Solution structure and dynamics of
Hui Wanga,1, Qi Zhangb,1, Bin Caic,1, Hongyan
Hou-Ming Wub,*, H
a Department of Chemistry and Open Laboratory of Chemical Biology, Th
b State Key Laboratory of Bio-organic and Natural Product
Chinese Academy of Sciences, Sh
c Department of Chemistry, Fudan Unive
Received 27 October 2005; revised 5 Decem
ne 9
ristia
rofibrillary tangles and extracellular amyloid plaques. Human
Metallothionein (MT) is a kind of small proteins (typically
6–7 kDa) with an array of conserved 20 cysteines and contains
polynuclear metal–sulfur coordination sites formed by metal
involved in the growth inhibitory activity [13,14]. Specifically,
MT-3 has been suggested to participate in the utilization of
FEBS Letters 580 (2006) 795–800correlation spectroscopyions with d10 configuration [1]. In mammalian MTs, the two
major metallothionein isoforms (MT-1 and MT-2), which are
zinc as a neuromodulator since it is massively expressed in neu-
rons that sequester zinc in their synaptic vesicles [15].
Three-dimensional structures of MT isoforms have been re-
ported previously, mostly on MT-1 and MT-2 [16–23]. Due to
lack of well-defined secondary structure elements, their overall
fold is dictated mostly by a clustered network of metal–thiolate
bonds among the sulfur atoms of the cysteine residues and the
metal ions, usually represented by zinc, copper, and cadmium
[24]. It was shown that MT-3, similar to the MT-1 and MT-2,
comprises two domains each wrapping around a metal–thio-
late cluster – M4S11 in the C-terminal a-domain and M3S9 in
the N-terminal b-domain by 113Cd NMR spectroscopy
[25,26]. Functional differences between MT-3 and the other
mammalian MT isoforms may implicate some structural
and/or dynamic differences. Therefore, it would be of interest
Abbreviations: AD, Alzheimer’s disease; GST, glutathione-S transfer-
ase; HMQC, heteronuclear multiple-quantum coherence; HSQC, het-
eronuclear single-quantum coherence; MTs, metallothioneins; MT-3,
metallothionein-3; NFT, neurofibrillar tangles; NOE, nuclear Overha-
user effect; NOESY, nuclear Overhauser enhancement spectroscopy;
RMSD, root mean square deviation; SP, senile plaques; TOCSY, total
*Corresponding authors. Fax: +852 2857 1586 (H. Sun).
E-mail addresses: hmwu@mail.sioc.ac.cn (H.-M. Wu),
hsun@hkucc.hku.hk (H. Sun).
1 These authors contributed equally to this work.neuronal growth inhibitory factor, classified as metallothionein-
3 (MT-3), was found to be related to the neurotrophic activity
promoting cortical neuron survival and dendrite outgrowth in
the cell culture studies. We have determined the solution struc-
ture of the a-domain of human MT-3 (residues 32–68) by multi-
nuclear and multidimensional NMR spectroscopy in combination
with the molecular dynamic simulated annealing approach. The
human MT-3 shows two metal–thiolate clusters, one in the N-
terminus (b-domain) and one in the C-terminus (a-domain).
The overall fold of the a-domain is similar to that of mouse
MT-3. However, human MT-3 has a longer loop in the acidic
hexapeptide insertion than that of mouse MT-3. Surprisingly,
the backbone dynamics of the protein revealed that the b-domain
exhibits similar internal motion to the a-domain, although the
N-terminal residues are more flexible. Our results may provide
useful information for understanding the structure–function
relationship of human MT-3.

2006 Federation of European Biochemical Societies. Published
by Elsevier B.V. All rights reserved.
Keywords: Dynamics; Growth inhibitory factor;
Metallothionein-3; NMR; Structures; Zinc
1. IntroductionAvailable onli
Edited by Ch
Abstract Alzheimer’s disease is characterized by progressive
loss of neurons accompanied by the formation of intraneural neu-0014-5793/$32.00
2006 Federation of European Biochemical Societies. Pu
doi:10.1016/j.febslet.2005.12.099human metallothionein-3 (MT-3)
Lia, Kong-Hung Szea, Zhong-Xian Huangc,
ongzhe Suna,*
e University of Hong Kong, Pokfulam Road, Hong Kong, PR China
s Chemistry, Shanghai Institute of Organic Chemistry,
anghai 200032, PR China
rsity, Shanghai 200433, PR China
ber 2005; accepted 22 December 2005
January 2006
n Griesinger
expressed in most organs, play a role in the homeostasis of
the essential trace metal (Zn2+ and Cu+) and in the heavy metal
detoxification (e.g. Cd2+ and Hg2+), and even in radical scav-
enging and stress response [2]. In the past decade, two tissue-
specific isoforms were discovered, MT-3 in the central nervous
system and MT-4 in epithelial cells [3].
Alzheimer’s disease (AD), a senile dementia, comprises dys-
function of neurons in brains, in which neurofibrillar tangles
(NFT) and senile plaques (SP) are the major neuropathological
characteristics [4]. It has been found that AD brain extract
shows enhanced neurotrophic activity promoting cortical neu-
ron survival and dendrite outgrowth in the cell culture studies
[5–8]. The enhanced activity was shown to be due to the down-
regulation of a neuronal growth inhibitory factor, which was
isolated from normal human brains, and was subsequently
shown to belong to the family of metallothioneins (MTs)
and designated as MT-3 [9,10]. Its characterization revealed
a metallo-protein of 68 amino acids containing four copper
and three zinc ions with its primary structure exhibiting
approximately 70% sequence identity with those of mamma-
lian metallothioneins (MT-1/MT-2 isoforms: 61 or 62 amino
acids), including the preserved array of 20 cysteine residues
[11]. In contrast to the amino acid sequences of mammalian
MT-1/MT-2, MT-3 contains two conserved insertions: a single
Thr in the N-terminal region and a glutamate-rich hexapeptide
(EAAEAE) near the C-terminus (Fig. 1). Additionally, all
known primary sequences of mammalian MT-3 contain the
conserved Cys6-Pro-Cys-Pro9 motif, which was shown to beblished by Elsevier B.V. All rights reserved.

added slowly into the protein solution under argon. The solution
was then titrated with 100 mM Tris base to pH 8.5 and then treated
T3
r iso
796 H. Wang et al. / FEBS Letters 580 (2006) 795–800with chelex-100 to remove excess metals.
2.2. NMR spectroscopy
All the NMR experiments were performed on a Bruker Avance 600
(and a Varian 600 Inova) spectrometer (except for 113Cd NMR), oper-
ating at a proton frequency of 600.13 MHz. Resonances of the Cd-
reconstituted human MT-3 were assigned using spectra acquired with
about 2 mM 15N-labeled hMT-3 in 15 mM sodium phosphate, pH 7.3,
90% H2O/10% D2O with addition of 0.02% (w/v) sodium azide. The
temperature of all the experiments was typically set to 298 K. Solvent
suppression was accomplished using the 3-9-19 WATERGATE pulse
sequence [30]. The backbone and side-chain resonances of MT-3 were
15to study the structural and dynamic properties of MT-3, in
spite of the structure of the a-domain of mouse MT-3 being re-
ported [22]. We have previously examined the uptake and re-
lease of Zn2+ (and Cd2+) ions from the human MT-3 [27]. In
the present study, we have determined the three-dimensional
structure of the a-domain of human Cd7-MT-3 by multinu-
clear and multidimensional NMR spectroscopy together with
molecular dynamic simulated annealing. The backbone
dynamics of both domains were also examined by 15N-relaxa-
tion experiments.
2. Materials and methods
2.1. Expression and purification of human Cd7-MT-3
The DNA sequence of human Cd7-MT-3 was cloned into a pGEX-
4T-2 (Pharmacia Biotech) vector. As described in our previous paper
[27], the resultant plasmid was transformed into Escherichia coli
BL21 and overexpressed as a glutathione-S transferase (GST)-tagged
fusion protein. The protein was purified by glutathione–Sepharose
4B affinity column (Amersham Pharmacia Biotech) with some modifi-
cations in the instructions [28], and was cleaved by thrombin at the site
between GST and the target sequences leaving additional Gly-Ser
dipeptide at the N-terminus. The protein was further purified and de-
salted by size-exclusion chromatography. Isotopically enriched MT-3
was prepared in the M9 minimal medium containing 1 g/L 15N-ammo-
nium chloride (15NH4Cl, Cambridge Isotopes Laboratory). The
113Cd
reconstitution of human MT-3 was carried out using a modified meth-
od described previously [29]. Briefly, the sample was incubated with ex-
cess 1,4-dithiothreitol (DTT) and lowered its pH to ca. 1–2 to remove
the metals. The sample was then desalted with gel filtration chromatog-
raphy at pH 2.0. Known amount of 113Cd2+ (8 mol equivalents) were
Fig. 1. Sequence alignment of human MT2A, human MT3, mouse M
conserved cysteines are shown in bold. The residues conserved in all fou
human MT-3 sequence.assigned by a combination of 2D N-decoupled total correlation spec-
troscopy (TOCSY), 15N-decoupled nuclear Overhauser enhancement
spectroscopy (NOESY), 1H–15N heteronuclear single-quantum coher-
ence (HSQC) and 3D 15N-edited-HSQC-TOCSY, 15N-edited-HSQC-
NOESY. Intramolecular distance restraints were obtained from 2D
15N-decoupled NOESY and 3D 15N-edited-HSQC-NOESY spectra
both using a mixing time of 150 ms. All the NMR data were processed
using nmrPipe and nmrDraw [31], and analyzed by Sparky software
[32].
The 113Cd NMR experiments were performed on a Bruker DRX-500
spectrometer. All the spectra were obtained at 310 K with ca. 1 mM
113Cd-reconstituted hMT-3 in 15 mM phosphate, pH 7.3, 90% H2O/
10% D2O containing 0.02% (w/v) sodium azide. The
113Cd NMR spec-trum was recorded using an inverse gated broad-band proton decou-
pling. A spectral width of 26667 Hz, an acquisition time of 0.31 s
and a pulse repetition rate of 0.5 s were used. The 113Cd resonances
are given in parts per million relative to an external standard
Cd(ClO4)2. Metal-to-ligand (
113Cd and cysteine residues) connectivities
were determined by 2D 1H–113Cd heteronuclear multiple-quantum
coherence (HMQC) NMR experiments. Each transient of 2048 com-
plex points was accumulated with 128 scans for 256 increments in t1
dimension. 3J(1H,113Cd) was varied over the range of 20–70 Hz.
The 1H and 15N chemical shifts of the a-domain have been deposited
in BioMagResBank (BMRB Accession No.: 6909).
2.3. Structure calculations
The structure of the a-domain in human MT-3 was calculated iter-
atively by CYANA (version 2.0) [33]. Inter-proton distance restraints
were derived from the 2D 15N-decoupled NOESY and 3D 15N-edi-
ted-HSQC-NOESY with mixing times of 150 ms using the automated
nuclear Overhauser effect (NOE) assignment strategy followed by a
manual check. NOE intensities and chemical shifts were extracted
using the Sparky3.0 [32], and served as an input for the CYANA.
The CYANA performs automated NOE-assignment, distance calibra-
tion of NOE intensities, removal of covalently fixed distance restraints,
structure calculation with torsion angle dynamics, and automatic NOE
upper distance limit violation analysis [34]. The four Cd2+ ions in the
a-domain were defined in CYANA as new patch residues (CD) con-
taining only the metal ions, and were linked to the end of the polypep-
tide chain of special linker residues (LLM). Information about the
cadmium–sulfur bond lengths and cadmium–CB distances in the me-
tal-ion-to-ligand connectivities, as well as the distances of sulfur–sulfur
of the metal cluster geometry, was taken from the 3D structure of rat
liver [Cd7]-metallothionein-2a [16]. Only upper distance limit of cad-
mium–sulfur cluster was incorporated into structural calculation.
More than 95% of the input NOE data are assigned during the struc-
tural calculation. Two hundred random conformers were annealed in
10000 steps using the distance restraints. The 30 structures with the
lowest target function and 2 NOE violations >0.2 A˚ (0.25 and
0.32 A˚) and no violations >5.0 for angle restraints were energy-mini-
mized under a second generation force field [35], using a non-bounded
cut off of 8.0 A˚ and a generalized Born solvent model [36] in AMBER7
package [37,38]. Structure generation and analysis were performed
with the MOLMOL [39].
2.4. 15N relaxation measurements
15N longitudinal relaxation times (T1),
15N spin–lattice relaxation
times (T2) and
1H–15N NOE measurements were performed, using
the pulse sequences described previously [40]. For T1 measurements,
the spectra were collected with relaxation delays of 10, 50, 100, 200,
300, 450, 600, and 850 ms. For T2 measurements, data were acquired
with delays of 16, 32, 48, 64, 96, 112, 128, 176, 320 ms. A recycle delay
and rat MT3 was created using the program CLUSTALX [12]. The
forms are highlighted in gray. The numbering is made according to theof 3 and 2 s was used for the T1 and T2 relaxation experiments, respec-
tively. A total of 2048 complex data with 200 complex increments were
collected for the relaxation experiments. The T1 and T2 were deter-
mined by fitting the cross-peak intensities to a mono-exponential func-
tion. Error estimates were determined using standard propagation of
errors treatment [40]. The 1H–15N steady-state NOE values were deter-
mined from a pair of spectra, recorded with and without proton satu-
ration. The NOE experiment with proton saturation was carried out
with a 3.5 s interscan relaxation delay followed by 2 s of proton satu-
ration, while the NONOE experiment used a 3.5 s interscan relaxation
delay without proton saturation. NOE values were determined as
the ratio of the peak volume with proton saturation to that without
saturation.