An iminodiacetic acid based lanthanide binding tag for paramagnetic exchange NMR spectroscopy

Article abstract of DOI:10.1002/anie.201007221

All the way with IDA! Attachment of iminodiacetic acid (IDA) to a protein helix creates a rigid lanthanide binding site that can be exploited for paramagnetic NMR spectroscopy (see picture). Pseudo-contact shifts (PCSs) larger than 8 ppm are achievable wi

Full text of DOI:10.1002/anie.201007221

DOI: 10.1002/anie.201007221
Lanthanide Tags
An Iminodiacetic Acid Based Lanthanide Binding Tag for
Paramagnetic Exchange NMR Spectroscopy**
James D. Swarbrick,* Phuc Ung, Sandeep Chhabra, and Bim Graham*
When bound to proteins, paramagnetic lanthanide ions
induce a range of effects that are observable by NMR
spectroscopy, including pseudo-contact shifts (PCSs), para-
magnetic relaxation enhancements (PREs), and residual
dipolar couplings (RDCs).[1] These effects provide valuable
constraints that can expedite protein structure refinement,[2]
the analysis of protein–protein[3] and protein–ligand interac-
tions,[4] and, potentially, the study of protein dynamics and
lowly populated encounter states of protein complexes.[5]
PCSs, measurable for nuclei beyond 60 ꢀ away from some
lanthanide ions, are especially useful for NMR structural
analysis of multidomain proteins and multiprotein com-
plexes.[6] These manifest as changes in chemical shifts between
paramagnetic and diamagnetic samples, with the difference in
shifts (DdPCS) dependent on the location of the nuclear (i.e.,
15N and 1HN) spins with respect to the anisotropic magnetic
susceptibility tensor (Dc) of the metal ion:
a paramagnetic lanthanide ion. If the exchange rate is
sufficiently high within the slow-exchange NMR timescale,
cross-peaks may be observed in exchange experiments
performed with a mixture of paramagnetic and diamagnetic
metal ions, greatly facilitating correlation of paramagnetic
and diamagnetic resonances. The attractiveness of this
approach is twofold. Firstly, it does not require prior knowl-
edge of protein structure, making it a potentially much more
broadly applicable method for assigning paramagnetic NMR
spectra than the iterative procedure. Secondly, by taking
advantage of the slowly relaxing heteronucleus and its
insensitivity toward paramagnetic relaxation, 15N longitudinal
exchange spectroscopy can extend PCS detection into the
usually PRE-broadened sphere close to the metal. While this
region can be probed with weakly anisotropic metals, these
yield few short-range PCSs, which makes the calculation of
the magnetic susceptibility tensor difficult for an individual
metal. Thus far, the exchange method has only been success-
fully demonstrated for a protein incorporating a natural metal
ion-binding site (the e186/q complex of DNA polymerase
III).[10,11] All reported synthetic lanthanide-binding tags bind
metal ions too tightly to allow exchange between the bound
and unbound states within the slow-exchange NMR regime.[1]
Here we present the first example of a small synthetic
lanthanide-binding tag for which the chemical exchange is
sufficiently fast to produce definitive exchange cross-peaks,
enabling the rapid assignment of both small and extraordi-
narily large PCSs by 15N heteronuclear exchange spectrosco-
py, without recourse to a structural model.
ꢀ
ꢁ
1
3
2
DdPCS
¼
Dcaxð3 cos2 q ꢀ 1Þ þ Dcrh sin2 q cos 2f
ð1Þ
12pr3
where Dcax and Dcrh are the axial and rhombic components of
the Dc tensor, r is the distance of the metal ion from the
nuclear spin, and q and f are angles that describe the
orientation the Dc tensor with respect to the protein.[7]
Assignment of PCSs provides access to a protein-anchored,
metal-centered coordinate system that can be used as a
reference frame to pinpoint the location of other nuclear spins
by virtue of their PCSs.
Provided both a structural model and the assignment of a
diamagnetic reference are available, assignment of para-
magnetic NMR spectra can be achieved by an iterative
procedure which involves minimization of the difference
between observed and back-calculated PCSs.[8] An alternative
strategy, developed by Otting and co-workers,[9–11] involves
the recording of NMR spectra under conditions in which both
paramagnetic and diamagnetic forms of the protein are
present and interconvert through binding and dissociation of
The tagging agent, 1 (Figure 1),[12] is a hybrid of the well-
known tridentate chelator iminodiacetic acid (IDA)[13] and l-
cysteine. The thiol group permits ready attachment of the tag
to a protein through disulfide bond formation with a surface-
exposed cysteine residue. Although IDA is small and flexible,
we reasoned that the stereocenter within the ligand, com-
[*] Dr. J. D. Swarbrick, P. Ung, S. Chhabra, Dr. B. Graham
Monash Institute of Pharmaceutical Sciences, Monash University
399 Royal Parade, Parkville, Victoria (Australia)
Fax: (+61)3-9903-9543
E-mail: james.swarbrick@monash.edu
bim.graham@monash.edu
[**] We thank Prof. Gottfried Otting for stimulating discussions and
recognizing the potential of 1 to be used for NMR exchange
spectroscopy, and Dr. Marcello Tellioni for coding the ZZ exchange
pulse sequences.
Figure 1. Left: Structure of the IDA-based tagging agent 1. Right:
Representation of rigid lanthanide ion chelation by the Cys-linked IDA
ligand in combination with an Asp residue within an a helix. Aquo
ligands are expected to occupy the remaining coordination sites about
the lanthanide ion.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201007221.
Angew. Chem. Int. Ed. 2011, 50, 4403 –4406
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4403

Communications
bined with judicious positioning of the tag near an additional
diamagnetic and paramagnetic lanthanide ions were added to
UbiqA28C–IDA (ca. 70 mm) until the intensity of the
resonances of the diamagnetic and paramagnetic species
were approximately equal in the 15N HSQC spectra (approx-
imately 130 mm of each). This process was performed for a
series of different paramagnetic lanthanide ions (Tb3+, Dy3+,
Ho3+, Er3+, Tm3+, and Yb3+), using La3+ as the diamagnetic
reference (Figures S2–S6).
Outstanding quality data was obtained using mixing times
of 200–300 ms, within a short data collection time (6–12 h)
(Figure 2). From this data, an exchange rate of approximately
15 sꢀ1 was estimated (Figure S7), which is almost twenty times
higher than that found for e186,[10] and consistent with the
three-fold shorter mixing time employed herein. Even using a
1:1 metal-to-protein ratio, good quality 15Nz exchange data
were recorded using mixing times of 300–400 ms (Figures S8
and S9).
Using Nz-exchange spectroscopy, both small and large
paramagnetically shifted resonances were easy to assign
unambiguously. Typically 80–90% of the PCS data obtained
from a single exchange spectrum were accurately measured
prior to first round tensor calculations within the program
Numbat,[15] using the high-resolution RDC-refined structure
of human ubiquitin.[16] A PCS of 4.3 ppm was measured in the
15N dimension that was otherwise broadened beyond detec-
tion in the standard 15N HSQC spectrum (Figure 2). The high
quality of the PCS data (Table S1), combined with facile
assignments and large magnetic anisotropies, meant that both
the metal position and the Dc tensor were accurately
determined from a single exchange experiment for each of
the metals Tb3+, Dy3+ and Ho3+, without recourse to addi-
tional metal-to-coordinating group distance restraints or
estimation of the position from the center of the alignment
tensor (Figures S10–S12).[17] The smaller magnetic anisotro-
metal ion-coordinating residue, might be sufficient to achieve
rigid complexation of a lanthanide ion in a single isomeric
form (or fast exchange between a limited number of isomers),
a prerequisite for the generation of useful PCS data. As
detailed below, attachment of the tag to a regular a-helix
motif in human ubiquitin containing an aspartic acid residue
at the i + 4 position (Figure 1) led to observation of PCSs for
90% of the protein residues. PCSs beyond 8 ppm in magni-
tude, and for nuclei as close as 8 ꢀ to the same metal center,
were measured using a series of different lanthanide ions.
The IDA-based tag was chemically attached to an
Ala28Cys mutant of human ubiquitin (UbiqA28C-IDA),
pre-activated with Ellmanꢁs reagent.[14] Cys28 is situated on
the solvent-exposed face of the a helix and is located between
two acidic residues, Glu24 and Asp32, which provide two
potentially additional coordination sites. Titration with the
diamagnetic lanthanide ion La3+ showed slow exchange on
the NMR timescale, and a single set of resonances in the 15
N
HSQC NMR spectrum after addition of excess metal, con-
sistent with a single-metal-bound isomer (Supporting Infor-
mation, Figure S1). Whilst significant resonance perturba-
tions were observed, these were for residues proximal to the
tag, particularly on the solvent exposed face of the helix
(Figure S1). Moreover, 15N-edited NOESY data showed that
the secondary structure is well maintained along the helix,
indicating that attachment of the tag and metal coordination
does not disrupt the native structure of the helix.
Addition of Dy3+ to UbiqA28C–IDA produced large
PCSs in slow exchange. This, combined with weak metal
binding from the La3+ titration (Kd ꢁ 10 mm), prompted us to
carry out a series of 15NZ-exchange experiments to optimize
the single-mixing time period for exchange-based assign-
ments.[10] For these, aliquots of an equimolar mixture of
Figure 2. 15Nz-exchange spectra for UbiqA28C–IDA. Left panel: Superposition of the 15N fast-HSQC spectrum (blue) of a mixture of 70 mm 15N-
labeled UbiqA28C–IDA and 130/130 mm La3+/Dy3+ with the 300 ms single-mixing 15Nz-exchange spectrum (red). Arrows highlight the extent and
direction of the observed PCS. The PCS for Gln41 is observed only in the 15N dimension. Right panel: Expanded region (boxed in left panel) of the
150 ms two-mixing 15Nz-exchange spectrum (red). Arrows highlight PCSs (values shown in parentheses) that are only observed in the 15N
dimension and are otherwise not observed in the single-mixing 15Nz experiment.
4404 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 4403 –4406

pies of Tm3+, Yb3+, and Er3+ (and thus smaller PCSs) led to
deviations from this position and instability in the Dc tensor
calculation. The metal position and Dc tensor for these
lanthanides were therefore each refined in turn, using
simultaneous refinement with the PCS data from Dy3+
within Numbat.[17] Final axial and rhombic components,
Euler angles, and metal position are presented in Table S2,
while correlations of the observed vs. calculated PCSs for all
tested metals are shown in Figure 3.
and co-workers reported significant deviation for the lantha-
nide-binding e186/q complex,[11] attributing this in part to the
presence of multiple metal-binding sites.
The Dc tensors show that the lanthanide ion is located
approximately 2.0 ꢀ from the side-chain dO of Asp24
ꢀ
(Figure S12). This Ln O distance is within the range typically
observed for lanthanide ions complexed by proteins or
organic ligands,[18] and is suggestive of a preference for
additional coordination of the ions by Asp32 over Glu24, both
located four residues away from the 1-tagged Cys
residue.
PCSs for the same residues measured with
different lanthanide ions lay along straight lines
in superimposed 15N HSQC spectra (Figure S14)
and displayed either good correlation (Tb3+/
Dy3+) or good anticorrelation (either Dy3+ or
Tb3+ with Tm3+, data not shown), indicating an
equivalent coordination geometry for each of
these lanthanides. The principal axes of the
susceptibility tensors were similarly oriented
(Figures S10) and in all cases, the axial compo-
nent of the anisotropy tensor, Dcax, lay approx-
imately perpendicular to the helix axis (Fig-
ure S11). As most of the protein lay within a
single lobe, the PCSs for each lanthanide tended
to shift in one direction (Figure 3). For La3+/
Dy3+-loaded UbiqA28C–IDA, Dcax was deter-
mined to be 32.4 ꢂ 10ꢀ32 m3, which would be
expected to yield well-detectable PCSs
(ꢂ0.08 ppm) at distances in excess of 60 ꢀ in
the axial direction.
The measured and calculated 1HN PCSs
Figure 3. Left panels: Correlations between calculated 1HN PCSs and those measured
in a single-mixing 15Nz-exchange experiment for the backbone amides of UbiqA28C–
IDA in the presence of an equimolar mixture of the specified paramagnetic lanthanide
ion and La3+. Right panels: Correlations between calculated PCSs and those measured
in a two-mixing, out-and-back 15Nz-exchange experiment for the amides of UbiqA28C–
IDA in the presence of equimolar La3+/Dy3+ or La3+/Tb3+. Additional close-range 15N
PCSs that were not observed in 15N HSQC experiments are circled.
correlated very well for spins close to the
lanthanide centers (Figure 3) and indicated that
large-amplitude tag motion was probably mini-
mal in Ln3+-loaded UbiqA28C–IDA (for a
mobile/flexible tag, PCSs for proximal spins are
much more sensitive to erroneous PCS fitting
due to an rꢀ3 averaging effect). This observation
1
encouraged us to measure DHN RDC data to
Given the success of the single-mixing exchange experi-
ment for rapid assignments, we conducted a two-mixing
period “out-and-back” 15Nz exchange experiment[11] for the
La3+/Dy3+- and La3+/Tb3+-coordinated UbiqA28C–IDA, to
enable the detection of substantially PRE-broadened nuclear
spins located close to the metal centers. Additional PCSs from
amides as close as 8 ꢀ to the strongly paramagnetic Dy3+ and
Tb3+ ions were measured in the 15N dimension (Figures 2 and
S13). These were in excess of 8 ppm, with the largest being d =
16.2 ppm for a PCS detected just above the noise threshold
for Tb3+. With the inclusion of these large PCSs, the
correlation between the measured PCSs and those back-
calculated from the alignment tensor remained very good
(Figure 3). The calculated metal position was unchanged, and
only a minor change (ca. 10%) in the magnitude of the Dc
tensor was observed (Table S2). This indicates that the
structure of the helix is essentially unperturbed by the
addition of the tag and/or metal, in agreement with the
previously collected 3D 15N NOESY data. In contrast, Otting
evaluate the tag motion in more detail. At 18.8 T, RDC values
up to 20 Hz were measured for La3+/Tb3+ bound to
UbiqA28C–IDA (Figure S15). Back-calculation of the
RDCs from the Dc tensors,[14,19,20] assuming an order param-
eter of 0.9, revealed that the observed RDCs were, on
average, 80% of the size of those calculated (Figure S16). The
orientations of the principal axes of alignment and the Dc
tensor were in very close agreement (Figure S17) as expected
for rigid metal coordination. Furthermore, the RDC-derived
axial and rhombic components of the Dc tensor for Tb3+-
loaded UbiqA28C–IDA (Table S2) were 80% and 87% of
the PCS-derived values, respectively, also indicative of highly
rigid lanthanide coordination. These values are comparable to
those obtained by Grzesiek,[19] Ubbink[20] and co-workers
using extremely tight metal-binding DOTA-based
tags
(DOTA = 1,4,7,10-tetraazacyclodecane-1,4,7,10-tetra-
acetic acid). The larger anisotropies observed in these latter
systems are therefore most likely attributable in part to
coordination geometry and ligand field effects, rather than
Angew. Chem. Int. Ed. 2011, 50, 4403 –4406
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4405

Communications
Biochemistry 2010, 49, 6846 – 6855; b) G. Pintacuda, A. Y. Park,
M. A. Keniry, N. E. Dixon, G. Otting, J. Am. Chem. Soc. 2006,
128, 3696 – 3702; c) I. Diaz-Moreno, A. Diaz-Quintana, M. A.
De La Rosa, M. Ubbink, J. Biol. Chem. 2005, 280, 18908 – 18915.
[4] See for example: a) T. Zhuang, H. S. Lee, B. Imperiali, J. H.
Prestegard, Protein Sci. 2008, 17, 1220 – 1231; b) M. John, G.
Pintacuda, A. Y. Park, N. E. Dixon, G. Otting, J. Am. Chem. Soc.
2006, 128, 12910 – 12916.
just differences in rigidity. Our results clearly demonstrate
that tight binding is not an essential requisite for rigid
lanthanide tagging of proteins.
In conclusion, an IDA-based lanthanide binding tag with
exceptional properties for rapid PCS determination has been
developed. Attachment of this tag to an a helix containing an
Asp residue at the i + 4 position creates a metal binding site
that complexes lanthanide ions in a rigid, yet kinetically
labile, fashion. Significant PCSs are observed for nuclei
located 8–60 ꢀ from single, strongly paramagnetic ions, which
are easily assignable with the aid of heteronuclear exchange
spectroscopy. The notion of a small, highly flexible ligand that
becomes rigid in situ upon complexation of a metal ion
represents a novel concept as compared to the traditional
approach of employing rigid, tight binding ligands for
immobilization of lanthanide ions. Given that a helices are
extremely ubiquitous and easily identified from backbone
chemical shifts alone,[21] we envisage that the methodology
described herein may potentially be transferable to the study
of other proteins. Moreover, the ability to generate PCS and
RDC data in the absence of a prior structural model could see
the use of tags such as 1 integrated with the recently reported
methodology of Baker,[22] Bax[23] and co-workers for rapid
NMR structure determination of proteins from backbone-
only data.
[5] M. A. S. Hass, P. H. J. Keizers, A. Blok, Y. Hiruma, M. Ubbink, J.
Am. Chem. Soc. 2010, 132, 9952 – 9953.
[6] See for example: a) R. R. Biekofsky, F. W. Muskett, J. M.
Schmidt, S. R. Martin, J. P. Browne, P. M. Bayley, J. Feeney,
FEBS Lett. 1999, 460, 519 – 526; b) T Saio, M Yokochi, H
Kumeta, F Inagaki, J. Biomol. NMR 2010, 46, 271 – 280.
[7] I. Bertini, M. B. L. Janik, Y. Lee, C. Luchinat, J. Am. Chem. Soc.
2001, 123, 4148 – 4188.
[8] I. Baig, I. Bertini, C. Del Bianco, Y. K. Gupta, Y.-M. Lee, C.
Luchinat, A. Quattrone, Biochemistry 2004, 43, 5562 – 5573.
[9] M. John, G. Otting, ChemPhysChem 2007, 8, 2309 – 2313.
[10] M. John, M. Headlam, N. E. Dixon, G. Otting, J. Biomol. NMR
2007, 37, 43 – 51.
[11] M. John, A. Y. Park, N. E. Dixon, G. Otting, J. Am. Chem. Soc.
2007, 129, 462 – 463.
[12] P. Siegmund, Hoppe-Seylerꢀs Z. Physiol. Chem. 1968, 349, 37 –
40.
[13] C. Kremer, J. Torres, S. Domꢄnguez, J. Mol. Struct. 2008, 879,
130 – 149.
[14] X. C. Su, B. Man, S. Beeren, H. Liang, S. Simonsen, C. Schmitz,
T. Huber, B. A. Messerle, G. Otting, J. Am. Chem. Soc. 2008, 130,
10486 – 10487.
Received: November 17, 2010
Revised: January 31, 2011
Published online: April 7, 2011
[15] C. Schmitz, M. Stanton-Cook, X.-C. Su, G. Otting, T. L. Huber, J.
Biomol. NMR 2008, 41, 179 – 189.
[16] A. Bax, N. Tjandrab, J. Biomol. NMR 1997, 10, 289 – 292.
[17] B. Man, X. C. Su, H. Liang, S. Simonsen, T. Huber, B. A.
Messerle, G. Otting, Chem. Eur. J. 2010, 16, 3827 – 3832.
[18] a) M. Nitz, M. Sherawat, K. J. Franz, E. Peisach, K. N. Allen, B.
Imperiali, Angew. Chem. 2004, 116, 3768 – 3771; Angew. Chem.
Int. Ed. 2004, 43, 3682 – 3685; b) D. Parker, R. S. Dickens, H.
Puschmann, C. Crossland, J. A. K. Howard, Chem. Rev. 2002,
102, 1977 – 2010.
[19] D. Hꢅussinger, J. Huang, S. Grzesiek, J. Am. Chem. Soc. 2010,
131, 14761 – 14767.
[20] P. H. J. Keizers, A. Saragliadis, Y. Hiruma, M. Overhand, M.
Ubbink, J. Am. Chem. Soc. 2008, 130, 14802 – 14812.
[21] G. Cornilescu, F. Delaglio, A. Bax, J. Am. Chem. Soc. 1998, 120,
6836.
[22] S. Raman, O. F. Lange, P. Rossi, M. Tyka, X. Wang, J. Aramini,
G. Liu, T. Ramelot, A. Eletsky, T. Szyperski, M. Kennedy, J.
Prestegard, G. T. Montelione, D. Baker, Science 2010, 327, 1014 –
1018.
[23] Y. Shen, O. Lange, F. Delaglio, P. Rossi, J. M. Aramini, G. Liu, A.
Eletsky, Y. Wu, K. K. Singarapu, A. Lemak, A. Ignatchenko,
C. H. Arrowsmith, T. Szyperski, G. T. Montelione, D. Baker, A.
Bax, Proc. Nat. Acad. Sci. USA 2008, 105, 4685 – 4690.
Keywords: lanthanides · NMR spectroscopy · protein structure ·
.
pseudo-contact shifts · residual dipolar couplings
[1] a) I. Bertini, C. Luchinat, G. Parigi, Prog. Nucl. Magn. Reson.
Spectrosc. 2002, 40, 249 – 273; b) G. Otting, Annu. Rev. Biophys.
2010, 39, 387 – 405; c) P. H. J. Keizers, M. Ubbink, Prog. Nucl.
Magn. Reson. 2011, 58, 88 – 96; d) F. Rodriguez-Castaꢃeda, P.
Haberz, A. Leonov, C. Griesinger, Magn. Reson. Chem. 2006, 44,
10 – 16.
[2] See for example: a) M. Allegrozzi, I. Bertini, M. B. L. Janik,
Y. M. Lee, G. Liu, C Luchinat, J. Am. Chem. Soc. 2000, 122,
4154 – 4161; b) B. Simon, T. Madl, C. D. Mackereth, M. Nilges,
M. Sattler, Angew. Chem. 2010, 122, 2011 – 2014; Angew. Chem.
Int. Ed. 2010, 49, 1967 – 1970; c) R. C. Page, S. Lee, J. D. Moore,
S. J. Opella, T. A. Cross, Protein Sci. 2009, 18, 134 – 146; d) L.
Banci, I. Bertini, G. Cavallaro, A. Giachetti, C. Luchinat, G.
Parigi, J. Biomol. NMR 2004, 28, 249 – 261.
[3] See for example: a) P. H. J. Keizers, B. Mersinli, W. Reinle, J.
Donauer, Y. Hiruma, M. Overhand, R. Bernhardt, M. Ubbink,
4406 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 4403 –4406