Báo cáo khoa học: Metal exchange in metallothioneins – a novel structurally significant Cd5 species in the alpha domain of human metallothionein 1a ppt

Metal exchange in metallothioneins – a novel structurally
significant Cd
5
species in the alpha domain of human
metallothionein 1a
Kelly E. Rigby Duncan, Christopher W. Kirby and Martin J. Stillman
Department of Chemistry, The University of Western Ontario, London, Canada
Cadmium is a known carcinogen that interferes with
cellular signaling and the regulation of gene expression
[1]. Metallothionein, a cysteine-rich metal-binding pro-
tein, has been shown to protect the cell from toxicity
by sequestering the cadmium ions via the cysteinyl
thiolate ligands [2–4]. Cellular response to cadmium is
dependent on the level of exposure, such that high
concentrations induce cytotoxicity whereas low to
moderate concentrations result in gene dysregulation
and uncontrollable growth. The mechanism of cad-
mium-induced toxicity is complex; however, evidence is
mounting that suggests a role for Cd
2+
in the inhibi-
tion of DNA repair processes [5]. Specifically, Cd
2+
is
thought to impair DNA damage recognition by inter-
fering with the interaction of key nucleotide excision
repair component proteins at the damage site by
Keywords
113
Cd NMR spectroscopy; circular dichroism
spectroscopy; ESI mass spectrometry;

metal exchange; metallothionein
Correspondence
M. J. Stillman, Department of Chemistry,
Chemistry Building, The University of
Western Ontario, London, ON,
Canada N6A 5B7
Fax: +1 519 661 3022
Tel: +1 519 661 3821
E-mail:
Website: />(Received 7 January 2008, revised 27
February 2008, accepted 4 March 2008)
doi:10.1111/j.1742-4658.2008.06375.x
Metallothioneins (MTs) are cysteine-rich, metal-binding proteins known to
provide protection against cadmium toxicity in mammals. Metal exchange
of Zn
2+
ions for Cd
2+
ions in metallothioneins is a critical process for
which no mechanistic or structural information is currently available. The
recombinant human a domain of metallothionein isoform 1a, which
encompasses the metal-binding cysteines between Cys33 and Cys60 of the
a domain of native human metallothionein 1a, was studied. Characteristi-
cally this fragment coordinates four Cd
2+
ions to the 11 cysteinyl sulfurs,
and is shown to bind an additional Cd
2+
ion to form a novel Cd
5

a-MT
species. This species is proposed here to represent an intermediate in the
metal-exchange mechanism. The ESI mass spectrum shows the appearance
of charge state peaks corresponding to a Cd
5
a species following addition
of 5.0 molar equivalents of Cd
2+
to a solution of Cd
4
a-MT. Significantly,
the structurally sensitive CD spectrum shows a sharp monophasic peak at
254 nm for the Cd
5
a species in contrast to the derivative-shaped spectrum
of the Cd
4
a-MT species, with peak maxima at 260 nm (+) and 240 nm
()), indicating Cd-induced disruption of the exciton coupling between the
original four Cd
2+
ions in the Cd
4
a species. The
113
Cd chemical shift of
the fifth Cd
2+
is significantly shielded (approximately 400 p.p.m.) when
compared with the data for the Cd

2+
ions in Cd
4
a-MT by both direct and
indirect
113
Cd NMR spectroscopy. Three of the four original NMR peaks
move significantly upon binding the fifth cadmium. Evidence from indirect
1
H-
113
Cd HSQC NMR spectra suggests that the coordination environment
of the additional Cd
2+
is not tetrahedral to four thiolates, as is the case
with the four Cd
2+
ions in the Cd
4
a-MT, but has two thiolate ligands as
part of its ligand environment, with additional coordination to either water
or anions in solution.
Abbreviations
MT, metallothionein; a-rhMT 1a, recombinantly prepared a domain of human metallothionein isoform 1a.
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2227
substitution for a Zn
2+
ion in the four-cysteine
Zn-finger protein xeroderma pigmentosum group A
protein. This is one of many examples where Cd

2+
has
been shown to replace Zn
2+
in Zn-finger proteins,
resulting in structural alterations and ultimately func-
tional inhibition. Indeed, substitution of Cd
2+
for the
Zn
2+
ions in the two-finger Tramtrack (TTK) peptide
reduces the affinity of this peptide for its DNA-binding
sequence [6]. However, incubation of the Cd-substi-
tuted peptide with Zn-substituted metallothionein
reverses the effect and restores DNA-binding ability.
Similarly, Cd
2+
coordination to the Zn-finger protein
TFIIIA was shown to inhibit DNA association at the
internal control region of the 5S ribosomal RNA gene;
however, metal exchange between Cd-TFIIIA and
Zn-MT resulted in reconstitution of the functional
Zn-finger protein. It is evident from these examples
that metallothionein is the primary defender against
Cd-induced toxicity, and that its role extends beyond
merely sequestering the ‘free’ ion upon cellular expo-
sure. Extraction of the Cd
2+
ion from the affected

protein results in liberation of the essential Zn
2+
ion
from the metallothionein pool; thus the metallothion-
ein exhibits a dual function with respect to metal
replacement.
The physiological effects described above indicate
that metal exchange or metal replacement in metallo-
thioneins is a critically important process that requires
mechanistic consideration; however, details of this pro-
cess are completely lacking. Based on examination of
the structures, we and others have previously proposed
that the domain crevice acts as the initiation site for
the exchange reaction due to exposure of one edge of
the metal-thiolate cluster to the surrounding environ-
ment [7–9]. Upon incorporation of the incoming Cd
2+
ion into the metal-thiolate cluster at the crevice site,
rearrangement of the cluster is proposed to take place,
resulting in expulsion of a previously coordinated
Zn
2+
ion from the domain to reduce the stoichiometry
back to four metal ions. This mechanism would
require an intermediate that includes the metals of the
domain plus the incoming metal. However, to date, no
experimental data have been published to support this
hypothesis.
In this paper, the first structural evidence to support
the formation of a cluster-expanded a domain is

described. CD, NMR spectroscopic and MS data show
the formation of a novel and structurally modified
Cd
5
a-MT 1a species upon titration of Cd
4
a-MT 1a
with a moderate excess of Cd
2+
. The additional Cd
2+
ion is proposed to coordinate to two cysteinyl sulfurs
positioned near the crevice site of the domain, with the
remainder of the ligand sphere probably completed
with either water or chloride ions based on indirect
1
H-
113
Cd HSQC NMR data. We propose that this
cluster-expanded Cd
5
a species represents a model for
the intermediate in the Cd ⁄ Zn metal-exchange reaction
pathway for this particular metallothionein isoform.
Results
The sequence of the thrombin-cleaved isolated a do-
main prepared recombinantly in Escherichia coli as an
S-tag fusion protein (herein referred to as a-rhMT 1a),
as used in this study, is shown in Fig. 1A. This
sequence encompasses the metal-binding cysteines

between Cys33 and Cys60 of the a domain of native
human metallothionein 1a but also includes amino
acids not found in the native protein. The four diva-
lent metal ions are labeled as 1, 5, 6 and 7 in accor-
dance with the original NMR numbering for
two-domain mammalian metallothionein (2, 3 and 4
are assigned to the three divalent metal ions in the
b domain) [10,11]. The 11 cysteinyl sulfurs are labeled
according to the residue number in the natural
two-domain human metallothionein 1a sequence [12].
Existence of the Cd
4
(S
cys
)
11
species is well docu-
mented for the a domain of mammalian metallothione-
ins as the result of structural characterization using a
variety of techniques including NMR spectroscopy
[11,13] and X-ray crystallography [14]. The isolated
Cd
4
(S
cys
)
11
cluster is shown in Fig. 1B: each cadmium
ion (green sphere) coordinates tetrahedrally to four
cysteinyl sulfurs (yellow spheres) such that five of the

11 cysteinyl sulfurs act as bridging ligands between
two metal centers and the remaining six act as terminal
ligands by coordinating to a single metal center. To
date, this is the maximum structurally characterized
Cd-to-cysteine stoichiometry observed for the single
a domain. These results are based largely on studies
carried out on a variety of mammalian MT species
including rabbit, rat and human [15–17].
The numbering of the cadmium ions and the cyste-
inyl sulfurs in Fig. 1B correspond with those in the
sequence shown in Fig. 1A. Figure 1C shows the
space-filling and ribbon model representations of
Cd
4
a-rhMT 1a, emphasizing the wrapping of the poly-
peptide backbone in a left-handed coil around the
metal-thiolate cluster, which is shown in the space-
filling model as located in the center of the domain.
Metal exchange of Zn
4
a-rhMT 1a with Cd
2+
The exchange reaction of the Zn-substituted a domain
with Cd
2+
was investigated by ESI mass spectrometry.
The Zn-substituted metallothionein was prepared by
A novel Cd
5
a metallothionein species K. E. Rigby Duncan et al.

2228 FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS
demetallation of recombinantly isolated Cd
4
a-rhMT
1a at low pH followed by removal of the Cd
2+
ions
using size-exclusion chromatography. Reconstitution
with Zn
2+
was achieved by raising the pH in the
presence of stoichiometric amounts of Zn
2+
. The top
spectrum of Fig. 2 shows that the Zn
4
a species is the
sole species formed in the reconstitution process. In all
the spectra, the measured charge states were +4 and
+5, with the +4 state predominant. Addition of 1.5
molar equivalents of Cd
2+
to the Zn
4
a sample results
in the formation of mixed-metal species, with the
Zn
3
Cd
1

a and Zn
2
Cd
2
a species predominating. The rel-
ative abundance shifted to primarily the Zn
1
Cd
3
a and
Cd
4
a species upon titration with 3.4 molar equivalents
of Cd
2+
. When 4.7 molar equivalents of Cd
2+
are
added, the Cd
4
a species is the predominant species,
indicating a near stoichiometric replacement of the
Zn
2+
ions with the incoming Cd
2+
ions in a non-
cooperative manner. This is consistent with other
reports for the Zn ⁄ Cd metal-exchange reaction [18,19].
However, titration with a moderate excess of Cd

2+
results in the appearance of a Cd
5
a species, which is
shown in Fig. 2 to be present as a minor contributor
upon addition of 4.7 molar equivalents, and is the
dominating species with the addition of 8.2 molar
equivalents of Cd
2+
. This newly identified Cd
5
a spe-
cies was further characterized by CD and UV absorp-
tion spectroscopy and ESI mass spectrometry by
titration of recombinantly prepared Cd
4
a-rhMT 1a
isolated directly from the E. coli source with excess
Cd
2+
.
Titration of Cd
4
a-rhMT 1a with excess Cd
2+
:CD
and ESI-MS results
The CD spectrum obtained for the Cd-coordinated
a domain as isolated from the recombinant prepara-
tion in E. coli is shown in Fig. 3A. A significant fea-

ture of the spectrum with no excess of Cd
2+
is the
biphasic, derivative-shaped signal, with positive
extrema at 260 and 220 nm and a negative extremum
at 240 nm. Many previous studies have reported this
CD spectrum as characteristic of the mammalian Cd
4
a
species, and it has been described as being due to exci-
ton splitting between the symmetric pairs of
[Cd(S
cys
)
4
]
2
groups in the Cd
4
(S
cys
)
11
binding site [18–
22]. This result confirms the correct folding and
domain stoichiometry of the recombinantly synthesized
a domain as being the Cd
4
a species. However, closer
inspection of the CD spectrum reveals a poorly

defined, weak and atypical shoulder at 254 nm, indi-
cating a coexisting secondary species of lower abun-
dance that lacks the exciton coupling property, as a
pure Cd
4
a sample results in a point of inflection at this
wavelength. Such a species was found during metal
replacement and cadmium-loading experiments for
A
BC
Fig. 1. (A) Recombinant sequence of the a domain of human MT 1a showing the connectivities of the four divalent metal cations to the 11
cysteinyl sulfurs. (B) Isolated Cd
4
(S
cys
)
11
cluster present in the a domain of human MT 1a. (C) Space-filling and ribbon representations of the
recombinant Cd
4
a-rhMT 1a. Numbering of metal ions is based on NMR numbering of mammalian MT [11] and cysteine numbering is based
on the natural human MT 1a sequence [12]. Atom legend: gray = C, blue = N, red = O, yellow = S, green = Cd.
K. E. Rigby Duncan et al. A novel Cd
5
a metallothionein species
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2229
Cd < 4 molar equivalents; however, as we show
below, the present peak is due to a species with
Cd > 4 molar equivalents. Based on the ESI mass
spectrometric data for the Zn ⁄ Cd metal exchange, this

species can be identified as a Cd
5
a species, and conse-
quently should increase in abundance upon titration
with excess Cd
2+
. We note similar broadening of the
CD spectrum for the Cd-substituted two-domain
ba-MT 1a as isolated from E. coli [23].
Addition of Cd
2+
to the solution of Cd
4
a, up to 5.0
molar equivalents, resulting in a total of 9.0 molar
equivalents of Cd
2+
in solution, results in a significant
shift in the CD spectrum, leading stepwise to a mono-
phasic peak at 254 nm and a reduction in peak inten-
sity of the band at 223 nm to negative DA values.
Despite the significant change observed in the CD
spectrum, the corresponding UV absorption spectrum
shows very little change upon addition of excess Cd
2+
(Fig. 3B). The loss of exciton coupling in the CD spec-
trum following titration of excess Cd
2+
into the
protein sample must be due to an alteration of the

metal-thiolate cluster arrangement, which we associate
Fig. 2. ESI mass spectra recorded for the titration of Zn
4
a-rhMT 1a
with Cd
2+
at pH 7.4. Spectral changes were recorded as aliquots of
Cd
2+
(3.3 mM) were titrated into a solution of Zn
4
a-rhMT 1a (15 lM)
at 22 °C. Spectra were recorded at Cd
2+
molar equivalent values of
0.0, 1.5, 3.4, 4.7 and 8.2.
A
B
C
Fig. 3. (A) CD and (B) UV absorption spectral changes observed
upon titrating Cd
4
a-rhMT 1a with an additional 5.0 molar equiva-
lents of Cd
2+
at pH 7.4 and 22 °C. (C) The ratio of CD peak inten-
sity at 254 nm per 264 nm versus molar equivalents of Cd
2+
added
to a sample of Cd

4
a-rhMT 1a as a measure of Cd
5
a-rhMT 1a
species formation.
A novel Cd
5
a metallothionein species K. E. Rigby Duncan et al.
2230 FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS
with loss of symmetry of the Cd
4
a structure. This spec-
trum is reminiscent of the CD spectrum recorded for
the a domain with up to three Cd
2+
ions [18,24].
Retention of the overall CD envelope shape indicates
that no significant changes in the wrapping of the
polypeptide backbone are induced by the additional
Cd
2+
. We propose that titration of excess Cd
2+
alters
only the metal-thiolate cluster stoichiometry to give
the lower-symmetry Cd
5
a species (see below).
Cycling the pH from neutral to acidic and then back
to neutral pH results in demetallation and subsequent

metallation of the protein, which can be monitored by
both CD spectroscopy and ESI mass spectrometry.
This reaction sequence, when applied to the Cd
4
a sam-
ple, results in restoration of the Cd
4
stoichiometry
upon raising the pH from 2 to 7 (data not shown).
The presence of an additional 5.0 molar equivalents of
Cd
2+
(total ratio of Cd
2+
: MT of 9.0) results in for-
mation of the Cd
5
a species, which also reforms repro-
ducibly upon cycling the pH (data not shown).
A plot of the ratio DA
254
⁄ DA
264
from the CD spec-
trum approximates the ratio of Cd
5
a ⁄ Cd
4
a. Figure 3C
shows that up to 5.0 additional molar equivalents are

required for nearly all the metallothionein species to
be converted into the Cd
5
a form.
The ESI mass spectrum obtained for the purified
Cd-coordinated a domain as isolated from the recom-
binant preparation in E. coli is shown in Fig. 4A. The
measured charge state distribution ranges from +3 to
+5, with the +4 charge state as the predominant
peak. Reconstruction of the mass spectrum results in a
single, principal species with a measured mass of
4526.4 Da, corresponding to the Cd
4
a-rhMT species
(calculated mass 4524.6 Da). Closer inspection of the
original mass spectrum shows the presence of a minor
peak with a measured m ⁄ z of 927, corresponding to
the +5 charge state of a Cd
5
a-rhMT species. This
result confirms the existence of a Cd
5
a species as a
minor contributor to the equilibrium of the Cd-coordi-
nated a domain of human MT 1a.
Addition of 5.0 molar equivalents of Cd
2+
to the
Cd
4

a-rhMT 1a solution results in the ESI mass spec-
trum shown in Fig. 4B. Peaks corresponding to the
Cd
4
a-rhMT species are no longer detected, and
instead a new set of peaks are observed that are
consistent with the formation of 100% Cd
5
a-rhMT
species with a reconstructed mass of 4633.2 Da (cal-
culated mass 4635.0 Da). The measured charge state
distribution remains +3 to +5, with the +4 charge
state predominating; however, the relative abundance
of the +5 charge state has increased significantly
compared to the corresponding peak in the mass
spectrum of the Cd
4
a-rhMT species. Previous ESI-MS
studies of globular proteins have shown a correlation
between the observed charge state distribution and
the solution polypeptide conformation [25–27]. In the
case of metallothionein, Palumaa et al. have reported
ESI-MS data showing a higher charge state distribu-
tion for the Cd
4
a domain of human MT 3 compared
with the Zn
4
a MT 3 by one unit [28]. Similarly, the
Cd

3
b domain of human MT 3 was shown to be sig-
nificantly more open in conformation compared with
the Zn
3
b domain of MT 3, indicating non-isostructur-
al replacement of Cd
2+
for Zn
2+
in this particular
MT isoform. This charge state distribution was inter-
preted by the authors as being due to a slight open-
ing of the polypeptide backbone to accommodate the
slightly larger cadmium ions. These data strongly sup-
port our interpretation that the increased relative
abundance of the higher +5 charge state species
observed in the ESI mass spectrum of the Cd
5
a pre-
sented here compared with the Cd
4
a species is due to
expansion of the metal binding domain to accommo-
date the fifth Cd
2+
. This suggests that the additional
Cd
2+
ion is inserted into the core of the domain,

becoming part of an expanded metal-thiolate cluster.
113
Cd NMR spectroscopy was therefore used to probe
the metal-thiolate cluster arrangement in the newly
identified Cd
5
a species.
A
B
Fig. 4. (A) ESI mass spectrum observed for the Cd-coordinated
a-rhMT 1a following isolation and purification of the recombinant
protein from E. coli. Reconstruction of the mass spectrum results
in a measured mass of 4526.4 Da corresponding to the Cd
4
a-rhMT
species (calculated mass 4524.6 Da). (B) ESI mass spectral
changes observed upon titrating the Cd
4
a-rhMT 1a sample from (A)
with an additional 5.0 molar equivalents of Cd
2+
at pH 7.4 and
22 °C. The reconstructed mass for Cd
5
a-rhMT 1a was 4633.2 Da
(calculated mass 4635.0 Da).
K. E. Rigby Duncan et al. A novel Cd
5
a metallothionein species
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2231

Titration of Cd
4
a-rhMT 1a with excess Cd
2+
:
113
Cd
NMR results
113
Cd NMR spectroscopy was used in this study to
further investigate the nature of the metal-thiolate
binding site in the a domain of human MT 1a both
before and after the addition of excess Cd
2+
.
Direct 1D
113
Cd NMR (
1
H-decoupled) spectroscopic
techniques to probe for formation of a novel
Cd
5
a-rhMT 1a species
The 1D
113
Cd NMR (
1
H-decoupled) spectrum of
Cd

4
a-rhMT 1a as isolated directly from recombinant
overexpression in E. coli and prepared in 10 mm
Tris ⁄ HCl buffer (pH 7.4) is shown in Fig. 5A. The
natural isotopic abundance of
113
Cd was used in this
experiment, despite the low value of 12.26%, in order
to observe the naturally occurring speciation. Six sig-
nals were observed in the Cd
4
a-rhMT 1a spectrum at
670, 633, 630, 626, 611 and 599 p.p.m. The chemical
shift values of the five most deshielded peaks observed
in the NMR spectrum are in the range of 670 to
611 p.p.m., and are in agreement with those reported
previously for the four cadmium ions in the a domain
of the native two-domain human MT isoform 1 [10].
These five peaks are labeled in Fig. 5A as 1, 5, 5¢,6
and 7, respectively, in accordance with the original
NMR numbering assignments. Splitting of the peak
assigned to the metal in site 5 has been noted previ-
ously and is attributed to heterogeneity in that particu-
lar site in the metal-thiolate cluster [10].
The sixth peak observed at 599 p.p.m. is more
shielded than the other peaks and has not been
reported previously for human MT 1 isoforms. Based
on the ESI-MS and CD spectroscopic results, this peak
is predicted to be due to the Cd
5

a-rhMT species, which
has been shown in this report to be a minor contribu-
tor to the equilibrium together with the Cd
4
a-rhMT
species.
Fig. 5. (A) Direct 1D
113
Cd NMR (
1
H-decoupled) spectrum (133 MHz) for a-rhMT 1a following isolation and purification of the recombinant
protein from E. coli with the natural isotopic abundance of
113
Cd, showing primarily the
113
Cd
4
a-rhMT 1a species. (B) Direct 1D
113
Cd
NMR (
1
H-decoupled) spectrum (133 MHz) for isotopically labeled
113
Cd
4
a-rhMT 1a titrated with an additional 10.0 molar equivalents of
113
Cd
2+

to form
113
Cd
5
a-rhMT 1a. The spectrum of
113
Cd
5
a-rhMT 1a (B) is a combination of two separate spectra acquired in the regions
585–705 p.p.m. and 220–245 p.p.m. Samples were prepared in 10 m
M Tris ⁄ HCl pH 7.4, and buffer-exchanged into 10% D
2
O for the Cd
4
a-
rhMT 1a sample and > 70% D
2
O for the
113
Cd
5
a-rhMT 1a sample. All spectra were acquired at 25 °C.
A novel Cd
5
a metallothionein species K. E. Rigby Duncan et al.
2232 FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS
A1D
113
Cd NMR (
1

H-decoupled) spectrum of iso-
topically enriched
113
Cd
4
a-rhMT 1a titrated with an
additional 10.0 molar equivalents of
113
Cd
2+
(total
ratio of Cd
2+
: MT of 14.0) is shown in Fig. 5B. For
easier viewing, the spectrum is divided into two parts,
the region on the left covers the chemical shift range
585–705 p.p.m. and that on the right covers the range
215–245 p.p.m. There are five relatively sharp signals
at 685, 647, 630, 599 and 224 p.p.m. The peak detected
at 599 p.p.m. confirms the presence of a small popula-
tion of the Cd
5
a species in the naturally isolated Cd
4
a
sample, as this peak was observed in the 1D spectrum
of this sample. The four most deshielded peaks
detected between 600–700 p.p.m. in the Cd
5
a spectrum

are assigned to the four
113
Cd
2+
ions that are known
to bind to the a domain of mammalian metallothion-
ein in a tetrahedral coordination to four thiolate
ligands. The relative assignment of these peaks to the
four cadmium sites is comparable to that of the
113
Cd
4
a-rhMT species in that the order is 1, 5, 6 and 7
for the peaks 685, 647, 630 and 599, respectively. How-
ever, the observed chemical shifts have changed signifi-
cantly with the addition of the fifth Cd
2+
ion. Peaks 1,
5 and 6 have shifted upfield by 15, 14 and 3 p.p.m.,
respectively, while peak 7 has shifted downfield by
12 p.p.m. In addition, the peaks labeled 5 and 5¢ in the
spectrum of Cd
4
a-rhMT 1a (Fig. 5A) have collapsed
into a single peak in the spectrum acquired with excess
113
Cd
2+
(Fig. 5B), indicating a loss of heterogeneity at
that site in the metal-thiolate cluster. This is expected

if a fifth Cd
2+
ion results in strain in the binding site,
reducing fluxionality of the metal cluster.
The signal detected at 224 p.p.m. is assigned to the
additional fifth Cd
2+
ion, confirming the CD spectro-
scopic and mass spectrometric data regarding the for-
mation of a Cd
5
a species. This peak is significantly
shielded compared to the other four peaks in the spec-
trum (approximately 400 p.p.m.), indicating that the
coordination environment around this additional Cd
2+
is not tetrahedral to four thiolate ligands. A previous
study reporting the chemical shifts of inorganic cad-
mium(II)-thiolate complexes correlated signals with
chemical shifts in the region of 224 p.p.m. with octahe-
dral complexes of the form Cd(RS)
2
(OH
2
)
4
[29,30].
Although the current data do not provide enough
information to verify this exact ligand field assignment,
as chloride ions are equally as likely as water molecules

to participate as ligands, it does provide support for
two thiolate groups acting as ligands for the fifth Cd
2+
ion and an increase in coordination number from four
to six. This proposed ligand field assignment suggests
partial insertion of the fifth Cd
2+
ion into the metal-
thiolate cluster in a manner that allows solvent access.
Given this restriction, the most likely position for the
fifth Cd
2+
ion is the crevice site of the domain in which
a number of the cysteinyl sulfurs that make up the
metal-thiolate cluster are solvent-exposed. To further
explore this possibility, indirect 2D NMR methods
were employed as a means of probing the coordination
environment around the additional Cd
2+
ion, with par-
ticular emphasis on identifying the two cysteinyl sulfurs
that are proposed to ligate the fifth Cd
2+
ion.
Indirect 2D
1
H–
113
Cd NMR spectroscopic techniques
for identification of the binding site for the fifth Cd

2+
ion in the Cd
5
a-rhMT 1a cluster
The indirect 2D
1
H–
113
Cd NMR approach exploits the
3
J scalar coupling between the cysteine b protons and
the coordinated cadmium ions as a means of mapping
out the metal-thiolate cluster connectivities and identi-
fying the cysteine residues that are coupled to the fifth
Cd
2+
ion. By focusing on the
1
H chemical shift range
of 2.3–3.6 p.p.m., corresponding to the cysteine H
b
protons only, the tetrathiolate connectivities of the
Cd(S
cys
)
4
units can be identified. Furthermore,
sequence assignment of the cysteine residues is possible
through identification of bridging versus terminal
cysteine ligands in the 2D spectrum.

Two-dimensional
1
H–
113
Cd HSQC NMR spectra
were acquired for the
113
Cd
5
a-rhMT 1a formed by
titration of
113
Cd
4
a-rhMT 1a with an additional 10.0
molar equivalents of
113
Cd
2+
(total ratio of
Cd
2+
: MT of 14.0). Figure 6 shows a combination of
two separate spectra acquired in the
113
Cd chemical
shift ranges of 585–705 p.p.m. and 215–245 p.p.m. to
allow visualization of the correlations between all five
Cd
2+

ions in the Cd
5
a cluster with the 11 cysteinyl sul-
fur residues. The H
b
–
113
Cd
3
J scalar couplings were set
to 66 and 40 Hz for acquisition between 585 and
705 p.p.m. and 215 and 245 p.p.m., respectively. Four
strong peaks were observed in the
113
Cd dimension of
the 2D spectrum of
113
Cd
5
a-rhMT 1a in the range
585–705 p.p.m. (Fig. 6), at chemical shift values that
are in agreement with those observed in the 1D
113
Cd
NMR (
1
H-decoupled) spectrum for the Cd
5
a species
(Fig. 5B, sites 1, 5, 6 and 7). The fifth Cd

2+
peak was
observed in the 2D spectrum acquired in the
113
Cd
chemical shift range of 215–245 p.p.m., with an exact
chemical shift of 224 p.p.m., which is also in agree-
ment with the observed chemical shift in the 1D
113
Cd
NMR (
1
H-decoupled) spectrum (Fig. 5B, site X).
Identification of the specific bridging versus terminal
cysteine residues is possible in this
113
Cd-decoupled
spectrum, and a nearly complete assignment of the
cysteine residues in the 2D spectrum has been
K. E. Rigby Duncan et al. A novel Cd
5
a metallothionein species
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2233
accomplished. The bridging cysteines are identified by
a solid line in Fig. 6, because a single H
b
chemical shift
correlates to two different
113
Cd atoms. The numbers

written beside each peak in Fig. 6 correspond to the
sequence number of the cysteine residues as labeled in
Fig. 1A. This assignment is the most probable solution
that satisfies the known connectivities in the metal-
thiolate cluster (Fig. 1A,B).
Although the H
b
–
113
Cd correlations are known for
the four tetrahedral thiolate-coordinated
113
Cd
2+
ions,
the unknown correlations of interest are those of the
fifth Cd
2+
. The two peaks correlating to the fifth
Cd
2+
at 224 p.p.m. were observed at
1
H chemical shift
values of 2.97 and 3.59 p.p.m., which are consistent
with the H
b
chemical shifts of cysteine residues 34 and
either 57 or 59, respectively, as shown by the dotted
lines in Fig. 6. This result substantiates our interpreta-

tion of cluster expansion to a Cd
5
(S
cys
)
11
species upon
titration with excess Cd
2+
, as opposed to the fifth
Cd
2+
ion attaching as an adduct to the surface of the
domain. The detection of only two correlations also
confirms that the coordination sphere of the fifth
Cd
2+
ion includes two cysteine residues as predicted
by the highly shielded chemical shift [30]. Figure 7
shows a space-filling model of Cd
4
a-rhMT 1a with a
view of the crevice site showing the exposed edge of
the metal-thiolate cluster. Cys34 is one of the sulfur
atoms exposed in this site (highlighted in purple),
which supports the NMR results indicating that this
sulfur atom is involved in coordination of the fifth
Cd
2+
ion. Cys57 and Cys59 are not present in the cre-

vice site, so it is not immediately obvious how the sec-
ond sulfur is involved in the coordination. One could
envision a potential cluster rearrangement upon coor-
dination to Cys34 that brings Cys57 or Cys59 into
position for coordination.
Discussion
The in vitro reactivity of metallothionein with Cd
2+
has been well documented by reports on the native
1
H (p.p.m.)
113
CD (p.p.m.)
Fig. 6. A combination of two indirect 2D
1
H–
113
Cd HSQC NMR spectra for isotopically enriched
113
Cd
4
a-rhMT 1a titrated with an additional
10.0 molar equivalents of
113
Cd
2+
to form the
113
Cd
5

a-rhMT 1a species. The spectra were recorded in the
1
H chemical shift range 2.3–
3.7 p.p.m. and the
113
Cd ranges 590–690 p.p.m. (
3
J = 66 Hz) and 220–245 p.p.m. (
3
J = 40 Hz). Both spectra were acquired at 25 °C using
an inverse single-axis z-gradient HCX probe with X tuned to
113
Cd.
Fig. 7. Space-filling model of Cd
4
a-rhMT 1a in two orientations
rotated by 90°, emphasizing the crevice site on the domain. One of
the proposed ligands of the fifth Cd
2+
ion, Cys34, is highlighted by
the arrow and the sulfur atom of this residue is shown in purple.
Atom legend: gray = C, blue = N, red = O, yellow = S, green = Cd.
A novel Cd
5
a metallothionein species K. E. Rigby Duncan et al.
2234 FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS
and ⁄ or recombinant two-domain protein, in addition
to the isolated fragments, from many mammalian spe-
cies including human, rat, rabbit and mouse [18,19,
31–33]. These spectroscopic studies involve reaction of

Zn-containing or metal-free forms of metallothionein
with sub-stoichiometric, stoichiometric or excess
amounts of Cd
2+
. Significantly, titration of Zn
7
-MT
with sub-stoichiometric molar equivalents of Cd
2+
resulted in a gradual red shift of the charge-transfer
band in the CD spectrum, indicating mixed-metal
speciation before saturation with seven Cd
2+
ions.
This result is consistent with the isolation of the
mixed-metal Cd
5
Zn
2
-MT species from in vivo sources
such as rabbit liver upon exposure of these animals to
Cd
2+
, indicating a non-cooperative mechanism of
metal replacement [19]. The available X-ray, NMR
and CD data are consistent with the Cd
4
(S
cys
)

11
cluster
in the a domain being adamantane-like in structure.
Thus, the Cd
4
a domain observed from many mamma-
lian species is characterized by a derivative-shaped CD
signal and a distinct NMR spectrum.
Despite the highly conserved positioning of the
cysteine residues in the mammalian sequence of metal-
lothionein and the structural consistencies, the behav-
ior of Cd
7
ba-MT towards excess Cd
2+
has been
shown to differ depending not only on the species but
also on the isoform or sub-isoform of the protein
within a particular species. Mouse MT 1, the sequence
of which is shown in Fig. 8, has been shown by CD
spectroscopy to have the capability of expanding
beyond the typical seven Cd
2+
ions per metallothion-
ein molecule [31,33]. Significant changes in the CD
spectrum of the isolated b domain of mouse MT 1
when binding additional Cd
2+
were interpreted by the
authors as the result of a Cd-induced, rearranged

peptide conformation. Unfortunately there were no
mass spectral data to confirm the actual number of
Cd
2+
ions bound. Reaction of the isolated a domain
of mouse MT 1 with excess Cd
2+
did not induce a
change in the CD spectroscopic profile, indicating that
this cluster did not expand beyond the Cd
4
(S
cys
)
11
stoi-
chiometry. The Cd-substituted two-domain human
MT 3 has been shown to coordinate additional Cd
2+
ions with stoichiometries of up to Cd
13
ba-MT; how-
ever, species with more than nine equivalents of Cd
2+
are reported as probably being due to adducts on the
surface of the protein. The additional two Cd
2+
ions
leading to the Cd
8

ba-MT and Cd
9
ba-MT species of
human MT 3 are proposed to bind into the cluster
regions; however, the ESI-MS results reported did not
provide the structural information necessary for locali-
zation of these metal ions [28,32,34]. ESI mass spectral
data have been reported for the single a domain of
human MT 2, prepared as a synthetic peptide, in the
presence of excess Cd
2+
, in which Cd
5
a was detected
as a minor species [35]. However, structural data on
this species were not provided, leaving open the possi-
bility that the fifth Cd
2+
acts as an adduct, as is com-
monly observed in mass spectroscopy and would lead
to the increased mass found in the ESI mass spectrum.
Interestingly, the structurally sensitive CD spectro-
scopic data reported for rabbit liver MT 2a and rat
liver MT 2, isolated from natural sources, showed no
change in the CD spectrum upon addition of excess
Cd
2+
, indicating that these particular metallothionein
species do not have the capability to expand to larger
metal clusters at reasonably low levels of excess Cd

2+
[18,36]. ESI mass spectral data for the two-domain
rabbit MT 1a also indicated that this protein was satu-
rated at seven Cd
2+
ions, similar to the data for the
rabbit MT 2a isoform [34].
While ESI-MS data have been reported previously
that do show Cd
2+
binding greater than seven for the
two-domain ba full chain of human MT 1a [23], no
structural data have been provided to indicate the sig-
nificance of the ‘over-loaded’ species. The experimental
data presented here show unambiguously that a Cd
5
a
species of recombinantly prepared human MT 1a is
Fig. 8. Sequence comparison of human MT 1a with other isoforms of human metallothionein as well as metallothionein from other mamma-
lian species for which spectroscopic data have been reported.
K. E. Rigby Duncan et al. A novel Cd
5
a metallothionein species
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2235
formed that involves a metal binding site that is differ-
ent from the Cd
4
a site. All three techniques used pro-
vide specific information. First, the ESI-MS data
confirm the increased metal binding stoichiometry and

indicate a slight unwinding of the peptide backbone.
Second, the CD spectroscopic data show a disruption
of the exciton coupling with molar equivalents of
Cd
2+
> 4, confirming disruption of the Cd
4
binding
site. Finally, the NMR data verify the coordination of
the fifth Cd
2+
ion by thiolate ligands within the clus-
ter, and, more specifically, within the crevice site of the
protein.
The biological significance of the Cd
5
a species is an
interesting question. We propose that the Cd
5
a species
described here may be a model of the intermediate in
the Cd ⁄ Zn metal-exchange reaction, a critically impor-
tant process for cadmium detoxification. However, the
data presented here only apply to the human metallo-
thionein isoform 1a, although a number of reports of
‘over-loaded’ metallothioneins for other mammalian
species are noted above. The ESI mass spectral data
obtained for the replacement reaction of the Zn-substi-
tuted a domain with Cd
2+

(Fig. 2) show a range of
mixed-metal species with a stoichiometry of no more
than four metal ions in any given species. The total
loading of four divalent metal ions that is observed
until all of the Zn
2+
ions have been replaced can be
explained in terms of the relative binding constants of
the two metals. The incoming metal by necessity has a
higher binding constant than the metals already pres-
ent, so the five-metal intermediate, when both Cd
2+
and Zn
2+
are present, is short-lived as Zn is immedi-
ately displaced. In the case of human MT 1a, when
four Cd
2+
ions are bound, we propose that the fifth
Cd
2+
ion is simply trapped at the exchange site as no
exchange can take place.
Experimental procedures
Materials
The chemicals used were cadmium sulfate (Fisher Scientific,
Ottawa, Canada), cadmium(113) chloride (Trace Sciences
International Inc., Richmond Hill, Canada), deuterium
oxide (Cambridge Isotopes Laboratories Inc., Andover,
MA, USA), ultrapure Tris buffer, tris(hydroxymethyl)ami-

nomethane (ICN Biomolecules, Irvine, CA, USA), zinc
sulfate (Caledon Laboratory Chemicals, Georgetown,
Canada), ammonium formate buffer (Sigma-Aldrich,
Oakville, Canada), ammonium hydroxide (BDH Chemi-
cals ⁄ VWR, Mississauga, Canada), formic acid (J. T. Baker
Chemical Co., Phillipsburg, NJ, USA) and hydrochloric
acid (Caledon Laboratory Chemicals). All solutions were
produced using > 16 MWÆcm deionized water (Barnstead
Nanopure Infinity, Van Nuys, CA, USA). HiTrapÔ SP HP
ion-exchange columns (Amersham Biosciences ⁄ GE Health-
care, Piscataway, NJ, USA), superfine G-25 Sephadex
(Amersham Biosciences), a stirred ultrafiltration cell (Am-
icon Bioseparations ⁄ Millipore, Bedford, MA, USA) with
YM-3 membrane (3000 molecular weight cut-off) and a Mi-
crocon YM-3 centrifugal filtration device (Amicon Biosepa-
rations ⁄ Millipore) were used in the protein purification
steps.
Protein preparation
The recombinant a domain of human metallothionein 1a
(sequence shown in Fig. 1A) was produced by overexpres-
sion in E. coli BL21(DE3) cells as an S-tag fusion protein
in the presence of Cd
2+
as described previously [24]. Fol-
lowing isolation and purification, the S-tag was cleaved
from the protein by incubation with thrombin.
Metal exchange of Zn
4
a-rhMT 1a with Cd
2+

Metal-free apo-a-rhMT was prepared by eluting the throm-
bin-cleaved Cd-bound protein from a G-25 column equili-
brated with deionized water that had been pH-adjusted
with HCOOH to pH 2.8. Elution of the protein using an
eluant of low pH effectively removes the metal ions from
the protein, which are then separated from the protein band
as a result of the size-exclusion processes. Preparation of
apo-MT by this method simultaneously desalts the solution
by the same size-exclusion process. The metal-free protein
was reconstituted by adding 4.0 molar equivalents of Zn
2+
(3.0 mm stock solution) and increasing the pH to 7.4. The
a-rhMT solution was determined to have a concentration
of 15 lm based on UV absorption at 220 nm
(e
220
= 40 000 LÆmol
)1
Æcm
)1
) and atomic absorption spec-
troscopy following complete metallation with Zn
2+
. Cad-
mium solutions were prepared in 25 mm ammonium
formate pH 7.4 (for MS studies) to a final concentration of
3.3 mm as determined by atomic absorption spectroscopy.
The final samples were thoroughly evacuated and argon-
saturated to remove the bulk of the oxygen from the solu-
tions in order to deter oxidation of the protein.

Cd
2+
was added incrementally to the solution of Zn
4
a to
8.2 molar equivalents, with thorough mixing after each
titration. Mass spectra were acquired at each addition after
a 2–5 min delay to allow the reaction to reach equilibrium
conditions. Mass spectra were acquired on a Micromass
LCT ESI-TOF mass spectrometer (Waters Micromass,
Mississauga, Canada) at room temperature (22 °C), and
recorded using the mass lynx software package version
4.0. The ESI-TOF spectrometer was calibrated with a solu-
tion of NaI. The scan conditions for the spectrometer were:
capillary, 3000.0 V; sample cone, 39.0 V; RF lens, 450.0 V;
A novel Cd
5
a metallothionein species K. E. Rigby Duncan et al.
2236 FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS
extraction cone, 11.0 V; desolvation temperature, 20.0 °C;
source temperature, 80.0 °C; cone gas flow, 51 LÆh
)1
; desol-
vation gas flow, 528 LÆh
)1
. The m ⁄ z range was 500.0–
1600.0, the scan mode was continuum, and the interscan
delay was 0.10 s. The observed spectra were reconstructed
using the max ent 1 program from the mass lynx ver-
sion 4.0 software package (Waters Micromass).

Titration of Cd
4
a-rhMT 1a with excess Cd
2+
A solution of Cd
4
a-rhMT 1a was prepared in 10 mm
Tris ⁄ HCl buffer (pH 7.4) to a concentration of 16 lm for
the CD spectroscopic studies. The same protein sample
was buffer-exchanged into ammonium formate (pH 7.4)
using a Microcon YM-3 centrifugal filtration device to
a final concentration of 16 lm for the mass spectro-
metric studies. The protein concentrations were confirmed
by UV absorption spectroscopy using the absorbance at
250 nm, which corresponds to the ligand-to-metal charge
transfer transition generated by the metal-thiolate bond
(e
250
=56000m
)1
Æcm
)1
) and atomic absorption spectros-
copy assuming a 4 : 1 ratio of Cd : MT. Cadmium solu-
tions were prepared as the sulfate salt in 10 mm Tris ⁄ HCl
(pH 7.4) for CD studies or 25 mm ammonium formate
(pH 7.4) for ESI-MS studies to a final concentration of
3.0–3.3 mm as determined by atomic absorption spectro-
scopy.
CD and UV absorption spectra were acquired for the

Cd
4
a-rhMT 1a solution (16 lm)in10mm Tris ⁄ HCl buffer
(pH 7.4) with no additional Cd
2+
and 5.0 molar equiva-
lents of Cd
2+
. The Cd
2+
was added as the sulfate salt dis-
solved in 10 mm Tris ⁄ HCl buffer (pH 7.4) to a final
concentration of 3.3 mm. CD spectra were acquired on a
Jasco J810 spectropolarimeter (Jasco, Easton, MD, USA)
in a 1 cm quartz cuvette at room temperature (22 °C) and
recorded using spectra manager version 1.52.01 software
(Jasco). The wavelength range of 200–300 nm was scanned
continuously at a rate of 50 nmÆmin
)1
with a band width of
2 nm. All spectra were baseline-corrected using 10 mm
Tris ⁄ HCl. The spectral data were organized and plotted
using originÒ version 7.0383 (OrginLab Corp., Northamp-
ton, MA, USA). The CD spectra are measured as DA,
which required conversion of the measured ellipticity in
degrees divided by a factor of 33. UV spectra were acquired
using a Cary 5G UV-Vis-NIR spectrophotometer (Varian
Canada, Mississauga, Canada) in a 1 cm quartz cuvette at
room temperature (22 °C) and recorded using the Cary win
uv scan software application. The wavelength range of

200–300 nm was scanned continuously. All spectra were
baseline-corrected using 10 mm Tris ⁄ HCl. The spectral data
were organized and plotted using originÒ version 7.0383.
An initial ESI mass spectrum was acquired for the
Cd
4
a-rhMT 1a solution (16 lm)in25mm ammonium for-
mate buffer (pH 7.4) with no additional Cd
2+
prior to
addition of 5.0 molar equivalents of Cd
2+
. The Cd
2+
was
added as the sulfate salt dissolved in 25 mm ammonium
formate buffer (pH 7.4) to a final concentration of
3.3 mm. Mass spectra were acquired on a Micromass LCT
ESI-TOF mass spectrometer (Waters Micromass) at room
temperature (22 °C) and recorded using the mass lynx
software package version 4.0. The observed spectra were
reconstructed using the max ent 1 program from the
mass lynx version 4.0 software package (Waters Micro-
mass).
NMR spectroscopic study
The Cd
4
a-rhMT product used for acquisition of the 1D
113
Cd NMR (

1
H-decoupled) spectrum was prepared using
the natural isotopic abundance of
113
Cd. For this unlabeled
NMR sample, protein products from two recombinant
preparations were pooled following thrombin cleavage of
the S-tag and concentrated to 5 mL using the Amicon
ultrafiltration cell with the YM-3 membrane, followed
by further concentration using the Microcon YM-3 centri-
fugal filtration device to 500 lL. To provide the field
frequency lock, 100 lLofD
2
O was added to the sample.
The final protein concentration was determined to be
5-6 mm based on atomic absorption spectroscopy
assuming a 4 : 1 Cd : MT ratio, as well as UV absorption
spectroscopy using the 250 nm peak (e
250
= 56 000 LÆ
mol
)1
Æcm
)1
). The buffer used for this sample was 10 mm
Tris ⁄ HCl (pH 7.4). The sample was argon-saturated and
sealed in a 5 mm NMR tube for analysis.
The
113
Cd

5
a-rhMT 1a sample (10.0 molar equivalents
113
Cd
2+
added to a sample of
113
Cd
4
a-rhMT 1a) used for
both the 1D and 2D NMR spectra was isotopically labeled
to increase sensitivity. The protein product from one
recombinant preparation was pooled following thrombin
cleavage of the S-tag and concentrated to a volume of
5 mL using the Amicon ultrafiltration cell with the YM-3
membrane. Reconstitution with
113
Cd was carried out by
first eluting the thrombin-cleaved Cd-bound protein from a
G-25 column equilibrated with 25 mm ammonium formate
buffer (pH 2.7) to generate the metal-free protein.
113
Cd
(> 95%) was added as the chloride salt to the low-pH pro-
tein fraction in a 4 : 1 Cd : MT molar ratio. The pH of the
reconstituted protein sample was raised by addition of
ammonium hydroxide to a final pH of 7.8. The
113
Cd-
reconstituted sample was concentrated to 5 ml using the

Amicon ultrafiltration cell with the YM-3 membrane, fol-
lowed by further concentration to 600 lL using the Micro-
con YM-3 centrifugal filtration device. The concentrated
sample was buffer-exchanged into D
2
O to 60–90% D
2
O
using the Microcon YM-3 centrifugal filtration device,
argon-saturated and sealed in a 5 mm NMR tube for anal-
ysis. The final protein concentration of the
113
Cd
4
a-rhMT
species was determined to be approximately 2-3 mm
by atomic absorption spectroscopy and UV absorption
spectroscopy.
K. E. Rigby Duncan et al. A novel Cd
5
a metallothionein species
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2237
Both direct 1D
113
Cd (
1
H-decoupled) and indirect 2D
1
H–
113

Cd HSQC (
3
J = 66 Hz in the range 585–
705 p.p.m. and 40 Hz in the range 215–245 p.p.m.) NMR
spectra were acquired for the isotopically labeled
113
Cd
5
a-
rhMT 1a (2-3 mm) following addition of 10.0 molar
equivalents of
113
Cd
2+
in the form of aliquots of solid
113
CdCl
2
added directly to the 600 lL
113
Cd
4
a-rhMT 1a
NMR sample. The
113
CdCl
2
was added to the NMR
sample in aliquots of 0.5, 1.0, 5.0 and 10.0 molar equiva-
lents, and the resultant solution monitored by CD spec-

troscopy. Following addition of the final aliquot of
113
CdCl
2
, the NMR sample was argon-saturated and
sealed for analysis. All of the NMR spectra were
acquired on a Varian Inova 600 NMR spectrometer
using vnmrj 1.1D software (Varian Canada Inc.). The
spectrometer was equipped with a single-axis z-gradient
HCX, probe with X tuned to
113
Cd (
1
H = 599.44 MHz,
113
Cd = 132.99 MHz) for both
113
Cd direct and indirect
experiments. All of the
113
Cd 1D spectra were acquired
using inverse-gated
1
H decoupling, a 30°
113
Cd flip angle,
a 0.2 s acquisition time, and a relaxation delay of 2 s.
The natural abundance
113
Cd NMR spectrum of the

Cd
4
a species was acquired using 70 000 scans from 590–
690 p.p.m. [37,38], but a much larger shift range was
studied (822 to )26 p.p.m.) for the
113
Cd-labeled Cd
5
species to find the fifth site, and only 25 000 scans were
required to obtain an adequate signal-to-noise ratio. All
113
Cd spectra were referenced to an external CdSO
4
solu-
tion. Once the
113
Cd shift ranges had been established,
1
H-
113
Cd HSQC were acquired to obtained connectivities.
Due to the large
113
Cd spectral range, it was not possible
to obtain a single
1
H–
113
Cd HSQC spectrum for the Cd
5

species, so it had to be acquired in two separate experi-
ments. For the high shift range (585–705 p.p.m.), 64 incre-
ments of 192 scans were acquired using a
3
J of 66 Hz and
GARP
113
Cd decoupling. Linear prediction to 256 points
and zero filling to 1024 were used to obtain good spectral
resolution. For the lower shift range (215–245 p.p.m.), 32
increments of 1024 scans were acquired using a
3
J of 40 Hz
and GARP
113
Cd decoupling. Linear prediction to 128
points and zero filling to 1024 were used to obtain good
spectral resolution. All of the NMR spectra shown here
were re-processed using the ACD Labs 2D NMR processing
package version 9.0 (Advanced Chemistry Development
Inc., Toronto, Canada) to a final size of 1 K·.
Acknowledgements
We thank Natural Sciences and Engineering Research
Council of Canada for financial support through oper-
ating and equipment grants (M. J. S.) and postgradu-
ate scholarships (K. E. R. D.). We also thank
Professor R. J. Puddephatt (Western Ontario) for use
of the ESI mass spectrometer funded by the Canada
Research Chair program, Doug Hairsine (Western
Ontario) for advice and discussion on operation of the

ESI mass spectrometer, and ACD Labs for an
academic trial license for the 2D NMR processing
program (version 9.0).
References
1 Byersmann D & Hechtenberg S (1997) Cadmium, gene
regulation, and cellular signalling in mammalian cells.
Toxicol Appl Pharmacol 144, 247–261.
2 Piotrowski JK, Trojanowska B & Sapota A (1974)
Binding of cadmium and mercury by metallothionein in
the kidneys and liver of rats following repeated adminis-
tration. Archiv Toxikol 32, 351–360.
3 Harford C & Sarkar B (1991) Induction of metallothi-
onein by simultaneous administration of cadmium(II)
and zinc(II). Biochem Biophys Res Commun 177, 224–
228.
4 Din WS & Frazier JM (1985) Protective effect of metal-
lothionein on cadmium toxicity in isolated rat hepato-
cytes. Biochem J 230, 395–402.
5 Kopera E, Schwerdtle T, Hartwig A & Bal W (2004)
Co(II) and Cd(II) substitute for Zn(II) in the zinc finger
derived from the DNA repair protein XPA, demonstrat-
ing a variety of potential mechanisms of toxicity. Chem
Res Toxicol 17, 1452–1458.
6 Roesijadi G, Bogumil R, Vasak M & Kagi JHR (1998)
Modulation of DNA binding of a tramtrack zinc finger
peptide by the metallothionein–thionein conjugate pair.
J Biol Chem 273, 17425–17432.
7 Fowle DA & Stillman MJ (1997) Comparison of the
structures of the metal-thiolate binding site in Zn(II)-,
Cd(II)-, and Hg(II)-metallothioneins using molecular

modeling techniques. J Biomol Struct Dyn 14, 393–406.
8 Presta A, Fowle DA & Stillman MJ (1997) Structural
model of rabbit liver copper metallothionein. J Chem
Soc Dalton Trans 6, 977–984.
9 Robbins AH & Stout CD (1992) Crystal structure of
metallothionein. In Metallothioneins: Synthesis, Struc-
ture and Properties of Metallothioneins, Phytochelatins
and Metal-thiolate Complexes (Stillman MJ, Shaw CF
III & Suzuki KT, eds), pp. 31–54. VCH Publishers Inc,
New York, NY.
10 Boulanger Y & Armitage IM (1982)
113
Cd NMR study
of the metal cluster structure of human liver metallothi-
onein. J Inorg Biochem 17, 147–153.
11 Otvos JD & Armitage IM (1980) Structure of the metal
clusters in rabbit liver metallothionein. Proc Natl Acad
Sci USA 77, 7094–7098.
12 Richards RI, Heguy A & Karin M (1984) Structural
and functional analysis of the human metallothionein-
1A gene: differential induction by metal ions and gluco-
corticoids. Cell 37, 263–272.
A novel Cd
5
a metallothionein species K. E. Rigby Duncan et al.
2238 FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS
13 Otvos JD & Armitage IM (1979)
113
Cd NMR of metallo-
thionein: direct evidence for the existence of polynuclear

metal binding sites. J Am Chem Soc 101, 7734–7736.
14 Robbins AH, McRee DE, Williamson M, Collett SA,
Xuong NH, Furey WF, Wang BC & Stout CD (1991)
Refined crystal structure of cadmium–zinc metallothion-
ein at 2.0 A
˚
resolution. J Mol Biol 221, 1269–1293.
15 Messerle BA, Schaffer A, Vasak M, Kagi JHR &
Wuthrich K (1992) Comparison of the solution confor-
mations of human [Zn
7
]-metallothionein-2 and [Cd
7
]-
metallothionein-2 using nuclear magnetic resonance
spectroscopy. J Mol Biol 225 , 433–443.
16 Messerle BA, Schaeffer A, Vasak M, Kagi JHR &
Wuthrich K (1990) Three-dimensional structure of
human [
113
Cd
7
]metallothoinein-2 in solution determined
by nuclear magnetic resonance spectroscopy. J Mol Biol
214, 765–779.
17 Okada Y, Ohta N, Yagyu M, Min K-S, Onosaka S &
Tanaka K (1985) Synthesis of a nonacosapeptide (beta-
fragment) corresponding to the N-terminal sequence
1–29 of human liver metallothionein II and its heavy
metal-binding properties. FEBS Lett 183, 375–378.

18 Stillman MJ, Cai W & Zelazowski AJ (1987) Cadmium
binding to metallothioneins. Domain specificity in reac-
tions of a and b fragments, apometallothionein, and
zinc metallothionein with Cd2+. J Biol Chem 262,
4538–4548.
19 Stillman MJ (1995) Metallothioneins. Coord Chem Rev
144, 461–511.
20 Willner H, Vasak M & Kagi JHR (1987) Cadmium-
thiolate clusters in metallothionein: spectrophotometric
and spectropolarimetric features. Biochemistry 26, 6287–
6292.
21 Rupp H & Weser U (1978) Circular dichroism of metal-
lothioneins: a structural approach. Biochim Biophys
Acta 533, 209–226.
22 Vasak M & Kagi JHR (1983) Spectroscopic properties
of metallothionein. Met Ions Biol Syst 15, 213–273.
23 Chan J, Huang Z, Watt I, Kille P & Stillman MJ
(2007) Characterization of the conformational changes
in recombinant human metallothioneins using ESI-MS
and molecular modeling. Can J Chem 85, 898–912.
24 Rigby DuncanKE & Stillman MJ (2007) Evidence for
non-cooperative metal binding to the alpha domain of
human metallothionein. FEBS J 274, 2253–2261.
25 Felitsyn N, Peschke M & Kebarle P (2002) Origin and
number of charges observed on multiply protonated
native proteins produced by ESI. Int J Mass Spectrom
219, 39–62.
26 Chowdhury SK, Katta V & Chait BT (1990) Probing
conformational changes in proteins by mass spectrome-
try. J Am Chem Soc 112, 9012–9013.

27 Ruotolo BT & Russell DH (2004) Gas-phase conforma-
tions of proteolytically derived protein fragments: influ-
ence of solvent on peptide conformation. J Phys Chem
B 108, 15321–15331.
28 Palumaa P, Njunkova O, Pokras L, Eriste E, Jornvall
H & Sillard R (2002) Evidence for non-isostructural
replacement of Zn
2+
with Cd
2+
in the b-domain of
brain-specific metallothionein-3. FEBS Lett 527, 76–
80.
29 Haberkorn RA, Que L, Gillium WO, Holm RH, Liu
CS & Lord RC (1976) Cadmium-113 Fourier transform
nuclear magnetic resonance and Raman spectroscopic
studies of cadmium(II)–sulfur complexes, including
[Cd
10
(SCH
2
CH
2
OH)
16
]
4+
. Inorg Chem 10, 2408–2414.
30 Sadler PJ, Bakka A & Beynon PJ (1978)
113

Cd nuclear
magnetic resonance of metallothionein: non-equivalent
CdS
4
sites. FEBS Lett 94, 315–318.
31 Cols N, Romero-Isart N, Capdevila M, Oliva B, Gonz-
alez-Duarte P, Gonzalez-Duarte R & Atrian S (1997)
Binding excess cadmium(II) to Cd
7
-metallothionein
from recombinant mouse Zn
7
-metallothionein 1.
UV-VIS absorption and circular dichroism studies and
theoretical location approach by surface accessibility
analysis. J Inorg Biochem 68 , 157–166.
32 Palumaa P, Eriste E, Njunkova O, Pokras L, Jornvall
H & Sillard R (2002) Brain-specific metallothionein-3
has higher metal-binding capacity than ubiquitous
metallothioneins and binds metals noncooperatively.
Biochemistry 41, 6158–6163.
33 Capdevila M, Cols N, Romero-Isart N, Gonzalez-
Duarte R, Atrian S & Gonzalez-Duarte P (1997)
Recombinant synthesis of mouse Zn
3
-beta and
Zn
4
-alpha metallothionein 1 domains and characteri-
zation of their cadmium(II) binding capacity. Cell Mol

Life Sci 53, 681–688.
34 Palumaa P, Eriste E, Kruusel K, Kangur L, Jornvall H
& Sillard R (2003) Metal binding to brain-specific
metallothionein-3 studied by electrospray ionization
mass spectrometry. Cell Mol Biol 49, 763–768.
35 Dabrio M, Vyncht GV, Bordin G & Rodriguez AR
(2001) Study of complexing properties of the alpha and
beta metallothionein domains with cadmium and ⁄ or
zinc using electrospray ionization mass spectrometry.
Anal Chim Acta 435, 319–330.
36 Zelazowski AJ, Szymanska JA, Law AYC & Stillman
MJ (1984) Spectroscopic properties of the a fragment of
metallothionein. J Biol Chem 259, 12960–12963.
37 Otvos JD & Armitage IM (1982) Elucidation of metal-
lothionein structure by
113
Cd NMR. In Biochemical
Structure Determination by NMR (Bothner-By AA,
Glickson JD & Sykes BD, eds), pp. 65–96. Marcel
Dekker Inc, New York, NY.
38 Armitage IM, Otvos JD, Briggs RW & Boulanger Y
(1982) Structure elucidation of the metal-binding sites
in metallothionein by
113
Cd NMR. Fed Proc 41, 2974–
2980.
K. E. Rigby Duncan et al. A novel Cd
5
a metallothionein species
FEBS Journal 275 (2008) 2227–2239 ª 2008 The Authors Journal compilation ª 2008 FEBS 2239