N-Salicylideneamino acidato complexes of oxovanadium(IV). the cysteine and penicillamine complexes

Article abstract of DOI:10.1039/b404305g

Oxovanadium(iv) complexes with ligands derived from the reaction of salicylaldehyde with L-cysteine and with D- and D,L-penicillamine are prepared. The compounds are characterised by elemental analysis, spectroscopy (UV-VIS, CD, EPR), TG, DSC and magnetic susceptibility measurements (9-295 K). We discuss several aspects related to the structure of these complexes in the solid state and in solution; in particular, the possibility of forming thiazolidine complexes, and their comparison with the characterised complexes is studied by molecular mechanics and density functional theory calculations. The solution structures depend on pH and solvent, and while with L-Cys the spectroscopic results show trends similar to those of the L-Ala and L-Ser systems up to ca. pH 8-9, where thiolate coordination starts being detected, the penicillamine system is quite distinct, namely thiolate coordination occurs for pH > 6.5. In the presence of salicylaldehyde and V^IVO the desulfydration of cysteine proceeds rapidly, but no similar reaction occurs with penicillamine, although its decomposition is also activated. The DFT calculations do not indicate any energetic basis for this distinct reactivity, which possibly results from different complexes present in the Cys and Pen systems. In the cysteine system, the N-salicylidene dehydroalanine-V^IVO complex V is believed to form in an intermediate stage of the desulfydration. Further, addition of several nucleophiles to the cysteine reaction mixtures produce amino acid derivatives by a Michael-type base-catalysed addition, a result compatible with the formation of V. The products of these reactions were analysed by TLC and HPLC, and in some cases isolated.

Full text of DOI:10.1039/b404305g

View Article Online / Journal Homepage / Table of Contents for this issue
F U L L P A P E R
N-Salicylideneamino acidato complexes of oxovanadium(IV). The
cysteine and penicillamine complexes†
João Costa Pessoa,*a Maria J. Calhorda,b,f Isabel Cavaco,a,c Paulo J. Costa,b,f Isabel Correia,a
Dina Costa,a,c Luís F. Vilas-Boas,a,d Vítor Félix,g Robert D. Gillard,e Rui T. Henriquesa,d and
Robert Wigginse
a
Centro Química Estrutural, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa,
Portugal
b
Instituto de Tecnogia Química e Biológica, Av. da República, Apt. 127, 2781-901 Oeiras,
Portugal
c
Now at Universidade do Algarve, Unidade de Ciências Exactas, Campus de Gambelas,
8000 Faro, Portugal
Instituto Tecnológico e Nuclear, Departamento de Química, Estrada Nacional 10,
d
2685 Sacavém, Portugal
Department of Chemistry, University of Wales, P.O. Box 912, Cardiff, UK CF1 3TB
Department of Chemistry, Faculdade de Ciências, Universidade de Lisboa, 1749-016 Lisboa,
e
f
Portugal
g
Departamento de Química, CICECO, Universidade de Aveiro, 3810-193 Aveiro, Portugal
Received 22nd March 2004, Accepted 9th July 2004
First published as an Advance Article on the web 24th August 2004
Oxovanadium(IV) complexes with ligands derived from the reaction of salicylaldehyde with L-cysteine and with D- and
D,L-penicillamine are prepared. The compounds are characterised by elemental analysis, spectroscopy (UV-VIS, CD,
EPR), TG, DSC and magnetic susceptibility measurements (9–295 K). We discuss several aspects related to the structure
of these complexes in the solid state and in solution; in particular, the possibility of forming thiazolidine complexes,
and their comparison with the characterised complexes is studied by molecular mechanics and density functional theory
calculations. The solution structures depend on pH and solvent, and while with L-Cys the spectroscopic results show trends
similar to those of the L-Ala and L-Ser systems up to ca. pH 8–9, where thiolate coordination starts being detected, the
penicillamine system is quite distinct, namely thiolate coordination occurs for pH > 6.5. In the presence of salicylaldehyde
and VIVO the desulfydration of cysteine proceeds rapidly, but no similar reaction occurs with penicillamine, although its
decomposition is also activated. The DFT calculations do not indicate any energetic basis for this distinct reactivity, which
possibly results from different complexes present in the Cys and Pen systems. In the cysteine system, the N-salicylidene
dehydroalanine–VIVO complex V is believed to form in an intermediate stage of the desulfydration. Further, addition of
several nucleophiles to the cysteine reaction mixtures produce amino acid derivatives by a Michael-type base-catalysed
addition, a result compatible with the formation of V. The products of these reactions were analysed by TLC and HPLC,
and in some cases isolated.
Introduction
Complexes of N-salicylideneamino acids have been the subject
of extensive research, typified by their vanadium compounds.1–16
With vanadium-(IV) and -(V) these Schiff base (SB) complexes
often have coordination geometries such as those in I or II. In a few
cases, dinuclear oxo-bridged VIV–O–VV or VV–O–VV compounds
III have been obtained4,5,11,12,16,17 and were characterised by X-ray
diffraction.
Upon standing in methanolic solution in air [VIVO(sal–aa)(H2O)]
complexes (sal–aa = N-salicylidene-amino acidato) are spontan-
eously oxidised to [VVO(sal–aa)(OMe)(OHMe)] II. If 2,2′-bipyri-
dine (bpy) or pyridine (py) is present in solution, they coordinate to
VIVO2+, forming complexes with binding modes such as IV, where
NN denotes bpy or 2 × py, and oxidation may be slowed down or
avoided.1,7,8,13,18
† Electronic supplementary information (ESI) available: ESI-1 Some TLC
In this work we isolate several oxovanadium(IV) complexes with
ligands derived from the reaction of salicylaldehyde with L-cysteine
and D- or D,L-penicillamine (hereafter designated by Pen), and we
study several aspects related to the structure of these complexes
in the solid state and in solution, including the use of molecular
mechanics and density functional calculations.
The change in ligand reactivity caused by the coordination of
amino acids to a wide variety of metal ions is well known. The
activating effect of the metal ion may be extended by coordinating
data for the VIVO/HSal/aa mixtures (aa = Cys, Pen, CysMe, CysEt) and the
corresponding complexes in ethanol. ESI-2 Preparation of complexes 2–9.
ESI-3 Racemization experiments. ESI-4 Synthesis of amino acids based on
the dehydroalanine complex V. ESI-5 Differential scanning calorimetric
curves. ESI-6 Circular dichroism spectra. ESI-7 CD and EPR spectra of
aqueous–ethanolic (3:1) solutions containing amino acid:salicylalde-
hyde:VIVO (1:1:1). ESI-8 Chiral-HPLC. ESI-9 General information about
the DFT-calculated structures. ESI-10 Molecular mechanics calculations.
ESI-11 Isomers of the thiazolidine complexes [VIVO(tz–Cys)(H2O)] and
[VIVO(tz–Pen)(H2O)]. See http://www.rsc.org/suppdata/dt/b4/b404305g/
T h i s j o u r n a l i s
©
T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 4
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6
2 8 5 5

View Article Online
benzylmercaptan (Aldrich), dimethylamine (Carlo Erba), and
solvents were used as purchased. S-hydroxyethylcysteine (see
ESI-4†) and 2-amino-3-(diethylamino)-propanoic acid30 were
prepared for use as standards (for TLC and HPLC). Oxygen was
removed from the solvents used for preparation or spectroscopic
studies by bubbling N2.
TLC experiments
Most preparations and syntheses were monitored by TLC, on Merck
TLC plates (Art. 5626, 10 × 20 cm or Art. 5721, 20 × 20 cm).
Samples of 1–5 l of the reaction mixture were applied to the
plates 20 mm from the bottom. In the experiments for synthesis
of Cys derivatives 12–19, normally two samples of the reaction
mixtures were applied: (i) sample acidified with a strong acid (1–3
M HCl, H2SO4 or HNO3), (ii) sample taken from the reaction flask
and diluted (or not) with the same volume of water as of acid in
(i). Elutions were carried out in Camag twin chambers with walls
covered with filter paper impregnated with the eluent. Eluents
used were: (A) ethanol/water (7:3), (B) butanol/ethanol/propionic
acid/water (10:10:2:5), (C) chloroform/methanol/ammonia (17%)
(2:2:1), (D) butanol/acetone/water/diethylamine (10:10:5:2).
When the eluents reached approximately 120 mm from the bottom,
the plates were removed and dried. The chromatogram was
developed: (i) with ninhydrin–collidine–copper solution prepared
according to Moffat and Lytle31 and (ii) with iodine vapours in an
enclosed chamber. In some cases the detection of sulfur-containing
compounds used a spray of a PdCl2/HCl solution. The presence
of particular products in samples was confirmed by comparing
their RF values with standards and by spiking the samples with the
corresponding standard, using the same TLC plate.
the amino acid as a Schiff base using a suitable carbonyl fragment
such as pyridoxal, salicylaldehyde or pyruvate.19,20 Snell21,22 showed
that many reactions involved in the metabolism of amino acids, e.g.
transamination, -decarboxylation, racemization, desulfydration
etc., catalysed by enzymes that require pyridoxal phosphate as a co-
factor, could be simulated using pyridoxal, a metal ion (Fe3+, Cu2+,
Al3+) and the amino acid. Vanadium is one of the most active metal
ions in -eliminations, a process involved in e.g. desulfydration.23–28
Bergel also showed that the desulfydration of cysteine is activated
in the presence of pyridoxal and vanadium salts,23,24 and this seems
to provide a good analogy to the action of cysteine desulfydrase,
a pyridoxal phosphate dependent enzyme.29 However, no similar
reaction occurred with penicillamine, although this also degrades.
The decomposition of the amino acids activated by vanadium could
explain the difficulty in isolating vanadium complexes involving
Cys and Pen.
The desulfydration of cysteine also occurs using salicylaldehyde
instead of pyridoxal. The N-salicylidenedehydroalanine complex V
is believed to form in an intermediate stage (Scheme 1),24 the
decomposition of the amino acid leading to the formation of NH3,
pyruvic acid and other products. In this work we also show that
addition of several nucleophiles (HNu) to the reaction mixture
containing vanadium salts, salicylaldehyde (Hsal) and Cys produce
amino acid derivatives by a Michael-type base-catalysed addition,
a result compatible with the formation of V (sal–DHAla) and VI.
The products of these reactions were analysed by means of TLC and
HPLC, and in some cases isolated.
TLC experiments to monitor the preparation of complexes
2–9. All preparations were monitored by TLC and ESI-1†
summarises the results. Distinct brown spots (Sp) corresponding
to the vanadium complexes were always detected. Distinct spots
(AA) were also detected at the RF of the free amino acids and the
brown spots Sp each tailed to the spot AA. This tailing was more
severe for eluent B (acid eluent). Before development, depending
on the sample, a pale blue spot (from VOSO4) could frequently be
detected at RF ≈ 0. In many cases, a weak bluish white continuous
tailing could be seen up to the RF of the complexes. All these TLC
results are similar to those obtained in the preparations of [VIVO(sal-
L-Ala)(X)]7 and [VIVO(dsal-L-aa)(X)] (X = H2O, py or bpy;
dsal = Hsal and derivatives).13 The [VIVO(sal-L-aa)(X)] complexes
were eluted in the TLC plates presumably with X = EtOH, butanol,
bpy or H2O, the RF values also being determined by the relative
amount of EtOH in the eluent. As complexes containing bpy
systematically presented lower RF values, in spite of the fact that
EtOH and H2O are present in much higher concentrations, we
conclude that the bpy–V bonds are the last to be hydrolysed during
elution. In the very first stages of preparation of 2–7 (above), before
the addition of the VOSO4 solution (and before precipitation of the
thiazolidines), three other weak spots were normally detected
besides the cysteine or penicillamine spots: one corresponding to
the aa disulfide, and two reddish/orange spots at RF values similar to
those of 2–7. These two spots possibly correspond to the free Schiff
bases and the thiazolidines.
TLC experiments to monitor the desulfydration of sulfur-
containing amino acids and the preparation of compounds
12–19
Scheme 1 Summary of the present reactions activated by
oxovanadium(IV). HNu represents a nucleophile which may add to the -
carbon atom of the coordinated amino acid, and the dehydroalanine ligand
in V will be designated by sal–DHAla.
All reactions were monitored by TLC and eluents B, C or D were
used (always at least two eluents). In some cases 20 × 20 cm plates
were used and were divided in two parts, 10 × 20 cm each, an
equal set of samples being applied on each part. After elution and
drying the plates, one half was covered with a 10 × 20 cm glass,
and the other half developed with the ninhydrin–collidine–copper,
followed, after ca. 2–5 h, by iodine vapours. The second half was
developed later with a spray of the PdCl2/HCl solution, to detect the
S containing compounds.
Experimental
L-cysteine (Sigma), D-penicillamine (Sigma), salicylaldehyde
(Aldrich), VOSO4·5H2O (Merck) or VOSO4·H2O (BDH), 2,2′-
bipyridine (Merck), sodium acetate (Merck) lanthionine (BDH),
S-methyl-L-cysteine (Sigma), S-ethyl-L-cysteine (Sigma),
S-benzyl-L-cysteine (Aldrich), 2-mercaptoethanol (Merck),
methanethiol (Aldrich), ethanethiol (Janssen), thiophenol (Merck),
2 8 5 6
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6

View Article Online
Table 1 Elemental analysis of VO(sal–aa)(X) compounds (aa = sulfur containing amino acids; X = H2O or bpy)a
Compound
%C
%H
%N
%S
%V
Colour
2 [VO(sal–L-Cys)(H2O)]
Calculated: for C10H11NSO5V
3 [VO(sal–L-Cys)(bpy)]·1·2H2O
39.2
38.97
51.3
51.33
44.7
45.00
45.8
45.54
53.6
53.92
53.6
53.92
54.0
53.66
53.0
3.6
3.60
3.9
4.18
4.7
4.72
5.0
4.67
4.7
4.95
4.7
4.95
4.9
4.71
4.9
4.5
4.54
8.7
8.98
4.0
4.37
3.9
3.90
7.8
7.99
7.8
7.99
8.2
8.53
8.1
10.5
10.40
6.6
6.85
9.7
10.01
9.3
8.94
6.1
6.10
6.1
6.10
6.6
6.51
6.1
16.6
16.53
n.a.
Greyish-blue
Orange
Green
Calculated for: C20H19.4N3SO5.2
V
4 [VO(sal–D-Pen)(H2O)]
15.3
15.9
13.6
14.2
9.5
9.69
9.6
9.69
n.a.
Calculated for: C12H15NSO5Vb,c
5 [VO(sal–D,L-Pen)]·0.2H2O·0.8EtOH
Calculated for: C13.6H18.2NSO5Vb
6 [VO(sal–D-Pen)(bpy)]·0.8H2O·0.8EtOH
Calculated for: C23.6H27.4N3SO5.6Vc
7 [VO(sal–D,L-Pen)(bpy)]·1.0H2O
Calculated for: C22H23N3SO5V
8 [VO(sal–CysSMe)(bpy)]·0.8EtOH
Calculated for: C22.6H22.2N3SO4.8
9 [VO(sal–CysSEt)(bpy)]·1.2H2O·0.3EtOH
Calculated for: C22.8H25.8N3SO5.6
Orange
Orange
Orange
Orange
Orange
White
V
10.0
9.90
V
53.23
52.5
52.69
54.3
4.90
4.9
5.00
6.0
8.17
6.0
6.14
5.3
6.23
14.3
14.07
12.3
12.06
10 (tz–L-Cys)·0.15H2O
Calculated for: C10H11NSO3·0.15H2O
11 (tz–D-Pen)·0.7H2O
White
Calculated for: C12H15NS O3·0.8H2O
54.20
6.22
5.27
a Compound 1 is brown and its formulation, as determined by X-ray diffraction,7 corresponds to [VO(sal–L-Ala)(bpy)]·1/8H2O·1/2EtOH. A few other
compounds of the type [VO(sal–L-aa)(bpy)] (aa = several) are known13 and their colour is brick-red. b The analytical results give better fits with formulations
including 0.1–0.2 further moles of H2O per V atom. c The CD signals of 4 and 6 were very weak, indicating that the amino acid partly racemized.
HPLC experiments
Experiments were carried out with stirring at 40 °C and at the pH
values: 3, 5.3, 6, 8, 9, in 250 ml 3/4 necked round flasks. First,
3–10 mmoles of L-Cys and 3–10 mmoles of VOSO4·5H2O were
dissolved in ca. 40–60 ml of water, and the solution deoxygenated
by bubbling nitrogen; this stream of nitrogen was maintained
throughout, except when solutions or solids were being added,
the pH measured, or during sampling for TLC. Ethanol (8–12 ml)
containing the required amount of Hsal (3–10 mmoles) was added
(tR = 0). These solutions normally corresponded to pH ~3. Sodium
acetate was added (moles = 3× those of L-Cys), and the pH adjusted
to pH 5.3 or 6 with acetic acid or base. For the experiments at pH 8
and 9, solid Na2CO3 was added after the acetate until the required
pH. The N2 entered through one of the necks and came out through
a 30 cm condenser and bubbled through 3 flasks, each containing
a 100 ml aqueous solution of 3CdSO4·8H2O (6.1 g, 8.0 mmole)
and 1 ml of concentrated H2SO4. The H2S produced in the reaction
mixture, and transferred to the flasks, precipitates as CdS (yellow).
In the experiments for quantitative determination of the H2S
produced, the reaction was stopped after ~24 h by adding acid
until pH ~3, and N2 was still kept bubbling for ~30 min. In other
experiments of longer duration, monitored by TLC, apparently no
more H2S was produced after ~24 h of reaction. The amount of
CdS was determined by adding a known amount of standard iodine
solution, and titrating the excess iodine either with a standard
thiosulfate solution, or with a standard arsenious oxide solution.
These used a Jasco HPLC system including Jasco 870-UV (ab-
sorbance) and 821-FP (fluorescence) detectors. A Rheodyne 7125
injection valve (20 l loop) and a reverse phase column (LiChrosorb
C18-5 m particles, 250 × 4 mm) were used. Eluents depended on the
samples to be analysed (see below); they were always filtered before
use (45 m filters) and degassed; its flow rate was 1.0 ml min−1.
Amino acids were analysed using three different precolumn
derivatizations: (A) the usual o-phthaldialdehyde/mercaptoethanol
procedure for amino acid analysis,32 or (B) the iodoacetate/o-
phthaldialdehyde/mercaptoethanol33 procedure for cysteine and
penicillamine and, (C) the o-phthaldialdehyde/chiral mercaptans
procedure34 to separate optically active amino acids: N-acetyl-L-
cysteine (NAC) and N-acetyl-D-penicillamine (NAP) were used for
this purpose. The presence of products in samples was confirmed
by comparing their retention times with standards, and by spiking
the samples with the corresponding standard.
Preparation of complexes 2–9
Complexes [VO(sal-L-Cys)(H2O)] 2, [VO(sal-L-Cys)(bpy)]·nH2O
3, [VO(sal-D-Pen)]·nH2O·mEtOH 4, VO(sal-D,L-Pen)·nH2O·mEtOH
5, [VO(sal-D-Pen)(bpy)]·nH2O·mEtOH 6, [VO(sal-D,L-Pen)-
(bpy)]·nH2O 7, [VO(sal-CysSMe)(bpy)]·mEtOH 8, [VO(sal-
CysSEt)(bpy)]·mH2O
9
(CysSMe = S-methyl-L-cysteine and
CysSEt = S-ethyl-L-cysteine) were obtained by the general pro-
cedure described below. The preparations were carried out with
deoxygenated solutions and under N2.
L-cysteine methyl ester. Similar experiments were performed
with L-CysOMe at pH ca. 6, 8 and 10. L-Cysteine methyl ester
evolved H2S soon after mixing, and the optical activity of these
solutions was lost within ca. 2 h (with L-Ala this takes several days,
see ESI-3†). No quantitative determination of the evolved H2S was
performed.
General procedure. The amino acid was dissolved in an aqueous
solution of sodium acetate and a solution of salicylaldehyde in
ethanol added. The solution became yellow and shortly afterwards
the white thiazolidine precipitated, except in the cases of 8 and 9.
An aqueous solution of vanadyl sulfate was added, and most of
the white solid dissolved. In the case of 3 and 6–9 solid bpy was
added either before (3) or after (6–9) the addition of VOSO4. The
mixture was filtered within 5–10 min and the remaining white
solid separated. After 20–30 min, the complexes precipitated, and
were separated by filtration and washed. Table 1 summarizes the
analytical results, ESI-1† the monitoring of the preparations by TLC
and ESI-2† the details of the preparations.
S-methyl-L-cysteine (CysSMe). Mixtures containing L-CysSMe,
VIVO and Hsal at pH 8 evolve a gas with the characteristic odour of
a thiol. By analogy to the L-Cys system, CH3SH would be produced
from the N-salicylidene S-methyl-L-cysteinato complex, but we did
not try to identify the exact nature of the gas evolved.
D,L-homocysteine, lanthione, penicillamine, methionine,
cystine. No H2S or thiol was evolved from these systems. This
was normally checked by bubbling a stream of N2 through a small
column containing filter paper impregnated with a CdSO4 solution.
If H2S (or presumably other thiol compounds) was produced, the
filter paper would turn yellow, but this did not happen. The mixtures
Desulfydration mixtures
Cysteine. The solutions used to study the desulfydrations
contained equimolar amounts of L-Cys, VOSO4·5H2O and Hsal.
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6
2 8 5 7

View Article Online
were monitored by TLC using 20 × 20 cm plates divided in two
10 × 20 cm parts, each with an equal sequence of samples. One
part was revealed with iodine vapours followed by a ninhydrin–
collidine–copper solution, and the other part with a PdCl2/HCl
solution.
outlined below. For details see ESI-4†. The preparations were
normally carried out with deoxygenated solutions and under N2.
General procedure. VOSO4 (often 3.0 mmol) was dissolved
in an aqueous sodium acetate solution (ca. 10 mmol), and an
ethanolic solution of salicylaldehyde (ca. 3.0 mmol) was added.
Green VO(sal)2 precipitated and Na2CO3 was added to pH 8.
The appropriate nucleophile reagent was added (see Scheme 2),
normally in a 5–10 fold excess relative to VOSO4. Either solid L-
cysteine or an aqueous solution was added, normally ca. 3.0 mmole,
but ca. 12 mmole in the case of the synthesis of lanthionine 14.
The pH of the reaction mixtures was normally kept at ~8 by adding
Na2CO3, and all reactions were monitored by TLC (see ESI-4†)
using at least two different eluents. The TLC samples were applied:
(i) directly, (ii) after being acidified to pH ~ 2, (iii) with acidified
samples spiked with standard. The acidified samples were frozen
for HPLC experiments, using a pre-column derivatization method
and a fluorescence detector (see above section on HPLC methods,
and ESI-4†for details). In several cases, bpy (ca. 3–9 mmol) was
added to the reaction mixtures after the addition of the nucleophile,
but no significant differences between the two experiments (with or
without bpy) were noticed in the TLC and HPLC results. In some
cases the products (lanthionine 14, benzylcysteine 19) were isolated
as solids and characterised by the usual analytical methods.
Detection of pyruvic acid
The mixture containing L-Cys, Hsal and VIVO2+, from which all H2S
had been evolved, was acidified at pH ~ 2. The precipitates formed
(e.g. lanthionine, cystine, VO(sal)2, or other), were filtered off and
a solution of 2,4-dinitrophenylhydrazine in HCl (2 M) was added.
The 2,4-dinitrophenylhydrazone was extracted according to the
methods of Case.35 Asolid 2,4-DNPproduct was recrystallized from
nitromethane (m.p. 213–215 °C (literature value for the pyruvic
acid derivative, 214 °C). Analysis: found C, 39.9%, H, 3.3%, the
product requires 40.31%, H, 3.01%.
Detection of ammonia. The presence of ammonia was detected
by adding NaOH to the green acidified solutions until basic, and
then blowing the ammonia formed into Nessler’s reagent36 with a
stream of N2. A brown precipitate was formed in Nessler’s reagent
indicating the presence of ammonia.
Racemization experiments. Equimolar solutions of 0.005 M in
vanadium, of composition L-Ala or L-Met:Hsal:VIVOSO4, were
deoxygenated, the pH adjusted to 9, and the solution kept at 20 °C.
The electronic spectrum and circular dichroism were measured at
intervals (see ESI-3†).
EPR spectra
The X-band EPR spectra were recorded at 77 K (on glasses made
by freezing solutions with liquid nitrogen), either with a Bruker
ESR-ER 200D connected to a Bruker B-MN C5 ESR spectrometer
and to a Bruker ESR data system, or with a Bruker ESP 300 E
spectrometer.
Reactions based on the dehydroalanine intermediate V (see
Schemes 1 and 2)
S-Hydroxyethylcysteine12,(2-amino-2-carboxy-3-dimethyl)sulfide
13, lanthionine 14, S-methylcysteine 15, S-ethylcysteine 16, 2-
amino-3-(diethylamino)propanoic acid 17, S-phenylcysteine 18
and S-benzylcysteine 19, were synthesised by the general procedure
Magnetic susceptibility
Magnetic susceptibility of polycrystalline samples of complexes
(36 mg of 2, 31 mg of 3, 25 mg of 4 and 31 mg of 6) were measured
in the range 9–295 K using a 7-T Faraday Oxford Instruments
system coupled to a Sartorius S3D-V microbalance, at 1 T and
applying forward and reverse gradients of 2.5 T m−1. Under these
conditions the magnetisation was found to be proportional to the
applied magnetic field.
Magnetic moment of the desulfydration mixture
The magnetic susceptibility of the mixtures were monitored
according to the method of Evans.37 A solution of composition
Cys:Hsal:VIVO2+ :OH− = 1:1:1:3, 0.1 M in vanadium, was kept
at 40 °C and the H2S evolved was removed by a stream of N2.
Samples (2.5 ml) of this solution were taken, tert-butanol (0.2 ml)
added, and the solution made up to 5.0 ml with water. The external
standard was a 4% aqueous solution of tert-butanol. These samples
were taken at 0, 2, 3, 5 and 22 h from the time of mixing, and the
resonance lines of tert-butanol measured. A constant frequency
separation of 7.7 Hz was obtained, this giving a constant eff of ~1.7
B per V atom.
Thermogravimetry and differential scanning calorimetry
Thermogravimetric (TG) and differential scanning calorimetric
(DSC) curves were measured with a Setaram TG-DSC111
thermobalance, normally in the range 20–600 °C. Some TG curves
were recorded on a Stanton Redcroft TG-750, connected to a
Venture Servoscribe 220.
Circular dichroism and isotropic absorption spectra
CD spectra were run either on a Jasco 720 spectropolarimeter
(either with a 170–800 nm or with a red-sensitive (400–1000 nm)
photomultiplier), or on a Roussel-Jouan Dicrographe MK III.
UV/VIS isotropic absorption spectra (UV-VIS) were run either on a
Perkin Elmer 9, or on a Cary 17 spectrophotometer.
Scheme 2 Amino acid derivatives obtained by a Michael-type base-
catalysed addition, upon addition of nucleophiles to solutions containing
[VIVO(sal–L-Cys)(X)] (X = solvent or bpy).
2 8 5 8
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6

View Article Online
Solid complexes 2–4, 6, 8, 9. The KBr disks or the Nujol mulls
for this purpose were prepared similarly to those for IR spectroscopy
but (i) the relative amounts of complex were ca. 50% higher, and
(ii) the optical path was as low and homogeneous throughout the
disk (or paste) as possible. The disk or the paste (one or two drops)
was placed between two microscope slides and placed in the sample
compartment in a fixed position. Depending on the complex and the
intensity of the CD signal obtained, one to three of such ‘samples’
were used (one for the KBr disks). A first CD spectrum was run,
the sample rotated ~70°–90° and another spectrum recorded; five
rotations were performed for each sample (14 rotations for 4, which
had an extremely weak signal) and the corresponding CD spectrum
recorded. With this type of solid sample, one does not know the
position of the base line, but the correct pattern of the spectrum
may be obtained if the spectrum recorded after each rotation of the
sample is always approximately the same.
by manipulation of the atomic co-ordinates of the available X-ray
structures and organic residues (e.g. –CH2–X,–CH3 or others) were
added when required. All model complexes were minimised using
the conjugate gradient algorithm and a high convergence criteria
with default parameters.
Density functional calculations
DFT calculations40 were carried out using the Amsterdam
Density Functional (ADF) program41 developed by Baerends
and co-workers (ADF-2002).42 Vosko, Wilk, and Nusair’s local
exchange correction potential was used,43 together with Becke’s
nonlocal exchange44 and Perdew’s correlation corrections.45,46
Unrestricted calculations were carried out on the paramagnetic
species. The geometry optimization procedure was based on the
method developed by Versluis and Ziegler,47,48 using the non-local
correction terms in the calculation of the gradients. Full optimisation
was carried out for all complexes. Full geometry optimisations
without any symmetry constraints of complexes [VIVO(sal-D-
Cys)(H2O)], [VIVO(sal-L-Pen)(H2O)], [VIVO(tz–Cys)(H2O)],
CD solution spectra. Before preparing the solutions, oxygen
was removed from the solvents by bubbling N2. The spectra were
recorded immediately after the preparation of the solutions: the cells
had their stoppers reinforced with parafilm® strips, but no special
care was taken to remove oxygen from them. The cells were placed
in thermostatted cell compartments.
[VIVO(tz–Pen)(H2O)],
[VIVO(sal-DHAla)(H2O)],
[VIVO(sal-
DHVal)(H2O)] were carried out. The starting structures were
based on structures of the corresponding isomers obtained from
molecular mechanic calculations. In the geometry optimisations,
core orbitals were frozen for V (1s, 2s, 2p), S (1s, 2s, 2p), and C,
N, O (1s), while triple  Slater-type orbitals (STO) were used for
the valence orbitals of H (1s), C, N, O (2s, 2p), and 3s, 3p, 4s, 4p,
and 3d of V. A set of polarisation functions was added for H (single
 p, d), V (single .p, f) and S, C, N, O (single , d, f). For the EPR
calculations, all electron basis sets, consisting of uncontracted
triple- STO functions, augmented by polarisation functions, were
used for all elements. The ZORA method49 was used to account for
relativistic effects and the spin orbit coupling. The A values were
obtained from an unrestricted calculation and the g values from a
spin restricted calculation with spin–orbit correction. The results of
the g and A calculations are included in ESI-9†. The agreement of
the A values with experiment is rather poor.
Complexes. The complexes were normally dissolved in methanol
in concentrations ca. 0.1–1 mM for experiments in the UV (230–
400 nm), or ca. 1–5 mM for experiments in the visible (400–800
or 400–1000 nm). The solutions were placed in a cell (1 cm optical
path) containing a small magnetic stirring bar. The spectra were run
immediately, and repeated after ca. 15, 25, 50 and 80 min of stirring
(total times after the preparation of the solution indicated). The cells
were opened and a few drops of water were added. After ca. 10 and
20 min of stirring, CD spectra were recorded. The cells were re-
opened, air gently bubbled into the solution for 10 min with stirring
and a final CD spectrum recorded. Variations of this sequence were
also done for the L-Cys system.
The geometries of [VIVO(sal–D-Cys)(H2O)] and [VIVO(tz–Cys)-
(H2O)] obtained from the ADF calculations were optimized again
using the Gaussian 98 package50 and the IR frequencies were calcu-
lated. TD-DFT51 calculations were performed for these two species
(10 states were requested). The unrestricted B3LYP52 formalism
was adopted. The standard LANL2DZ53 basis set, along with the
associated ECP, was used for V and S, while a standard 6–31G(d)
basis set54 was used for C, O, N, and H. Graphical representations
of molecular orbitals were drawn with MOLEKEL.55
Other experiments. CD and visible spectra were also recorded
in aqueous–ethanolic solutions (2:1) containing amino acid:s
alicylaldehyde:VIVO:acetate (1:1:1:2), starting at pH ~ 4–5,
recording the CD spectrum (400–800 nm or 400–1000 nm), and
taking a sample for EPR (sample immediately frozen at 77 K). The
pH was increased about 0.8 pH units, the CD spectrum recorded
and another sample taken for EPR. This was normally repeated up
to pH ~ 13.5. Such experiments were done with L-alanine, L-serine,
L-cysteine, D-penicillamine, L-cysteine ethyl ester (EPR only) and
S-methyl-L-cysteine (EPR only).
Results and discussion
CD spectra of solutions containing Hsal and L-Cys, or Hsal
and D-Pen (both reagents ~10−4 M) were also recorded. Here the
corresponding sal–Cys or sal–Pen thiazolidines (tz–Cys or tz–Pen)
presumably predominate.
Characterisation of complexes
The formation and precipitation of the thiazolidines, and the
decomposition of the amino acids activated by vanadium
partly explain the difficulty in isolating complexes from mix-
tures containing VIVO2+, Hsal and Cys or Pen. The presence
of amounts of water and/or ethanol may introduce further dif-
ficulties in establishing correct formulations. This was so with
[VO(sal–L-Ala)(bpy)] 1, corresponding to structure IV, which by
X-ray diffraction was found to crystallise as [VO(sal–L-Ala)(bpy)]·-
1/8H2O·1/2EtOH.7 Table 1 summarises the new complexes 2–9,
their C, H, N, S analyses and the formulations proposed. In some
cases, the analytical results fit better with dinuclear (L)VV–O–VV(L)
or Na[(L)VIV–O–VV(L)] (L = sal–aa) formulations. However, the
magnetic properties measured for 2, 3, 4 and 6 (see below) indi-
cate that they are monomeric, and the amount of Na+ found was
always <0.2%, so these dimeric formulations are ruled out. As
mentioned, upon mixing Hsal and Cys or Pen, the corresponding
thiazolidines precipitate rapidly as white powders. Presumably
these may also precipitate with variable amounts of solvent as
somewhat different analytical results were obtained in different
batches. In Table 1 compounds 10 and 11 correspond to two of the
products obtained.
Molecular mechanics calculations
Molecular mechanics calculations were carried out using the Uni-
versal Force Field (UFF),38 within the CERIUS2 software.39 This
force field is parameterized for full periodic, but the minimization
of a few X-ray structures revealed clearly that the default param-
eters found in the field were inappropriate to reproduce accurately
the co-ordination spheres of the complexes. Therefore a specific
set of parameters comprising the L–M bond lengths and L–M–L
angles was developed empirically following a experimental pro-
cedure identical to that described in ref. 16. Partial charges were
not included as they were difficult to calculate accurately and
only have marginal impact on relative strain energies in metal
complexes. The atom types O_R2, N_R3 and N_R2 have the same
UFF properties of O_R (angular oxygen) and N_R (triangular ni-
trogen) respectively, in order to allow the individual assignment
of the ideal distances and/or bending angles. The values of the
ideal and L–M–L angles are listed in ESI-10†, Tables 1 and 2. The
starting co-ordinates for the thiazolidine isomers were obtained
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6
2 8 5 9

View Article Online
The complexes may precipitate with the ligand either in the form
of a Schiff base [as in VII: VIVO(SB)(X)] or of a thiazolidine (as in
VIII: VIVO(tz)(X)). As the number of the C, H, N, S, O and V atoms
in VII and VIII are identical, they may be considered tautomers, and
it is not possible to distinguish them by the analytical results. This
was performed by spectroscopy and by theoretical calculations.
strong bands corresponding to (CN) and as(COO) centred
around 1610–1620 and 1640–1650 cm−1 (~1690 cm−1 for 8 and 9).
The symmetric carboxylate stretch, s(COO), probably corresponds
to the medium/strong bands in the range 1335–1345 cm−1. Medium
to strong bands in the range 1300–1315 cm−1 probably correspond
to (O–Ph). The (VO) band appears in the range 955–1000 cm−1.
Besides these general features, some others appear or become more
intense, e.g., in complexes containing bpy, bands at 3070–80 cm−1
due to the aromatic C–H stretch.
Overall the IR of compounds 2–7 agree with their formulation as
VIVO (Schiff base) complexes. However, IR spectra alone cannot
rule out the presence of the ligands as thiazolidines.
The calculated IR spectra for the two tautomers of 2 differ. They
were obtained from a B3LYP51 calculation (Gaussian 9850). The
geometry was fully optimised and the results agreed with those
obtained from theADF calculation. In the Schiff base form, the S–H
stretching is calculated at 2675 cm−1, a very strong VO stretching
at 1086 cm−1, and an even stronger CN vibration at 1686 cm−1.
Conversely, in the thiazolidine form, no such well defined and
strong VO stretching is observed in the expected energy range.
These results definitely favour the Schiff base formulation. The
agreement between calculated and experimental frequencies could
be improved if scale factors were applied, which are available
from Gaussian 98 for certain conditions and usually take values
about 0.96.
MM and DFT calculations. In the VIVO(tz)(X) complexes (see
VIII), there are several stereogenic centres: the -C (C2), the Namine
,
the C3 and the V atom, and the S/R convention will be used for
the ligand atoms (for amino acids and the corresponding SB, the
D,L convention is used in this work). MM calculations were done
in order to screen the possible isomers of the thiazolidine form of
complexes 2 and 4 and look for the most sterically favoured ones.
The results are compiled in ESI-11†. For the isomers of both [VIVO-
(tz–Cys)(X)] and [VIVO(tz–Pen)(X)], the gas-phase strain energies
follow the same trend, and the lowest energy isomers correspond to
S,S,S,A configurations (A for the V atom).56 These two most stable
isomers were taken as models for DFT calculations (ADF program;
see Experimental for details). These were carried out in order to find
the lowest energy isomer obtained in the reaction of the vanadium
precursor and the amino acids, namely, whether a Schiff base or a
thiazolidine derivative of cysteine or penicillamine is more likely
to be formed. The Schiff base form of complex 2 is more stable
by 38 kJ mol−1 than the thiazolidine, while the difference favours
the equivalent form of 4 by only 2 kJ mol−1. Interestingly, another
Schiff base type isomer, where the VO and the side chain are on
the same side (contrary to being on opposite sides in the preferred
isomer) is only destabilised by 2 kJ mol−1 in the cysteine case, but by
44 kJ mol−1 for penicillamine. This suggests that the two complexes
may exhibit different reactivity. Besides the energies, a deeper com-
parison between the Schiff base and the thiazolidine forms (depicted
in Fig. 1) was carried out for the cysteine derivatives, by calculating
the infrared (Gaussian 98) and the EPR spectra (ADF), as well as
the electronic excitation energies (Gaussian 98) for both forms, and
comparing them with the experimental data, as will be described in
the following sections. The geometry of the desulfydrated species
[VIVO(sal–DHAla)(H2O)] and [VIVO(sal–DHVal)(H2O)] were also
fully optimised.
Magnetic moments. The magnetic susceptibilities of 2, 3, 5,
and 6 were measured by the Faraday method at 1 T between 9 K
and room temperature. The paramagnetic molar susceptibilities
were obtained from the results after subtracting the diamagnetic
contributions estimated from tabulated Pascal’s constants. Fig. 2
shows the results for 2, and similar curves were obtained with 3,
4, and 6. At 20 °C the eff values are 1.73, 1.87 and 1.70, and 1.71
B for 2, 3, 5 and 6, respectively, and they decrease very slightly
upon cooling. If eff = 2.839 (PT + )1/2 is used and/or a TIP term
is included, the eff values may become slightly smaller. The P and
eff values, and their change with temperature are typical of VIVO
compounds with a spin 1/2 per formula unit, indicating that the
complexes are monomeric with no significant antiferromagnetic
interactions.
Fig. 2 Temperature dependence of the magnetic susceptibilities (, left
scale), and eff values (*, right scale) of compound 2 in the range 9–295 K.
The eff values were calculated by the formula 2.839 (PT)1/2
.
Fig. 1 Optimised structures of (A) the C-[VIVO(sal–D-Cys)(H2O)] Schiff
base and (B) A-[VIVO(tz–L-Cys)(H2O)] (R,R,R,C-configuration) thiazolidine
complexes.
Thermogravimetric analysis
Thermogravimetric (TG) and differential scanning calorimetric
(DSC) curves were obtained for 2 and 3. TG curves were also mea-
sured for 4 and 6. The TG/DSC curves show trends similar to those
measured for [VO(sal–GlyGly)(H2O)n].59 For e.g. 2 (Fig. ESI-5†)
the weight loss up to 200 °C is compatible with the loss of 1–1.2
molecules of H2O per vanadium. The final weight, if assigned to
V2O5, indicates a molecular weight of 315, which compares well
with the 308.2 for the formulation [VO(sal–L-Cys)(H2O)]. At
IR spectra. All complexes present a broad band due to water
centred at ~3450 cm−1, less pronounced in the [VIVO(sal–aa)(bpy)]
compounds. In the IR of 2 a clear band at 2550 cm−1, is due to
(S–H). No such band is seen in complexes 3–9. A medium/strong
band at 1540–1560 cm−1, always present, may originate from the
vibration of the (Ph–)C–C(N) bond57 and typifies complexes
derived from salicylaldehyde.7,11,13,16,58 All complexes present very
2 8 6 0
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6

View Article Online
Table 2 EPR parameters obtained from the experimental EPR spectra of frozen (77 K) methanolic solutions of complexes obtained by simulation of the
spectra, back-calculated A(Nimine) values for each complex and (*)2 values.a
Compound
g
A × 104/cm−1
Back-calculatedA(Nimine) × 104/cm−1
g⊥(A⊥ × 104 /cm−1)
(*)2
2: VO(sal–L-Cys)(H2O)
3: VO(sal–L-Cys)(bpy)
6: VO(sal–D-Pen)(bpy)
8: VO(sal–L-CysSMe)(bpy)
9: VO(sal–L-CysSEt)(bpy)
1.933
1.942
1.935
1.950
1.951
164.8
161.2
159.7
162.0
162.1
38.18
39.52
38.02
40.32
40.42
1.977 (62.8)
1.977 (59.2)
1.974 (58.2)
1.981 (58.5)
1.981 (58.2)
0.845
0.855
0.844
0.877
0.881
a The square of the orbital coefficient, (*)2, may be obtained from the equation: (*)2 = 7/6 [(A − A⊥)/P − g + 5/14 g⊥)], with g = 2.0023 − g and
g⊥ = 2.0023 − g⊥, taking P = 0.0130 cm−1. The quantity (*)2 corresponds to the coefficient of the vanadium 3dxy orbital in the SOMO molecular orbital.
least three further weight losses start at ~190, ~230 and ~330 °C.
Comparing these with TG (and DSC) results obtained with
[VO(sal–GlyGly)(H2O)n],59 [VO(sal–L-Ala)(H2O)],5 and Na[V2O3-
(sal–D,L-Ser)2],5 the first two losses probably correspond to decar-
boxylation followed by oxidation of the remaining Cys moiety, and
the weight loss from ~330 °C probably involves the oxidation of
the benzene moiety. For 3, 4 and 6, the TG (and DSC for 3) show
similar trends.
ing from a reduced effective charge on the metal and some delocal-
ization of unpaired spin density onto the ligands.
The EPR parameters were also obtained for A–[VIVO(sal–L-Cys)-
(H2O)] and C–[VIVO(tz–Cys)(H2O)] (R,R,R,C isomer) complexes
from DFT calculations (see Experimental). These are also included
in Table 2. The calculated g values for A-[VIVO(sal–L-Cys)(H2O)]
give better agreement with the experimental g⊥ and g values than
those of C-[VIVO(tz–Cys)(H2O)]. Constraints in the calculations,
such as the functionals used,62 spin restrictions, accounting for
spin–orbit effects, may be responsible for the deviations between
calculated and experimental A values.[16] The order of magnitude
of the Az, Ay, Ax are reasonable, but the relative error is high.
EPR spectra
The EPR spectra may help to elucidate which groups co-ordinate
in equatorial position in solution. Table 2 summarises results from
frozen (77 K) methanolic solutions. In the spectra of 3, 6, 7 and 8 a
second minor species could be detected corresponding to ca. 5–8%
(3 and 6) and 3–5% of the main ones. Apparently these correspond
to EPR parameters close to those of 2 so the minor components
correspond to the [VO(sal–L-aa)(X)] (X = solvent) complexes.
Table 2 also includes the unpaired ground state (dxy) orbital
population (*)2 estimated from the EPR spectra in methanol.
The (*)2 values are similar to those obtained previously from
[VIVO(sal–L-Ala)(X)]7,16 complexes (X = H2O, EtOH or bpy),
and indicate that the unpaired electron is mostly localised in the d
orbital. The d orbital according to the ADF calculations is shown
in Fig. 3 for the A,L isomer of complex 2. It is essentially localized
in the vanadium dxy orbital (85.3%), with small * contributions
CD spectra. UV range. Circular dichroism generally gives
more useful structural information on vanadyl complexes than
do isotropic absorption spectra.7,11,13 In solutions containing Hsal
and L-Cys (both 10−4 M), where the cysteine thiazolidine (tz–Cys)
presumably predominates, two intense positive CD bands are clear
at ~220 and 255 nm, with a weaker band at 290 nm.Asimilar pattern
but with opposite signs is found for solutions containing Hsal and
D-Pen (see Fig. 3 in ESI-6†). As the absolute configuration of the
penicillamine used is D-, to compare these CD spectra with those
for corresponding solutions involving L-amino acids, the  (or )
values must be multiplied by −1. Unless otherwise specified, this
factor will be implicit throughout the discussion below.
In general the VIVO complexes derived from Hsal possess a low-
energy absorption band around 373 nm (CD) or 375 (absorption),
which can be attributed to a  → * transition originating mainly
in the azomethine chromophore. For L-amino acids, these bands
display Cotton effects of negative sign in the CD spectra, as found
for the related zinc,58 copper63–65 and cobalt(II)66 chelates. This CD
band was found at 376 nm in 2, 380 nm in 3 and 385 nm in 6. For 2
the absorption band in H2O/MeOH solutions was found at ~375 nm,
and at ~370 nm in a MeOH solution of [VIVO(sal–L-Ala)(H2O)].
TD-DFT calculations using Gaussian 98 showed two absorption
bands in this region for the Schiff base complex 2 (isomer A-
[VIVO(sal–L-Cys)(H2O)]). The more intense is observed at 378 nm
and has two main components (40 and 45%); both consist of two
transitions from * (phenol) to a  V–N bonding, also spread over
benzene *, dxz orbital, and they can be considered as LMCT. A less
intense band in the visible region (at ~650 nm) can be assigned to
the classic dxy → dxz,dyz transition (band I); indeed, one component
(40%) is dxy → dxz, where the dxz orbital belongs to a  antibonding
V–Ooxo orbital, and dxy is almost non bonding, as described in the
EPR section (figure 2); the second contribution (48%) to this band
arises again from dxy and also ends in  V–N bonding, also spread
over benzene *, dxz orbital, giving some charge transfer character to
the band. The other calculated transitions, which might correspond
to other d–d transitions are extremely weak (oscillator strength
close to zero) and therefore were not considered (the pictures of
orbitals and band assignments are given in ESI-9†).
mainly from the oxygen atoms in the equatorial plane (1% Ocarboxylate
3.9% Ophenolate; 0.5% Ooxo), this being in agreement with the EPR
results.
;
Fig. 3 The singly occupied dxy orbital of complex 2, calculated with ADF.
This isomer, with A configuration on V and L on the cysteine -carbon atom,
is more stable than the A,D-isomer (enantiomer of C,L-) by ~2 kJ mol−1.
It is well known that the EPR spectra may help to elucidate which
groups co-ordinate in equatorial position in solution: for this, the
so-called additivity relation60 is often used. The A and g values
obtained are in the range expected for complexes where sal–L-aa
coordinates equatorially as a tridentate ligand through Ophenolate
,
The pattern and band intensities of the CD spectrum of solutions
of 2 and 4 in methanol (ca. 10−4 M) (Fig. 4), differ from those
of the thiazolidines, and are very similar to those found for
[VO(sal–L-Val)(H2O)] or [VO(sal–L-Ile)(H2O)].11 These intense
CD bands with max at ~255 and ~220 nm are probably associated
with benzene ring  → * and charge transfer transitions.66,67 The
imine bands are flanked by a prominent band/shoulder at 290 nm,
Nimine and Ocarboxylate, as found for other [VO(sal–L-aa)(H2O)] and
[VO(sal–L-aa)(bpy)]16 complexes (sal–aa = Schiff bases derived
from the reaction of Hsal and Gly, Ala, Val, Met, Phe, Ile, i.e.,
amino acids with no coordinating side chains). The presence of bpy
gives rise to a slight decrease of A and A⊥. These back-calculated
A (Nimine) values are among the lowest for aromatic imine donors,61
and reflect a reduced electron-nuclear hyperfine interaction result-
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6
2 8 6 1

View Article Online
but this apparently corresponds to relatively weak CD activity. The
CD spectra of 3 and 6 (Fig. 4), and those of 8 and 9 in methanol are
similar to those of other [VO(sal–L-aa)(bpy)] complexes11 (aa = Ala,
Val, Met, Ile, Phe). Overall these CD spectral results also indicate
that in the present methanolic solutions of the complexes containing
Cys and Pen, the ligands are present as Schiff bases.
Fig. 6 Circular dichroism spectra of methanolic solutions of several VIVO
complexes prepared in this work, and those of [VIVO(sal–L-Ala)(H2O)] and
[VIVO(sal–L-Ala)(bpy)] 1. Fig. 4 in ESI-6†shows a comparison of the solid
state and solution CD spectra of 2.
of the solid state spectrum and that of the methanolic solution is the
same, but on standing in air a positive band at ca. 440 nm appeared,
indicating significant oxidation of V(IV). The band at ca. 440 nm is
possibly due to a ligand-to-metal charge transfer transition, but as
the band is superimposed on the negative imine band, its max cannot
be determined accurately.
Fig. 4 CD spectra (UV range) of methanolic solutions of (A) 2 and 4, (B)
3 and 6. The CD spectra of 3 and 6 were multiplied by −1.
CD spectra. Visible range. In the visible isotropic absorption
spectra of complexes 2–9, band I (dxy → dxz,dyz) normally appears
broad, between approximately 650 and 900 nm (at least), and band
II (dxy → dx2 − y2) with maxima or shoulder at ~550 nm. The CD
spectra shows bands I, II and the imine band more distinctly.
Fig. 5 includes CD spectra of solids 2, 3, 6, 8 and 9. The pattern of
the spectra is the same in all cases and similar to that of [VO(sal–L-
Ala)(H2O)].7 Therefore, the structural factors that dominate the CD
signals are similar, thus indicating that for all complexes in the solid
state, namely those containing Cys and Pen, the ligand is the Schiff
base and not the thiazolidine. Although the band pattern of 3 and 6
are similar, the max for bands II, IB and IA differ significantly: 560,
690, 900 nm, and 560, 750, ~1000 nm for 3 and 6, respectively.
Desulfydration of cysteine. In a stirred mixture of equimolar
amounts of a VIVO2+ salt, L-Cys and an ethanolic solution of
salicylaldehyde in buffered aqueous medium, ready evolution of
H2S occurred at room temperature in the pH range 5–9. This system
seems to provide an analogy with the action of cysteine desulfydrase,
a pyridoxal dependent enzyme. Bergel et al.23,24 studied similar
reactions in the presence of pyridoxal instead of Hsal.
Fig. 7 shows the evolution of the Cys concentration with time
as obtained by TLC. At pH 3, no H2S was detected and almost no
cysteine decomposes. At pH 5.3, the desulfydration occurs. At pH 6
and at 40 °C, within 24 h the amount of H2S formed corresponded
to 50–60% of the L-Cys initially present. On increasing the pH,
the rate of the H2S evolution is accelerated, but the total amount
released within the first 24 h is approximately the same in the pH
range 6–9. However, at pH ≥ 8 the desulfydration proceeds very
fast and after ca. 20 h no Cys spot is detected by TLC. The solutions
after desulfydration show no optical activity. From the acidified
reaction mixtures pyruvic acid was identified through its 2,4-
dinitrophenylhydrazone; the presence of ammonia was confirmed
by the use of Nessler’s reagent.
H2S is quite soluble in aqueous media, and at pH < 8 (though not
at pH ≥ 8) small amounts of cystine and of lanthionine (which are
only slightly soluble in this medium) were detected by TLC. This
may explain why the H2S measured after flushing with N2 or argon
is much less than 100% of the initial L-Cys. At pH ≥ 8 several non-
identified spots were detected by TLC using a PdCl2/HCl solution,
Fig. 5 Circular dichroism spectra of several complexes dispersed in KBr
disks (8, 9) or Nujol mulls (2, 3 and 4). 2 (−−−); 3 (); 4 (); 8 (ooo);
9 (++) The spectrum of 6 was extremely weak, indicating that the amino
acid partly racemized.
Fig. 6 shows spectra of methanolic solutions of the same
complexes, and of [VO(sal–L–Ala)(X)] (X = H2O, bpy) within
ca. 5–15 min. after their dissolution. The pattern of the CD spectra
in the solid state and in methanol is the same, so the coordination
geometry does not change upon dissolution. It is also clear that
the pattern is similar in all cases, so in these solutions all ligands
coordinate similarly to sal–L-Ala.
Fig. 4 (in ESI-6† ) includes the CD spectrum of 2, and of its meth-
anolic solution, immediately after its preparation and after several
hours (ca. 2.2 and 3.4 h) standing in air. These spectra confirm that
L-Cys did not racemize significantly within this period. The pattern
Fig. 7 Approximate change in the concentration of free cysteine, as
estimated by TLC, in aqueous/ethanolic solutions containing L-Cys, Hsal
and VOSO4 (relative molar concentrations, 1:1:1). As cysteine may be
in the form of other compounds (e.g., the Schiff base, which partially
hydrolyses during sample application and elution), this only gives indirect
information about the cysteine that reacted to produce H2S.
2 8 6 2
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6

View Article Online
Table 3 Spin-Hamiltonian parameters calculated from the EPR spectra of frozen (77 K) aqueous:ethanolic (2:1) solutions containing amino
acid:Hsal:VIVO2+ :acetate (1:1:1:2),a and plausible equatorial donor atoms (these based on the EPR parameters and UV-Vis and CD spectra recorded). In
the 3rd column, the estimated A values60 are also included in bracketsb
Species
g
A × 104 /cm−1
Predicted equatorial donor atoms
A
B
1.949
1.953
1.951
1.954
1.955
1.957
~1.959
1.959
~1.969
~1.955
1.967
170 (171)
166 (166)
168 (166)
164 (162)
164 (165)
157 (159)
~155 (152)
156 (152)
~154 (152)
~157 (155)
145 (148)
~155 (152)
160 (162)
Ophenolate, Oaldehyde, Oacetate, H2O
Ophenolate, Nimine, Ocarboxylate, X (X = H2O ?)
Ophenolate, Oaldehyde, Namine, Ocarboxylate (?)
Ophenolate, Nimine, 2 × Oacetate (?)
Ophenolate, Ocarboxylate, H2O, OH− ?
Ophenolate, Nimine, O−alcoholate, X (X = H2O ?)
Ophenolate, Nimine, S−, Oacetate (?)
Ophenolate, Nimine, S−, Oacetate
B1 (Pen)
B2 (CysOEt)
C
D (Ser)
F (CysOEt)
E
E1
E2
G (Pen)
H (Pen)
T
Ophenolate, Nimine, S−, Oacetate (?)
Ophenolate, Nimine, S−, H2O (?)
Ophenolate, Namine, S−, OH− ?
~1.966
1.957
Ophenolate, Namine, S−, Oacetate (?)
3 × OH−, H2O
a The spin-Hamiltonian parameters of the species T were obtained by simulation of experimental spectra at pH ~ 13.5. All others were calculated following
the method described by Chasteen,60 by an iterative procedure using the corrected equations.3 b To estimate the A values, the following equatorial contribu-
tions were considered: Ophenolate (38.88), Nimine (39), Namine (40.08), H2O (45.65), Oaldehyde (44.47), Oacetate = Ocarboxylate (42.1), S− (31.92), OH− (38.68), O−
alcoholate
(35.32).
indicating that other sulfur-containing compounds formed. While
at pH ~ 5 cystine precipitated from the mixture, identified by TLC,
IR and by elemental analysis, for pH > 8 no cystine was detected by
TLC. In basic medium, cystine reacts with sulfide to cysteine;68 this
could explain its non-detection.
Cystine did not desulfydrate in the presence of V(IV) salts. With
V(V) salts instead of V(IV), or in solutions kept under oxygen the
desulfydration of Cys is much slower. The V(V) salts catalyse the
oxidation of cysteine to cystine, Cys is removed form the solution
and the amount of H2S produced decreases. Depending on the pH
and on the amount of oxygen present, H2S may not be detected.
In solutions containing VOSO4, L-Cys and Hsal, in an inert
atmosphere, the magnetic moment of the solutions were measured
by Evans’ method as the desulfydration of Cys proceeded; the eff
remained constant at 1.7 B per V atom. This indicates that the
desulfydration involves no valence change at the V atom, i.e., it
remains as V(IV).
Solutions of complex 8 and those containing the L-cysteine
methyl ester (L-CysOMe) also produced H2S, but no desulfydration
was observed in the case of methionine, homocysteine, cystine (Cis)
or penicillamine. -eliminations have been extensively studied in
the amino acid/pyridoxal/metal ion systems. The existence of the
-H atom is essential for reaction, and in some systems its removal
has been found to be the rate-limiting step. The electronegativity
of the leaving group is another important factor.69,70 These and
other observations are compatible with the mechanism outlined
in Scheme 3, which involves the formation of V. The presence
Scheme 3 Desulfydration of cysteine and formation of the dehydroala-
of extra groups e.g. –CH2–, –CH2SR as in the case of methionine,
homocysteine, lanthionine or cystine, hinders the -elimination
(no good leaving group may be formed). However, why no
desulfydration occurs with penicillamine, neither in the presence
of salicylaldehyde nor pyridoxal,24 remains to be explained (see
below).
nine complex V, i.e. the complex designated as [VIVO(sal–DHAla)(X)]
(X = solvent).
signal, but the CD signal differed (see ESI†, Fig. 8). For pH >
13 the CD signal was ~0 and the EPR corresponded to that of
−
VIVO(OH)3 (species T).
In the L-CysSMe system, in the pH range 5.0–12 the EPR were
similar to those of the L-Ala system.
Spectroscopic studies of aqueous/ethanolic solutions contain-
ing amino acid, salicylaldehyde, VIVO and acetate (1:1:1:2). In
the case of L-Ala, in the pH range 5.0–6.3, the EPR spectra (Fig. 5
in ESI-7†) showed two distinct signals A and B. The A(A) value
is compatible with VIVO(sal)(X), with X = Oacetate, while the A(B)
corresponds to VIVO(sal–L-Ala)(X) (see Table 3). The intensity ra-
tios B:A increased with pH. Fig. 6 (see ESI†) shows a typical CD
spectrum in this pH range (at pH 5.9).
For pH > 7.0 only signal B, which corresponds to the Schiff base
complex, was detected in the EPR. As the pH was increased, this
slightly shifted to lower A values, but the CD pattern did not change
much up to pH 10.5. In ESI†Fig. 7 shows a CD spectrum typical of
those recorded in the pH range 7–10. For pH > 8.3 the || values
decreased moderately, but in the range 10.5–11.6 the CD pattern
changed. This new species C, did not yield a clearly distinct EPR
In the pH range 5.0–10 the EPR and CD spectra of the L-Ser
system were similar to those of the L-Ala system (see Fig. 8 and
ESI-7†Figs. 6–8). For pH > 10 a new EPR signal D, with signifi-
cantly lower A values, became clear, and in the pH range 10.8–12.5
only this EPR signal could be detected. The EPR parameters of D
are consistent with the coordination of –CH2O− (Oalcoholate) in equato-
rial position. For pH > 12.5 the EPR signal shifted gradually until at
−
pH 13.6 it corresponded to that of VO(OH)3 .
In the case of L-Cys, in the pH range 5.0–8.5 the EPR and CD
spectra were similar showing the same pattern as those of the
L-Ala and L-Ser systems. The binding mode of the species pres-
ent is therefore the same. For pH > 8.5 the EPR showed a new
signal (E), corresponding to a much lower A value (Table 3). The
relative intensity of species E in the EPR progressively increased
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6
2 8 6 3

View Article Online
(1:1:1) at pH 4–6 differed totally from those of the other systems,
and for pH > 7 they resemble (in pattern and max of the bands) those
recorded for the binary system VIVO + D-Pen. Overall the spectro-
scopic results in aqueous/ethanolic solution indicate that the SB
complexes with binding mode (Ophenolate, Nimine, Ocarboxylate, X)equatorial
do not form significantly. Therefore, a possible explanation for the
distinct behaviour of D-Pen in these systems, namely, the non-desul-
fydration of this amino acid, may be that in these solutions D-Pen
is mostly present in the form of mixed-ligand complexes e.g. as
VIVO(sal)(D-Pen) species, the coordination of D-Pen changing with
pH, or with binding modes (Ophenolate, Nimine, S−, X)equatorial. It is possible
that VIVO(tz–Pen)(X) (X = Oacetate, H2O, OH−) complexes also form.
Fig. 8 EPR spectra at 77 K (high field region) of aqueous/ethanolic solu-
tions containing amino acid:salicylaldehyde:VIVO:acetate (1:1:1:2), at
the pH values indicated (aa = L-Ser, L-Cys or D-Pen) and CVO ~ 0.005 M
(see also Table 3).
until ca. pH 12, and the pattern of the CD signal also changed (see
ESI-7†Figs. 6–8). These observations are consistent with the for-
mation of the new SB complex involving equatorial coordination
of –CH2–S−. The spin-Hamiltonian parameters agree with a bind-
ing mode for this complex involving (Ophenolate, Nimine, S−, X)equatorial
(X = H2O or Oacetate). For pH > 12.5 the EPR signal of species T
Syntheses of the amino acid derivatives. The syntheses of
several amino acids by the procedure outlined in Scheme 1 was
described in the Experimental section. This is summarized in
Scheme 2, where the structural formulae of the amino acids ob-
tained are included. In most cases the nucleophile (Nu) is a thiol
compound, but it may be e.g. an amine group as in the synthesis of
17. The formation of amino acids 12–19 was confirmed by TLC and
HPLC. In some cases they were isolated as in the case of 14 and 19.
As judged by the TLC and HPLC results the presence of bpy had no
significant effect on the type of products nor yields.
As the ligand in the dehydroalanine complex V is not optically ac-
tive, no enantioselectivity is expected in the reactions in Schemes 2
and 3. Besides the vanadium atom is a stereogenic centre and both
A- and C-isomers56 are expected to form in similar concentrations.16
This was confirmed in the case of S-ethylcysteine (see Experimen-
tal section and ESI-7†). Further, the amino acids may racemize in
these systems; this was observed for L-Ala, L-Met, L-CysOMe in
this work, and for L-Asn.18
The success in obtaining amino acids 12–19 indicates that a
significant concentration of the vanadium complex VII with the
N-salicylidenecysteinato ligand is present in solution (a similar
reaction cannot occur with the thiazolidine complex VIII). The
reason why the desulfydration of cysteine proceeds rapidly, but
no H2S is detected in aqueous/ethanolic solutions containing Pen,
salicylaldehyde and oxovanadium(IV/V) is now discussed. Three
possible reasons are explored:
−
(corresponding to VO(OH)3 ) gradually increased, this being the
only signal detected at pH 13.5 (Fig. 8).
In the case of L-CysOEt (L-cysteine ethyl ester), in the pH range
5.0–6.4 the EPR showed two species: A + B2. The A values of
B2 are slightly lower than those of B, and its intensity increased
with pH (about 50% at pH = 6). In the pH range 7–8 only B2 was
detected. Species B2 possibly corresponds to a binding mode
(Ophenolate, Nimine, 2 × Oacetate)equatorial. For pH > 8 another species (F)
was detected, this being the only distinct species recorded in the pH
range 9–11.8. The A values of F are slightly lower than those of E
(Cys system), and are in agreement with a binding mode involving
(Ophenolate, Nimine, S−, X)equatorial (X = H2O or Oacetate; OH− corresponds
−
to a too low A). For pH > 12 VIVO(OH)3 began to be detected,
being ~60% at pH ~ 13.
In the case of D-Pen, the EPR and CD spectra recorded in the
pH range 4.0–13 differed from those of all systems previously
mentioned, namely those of the L-Cys system. In particular, the
EPR spectra were very complex (see Fig. 8 and ESI†, Fig. 5);
several species always co-existed, except at pH > 13.3 where only
VIVO(OH)3− was detected.
The only species with spin-Hamiltonian parameters similar to
those of B, was detected at ca. pH 5.8 ± 0.8 (see Fig. 8 and ESI†,
Fig. 5). This could correspond to a VIVO(sal–L-Pen)(X) complex,
(i): In the cysteine system the Schiff base complex forms while
in the penicillamine system the thiazolidine complexes are more
stable. (ii): The desulfydrated complex V formed with penicil-
lamine, [VIVO(sal–DHVal)(H2O)], is not stable (or its activation
energy too high). (iii): Some structural or other factors are present
in the penicillamine system in solution that hinders the desulfydra-
tion of this amino acid.
where the SB sal–L-Pen coordinates equatorially via Ophenolate, Nimine
,
Ocarboxylate. However the CD spectra differ totally from those with
similar SBs with L-Ala, L-Ser or L-Cys, indicating that the binding
mode differs.
A species with EPR parameters similar to those of E, detected in
the L-Cys system, is seen in the pH range 6–10, i.e. in the penicilla-
mine system the equatorial coordination of S− starts about 3 pH units
lower than in the cysteine system. In the binary system VIVO + D-
Pen the equatorial coordination of S− starts about 1.3 pH units lower
than in the VIVO + L-Cys system.71 Several other species were de-
tected in the EPR spectra (see Fig. 8), and the most relevant one,
designated by G, corresponds to (g = 1.967, A = 145 × 10−4 cm−1).
Calculations can shed some light on the thermodynamics of the
desulfydration processes, but it must be emphasized that our MM
and DFT calculations apply to the gas-phase structures, i.e., entropic
contributions, solvation, ion-pairing, intermolecular H-bonding and
electrostatic effects were not taken into account, as is often done for
coordination compounds.72,73 Therefore, the calculations do not give
definite answers and what follows should be read under this under-
standing.As concluded above, the Schiff base is more stable than the
thiazolidine for both the cysteine and the penicillamine derivatives,
but the energy is much closer for the latter one (only 2, rather than
38 kJ mol−1). However, the rather similar energies of the thiazolidine
and SB complexes rule out explanation (i). We can also calculate
the energy differences between the Schiff base complexes and the
sum of the energies of the desulfydrated species and SH2 (HCys and
HPen, for cysteine and the penicillamine, respectively):
This is compatible with a binding mode involving (Ophenolate, Nimine
,
S−, OH−)equatorial, which corresponds to Aest = 148.5 × 10−4 cm−1. This
could correspond to a complex such as IX.
In the D-Pen system, the EPR spectra showed a much greater num-
ber of species than expected for coordination of the sal/Pen ligand
either only as a Schiff base or only as a thiazolidine. Moreover, the
CD spectra recorded for solutions containing VIVO, D-Pen and Hsal
2 8 6 4
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6

View Article Online
HCys = [E(VIVO(sal–DHAla)(H2O) + H2S)
− E(Schiff base)] = 55 kJ mol−1 (1)
3 L. Casella, M. Gullotti, A. Pintar, S. Collonna and A. Manfredi, Inorg.
Chim. Acta, 1988, 144, 89–97.
4 K. Nakajima, M. Kojima, K. Toriumi and K. Saito, Bull. Chem. Soc.
Jpn., 1989, 62, 760–767.
5 J. Costa Pessoa, J. A. L. Silva, A. L. Vieira, L. Vilas-Boas, P. O’Brien
and P. Thornton, J. Chem. Soc., Dalton Trans., 1992, 1745–1749.
6 V. Vergopoulos, W. Priebsch, M. Fritzsche and D. Rehder, Inorg. Chem.,
1993, 32, 1844–1849.
7 I. Cavaco, J. Costa Pessoa, D. Costa, M. T. Duarte, R. D. Gillard and
P. M. Matias, J. Chem. Soc., Dalton Trans., 1994, 149–157.
8 S. Dutta, S. Mondal and A. Chakravorty, Polyhedron, 1995, 14,
1163–1168.
9 S. Mondal, S. Dutta and A. Chakravorty, J. Chem. Soc., Dalton Trans.,
1995, 1115–1120.
10 R. Fulwood, H. Schmidt and D. Rehder, J. Chem. Soc., Chem. Commun.,
1995, 1443–1444.
11 I. Cavaco, J. Costa Pessoa, M. T. Duarte, R. T. Henriques,
P. M. Matias and R. D. Gillard, J. Chem. Soc., Dalton Trans., 1996,
1989–1996.
12 S. Mondal, P. Ghosh and A. Chakravorty, Inorg. Chem., 1997, 36,
59–63.
13 J. Costa Pessoa, I. Cavaco, I. Correia, M. T. Duarte, R. D. Gillard,
R. T. Henriques, F. Higes, C. Madeira and I. Tomaz, Inorg. Chim. Acta,
1999, 293, 1–11.
14 K. K. Rajak, S. P. Rath, S. Mondal and A. Chakravorty, Inorg. Chem.,
1999, 38, 3283–3289.
15 K. K. Rajak, S. P. Rath, S. Mondal and A. Chakravorty, J. Chem. Soc.,
Dalton Trans., 1999, 2537–2540.
16 J. Costa Pessoa, M. J. Calhorda, I. Cavaco, I. Correia, M. T. Duarte,
V. Felix, R. T. Henriques and M. F. M. Piedade, J. Chem. Soc., Dalton
Trans., 2002, 4407–4415.
HPen = [E(VIVO(sal–DHVal)(H2O) + H2S)
− E(Schiff base)] = 6 kJ mol−1 (2)
Assuming that the gas phase HCys and HPen are good
approximations of the solution ones, explanation (ii) may also be
ruled out, the reaction is not driven by thermodynamic factors, as
the process is endothermic, being even less favourable for cysteine,
contrary to what is apparently observed.
As discussed above, in aqueous–ethanolic solutions, while the
CD and EPR of the aa/sal/VO systems (aa = Ala, Ser, Cys, pH ca.
5–9, 1:1:1 ratio) have similar patterns, the CD and EPR spectra for
the D-penicillamine/sal/VIVO system (1:1:1 ratio) show important
differences: while with Pen for pH ≥ 5–6 the sulfur atom coordinates
equatorially, with cysteine this occurs only for pH ≥ 8.8. So, the
speciation differs in the two systems and complexes are formed with
different binding modes. Therefore, in the penicillamine system, the
much earlier coordination of the thiolate indicates a greater stability
of the isomers with equatorially coordinated S−. If in the fraction of
complexes where the Pen containing Schiff base is the ligand, this
coordinates through its S− in equatorial position (e.g. structure IX),
its desulfydration cannot be activated. This is certainly a relevant
factor to explain the non-desulfydration of this amino acid, but
as discussed above, as explanations (i) and (ii) cannot be fully
accounted for, they cannot be ruled out.
17 C. Grüning, H. Schmidt and D. Rehder, Inorg. Chem., Chem. Commun.,
1999, 2, 57–59.
18 I. Cavaco, J. Costa Pessoa, M. T. Duarte, P. M. Matias and R. D. Gillard,
J. Chem. Soc., Chem. Commun., 1996, 1365–1366.
Conclusions
Oxovanadium(IV) complexes with ligands derived from the reaction
of salicylaldehyde with L-cysteine and with D- and D,L-penicillamine
have been prepared and characterised. The spectroscopic studies
in methanolic solution and in the solid state indicate that in the
complexes prepared the ligands coordinate as Schiff bases. The
solution structures of the complexes depend on pH and solvent,
and while with L-Cys the spectroscopic results show trends similar
to those of the L-Ala and L-Ser systems up to ca. pH 8–9, where
thiolate coordination starts, the penicillamine system behaves quite
distinctly, namely, thiolate coordination occurs for pH > 6.
In the presence of salicylaldehyde and VIVO, the desulfydration
of cysteine proceeds rapidly, but no similar reaction occurs with
penicillamine. In the cysteine system, the N-salicylidenedehy-
droalanine–VIVO complex V is believed to form in an intermediate
stage of the desulfydration, and addition of several nucleophiles to
the cysteine reaction mixtures produced amino acid derivatives by
a Michael-type base-catalysed addition, a result compatible with
the formation of V.
The desulfydration may only be activated by the expected -
elimination path, modelling the action of cysteine desulfydrase,
if the sulfur-containing amino acid is present as a Schiff base
complex. However, DFT calculations for both types of tautomers
(Schiff base and thiazolidine complexes) give somewhat similar
energies, so no clear energetic basis for this distinct reactivity
was found. Therefore, the non-desulfydration of penicillamine is
probably the result of distinct speciation and/or binding modes
in the cysteine and penicillamine systems in aqueous/ethanolic
solutions, particularly the much greater importance of binding
modes involving its S− in equatorial position.
19 D. A. Phipps, J. Mol. Catal., 1979, 5, 81–107.
20 R. D. Gillard and R. Wootton, J. Chem. Soc. B, 1970, 364–371.
21 D. E. Metzler and E. E. Snell, J. Biol. Chem., 1952, 198, 353–361.
22 D. E. Metzler, M. Ikawa and E. E. Snell, J. Am. Chem. Soc., 1954, 76,
648–652.
23 F. Bergel, R. C. Bray and K. R. Harrap, Nature, 2003, 181, 1654–1655.
24 F. Bergel, K. R. Harrap and A. M. Scott, J. Chem. Soc. B, 1962,
1101–1112.
25 Y. Murakami, H. Kondo and A. E. Martell, J. Am. Chem. Soc., 1973, 95,
7138–7145.
26 H. Nakano, M. Nishioka, O. Sangen and Y. Yamamoto, J. Polym. Sci.,
Polym. Chem. Ed., 1981, 19, 2919–2928.
27 H. Nakano, R. Yamane, O. Sangen and Y. Yamamoto, J. Polym. Sci.,
Polym. Chem. Ed., 1982, 20, 2335–2339.
28 J. Costa Pessoa, M. T. Duarte, R. D. Gillard, C. Madeira, P. M. Matias
and I. Tomaz, J. Chem. Soc., Dalton Trans., 1998, 4015–4020.
29 A. L. Fluharty, The Chemistry of the Thiol Group, J. Wiley & Sons,
London, 1974, Chapter 13, pp. 589–668.
30 Y. N. Belokon, A. S. Sagyan, S. M. Djamgaryan, V. I. Bakhmutov and
V. M. Belikov, Tetrahedron, 1988, 44, 5507–5514.
31 E. D. Moffat and R. I. Lytle, Anal. Chem., 1959, 1979, 926–928.
32 J. D. H. Cooper and D. C. Turnell, J. Chromatogr., 1982, 227,
158–161.
33 P. Lindroth and K. Mopper, Anal. Chem., 1979, 51, 1667–1674.
34 R. H. Buck and K. Krummen, J. Chromatogr., 1987, 387, 255–265.
35 E. M. Case, Biochem. J., 1932, 26, 753.
36 A. I. Vogel, J. Basset, R. C. Denney, G. H. Jeffery and J. Mendham,
Vogel’s Textbook of Quantitative Inorganic Analysis, Longman, London,
1978, pp. 730–731.
37 D. F. Evans, J. Chem. Soc., 1959, 2003–2005.
38 A. K. Rappé, C. Casewit, K. S. Colwell, W. A. Goddard, III and
W. M. Skiff, J. Am. Chem. Soc., 1992, 114, 10024.
39 CERIUS 2, version 3.5, Molecular Simulations Inc., San Diego,
1997.
40 R. G. Parr and W. Yang, Density Functional Theory of Atoms and Mol-
ecules, Oxford University Press, New York, 1989.
Acknowledgements
41 (a) ADF (2000) E. J. Baerends, A. Bérces, C. Bo, P. M. Boerrigter,
L. Cavallo, L. Deng, R. M. Dickson, D. E. Ellis, L. Fan, T. H. Fischer,
C. Fonseca Guerra, S. J. A. van Gisbergen, J. A. Groeneveld, O. V.
Gritsenko, F. E. Harris, P. van den Hoek, H. Jacobsen, G. van Kessel,
F. Kootstra, E. van Lenthe, V. P. Osinga, P. H. T. Philipsen, D. Post,
C. C. Pye, W. Ravenek, P. Ros, P. R. T. Schipper, G. Schreckenbach,
J. G. Snijders, M. Sola, D. Swerhone, G. te Velde, P. Vernooijs,
L. Versluis, O. Visser, E. van Wezenbeek, G. Wiesenekker, S. K. Wolff
and T. K. Woo, T. Ziegler; (b) C. Fonseca Guerra, O. Visser, J. G.
Snijders, G. te Velde and E. J. Baerends, Parallelisation of the
Amsterdam Density Functional Programme in Methods and Techniques
We thank Fundo Europeu para o Desenvolvimento Regional and
Fundação para a Ciência e Tecnologia (project POCTI/35368/QUI/
2000) for financial support.
References
1 J. J. R. Fraústo da Silva, R. Wootton and R. D. Gillard, J. Chem. Soc. A,
1970, 3369–3372.
2 R. Hämäläinen and U. Turpeinen, Acta Crystallogr., Sect. C: Cryst.
Struct. Commun., 1985, 41, 1726–1728.
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6
2 8 6 5

View Article Online
for Computational Chemistry, ed. E. Clementi and C. Corongiu, 1995,
STEF, Cagliari, pp. 303–395; (c) C. Fonseca Guerra, J. G. Snijders,
G. te Velde and E. J. Baerends, Theor. Chem. Acc., 1998, 99, 391.
42 (a) E. J. Baerends, D. Ellis and P. Ros, Chem. Phys., 1973, 2, 41;
(b) E. J. Baerends and P. Ros, Int. J. Quantum Chem., 1978, S12, 169;
(c) P. M. Boerrigter, G. te Velde and E. J. Baerends, Int. J. Quantum
Chem., 1988, 33, 87; (d) G. te Velde and E. J. Baerends, J. Comput.
Phys., 1992, 99, 84.
pp. 1–27; (b) P. J. Hay and W. R. Wadt, J. Chem. Phys., 1985, 82,
270–283; (c) P. J. Hay and W. R. Wadt, J. Chem. Phys., 1985, 82,
299–310.
54 (a) R. Ditchfield, W. J. Hehre and J. A. Pople, J. Chem. Phys., 1971,
54, 724–728; (b) W. J. Hehre, J. A. Ditchfield and J. A. Pople, J. Chem.
Phys., 1972, 56, 2257–2261; (c) P. C. Hariharan and J. A. Pople, Theor.
Chim. Acta, 1973, 28, 213–222; (d) P. C. Hariharan and J. A. Pople,
Mol. Phys., 1974, 27, 209–214; (e) M. S. Gordon, Chem. Phys. Lett.,
1980, 76, 163–168.
55 H. P. Lüthi and S. Portmamn, Chimia, 2000, 54, 766.
56 G. J. Leigh, Nomenclature of Inorganic Compounds. Recommendations
1990, Blackwell, Oxford, 1990, pp. 183–189.
43 S. H. Vosko, L. Wilk and M. Nusair, Can. J. Phys., 1980, 58, 1200.
44 A. D. Becke, J. Chem. Phys., 1987, 88, 1053.
45 J. P. Perdew, Phys. Rev. B, 1986, 33, 8822.
46 J. P. Perdew, Phys. Rev. B, 1986, 34, 7406.
47 L. Versluis and T. Ziegler, J. Chem. Phys., 1988, 88, 322.
48 L. Fan and T. Ziegler, J. Chem. Phys., 1991, 95, 7401.
49 E. van Lenthe, A. Ehlers and E. J. Baerends, J. Chem. Phys., 1999, 110,
8943.
57 J. W. Ledbetter, Jr., J. Phys. Chem., 1977, 81, 54–59.
58 L. Casella and M. Gullotti, J. Am. Chem. Soc., 1981, 103, 6338–6347.
59 I. Cavaco, J. Costa Pessoa, S. M. Luz, M. T. Duarte, P. M. Matias,
R. T. Henriques and R. D. Gillard, Polyhedron, 1995, 14, 429–439.
60 N. D. Chasteen, Biological Magnetic Resonance, Plenum, New York,
1981, ch. 2, pp. 53–119.
61 T. S. Smith, C. A. Root, J. W. Kampf, P. G. Rasmussen and
V. L. Pecoraro, J. Am. Chem. Soc., 2000, 122, 767–775.
62 F. Neese, J. Inorg. Biochem., 2001, 86, 357.
63 L. Casella, M. Gullotti and G. Pacchioni, J. Am. Chem. Soc., 1982, 104,
2386.
64 L. Casella, M. Gullotti, A. Pasini and A. Rockenbauer, Inorg. Chem.,
1979, 18, 2825.
50 Gaussian 98, Revision A.11.3, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski,
J. A. Montgomery, Jr., R. E. Stratmann, J. C. Burant, S. Dapprich,
J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas,
J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli,
C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui,
K. Morokuma, N. Rega, P. Salvador, J. J. Dannenberg, D. K. Malick,
A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski,
J. V. Ortiz, A. G. Baboul, B. B. Stefanov, G. Liu, A. Liashenko, P. Pis-
korz, I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith,
M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe,
P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L. Andres, C. Gon-
zalez, M. Head-Gordon, E. S. Replogle, and J. A. Pople, Gaussian, Inc.,
Pittsburgh PA, 2002.
65 M. R. Wagner and F. A. Walker, Inorg. Chem., 1983, 22, 3021.
66 L. Casella and M. Gullotti, Inorg. Chem., 1986, 25, 1294.
67 C. J. Ballhausen and H. B. Gray, Inorg. Chem., 1962, 1, 111.
68 D. K. Liu and S. G. Chang, Can. J. Chem., 1987, 65, 770–774 and
references therein.
51 (a) R. E. Stratmann, G. E. Scuseria and M. J. Frisch, J. Chem. Phys.,
1998, 109, 8218; (b) R. Bauernschmitt and R. Ahlrichs, Chem. Phys.
Lett., 1996, 256, 454; (c) M. E. Casida, C. Jamorski, K. C. Casida and
D. R. Salahub, J. Chem. Phys., 1998, 108, 4439.
52 (a) C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785–789;
(b) A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
69 J. H. Thomas, K. S. Dodgson and N. Tudball, Biochem. J., 1968, 110,
687–692.
70 H. Reiber, Biochem. Biophys. Acta, 1976, 444, 734–755.
71 J. Costa Pessoa, L. F. Vilas Boas and R. D. Gillard, Polyhedron, 1990, 9,
2101–2125.
72 P. Comba, Coord. Chem. Rev., 1999, 182, 343–371.
73 P. Comba and T. W. Hambley, Molecular Modelling of Inorganic
Compounds, Wiley-VCH, Weinheim, Germany, 1995.
53 (a) T. H. Dunning, Jr. and P. J. Hay, in Modern Theoretical Chemistry,
vol. 3, ed. H. F. Shaefer, III, Plenum Press, New York, 1977, vol. 3,
2 8 6 6
D a l t o n T r a n s . , 2 0 0 4 , 2 8 5 5 – 2 8 6 6