward defining the spatial positions of the OH protons, and thus the hydrogen-bonding acceptor atoms.

Full text of DOI:10.1021/ja206799v

ARTICLE
pubs.acs.org/JACS
Hydrogen Exchange Study on the Hydroxyl Groups of Serine and
Threonine Residues in Proteins and Structure Refinement Using NOE
Restraints with Polar Side-Chain Groups
Mitsuhiro Takeda,† JunGoo Jee,‡ Akira M. Ono,‡ Tsutomu Terauchi,‡ and Masatsune Kainosho*,†,‡
†Structural Biology Research Center, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan
‡Center for Priority Areas, Graduate School of Science and Technology, Tokyo Metropolitan University, 1-1 minami-ohsawa,
Hachioji 192-0397, Japan
S Supporting Information
b
ABSTRACT: We recently developed new NMR methods for
monitoring the hydrogen exchange rates of tyrosine hydroxyl
(TyrÀOH) and cysteine sulfhydryl (CysÀSH) groups in
proteins. These methods facilitate the identification of slowly
exchanging polar side-chain protons in proteins, which serve as
sources of NOE restraints for protein structure refinement.
Here, we have extended the methods for monitoring the
hydrogen exchange rates of the OH groups of serine (Ser)
and threonine (Thr) residues in an 18.2 kDa protein, EPPIb,
and thus demonstrated the usefulness of NOE restraints with slowly exchanging OH protons for refining the protein structure. The
slowly exchanging Ser/ThrÀOH groups were readily identified by monitoring the 13Cβ-NMR signals in an H2O/D2O (1:1)
mixture, for the protein containing Ser/Thr residues with 13C, 2H-double labels at their β carbons. Under these circumstances, the
OH groups exist in equilibrium between the protonated and deuterated isotopomers, and the 13Cβ peaks of the two species are
resolved when their exchange rate is slower than the time scale of the isotope shift effect. In the case of EPPIb dissolved in 50 mM
sodium phosphate buffer (pH 7.5) at 40 °C, one Ser and four Thr residues were found to have slowly exchanging hydroxyl groups
(kex < ∼40 sÀ1). With the information for the slowly exchanging Ser/ThrÀOH groups in hand, we could collect additional NOE
restraints for EPPIb, thereby making a unique and important contribution toward defining the spatial positions of the OH protons,
and thus the hydrogen-bonding acceptor atoms.
’ INTRODUCTION
realize this, we have recently developed NMR methods for
monitoring the proton/deuterium hydrogen exchange rates of
TyrÀOH and CysÀSH groups in proteins.10,11 In these meth-
ods, the NMR resonances of the carbon atoms attached to
the polar groups are observed in a 1:1 mixture of H2O and
D2O, where the side-chain polar groups exist in equilibrium between
the protonated and deuterated isotopomers. Because the chemical
shifts of the carbon resonances are slightly different between the
two isotopomers, due to an isotope shift effect,12 the isotopomer-
resolved carbon resonances are separately observed when the
exchange rate is slower than the inverse of the isotope shift.13 The
key behind these methods is that the line-widths of the target
carbon atoms, 13Cζ of Tyr and 13Cβ of Cys, should be reduced
by using Tyr and Cys with optimized isotope labeling patterns,
(2S,3R)-[0,α,β,ζ-13C4; β2,ε1,2-2H3; 15N]-Tyr (ζ-SAIL Tyr)14 and
L-[β-13C;β,β-2H2] Cys. In the case of an 18.2 kDa protein, E. coli
peptidyl prolyl cisÀtrans isomerase b (EPPIb), the exchange rates
of two TyrÀOH groups and the two CysÀSH groups were readily
found to be slower than the order of 10 sÀ1 by this method.10,11
The structure of a protein is stabilized by a large number
of hydrogen bonds between polar atoms in the polypeptide
chain.1À3 Therefore, it is quite important to investigate these
specific interactions to advance our understanding of the nature
of proteins.
Protein hydrogen exchange studies have played an important
role in investigations of hydrogen bonds.4 Especially, NMR
studies have made large contributions, because they can provide
atom-resolved information.5 However, because a conventional
NMR hydrogen exchange study involves the observation of
exchanging protons, this method is not optimal to study polar
groups with intrinsically rapid exchange rates, which is the case
for side-chain OH and SH groups.6À9 In addition, even when
a slowly exchanging OH/SH proton exists in a protein, it is
difficult to observe, because conventional 2D experiments, such
15
13
as 1HÀ N and 1HÀ C HSQC, cannot be applied. Therefore, a
large number of such observable OH/SH proton peaks probably
have been overlooked or mistakenly interpreted.
The situation would undoubtedly be improved, if the respec-
tive hydrogen exchange rates of OH/SH groups could be
monitored independently by another method. In an effort to
Received: July 27, 2011
Published: September 28, 2011
r
2011 American Chemical Society
17420
dx.doi.org/10.1021/ja206799v J. Am. Chem. Soc. 2011, 133, 17420–17427
|

Journal of the American Chemical Society
ARTICLE
isotope-labeled Ser and Thr were produced by using an E. coli cell-free
system, as described previously.15,16 The cell-free reactions were per-
formed in a vessel composed of 5 mL of an inner reaction solution and
20 mL of an outer feeding solution, with shaking for 8 h at 37 °C. The
synthesized Ser and Thr were each present at 0.5 mM, and all of the
other unlabeled amino acids were present at 1 mM, for both the inner
and the outer solutions. The produced EPPIb protein was purified
according to the previously reported method.18 Each NMR sample
contained 0.2À0.7 mM EPPIb in 50 mM sodium phosphate buffer,
containing 100 mM NaCl and 0.1 mM NaN3, pH 7.5 (meter reading).
NMR Measurements and Analysis. All NMR experiments
were recorded on a DRX600 spectrometer (Bruker Biospin; 600.3 MHz
for 1H) equipped with a TCI cryogenic triple resonance probe at 40 °C,
unless otherwise stated. The 13C direct observations of EPPIb samples
were performed with proton and/or deuterium decoupling by a
WALTZ16 1H and/or 2H decoupling scheme.19 The carrier frequencies
of carbon were 62 ppm for the Ser 13Cβ observation and 72 ppm for the
Thr 13Cβ observation. The sweep widths were both 4500 Hz, and the
repetition time was 5 s. In the experiment in the 100% H2O solution, a
4.1 mm o.d. Shigemi tube containing the protein solution was inserted
into a 5 mm o.d. outer tube containing pure D2O, for the 2H lock signal.
All of the acquired 1D-13C NMR data were processed without a window
function, to facilitate the comparison of the line-widths of the proto-
nated and deuterated β-carbon signals. All of the protein samples
dissolved in H2O/D2O (1:1) and 100% D2O solutions were incubated
for at least 1 day at 40 °C to ensure that the hydrogenÀdeuterium
equilibrium was completed before the NMR measurements.
Figure 1. Chemical structures of L-[β-13C;β,β-2H2] serine and
L-[β-13C;β-2H] threonine.
Scheme 1. Preparation of L-[β-13C;β,β-2H2]Sera
a Reaction conditions: (i) see ref 11, (ii) H+, Δ, then ion exchange.
Scheme 2. Preparation of L-[β-13C;β-2H] Thra
Collection of NOEs with Slowly Exchanging Hydroxyl
Protons. The assignments of the Ser/Thr hydroxyl protons and their
NOE peaks were performed by using 13C-edited NOESY-HSQC and
15N-edited NOESY-HSQC data, which were acquired with 0.3À0.5 mM
of the SAIL-EPPIb sample, exclusively composed of SAIL amino acids20
(Figure S1). The SAIL-EPPIb protein contains (2S,3S)-[ul-13C;15N;-
β-2H] Ser and L-[ul-13C;15N;γ,γ-2H2] Thr. The 15N-edited 3D-NOESY-
HSQC spectra were measured with a Bruker DRX600 spectrometer
equipped with a TXI cryogenic probe. The data points and the spectral
width of the 15N-edited NOESY-HSQC were 256 (t1) Â 48
(t2) Â 1024 (t3) points and 8400 Hz (ω1, 1H) Â 1800 Hz (ω2, 15N) Â
8400 Hz (ω3, 1H), respectively. The number of scans/FID was 32, and the
NOE mixing time was 200 ms. The repetition time was 1 s. The 13C-edited
3D-NOESY-HSQC experiments were performed with a DRX800 spectro-
meter equipped with a normal TXI probe, and the experiments were
acquired for two different regions, the aliphatic and aromatic regions. In the
first set for the aliphatic region, the data points and the spectral width were
210 (t1) Â 58 (t2) Â 1024 (t3) points and 12 000 Hz (ω1, 1H) Â 6700 Hz
(ω2, 13C) Â 10 000 Hz (ω3, 1H), respectively. The number of scans/FID
was 32. The repetition time was 1 s. The carrier frequency of carbon was set
to 40 ppm. In the second set for the aromatic region, the data points and the
spectral width were 200 (t1) Â 40 (t2) Â 1,024 (t3) points and 12 000 Hz
(ω1, 1H) Â 5800 Hz (ω2, 13C) Â 10 000 Hz (ω3, 1H), respectively. The
number of scans/FID was 48. The repetition time was 1 s. The carrier
frequency of carbon was set to 125.5 ppm. The mixing time for both of the
13C-edited NOESY-HSQC spectra was 145 ms.
a Reaction conditions: (i) phthaloyl dichloride, (ii) Bu3SnD, Pd(PPh3)4,
(iii) Ni-(S)-BPB-Gly complex, MeONa, MeOH, (iv) H+, Δ, then ion
exchange.
In this study, we have extended the method for monitoring the
hydrogen exchange rates of the SerÀOH and ThrÀOH groups.
To reduce the 13Cβ line-widths of Ser and Thr, which are
sensitive reporters for the OH/OD states of hydroxyl groups,
we synthesized L-[β-13C;β,β-2H2] Ser and L-[β-13C;β-2H] Thr
(Figure 1) and incorporated them into EPPIb by an E. coli cell-
free expression system.15,16 In the case of EPPIb dissolved in
50 mM sodium phosphate buffer (pH 7.5) at 40 °C, we could
identify one Ser and four Thr residues with slowly exchanging
hydroxyl groups. Subsequently, NOE restraints involving the
slowly exchanging OH and SH protons, identified in this (Ser,
Thr) and previous studies (Tyr, Cys),10,11 were implemented in
restrained simulated annealing molecular dynamics for the NMR
structure calculation of EPPIb, based on EPPIb composed
exclusively of SAIL amino acids. The effect of the supplemental
restraints on the quality of the NMR structure was evaluated.
’ MATERIAL AND METHODS
Structure Calculation with Distance Restraints Involving
Hydroxyl and Sulfhydryl Protons. In total, 34 NOE peaks invol-
ving SerÀOH, ThrÀOH, and TyrÀOH and 10 CysÀSH peaks were
assigned, and their peak intensities were converted to upper distance
restraints by using a calibration protocol implemented in CYANA.21
In the conversion of the peak intensity to the upper distance restraint,
the exchange-relaxation effect during the NOE mixing time was not
considered, assuming that exchange processes on the order of 10À1 sÀ1
did not appreciably affect the intensity.6
Chemical Syntheses of L-[β-13C;β,β -2H2] Serine and
L-[β-13C;β -2H] Threonine. L-[β-13C;β,β-2H2]Ser (Scheme 1) was
prepared by acid hydrolysis of the intermediate for the synthesis of
L-[β-13C;β,β-2H2]Cys.11 The synthesis of L-[β-13C;β-2H]Thr (Scheme 2)
was accomplished via aldol condensation of acetaldehyde, which was
prepared from sodium [1-13C]acetate, with a Ni-BPB-Gly complex17 to
obtain the diastereomerically pure threonine.
Preparation of E. coli Peptidyl-Prolyl CisÀTrans Isomerase
b, EPPIb, Selectively Labeled with Isotope-Labeled Ser and
Thr. The protein samples of EPPIb selectively labeled with each of the
The structural calculation of the EPPIb protein was performed in the
presence and absence of NOE restraints involving the OH and SH
17421
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427

Journal of the American Chemical Society
ARTICLE
Figure 2. 150.9 MHz [2H,1H]-decoupled 1D 13C NMR spectrum of EPPIb, selectively labeled with L-[β-13C] serine (A), L-[β-13C;β,β-2H2] serine (B),
L-[β-13C] threonine (C), and L-[β-13C;β-2H] threonine (D). These spectra were acquired at 40 °C at 150.9 MHz of 13C frequency. In (B) and (D),
proton and deuterium decoupling was applied during acquisition. The chemical shift differences between L-[β-13C] serine and L-[β-13C;β,β-2H2] serine
and those between L-[β-13C] threonine and L-[β-13C;β-2H] threonine are due to the deuterium isotope shift effect arising from the proton/deuterium
substitution of methylene protons. Each peak is labeled with its assignment.
protons. The structural restraints common to the two calculations were
the NOE restraints previously collected from SAIL-EPPIb samples,20
which is exclusively composed of SAIL amino acids, as listed in Support-
ing Information Figure 1, and the backbone torsion angle restraints
obtained from database searches with the program TALOS.22 Torsion
angle dynamics with 20 000 steps per conformer by CYANA generated
100 structures that lacked significant violations against the experimental
restraints. These structures were further refined by the AMBER 9
software package23 with an all-atom force field (ff99SB)24 and a general-
ized Born (GB) implicit solvent model.25 The refinement was comprised
of three stages: minimization, molecular dynamics, and minimization.
Minimization and molecular dynamics consisted of 1500 steps and a
20 ps duration, respectively. We used the force constants of 50 kcal/
(mol  Å2) and 200 kcal/(mol  rad2) for the distance and angle
restraints, respectively. Among the structures that did not signifi-
cantly violate the experimental restraints, we selected the 20 lowest
energy structures for analysis. The hydrogen acceptor atoms of the
side-chain hydrogen bonds were searched using the program
MOLMOL26 with the following criteria: for the distance between
the OH/SH protons, maximum distance between the proton and
acceptor atom, 2.4 Å; the maximum angle O(S)ÀH
(acceptor
3 3 3
atom), 45°; and the minimum number of conformers that satisfy the
criteria, 10 (out of 20).
The structural quality was evaluated with PROCHECK-NMR.27 The
program MOLMOL26 was used to visualize the structures. The co-
ordinates of the 20 energy-refined conformers of SAIL-EPPIb have been
deposited in the Protein Data Bank (accession code 2rs4). The chemical
shifts of SAIL-EPPIb, including the assignments of the OH and SH
17422
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427

Journal of the American Chemical Society
ARTICLE
These Ser and Thr residues were efficiently incorporated
into the EPPIb protein by using an E. coli cell-free expression
system.15,16 The prepared sample was dissolved in 100% H2O
buffer, and 1D-13C NMR spectra were acquired at 40 °C. The
13Cβ peaks of 6 Ser and 12 Thr residues were observed at
chemical shifts ranging from 63 to 66 ppm and 66 to 73 ppm,
respectively (Figure 2B,D). These peaks were assigned according
to previously established assignments.18 Overall, the 13Cβ reso-
nances were very sharp, about 4 Hz, due to the optimized isotope
labeling pattern. Especially, the deuteration of the attached
protons greatly contributed to the line-narrowing, which was
confirmed through a comparison with the line-widths of EPPIb
samples selectively labeled with L-[β-13C] Ser and L-[β-13C] Thr,
about 10 Hz (Figure 2A,C) (Tables S1,S2).
Observation of Isotope Shift Effects on the 13Cβ Atoms in
Ser/Thr Residues. The line-narrowing achieved by the special
stable isotope labeling pattern facilitated the detection of iso-
topomer-resolved 13Cβ peaks in an H2O/D2O (1:1) solution.
However, caution should be taken in the analysis, because the
two-bond isotope shift effect on the side-chain OH group and the
three-bond isotope shift effect on the backbone amide (NH)
group are observed for the 13Cβ resonances of Ser/Thr residues.
Therefore, their line-shapes in an H2O/D2O mixture are gov-
erned by the exchange rates of the two polar groups, as in the case
for the 13Cβ peak of a Cys residue.11
In the case of EPPIb, four types of line-shapes were observed
for the 13Cβ peaks of Ser/Thr residues in an H2O/D2O solution:
a four-line peak (Thr-23), a two-line peak with relatively large
splitting, 0.08À0.11 ppm (Thr-6, Thr-39, Thr-82, and Ser-93), a
two-line peak with relatively small splitting, 0.02 ppm (Thr-161-
Ser-163), and a single-line peak (others) (Figures 3, 4; Table S3).
Because the size of a three-bond isotope shift effect is often
smaller than that of a two-bond shift effect,11,28,29 and the splitting
width of the second group is close to the typical value of the two-
bond isotope shift effect on OH groups in sugar rings, about
0.09À0.12 ppm,28 we concluded that the splits observed for the
second and third groups arose from side-chain OH and backbone
amide groups, respectively. Taken together, both OH/OD and
NH/ND, only OH/OD, and only NH/ND isotopomers were
resolved in the first, second, and third groups, respectively. It
should be noted that the line-width of the fourth group is slightly
broader than that of 100% H2O or 100% D2O buffer, which
reflects the unresolved isotope effect. Therefore, the exchange
rates of the OH groups of Thr-6, Thr-23, Thr-39, Thr-82, and
Ser-93 are relatively slower, as compared to the size of the isotope
shift effect.
As judged from the spectral simulation, the exchange rate at
which the resolution of the isotopomer is observed for 13Cβ
resonances is slower than the order of 10 sÀ1. On the basis of the
quantitative estimation by a 13C EXSY experiment10,30 or a line-
shape analysis for the split peaks, the exchange rates of the slowly
exchanging OH groups are different from each other, which is
correlated with the frequency of the solvent exposure, governed
by the dynamic behavior of a protein (Thr-6, <0.2 sÀ1; Thr-23,
0.5 ( 0.2 sÀ1; Thr-39, 0.3 ( 0.1 sÀ1; Thr-82, <0.2 sÀ1; Ser-93,
26 ( 4 sÀ1) (Figure S2).
Figure 3. 150.9 MHz [2H,1H]-decoupled 1D 13C NMR spectra of
EPPIb, selectively labeled with L-[β-13C;β,β-2H2] serine. The labeled
EPPIb was dissolved in H2O (A), H2O/D2O (1:1) (B), and D2O (C)
buffers to concentrations ranging from 0.2 to 0.7 mM for each solution at
pH 7.5. Peaks are labeled with their assignments. These NMR experi-
ments were performed at 40 °C. The 13Cβ resonances for Ser-93 were
observed as two signals in an H2O/D2O (1:1) buffer, due to the isotope
effect from the side-chain hydroxyl groups. Overall, the line-width of
each peak in an H2O/D2O (1:1) buffer is slightly larger than that in H2O
or D2O, due to the unresolved isotope effect from the backbone
amide group.
protons and the distance restraints, have been deposited in the BioMa-
gResBank (accession code 11451).
’ RESULTS AND DISCUSSION
Optimization of the Isotope Labeling Patterns in Ser and
Thr Residues. We previously demonstrated for the EPPIb
protein that the exchange rates of the TyrÀOH and CysÀSH
groups can be estimated in an aqueous solution, through NMR
observations of the carbon atoms attached to the OH/SH groups
in an H2O/D2O (1:1) solution.11,12 The success of the previous
studies motivated us to apply this approach to an analysis of the
SerÀOH and ThrÀOH groups in EPPIb.
In Ser and Thr residues, a practically measurable isotope shift
effect on their OH group is observed for their 13Cβ atoms. Thus, a
high-resolution observation sufficient to resolve the isotopomer-
lines of the 13Cβ resonance was a key for success. For this
reason, we designed and synthesized L-[β-13C;β,β-2H2] Ser and
L-[β-13C;β-2H] Thr (Scheme 1). In these amino acids, the
carbon atom at the β position is selectively enriched by 13C,
and the hydrogen atoms attached to the 13Cβ atom are deuter-
ated, such that the dipole and scalar couplings involving the 13Cβ
atom are extensively eliminated.
Assignment of Slowly Exchanging OH Protons in Ser/Thr
Residues. The exchange rates of slowly exchanging OH groups
that give isotopomer-resolved peaks in an H2O/D2O solution are
slow enough to allow their OH protons to generate NMR-
1
observable H resonances.10 We thus searched for the OH
proton signals by NOE and scalar coupling-based experiments.
17423
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427

Journal of the American Chemical Society
ARTICLE
Figure 4. 150.9 MHz [2H,1H]-decoupled 1D 13C NMR spectra of EPPIb, selectively labeled with L-[β-13C;β-2H] threonine in H2O (A), H2O/D2O
(1:1) (B), and D2O (C) buffers. Sample concentrations were 0.2À0.7 mM, and all experiments were performed at 40 °C. Peaks are labeled with their
assignments. The 13Cβ resonances for Thr-6, Thr-39, and Thr-82 were observed as two signals in an H2O/D2O (1:1) buffer, due to the isotope effect
from the side-chain hydroxyl groups. The 13Cβ resonance for Thr-23 was observed as the four-line signals in H2O/D2O (1:1), due to the dual isotope
effect from the backbone amide and side-chain hydroxyl groups.
Figure 5. The 20 solution structures of SAIL-EPPIb, determined without (A) and with (B) distance restraints involving observable hydroxyl/sulfhydryl
protons, superimposed on a well-defined region (a.a. 1À87, 92À110, 118À136, and 153À162). The side-chains of the Ser, Thr, Cys, and Tyr residues
are colored blue, cyan, orange, and red, respectively.
Through the strong intraresidue NOE peaks between the OH
protons and the β protons, the discrete OH proton peaks could
readily be found at chemical shift values of 6.29, 8.22, 5.67, 6.00,
and 6.02 ppm for Thr-6, Thr-23, Thr-39, Thr-82, and Ser-93,
respectively (Figure S3). Furthermore, for the scalar-based
experiment, 2D HSQC-TOCSY was employed, and correlation
peaks between 13Cβ and OH protons were detected for Thr-6,
Thr-39, and Thr-82 (Figure S4). However, the correlation peaks
for Thr-23 and Ser-93 could not be detected, possibly due to the
smaller scalar coupling between Hβ and OH, the side-chain
motion, and/or the relatively faster hydrogen exchange. While
the assignment of the OH protons was performed using SAIL
proteins in this case, the assignment could also be accomplished
with a conventional 13C, 15N-double labeled protein.
Structure Refinement by NOE Restraints with Hydroxyl
and Sulfhydryl Protons. The identification of slowly exchan-
ging OH/SH protons in EPPIb10,11 prompted us to refine the
NMR structure of EPPIb by applying NOE restraints with
them. Through an analysis of the NOESY data of SAIL EPPIb
proteins,20 a total of 44 NOE peaks involving OH/SH protons
were assigned, and the corresponding NOE restraints were
created. It is noteworthy that, based on a spectral simulation, an
OH/SH proton with an exchange rate on the order of 10À103 sÀ1
gives rise to a discrete proton signal, while the isotopomer peaks
17424
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427

Journal of the American Chemical Society
ARTICLE
Figure 7. The 10 lowest energy structures of SAIL-EPPIbEPPIb,
superimposed for the regions including Tyr-30 (A), Tyr-36 (B), Cys-
31 (C), and Cys-121 (D). For each panel, structures determined without
(left) and with (right) distance restraints with OH/SH are shown. For
clarity, conformers were again superimposed for each displayed region.
Side-chain hydroxyl and sulfhydryl groups are circled or, when hydrogen
bonds are explicitly formed, enclosed by a dashed box together with its
hydrogen-bond acceptor atom.
Figure 6. The 10 lowest energy structures of SAIL-EPPIb, super-
imposed for regions including Thr-6 (A), Thr-23 (B), Thr-39 (C),
Thr-82 (D), and Ser-93 (E). For each panel, structures determined
without (left) and with (right) NOE distance restraints involving OH
are shown. For clarity, conformers were again superimposed for each
displayed region. Hydrogen, carbon, nitrogen, and oxygen atoms are
colored white, black, blue, and red, respectively. Side-chain hydroxyl
groups are circled or, when hydrogen bonds are explicitly formed,
enclosed by a dashed box together with its hydrogen-bond acceptor
atom. These figures were produced with the MolMol software.26
regard, the accurate determination of the side-chain conforma-
tions is also important.
The solution structure of EPPIb obtained by the SAIL
method, with the supplementary NOE information associated
with the side-chain exchangeable OH and SH groups, showed
only a slight improvement with these supplementary NOEs,
because the original structure was very accurately determined by
the SAIL method. The final structure was quite similar to that
obtained by X-ray crystallography (rmsd for the core region is
0.85 Å) (Figure S5). Once the positions of the OH/SH protons
are determined, the hydrogen-bond acceptor atoms could readily
be identified, based on the relative positions of the atoms in the
structure (see Materials and Methods). In the case of EPPIb,
seven side-chain hydrogen bonds were determined, and all of
them are embedded in the structure of EPPIb (Figure 8, Table 1).
A large proportion of the side-chain hydrogen bonds are local,
and some of them belong to recurring structural elements
(Figure 8CÀE,G). For example, the OH groups of Thr-39,
Thr-82, and Ser-93 are hydrogen bonded to the backbone
carbonyl carbons three residues behind them in the primary
sequence (Figure 8CÀE). In these cases, the intervening resi-
dues are likely to adopt a right-handed α helix-like structure, and
the χ1 angle of the Ser/Thr residue is almost always restricted to
the g+ conformation.31 The hydrogen bond between the SH
group of Cys-31 and the carbonyl carbon three residues behind in
the α-helix is also regarded as a typical pattern of an SH-involved
in a local hydrogen bond (Figure 8G).3 On the other hand,
long-range hydrogen bonds seem to play an important role in
maintaining the tertiary arrangement of the structural elements,
as in the case of the hydrogen bond between TyrÀOH and the
backbone carbonyl carbon of Ala-96 (Figure 8F). It is particularly
interesting that bifurcated hydrogen bonds are formed between
Thr-23 and Glu-71 (Figure 8B). In this situation, the side-chain
carboxylate of Glu-71 serves as an N-terminal cap, through the
are unresolved in the observation of the 13Cβ signal.6,7,10 There-
fore, we searched for observable NOE peaks involving rapidly
exchanging OH groups. In the present case, however, no obser-
vable OH protons could be found in the NOESY spectra.
To evaluate the effect of the OH/SH restraints, we performed
the restrained simulated annealing with the AMBER force field24
for the structure of EPPIb in the presence and absence of the
OH/SH restraints. By virtue of the large number of NOE
restraints,20 both calculations produced high-quality NMR struc-
tures, where the positions of the oxygen/sulfur atoms in the
OH/SH groups were fairly well-defined (Figure 5; Table S4).
However, the locations of the attached OH/SH protons were
ambiguous in the structure calculated without the OH/SH
restraints, due to the peculiar hybrid orbital of the oxygen/sulfur
atom, which allows three rotamers about the CÀO/CÀS bond.
On the other hand, the positions of the OH/SH protons were
well-defined in the structures calculated with the OH/SH
restraints, in most cases (Figures 6,7; Table 1).
In a conventional structure calculation, the determination of
the rotameric conformations of the OH/SH groups depends
only on structural packing effects and/or energy minimization,
using a sophisticated force field. The present study highlights the
unique and important contribution of the OH/SH restraints to
determine the rotameric conformations of the side-chain polar
groups (Table 1). Of course, the determination of the con-
formation by the OH/SH restraints only works well when the
positions of the surrounding atoms are precisely defined. In this
17425
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427

Journal of the American Chemical Society
ARTICLE
Table 1. Conformations of Side-Chain Hydroxyl and
Sulfhydryl Groups in SAIL-EPPIb Structures Determined in
the Presence and Absence of Distance Restraints Involving
OH/SH
is small, they are probably involved in structurally or functionally
important hydrogen bonds in proteins. It is especially noteworthy
that the hydrogen bonds involving the OH groups of Thr-23, Thr-
39, and Ser-93 in EPPIb are conserved in structurally homologous
proteins (Figure S6), raising the possibility that long-lived side-chain
hydrogen bonds are likely to be conserved in homologous proteins,
due to their structural importance. Such conservation of water-
inaccessible side-chain polar groups in the interiors of homologous
proteins has been reported for proteins that satisfy the hydrogen-
bonding potential.33
conformation of side-chain polar groupb
relative
polar
solvent
accessibility (%)a
without restraints
with OH/SH
with restraints
with OH/SH
group
Ser-93 OH
Thr-6 OH
Thr-23 OH
Thr-39 OH
Thr-82 OH
Tyr-30 OH
Tyr-36 OH
Cys-31 SH
Cys-121 SH
0
g+, 0; gÀ, 16; t, 0c
g+, 0; gÀ, 12; t, 8
g+, 10; gÀ, 4; t, 6
g+, 3; gÀ, 17; t, 0
g+, 0; gÀ, 16; t, 4
E,d 4; Z, 16
g+, 0; gÀ, 20; t, 0
g+, 0; gÀ, 20; t, 0
g+, 20; gÀ, 0; t, 0
g+, 0; gÀ, 20; t, 0
g+, 0; gÀ, 20; t, 0
E, 6; Z, 14
’ CONCLUSION
0
0
We have developed a method for monitoring the hydrogen
exchange rates of SerÀOH and ThrÀOH groups. The method
only requires a protein selectively labeled with [β-13C; β,β-2H2]-
Ser and/or [β-13C; β-2H]Thr, which can be prepared by con-
ventional cellular expression systems, making it feasible for most
NMR laboratories. Using EPPIb, we demonstrated that NOE
restraints with NMR-observable OH and SH protons are quite
valuable for the refinement of the NMR structure, especially for
the residues bearing such OH and SH groups. Because the
chemical shift regions of 13Cβ in Ser, Thr, and Cys, and 13Cζ
of Tyr do not overlap each other, it is possible to prepare proteins
simultaneously labeled with the four types of amino acids and
to perform 13C NMR analyses.10,11 It should also be pointed out
that our approach can be used to detect intermolecular side-chain
hydrogen bonds.
0
0
0
2.3
0
E, 15; Z, 5
E, 20; Z, 0
g+, 18; gÀ, 0; t, 2
g+, 2; gÀ, 9; t, 9
g+, 20; gÀ, 0; t, 0
g+, 1; gÀ, 12; t, 7
0
a The relative solvent accessibility was evaluated for each oxygen/sulfur
atom by using EPPIb crystal structure34 (PDB code: 2NUL). b The
definitions of g+, gÀ, and t for the side-chain conformations are identical
to those reported by Ippolito.35 c Four out of 20 conformers have
different χ1 angles from the others and are not included. d Because the
Tyr OH protons are prone to be located in the aromatic ring plane,33 the
conformation of the OH proton is defined by an E/Z designation. When
the Tyr OH proton is proximal to Cε1, as compared to Cε2, it is defined
as the Z conformation; otherwise, it is the E conformation. Here,
however, the Cδ1 is defined as carbons for which the CαÀCβÀCγÀCδ
dihedral angle has a positive value, 0° to +180°, and Cδ2 has a negative
value, À180° to 0°. This definition differs from the previous definition,
in which the Cδ1 is defined as the carbon for which the absolute value of
the CαÀCβÀCγÀCδ dihedral angle is smaller than that involving the
other Cδ carbon.36
’ ASSOCIATED CONTENT
Supporting Information. 13C NMR EXSY experiment
S
b
for Ser/ThrÀOH groups; structural statistics of the SAIL-EPPIb
structure determined with/without distance restraints with OH
and SH protons; 13Cβ chemical shifts of Ser/Thr in different
isotope-labeling patterns and in 100% H2O, H2O/D2O (1:1),
100% D21O; experiment for detecting OH and SH protons based
1
on HÀ H vicinal couplings; side-chain hydrogen bonds in
homologous proteins; chemical structures of SAIL amino acids
used to prepare SAIL-EPPIb proteins; and comparison between
the NMR and crystal structures of EPPIb. This material is
available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
kainosho@tmu.ac.jp
Figure 8. The side-chain hydrogen bonds in EPPIb. The side-chain
hydrogen bonds with the hydroxyl groups of Thr-6 (A), Thr-23 (B),
Thr-39 (C), Thr-82 (D), Ser-93 (E), Tyr-36 (F), and the sulfhydryl
group of Cys-31 (G), those were determined in this study, are depicted
as black dashed lines on the lowest-energy SAIL-EPPIb NMR structure.
In (B), the hydrogen bond between the Thr-23 amide and the Glu-71
carboxylate is shown by a gray dashed line, to clarify the bifurcated
hydrogen bonds.
’ ACKNOWLEDGMENT
This work was supported by the Targeted Protein Research
Program (MEXT) to M.K., a Grant-in-Aid for Young Scientists
(B) (23770109) to M.T., and a Grant-in-Aid in Innovative Areas
(4104) to M.K. and J.J.
formation of a hydrogen bond with the amide of Thr-23. The
hydrogen bond between the Thr-23 OH group and the Glu-71
carboxylate stabilizes the interaction.
Side-Chain Hydrogen Bonds Involving Slowly Exchan-
ging OH Groups Are Likely To Play an Important Role in
Stabilizing the Structure of a Protein. As shown for EPPIb,
proteins sometimes contain slowly exchanging and NMR-observa-
ble OH/SH protons in their interiors.6,32 Although their population
’ REFERENCES
(1) (a) Baker, E. N.; Hubbard, R. E. Prog. Biophys. Mol. Biol. 1984,
44, 97–109. (b) Sticke, D. F.; Presta, L. G.; Dill, K. A.; Rose, G. D. J. Mol.
Biol. 1992, 226, 1143–1159.
(2) Eswar, N.; Ramakrishnan, C. Protein Eng. 2000, 13, 227–238.
(3) (a) Gregoret, L. M.; Rader, S. D.; Fletterick, R. J.; Cohen, F. E.
Proteins 1991, 9, 99–107. (b) Zhou, P.; Tian, F.; Lv, F.; Shang, Z. Proteins
2009, 76, 151–63.
17426
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427

Journal of the American Chemical Society
ARTICLE
(4) (a) Havdt, A.; Nielsen, S. O. Adv. Protein Chem. 1966,
21, 287–386. (b) Englander, S. W.; Downer, N. W.; Teitelbaum, H.
Annu. Rev. Biochem. 1972, 41, 903–924. (c) Woodward, C.; Simon, I.;
Tuchsen, E. Mol. Cell. Biochem. 1982, 48, 135–160. (d) Englander, S. W.;
Mayne, L. Annu. Rev. Biophys. Biomol. Struct. 1992, 21, 243–265. (e)
Baldwin, R. L. Curr. Opin. Struct. Biol. 1993, 3, 84–91. (f) Englander,
S. W.; Kallenbach, N. Q. Rev. Biophys. 1984, 16, 521–655. (g) Bai, Y.;
Sosnick, T. R.; Mayne, L.; Englander, S. W. Science 1995, 269, 192–197.
(h) Englander, S. W.; Sosnick, T. R.; Englander, J. J.; Mayne, L. Curr.
Opin. Struct. Biol. 1996, 6, 18–23. (i) Li, R.; Woodward, C. Protein Sci.
1999, 8, 1571–1590. (j) Krishna, M. M. G.; Hoang, L.; Lin, Y.;
Englander, S. W. Methods 2004, 34, 51–64.
(5) (a) Wagner, G.; W€uthrich, K. J. Mol. Biol. 1979, 130, 31–37. (b)
Wagner, G.; W€uthrich, K. J. Mol. Biol. 1979, 134, 75–94. (c) Wagner, G.;
W€uthrich, K. J. Mol. Biol. 1982, 160, 343–361. (d) Roder, H.; Wagner,
G.; W€uthrich, K. Biochemistry 1985, 24, 7396–7407.
(6) Liepinsh, E.; Otting, G.; W€uthrich, K. J. Biomol. NMR 1992,
2, 447–465.
(25) Tsui, V.; Case, D. A. Biopolymers 2000, 56, 275–291.
(26) Koradi, R.; Billeter, M.; W€uthrich, K. J. Mol. Graphics 1996,
14, 51–55.
(27) Laskowski, R. A.; Rullmann, J. A. C.; MacArthur, M. W.;
Kaptein, R.; Thornton, J. M. J. Biomol. NMR 1996, 8, 477–486.
(28) Bosco, M.; Picotti, F.; Radoicovich, A.; Rizzo, R. Biopolymers
2000, 53, 272–280.
(29) Ottiger, M.; Bax, A. J. Am. Chem. Soc. 1997, 119, 8070–8075.
(30) (a) Fischer, M. W. F.; Zeng, L.; Zuiderweg, E. R. P. J. Am. Chem.
Soc. 1996, 118, 12457–12458. (b) Bertini, I.; Felli, I. C.; K€ummerle, R.;
Moskau, D.; Pierattelli, R. J. Am. Chem. Soc. 2004, 126, 464–465.
(31) Eswar, N.; Ramakrishnan, C. Protein Eng. 1999, 12, 447–455.
(32) (a) L€ohr, F.; Mayhew, S. G.; R€uterjans, H. J. Am. Chem. Soc.
2000, 122, 9289–9295. (b) Plevin, M. J.; Hayashi, I.; Ikura, M. J. Am.
Chem. Soc. 2008, 130, 14918–14919.
(33) Worth, C. L.; Blundell, T. L. Proteins 2009, 75, 413–429.
(34) Edwards, K. J.; Ollis, D. L.; Dixon, N. E. J. Mol. Biol. 1997,
271, 258–265.
(7) Liepinsh, E.; Otting, G. Magn. Reson. Med. 1996, 35, 30–42.
(8) Otting, G.; W€uthrich, K. J. Am. Chem. Soc. 1989, 111, 1871–1875.
(9) Otting, G.; Liepinsh, E.; W€uthrich, K. J. Am. Chem. Soc. 1991,
113, 4363–4364.
(35) Ippolito, J. A.; Alexandera, R. S.; Christianson, D. W. J. Mol. Biol.
1990, 215, 457–471.
(36) Markley, J. L.; Bax, A.; Arata, Y.; Hilbers, C. W.; Kaptein, R.;
Sykes, B. D.; Wright, P. E.; W€uthrich, K. J. Mol. Biol. 1998, 280, 933–952.
(10) Takeda, M.; Jee, J.; Ono, A. M.; Terauchi, T.; Kainosho, M.
J. Am. Chem. Soc. 2009, 131, 18556–18562.
(11) Takeda, M.; Jee, J.; Terauchi, T.; Kainosho, M. J. Am. Chem. Soc.
2010, 132, 6254–6260.
(12) (a) Hansen, P. E. Annu. Rep. NMR Spectrosc. 1983, 15, 105–234.
(b) Dziembowska, T.; Hansen, P. E.; Rozwadowski, Z. Prog. Nucl. Magn.
Reson. Spectrosc. 2004, 45, 1–29.
(13) (a) Hawkes, G. E.; Randall, E. W.; Hull, W. E.; Gattegno, D.;
Conti, F. Biochemistry 1978, 17, 3986–3993. (b) Feeney, J.; Partington,
P.; Roberts, G. C. K. J. Magn. Reson. 1974, 13, 268–274. (c) Kainosho,
M.; Tsuji, T. Biochemistry 1982, 21, 6273–6279. (d) Kainosho, M.;
Nagao, H.; Tsuji, T. Biochemistry 1987, 26, 1068–1075. (e) Markley,
J. L.; Kainosho, M. In Stable Isotope Labeling and Resonance Assignments
in Larger Proteins: NMR of Macromolecules; Roberts, G. C. K., Ed.;
Oxford University Press: New York, 1993; pp 101À152. (f) Uchida, K.;
Markley, J. L.; Kainosho, M. Biochemistry 2005, 44, 11811–11820.
(14) Kainosho, M.; Torizawa, T.; Iwashita, Y.; Terauchi, T.; Ono,
A. M.; G€untert, P. Nature 2006, 440, 52–57.
(15) Takeda, M.; Ikeya, T.; G€untert, P.; Kainosho, M. Nat. Protoc.
2007, 2, 2896–2902.
(16) Torizawa, T.; Shimizu, M.; Taoka, M.; Miyano, H.; Kainosho,
M. J. Biomol. NMR 2004, 30, 311–325.
(17) Yuri, N.; Belokon, Y. N.; Bulychev, A. G.; Vitt, S. V.; Struchkov,
Y. T.; Batsanov, A. S.; Timofeeva, T. V.; Tsyryapkin, V. A.; Ryzhov,
M. C.; Lysova, L. A.; Bakhmutov, V. I.; Belikov, V. M. J. Am. Chem. Soc.
1985, 107, 4252–4259.
(18) Kariya, E.; Ohki, S.; Hayano, T.; Kainosho, M. J. Biomol. NMR
2000, 18, 75–76.
(19) Shaka, A. J.; Keeler, J.; Frenkiel, T.; Freeman, R. J. Magn. Reson.
1969, 52, 335–338.
(20) Takeda, M.; Terauchi, T.; Ono, A. M.; Kainosho, M. J. Biomol.
NMR 2010, 46, 45–49.
(21) (a) Mumenthaler, C.; G€untert, P.; Braun, W.; W€uthrich, K.
J. Biomol. NMR 1997, 10, 351–362. (b) G€untert, P. Methods Mol. Biol.
2004, 278, 353–378. (c) Herrmann, T.; G€untert, P.; W€uthrich, K. J. Mol.
Biol. 2002, 319, 209–227.
(22) Cornilescu, G.; Delaglio, F.; Bax, A. J Biomol. NMR 1999,
13, 289–302.
(23) Case, D. A.; Cheatham, T. E.; Darden, T.; Gohlke, H; Luo, R.;
Merz, K. M.; Onufriev, A.; Simmerling, C.; Wang, B.; Woods, R. J.
J. Comput. Chem. 2005, 26, 1668–1688.
(24) (a) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz,
K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.;
Kollman, P. A. J. Am. Chem. Soc. 1995, 117, 5179–5197. (b) Hornak, V.;
Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Proteins
2006, 65, 712–25.
17427
dx.doi.org/10.1021/ja206799v |J. Am. Chem. Soc. 2011, 133, 17420–17427