Sensing of peptide hormones with dynamic combinatorial libraries of metal-dye complexes: The advantage of time-resolved measurements

Article abstract of DOI:10.1039/b912400d

A dynamic combinatorial library of metal-dye complexes was obtained by reacting aqueous solutions of the dyes Methyl Calcein Blue, Arsenazo I, and Xylenol Orange with CuCl₂ and NiCl₂. The mixture gave a characteristic UV-Vis response

Full text of DOI:10.1039/b912400d

View Article Online / Journal Homepage / Table of Contents for this issue
www.rsc.org/obc
Volume 7 | Number 22 | 21 November 2009 | Pages 4549–4800
ISSN 1477-0520
FULL PAPER
Kay Severin et al.
Highlightsin
Sensing of peptide hormones with
dynamic combinatorial libraries of
metal–dye complexes: the advantage
of time-resolved measurements
ChemicalScience
In this issue...
1477-0520(2009)7:22;1-C

PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Sensing of peptide hormones with dynamic combinatorial libraries
of metal–dye complexes: the advantage of time-resolved measurements†
Friederike Zaubitzer, Thomas Riis-Johannessen and Kay Severin*
Received 23rd June 2009, Accepted 24th July 2009
First published as an Advance Article on the web 3rd September 2009
DOI: 10.1039/b912400d
A dynamic combinatorial library of metal–dye complexes was obtained by reacting aqueous solutions
of the dyes Methyl Calcein Blue, Arsenazo I, and Xylenol Orange with CuCl2 and NiCl2. The mixture
gave a characteristic UV-Vis response upon addition of the peptide hormones angiotensin I and
angiotensin II. This allowed distinguishing pure samples of peptide hormones from mixtures. The
discriminatory power of the sensor was enhanced when the several UV/Vis measurements were
performed during the equilibration process of the library.
independent sensor units, whereas a single UV/Vis or fluorescence
Introduction
measurement is sufficient for a DCL sensor.
Adaptive chemical networks are formed by mixing molecular
The utilization of DCLs for sensing purposes is a relatively new
building blocks, which are able to bind to each other via reversible
interactions. If there is cross-reactivity between the various
concept and so far there are only few studies in this direction.
Guiseppone and Lehn have investigated constitutionally dynamic
polymers (“dynamers”),3 which were obtained by polycondensa-
tion of fluorescent and non-fluorescent diamine monomers with
2,7-fluorenebiscarbaldehyde in the presence of variable amounts
of Zn2+ ions.4 The polymer composition—and consequently the
fluorescence—was found to depend on the concentration of Zn2+.
The dynamic polymer can therefore be regarded as a system
which is able to sense Zn2+ by an analyte-induced constitutional
rearrangement.
The group of Anslyn has investigated DCL sensors based on
two synthetic receptors for oxoanions and two dyes.5 Both dyes can
bind to both receptors and four different receptor–dye complexes
were formed. This system was used to simultaneously determine
the concentration of malate and tartrate in aqueous solutions. A
related approach was employed to determine the concentration of
Ca2+ and citrate in flavored vodkas.6
building blocks, the assemblies are formed in a combinatorial
fashion and the resulting network is referred to as a dynamic
combinatorial library (DCL).1 The concentrations of the different
members of a DCL depend on the environment of the respective
system (pH, solvent, temperature, etc.). It is possible to alter
the library composition by adding molecules that interact with
some members of the library. If the library composition can be
transduced into a signal output, the DCL can be used as a sensor
(Scheme 1).
The Lavigne group has studied the analyte induced aggregation
of carboxylic acid functionalized poly(thiophene)s. The system
was used to identify low millimolar concentrations of a,w-
diamines.7 In a subsequent study, they have shown that a sensor
of this kind can be used to quantify histamine in a fish sample.8
The most recent investigation about sensing with adap-
tive chemical systems was published by Margulies and
Hamilton.9 As sensors, they used dynamic libraries of fluorescent
G-quadruplexes. The interaction with proteins resulted in charac-
teristic fluorescence emission patterns, which allowed identifica-
tion of the respective protein.
Our group has shown that combinatorial mixtures of transition-
metal complexes and commercially available dyes can be used
as powerful sensors for small peptides10 and nucleotides.11 In
extension of this work, we have reported recently a sensor that is
able to determine in retrospect the history of analyte variations.12
The study demonstrated that the adaptation process of a DCL
contains useful information. This inspired us to explore whether
time-resolved measurements with DCL sensors are beneficial
for the resolution. In the following we will demonstrate that
this is indeed the case. A DCL sensor, which is analyzed in a
Scheme 1 Basic principle of a DCL sensor: a dynamic mixture of
colored (or fluorescent) compounds A–G undergoes an analyte-induced
re-equilibration. The change in color (or fluorescence) can be used to
obtain information about the identity, the quantity, or the purity of the
analyte.
For a sensor of this kind, the information about the analyte is
distributed over the spectrum. The spectrum therefore represents
a ‘fingerprint’ of the analyte. To correlate the changes at different
wavelengths with the analyte properties of interest (identity,
quantity, purity), it is advantageous to use multivariate analyses.
In this regard, a DCL sensor is related to sensor arrays.2 However,
sensor arrays are analyzed by measuring the response of the
´
Institut des Sciences et Inge´nierie Chimiques, Ecole Polytechnique Fe´de´rale
de Lausanne (EPFL), 1015, Lausanne, Switzerland. E-mail: kay.severin@
epfl.ch; Fax: +41(0)21 6939305
† Electronic supplementary information (ESI) available: Mass spectrom-
etry and UV/Vis data; details about the multivariate analyses. See DOI:
10.1039/b912400d
4598 | Org. Biomol. Chem., 2009, 7, 4598–4603
This journal is
The Royal Society of Chemistry 2009
©

Table 1 Fitted cumulative formation constants (log bm,l ) for the complexes [Mm(dye)l ] (M = Cu2+, Ni2+) observed during UV/Vis titrations in buffered
aqueous solution (CHES buffer, 0.1 M, pH 8.4, 298 K). The numbers in parentheses give the calculated error of the fitting procedure
Xylenol Orange
Arsenazo I
Methyl Calcein Blue
M2+
log b1,1
log b2,1
log b1,2
log b1,1
log b2,1
log b1,2
log b1,1
Cu2+
Ni2+
8.75(3)
7.75(5)
15.98(6)
13.86(5)
—
12.80(8)
8.69(4)
7.53(4)
15.53(7)
—
—
12.52(3)
7.34(3)
7.19(2)
time-resolved fashion, is able to distinguish mixtures of the
hormones angiotensin I and angiotensin II with a significantly
improved resolution compared to a sensor that is analyzed under
pseudo-equilibrium conditions.
bis(N-carboxylate)amine binding sites on the dye. The spectral
variations ceased at metal/dye ratios of ≥ 2.0 and, accordingly,
the absorbance changes for integer wavelengths in the range
400–650 nm could be satisfyingly modeled by the two equilibria
leading to the formation of [M(XO)] and [M2(XO)] (M =
Cu2+, Ni2+) complexes (Table 1 and Fig. S1 and S2, ESI†).
Electrospray ionization mass spectra (ESI-MS) of solutions
containing [XO]tot = 0.1 mM (pH 8.4) and metal/dye ratios in
the range 0.5–2.0 reflect these observations. With low relative
concentration of metal, the spectra are dominated by peaks
corresponding to complexes of the type [M(XO)]. Increasing
the metal concentration to [M]tot = 0.2 mM then results in
the appearance of peaks for dinuclear species with [M2(XO)]
formulations (M = Cu2+, Ni2+; Table S1, ESI†).
Results and discussion
First, a small screening was performed in buffered aqueous so-
lution (2-(N-cyclohexylamino)ethanesulfonic acid (CHES) buffer,
0.1 M, pH 8.4), in which the stability and solubility of fourteen
commercially available dyes (50 mM) and their binding towards
Ni2+ and Cu2+ (0.1 mM each) were investigated by UV/Vis
spectroscopy (for details see experimental section). In order to
obtain optimal signal output from a dynamic combinatorial
library consisting of a mixture of dyes and metals, complexation
induced absorbance changes must be in different regions of the
UV/Vis spectrum. Methyl Calcein Blue (MCB, Amax = 360 nm),
Arsenazo I (AI, Amax = 500 nm) and Xylenol Orange (XO,
Amax = 580 nm) best fulfilled these requirements under the chosen
experimental conditions (Scheme 2).
Titrations with AI suggest markedly different speciation be-
havior for the two metals. For Cu2+, clear inflections occur at
metal/dye ratios of 1.0 and 2.0 (Fig. 1), and the data can
be excellently modeled by considering similar [Cu(AI)]- and
[Cu2(AI)]-type complexes to those observed for XO (Table 1). For
Ni2+, however, although perhaps less apparent to the naked eye,
inflection points are found at metal/dye ratios of 0.5 and 1.0.
Formation of the implied [Ni(AI)2] species is also supported by
evolving factor analysis,13,14 which strongly invokes the presence
of three absorbing species to account for the spectral variations at
metal/dye ratios of ≤ 1.0. After a ratio of ~1.0 is attained, however,
no further changes occur and the data can thus be modeled by
sequential formation of the binary complex [Ni(AI)2] and the 1 : 1
complex [Ni(AI)] (Table 1 and Fig. 1). Parallel ESI-MS titrations
showed intense peaks for the free dye and 1 : 1 complexes of
AI with Cu2+ and Ni2+, but we were unable to observe either
the higher nuclearity [Cu2(AI)] complex or the binary complex
Scheme 2 Molecular structures of the three dyes used to construct the
sensor.
Binding of the three selected dyes to Cu2+ and Ni2+ cations
was then further investigated by performing a series of UV/Vis
spectrophotometric titrations. In all cases, adding sequential
aliquots of aqueous metal solutions to buffered aqueous solutions
of the dyes caused marked absorbance changes in distinct regions
of the UV/Vis spectrum (overall range 300–700 nm). Plots of
absorbance vs. metal/dye ratio for selected wavelengths were
examined to identify possible stoichiometries for the participating
species. In the case of XO, clear inflection points at metal/dye
ratios of 1.0 and 2.0 were observed for both Cu2+ and Ni2+,
indicating the successive fixation of two ions to the two pendant
Fig. 1 Changes in absorbance observed during titrations of Cu2+ (left)
and Ni2+ (right) into solutions of AI ([dye]tot = 15 mM, CHES buffer,
0.1 M, pH 8.4, 298 K). Below the spectral overlays are corresponding
plots of absorbance vs. metal/dye ratio for three selected wavelengths (the
lines represent best fits to the models discussed in the text).
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 4598–4603 | 4599
©

[Ni(AI)2] suggested by the UV/Vis data (Table S2, ESI†). We
note, however, the persistence of intense peaks corresponding to
the free dye, even in the presence of excess metal. It is thus plausible
that observation of the latter complexes is impeded by gas-phase
fragmentation processes operating in the spectrometer.
good ligand for transition metals, it was expected that the two
peptides bind differently to Cu2+ and Ni2+.
The basic concept of the DCL sensor is illustrated in Scheme 3.
MCB (50 mM), AI (25 mM) and XO (12.5 mM)17 were mixed with
either one or both peptides ([peptide]tot = 20 mM), followed by the
addition of a 1 : 1 mixture of Ni2+ and Cu2+ ([M2+] = 0.2 mM) at
t = 0 min (CHES, 0.1 M, pH 8.4). The solutions were analyzed
by UV/Vis spectroscopy at different times and the data were
evaluated by multivariate analyses.
Analogous titrations with MCB and Cu2+ showed the exclusive
formation of a mononuclear [Cu(MCB)] complex with one
absorbance inflection occurring at a metal/dye ratio of 1.0 (Fig.
S3, ESI†). Once again, however, the same titration with Ni2+
showed biphasic behavior and the data could only be modeled
satisfactorily by considering the formation of a binary complex
[Ni(MCB)2], in addition to the 1 : 1 complex [Ni(MCB)] (Table 1
and Fig. S4, ESI†). The mass spectra show complexes of the latter
formulation in abundance for both metals (Table S3, ESI†), but as
for the AI/Ni2+ system, no binary complexes were detected in the
gas phase for solutions containing e.g. a 1 : 2 ratio of Ni2+ : MCB
([MCB]tot = 0.1 mM, pH 8.4).
We stress that these complex formulations are limited to the
basic stoichiometries of the metal/dye complexes: no concrete
conclusions regarding chemical structure, charge and/or proto-
nation state of the various species can be drawn from the current
observations. The fitted values listed in Table 1 may thus be
considered effective stability constants, whose significance hold
strictly within the conditions chosen for the current investigation.
They are also subject to error arising from e.g. salt impurities in
the dye precursors and (in some cases) high correlation between
the calculated spectra of the participating species.
These considerations aside, combining all three dyes with both
Cu2+ and Ni2+ will clearly yield a dynamic combinatorial library of
complexes, metal ions and free dyes, with the metal–dye complexes
being not only homometallic and homoleptic, but ideally compris-
ing mixed species. To provide evidence of the variety of the species
in this DCL, we studied an aqueous mixture (pH 8.4) of the three
dyes ([MCB] = 50 mM, [AI] = 25 mM and [XO] = 12.5 mM) and
Ni2+ and Cu2+ (0.1 mM each) by ESI mass spectrometry. Homo-
and heterometallic, as well as homo- and heteroleptic compounds
(Table S4, ESI†) comprising all dyes and metal ions could indeed
be identified. These results give an insight into the complexity of
such a system. In fact, it is very likely that in reality this DCL
contains even more species, as due to instability in the gas-phase
some metal complexes might not be detectable by mass spectrom-
etry. It should be noted that the formation of heterometallic and
heteroleptic complexes is a unique feature of the DCL sensor
approach. In a more classical sensor array with individual metal–
dye combinations as sensor units, such species are not formed and
can thus not contribute to the resolution of the sensor.15
Scheme 3 Experimental procedure for the sensing of the peptide hor-
mones angiotensin I and angiotensin II. At t = 0 min the metal salts were
added to a solution containing the three dyes and either one or a mixture
of both peptides. The data from UV/Vis measurements at different times
were evaluated by multivariate analysis.
Time-resolved UV/Vis studies of the sensing ensembles showed
that the major absorbance changes occurred in the first 20 min
following addition of the metal ions (Fig. 2). After this initial
rapid phase, the system took a further 4 days to fully equilibrate.
The slow equilibration kinetics of the DCLs contrast with
those characterizing the bimolecular reactions of any two of
The DCL was subsequently studied for sensing of the peptide
hormones angiotensin I (Asp-Arg-Val-Tyr-Ile-His-Pro-Phe-His-
Leu) and angiotensin II (Asp-Arg-Val-Tyr-Ile-His-Pro-Phe). Hu-
man angiotensin I and angiotensin II are ten- and eight-amino acid
peptides respectively, with angiotensin I being the physiologically
inactive precursor of angiotensin II. Angiotensin II is involved in
several physiological and pathophysiological effects in the human
body, such as vasoconstriction, which leads to the increase of blood
pressure. Permanent hypertension can be the cause for chronic
renal failure and cardiac failure, amongst others.16 Angiotensin I
is transformed to angiotensin II through removal of the terminal
histidine and leucine residues by the exopeptidase Angiotensin-
converting enzyme (ACE). Since the side chain of histidine is a
Fig. 2 UV/Vis absorbance data at different wavelengths showing the
equilibration over time of DCLs containing Ni2+, Cu2+, Methyl Calcein
Blue, Arsenazo I, Xylenol Orange and angiotensin I (filled symbols) or
angiotensin II (empty symbols).
4600 | Org. Biomol. Chem., 2009, 7, 4598–4603
This journal is
The Royal Society of Chemistry 2009
©

the (metal/dye) components, which are typically complete within
minutes. Such effects are, however, to be expected since the true
energy minimum of the complex network of reactions operating in
the DCL is undoubtedly reached via numerous close-lying minima,
each with a corresponding kinetic barrier to be overcome. In
this respect, equilibration of the DCL is therefore slow enough
to be followed by simple UV/Vis measurements. Furthermore,
the kinetic profile of the equilibrating system was found to vary
according to the analyte. This phenomenon is especially appealing
for systems where different analytes give rise to the same signal
when the sensor is read out at thermodynamic equilibrium.
Our DCL sensor was first implemented to differentiate between
angiotensin I, angiotensin II and a 1 : 1 mixture of the two
peptides. The absorbance changes (300–800 nm) were recorded at
different times (t = 2, 5, 10, 20, 40, 60 min) until the system reached
a pseudo-equilibrium defined as t = 24 h. For each peptide, the
experiment was repeated five times. The data were then evaluated
by linear discriminant analysis (LDA).18 The results of the LDA
suggested that the data from four wavelengths and at t ≤ 10 min
was enough to tell apart the three sample types.
The DCL sensor was then used to distinguish between 11
mixtures (six repetitions each) of angiotensin I and II ([an-
giotensin I] = (20 - x) mM; [angiotensin II] = x mM; with x =
0, 2, 4, 6,. . ., 20). The sensing experiments were performed under
the conditions described above and the samples were analyzed
by UV/Vis spectroscopy (330–650 nm) at t = 2, 5, 10 min and
24 h. The absorbance data at l = 590, 560, 480 and 340 nm was
then evaluated by a LDA. In order to determine to what extent
the kinetic profiles contributed to the discriminative power of the
DCL, the data was analyzed twice. First, only the data obtained
from the DCL at pseudo equilibrium (t = 24 h) was included in
the LDA (Fig. 3, top). In a succeeding analysis, all the data was
used for evaluation (Fig. 3, bottom). The corresponding results
are displayed as score plots.
Fig. 3 Score plots of LDA treating only the data after 24 h at
pseudo-equilibrium (top) and including the data after 2, 5, 10 min and
24 h (bottom): [angiotensin I]tot = (20 - x) mM, [angiotensin II]tot = x mM;
x = 20 (maroon, ᭜), 18 (blue, ᭹), 16 (grey, ꢀ), 14 (orange, +), 12 (green,
ꢀ), 10 (yellow, ᭢), 8 (light blue, ¥), 6 (light green, ꢀ), 4 (light blue, ꢁ), 2
(green, ᭜), 0 (dark blue, +).
For both analyses, most of the variance of the data was
contained in score 1 (94.8 and 93.5%, respectively). Important
differences were found for the resolution. Analysis (a) including
only the data obtained at pseudo-equilibrium yields a score plot
with considerable overlap of most of the scores of the mixtures.
With a correctness of classification of 83%,19 the data is not
sufficient to tell apart all sample types. It should be noted, however,
that the scores for the pure samples (angiotensin I: dark blue
crosses, angiotensin II: maroon diamonds) are well clustered
and separated. Therefore, this method is still good enough to
discriminate the pure samples. In analysis (b) where the entire
data set, i.e. also the data from the equilibration kinetics of the
DCL, is considered, the scores of all sample types are properly
classified and well separated. In spite of some score clusters being
in close proximity, the correctness of classification is 100%.20 The
supplementary data used in analysis (b) led to an improvement of
the classification of the scores and consequently an increase in the
discriminative power of the DCL sensor.
provided insight into the complexity of the resulting chemical
network. The DCL was used as a sensor for the peptide hormones
angiotensin I and angiotensin II. Despite its simplicity, the DCL
sensor was found to display a remarkable discriminative power:
mixtures of the two peptides were distinguished from pure samples
at low micromolar concentrations (20 mM). The resolution of
the sensor was significantly enhanced when the measurements
were performed in a time-resolved fashion. It should be noted
that so far, work with DCLs has primarily focused on systems
under thermodynamic control.1 The findings described above are
evidence that the kinetics of a DCL adaptation process may
contain useful information, in particular if dynamic networks are
used for challenging sensing problems.
Experimental
General
Alizarin Red S (Fluka), Arsenazo I (Lancaster), Azophloxine
(Fluka), Gallocyanine (Acros Organics), Methyl Calcein Blue
(Sigma), Methylene Blue (Sigma), Methyl Orange (Ciba), Methyl
Red (Acros Organics), Mordant Blue 9 (Acros Organics), Mordant
Blue 13 (Acros Organics), Naphthol Blue Black (Fluka), Orange
G (Acros Organics), Pyrocatechol Violet (Acros Organic), Xylenol
Conclusions
A DCL of metal–dye complexes was prepared by mixing the
dyes Methyl Calcein Blue, Arsenazo I and Xylenol Orange with
CuCl2 and NiCl2. Detailed UV/Vis titration experiments and
ESI mass spectrometry studies of the library and its components
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 4598–4603 | 4601
©

Orange (Fluka), NiCl2·6H2O (Strem Chemicals), CuCl2·2H2O
(Fluka), CHES buffer (Fluka), and angiotensin I (Asp-Arg-Val-
Tyr-Ile-His-Pro-Phe-His-Leu) and angiotensin II (Asp-Arg-Val-
Tyr-Ile-His-Pro-Phe) (both Bachem) were used as received. Stock
solutions of all dyes (1 mM, except Xylenol Orange: 0.5 mM),
the metal salts (10 mM), the peptides (0.1 mM) and CHES buffer
(pH 8.4, 0.2 M) were prepared in bidistilled water. All UV/Vis
spectra were recorded on a Lambda 35 spectrometer (Perkin
Elmer). Electrospray-ionisation MS data were acquired on a
Q-Tof Ultima mass spectrometer (Waters) fitted with a standard
Z-spray ion source and operated in the negative ionization mode.
Experimental parameters were set as follows: capillary voltage:
residuals of < 0.005, indicating that the data was excellently
modeled. Complete spectral overlays, calculated spectra and
speciation diagrams for all titrations are shown in the ESI.† The
standard deviations for the fitted parameters (listed in parentheses
in Table 1) were estimated using the sum-of-squared residuals and
inverted curvature matrices as obtained from the iterative fitting
processes (see ref. 13 for details).
Mass spectrometric titrations
Samples were prepared by mixing appropriate amounts of stock
solutions of the dye ([dye]tot = 0.1 mM) and the metal salt (0.05 ≤
[M2+]tot ≤ 0.2 mM) in water. The pH was then adjusted to 8.4 ( 0.2)
using NaOH(aq) (0.01 M).
◦
3 kV, sample cone: 50 V, source temperature: 80 C, desolvation
temperature: 200 ◦C, acquisition window: m/z 50-1000 in 1 s.
5 mL of the sample was introduced into the mass spectrometer by
infusion at a flow rate of 20 mL min-1 with a solution of ACN–
H2O 50 : 50 (v/v). External calibration was carried out with a
solution of phosphoric acid at 0.01%. Data were processed using
the MassLynx 4.1 software.
Mass spectrometry of the DCL
For sample preparation appropriate amounts of stock solutions of
the dyes and the metal salts were mixed to give final concentrations
of: [MCB]tot = 50 mM, [AI]tot = 25 mM and [XO]tot = 12.5 mM;
[Ni2+]tot = [Cu2+]tot = 0.1 mM. NaOH(aq) (0.01 M) was then added
to adjust the pH to 8.4 ( 0.2).
Dye screening
For each dye, UV/Vis spectra of the free dye and the metal–dye
combinations (final conc.: [dye]tot = 50 mM; [M2+]tot = 0.1 mM) in
buffered aqueous solution (CHES, 0.1 M, pH 8.4) were recorded
after standing for 1 h at room temperature. The following dyes were
studied and not considered for further investigation due to precip-
itation with either one of the metal ions: Azophloxine, Methylene
Blue, Naphthol Blue Black. Orange G and Pyrocatechol Violet
were not suitable, as they decompose in the presence of one of
the metal ions. Methyl Orange and Methyl Red do not display
sufficient binding affinity towards the metal ions and Alizarin Red
S, Mordant Blue 9 and Mordant Blue 13 gave weak signals, when
tested in a dynamic combinatorial library.
Sensing experiments
The DCL sensors were prepared by mixing appropriate amounts
of stock solutions containing the dyes, the metal salts and the
buffer. After addition of either one peptide or a mixture of
both peptides, UV/Vis spectra were recorded at t = 2, 5, 10,
20, 40, 60 min and 24 h or t = 2, 5, 10 min and 24 h. Each
series of measurements was repeated five or six times. The final
concentrations were: [MCB]tot = 50 mM, [AI]tot = 25 mM and
[XO]tot = 12.5 mM, [Ni2+]tot = [Cu2+]tot = 0.1 mM, and [CHES]tot
=
0.1 M. The peptide concentration was varied in 0.1 eq. increments
with [angiotensin I]tot = (20 - x) mM and [angiotensin II]tot
=
x mM. The data was analyzed with the commercially available
statistics program SYSTAT (version 11.0) by a linear discriminant
analysis algorithm.
Spectrophotometric titrations
For titrations of Xylenol Orange with CuCl2·2H2O and
NiCl2·6H2O, two series of measurements were performed: one in
which solutions of [M2+]tot = 5 mM were added to solutions of
[XO]tot = 7.5 mM (3 mL, + CHES, 0.1 M, pH 8.4, 298 K) in 1 mL
increments, and one in which solutions of [M2+]tot = 1 mM were
added to solutions of [XO]tot = 15 mM (1.5 mL, + CHES, 0.1 M,
pH 8.4, 298 K) in 5 mL increments. For titrations of Arsenazo I
with CuCl2·2H2O and NiCl2·6H2O, solutions of [M2+]tot = 10 mM
were added to solutions of [AI]tot = 15 mM (3 mL, + CHES, 0.1 M,
pH 8.4, 298 K) in 1 mL increments, whereas for Methyl Calcein
Blue, solutions of [M2+]tot = 10 mM were added to solutions of
[MCB]tot = 25 mM (3 mL, + CHES, 0.1 M, pH 8.4, 298 K) in
1 mL increments. After each addition, solutions were equilibrated
for 10 min at 298 K before measuring the absorption spectra in
the range 300–700 nm (XO and AI) and 300–500 nm (MCB).
The speciation models described in the main text were fitted to
Acknowledgements
We would like to thank Dr Laure Menin for performing mass
spectrometry and Dr Geoffrey Wood for help with data treatment.
The work was supported by the Swiss National Science founda-
tion, by the COST action CM0703 on Systems Chemistry, and by
the EPFL.
References
1 (a) S. Ladame, Org. Biomol. Chem., 2008, 6, 219–226; (b) R. F. Ludlow
and S. Otto, Chem. Soc. Rev., 2008, 37, 101–108; (c) J.-M. Lehn, Chem.
Soc. Rev., 2007, 36, 151–160; (d) P. T. Corbett, J. Leclaire, L. Vial, K. R.
West, J.-L. Wietor, J. K. M. Sanders and S. Otto, Chem. Rev., 2006,
106, 3652–3711.
2 (a) A. T. Wright and E. V. Anslyn, Chem. Soc. Rev., 2006, 35, 14–28;
(b) K. L. Albert, N. S. Lewis, C. L. Schauer, G. A. Sotzing, S. E. Stizel,
T. P. Vaid and D. R. Walt, Chem. Rev., 2000, 100, 2595–2626.
3 J.-M. Lehn, Prog. Polym. Sci., 2005, 30, 814–831.
4 (a) N. Guiseppone, G. Fuks and J.-M. Lehn, Chem.–Eur. J., 2006, 12,
1723–1735; (b) N. Guiseppone and J.-M. Lehn, J. Am. Chem. Soc.,
2004, 126, 11448–11449.
5 S. L. Wiskur, P. N. Floriano, E. V. Anslyn and J. T. McDevitt, Angew.
Chem., Int. Ed., 2003, 42, 2070–2972.
R
the spectral data using an in-house built routine in MATLABꢀ
which implements evolving factor analysis and a Newton–Gauss
multi-non-linear least squares fitting algorithm.13,14 For XO, the
speciation models were fitted to the data from both series of
measurements (for each metal) in a global multi non-linear least-
squares fit. In all cases the fits converged with sum-of-squared
4602 | Org. Biomol. Chem., 2009, 7, 4598–4603
This journal is
The Royal Society of Chemistry 2009
©

6 S. C. McCleskey, P. N. Floriano, S. L. Wiskur, E. V. Anslyn and J. T.
McDevitt, Tetrahedron, 2003, 59, 10089–10092.
7 T. L. Nelson, C. O’Sullivan, N. T. Greene, M. S. Maynor and J. J.
Lavigne, J. Am. Chem. Soc., 2006, 128, 5640–5641.
8 M. S. Maynor, T. L. Nelson, C. O’Sullivan and J. J. Lavigne, Org. Lett.,
2007, 9, 3217–3220.
14 (a) H. Gampp, M. Maeder, C. J. Meyer and A. D. Zuberbu¨hler, Talanta,
1985, 23, 1133–1139; (b) H. Gampp, M. Maeder, C. J. Meyer and A. D.
Zuberbu¨hler, Talanta, 1986, 33, 943–951.
15 For a discussion about the advantage of complex stoichoimetries for
sensors based on metal dye complexes, see: M. Kitamura, S. H. Shabbir
and E. V. Anslyn, J. Org. Chem., 2009, 74, 4479–4489.
16 M. G. Nicholls and J. I. S. Robertson, J. Hum. Hypertens., 2000, 14,
649–666.
9 D. Margulies and A. D. Hamilton, Angew. Chem., Int. Ed., 2009, 48,
1771–1774.
10 (a) A. Buryak and K. Severin, J. Comb. Chem., 2006, 8, 540–543; (b) A.
Buryak and K. Severin, Angew. Chem., Int. Ed., 2005, 44, 7935–7938.
11 A. Buryak, A. Pozdnoukhov and K. Severin, Chem. Commun., 2007,
2366–2368.
17 Different dye concentrations were chosen to account for differences in
the absorption coefficient.
18 P. C. Jurs, G. A. Bakken and H. E. McClelland, Chem. Rev., 2000, 100,
2649–2678.
12 A. Buryak, F. Zaubitzer, A. Pozdnoukhov and K. Severin, J. Am. Chem.
19 A standard jackknifed validation routine, in which one measurement at
a time was randomly omitted was performed: 70% of the measurements
were classified correctly.
20 82% of the measurements were classified correctly in a jackknifed
validation routine.
Soc., 2008, 130, 11260–11261.
13 M. Maeder and Y.-M. Neuhold, in Practical Data Analysis in Chem-
istry, ed. S. Rutan and B. Walczak, Elsevier Publishing Co, Amsterdam,
2007.
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 4598–4603 | 4603
©