Kinetic analysis of arsenic-metalation of human metallothionein: Significance of the two-domain structure

Article abstract of DOI:10.1021/ja8060326

Metallothionein (MT) is ubiquitous in Nature, underlying MT's importance in the cellular chemistry of metals. Mammalian MT consists of two metal-binding domains while microorganisms like cyanobacteria consist of a single metal-binding domain MT. The evolution of a two-domain protein has been speculated on for some time; however, no conclusive evidence explaining the evolutionary necessity of the two-domain structure has been reported. The results presented in this report provide the complete kinetic analysis and subsequent mechanism of the As³⁺-metalation of the two-domain βαhMT and the isolated single domain fragments using time- and temperature-resolved electrospray ionization mass spectrometry. The mechanism for βαhMT binding As³⁺ is noncooperative and involves six sequential bimolecular reactions in which the ∞ domain binds As ³⁺ first followed by the β domain. At room temperature (295 K) and pH 3.5, the sequential individual rate constants, k_n (n = 1-6) for the As³⁺-metalation of βαhMT starting at k _1βα are 25, 24, 19, 14, 8.7, and 3.7 M^-1s ^-1. The six rate constants follow an almost linear trend directly dependent on the number of unoccupied sites for the incoming metal. Analysis of the temperature-dependent kinetic electrospray ionization mass spectra data allowed determination of the activation energy for the formation of As ₁-H₁₇-βαhMT (14 kJ mol^-1) and As_2-6-βαhMT (22 kJ mol^-1). On the basis of the increased rate of metalation for the two-domain protein when compared with the isolated single-domain, we propose that there is an evolutionary advantage for the two-domain MT structures in higher organism, which allows MT to bind metals faster and, therefore, be a more efficient metal scavenger.

Full text of DOI:10.1021/ja8060326

Published on Web 11/19/2008
Kinetic Analysis of Arsenic-Metalation of Human
Metallothionein: Significance of the Two-Domain Structure
Thanh T. Ngu, Alexandria Easton, and Martin J. Stillman*
Department of Chemistry, The UniVersity of Western Ontario, London,
Ontario, Canada, N6A 5B7
Received July 31, 2008; E-mail: Martin.Stillman@uwo.ca
Abstract: Metallothionein (MT) is ubiquitous in Nature, underlying MT’s importance in the cellular chemistry
of metals. Mammalian MT consists of two metal-binding domains while microorganisms like cyanobacteria
consist of a single metal-binding domain MT. The evolution of a two-domain protein has been speculated
on for some time; however, no conclusive evidence explaining the evolutionary necessity of the two-domain
structure has been reported. The results presented in this report provide the complete kinetic analysis and
subsequent mechanism of the As3 -metalation of the two-domain ꢀRhMT and the isolated single domain
fragments using time- and temperature-resolved electrospray ionization mass spectrometry. The mechanism
+
3+
for ꢀRhMT binding As is noncooperative and involves six sequential bimolecular reactions in which the
R domain binds As3 first followed by the ꢀ domain. At room temperature (295 K) and pH 3.5, the sequential
+
+
individual rate constants, k
8
n
(n ) 1-6) for the As3 -metalation of ꢀRhMT starting at k1ꢀR are 25, 24, 19, 14,
-
1 -1
.7, and 3.7 M
of unoccupied sites for the incoming metal. Analysis of the temperature-dependent kinetic electrospray
ionization mass spectra data allowed determination of the activation energy for the formation of As -H17-
s . The six rate constants follow an almost linear trend directly dependent on the number
1
-
1
-1
ꢀ
RhMT (14 kJ mol ) and As2-6-ꢀRhMT (22 kJ mol ). On the basis of the increased rate of metalation for
the two-domain protein when compared with the isolated single-domain, we propose that there is an
evolutionary advantage for the two-domain MT structures in higher organism, which allows MT to bind
metals faster and, therefore, be a more efficient metal scavenger.
Introduction
the differential selectivity for metalation of the two domains.
1
-8
For example, it has been reported that the ꢀ domain preferen-
Metalated and demetalated metallothioneins (MT)
have
+
2+
tially binds Cu while the R domain preferentially binds Zn
under these conditions and that each cluster forms independently
been documented since 1957 when Vallee and Margoshes first
described horse kidney MT. Mammalian MT consists of 20
cysteines structured into two domains, often described as
dumbbell-shaped from the NMR and X-ray structures.
9
12-15
of constraints or influences from the other cluster.
However,
2
-5,10,11
another study noted that while the two domains assembled
independently, the cumulative properties of each domain were
A 3-metal, 9-cysteine ꢀ cluster and a 4-metal, 11-cysteine R
1
3
not sufficient to describe MT structurally or functionally. Yet
studies of the evolutionary distant sea urchin MT have shown
a conservation of the two-domain structure, supporting previous
conclusions regarding the functional importance of this structural
cluster form following metalation with the divalent Group 12
metals Zn2 and Cd . The significance of the two domains in
+
2+
MT has long been discussed with frequent suggestions invoking
1
6,17
motif.
We report in this paper new kinetic data that may
(
(
(
(
(
(
1) Ngu, T. T.; Stillman, M. J. J. Am. Chem. Soc. 2006, 128, 12473–
1
2483.
account for the evolution of a two-domain protein in higher
life-forms rather than the single-domain proteins that are
2) Kagi, J. H. R.; Nordberg, M. Metallothionein, 1st ed.; Birkhauser:
Boston, 1978; Vol. 1.
3) Kagi, J. H. R.; Kojima, Y. Metallothionein II, 1st ed.; Birkhauser:
Boston, 1985; Vol. 2.
4) Stillman, M. J.; Shaw, C. F., III.; Suzuki, K. T. Metallothioneins, 1st
ed.; VCH Publishers: New York, 1992.
5) Suzuki, K. T.; Imura, N.; Kimura, M. Metallothionein III, 1st ed.;
Birkhauser: Boston, 1993; Vol. 3.
6
18,19
observed in lower life-forms like cyanobacteria, E. fetida,
(12) Nielson, K. B.; Winge, D. R. J. Biol. Chem. 1984, 259, 4941–4946.
(13) Jiang, L.-J.; Vasak, M.; Vallee, B. L.; Maret, W. Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 2503–2508.
6) Blindauer, C. A.; Polfer, N. C.; Keiper, S. E.; Harrison, M. D.;
Robinson, N. J.; Langridge-Smith, P. R. R.; Sadler, P. J. J. Am. Chem.
Soc. 2003, 125, 3226–3227.
(14) Nielson, K. B.; Winge, D. R. J. Biol. Chem. 1985, 260, 8698–8701.
(15) Li, H.; Otvos, J. D. J. Inorg. Biochem. 1998, 70, 187–194.
(16) Riek, R.; Precheur, B.; Wang, Y.; Mackay, E. A.; Wider, G.; Guntert,
P.; Liu, A.; Kagi, J. H. R.; Wuthrich, K. J. Mol. Biol. 1999, 291,
417–428.
(
(
7) Krezel, A.; Maret, W. J. Am. Chem. Soc. 2007, 192, 10911–10921.
8) Dabrio, M.; Van Vyncht, G.; Bordin, G.; Rodriguez, A. R. Anal. Chim.
Acta 2001, 435, 319–330.
(17) Vasak, M.; Hasler, D. W. H. Curr. Opin. Chem. Biol. 2000, 4, 177–
183.
(
9) Margoshes, M.; Vallee, B. L. J. Am. Chem. Soc. 1957, 79.
(
10) Wehrli, S.; Engeseth, H. R.; Otvos, J. D. Biochemistry 1985, 24, 6735–
(18) Gehrig, P. M.; You, C.; Dallinger, R.; Gruber, C.; Brouwer, M.; Kagi,
J. H. R.; Hunziker, P. E. Protein Sci. 2000, 9, 395–402.
(19) Gruber, C.; Sturzenbaum, S. R.; Gehrig, P. M.; Sack, R.; Hunziker,
P. E.; Berger, B.; Dallinger, R. Eur. J. Biochem. 2000, 267, 573–582.
6
740.
11) Otvos, J. D.; Armitage, I. M. Proc. Natl. Acad. Sci. U.S.A. 1980, 77,
094–7098.
(
7
1
7016 9 J. AM. CHEM. SOC. 2008, 130, 17016–17028
10.1021/ja8060326 CCC: $40.75  2008 American Chemical Society

Metal-Induced Folding of Metallothionein
A R T I C L E S
2
0-22
and funga.
These single-domain proteins have been shown
structure because for the first time we report a property of MT
that directly relates to the presence of the two domains. We
report the rate constant for the first As3 bound is 65% faster
for the two-domain protein than for the single domain fragments
and that there is an almost linear correlation between the binding
rates for the incoming metal and the number of available sites.
Hence, a two-domain protein binds metals dramatically faster
than the individual domains. Assuming a constant koff, we can
for the first time, contrast that the k1ꢀR (for the first metal bound)
is 6.8× greater than k6ꢀR (for the last metal bound).
to contain a highly conserved central segment that correlate to
the domain center of mammalian MT and may be considered
the core upon which many MTs have evolved.
Metal homeostasis, which is necessary for survival of all
organisms, requires buffering of metals via both metalation
reactions that store the metals and metal exchange reactions
into apoproteins. One of MT’s suggested functions involves
metal homeostasis; MT appears to play a role in zinc homeo-
stasis as the Zn2 buffering capacity of MT may control cellular
+
2
3
+
2
4
Zn availability. This buffering capacity of MT is based on
MT’s ability to undergo metalation, demetalation and metal
exchange reactions. The steady state metalation/demetalation
equilibria for MTs are well-known for a large number of
Experimental Methods
Materials and Methods. Experimental procedures have previ-
1
ously been published; please refer to Ngu et al. for further details.
7
,15,25-28
metals
and numerous metal exchange studies with MT
Recombinant human metallothionein (hMT) was expressed in
BL21(DE3) Escherichia coli cells that were transformed using a
pP-1 plasmid which contains an N-terminal S-tag (MKETAAAKFE
6
,13,29,30
have been documented.
While metal exchange in MT
has been studied extensively, the detailed mechanism of
metalation showing all intermediate species has been hard to
determine, mainly because of a lack of probes that can
differentiate the binding of more than one metal in dilute
solutions. Recent studies have shown that mass spectrometry
is able to discriminate between different intermediates and
further, quantify the relative concentrations of all species
3
8,39
RQHMDSPDLG TLVPRGS) as previously described.
The
S-tag was removed using a Thrombin CleanCleave Kit (Sigma).
The RhMT, ꢀhMT, and ꢀRhMT proteins used in this study were
based on the 43-residue, 40-residue, and 74-residue sequences,
respectively. The sequence for RhMT is GSMGKAAAAC CSC-
CPMSCAK CAQGCVCKGA SEKCSCCKKA AAA; for ꢀhMT
is GSMGKAAAAC SCATGGSCTC TGSCKCKECK CNSCK-
KAAAA, and for ꢀRhMT is GSMGKAAAAC SCATGGSCTC
TGSCKCKECKCNSCKKAAAACCSCCPMSCAKCAQGCVCKG
ASEKCSCCKK AAAA. There are 11, 9, and 20 cysteine residues
present in RhMT, ꢀhMT and ꢀRhMT, respectively, and no disulfide
bonds. The protein was further purified and demetalated by elution
through a Sephadex G25 column with a 20 mM ammonium formate
buffer at pH 2.7 (Fisher). Elution was monitored by UV-visible
absorption spectroscopy at the characteristic metal-free (apo) MT
wavelengths of 300 to 200 nm. Fractions that contained metal-free
MT were collected and purity was checked using ESI-MS. Previous
reports have suggested or shown the existence of apoprotein in
3
1-35
36
present
for kinetic experiments.
Shaw and Petering have reported that Cd and Zn2+
metalation is largely complete at pH 7 and room temperature
within the 4 ms dead-time of the stopped-flow UV-visible
spectometer used. On the other hand arsenic-metalation of
MT occurs on a time-scale of minutes and is observable by
2+
3
7
1
mass spectrometry making it an ideal reaction to study to obtain
a detailed understanding of the metal-induced folding reactions
3+
of metallothionein. We report the first detailed As -metalation
study on the two-domain ꢀR human metallothionein showing
all intermediate As-species with correlations to the native
individual single domain fragments. The data analysis reported
here provides insight into the evolution of the two-domain
2
4,40-48
ViVo
making the study of apoprotein relevant as a starting
3+
point in exploring the metalation reactions of As . The apo-RhMT
protein concentrations were determined from the extinction coef-
ficients of 40,000 Lmol cm at 220 nm, the Cd
concentrations were determined from the extinction coefficient of
36,000 Lmol cm at 250 nm, and the Cd
was determined from the extinction coefficient of 115,000
Lmol cm at 250 nm. The protein concentration for the kinetic
ESI-MS experiment using the thermostatted-mixing tee was 19.5
µM of apo-ꢀRMT and concentrations for the timed-resolved ESI-
MS experiments were 18 µM for apo-ꢀRMT, 23 µM for apo-RMT
and a 30 µM for apo-ꢀMT. Oxidation is a significant problem with
solutions of MT in these experiments and the apo-proteins were
maintained in their reduced state by carefully deoxygenating the
-
1
-1
3
-ꢀhΜΤ protein
(
20) Claderone, V.; Dolderer, B.; Hartmann, H.-J.; Echner, H.; Luchinat,
-1
1
C.; Bianco, C. D.; Mangani, S.; Weser, U. Proc. Natl. Acad. Sci. U.S.A.
7
-ꢀR protein concentration
2
005, 102, 51–56.
(
21) Munger, K.; Germann, U. A.; Lerch, K. EMBO J. 1985, 4, 2665–
-1
-1
2
668.
(
22) Munger, K.; Lerch, K. Biochemistry 1985, 24, 6751–6756.
23) Nemer, M.; Wilkinson, D. G.; Travaglini, E. C.; Sternberg, E. J.; Butt,
T. R. Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 4992–4994.
(
(
(
(
24) Krezel, A.; Maret, W. J. Biol. Inorg. Chem. 2008, 13, 401–409.
25) Rigby Duncan, K. E.; Stillman, M. J. FEBS J. 2007, 274, 2253–2261.
26) Chan, J.; Huang, Z.; Watt, I.; Kille, P.; Stillman, M. J. Can. J. Chem.
2
007, 85, 898–912.
(
(
(
(
(
(
(
(
(
(
(
27) Shaw, C. F.; He, L.; Munoz, A.; Savas, M. M.; Chi, S.; Fink, C. L.;
Gan, T.; Petering, D. H. J. Biol. Inorg. Chem. 1997, 2, 65–73.
28) Palumaa, P.; Eriste, K.; Kruusel, K.; Kangur, L.; Joernvall, H.; Sillard,
R. Cell. Mol. Biol. 2003, 49, 763–768.
(38) Merrifield, M. E.; Huang, Z.; Kille, P.; Stillman, M. J. J. Inorg.
Biochem. 2002, 88, 153–172.
(39) Chan, J.; Huang, Z.; Merrifield, M. E.; Salgado, M. T.; Stillman, M. J.
Coord. Chem. ReV. 2002, 233-234, 319–339.
29) Salgado, M. T.; Bacher, K. L.; Stillman, M. J. J. Biol. Inorg. Chem.
2
007, 12, 294–312.
30) Nettesheim, D. G.; Engeseth, H. R.; Otvos, J. D. Biochemistry 1985,
4, 6744–6751.
31) Hathout, Y.; Fabris, D.; Fenselau, C. Int. J. Mass Spectrom. 2001,
04, 1–6.
32) Yu, X.; Wojciechowski, M.; Fenselau, C. Anal. Chem. 1993, 65, 1355–
359.
33) Zaia, J.; Fabris, D.; Wei, D.; Karpel, R. L.; Fenselau, C. Protein Sci.
998, 7, 2398–2404.
34) Daneshfar, R.; Kitova, E. N.; Klassen, J. S. J. Am. Chem. Soc. 2004,
(40) Krezel, A.; Maret, W. Biochem. J. 2007, 402, 551–558.
(41) Pattanaik, A.; Shaw, C. F. I.; Petering, D. H.; Garvey, J. S.; Kraker,
A. J. J. Inorg. Biochem. 1994, 54, 91–105.
2
(42) Rigby, K. E.; Chan, J.; Mackie, J.; Stillman, M. J. Proteins: Struct.,
Funct., Bioinf. 2006, 62, 159–172.
2
(43) Rigby, K. E.; Stillman, M. J. Biochem. Biophys. Res. Commun. 2004,
325, 1271–1278.
1
(44) Hase, H.; Maret, W. Anal. Biochem. 2004, 333, 19–26.
(45) Yang, Y.; Maret, W.; Vallee, B. L. Proc. Natl. Acad. Sci. U.S.A. 2001,
98, 5556–5559.
1
1
26, 4786–4784.
(46) Shapiro, S. G.; Squibb, K. S.; Markowitz, L. A.; Cousins, R. J.
Biochem. J. 1978, 175, 833–840.
35) Shoemaker, G. K.; Kitova, E. N.; Palcic, M. M.; Klassen, J. S. J. Am.
Chem. Soc. 2007, 129, 8674–8675.
(47) Krezoski, S. K.; Villalobos, J.; Shaw, C. F. I.; Petering, D. H. Biochem.
J. 1998, 255, 483–491.
36) Wang, W.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2003, 75, 4945–
4
955.
(48) Petering, D. H.; Zhu, J.; Krezoski, S. K.; Meeusen, J.; Kiekenbush,
C.; Krull, S.; Specher, T.; Dughish, M. Exp. Biol. Med. 2006, 231,
1528–1534.
37) Ejnik, J.; Robinson, J.; Zhu, J.; Forsterling, H.; Shaw, C. F.; Petering,
D. H. J. Inorg. Biochem. 2002, 88, 144–152.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008 17017

A R T I C L E S
Ngu et al.
samples using evacuation and then saturation with argon gas such
that the proteins were contained within a sealed, inert environment.
All protein samples were buffered in 20 mM ammonium formate
at pH 2.7.
Caution: Arsenic is a highly toxic reagent. Caution should be
used when handling arsenic and its deriVatiVes. Solutions of 5 mM
3
+
2 3
As at pH 3.5 were prepared by dissolving As O (AnalaR) in
concentrated HCl (Caledon) and diluting with ultra pure deionized
Figure 1. Schematic diagram of the continuous flow mixing set up for
temperature-resolved ESI-MS measurements. Refer to the text for details.
4
water (Barnstead). Aliquots of 7.4 M NH OH (Fisher) were used
to raise the pH of the As3 solutions. As solutions were
deoxygenated and contained within a sealed inert environment.
Reactions were carried out at a pH range of 3-4 as this was the
optimum range for product formation and stabilization; this pH was
+
3+
Data Analysis. We will briefly describe the rationale behind the
data analysis used in this study; further information can be found
1
in Ngu et al. The reactions shown in Figures 2, S2 and S3, all
4
9
50
also suggested by Rey et al. and Serves et al. in order to stabilize
proceed to completion with the formation of As
As -H -RhMT or As -ꢀhMT, respectively, with excess As added.
Under the nonequilibrium conditions of excess As (greater than
0X excess) and the short reaction times used for the kinetic
6
-H
2
-ꢀRhMT,
3
+
the As . We should mention that the structure of apo-MT in the
pH range of 3-4 may have a structure different from that observed
at neutral pH, which will require further study. It has been shown
that MT is located in the acidic lysosomes of kidneys
immunostaining in earthworms indicates that Cd-MT is located in
3+
3
2
3
3+
1
5
1,52
and MT
measurements reported in Figures 2-6, the reverse reactions are
not significant. A series of sequential second order reactions,
Scheme 1, are proposed to account for the observed kinetic data in
53
acidic subcellular structures similar to lysosome, making the study
of As3 binding at low pH of interest. High salt concentrations
could not be used in conjunction with the ESI-MS measurements.
ESI-MS Procedures. All data were collected using a Micro-
mass LCT mass spectrometer in the positive ion mode. The mass
spectrometer was operated using the parameters: 3000.0 V capillary,
+
3+
Figures 3, 6, and 7. Scheme 1C shows the proposed order of As
binding to the domains in ꢀRhMT. Data analysis was carried out
by modeling a second order, bimolecular, irreversible mechanism
56-58
with the program Gepasi.
The kinetic data for ꢀRhMT were
fitted using the six sequential bimolecular reactions of this scheme.
This approach will be discussed in greater detail below. It is clear
from inspection of the modeled data compared with the experi-
mental data that the second order mechanism proposed for each
step of the sequential metalation, Scheme 1, fits the experimental
data exceptionally well (vide infra), Figure 3.
4
2.0 V sample cone, 4.0 V extraction cone, acquisition scan time
of 4 s, and interscan delay time of 0.1 s. The ESI-MS data were
processed and deconvoluted using the MaxEnt I software (Micro-
mass). ESI-MS measurements were obtained using a wide scan
range (700-2000 m/z) and MS mass calculations were integrated
from a minimum of 10 scans.
The measured charge state data or the deconvoluted mass spectral
data were used for the kinetic analyses. Yu et al. have shown that
relative concentration of various Cd-MT species can be estimated
Time-Resolved ESI-MS Measurements. Kinetic data were
3
+
obtained by adding 10×, 15×, and 11× excess As to solutions
of apo-RhMT, apo-ꢀhMT, and apo-ꢀRhMT, respectively. The
reaction was monitored by ESI-MS for up to110 min at 25 ( 1
3
2
from the relative abundances of the ions in the ESI spectrum,
and this approach was previously shown to be reliable for analysis
1
°
C.
Temperature-Resolved ESI-MS Measurements. Kinetic data
of kinetic data of the isolated domains. The relative abundances
of all the MT species observed in each mass spectrum were summed
and normalized. In the time-resolved experiments, the relative
abundances were plotted against time and fitted to calculate the kn
(n ) 1–6 for each step in the complete metalation reaction). In the
temperature-resolved experiments, the normalized relative abun-
dances in each spectrum, following a specific reaction mixing time,
were obtained using a thermostatted mixing “tee” attached directly
to the input capillary of the ESI-MS instrument via a reaction
1
capillary. The reaction mixing time was dependent on the length,
5
4,55
diameter and flow rate within the reaction capillary
and the
short length of the input capillary. The exchangeable reaction
capillary used had an inner diameter of 75 µm, with lengths ranging
from 30 to 148 cm, and together with the input capillary gave
reaction times of 99 to 498 s, all of which was contained within a
temperature-controlled water bath. The input capillary did affect
the average temperature of the reaction, particularly at the extreme
high or low temperatures and short reaction times, leading to those
data sets including slightly greater uncertainties. The temperature
range used was 273 to 344 K. Kinetic data were obtained by mixing
the two streams of solutions as shown in Figure 1 with a flow rate
of 2 µL/min for the 2.5 mM As3 solutions and 8 µL/min for the
protein solutions, for a total flow rate of 10 µL/min, which results
in a final concentration after mixing of 15.6 µM for ꢀRhMT and a
-
1
were then plotted versus 1/T (K ), with T ranging from 273 to
344 K for each solution. As kinetic data are more readily analyzed
from concentration data as a function of time at a constant
temperature, our kinetic data were rearranged into this format. This
serves two purposes: first, the analysis follows traditional procedures
and second, the data may be directly compared to kinetic data
obtained as a function of time at a fixed temperature. In addition,
the data used in the subsequent analyses are averages of many
different and independent data sets, which greatly improves the
confidence in the calculated results. The data sets were combined
in terms of ln(Relative Abundance) and appropriate interpolation
points were calculated for construction of 3-dimensional graphs
with the axes ln(Relative Abundance) (y-axis), time (s) (z-axis) and
+
3
+
ratio for As :ꢀRhMT of 32:1. The temperatures reported are an
average with uncertainties of (1 K.
-
1
1/T (K )(x-axis). These 3D plots are shown below in Figure 4.
From these 3D plots, slices were obtained of relative abundance
as a function of time at fixed temperatures. These data sets were
fitted according to Scheme 1 to determine the specific temperature-
dependent rate constants. Finally, the reliability of the method was
confirmed by back-calculating the relative abundances of all species
as a function of reaction time and temperature for comparison
directly with the measured data, Figures 5 and 6.
(
(
(
49) Rey, N. A.; Howarth, O. W.; Pereira-Maia, E. C. J. Inorg. Biochem.
004, 98, 1151–1159.
2
50) Serves, S. V.; Charalambidis, Y. C.; Sotiropoulos, D. N.; Ioannou,
P. V. Phoshorus, Sulfur Silicon Relat. Elem. 1995, 105, 109–116.
51) Klein, D.; Lichtmannegger, J.; Heinzmann, U.; Muller-Hocker, J.;
Michaelsen, S.; Summer, K. H. Eur. J. Clin. InVest. 1998, 28, 302–
3
10.
Molecular Models. MM3/MD calculations were carried out
using CAChe Workstation Pro (6.1.1) Software (Fujitsu America)
(
(
52) Freedman, J. H.; Powers, L.; Peisach, J. Biochemistry 1986, 25, 2342–
2
349.
53) Morgan, A. J.; Sturzenbaum, S. R.; Winters, C.; Grime, G. W.; Aziz,
N. A. A.; Kille, P. Ecotoxicol. EnViron. Saf. 2004, 57, 11–19.
(56) Mendes, P. Comput. Applic. Biosci. 1993, 9, 563–571.
(57) Mendes, P. Trends Biochem. Sci. 1997, 22, 361–363.
(58) Mendes, P.; Kell, D. B. Bioinformatics 1998, 14, 869–883.
(
(
54) Attwood, P. V.; Geeves, M. A. Anal. Biochem. 2004, 334, 382–389.
55) Wilson, D. J.; Konermann, L. Anal. Chem. 2004, 76, 2537–2543.
1
7018 J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008

Metal-Induced Folding of Metallothionein
A R T I C L E S
Figure 2. Time dependence of the ESI mass spectra of 18 µM apo-ꢀRhMT in the presence As in a stoichiometric ratio of 11:1 for As3+: ꢀRhMT. The
3+
sample was continuously monitored using ESI-MS for over 100 min at 296 K and pH 3.5.
and parametrized using the modified force field described by Chan
sequentially along the residue sequence starting with the first 3 Cys
residues on the C-terminal which for RhMT would be the terminal
most unlikely to be hindered in the ꢀRhMT protein and for ꢀRhMT
was the most reasonable based on the data presented later.
2
6,59
et al.
with the dielectric constant of 80 for water to obtain
energy minimized structures of the apo- and As-bound ꢀhMT,
RhMT and ꢀRhMT proteins. An average bond length for an As-
SR (R ) any carbon atom) bond of 2.196 Å and an average RS-
As-SR bond angle of 107.91° were used for the modeling
calculation and which are consistent with As-SR crystal structures
Results
6
0
Figures 2, S1, and S2 (Supporting Information) show the
time-resolved ESI-MS spectra (with the charge states stated
below the topmost spectra) measured between 2 min and up to
for bond length and bond angle. The original apo MT structures
26
were obtained from Chan et al. and a cycle of MM3 minimizations
followed by MD calculations gave the energy-minimized apo MT
3
+
3+
structures reported here. Sequential addition of As provided the
100 min for the As -metalation of apo-ꢀRhMT, apo-ꢀhMT and
3
+
As -bound structures as follows: the structures were first energy-
minimized using the MM3 calculation followed by an MD
simulation at 50 K for 100 ps and then another two MD simulations
apo-RhMT. Analyses of the ESI-MS data shown in Figures 2,
S1, and S2 (Supporting Information) show a maximum of three
3+
3+
As bound to each isolated domain and six As bound to the
ꢀ
3
at 300 K for 1000 ps. Arsenic atoms were modeled using an sp
1,61,62
RhMT protein, which corresponds to previous reports.
coordination and were bound sequentially as shown in Scheme 1.
For ꢀhMT the As atoms were bound sequentially along the residue
sequence starting with the first 3 Cys residues on the N-terminal
which would be the terminal most unlikely to be hindered in the
Most proteins, including metallothionein, exhibit a range of
charge states due to the presence of a range of multiply charged
ions. The number of charges is dependent on the number of
ꢀRhMT protein. For RhMT and ꢀRhMT, the As atoms were bound
(
61) Jiang, G.; Gong, Z.; Li, X.-F.; Cullen, W. R.; Le, X. C. Chem. Res.
Toxicol. 2003, 16, 873–880.
(
(
59) Chan, J.; Merrifield, M. E.; Soldatov, A. V.; Stillman, M. J. Inorg.
Chem. 2005, 44, 4923–4933.
60) Iyer, R. G.; Kanatzidis, M. G. Inorg. Chem. 2004, 43, 3656–3662.
(62) Toyama, M.; Yamashita, M.; Hirayama, N.; Murooka, Y. J. Biochem.
(Tokyo) 2002, 132, 217–221.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008 17019

A R T I C L E S
Ngu et al.
Scheme 1. Sequential Binding Mechanism Proposed for As3
+
a
Binding to the (Α) RhMT, (B) ꢀhMT, and (C) ꢀRhMT Proteins
a
The rate constants for each step are indicated by k1-3 (R or ꢀ) or k1-6 (ꢀR)
3
+
which are defined as the rate of adding a single As to the metallothionein
proteins to form a product with n (1-6) As bound. This same
nomenclature is used in Table 1 where the experimental values are reported.
3
+
The kinetic data were fitted for ꢀRhΜΤ (C) as six sequential bimolecular
reactions without consideration of which domain was binding the As3
+
.
2
6,63,64
in conformation of the protein
and suggests that the two
protein fragments have rearranged into a less solvent-accessible
6
5
conformation. Figure 2 shows that apo-ꢀRhMT has a charge
state distribution of +4 to +8 with two dominant charge state
maxima, at +5 and +7, suggesting two conformations exist in
3+
solution. Upon As -metalation, the appearance of a dominant
charge state of +5 for As1-4-ꢀRhMT suggests a more closed,
solvent-inaccessible structure. Further As-metalation results in
a charge state shift to +6 for As5-6-ꢀR-hMT which is consistent
with the trend observed in Cd2 metalation of ꢀRhMT.
+
66
Figure 3. Time-resolved ESI-MS relative abundances for apo-RMT (A),
3
+
apo-ꢀMT (B), and apo-ꢀRMT (C) following reaction with As at 25 °C
The deconvoluted measured masses correlate with the theo-
retical masses based on the sequence shown in the Experimental
Methods above. The estimated masses based on the metal
and pH 3.5 to form As -MT (n ) 1-3 for the fragments and n ) 1-6 for
n
3+
ꢀ
RhMT). Reaction was carried out with an As :MT stoichiometric ratio
of 10:1 for the RhMT, a ratio of 15:1 for the ꢀhMT, and a ratio of 11:1 for
the ꢀRhMT. Solutions of 23 µM apo-RhMT, 30 µM apo-ꢀhMT, 18 µM
apo-ꢀRhMT and a 5 mM As solution were used. The relative abundances
for each species (apo-RhMT, apo-ꢀhMT, and apo-ꢀRhMT) are shown as
data points on the graphs with different symbols indicating each component
species. The smooth lines are calculated based on the complete analysis of
the kinetic data for the relative abundance at the specified times. In each of
content and amino acid sequence for H11-RhMT, As
RhMT, As -H -RhMT, and As -H -RhMT are 4083.0, 4154.9,
226.8, and 4298.8 Da, respectively, and the estimated masses
based on the metal content and amino acid sequence for H
hMT, As -H -ꢀhMT, As -H -ꢀhMT, and As -ꢀhMT are
1 8
-H -
3+
2
5
3
2
4
9
-
ꢀ
1
6
2
3
3
(A), (B), and (C), the lines are connected by the series of sequential reactions
3753.5, 3825.4, 3897.3, and 3969.2 Da, respectively, and last
the estimated masses based on the metal content and amino acid
shown in Scheme 1 and described in the text.
protonatable sites, the conformation of the protein, and the pH
of the solution. Figures S1 and S2 (Supporting Information)
(
(
63) Dobo, A.; Kaltashov, I. A. Anal. Chem. 2001, 73, 4763–4773.
64) Griffey, R. H.; Sannes-Lowery, K. A.; Drader, J. J.; Mohan, V.;
Swayze, E. E.; Hofstadler, S. A. J. Am. Chem. Soc. 2000, 122, 9933–
9938.
1
show charge states of +3 and +4 for apo-ꢀhΜΤ, and +3, +4,
3+
and +5 for apo-RhMT. Following As -metalation there is a
dramatic shift to a nearly exclusive charge state of +4 for both
species. This change in dominant charge state reflects a change
(
65) Fenn, J. B. J. Am. Soc. Mass Spectrom. 1993, 4, 524–535.
66) Sutherland, D. E. K.; Stillman, M. J. Biochem. Biophys. Res. Commun.
2008, 372, 840–844.
(
1
7020 J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008

Metal-Induced Folding of Metallothionein
A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008 17021

A R T I C L E S
Ngu et al.
Figure 5. Comparison of experimental and theoretical relative abundances
calculated for ꢀRhMT at 99 s, 287 K (A) and 402 s, 323 K (B) based on
values from Table 1 and using the model shown in Scheme 1 to provide
3
+
the predicted relative abundances. Conditions used were an As : ꢀRhMT
stoichiometric ratio of 32:1. Solutions containing 15.6 µM apo-ꢀRhMT and
.5 mM As3 were used for the measurements. The theoretical values from
+
0
the simulation are within 10% of the experimental values. The experimental
values are subject to an estimated 10% error.
sequence for H20-ꢀR hMT, As
As -H11-ꢀRhMT,As -H -ꢀRhMT,As
RhMT are 6874.4, 6946.3, 7018.2, 7090.1, 7162.0, 7233.9,
1
-H17-ꢀRhMT, As
2
-H14-ꢀRhMT,
3
4
8
5
-H -ꢀRhMT,andAs
5
6
-H -
2
ꢀ
and 7305.9 Da, respectively. The masses calculated for ꢀRhMT
are based on a truncated sequence resulting from the loss of
the N-terminal G residue and the C-terminal AA residues
without any loss of cysteine residues or change in the metal-
binding capabilities of the protein. The ESI-MS measured
masses closely match the theoretical masses for these species.
The mass difference between the peaks in the same charge state
is approximately equal to the mass of one As atom minus three
3
+
protons, once again indicating that each As binds three
thiolates and displaces three protons.
1
3+
Figure 3 shows the As -metalation of apo-RhMT (A), apo-
ꢀhMT (B) and apo-ꢀRhMT (C) in terms of the time-dependence
3
+
of the relative abundances for the apo- and As -bound species
in the ESI-mass spectra at 25 °C. The relative abundance data
points shown were obtained following normalization of the
data for all species present at any single time. The experimental
data were fitted (Figure 3 smooth lines) using Scheme 1 with
the calculated rate constants reported in Table 1. The reaction
is complete for the individual domains before 5000 s, while
the 2-domain protein saturates with As3+ after 6000 s. The
Figure 6. Temperature-resolved relative abundances of apo-ꢀRhMT and
n
3+
As -ꢀRhMT (n ) 1-6) following reaction of apo-ꢀRhMT with As at
fixed reaction times. ESI-MS data were obtained for apo-ꢀRhMT at a series
of fixed reaction times with increasing temperatures (273 - 344 K) in the
3+
presence of excess As . The fixed reaction times were 99 s (A), 191 s
+
(
B), 402 s (C), and 498 s (D). Reaction was carried out with an As3
:
ꢀ
RhMT stoichiometric ratio of 32:1. Solutions containing 15.6 µM apo-
3+
ꢀ
RhMT and 0.5 mM As were used for the measurements. The smooth
lines were calculated based on the complete analysis of the kinetic data for
the relative abundance of the individual component species for each
temperature at the specified times. The lines are connected by a series of
six sequential reactions shown in Scheme 1 and described in the text. The
data points that comprise the theoretical lines were calculated from analysis
of data sets measured at all reaction times and temperatures to obtain the
a
3+
reaction of the two-domain apo-ꢀRhMT with excess As was
fitted using six sequential bimolecular reactions without con-
3+
sideration as to which domain the As was binding to. The
identity of the binding domain cannot be determined readily
by ESI-MS as the technique can only define the number of
arsenics bound based on the mass.
kinetic parameters: ktemp, A, E , which were used with the experimental
3
+
concentrations of the protein and the As to predict the concentration of
each species at the specified temperature and time. There will always be
considerable uncertainty in any simulation that uses parameters extracted
from the entire temperature-time-relative abundance data set that comprise
many different experiments. The experimental values are subject to an
estimated 20% error.
3+
It has previously been shown by our group that the As -
metalation of metallothionein is time and temperature dependent
for the ꢀhMT and RhMT protein fragments containing the
1
additional S-tag peptide. Further, using ESI-MS and a
1
7022 J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008

Metal-Induced Folding of Metallothionein
A R T I C L E S
q
q
A
Table 1. Rate Constants, Activation Energies (E ), Arrhenius Factor (A), Activation Enthalpies (∆H ), Activation Entropies (∆S ), and
q
3+
Activation Free Energies (∆G ) for the As -Induced Metalation of the ꢀRhMT
a
a
n
-1 -1
b
A
-1
b
A
4
-1
q
-1
q
-1
-1
q
-1
Arsenic species formed
n
k
at 298 K M
s
E
kJmol
× 10 s
∆H kJmol
∆S JK mol
∆G at 298 K kJmol
As
1
As
2
As
3
As
4
As
5
As
6
-H17-ꢀRhMT
-H14-ꢀRhMT
-H11-ꢀRhMT
1ꢀR
2ꢀR
3ꢀR
4ꢀR
5ꢀR
6ꢀR
25.2 ( 0.7
24.1 ( 0.8
19.5 ( 0.6
14.1 ( 0.3
8.7 ( 0.2
3.7 ( 0.1
14 ( 1
22 ( 1
22 ( 1
22 ( 1
22 ( 1
22 ( 1
0.87 ( 0.05
20 ( 1
18.0 ( 0.9
14.0 ( 0.7
8.9 ( 0.4
4.0 ( 0.2
12 ( 1
16 ( 1
20.1 ( 0.4
25 ( 2
20 ( 2
17 ( 2
-178 ( 37
-164 ( 14
-150 ( 5
-138 ( 13
-158 ( 19
-172 ( 53
65 ( 15
64 ( 7
65 ( 2
66 ( 8
67 ( 10
69 ( 23
-H
-H
-H
8
-ꢀRhMT
5
-ꢀRhMT
3
-ꢀRhMT
a
refers to the rate constant for a single step that involves addition of a single As3+ forming a product
are equivalent and the average is shown.
For reactions as shown in Scheme 1, where k
n
with n As3 bound. Inspection of the lines of best fits show a clear trend for k2-6ꢀR for which these E
+
b
A
comparison of the relative abundances of As
n
-ꢀRhMT (n )
0
-6) for two simulations (99 s at 287 K and 402 s at 327K)
with the respective experimental data. In both Figure 5A and
B, the predicted relative abundances in the simulation are within
8
% of the experimental relative abundances. The predominant
species present at 99 s and 287 K (Figure 5A) are apo- and
As -ꢀRhMT; upon increasing the reaction times and temperature
to 402 s and 327 K (Figure 5B) the predominant species shifted
1
3+
5 6
to the more As -loaded species, As - and As -ꢀRhMT. This
test uses fitted parameters from analysis of all the experimental
data for each species, for example, 27 data sets were used to
calculate the parameters.
To further test the validity of the method and the accuracy
of the parameters derived in Table 1, we simulated the
normalized abundance of each species for the fixed reaction
times of 99, 191, 402, 498 s at increasing temperatures, for all
experimental data used for the temperature-resolved experiments
Figure 7. Comparison of the rate constants calculated from the time-
resolved ESI-MS measurements for As -metalation of RhMT, ꢀhMT, and
3+
ꢀ
RhMT, the trend in rate constant values for six equivalent sites where k
1
(
Figure 6). The smooth lines in Figure 6 are simulations of the
-
1 -1
)
28.8 M
s
and the rate constants reported for RhMT-S-tag and ꢀhMT-
six simultaneous bimolecular reactions occurring over the range
of temperatures and times shown, which are described by 27
individual data sets. The simulations are dependent on the
calculated specific rate constants for the specified temperature
and the activation energies and Arrhenius constant, A, shown
in Table 1. Considering the inherent uncertainties of the
experimental data and the extent to which each component
interconnects, such that errors in the relative abundance of one
species for the data set will introduce compensating errors in
the relative abundance of the other species, we suggest the
simulations are in remarkably good agreement with the experi-
mental normalized concentrations. We assess the uncertainties
as up to 20%.
1
S-tag. The red dashed line represents rate constant data for the ꢀ hΜΤ
redrawn with the value of n shifted by three to illustrate the similarity to
the rate constant trend for the final three As binding to ꢀRhMT.
3
+
thermostatted mixing tee, it was possible to quantitatively
measure the temperature-, reaction time-, and concentration-
3+
dependence of the reaction of As with these ꢀhMT and RhMT
1
protein fragments. In this present report, kinetic data were
determined for the reaction of the more complicated, two-
3+
domain ꢀRhMT protein with excess As for a series of fixed
reaction times as a function of temperature. As described in
the Experimental Methods section, the data were rearranged into
a 3D representation showing the relative abundance for each
of the species observed in the mass spectra: apo, As
1
-, As
2
-,
Taking a closer look at Figure 6A shows that at 99 s and
low temperatures, the dominant species present was apo-
ꢀRhMT, but following heating, the apoprotein is consumed to
form As -H -ꢀRhMT, which in turn is consumed to form
As -, As -, As -, and As -ꢀRhMT as a function of temperature
3
4
5
6
and at each reaction time. Figure 4 shows seven 3D plots, which
allow for the visualization of the relative abundance of each
1
17
further intermediates with greater As3 content. Figure 6D shows
that at 498 s and low temperatures, there is a mixture of As-
intermediates already formed compared with the situation at
99 s, hence the reaction has proceeded further toward comple-
tion, as expected. At high temperatures and for the reaction time
of 498 s, nearly all the apo-ꢀRhMT has reacted to form the
final product, As -H -ꢀRhMT
+
As -ꢀRhMT (n ) 0-6) species as function of temperature and
n
reaction time. The contours included in Figure 4 also allow for
each species’ relative concentration to be determined as a
function of temperature and reaction time. Multiple sets of data
at a fixed temperature were extracted from these 3D plots and
analyzed using the sequential second-order reaction mechanism
shown in Scheme 1. The temperature-dependent rate constants
6
2
-1
were obtained and plotted as ln(k) versus 1/T (K ) (Figure S3A,
Discussion
-1
Supporting Information) and as ln(k/T) versus 1/T (K ) (Figure
S3B, Supporting Information) to evaluate activation energies,
Arrhenius factors, and the Erying transition state parameters
Metalloproteins are considered to represent 30% of all
proteins, yet there is a lack of detailed understanding about the
metalation process itself. In general, one can separate the
metalation process into two categories, (i) metalation of an
apoprotein that is already folded and in its native conformation,
and (ii) metalation that results in significant folding of the
apoprotein that may be identified as metal-induced folding. It
is clear from all previous work, that in metallothioneins, metal-
(
Table 1).
The activation energies and Arrhenius factors in Table 1 were
used to simulate the theoretical relative abundance of each As
n
-
ꢀ
RhMT (n ) 0-6) for every experimental data point obtained
at fixed times and varying temperatures during the temperature-
resolved experiments (Figures 5 and 6). Figure 5 shows a
J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008 17023

A R T I C L E S
Ngu et al.
induced folding is the major force controlling folding to the
native state, playing a more important role than the typical
interaction forces of H-bonding and hydrophobicity. In this
present study, we report on the mechanism of metalation for
the two-domain, human metallothionein, which is a model for
metalloproteins with multiple metal-binding sites for which
metal-induced folding applies.
Metallothionein is ubiquitous in Nature, highlighting its
significance in the cellular chemistry of metals. As one of the
most prominent multiple-metal-binding proteins in the natural
world, an understanding of the mechanism for metalation of
metallothioneins is needed and this mechanism will serve as
an excellent model for other metalloproteins where the metal
or metals dominate the formation of the secondary and tertiary
structure.
of the experimental data using this model provided good fits to
the experimental time- and temperature-dependent data, which
we consider to support the validity of the mechanism in Scheme
1. This study confirms the binding stoichiometry of three and
six As3 bound to the single fragments and the two-domain MT,
+
1,61,62
respectively, as reported in previous studies.
The presence
3+
3+
of As -species with fewer than these numbers of As bound
indicates noncooperative sequential binding rather then a
cooperative mechanism. There was no evidence of thiolate-
bridging in any of the three protein species, because if thiolate-
bridging were to occur then we believe the remaining two
cysteines in the 11 Cys-RhΜΤ and 20 Cys-ꢀRhMT would bind
an additional As3 ion, bringing the saturated levels to four and
seven, respectively, for the fragments and two-domain protein.
This also implies that there is no cluster formation as typically
+
2+
Our mechanistic studies reported in this paper extend our
observed with other metals like Cd . Based on the masses
obtained from the ESI-MS data, we can determine that each
3+
previously reported slow metalation reaction of As with the
1
3+
isolated metallothionein fragments. In this paper, we report the
As binds to three sulfurs, displacing three protons, implying
metalation of the two-domain human metallothionein, comparing
a distorted trigonal conformation based on arsenic’s typical
coordination chemistry with a nonbonding electron pair blocking
the fourth site of the tetrahedron.
3+
the rates of binding up to six As into the two-domains with
3+
72
the rates for binding up to three As into the two isolated
fragments. Mammalian metallothionein exists only as the two-
domain protein yet no conclusive explanation of the evolutionary
advantage for the two-domains has been put forward. Com-
parisons between the metalation data of the isolated fragments
and the two-domain protein have previously been difficult
because the rates of metalation are typically too fast to measure
even by stopped-flow techniques. In the experimental methods
used in the present study, the metal binding rates are slow
enough to allow complete determination of all the kinetic
parameters for each metal and further, we have available the
complete data for both the isolated fragments and the two-
domain protein.
Comparison and Implications of the Values of the Rate
Constants. Figure 7 shows a comparison of the rate constants
3+
calculated for As -metalation of ꢀhMT, ꢀhMT-S-tag, RhMT,
RhMT-S-tag, and ꢀRhMT. The ꢀhMT-S-tag and RhMT-S-tag
1
protein from Ngu et al. contained an additional short amino
acid sequence (S-tag) in the MT residue sequence, which was
part of the protein expression system. It is clear, now that kinetic
data for the thrombin-cleaved fragments have been obtained,
that the presence of the S-tag slowed the binding rates of the
3+
protein but did not affect the overall mechanism of As binding.
MD calculations have previously shown the S-tag is quite
fluxional and does not adopt a fixed or even sticky association
3
+
Mechanism of Binding for As . A recent paper reports the
potential use of seaweed MT, expressed in Escherichia coli cells
26
3+
with MT. However, we believe the S-tag acts in these As -
metalation studies by periodically blocking the binding site. This
blocking action slows the metalation reaction rates but our
to remediate As3 in drinking water, which will require an
+
67
3+
understanding of the As metalation chemistry for the develop-
ment of the remediation technology. As -metalation of MT in
the mammalian system is also important in understanding the
in vivo chemistry of As and its metabolic pathway following
ingestion from seafood and drinking water. Studies by
3+
analysis shows that it does not change the mechanism of As -
3+
metalation.
To compare the rates of metalation of the previous S-tag-
containing fragments with the data reported in the study for
which the S-tag had been cleaved, as shown in Figure 7, we
have scaled the rate constants for the S-tag containing protein
by an empirically determined factor of 2.8. The trends now
displayed in Figure 7 are similar for both sets of fragment data
and confirms that the rate constants in this paper are consistent
3+
6
8-70
71
Graham
and Rosen have shown that there are two
potential mechanisms by which arsenic may be detoxified from
the human body. Both detoxification pathways involve thiol-
3+
containing proteins and peptides binding As and in the case
5+
3+ 71
of the arsenic reductases, the reduction of As to As .
1
with those reported earlier.
Clearly, the cysteine-rich metallothioneins may play an impor-
tant role in As3+ metabolism.
Figure 7 summarizes all the trends observed in the rate
constants for the two-domain protein and the individual single
fragments. The significant and novel result reported here is that
the two-domain protein binds the initial As3 ions considerably
faster than the isolated domains and that the rate constants
systematically diminish as a function of the remaining sites.
The trends in rate constants for each species guide our
interpretation of the overall metal-binding properties of the
While previous mechanistic results were for the isolated
single-domain fragments, the data presented here are for the
complete, two-domain protein. Scheme 1 shows a mechanism
+
3+
for the sequential binding of six As to both domains. Analysis
(
67) Singh, S.; Mulchandani, A.; Chen, W. Appl. EnViron. Microbiol. 2008,
7
4, 2924–2927.
ꢀ
RhMT protein. For each fragment, the reduction in rate
(
68) Gailer, J.; George, G. N.; Pickering, I. J.; Prince, R. C.; Ringwald,
S. C.; Pemberton, J. E.; Glass, R. S.; Younis, H. S.; DeYoung, D. W.;
Aposhian, H. V. J. Am. Chem. Soc. 2000, 122, 4637–4639.
3+
constants as a function of As -bound also follows the same
dependence on the number of available unoccupied binding sites.
Additionally, the nonlinear behavior of the rate constants for
the first As3 bound to apo-RhMT and apo-ꢀRhMT, we propose,
(
69) Gailer, J.; George, G. N.; Pickering, I. J.; Prince, R. C.; Younis, H. S.;
Winzerling, J. J. Chem. Res. Toxicol. 2002, 15, 1466–1471.
+
(
70) Manley, S. A.; George, G. N.; Pickering, I. J.; Glass, R. S.; Prenner,
E. J.; Yamdagni, R.; Wu, Q.; Gailer, J. Chem. Res. Toxicol. 2006, 19,
6
01–607.
(
71) Demel, S.; Shi, J.; Martin, P.; Rosen, B. P.; Edwards, B. F. P. Protein
Sci. 2004, 13, 2330–2340.
(72) Housecroft, C. E.; Sharpe, A. G. Inorganic Chemistry, 2nd ed.; Pearson
Prentice Hall: Harlow, England, 2005.
1
7024 J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008

Metal-Induced Folding of Metallothionein
A R T I C L E S
arises from structural properties of the metal-free protein R
domain. We will now address the following key features from
Figure 7:
tion) shows the Arrhenius plot and Erying plot for k1-6ꢀR. It
can be seen in Figure 3A that with the exception of the 1/T
dependence of k1ꢀR, the trend in 1/T dependence of k
n
are very
(
i) K1ꢀ Is Less for S-Tag-Containing ꢀhMT than the
similar, hence to calculate the E we set the slope of the 1/T
A
S-Tag-Free Fragment. The first rate constant, k1ꢀ for the first
dependence of k2-6ꢀR to the average value for these five
reactions. The data in Figure S3A (Supporting Information)
show that the E is essentially the same for all but the first
A
3+
As bound to the ꢀhMT-S-tag-free protein is greater than that
of the adjusted k1ꢀ for the S-tag bound ꢀhMT protein. We
believe this is because the S-tag-free protein can bind the first
3
+
3+
As bound. We conclude, therefore, that the As -metalation
of apo-ꢀRhMT is controlled by the Arrhenius factor, in other
words, the statistical frequency of collisions that occur with the
correct orientation and result in product formation.
3+
As ion from both the N- and C-terminals whereas the
N-terminal bound S-tag containing protein can only readily bind
the first As3 from the C-terminal, statistically reducing the rate.
+
q
q
q
(
ii) k1r < k2r for rhMT. The rate constants for the first two
The Erying Transition State Parameters ∆H , ∆S , and ∆G
1
3+
As -metalation steps of apo-RhMT exhibit the reverse order
than expected, that is k1R < k2R; this was observed in the data
for both this report and the previous report. Duncan et al. have
correspond to those calculated by Ngu et al. and provides
3+
further evidence that the underlying chemistry for As -binding
to ꢀRhMT is the same as that observed in the isolated, individual
R and ꢀ domains.
1
proposed that the apo-RhMT adopts a structure as a result of
metal templating instead of existing as a simple random
Molecular Modeling of As-Metalated MT. Figures 8 and 9
shows molecular models constructed from the energy-minimized
4
2,43,73
3+
coil.
If this was the case, then the first As to bind apo-
3+
RhMT may need to disrupt this structure, thereby allowing the
structures for the metal-free and the As -loaded ꢀhMT, RhMT
and ꢀRhMT proteins. In the absence of any As-MT X-ray crystal
or NMR structures, these molecular models provide an initial
succeeding As3 to bind more readily and resulting in k1R
k
+
<
3+
2R. The data suggest the first As binds to the C-terminal
3+
because the reduction in k1R is similar for both the S-tag-free
visual framework to aid in our understanding of the As -
metalation mechanism and possible final structures. The apo
structures of all three constructs shown in Figure 8 and 9 are
and S-tag containing fragments.
(
iii) The Rate Constant Values for Each Successive As3+
Bound in ꢀrhMT Decrease Approximately by a Factor of 1/6
per As3 Bound. The six experimentally determined rate
constants for the two-domain ꢀRhMT protein are shown as
function of the number of As bound in Figure 7. With the
exception of k1ꢀR the trend in the rate constant values is that
predicted for the presence of six equivalent binding sites. The
somewhat globular with the sulfurs exposed to the solvent as
+
42,43
first reported by Rigby et al.
Refering to the ESI-MS data
we see that addition of the first As3 ion to the RhMT or the
ꢀhMT protein resulted in a collapse of multiple charges for the
metal-free proteins to a single dominant charge state (Figures
S1-S2, Supporting Information). The molecular models show
that a single addition of one As3 atom to apo-ꢀhMT or apo-
RhMT results in formation of another globular structure but
where the sulfurs are more shielded by the peptide chain. Further
+
3+
+
predicted magnitude of each k
sites diminishes as a function of the number of unoccupied sites
available for the incoming metal, that is, k ) 6/5k ) 3/2k
, where the predicted value of k is 28.8 M s .
v) The Trend in k1ꢀr Is Reminiscent of the Trend in k1r
n
(n ) 1-6) for six equivalent
1
2
3
)
-1
-1
3+
2k
3
) 3k
5
) 6k
6
1
additions of As to the ꢀhMT or RhMT proteins do not change
(
.
the charge state distributions observed (Figures S1 and S2,
Supporting Information), which may be the result of formation
of similar globular structures predicted by the molecular
modeling (Figure 8).
The molecular models for ꢀRhMT are shown in Figure 9.
The MS spectra for As-metalation of apo-ꢀRhMT (Figure 2)
show apo-ꢀRhMT has two dominant charge states (+7 and +5),
which give way to the single dominant charge state of +5 when
The reduced k1ꢀR is also found for the As-metalation of apo-
RhMT and apo-RhMT-S-tag (Figure 7). Our interpretation of
the reduction in k1ꢀR is based on the same argument used in
point (ii) in which hindered access to the binding sites for the
metal-template structure slows the metal-binding process. We
propose from this analysis that the majority of the first As3 is
bound to the C-terminus of the R domain in ꢀRhMT based on
the trend observed for k1-2ꢀR. We suggest, however, that the
small increase between k1ꢀR and k2ꢀR, arises from a small fraction
of As3 binding to the N-terminal ꢀ domain in the ꢀRhMT.
+
As
dominant charge states of +6 and +5 for As
Similarly to the As -R/ꢀhMT (n ) 1-3) molecular models, the
4
-H
8
-ꢀRhMT is most abundant and then shifts to the
6
-H -ꢀRhMT.
2
+
n
MD energy-minimized ꢀRhMT structures show that the metal-
free protein coalesces to a globular structure in which the two
domains intermingle and the sulfurs are exposed to the solvent.
(
vi) k1-3ꢀ Are Relatively Close in Value to k4-6ꢀr. In Figure
7
, the rate constants, k1-3ꢀ for ꢀhMT are also shown shifted by
n ) 3 places to align them with k4-6ꢀR for ꢀRhMT and to
emphasize the relative closeness of these values to the last three
rate constants for ꢀRhMT. On the basis of this similarity in
values between k1-3ꢀ and k4-6ꢀR, we propose that the last three
3+
Following binding of 1-4 As atoms, the structure of the
protein, though still globular, is more compact with the unbound
sulfurs still exposed on the surface. The binding of the fifth
and sixth As3 ion results in another globular conformation
+
3+
As are bound to the ꢀ domain of ꢀRhMT.
3
where the As-Cys units are buried inside the protein
Implications of the Values of the Kinetic and Erying
Transition State Parameters. Table 1 lists the activation energies
Why Is Mammalian MT a Two-Domain Structure? The two-
domain evolutionary advantage of the mammalian MT has been
speculated on for a long time, particularly compared to single-
domain MT. In this present study, we analyzed kinetic data for
for the reaction forming each As
n
-ꢀRhMT species (n ) 1-6).
increases from 14 kJmol for H -ꢀ, As -H -ꢀRhMT to 22
-ꢀRhMT species (n ) 2-6). For
-ꢀRhMT (n ) 2-6), the activation energies listed are an
average of all the E ’s for k2-6ꢀR calculated from the lines of
best fit of the Arrhenius plots. Figure S3 (Supporting Informa-
-1
E
A
9
1
8
-1
kJmol for the remaining As
As
n
3+
the As -metalation of the two-domain human MT and com-
pared these results to the isolated domain fragments. The
metalation involved a noncooperative sequential binding mech-
anism where, we propose, the data show the R domain and then
the ꢀ domain bind As3 atoms sequentially along the amino
acid sequence starting with the first three cysteines from the
n
A
+
(
73) Rigby Duncan, K. E.; Stillman, M. J. J. Inorg. Biochem. 2006, 100,
101–2107.
2
J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008 17025

A R T I C L E S
Ngu et al.
1
7026 J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008

Metal-Induced Folding of Metallothionein
A R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008 17027

A R T I C L E S
Ngu et al.
C-terminal. We further showed that in ꢀRhMT the metalation
rate of the five equivalent binding sites are controlled by the
statistical probability of a collision with the correct orientation
occurring.
always be sufficient MT to reduce metal concentrations quite
reasonably resulting in the presence of partially metalated MT.
Progress in Understanding Metalation Reactions of
Metalloproteins and Metallothionein. A review by Gray in 2003
states that one of the major growth areas for inorganic
biochemistry will be the study of how peptides/proteins fold
The key result we report is that the initial rate constant values
for the separate single domain fragments are significantly slower
3+
than the initial rate constant of the two-domain protein for As -
metalation. This observation, we propose, is a consequence of
75
around metals. Even now, few metalation/folding studies of
76
metalloproteins or peptides have been reported. To the best
ꢀ
RhMT having six binding sites compared to the three binding
of our knowledge there are only a small number of reports on
sites in the individual fragments and reflects how an increase
in the number of binding sites directly results in an increase in
the rate constants values for the two-domain protein. For a
protein, increasing the number of equivalent binding sites,
therefore, increases the rate at which the protein will bind the
first few metals. If the dissociation reaction occurs with the same
rate for each site then this differential in binding rates results
in significantly different binding constants for the first and last
metal bound. For MT this implies that the change from a single-
domain protein to a two-domain protein, results in the two-
domain MT being able to bind metals faster and more efficiently,
hence becoming a superior metal scavenger; one of MT’s
proposed functions.
the kinetics of metalation or demetalation reactions of
1,28,31,33,37,77,78
metallothionein.
We have described here for the
first time the complete kinetic analysis for the arsenic-induced
metalation of the two-domain ꢀRhMT and show the sequential
rate constants for each step of the metalation reaction. These
data show that human metallothionein binds arsenic sequentially
rather then cooperatively and the order of domain loading
3+
appears to be R then ꢀ with respect to As . Our results provide
direct evidence for the evolution of the two-domain metal-
lothionein protein.
Acknowledgment. We thank NSERC of Canada for financial
support through operating funds (to M.J.S.) and the NSERC
Canadian Graduate Scholarship (to T.T.N.). We thank Prof. R. J.
Puddephatt for use of the ESI-MS funded by the Canada Research
Chair program. We are grateful to Doug Hairsine for advice and
discussion on the operation of the ESI-MS.
A number of reports have been published that describe
properties of MT that may be explained by our proposed
explanation for metallothionein’s two-domain structure. We
7
,24
outline one such property now. Partially metalated MT
has
2
+
been reported and noncooperative binding mechanisms for Zn
and Cd2+
25,66,74
Supporting Information Available: Three figures (S1-S3)
with descriptive text are provided that show timed-resolved
ESI-mass spectral data, the Arrhenius plot, and the Erying plot
referred to in the text. This material is available free of charge
via the Internet at http://pubs.acs.org.
have been demonstrated. On the basis of our
proposal that an increase in the number of binding sites will
increase the rate constant values (and the binding constants)
for noncooperative binding, we can explain why partially
metalated MT may exist. From a coordination chemistry point
of view, with a large number of equivalent binding sites, it is
quite reasonable that one or more of these sites may be
unoccupied. Only in the presence of a large excess of metals
will the last binding site be occupied. But it is well-known that
MT induction takes place rapidly so that in the cell there will
JA8060326
(75) Gray, H. B. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3563–3568.
(
(
76) Ghosh, D.; Pecoraro, V. L. Inorg. Chem. 2004, 43, 7902–7915.
77) Krepkiy, D.; Antholine, W. E.; Petering, D. H. Chem. Res. Toxicol.
2
003, 16, 750–756.
(
74) Palumaa, P.; Eriste, E.; Njunkova, O.; Pokras, L.; Joernvall, H.; Sillard,
(78) Salgado, M. T.; Stillman, M. J. Biochem. Biophys. Res. Commun. 2004,
318, 73–80.
R. Biochemistry 2002, 41, 6153–6158.
1
7028 J. AM. CHEM. SOC. 9 VOL. 130, NO. 50, 2008