www.elsevier.

nl/locate/jinorgbio

Journal of Inorganic Biochemistry 79 (2000) 225–229

Bacterial metal-resistance proteins and their use in biosensors for the

detection of bioavailable heavy metals
Ibolya Bontidean b, Jon R. Lloyd a, Jon L. Hobman a, Jon R. Wilson a, Elisabeth Csoregi
¨ b
,
Bo Mattiasson b, Nigel L. Brown a,*
a
The University of Birmingham, School of Biological Sciences, Edgbaston, Birmingham B15 2TT, UK
b
Lund University, Centre for Chemistry & Chemical Engineering, Department of Biotechnology. PO Box 124, S-221 00 Lund, Sweden

Received 17 April 1999; received in revised form 12 September 1999; accepted 22 September 1999

Abstract

We have expressed and purified metal-resistance and metal regulatory proteins from the bacterial determinants of resistance to heavy metals
and utilised these in the development of biosensors for heavy metals. Both the metallothionein from the cyanobacterium Synechococcus PCC
7942 and the MerR regulatory protein from transposon Tn501 allow the detection of non-specific metal binding down to 10y15 M concentrations
of Hg(II), Cu(II), Zn(II) and Cd(II) in pure solution. Differential effects of the metals can be detected at both low and high concentrations,
and the shape of the capacitance curves may reflect biologically relevant responses of the proteins to metals. Further work is required to
establish the relationship between the detected binding of metal and the biological response of the protein, or to provide biosensors of use in
the natural environment. q2000 Elsevier Science Inc. All rights reserved.

Keywords: Heavy metals; Biosensors; Metal-binding proteins; MerR; SmtA; Capacitance; Electrode

1. Introduction loregulatory systems which have been identified to date

Bacteria may exhibit resistance to a wide variety of heavy
metals, usually because they carry determinants conferring
resistance to one or to a small number of heavy metals. These
include Ag, Bi, Cd, Co, Cu, Ge, Hg, Pb, Ni, Tl or Zn cations
and the oxyanions of As, Cr, Sb, Te or W. Mechanisms of
bacterial metal resistance have been reviewed several times,
e.g. [1–4]. Possibly because bacteria are simple cells with
limited compartmentalisation, resistance is generally specific
to one or to a few metals, and the mechanisms of resistance
include efflux of the metal, modification of the speciation of
the metal, sequestration of the metal, or a combination of
these mechanisms [2,4].
The majority of metal-resistance determinants are induci-
ble by the cognate metal ion, and the structural genes for
metal resistance are under the control of a specific transcrip-
tional regulatory mechanism which enables expression of the
resistance genes to occur only at times of metal stress. A large
number of such regulatory mechanisms have been identified
and characterised [5], and many of these will respond to
metal stress outside their normal genetic location. The metal-
* Corresponding author. Tel: q44-121-414-5465; fax: q44-121-414-
5907; e-mail: n.l.brown@bham.ac.uk
include those for the cations or oxyanions of Ag, As, Cd, Co,
Cr, Cu, Fe, Hg, Ni, Sb, Zn [5,6]. These regulatory systems
have begun to be exploited in the development of biosensors
for metals, in which the response to metal is used to express
reporter genes, such as the lux genes of Vibrio fischeri [7,8]
the product of which can be detected by the amount of light
emitted due to the action of luciferase. Several groups hold
patents on biosensors constructed by linking the mercury-
resistance regulator region to the lux genes. The Flemish
Institute for Technological Research (VITO) produces a kit
containing a series of organisms that emit light in response
to different metals. The advantage of such biosensors over
classical analytical methods such as inductively coupled
plasma atomic electron spectrometry (ICP/AES), mass
spectrometry (ICP/MS), flow injection atomic absorption
(FIAAS) or electrochemical methods, is that the samples
often require little pretreatment and the bioavailable concen-
tration of the toxic metal is measured, rather than the total
concentration. However, the difficulty of using such gene-
based biosensors is that the biological component is a viable
cell, and therefore one is limited to conditions of measure-
ment which allow survival of the cell (narrow pH and tem-
perature ranges, lack of other toxic compounds, etc.). In

0162-0134/00/$ - see front matter q2000 Elsevier Science Inc. All rights reserved.
PII S 0 1 6 2 - 0 1 3 4 ( 9 9 ) 0 0 2 3 4 - 2

Friday Apr 07 10:59 AM StyleTag -- Journal: JIB (Journal of Inorganic Biochemistry) Article: 6328
226 I. Bontidean et al. / Journal of Inorganic Biochemistry 79 (2000) 225–229
addition, the response of the genetic regulatory system to
metal may not allow easy quantitation of the metal ion [9].
Our approach was to use metal-binding proteins them-
selves as the biological part of the biosensor. In principle, this
should confer the high biological specificity of the biological
system with a robustness of a (bio)chemical reaction that
may occur outside the physiological parameters required by
a living cell [10]. We have shown that capacitance changes
occurring on the surface of an electrode due to binding of a
metal ion by a protein can be used to detect the presence of
femtomolar to millimolar concentrations of heavy-metal ions
[8,11]. In this paper we describe the initial characterisation
of this system using, as examples of metal-binding proteins,
the prokaryotic metallothionein SmtA (as a fusion with glu-
tathione-S-transferase) [12] and the mercury-responsive
regulatory protein MerR [13].
2. Experimental
2.1. Overexpression and purification of heavy-metal
binding proteins
The mercuric ion-binding regulatory protein, MerR from
transposon Tn501 [13], was overexpressed in E. coli and
purified as described elsewhere [11,14]; 10 mg of MerR
protein was obtained. The synechococcal metallothionein
protein, SmtA, was overexpressed as a fusion with glutathi-
one-S-transferase and the GST-SmtA fusion protein was puri-
fied by published methods [11,12]; 4 mg of protein was
obtained.
2.2. Protein immobilisation and capacitance measurements
The fusion proteins GST-SmtA and MerR were dissolved
to a final concentration of 1 mg/mL protein in phosphate
buffered saline (PBS; 70 mM NaCl, 1.3 mM KCl, 5 mM
Na2HPO4, 0.9 mM KH2PO4, pH 7.3) containing 50% glyc-
erol. Thioctic acid was purchased from Sigma (St. Louis,
MO) and 1-(3-dimethylaminopropyl)-3-ethyl-carbodiimide
hydrochloride (EDC) was obtained from Fluka AG (Buchs,
Switzerland). 1-Dodecanethiol and the gold rods (99.99%,
3 mm in diameter) used for the electrodes came from Aldrich
Chemicals (Milwaukee, WI). Dithiothreitol (DTT) was pur-
chased from Chemicon (Malmo, ¨ Sweden). Heavy-metal
salts CuCl2P2H2O, ZnCl2, HgCl2 and Cd(NO3)2P4H2O were
from Merck (Darmstadt, Germany). All other reagents were
of analytical grade and solutions were, if not otherwise men-
tioned, prepared with water obtained from a Milli-Q system
preceded by a reverse-osmosis step, both from Millipore
(Bedford, MA). All glassware was soaked in 3 M HNO3 for
3 days and then 1 day in Millipore water before use.
The biosensors were prepared by immobilising fusion pro-
teins on the surface of gold rods by pretreating with thioctic
acid and coupling with EDC as described elsewhere [11,15]
and illustrated in Fig. 1. The dissolved fusion proteins were
Friday Apr 07 10:59 AM
Fig. 1. Scheme showing a metal-binding protein immobilised on the surface
of a gold electrode. The thioctic acid linker and the blocking of unused
binding sites with 1-dodecanethiol are shown, but the EDC coupling of the
protein to the linker is not.
diluted in 100 mM borate buffer at pH 8.75 and the solutions
were filtered through a micro-filter (Amicon, Beverly, MA)
with a molecular cut-off of 3000 D. After ultrafiltration, the
fusion protein concentration was adjusted to 0.04 mg/mL in
borate buffer. In order to stabilise the oxygen-sensitive MerR
electrodes, DTT was added to this protein solution to a final
concentration of 2 mM.
The biosensor was placed as the working electrode in an
in-house-built three/four-electrode containing electrochem-
ical cell (with a dead volume of 10 mL) connected to a fast
¨ Elektronik, Lund, Sweden) [11,15]. A Pt
potentiostat (Zata
foil served as the auxiliary and a Pt wire as the reference
electrode. An additional reference electrode (AgNAgCl) was
placed in the outlet stream. The cell was arranged in a flow
injection (FI) system pumping the buffer solutions (10 mM
borate or HEPES, both containing 0.02% sodium azide) with
a peristaltic pump (Alitea AB, Stockholm, Sweden) at a flow
rate of 0.5 mL/min. Samples were injected into the flow via
a 250 mL sample loop. The carrier buffers were filtered
through a 0.22 mm Millipore filter and degassed before use.
3. Results and discussion
Earlier experiments with the metal-responsive capacitance
biosensors [8,11] showed that the capacitance system would
respond from 10y15 to 10y1 M Cu(II) using GST-SmtA as
the biological component. The relative response of the GST-
SmtA and Tn501 MerR biosensors to Cu(II), Zn(II), Hg(II)
and Cd(II) was tested across a 105-fold range from 10y15
M. Cu(II) was used rather than Cu(I) due to our interest in
metal species in oxic natural environments. In this work we
examine the effects of pH and buffer on the sensitivity of the
GST-SmtA electrode and compare the response of both SmtA
and MerR electrodes to four metal ions across the range 10y15
to 10y3 M.
StyleTag -- Journal: JIB (Journal of Inorganic Biochemistry) Article: 6328
 I. Bontidean et al. / Journal of Inorganic Biochemistry 79 (2000) 225–229
Fig. 2. Capacitance changes of the GST-SmtA electrode to Hg(II) and
Cu(II) at different pH values. The electrodes were constructed by EDC-
coupling in borate buffer at pH 8.75 as described in the text and in Refs.
[11,15] and subsequently equilibrated with borate or HEPES buffer (both
10 mM) as shown. Open and cross-hatched columns represent Cu(II) and
Hg(II), respectively, in HEPES buffer; vertically lined and solid columns
represent Cu(II) and Hg(II), respectively, in borate buffer.
The capacitance change of the GST-SmtA biosensor on
exposure to 10y4 M Cu(II) or Hg(II) at pH 8.75 and pH
8.0, in HEPES and borate buffers, is shown in Fig. 2. At pH
8.0, Hg(II) gives a better response in borate buffer than in
HEPES, and both Cu(II) and Hg(II) in borate buffer give a
better response at pH 8.75 than at pH 8.0. Using HEPES
buffer, the response at pH 7.0 is lower than that at pH 8.0 for
Hg(II). Borate, therefore, seems to allow the optimum
response.
The robustness of a biosensor depends on the number of
times it can be regenerated following a measurement. In order
to investigate this, a GST-SmtA electrode was subjected to
repeated cycles of exposure to 10y4 M Hg(II) at pH 8.0 in
Fig. 3. Regeneration with 1 mM EDTA of a GST-SmtA electrode exposed to 0.1 mM Hg(II). Total capacitance measurements were carried out at pH 8.0 in
HEPES (upper line) and borate buffer (lower line): 1, injection of 1 mM EDTA; 2, injection of 0.1 mM Hg(II).
Friday Apr 07 10:59 AM
227
both HEPES and borate buffer. The absolute capacitance was
measured. Fig. 3 shows that the capacitance of the electrode
in the presence of HEPES is relatively constant across three
cycles of exposure to 10y4 M Hg(II) followed by 1 mM
EDTA, whereas the capacitance of the electrode in borate
buffer at the same pH drifts upwards slightly. Therefore,
either buffer condition would allow the electrode to be used
several times. The full extent of reusability of the electrode,
optimal storage and washing conditions, and the effects on
regeneration due to other metals remain to be determined.
When the electrode was stored in the presence of heavy metal
and reactivated with EDTA immediately before use, the
response was stable for about a week [11], although the
electrode deteriorated rapidly if stored in EDTA. This is prob-
ably due to the metal ions protecting the protein from
oxidation.
Fig. 4(A) shows the response of the GST-SmtA electrode
at pH 8.75 in borate buffer to Cu(II), Zn(II), Cd(II) and
Hg(II) across the full range of concentrations tested. At low
concentrations (-10y11 M), the electrode is most sensitive
to Cu(II), but it becomes least sensitive to Cu(II) and most
sensitive to Hg(II) at higher concentrations ()10y7 M).
The shapes of the capacitance versus concentration curves
for each metal are similar, but are displaced. The shape of the
capacitance curves for each metal at concentrations below
10y10 M is compatible with a mechanism in which the metal-
binding sulfydryl and histidine groups of SmtA are titrated
at different rates and the shape of the curve represents the
sum of the titration events for each metal. At varying con-
centrations for each metal, the slope of the curve changes
dramatically, and this may represent the folding of the SmtA
StyleTag -- Journal: JIB (Journal of Inorganic Biochemistry) Article: 6328
228 I. Bontidean et al. / Journal of Inorganic Biochemistry 79 (2000) 225–229

Fig. 4. Capacitance changes of (A) a GST-SmtA electrode and (B) a MerR electrode on exposure to metal solutions: circles, Hg(II); diamonds, Cd(II);
triangles, Zn(II); and squares, Cu(II). The electrodes were prepared as described for Fig. 2 and all measurements were done at room temperature. Data are
expressed as capacitance per unit area of electrode surface. The insets are enlargements of the lower end of the graphs, with the Y-axis amplified.

region to form the typical metallothionein cage structure.
NMR analysis of SmtA suggests that histidine residues, as
well as cysteines, may be involved in metal binding [16].
The GST-SmtA electrode was constructed as SmtA binds
a variety of metals [16]. Part of its biological function was
thought to protect the cyanobacterium against toxic metals,
although more recent interpretations suggest that it may be
concerned primarily with zinc homeostasis in Synechococcus
[16]. Tn501 MerR, on the other hand, is highly specific in
its biological function (induction of transcription of mercury-
resistance genes [13,14]), having an in vitro response to
Hg(II) in the initiation of transcription about 103-fold more
sensitive than its response to Cd(II) [17]. Fig. 4(B) shows
the results of capacitance change against metal-ion concen-
tration in borate buffer at pH 8.75. At low metal-ion concen-
trations (10y15 to 10y9 M) the response to all metals is
approximately linear, but the biosensor responds in the order,
Hg(II))Cu(II))Zn(II)sCd(II). The slope of the capac-
itance change against concentration curve changes at approx-
imately 10y7 M for Hg(II), Zn(II) and Cd(II), and at about
10y5 M for Cu(II). Again this is compatible with metal
titration of non-specific binding sites (e.g. cysteine and his-
tidine residues) on the protein at low metal concentrations.
The change above 10y7 M for Hg(II), etc., may coincide
with the change to specific binding of the metals to the func-
tional mercury-binding site. This is known to consist of cys-
teines Cys82, Cys117 and Cys126 of MerR, and Hg(II) is
bound as a single ion in a tri-coordinate complex [18,19].

Friday Apr 07 10:59 AM StyleTag -- Journal: JIB (Journal of Inorganic Biochemistry) Article: 6328
 I. Bontidean et al. / Journal of Inorganic Biochemistry 79 (2000) 225–229 229

Throughout the binding curve, Hg(II) shows the greatest relate these capacitance changes to relevant interpretations of
effect. As with GST-SmtA, the capacitance change caused properties in vivo.
by other metal ions differs in relative order depending on the
concentration of the cation. At high concentration (10y3 M)
there is little difference between Hg(II), Zn(II) and Cu(II), Acknowledgements
but Cd(II) shows a lower capacitance change.
The response of the capacitance change of MerR-coated Plasmid pGEX3X-Smta was a kind gift of Professor Nigel
electrodes to metal-ion concentration may partially be Robinson (Newcastle). Ken Jakeman provided technical
explained in terms of the biological properties of the protein. assistance in protein purification. The work was supported
Since we first embarked on these experiments, we have iden- by the European Commission as part of Project ENV4-CT95-
0141 and by grants from the UK Biotechnology and Biolog-
tified a protein ZntR in E. coli, which has a similar sequence
ical Sciences Research Council (No. G07943 to N.L.B.) and
to MerR and which responds to Zn(II) [20]. ZntR contains
the Swedish Natural Research Council (to E.C.). I.B. and
cysteines corresponding to those required for Hg(II) binding.
J.R.W. were funded in part by the Swedish Institute.
Although there is little biological cross-reactivity between
MerR and ZntR, they may still bind both metals, and we
assume that the conformational change required in MerR to References
cause activation of transcription only occurs in the presence
of Hg(II) [21]. The change in slope of the binding curve [1] M. Mergeay, Trends. Biotechnol. 9 (1991) 17–24.
cannot simply be the biologically relevant conformational [2] S. Silver (Ed.), Plasmid 27 (1992) — special issue.
[3] S. Silver, Gene 179 (1996) 1–19.
change, as this occurs at 10y7 M for Hg(II) and 10y6 M for [4] S. Silver, L.T. Phung, Annu. Rev. Microbiol. 50 (1996) 753–789.
Zn(II), whereas the in vitro response for activation of tran- [5] S Silver, W. Walden, Metal Ions in Gene Regulation, Chapman and
scription shows at least 105-fold difference [16]. We assume Hall, New York, 1997.
that a similar conformational change is required in ZntR on [6] N.L. Brown, K.R. Brocklehurst, B. Lawley, J.L. Hobman, in: S.J.W.
Busby, C.M. Thomas, N.L. Brown (Eds.), Molecular Microbiology,
the binding of Zn(II), although such a conformational change
Springer, Berlin, 1998, pp. 159–173.
has not yet been demonstrated in this case. Recently, a third [7] P. Corbisier, G. Ji, G. Nuyts, M. Mergeay, S. Silver, FEMS Microbiol.
member of the MerR family has been identified which Lett. 110 (1993) 231–238.
responds to heavy metals; this is PbrR from Alcaligenes [8] P. Corbisier, D. van der Lelie, B. Borremans, A. Provoost, V. de
eutrophus CH34, which regulates the expression of lead- Lorenzo, N.L. Brown, J.R. Lloyd, J.L. Hobman, E. Csoregi, ¨ G.
Johansson, B. Mattiasson, Anal. Chim. Acta 387 (1999) 235–244.
resistance genes. This is highly specific in vivo for Pb(II) [9] D.A Rouch, J. Parkhill, N.L. Brown, J. Ind. Microbiol. 14 (1995)
and does not respond to Zn(II), Cd(II), Cu(II) or Hg(II) 249–253.
[8,22]. We have recently purified the protein and it will also [10] N.L. Brown, Proceedings of the 2nd International Symposium on
be tested on the affinity capacitance biosensor. Environmental Biotechnology, Institute of Chemical Engineers,
The data presented in Fig. 4 suggest that the use of EDC- Rugby, UK, 1994, pp. 1–3.
[11] ¨
I. Bontidean, C. Berggren, G. Johansson, E. Csoregi, B. Mattiasson,
coupled metal-resistance proteins may provide useful infor- J.R. Lloyd, K.J. Jakeman, N.L. Brown, Anal. Chem. 70 (1998) 4162–
mation on the relative binding of metals by the proteins and 4169.
allow these to be interpreted in a biological context. However, [12] J.G. Shi, W.P. Lindsay, J.W. Huckle, A.P. Morby, N.J. Robinson,
such data must be interpreted with caution, as the exact bio- FEBS Lett. 303 (1992) 159–163.
[13] P.A. Lund, S.J. Ford, N.L. Brown, J. Gen. Microbiol. 132 (1986)
logical responses in vivo and in vitro [7,9,10,16] differ from
465–480.
the electrode responses obtained here. There are also differ- [14] J. Parkhill, A.Z. Ansari, J.G. Wright, N.L. Brown, T.V. O’Halloran,
ences between electrodes prepared under nitrogen (giving a EMBO J. 12 (1993) 413–421.
higher capacitance change [11]) and those prepared in the [15] C. Berggren, G. Johansson, Anal. Chem. 69 (1997) 3651–3657.
presence of 2 mM DTT (this work), indicating that some [16] M.J. Daniels, J.S. Turner-Cavet, R. Selkirk, H.Z. Sun, J.A. Parkinson,
P.J. Sadler, N.J. Robinson, J. Biol. Chem. 273 (1998) 22957–22961.
oxidation of the protein may occur even in the presence of [17] J.D. Helmann, B.T. Ballard, C.T. Walsh, Science 247 (1990) 946–
sulfydryl protective agents. With the proteins that have so far 948.
been tested in the system, the discrimination between metals [18] D.L. Huffman, L.M. Utschig, T.V. O’Halloran, Met. Ions Biol. Syst.
is not sufficient to be of use directly as biosensors in environ- 34 (1997) 503–526.
[19] K.R. Brocklehurst, J.L. Hobman, B. Lawley, L. Blank, S.J. Marshall,
mental situations, although this was one of the original pre-
N.L. Brown, A.P. Morby, Mol. Microbiol. 31 (1999) 893–902.
dictions for the use of such biosensors [10,23]. Natural [20] J. Parkhill, B. Lawley, J.L. Hobman, N.L. Brown, Microbiology 144
samples would be likely to contain a mixture of metals and (1998) 2855–2864.
the complex responses that would result may not readily be [21] D. Ralston, T.V. O’Halloran, Proc. Natl Acad. Sci. USA 87 (1990)
interpretable. We predict that the capacitance electrode may 3846–3850.
[22] B. Borremans, A. Provoost, P. Corbisier, M. Mergeay, J.L. Hobman,
be of more use in adding to the range of physical techniques, N.L. Brown, D. van der Lelie, Proc. VI Int. Congr. Pseudomonas:
whereby the binding of metals to metal-responsive proteins Mol. Biol. Biotechnol. (1997) IX.
can be studied, but considerable further work is needed to [23] Metal Ion Specific Affinity Sensors, Swedish Patent No. 9703315-3.

Friday Apr 07 10:59 AM StyleTag -- Journal: JIB (Journal of Inorganic Biochemistry) Article: 6328