DOTA-M8: An extremely rigid, high-affinity lanthanide chelating tag for PCS NMR spectroscopy

Article abstract of DOI:10.1021/ja903233w

A new lanthanide chelating tag (M8) for paramagnetic labeling of biomolecules is presented, which is based on an eight-fold, stereoselectively methyl-substituted DOTA that can be covalently linked to the host molecule by a single disulfide bond. The steric overcrowding of the DOTA scaffold leads to an extremely rigid, kinetically and chemically inert lanthanide chelator. Its steric bulk restricts the motion of the tag relative to the host molecule. These properties result in very large pseudocontact shifts (>5 ppm) and residual dipolar couplings (>20 Hz) for Dy-M8 linked to ubiquitin, which are unprecedented for a small, single-point-attachment tag. Such large pseudocontact shifts should be well detectable even for larger proteins and distances beyond ～50 A. Due to its exceptionally high stability and lanthanide affinity M8 can be used under extreme chemical or physical conditions, such as those applied for protein denaturation, or when it is undesirable that buffer or protein react with excess lanthanide ions.

Full text of DOI:10.1021/ja903233w

Published on Web 09/28/2009
DOTA-M8: An Extremely Rigid, High-Affinity Lanthanide
Chelating Tag for PCS NMR Spectroscopy
Daniel Ha¨ussinger,*,† Jie-rong Huang,‡ and Stephan Grzesiek*,‡
Department of Chemistry, UniVersity of Basel, St. Johanns-Ring 19, 4056 Basel, Switzerland,
and DiVision of Structural Biology, Biozentrum, UniVersity of Basel, Klingelbergstrasse 50/70,
4056 Basel, Switzerland
Received April 22, 2009; E-mail: daniel.haeussinger@unibas.ch; stephan.grzesiek@unibas.ch
Abstract: A new lanthanide chelating tag (M8) for paramagnetic labeling of biomolecules is presented,
which is based on an eight-fold, stereoselectively methyl-substituted DOTA that can be covalently linked
to the host molecule by a single disulfide bond. The steric overcrowding of the DOTA scaffold leads to an
extremely rigid, kinetically and chemically inert lanthanide chelator. Its steric bulk restricts the motion of
the tag relative to the host molecule. These properties result in very large pseudocontact shifts (>5 ppm)
and residual dipolar couplings (>20 Hz) for Dy-M8 linked to ubiquitin, which are unprecedented for a small,
single-point-attachment tag. Such large pseudocontact shifts should be well detectable even for larger
proteins and distances beyond ∼50 Å. Due to its exceptionally high stability and lanthanide affinity M8 can
be used under extreme chemical or physical conditions, such as those applied for protein denaturation, or
when it is undesirable that buffer or protein react with excess lanthanide ions.
Introduction
site, the attachment has turned out to be quite a severe
stereochemical and physicochemical challenge, because lan-
Accurate determination of three-dimensional structures of
proteins in solution is one of the strongholds of modern NMR
spectroscopy. An even more demanding task is the precise
description of interaction sites and mechanisms in protein-protein
and protein-ligand complexes. All of these endeavors benefit
tremendously from recent developments in NMR techniques to
obtain long-range structural information. In addition to residual
dipolar couplings (RDCs), paramagnetic relaxation enhancement
(PRE) and pseudocontact shifts (PCS) caused by the interaction
of unpaired electrons in transition metal ions and nuclear spins
have recently attracted much attention.1-5
Due to their 1/r3 distance dependence, PCS yield particularly
valuable long-range distance and angular information. In many
cases, basic, very sensitive 2D-NMR experiments are sufficient
to extract precise PCS data, and as a consequence, PCS
determination is possible even for large biopolymers at the size
limit of current solution NMR techniques.5 To generate the PCS
effect, preferably lanthanide ions are attached to the protein,
since these transition metals have a particularly large magnetic
anisotropy. For the specialized case of metal-binding proteins,
this can easily be performed by exchanging the native metal
with lanthanide ions.6 In the absence of such a natural binding
thanides prefer a nine-fold coordination sphere, giving rise to
isomer formation.7-10 In addition, the metal ions need to be
positioned rigidly with respect to the protein; otherwise the PCS
is drastically reduced due to motional averaging. Several
lanthanide chelating tags (LCTs) based on EDTA,10-12 DTPA,7
and DOTA9,13 have been described that use convenient binding
to a single free cysteine on the surface of the protein. However,
thus far the observed PCS for these LCTs were much smaller
than for native metal-binding proteins, presumably due to the
motional freedom within the tag and/or of the tag relative to
the protein. Other approaches used zinc finger14 or EF hand15
domains fused to the protein of interest, which resulted in
similarly modest PCS and RDCs. More efficient tags have been
found in artificial lanthanide-binding peptides16-19 and in a rigid
(7) Prudencio, M.; Rohovec, J.; Peters, J. A.; Tocheva, E.; Boulanger,
M. J.; Murphy, M. E. P.; Hupkes, H. J.; Kosters, W.; Impagliazzo,
A.; Ubbink, M. Chem.sEur. J. 2004, 10, 3252–3260.
(8) Vlasie, M. D.; Comuzzi, C.; van den Nieuwendijk, A. M. C. H.;
Prudencio, M.; Overhand, M.; Ubbink, M. Chem.sEur. J. 2007, 13,
1715–1723.
(9) Keizers, P. H.; Desreux, J. F.; Overhand, M.; Ubbink, M. J. Am. Chem.
Soc. 2007, 129, 9292–3.
(10) Ikegami, T.; Verdier, L.; Sakhaii, P.; Grimme, S.; Pescatore, B.;
Saxena, K.; Fiebig, K. M.; Griesinger, C. J. Biomol. NMR 2004, 29,
339–349.
† Department of Chemistry, University of Basel.
‡ Division of Structural Biology, Biozentrum, University of Basel.
(1) Bertini, I.; Luchinat, C. Curr. Opin. Chem. Biolo. 1999, 3, 145–151.
(2) Rodriguez-Castaneda, F.; Haberz, P.; Leonov, A.; Griesinger, C. Magn.
Reson. Chem. 2006, 44, S10–6, Spec No.
(11) Gaponenko, V.; Altieri, A. S.; Li, J.; Byrd, R. A. J. Biomol. NMR
2002, 24, 143–148.
(12) Gaponenko, V.; Sarma, S. P.; Altieri, A. S.; Horita, D. A.; Li, J.; Byrd,
R. A. J. Biomol. NMR 2004, 28, 205–212.
(3) Mittag, T.; Forman-Kay, J. D. Curr. Opin. Struct. Biol. 2007, 17, 3–
14.
(13) Ha¨ussinger, D. In Book of Abstracts; First European Conference on
Chemistry for Life Sciences, Rimini, Italy, October 4-8, 2005.
(14) Gaponenko, V.; Dvoretsky, A.; Walsby, C.; Hoffman, B. M.; Rosevear,
P. R. Biochemistry 2000, 39, 15217–24.
(4) Pintacuda, G.; John, M.; Su, X. C.; Otting, G. Acc. Chem. Res. 2007,
40, 206–212.
(5) Otting, G. J. Biomol. NMR 2008, 42, 1–9.
(15) Ma, C.; Opella, S. J. J. Magn. Reson. 2000, 146, 381–384.
(16) Wohnert, J.; Franz, K. J.; Nitz, M.; Imperiali, B.; Schwalbe, H. J. Am.
Chem. Soc. 2003, 125, 13338–13339.
(6) Allegrozzi, M.; Bertini, I.; Janik, M. B. L.; Lee, Y. M.; Lin, G. H.;
Luchinat, C. J. Am. Chem. Soc. 2000, 122, 4154–4161.
9
10.1021/ja903233w CCC: $40.75  2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 14761–14767 14761

A R T I C L E S
Ha¨ussinger et al.
chemical tag based on dipicolinic acid,20 a tridentate ligand.
Lanthanide-loaded tags of this type resulted in large PCS and
RDCs, which are comparable to metal-binding proteins. The
drawback of these tags is their relatively low lanthanide affinity
(typically on the order of 10-6 to 10-8 molL-1), requiring
considerable excess of lanthanide ions in the NMR samples.19,20
Another very promising approach was demonstrated recently
by the Ubbink group, where a lanthanide chelator based on
DOTA8 or a cyclen-bis-pyridine oxide9 is attached to the target
protein via two amido-disulfide bridges. This results in very
rigid attachment, high affinity toward the metal, and very large
PCS and RDC effects. The drawback of this method is the need
for two solvent-accessible cysteine residues at a specific distance
from each other, which usually requires structural knowledge
beforehand and may also lead to unwanted disulfide formation
in certain cases.
Because of its extremely high affinity toward lanthanides on
21
the order of 10-25 to 10-27 molL-1
,
the cyclen derivative
DOTA (1,4,7,10 tetraaza-cyclododecane-tetraacetic acid) is
commonly used for caging lanthanides and other heavy metal
ions in magnetic resonance imaging and medical radioactive
applications. This high affinity also makes it an ideal lanthanide
chelating tag (LCT) for protein applications. Its extended polar
and hydrophobic surfaces are expected to engage in interactions
with the protein surface, thereby restraining its relative motion.
In this study we describe the syntheses of several new DOTA-
based LCTs. The particularly strong power of an eight-fold,
stereospecifically methylated DOTA (M8) to induce PCS is
demonstrated for two ubiquitin mutants in the folded state.
Furthermore, M8 can also be used to induce PCS in the unfolded
state under harsh conditions of denaturation.
Results
1
Figure 1. Part of the aliphatic region of the H NMR spectra (14.1 T) of
(A) [Lu(DOTA-SPy)] at 333 K and at 280 K and (B) [Lu(M8-SPy)] at 333
K and at 280 K.
Syntheses of Tags, Metalation and Ligation to Proteins.
Starting from a commercially available precursor, we initially
synthesized a DOTA-based chelating tag (DOTA-SPy)13 con-
nected to the protein by a covalent disulfide bond (see formula
in Figure 1A). However, the dysprosium complex of this new
tag [Dy(DOTA-SPy)] resulted in rather moderate PCS of only
a few hundred ppb and correspondingly low RDCs of maximally
5 Hz when covalently linked to the ubiquitin mutant S57C
(Supporting Information, Figure S1). We reasoned that the small
paramagnetic effects were related to the internal mobility of
the metal chelator. This internal mobility became evident from
PCS and RDCs. The stereochemistry of DOTA is very
complex,22-26 and simultaneous motions of the four fused five-
membered rings (δδδδ or λλλλ) and of the acetic acid side arms
can yield two stereochemically nonequivalent configurations (∆
or Λ) around the metal center. When the metal-loaded tag is
covalently linked to a chiral molecule, e.g. a protein, a total of
four magnetically inequivalent stereoisomers can be formed. For
some residues in ubiquitin S57C tagged with [Dy(DOTA)] up
to three resonances were observed. Therefore, the interpretation
of the data and also the sensitivity of the experiments are
problematic, and DOTA-SPy appears of little practical use. A
similar stereochemical averaging problem is expected to affect
previously described LCTs based on EDTA and DTPA.
To avoid averaging of the PCS from different conformations,
we have followed a recently described route27,28 to rigidify the
1
the temperature dependence of the H NMR spectrum of the
diamagnetic lutetium complex [Lu(DOTA-SPy)] (Figure 1A).
At 280 K, a complex spectrum is obtained corresponding to
the broken four-fold symmetry of the DOTA framework.
However, heating to slightly elevated temperatures leads to
coalescence of the spectrum at 333 K. Besides this internal
mobility, the flexibility of the linker between the protein and
the cyclen ring may also contribute to the partial averaging of
(22) Csajbok, E.; Baranyai, Z.; Banyai, I.; Brucher, E.; Kiraly, R.; Muller-
Fahrnow, A.; Platzek, J.; Raduchel, B.; Schafer, M. Inorg. Chem. 2003,
42, 2342–9.
(17) Martin, L. J.; Hahnke, M. J.; Nitz, M.; Wohnert, J.; Silvaggi, N. R.;
Allen, K. N.; Schwalbe, H.; Imperiali, B. J. Am. Chem. Soc. 2007,
129, 7106–7113.
(23) Csajbok, E.; Banyai, I.; Brucher, E. Dalton Trans. 2004, 2152–6.
(24) Aime, S.; Barge, A.; Botta, M.; Fasano, M.; Ayala, J. D.; Bombieri,
G. Inorg. Chim. Acta 1996, 246, 423–429.
(18) Su, X. C.; McAndrew, K.; Huber, T.; Otting, G. J. Am. Chem. Soc.
2008, 130, 1681–7.
(25) Aime, S.; Botta, M.; Fasano, M.; Marques, M. P. M.; Geraldes,
C. F. G. C.; Pubanz, D.; Merbach, A. E. Inorg. Chem. 1997, 36, 2059–
2068.
(19) Su, X. C.; Huber, T.; Dixon, N. E.; Otting, G. ChemBioChem 2006,
7, 1599–604.
(20) Su, X. C.; Man, B.; Beeren, S.; Liang, H.; Simonsen, S.; Schmitz, C.;
Huber, T.; Messerle, B. A.; Otting, G. J. Am. Chem. Soc. 2008, 130,
10486–7.
(26) Aime, S.; Barge, A.; Borel, A.; Botta, M.; Chemerisov, S.; Merbach,
A. E.; Muller, U.; Pubanz, D. Inorg. Chem. 1997, 36, 5104–5112.
(27) Ranganathan, R. S.; Pillai, R. K.; Raju, N.; Fan, H.; Nguyen, H.;
Tweedle, M. F.; Desreux, J. F.; Jacques, V. Inorg. Chem. 2002, 41,
6846–55.
(21) Rocklage, S. M.; Watson, A. D. J. Magn. Reson. Imaging 1993, 3,
167–78.
9
14762 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

DOTA-M8: Lanthanide Chelating Tag for PCS NMR
A R T I C L E S
Scheme 1. Synthesis of [Dy(M8-SPy)]a
a Conditions and reagents: a) R-Bn-lactate-OTf (0.7 equiv), CH2Cl2, 25 °C, 45 min, then NEt3, 57%; b) R-tBu-lactate-OTf (4.0 equiv), MeCN, K2CO3, 25
°C, 6 h, then NEt3, 90%; c) H2, Pd/C, MeOH, 25 °C, 14 h, 87%; d) PyBOP/NMM/DMF, then H2NC2H4SSpy, 25°C, 2 h; e) TFA/PhSMe/H2O (92/6/2), 25
°C, 5 h, 40%; f) DyCl3 (4 equiv), H2O, pH ) 5.50, 75 °C, 10 h, >90%.
DOTA moiety by the attachment of eight methyl groups. For
convenient linking to a free cysteine on the protein, this octa-
methyl derivative of DOTA was attached to a pyridine-thione
activator. We term this compound M8-SPy (see formula in
Figure 1B).
reverse decomplexation reaction. In fact, we have not been able
to remove dysprosium or lutetium from M8-SPy once it is
coordinated, not even with 250 mM potassium oxalate at pH
5.6 (the solubility product of dysprosium oxalate is 2.0 × 10-31
M5). Thus, M8 allows complete freedom in the choice of buffer,
including the presence of other metals and/or chelates without
interference as well as very high to very low pH conditions
(we have tested pH 9 and 2.5, but more extreme values should
be possible). As a consequence, the metal-loaded tag can be
conveniently purified by aqueous HPLC without detectable (ESI-
MS) loss of metal (see Supporting Information page S8).
In contrast to the diamagnetic parent [Lu(DOTA-SPy)]
complex, no dynamic processes could be detected by 1H NMR
for the new [Lu(M8-SPy)] complex in the temperature range
between 280 and 333 K (Figure 1B). Accordingly and unlike
[Lu(DOTA-SPy)], no exchange peaks were observed in an
EXSY spectrum of [Lu(M8-SPy)] with a mixing time of 1.0 s
at 295 K (data not shown), consistent with the complete rigidity
of the M8 LCT. The stereochemistry was investigated in detail
The M8 framework combines extremely strong affinity for
lanthanides with kinetic inertness, steric bulk, and the unique
property of forming only one stereoisomer when complexed with
trivalent lanthanides. A synthesis for the new LCT, M8-SPy,
comprising four S-configured 1-carboxy-ethyl side arms, one
of which is amide-coupled to a short C2H4-linker and an
activated disulfide, was developed as indicated in Scheme 1.
The starting material, 2S,5S,8S,11S-tetra-methyl-cyclen, was
synthesized from L-alanine via alaninol and 1-benzyl-2-methy-
laziridine following a method described previously.27 For the
N-alkylation of the parent macrocycle, triflate reagents had to
be used to prevent racemization. In a first step, monoalkylation
with benzyl-protected lactate was performed, followed by 3-fold
N-alkylation with tert-butyl-protected lactate. Deprotection of
the benzylester with Pd/H2 yields the monoacid which is amide-
coupled to the pyridinethione activated thiol. A second depro-
tection step affords the tricarboxylate, which is subsequently
metalated in aqueous solution. In-depth details of the syntheses
and stereochemistry of M8-SPy and of its diastereoisomer with
four R-configured side arms are reported in the Supporting
Information. The formation of metal complexes of M8-SPy
required temperatures of 80 °C and reaction times of several
hours at pH 5, which is in accordance with the highly strained
and inert nature of the ligand. The metalation reaction was
followed by NMR for the lutetium complex and is considerably
slower than for the parent compound DOTA-SPy. This indicates
a high kinetic barrier for metal incorporation into the sterically
overcrowded M8 framework, and the same is expected for the
1
for the diamagnetic [Lu(M8-SPy)] complex. H-1H-coupling
constants and ROE correlations (data not shown) indicate a
∆(δδδδ)-configuration of the complex and therefore a twisted
square antiprism as coordination polyhedron (Figure 2).
The 1H NMR spectrum of paramagnetic [Dy(M8-SPy)] shows
resonances between +365 ppm and -218 ppm (Supporting
Information, Figure S2) that are heavily broadened by para-
magnetic relaxation enhancement. Due to the resulting low
sensitivity and the enormous spectral range, it was impractical
to obtain 2D-NMR spectra of the paramagnetic dysprosium
complex.
To attach paramagnetic [Dy(M8-SPy)] to ubiquitin, two
mutants (K6C and S57C) were produced, which showed 1H-15N
HSQC spectra for untagged protein under reducing conditions
(1 mM TCEP) that were nearly identical to the native protein.
In particular, the line widths and intensities are uniform
throughout the protein, and larger chemical shift changes are
(28) Ranganathan, R. S.; Raju, N.; Fan, H.; Zhang, X.; Tweedle, M. F.;
Desreux, J. F.; Jacques, V. Inorg. Chem. 2002, 41, 6856–66.
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14763

A R T I C L E S
Ha¨ussinger et al.
Figure 3. 1H-15N HSQC spectra of Ubi-S57C tagged with diamagnetic
[Lu(M8)] (black) and Ubi-S57C tagged with [Dy(M8)] (red), both recorded
at 298 K and 14.1 T. Resonances in less crowded regions of the spectrum
are labeled with assignment information. Green asterisks indicate resonances
of a minor spectral species for the [Dy(M8)] complex (see text).
Figure 2. Schematic space-fill representation of the [Dy(M8)] complex;
(A) view from the nitrogen side; (B) view from the carboxy side of the
complex; the ninth coordination site of the dysprosium center remains
accessible for one additional coordinating atom, e.g. water oxygen.
only found in the immediate vicinity of the spin-label sites ((2
residues), indicating that the conformation of these mutants is
very close to that of the wild-type. For the coupling reaction
freshly reduced, uniformly 15N-labeled solutions of the ubiquitin
mutants (80 µM, 10 mM phosphate, pH 6.0) were reacted with
a 2.5-fold excess of LCT. The progress of the coupling reactions
could be monitored conveniently by ESI-MS spectroscopy,
which shows the addition of 733 Da to the protein mass. For
both mutants, the reactions went to completion after 75 min,
and pure protein samples were obtained by ultrafiltration with
excess reaction buffer (3 × 12 mL 10 mM phosphate, pH 6.0
or 5.0).
1
Figure 4. Regions of H-15N HSQC spectra (14.1 T) of (A) Ubi-S57C
and (B) Ubi-K6C, both tagged with [Dy(M8)] at different temperatures
(blue: 288 K, green: 298 K, orange: 313 K, red: 323 K) and tagged with
diamagnetic [Lu(M8)] (black: 298 K).
be detected, whereas only 24 resonances in the vicinity of S57
were broadened beyond detection in this comparatively small
protein. The PCS have a pronounced temperature dependence,
as paramagnetically shifted resonances move toward the dia-
magnetic position when increasing the temperature from 288
to 323 K (Figure 4). Thus at higher temperatures, the apparent
magnitude of the susceptibility tensor becomes smaller, which
is presumably caused by increased flexibility of the linker. A
similar temperature behavior has been observed for Calbindin
by Bertini and co-workers and used for assignments.6 We have
used this strategy to obtain initial assignments for a number of
strongly shifted, isolated peaks. On the basis of these data, an
initial magnetic anisotropy tensor and dysprosium position were
calculated, and then iteratively used to assign more of the shifted
peaks. The final assignments were validated by data from 15N-
edited 3D NOESY and TOCSY spectra. Although not necessary
for assignment, we have also successfully recorded other
standard triple resonance 3D experiments (HNCA, HNCO, etc.),
which demonstrate that classical assignment strategies could be
used for more difficult cases. As suggested by Otting and co-
workers,19 a further assignment strategy could consist in the
For obtaining a diamagnetic reference, [Lu(M8-SPy)] was
also coupled to both mutants by the same procedure. Intensities
and line widths of these reference samples were very similar to
those of the untagged proteins, excluding a destabilization by
the label. A comparison of the chemical shifts of the [Lu(M8)]
coupled to the respective untagged mutant proteins in the
presence of reducing agent TCEP (1 mM) showed that perturba-
tions by the label only occur in its immediate vicinity. The
perturbation of amide resonances was significantly smaller than
the observed PCS, i.e. <20 ppb for 1HN (<100 ppb for 15N) for
all residues that are further away than ∼13 Å from the cysteine
Cꢀ position. Virtually all amides within this radius were bleached
out in the dysprosium-labeled mutants. Thus for practical
convenience, untagged protein may also be usable as a dia-
magnetic reference.
1
PCS and RDC Measurements. The H-15N HSQC spectra
(Figures 3 and 4) of both ubiquitin mutants tagged with
[Dy(M8)] show significant shifts for all 1H and 15N resonances
amounting to PCS of >5 ppm in extreme cases. For ubiquitin
S57C-[Dy(M8)] (Figure 3), a total of 46 NH resonances could
9
14764 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

DOTA-M8: Lanthanide Chelating Tag for PCS NMR
A R T I C L E S
Table 1. Magnetic Susceptibility Tensorsa for Ubiquitin Induced by [Dy(M8)]
protein
effect
temp/K
∆ꢁax/10-31 m3
∆ꢁrh/10-31 m3
Qb
X/Åc
Y/Åc
Z/Åc
r/Åd
Ubi-S57C
Ubi-S57C
Ubi-S57C
Ubi-S57C
Ubi-S57C
Ubi-S57C
Ubi-S57C
PCS
PCS
PCS
PCS
288
298
313
323
298
323
298
3.2
2.8
1.9
1.6
3.5
2.4
2.1
1.3
1.7
1.1
0.9
1.7
1.0
0.6
0.08
0.07
0.10
0.13
0.12
0.12
0.25
58.1
59.0
60.0
60.2
64.5
64.3
-80.4
-80.0
-78.9
-78.8
-79.7
-81.3
17.0
16.5
14.7
14.8
16.3
14.5
7.9
7.3
5.8
5.9
8.4
6.8
PCS seconde
PCS seconde
RDC
Ubi-K6C
Ubi-K6C
Ubi-K6C
Ubi-K6C
Ubi-K6C
PCS
PCS
PCS
PCS
RDC
288
298
313
323
298
4.0
4.4
2.6
2.0
2.2
1.4
1.9
0.9
0.8
1.2
0.12
0.11
0.13
0.12
0.24
37.1
36.6
37.7
37.9
-80.3
-81.5
-80.6
-79.9
3.7
5.6
4.8
3.6
7.2
9.1
7.8
6.6
calbindinf
PCS
300
3.5
2.1
a Calculated by fitting the experimental data to the first model of the NMR structural ensemble (PDB-code 1D3Z). b Q-factor calculated as
root-mean-square deviation between measured and predicted PCS (RDCs) divided by the root-mean-square of the measured PCS (RDCs). c Position of
d
the paramagnetic center in coordinates of the PDB file. Distance of paramagnetic center from the position of the Cꢀ position of S57 or K6 in the PDB
file. e Data corresponding to the minor spectral species (see text). f Amplitudes of the susceptibility tensor of dysprosium-loaded calbindin6 are shown for
comparison.
use of several different paramagnetic lanthanide ions. For both
assigned ubiquitin mutants, very different PCS are observed,
which correspond to large differences in tensor orientations and
dysprosium positions relative to the protein. In the case of S57C
all resonances are shifted upfield (Figures 3 and 4A), indicating
that all protein atoms are located within a single tensor cone.
For K6C, however, the sign of the PCS alternates (Figure 4B)
since the magic angle intersects the protein.
The final tensor parameters and dysprosium positions were
obtained from a fit of the PCS at each temperature to ubiquitin’s
NMR structure29 (PDB-code 1D3Z) and are listed in Table 1.
The excellent quality of the fit is evident from the close
agreement between measured and predicted data (Figure 5A and
B) and the corresponding low Q-factors of 8-13% (Table 1).
The axial components of the anisotropy tensor amount to 2.8
× 10-31 m3 for the S57C [Dy(M8)]-tagged ubiquitin mutant at
298 K and to 4.4 × 10-31 m3 in the case of the K6C mutant.
These values are similar in size to dysprosium-loaded calbindin6
and would correspond to a well detectable PCS of g0.07 ppm
even at a distance of about 70 Å in the axial direction. Compared
to the two-point-attachment LCTs, the position of the metal
center in the single-point-attachment M8-LCT is less predictable.
However, it is noted that the metal center was always found at
a distance of 6-9 Å from the position of the Cꢀ atom of the
amino acid that was mutated to cysteine (Table 1). The fit of
the position (three parameters) is rather well-defined from the
high number of measured PCS values (92). A more serious
problem is the unknown flexibility of the linker, which is
expected to cause residual mobility of the M8 moiety relative
to the protein. We attribute the reduction of the PCS at higher
temperature to an increase in this mobility and hence increased
Figure 5. Comparison of experimental PCS at 14.1 T for (A) Ubi-S57C
averaging (see above). In the fit procedure, the mobility is
absorbed as a scaling factor (order parameter) within the tensor
amplitudes and angles, and we attribute the observed variations
in the tensor amplitudes of the S57C and K6C mutants to this
effect. More elaborate fitting procedures using an ensemble of
and (B) Ubi-K6C with predictions according to the NMR structure29 (PDB-
access code: 1D3Z). Filled (open) symbols correspond to experimental
(predicted) PCS with color coding: 323 K (red), 298 K (green), 288 K (blue).
label conformers are possible but are expected to yield less
robust results due the higher number of parameters.
In accordance with this strong PCS we also observed large
residual dipolar couplings (RDCs) that are induced by the partial
alignment of the paramagnetic tag in the B0-field. At 18.8 T
and 298 K, the largest RDCs exceed 20 Hz for the S57C mutant
(Figure 6). The total magnitude of the RDC-derived susceptibil-
ity tensors (Table 1) is reduced to 81% (S57C) and 52% (K6C)
(29) Cornilescu, G.; Marquardt, J. L.; Ottiger, M.; Bax, A. J. Am. Chem.
Soc. 1998, 120, 6836–6837.
(30) Sass, J.; Cordier, F.; Hoffmann, A.; Cousin, A.; Omichinski, J. G.;
Lowen, H.; Grzesiek, S. J. Am. Chem. Soc. 1999, 121, 2047–2055.
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14765

A R T I C L E S
Ha¨ussinger et al.
bond. Efforts are currently underway to further reduce the
conformational freedom in the linker by chemical modifications.
Application to Urea-Denatured Ubiquitin. One distinctive
feature of the M8 ligand is its extremely high affinity toward
lanthanide ions. Although hard to determine exactly, it should
be similar or higher than that of DOTA having a pK > 25. Thus,
M8 metal complexes are extremely stable. As an application
we show detection of PCS under the harsh conditions of protein
denaturation. Figure 7 (left) depicts the 1H-15N HSQC of
ubiquitin mutant K6C at 8 M urea and pH 2.5 under diamag-
netic, i.e. without attached M8, and paramagnetic, i.e. complexed
to [Dy(M8)], conditions. Clearly visible is the disappearance
of many amino acid resonances in the vicinity of C6 by PRE,
which affects all residues from Q2 up to V17 in the first
ꢀ-hairpin of the native structure. However, for residues E18 to
A28 at the end of the second ꢀ-strand and at the beginning of
the native R-helix strong shift changes induced by [Dy(M8)]
of up to 0.2 ppm are observed. As expected for PCS, the shift
changes are collinear in 1HN and 15N and decrease in the
C-terminal direction with increasing distance from C6. This
indicates that this part of ubiquitin is in close contact to re-
sidue C6 under our unfolding conditions and that the relative
orientations between the affected residues and residue 6 have
nonisotropic distributions; otherwise the PCS would average
to zero. This is consistent with the finding that the first ꢀ-hairpin
of ubiquitin is already formed to 10-25% under urea-denaturing
conditions,34 since hairpin formation reduces the distance
between residue C6 and the beginning of the R-helix and
restricts the freedom of their relative orientation (Figure 7, right).
1
Figure 6. Comparison of experimental DHN RDCs for Ubi-S57C with
predictions according to the 1D3Z NMR structure (298 K, 18.8 T).
relative to the PCS-derived tensors. The linear correlation30
between the irreducible tensor elements for the RDC-derived
and PCS-derived tensor elements is better than 90% for both
mutants at 298 K, whereas the Euler angles agree within about
10° (data not shown). We attribute the reduction of the RDC
susceptibility tensors relative to the PCS tensors to the remaining
flexibility of the cysteinamine-linker. As pointed out by Bertini
and co-workers,31,32 the long-range nature of PCS makes them
much less sensitive than RDCs to both local mobility and local
structural inaccuracy. Thus, if the protein locally diffuses around
an axis without changing the position and orientation of the
paramagnetic center, the orientation of the short-range RDC
internuclear dipolar vectors changes much more than the
orientation of the electron-nuclear PCS dipolar vectors.
Despite this residual flexibility, the size of the measured PCS
and RDC is unprecedented for a single-point-attachment LCT
and resembles the paramagnetic effects found in proteins that
contain metal binding site(s) or the doubly tethered LCT
described by Ubbink.33 Besides the high stereochemical rigidity
of the M8 framework, we attribute the strong paramagnetic
effects to the steric bulk and to interactions between the polar
(carbonyl) or hydrophobic (methyl) face of M8 (Figure 2) and
the protein surface, which reduce the mobility of the tag.
In addition to the main species of paramagnetically shifted
resonances, a second shifted species is observable at 298 K for
both the K6C (marked by green asterisks in Figure 3) and the
S57C complex, which amounts to 15-20% of the main spectral
species. This second set of resonances was assigned for S57C.
The amplitude and rhombicity of the respective susceptibility
tensor are comparable to the main species (Table 1), but the
tensor is rotated by about 78° (data not shown), and the position
of the metal center is shifted by 5.5 Å. When heating the sample
to 323 K, the relative contribution of the second species
increases to about 50%. The effect is fully reversible on cooling
with a time constant for reaching the equilibrium at 298 K on
the order of several hours. We speculate that this slow rate is
caused by the cis-trans isomerization of the M8-SPy peptide
Discussion
We have developed a new LCT, M8-SPy, which has a highly
rigid chelating scaffold even at elevated temperatures. The
conformational stability is in marked contrast to its nonmethy-
lated parent compound, DOTA-SPy, which suffers from ex-
change between different ring conformers leading to exchange
broadening of its ring and side arm resonances and a reduction
of the paramagnetic susceptibility. Similar effects are expected
for other nonstabilized DOTA-based LCTs. Besides conforma-
tional rigidity, the second desired feature of an ideal LCT is a
stable metal position with respect to the protein. This is achieved
to a good extent for M8-SPy by its bulky nature (cf. Figure 2)
and its comparatively short linker, which results in steric
hindrance and presumably attractive interactions with the protein
surface. The PCS observed for two ubiquitin mutants tagged
with [Dy(M8)] exceed the PCS previously reported2,18,20 for
single attachment point LCTs by a factor of 2 or more.
Comparable PCS have only been described in metal-binding
proteins such as calbindin35 or in a two-point-attachment LCT.33
Lanthanide complexes of M8 have extremely high affinity
and chemical, physical, and conformational stability. This
contrasts with most metal-binding proteins, and all other
described single-point-attachment LCTs, which due to their
relatively low metal affinities readily exchange the bound metal
with buffer. The quasi-covalent nature of the lanthanide chelate
allows the use of M8 under extreme chemical or physical
conditions, such as those used for protein denaturation, or where
it is undesirable that buffer (EDTA) or protein react with excess
lanthanide ions. In particular, this may be useful for mechanistic
(31) Banci, L.; Bertini, I.; Huber, J. G.; Luchinat, C.; Rosato, A. J. Am.
Chem. Soc. 1998, 120, 12903–12909.
(32) Bertini, I.; Kursula, P.; Luchinat, C.; Parigi, G.; Vahokoski, J.;
Wilmanns, M.; Yuan, J. J. Am. Chem. Soc. 2009, 131, 5134–44.
(33) Keizers, P. H.; Saragliadis, A.; Hiruma, Y.; Overhand, M.; Ubbink,
M. J. Am. Chem. Soc. 2008, 130, 14802–12.
(34) Meier, S.; Strohmeier, M.; Blackledge, M.; Grzesiek, S. J. Am. Chem.
Soc. 2007, 129, 754–5.
(35) Bertini, I.; Kowalewski, J.; Luchinat, C.; Parigi, G. J. Magn. Reson.
2001, 152, 103–8.
9
14766 J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009

DOTA-M8: Lanthanide Chelating Tag for PCS NMR
A R T I C L E S
1
Figure 7. Detection of PCS in unfolded K6C ubiquitin. (Left) H-15N HSQC of K6C ubiquitin in 8 M urea, 10 mM glycine-HCl, pH 2.5, 298 K, 18.8 T.
The diamagnetic spectrum (200 µM K6C ubiquitin without attached M8, 1 mM TCEP) is shown in blue, the paramagnetic spectrum (100 µM [Dy(M8)]-K6C
ubiquitin) is shown in red. Resonances that experience strong paramagnetic effects are labeled with assignment information: PCS (green), PRE (black).
(Right) First half of ubiquitin’s topology showing the detectable PCS from position C6 toward residues in the range between E18 and A28.
PIPP.39 Susceptibility tensors were fitted to the experimental PCS
studies of calcium-binding proteins, e.g. the calcium-dependent
oligomerization of cadherins,36 where low-affinity LCTs would
and RDC data by in-house written MATLAB (The MathWorks,
Inc.) routines.
compete with the native calcium binding sites for lanthanide
and calcium ions.
Acknowledgment. We thank M. Rogowski for the expression
of ubiquitin mutants and C. Palivan for EPR measurements of
Experimental Section
[Dy(M8-SPy)]. J. Desreux, I. Bertini, and M. Ubbink are gratefully
Details of the syntheses of DOTA-SPy and M8-SPy, their
acknowledged for helpful discussions. This work was supported
metalation and protein attachment are described in the Supporting
by SNF grant 31-109712 (S.G.).
Information. Ubiquitin plasmids for S57C and K6C were prepared
using a site-directed mutagenesis kit (Stratagen). 15N-Labeled
Supporting Information Available: Details of the syntheses
ubiquitin was prepared as described previously.30
of DOTA-SPy and two diastereomeric forms of M8-SPy, their
NMR experiments were performed on mutant and wild-type
ubiquitin (10 mM phosphate, pH 6.0) on Bruker Avance DRX 800,
1
metalation and attachment to the protein, H-15N HSQC of
ubiquitin S57C-[Dy(DOTA)], 1D 1H spectrum of [Dy(DOTA-
600 or DPX 400 spectrometers equipped with a TCI cryoprobe, a
SPy)]. This material is available free of charge via the Internet
at http://pubs.acs.org.
standard BBI or QNP probe, respectively. PCS were determined
from standard 15N HSQC spectra, RDC data were extracted from
IPAP-HSQC experiments.37 NMR data were processed using the
NMRPipe software package38 and analyzed with the program
JA903233W
(36) Ha¨ussinger, D.; Ahrens, T.; Aberle, T.; Engel, J.; Stetefeld, J.; Grzesiek,
S. EMBO J. 2004, 23, 1699–708.
(38) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax,
A. J. Biomol. NMR 1995, 6, 277–293.
(39) Garrett, D. S.; Powers, R.; Gronenborn, A. M.; Clore, G. M. J. Magn.
Reson. 1991, 95, 214–220.
(37) Ottiger, M.; Delaglio, F.; Bax, A. J. Magn. Reson. 1998, 131, 373–
378.
9
J. AM. CHEM. SOC. VOL. 131, NO. 41, 2009 14767