Solution properties, electrochemical behavior and protein interactions of water soluble triosmium carbonyl clusters

Article abstract of DOI:10.1016/j.jorganchem.2004.02.036

The synthesis and solution chemistry of the water soluble clusters [Os₃(CO)₉(μ-η²-Bz)(μ-H) L⁺] (HBz=quinoxaline, L⁺=[P(OCH₂ CH₂NMe₃)₃I₃], 1) along with its negatively charged analog [Os₃(CO)₉ (μ-η²-Bz)(μ-H)L^-] (L^-= [P(C₆H₄SO₃)₃ Na₃], 2) are reported. In addition, we have examined the reduction potentials of the complexes [Os₃(CO)₉ (μ-η²-Bz)(μ-H)L] (HBz=phenanthridine, L=L⁺ (3); HBz=5,6 benzoquinoline, L=L⁺ (4); HBz=3-amino quinoline, L=L⁺ (5); HBz=3-amino quinoline, L=L^- (6). The neutral analog of 1 and 2 [Os₃(CO)₉(μ-η²-Bz)(μ-H) PPh₃] (Bz=quinoxaline, 7) was also examined for comparison. Both compounds 1 and 2 show pH dependent NMR spectra that are interpreted in terms of the extent of protonation of the uncoordinated quinoxaline nitrogen which impacts the degree of aggregation of the clusters in aqueous solution. Compound 1 undergoes a reversible 1e^- reduction in water while 2 undergoes a quasi-reversible 1e^- reduction at more negative potentials as expected from the difference in charge on the phosphine ligand. Compound 7 undergoes a marginally reversible CV in methylene chloride at a potential intermediate between the positively and negatively charged clusters. The overall stability of the radical anions of 1, 2 and 7 is somewhat less than the corresponding decacarbonyl [Os₃(CO)₁₀ (μ-η²-Bz)(μ-H)] (HBz=quinoxaline). While complexes 1 and 2 show reversible 1e^- reductions, all the other complexes examined show 1e^- and/or two 1e^- irreversible reductions in aqueous and non-aqueous solvents. The potentials for these complexes follow expected trends relating to the charge on the phosphine and the pH of the aqueous solutions. The ligand dependent trends are compared with those of the previously reported corresponding decacarbonyls. The interactions of the positively and negatively charged clusters with albumin have been investigated using the transverse and longitudinal relaxation times of the hydride resonances as probes of binding to the protein. Evidence of binding is observed for both the positive and negative clusters but the positive and negative clusters exhibit distinctly different rotational correlation times. Two additional complexes [Os₃(CO)₉ (μ-η²-Bz)(μ-H)L] (HBz=2-methylbenzimidazole, L=L⁺ (8); L=L^- (10) and HBz=quinoline-4-carboxaldehyde, L=L⁺ (9); L=L^- (11)) are reported in connection with these studies.

Full text of DOI:10.1016/j.jorganchem.2004.02.036

Journal of Organometallic Chemistry 689 (2004) 1796–1805
www.elsevier.com/locate/jorganchem
Solution properties, electrochemical behavior and protein interactions
of water soluble triosmium carbonyl clusters
a
a,
a
a
Carlo Nervi , Roberto Gobetto *, Luciano Milone , Alessandra Viale ,
b,
b
b
c
Edward Rosenberg *, Fabrizio Spada , Dalia Rokhsana , Jan Fiedler
a
Dipartimento di Chimica IFM, Universitꢀadi Torino Via P. Giuria 7, 10125 Torino, Italy
Department of Chemistry, University of Montana, Missoula, MT 59812, Bldg. 14, USA
b
c
J. Heyrovskꢁy Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Dolejꢂskova 3, 18223 Prague, Czech Republic
Received 23 January 2004; accepted 27 February 2004
Abstract
The synthesis and solution chemistry of the water soluble clusters [Os3(CO)9(l-g2-Bz)(l-H)Lþ] (HBz ¼ quinoxaline,
Lþ ¼ [P(OCH2CH2NMe3)3I3], 1) along with its negatively charged analog [Os3(CO)9(l-g2-Bz)(l-H)Lꢀ] (Lꢀ ¼ [P(C6H4SO3)3Na3],
2) are reported. In addition, we have examined the reduction potentials of the complexes [Os3(CO)9(l-g2-Bz)(l-H)L]
(HBz ¼ phenanthridine, L ¼ Lþ (3); HBz ¼ 5,6 benzoquinoline, L ¼ Lþ (4); HBz ¼ 3-amino quinoline, L ¼ Lþ (5); HBz ¼ 3-amino
quinoline, L ¼ Lꢀ (6). The neutral analog of 1 and 2 [Os3(CO)9(l-g2-Bz)(l-H) PPh3] (Bz ¼ quinoxaline, 7) was also examined for
comparison. Both compounds 1 and 2 show pH dependent NMR spectra that are interpreted in terms of the extent of protonation
of the uncoordinated quinoxaline nitrogen which impacts the degree of aggregation of the clusters in aqueous solution. Compound 1
undergoes a reversible 1eꢀ reduction in water while 2 undergoes a quasi-reversible 1eꢀ reduction at more negative potentials as
expected from the difference in charge on the phosphine ligand. Compound 7 undergoes a marginally reversible CV in methylene
chloride at a potential intermediate between the positively and negatively charged clusters. The overall stability of the radical anions
of 1, 2 and 7 is somewhat less than the corresponding decacarbonyl [Os3(CO)10(l-g2-Bz)(l-H)] (HBz ¼ quinoxaline). While com-
plexes 1 and 2 show reversible 1eꢀ reductions, all the other complexes examined show 1eꢀ and/or two 1eꢀ irreversible reductions in
aqueous and non-aqueous solvents. The potentials for these complexes follow expected trends relating to the charge on the
phosphine and the pH of the aqueous solutions. The ligand dependent trends are compared with those of the previously reported
corresponding decacarbonyls. The interactions of the positively and negatively charged clusters with albumin have been investigated
using the transverse and longitudinal relaxation times of the hydride resonances as probes of binding to the protein. Evidence of
binding is observed for both the positive and negative clusters but the positive and negative clusters exhibit distinctly different
rotational correlation times. Two additional complexes [Os3(CO)9(l-g2-Bz)(l-H)L] (HBz ¼ 2-methylbenzimidazole, L ¼ Lþ (8);
L ¼ Lꢀ (10) and HBz ¼ quinoline-4-carboxaldehyde, L ¼ Lþ (9); L ¼ Lꢀ (11)) are reported in connection with these studies.
Ó 2004 Elsevier B.V. All rights reserved.
Keywords: Osmium; Clusters; Electrochemistry; Water soluble; Albumin
1. Introduction
electron microscopy. Principal handicaps in the use of
organometallic carbonyl metal clusters as biological
probes are their instability and low solubility in aqueous
solutions at physiological pH.
We have developed synthetic procedures for a class
of electron deficient complexes of general formula
Metal complexes have been widely used to elucidate
the structure and the function of biological systems [1,2].
In particular, polymetallic complexes can play a singu-
larly active role due to their direct visualization using
[Os3(CO)9(l3-g2-Bz)(l-H)]
ligand) [3]. Recently, we reported the electrochemical
properties of these triosmium clusters where the electron
deficiency, according to the effective atomic number
(HBz ¼ benzoheterocyclic
* Corresponding authors. Tel.: +1-406-243-2592; fax: +1-406-243-
4227.
E-mail address: rosen@selway.umt.edu (R. Gobetto).
0022-328X/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2004.02.036

C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
1797
rule, arises from the presence of a three center two
electron bond b to the coordinated pyridinyl nitrogen
(Fig. 1) [4]. All of the compounds proved to be elec-
trochemically active but in most of the complexes elec-
trochemically reversible but chemically irreversible
reductions were observed. The phenanthridine, 5,6-
benzoquinoline electron deficient complexes (Fig. 1) and
one electron precise decacarbonyl clusters, [Os3(CO)10-
(l-g2-Bz)(l-H)] (HBz ¼ quinoxaline) (Eq. (1)) show
electrochemically and chemically reversible 1eꢀ reduc-
tions. The corresponding stable monoanion radicals
have been studied in detail both from the experimental
as well as from the theoretical point of view [5,6]. While
the LUMO of organometallic clusters is in general lo-
calized on the metal core [7], in the case of the clusters
mentioned the ligand plays a significant role in the sta-
bility of the radical anions. In fact, the LUMO of
[Os3(CO)10(l-g2-Bz)(l-H)] (HBz ¼ quinoxaline) is com-
pletely centered on the organic ligand.
[1–3]. Recently, we have shown that these clusters,
modified with positively charged phosphines, show
binding affinities to plasmid DNA that are under-
standable in terms of the intercalating ability of the li-
gand bound to the cluster [8]. There are several
motivations for developing biomacromolecular labels
using polymetallic species containing bioactive ligands
besides direct visualization by electron microscopy and
the ease with which X-ray phase information can be
obtained from diffraction experiments. In addition, it
has been pointed out that electrochemically active
clusters could combine these features with a sensitive
method of detection and/or electron transfer induced
cleavage of the biomacromolecules [8,9]. Furthermore,
in the case of the clusters reported herein the hydride
ligand, which resonates at )10 to )15 ppm, is in a region
devoid of all other protons, and represents a potentially
useful probe for monitoring cluster binding to biomac-
romolecules. Therefore, we thought it would be useful to
investigate the properties of these water soluble trios-
mium clusters. We report here the electrochemical be-
havior of the water soluble clusters and their binding
interaction with albumin as probed by the measurement
of the spin–lattice and spin–spin relaxation times of the
hydride ligand.
N
N
N
N
-
e
H
H
ð1Þ
Os
Os
Os
Os
-
- e
Os
Os
Most recently, we have become interested in the
aqueous chemistry of the complexes illustrated in Fig. 1,
modified with water solubilizing ligands. We placed li-
gands bearing positive Lþ ¼ [P(OCH2CH2NMe3)3I3] or
negative Lꢀ ¼ [P(C6H4SO3)Na3] charges on the metal
core to provide water solubility. The aromatic ligands in
this family of clusters are known intercalators as well as
agonists and antagonists for neurotransmitter receptors
2. Experimental
2.1. General
The benzoheterocycle triosmium complexes of general
formula [Os3(CO)9(l-g2-Bz)(l-H)] (HBz ¼ quinoxaline,
phenanthridine, 5,6-benzoquinoline, 3-amino- quinoline)
N
N
N
N
N
Os
H
Os
Os
H
Os
H
Os
H
Os
Os
Os
Os
Os
Os
Os
Quinoxaline
Quinoline
Phenanthridine
5,6-Benzoquinoline
H
N
N
N
N
S
O
CH
3
R
CH
R
3
N
N
N
Os
Os
Os
H
Os
Os
H
Os
H
Os
Os
H
Os
Os
Os
Os
2-Me-Benzimidazole
2-R-Benzoxazole
2R-Benzothiazole
2-Me-Benzotriazole
R= H, CH
3
R=H, CH
3
Fig. 1. Structures of the family of electron deficient benzoheterocycle triosmium clusters.

1798
C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
were synthesized according to published procedures [3].
The benzoheterocycle triosmium complex [Os3(CO)9-
(l3-g2-Bz)(l-H)] (HBz ¼ quinoline-4-carboxaldehyde)
was prepared according to published literature proce-
dures [6]. The water soluble phosphine derivatives 3, 5, 6
and 9 and the positively charged phosphine ligand,
[P(OCH2CH2NMe3)I3], were synthesized according to
published literature procedures [8,9]. Osmium carbonyl
was purchased from Strem Chemicals and used as re-
ceived. Sodium triphenylphosphane sulfonate was pur-
chased from Aldrich and used as received.
thaw techniques. Temperature calibration was carried
1
out with a methanol standard H thermometer. Errors
in the reported T1 and T2 values were estimated to be in
the range of ꢁ10%. Samples for the experiments with
albumin were prepared by dissolving 1 mg (0.001 mmol)
of cluster in 0.5 mL of D2O (no buffer). Precipitation
does not occur at any albumin concentration. Clusters
used: 3-aminoquinoline negative (6); 2-me-benzimid-
azole negative (10); 2-me-benzimidazole positive (8);
quinoline-4-carboxaldehyde negative (11); quinoline-4-
carboxaldehyde positive (not completely water soluble)
(9).
2.2. Electrochemical measurements
2.4. Synthesis of the water soluble clusters [Os3-
(CO)9(l-g2-Bz)(l-H)(Lþ or Lꢀ)] (HBz ¼ quinoxaline,
Lþ (1); Lꢀ (2); 5,6-benzoquinoline, Lþ (4); 2-methyl-
benzimidazole, Lþ (8); Lꢀ (10) and quinoline-4-carbox-
aldehyde, Lꢀ (11) (Lþ ¼ [P(OCH2CH2 NMe3)3I3]
Lꢀ ¼ [P(C6H4SO3)3Na3])
An EG&G Princeton Applied Research Potention-
stat/Galvanostat Model M273, connected to a PC with
EG&Princeton M270 software, was used. The working
electrodes were a dropping mercury electrode (DME)
and a glossy carbon electrode (GCE); the reference
electrode was a self-made calomel saturated electrode,
the counter electrode was a platinum wire. The GCE
was polished with wet alumina before use. All solutions
used for electrochemical measurements solutions were
deoxygenated with an argon purge and were kept under
argon flux during the measurements. THF was freshly
distilled from benzophenone ketyl and CH2Cl2 from
phosphorus pentoxide. The pH-measurements were
performed with an Amel 334-B pH-meter. Aqueous
electrochemical measurements were performed using the
following buffer: 0.72 mL of 80% H3PO4, 0.57 mL of
glacial CH3COOH, 0.62 g of H3BO3 dissolved in water
to make 250 mL of stock solution. By adding controlled
amounts of 4 M NaOH, pH values between 1.8 and 12.0
can be obtained [10,11]. Electrochemical measurements
in organic solvents were done using 0.1 M tetrabuty-
lammonium hexafluorophosphate as the supporting
electrolyte. In aqueous solutions, measurements were
made against a SCE reference electrode. The non-
aqueous reduction potentials are reported with reference
to the ferrocene/ferrocenium ion redox couple in the
appropriate solvent.
In
a 25 mL round bottom flask 50 mg of
[Os3(CO)9(l3-g2-Bz)(l-H)] (0.05 mmol) was dissolved in
10 mL of methanol (or acetone), then an equimolar
amount of Lþ (38 mg) or Lꢀ (30 mg) dissolved in a few
drops of water was added to the cluster solution. The
reaction is quantitative and yields 80–88 mg (100%) of
yellow to orange product as a precipitate. Analytical
and spectroscopic data of the compounds are reported
below.
Compound 1: Anal. Calc. for [C32H45N5O12Os3PI3]:
C, 22.7; H, 2.70; N, 4.15. Found: C, 22.01; H, 2.79; N,
3.84%. IR m(CO) in D2O: 2094 (m), 2052 (s), 1998 (s),
1
1937 (sh) cmꢀ1. H NMR in D2O: d 9.41 (br, 1H), 8.40
(br, 1H), 8.33 (d, 1H), 7.51 (dd, 2H), 3.40 (m, 6H), 3.07
(s, 3H), 3.04 (s, 3H), 2.90 (s, 27H) and )13.15 (d, 1H,
JPH ¼ 16:47 Hz).
Compound 2: Anal. Calc. for [C35H18N2O18Os3
PS3Na3]: C, 27.65; H, 1.85; N, 1.19%. Found: C, 27.03;
H, 1.09; N, 1.84. IR ( CO) in D2O: 2089 (sh), 2046 (sh),
1
1965 (w), 1926 (s) cmꢀ1. H NMR in D2O: d 9.47 (s,
1H), 8.22 (s, 1H), 7.73 (d, 1H), 7.60 (m, 6H), 7.2 (s, 6H),
7.03 (d, 1H), 6.75 (s, 1H) and )12.26 (d, 1H, JPH ¼ 16:09
Hz).
2.3. NMR measurements
Compound 4: Anal. Calc. for [C37H48N4O12Os3PI3]:
C, 24.82; H, 2.78; N, 3.23. Found: C, 24.36; H, 2.03; N,
3.25%. IR m(CO) in D2O: 2090 (w), 2076 (sh), 2045 (m),
The NMR spectra were recorded on Varian Unity
Plus 400 MHz, JEOL EX 400 and Bruker Avance 600
MHz spectrometers. The NMR solvents (Aldrich) were
dried over activated molecular sieves (Type 4A).
Chemical shifts were referenced internally relative to the
residual protons in the deuterated solvents used. The 2D
NMR measurements were performed using 1024 data
points in t2 and 256 in t1 with a pulse repetition of 3 s.
1
2010 (sh), 1971 (s), 1938 (sh) cmꢀ1. H NMR in D2O: d
9.40 (br, 1H), 8.92 (br, 1H), 8.55 (s, 1H), 8.35 (br, 1H),
7.80 (br, 1H), 7.58 (br, 1H), 7.41 (br, 1H), 7.11 (br, 1H),
3.40 (m, 6H), 3.07 (s, 3H), 3.04 (s, 3H), 2.90 (s, 27H) and
)13.07 (d, 1H, JPH ¼ 16:7 Hz).
Compound 8: Anal. Calc. for [C32H46N5O12Os3PI3]:
C, 22.93; H, 2.74; N, 4.18. Found: C, 22.29; H, 3.60; N,
4.18. IR m(CO) in D2O: 2087 (w), 2041 (m), 1997 (s),
1
Non-selective inversion recovery was used to obtain H
T1 values. Samples for T1 measurements were prepared
in the absence of O2 by using standard freeze–pump–
1
1949 (sh) and 1921 (sh) cmꢀ1. H NMR in D2O: d 7.40
(t, 1H), 6.90 (d, 1H), 6.80 (d, 1H), 3.40 (m, 6H), 3.07 (s,

C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
1799
3H), 3.04 (s, 3H), 2.90 (s, 27H) and )13.00 (d, 1H,
JPH ¼ 16:88 Hz).
Compound 7: Anal. Calc. for [C32H21N2O9Os3P]: C,
34.57; H, 1.73; N, 2.30. Found: C, 34.38; H, 1.68; N,
2.29%. IR m(CO) in CH2Cl2: 2089 (s), 2048 (s), 2010(s),
1992 (m) 1962 (m) and 1929 (w) cmꢀ1 UV/Vis in CH2Cl2
(kmax): 395 and 530 nm. H NMR in CD2Cl2: d 9.21 (d,
1H), 8.45 (d, 1H), 7.85 (d, 1H), 7.31 (m, 4H), 7.20(m,
Compound 10 (in the hexahydrate form): Anal. Calc.
for [C35H19N2O18Os3PS3Na3]: C, 26.83; H, 1.99; N,
0.85. Found: C, 26.97; H, 1.99; N, 0.82. IR m(CO) in
D2O: 2087 (w), 2043 (m), 1998 (s), 1982 (sh) and 1954
(sh) cmꢀ1. 1H NMR in D2O: d 7.60 (t, 1H), 7.50 (d, 1H),
7.40 (m, 12H), 7.20 (d, 1H), 2.40 (s, 3H) and )12.00 (d,
1H, JPH ¼ 16:37 Hz).
1
12H), and )11.92 (d, 1H, JPH ¼ 15:63 Hz).
Compound 11: Anal. Calc. for [C37H19NO19-
Os3PS3Na3]: C, 25.84; H, 3.05; N, 3.27. Found: C, 25.76;
H, 2.85; N, 3.35%. IR m(CO) in D2O: 2088 (w), 2045 (m),
3. Results and discussion
3.1. Electrochemical behavior of [Os3(CO)9(l-g2-
Bz)(l-H)(P(OCH2CH2NMe3)3I3] (1, HBz ¼ quinoxa-
line) in aqueous media
1991 (s), 1955 (sh) and 1926 (sh) cmꢀ1 1H NMR in
.
D2O: d 9.90 (s, 1H), 9.45 (d, 1H), 8.93 (d, 1H), 7.80 (d,
1H), 7.36(m, 12H), 7.30 (d, 1H), 7.02 (t, 1H) and )12.00
(d, 1H, JPH ¼ 16:22 Hz).
The complex, [Os3(CO)9(l-g2-Bz)(l-H)(P(OCH2
CH2NMe3)3I3)] (1, HBz ¼ quinoxaline, Fig. 2) shows a
reversible 1eꢀ reduction. An aqueous solution of 1 (pH
4–5) shows a reversible half-wave potential at )0.62 V
vs. SCE at a scan rate of 500 mV/s. The free ligand,
quinoxaline, shows irreversible electrochemical behavior
on the cyclic voltammetry time scale in both aqueous
and non-aqueous media [4,12]. Slight variations in the
value of this half-wave potential and the observation of
a second partially overlapping wave at more negative
potential prompted us to make a more detailed investi-
gation of 1 in aqueous media.
2.5. Synthesis of [Os3(CO)9(l-g2-Bz)(l-H)L] (HBz ¼
quinoxaline, L ¼ PPh3) (7)
In a 25 mL round bottom flask 50 mg of [Os3(CO)9-
(l3-g2-Bz)(l-H)] (HBz ¼ quinoxaline, 0.05 mmol) was
dissolved in 10 mL methylene chloride then an equi-
molar amount of PPh3 (13 mg) was added to the cluster
solution. The reaction is quantitative and yields ꢂ60 mg
(100%) of red product after evaporation of the solvent
and crystallization from CH2Cl2–hexane.
N
N
N
N
N
N
L+
L+
L-
H
L+
H
H
H
Os
Os
Os
Os
Os
Os
Os
Os
Os
Os
Os
Os
4
1
2
3
N
NH2
NH2
N
N
N
L-
L+
H
L
H
H
Os
Os
Os
Os
Os
Os
Os
Os
Os
7
6
5
CHO
CHO
N
N
CH3
CH3
N
N
N
N
L+
H
L-
L-
L+
H
H
H
Os
Os
Os
Os
Os
Os
Os
Os
Os
Os
11
Os
Os
8
9
10
L+ = [P(OCH2CH2NMe3)3]+3
L = P(C6H5)3
L- = [P(C6H4SO3)3]-3
Fig. 2. Structures of the water soluble clusters.

1800
C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
Polarography of 1 in water/0.1 M KCl or phosphate
900.0
700.0
500.0
300.0
100.0
-100.0
buffer shows two overlapping cathodic waves. Half-
wave potentials of both waves are dependent on the pH
value, which indicates participation of protonation
equilibria in the reduction process. The wave at more
positive potential decreases in height on going from
acidic to alkaline solution but the overall height of the
overlapped waves remains constant up to pH 10. Con-
trolled potential coulometry performed either at the
potential of the limiting current plateau of the first wave
or at the potential behind the second wave gives ap-
proximately the same value of the charge consumption:
1.5 F/mol. Control of the pH can be achieved by means
of a H3PO4/CH3COOH/H3BO3 buffer solution. At pH
values lower than 7, CV performed on a glassy carbon
(GC) electrode shows that the first reduction at less
negative potentials is reversible on the CV time scale
(Fig. 3), at scan rates as low as 50 mV/s. The second
reduction process overlaps with the first, as the pH
values are raised by stepwise addition of a concentrated
solution of NaOH (Fig. 4). Also, the intensity of both
cathodic peaks is pH dependent, and the decrease in
current associated with the first reduction peak corre-
lates with an increase of current for the second reduction
peak. This can be easily visualized qualitatively by
square wave voltammetry (Fig. 5).
pH 6,1
pH 11,0
pH 8,7
-750.0
-850.0
-950.0
-1050.0
-1150.0
E (mV) vs SCE
Fig. 4. Cyclic voltammetry of buffered solutions of 1 at 200 mV/s, GC
electrode, at values of pH of 6.1, 8.7 and 11.0.
12.000
pH 2,0
pH 4.1
10.000
8.000
6.000
4.000
2.000
pH 6.25
I
( u A )
These results lead to the conclusion that the reduction
processes in both waves proceed with proton con-
sumption and lead to the same reduction products. This
interpretation is supported by polarographic monitoring
in non-buffered solutions during electrolysis and for-
mation of hydroxyl ions as detected by the appearance
of an anodic wave on the mercury electrode at ca. 0.08 V
(vs. SCE). Using a standard addition of NaOH it could
be determined that the consumption of one electron
results in the formation of one hydroxyl ion [11].
-500.0
-600.0
-700.0
E (mV)
-800.0
-900.0
-1000.0
Fig. 5. Square wave voltammetry (SWV) of buffered solutions of 1 at
30 Hz, at values of pH of 2.0, 4.1 and 6.25.
protonated complex and the second wave to the reduc-
tion of non-protonated species followed by protonation
of the reduced form. This is supported by the fact that
the aromatic proton resonances of 1 sharpen and show
increasing down field shifts as acid is added to a solution
of 1 in D2O. The sharpening is an indication of a de-
creasing tendency to aggregate as positive charge is
placed on the aromatic ring [8]. The hydride signal lo-
cated at )12.97 ppm is not pH dependent; no other
hydride resonance appears during the whole titration
ruling out the possibility of a protonation process at the
metal triangle. Lowering the pH from 7 to 1.4 results in
small downfield shifts for the aromatic resonances with
the signal at 9.7 ppm attributable to the proton adjacent
to the uncoordinated nitrogen undergoing the largest
downfield shift due to aromatic ring protonation at that
atom.
The more positive reduction wave, which is larger in
acidic solutions, can be ascribed to the reduction of the
pH 2,2
500.0
pH 3,0
300.0
pH 4,3
)
A
n
(
100.0
I
-100.0
-300.0
-450.0
-550.0
-650.0
-750.0
That 1.5 F/mol are consumed as observed by cou-
lometry could be the result of competitive follow-up
reactions of the reduced, protonated complex; hydrogen
evolution with regeneration of the oxidized form of the
E (mV) vs SCE
Fig. 3. Cyclic voltammetry of buffered solutions of 1 at 200 mV/s, GC
electrode, at values of pH of 2.2, 3.0 and 4.3.

C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
1801
-1
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
complex and a reaction leading to degradation of the
cluster structure.
The pH dependence of the polarographic half-wave
potential in the case of the 1eꢀ reduction process
(Ox + 1eꢀ ꢀ Red), with protonation of both oxidized
and reduced forms (Ox + Hþ ꢀ OxHþ and Red-
+ Hþ ꢀ RedHþ), should follow the equation (assum-
ing equal diffusion coefficients) [10,11]
E1/2
E1=2 ¼ E0 þ ðR=F Þ lnðKa=Ka0 Þ
þ ðRT=F Þ ln½ðKa0 þ ½HþꢃÞ=ðKa þ ½HþꢃÞꢃ;
ð10 Þ
0
2
4
6
8
10
12
14
pH
where Ka, Ka0 are the equilibrium dissociation con-
stants of the protonated oxidized and reduced forms,
respectively
Fig. 6. Dependence of the polarographic half-wave potential of the
reduction wave of (1) on pH; experimental values (j) and calculated
(Eq. (1)) curve for pKa ¼ 2, pKa0 ¼ 10.
Ka ¼ ð½Oxꢃ½HþꢃÞ=½OxHþꢃ; Ka0 ¼ ð½Redꢃ½HþꢃÞ=½RedHþꢃ:
Three limiting cases can be considered:
anodic peak couple. The reoxidation after the cathodic
scan going to the second reduction peak proceeds only
at the first, more positive wave due to a fast proton-
ation/slow deprotonation of the reduced cluster.
The polarographic limiting current behind the re-
duction wave(s) significantly decreases (by ca. 30%) be-
tween the pH values 11–12 and the solution color
changes from red to green. Obviously, this unknown
transformation of the complex stops the side reaction of
proton reduction, which increases the polarographic
current in more acidic solutions.
1. In very acid solutions, where ½Hþꢃ ꢄ Ka, Ka0 , the half-
wave potential is independent of pH and is shifted
versus standard potential according to relation
E1=2 ¼ E0 þ ðRT =F Þ lnðKa=Ka0 Þ.
2. In the pH range pKa < pH < pKa0 , i.e., Ka ꢄ ½Hþꢃ ꢄ
Ka0 , the equation (1) is simplified to
E1=2 ¼ E0 þ ðRT=F Þ lnðKa=Ka0 Þ þ ðRT=F Þ lnð½Hþꢃ=KaÞ;
ð2Þ
i.e., at 25 °C
E1=2 ¼ E0 þ 0:059 pKa0 ꢀ 0:059 pH
ð20 Þ
3.2. Electrochemical behavior of [Os3(CO)9(l-g2-Bz)-
(l-H)(P(OCH2CH2NMe3)3I3)](1, HBz ¼ quinoxaline)
in non-aqueous media
and a negative potential shift of 59 mV/pH is
expected.
3. In alkaline solutions, ½Hþꢃ ꢅ Ka, Ka0 , the reduction
proceeds at the standard potential, E1=2 ¼ E0.
In CH3CN/0.1 M tetrabutylammoniun hexafluoro-
phosphate 1 exhibits two polarographic reduction
waves, E1=2 ¼ ꢀ1:52 and )2.01 V (vs. FeCp2/FeCpþ2 ), of
approximately the same height. The limiting current
both behind the first and second wave grows irregularly
which indicates formation of decomposition products
from the reduced complex. Cyclic voltammetry shows
reversibility of the first reduction at higher scan rates (5
V/s). The second reduction appears to be irreversible
with only a small anodic counter peak. Coulometry
gives 1eꢀ/mol for the first wave and after the second 1eꢀ
reduction both waves disappear and several new waves
of decomposition products appear at more negative
potential (which can be reduced with further charge
consumption). Reduction directly at the potential be-
hind the second wave gives 2–2.5 F/mol of consumption
and a precipitate is formed in the solution. This rather
complex electrochemical behavior aside, it is important
to note that for complex 1 the initial 1eꢀ reduction gives
a more stable reduction product at less negative poten-
tials in aqueous media rather than in non-aqueous
media.
The experimentally observed potential shift is linear
in the measured range of pH 2–10 and slopes 61 mV/pH
are observed. Therefore, from the theoretical relations
given the dissociation constants Ka ꢂ 10ꢀ2 and
Ka0 ꢂ 10ꢀ10, for oxidized and reduced complex respec-
tively can be estimated (Fig. 6). The plotting of Fig. 6 as
an S curve instead of a straight line is based on the
calculated values of E1=2 at high pH using the above
analysis. Unfortunately, it was not possible to check this
because of precipitation in the solutions examined above
pH 10. In any case, the estimates of Ka and Ka0 are taken
from the linear region and should be not be significantly
impacted by this approximation.
The splitting of the polarographic wave into two
overlapping waves and the variation of their heights
with pH can be ascribed to an insufficiently mobile
equilibrium due to a low value for the rate constant of
protonation of the complex in its oxidized form [11].
The cyclic voltammetry measurements are in agree-
ment with the above-proposed mechanism. The reduc-
tion in the first wave appears as a reversible cathodic/

1802
C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
3.3. Electrochemical behavior of [Os3(CO)9(l-g2-Bz)-
(l-H)(Lþ or Lꢀ)] (HBz ¼ quinoxaline, Lꢀ (2); HBz ¼
phenanthridine, Lþ (3); HBz ¼ 5,6-benzoquinoline, Lþ,
4; HBz ¼ 3-NH2-quinoline, Lþ (5); HBz ¼ 3-NH2-quin-
oline, Lꢀ (6) and the neutral analog [Os3(CO)9(l-g2-Bz)-
(l-H)L](HBz ¼ quinoxaline, L ¼ [P(C6H5)3], 7)(Lþ ¼
[P(OCH2CH2NMe3)3I3] and Lꢀ ¼ [P(C6H4SO3)3-
Na3])
Table 2
Polarographic half-wave potential in aqueous media, in V vs. SCE
Compound
pH
E1=2
(1)
4.5a
8.7a
4.5a
8.7a
4.5a
8.7a
8.7a
8.7a
)0.62
)0.85
)0.71
)0.90
)1.09
)1.32
)1.45c
)1.20c
)1.65c
)0.76
)0.97
)0.99
)1.09
)1.51
)1.78
(2)
(3)
c
(4)
(5)
(6)
–
The complex 2 in dimethylformamide solution is re-
duced in a 1eꢀ reversible polarographic wave at )1.74 V
(vs. FeCp2/FeCpþ2 ). Reversibility was confirmed by cy-
clic voltammetry and the charge consumption by con-
trolled potential coulometry. In contrast, the complexes
3–5 in acetonitrile or dimethylformamide solutions do
not exhibit any anodic counter peak on cyclic voltam-
metry within the range of scan rates employed (0.05–40
V/s), although the corresponding 1eꢀ polarographic
waves appear as reversible (2) or semi-reversible (slightly
decreased slope due to slower electrode reaction; 3–5).
This indicates a fast chemical reaction following the
charge transfer. Instability of the primary reduction
products and a conversion to electro-inactive (non-oxi-
dizable) species is also observed in bulk electrolysis ex-
periments. Reduction of the complex (6) is irreversible
both in polarography and voltammetry.
The behavior of the complexes 2–6 in aqueous solu-
tion seems to be analogous in some respects to the be-
havior of 1. The polarographic reduction potentials of
2–6 in polar organic solvents and in aqueous solution
are given in Tables 1 and 2. Two overlapping polaro-
graphic waves are shifted to negative potential with
increased pH values, thus suggesting a consumption of
protons in the reduction process. A reversible electro-
chemical response is observed only with 2, at the more
positive wave, but only at very fast scan rates. In the
case of complexes 3–6 the reduction processes are irre-
versible and furthermore, the waves are ill defined, ei-
ther obscured by polarographic maxima or overlapping
with strongly growing current which possibly originates
from hydrogen evolution, catalyzed by the complexes or
by decomposition fragments from them. It should be
noted that in general the reduction potentials of all the
clusters is less negative in water than in non-polar sol-
)1.90c
b
–
)1.82c
a Phosphate buffer.
b 0.1 M tetrabutylammonium chloride (TBACl).
c Ill defined.
vents. This is most likely due to the higher dielectric
constant and corresponding lower resistivity of the
aqueous solutions employed which contained the high
salt concentrations required by the buffer system em-
ployed. As expected, the positively charged clusters 1
and 5 with the same heterocycle as the negatively
charged species 2 and 6 are reduced at less negative
potentials. More interestingly, the positively charged
clusters 3–5 are reduced at more negative potentials
than the negatively charged 2 indicating that it is the
nature of the heterocycle rather the overall charge on
the cluster that determines the reduction potential. This
is further corroborated by the electrochemical behavior
of the neutral analog, 7, which shows a reversible 1eꢀ
reduction at )1.84V vs. FeCp2/FeCpþ2 in CH2Cl2. As
expected the reduction product is less stable than the
corresponding decacarbonyl as evidenced by decreasing
current of the half-wave with decreasing scan rate. The
first reduction potential of 7 is slightly more negative
than that of 1 and 2 measured in more polar organic
solvents. The limited solubility of 7 in DMF prevented a
direct comparison with 1 and 2 but the similarity of the
potentials relative to the other complexes certainly
supports the idea that the nature of the heterocycle and
the substitution of phosphine for CO (i.e. the corre-
sponding decacarbonyl has a reversible wave at )1.16 V
in CH2Cl2 vs. FeCp2/FeCpþ2 ) are the dominant influ-
ences on the reduction potential in these complexes ra-
ther than the charge on the complex.
3.4. Interaction of the water soluble clusters with albumin
Table 1
Polarographic half-wave potential in non-aqueous media, in V vs.
FeCpþ2 /FeCp2
When a 0.001 M solution of the negatively charged
cluster [Os3(CO)9(l-g2-Bz)(l-H)Lꢀ] (HBz ¼ 2-methyl-
benzimidazole, Lꢀ ¼ [P(C6H4SO3)3Na3] (10)) is titrated
with successively increasing amounts of albumin the
cluster’s 1H NMR signals progressively broaden. The
resonances are detectable until the ratio of albumin/
cluster ¼ 0.04, at higher protein concentrations, the
signals become too broad to be observed (Fig. 7). Sim-
ilar behavior is observed for complex 6 and for
Compound
Solvent
E1=2
(1)
CH3CN
DMF
DMF
)1.52
)1.56
)1.74
)1.89
)1.81
)1.98
)2.24
)2.01
)2.01
(2)
(3)
(4)
(5)
(6)
CH3CN
CH3CN
CH3CN
DMF

C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
1803
Fig. 7. The 1H NMR spectrum of 10 in the presence of various
albumin to cluster ratios measured in water at 600 MHz.
Fig. 8. The relationship between the rotational correlation time and the
longitudinal (T1) and transverse (T2) relaxation times.
[Os3(CO)9(l-g2-Bz)(l-H) Lꢀ] (HBz ¼ quinoline-4-car-
boxaldehyde, Lꢀ ¼ [P(C6H4SO3)3Na3] (11)). The latter
complex was investigated because it was thought the
aldehyde group might covalently attach to the primary
amines of the protein and thus show different behavior
than the other clusters where only an electrostatic and or
p–p stacking interactions are possible. The longitudinal
ðT1Þ and transverse ðT2Þ relaxation times of the hydride
signal before and after the addition of albumin at two
different ratios have been measured. The results of these
measurements for the three clusters are summarized in
Table 3. In all three cases the T1 increases while the T2
decreases.
energy transfer is most efficient. Thus, increases in the
correlation time sc can result in either an increase or a
decrease in T1 depending on whether or not a decrease in
the molecular rotation increases the efficiency of energy
transfer to the lattice. In the case of the negative clusters
6, 10 and 11 adduct formation would slow molecular
tumbling leading to the conclusion that the correlation is
on the positive slope side of the T1 curve and that
therefore the correlation time of the hydride has signif-
icantly increased [13]. We do not know what the effect of
the phosphine and of the heterocycle on T1 is, but the
relatively large increase in sc is consistent with adduct
formation. As expected the increase in sc is always as-
sociated with a decrease in T2. In the particular case of
11 the albumin cluster solution was heated to 90 °C for 1
h to see if evidence of covalent bonding of the cluster to
the protein could be observed [14]. As can be seen from
Table 3 there was an increase in T1 but it is not large
enough to attribute irreversible covalent binding of the
cluster to the protein [15]. In a separate experiment it
was shown that at elevated temperatures 11 isomerizes
to a hydride with a different chemical shift and this most
likely accounts for the change in T1.
In order to be sure that the variations observed in
the relaxation rates are due to a real interaction with the
protein and not simply to the increased viscosity of the
solution after albumin addition, the measurements have
been repeated on 6 in the presence of different quantities
of polyethylene glycol (PEG, M.W. ¼ 1500), which
should enhance viscosity presumably without interact-
ing with the cluster. Significantly, no appreciable line
broadening of the hydride resonance is observed
throughout the concentration range examined and the
changes in T1 and T2 are relatively small compared with
These data can be interpreted in terms of the expected
relationship between rotational correlation and the spin
lattice relaxation time T1 (Fig. 8) [13]. According to this
relationship there is a minimum value for T1 that cor-
responds to a specific rate of molecular rotation where
Table 3
Transverse (T2) and longitudinal (T1) relaxation times for the nega-
tively charged clusters [Os3(CO)9(l-g2-Bz)(l-H)(P(C6H4SO3)3Na3)]a;b
Albumin/cluster
T1 (s)
T2 (ms)
(A) HBz ¼ 2-me-benzimidazole (10)
0
1.44/1.42 358/383
2.05/2.14 67/53
0.02
0.04
2.03
ffi34
(B) HBz ¼ 3-NH2-quinoline (6)
0
1.45/1.50 325/704
1.59/1.47 92/92
0.02
0.04
1.74
50
(C) HBz ¼ quinoline-4-carboxaldehyde (11)
0
0.02
1.52/1.52 641/487
1.75
1.93
200
120
0.02 after 90°
a The two values refer to the two components of the doublet.
b Values reported with an error of ꢁ10%.

1804
C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
Table 4
Transverse ðT2Þ and longitudinal ðT1Þ relaxation times for the posi-
those observed with albumin (Fig. 9). The vertical line in
Fig. 9 represents the concentration ratio where PEG/
cluster ¼ 0.04 the value at which maximum yet ob-
servable line broadening is seen with albumin. It seems
that up to this value (and even for higher concentration),
no effect on the relaxation is observed and this means
that the variations found in the albumin case are due
to the interaction between the cluster and the pro-
tein. No significant line broadening of the hydride res-
onance is observed with the positively charged clusters
[Os3(CO)9(l-g2–Bz)(l-H)Lþ] (HBz ¼ 2-me-benzimid-
azole, Lþ ¼ [P(OCH2CH2NMe3)3I3] (8); HBz ¼ quino-
line-4-carboxaldehyde, Lþ ¼ [P(OCH2CH2NMe3)3I3] (9))
at albumin/cluster ratios lower than 0.04, while in the
negative case broadening is extensive at ratios ¼ 0.02. At
ratios of 0.1 the negatively charged cluster hydride res-
onances in 6, 10 and 11 are broadened into the base line
while in the case of 8 and 9 the hydride resonance is still
relatively sharp (Figs. 7 and 10).
The T1 measurements were carried out on 8 and 9 but
at higher albumin concentrations. The results are sum-
marized in Table 4 and it can be seen that here the ad-
dition of albumin to the cluster solution results in
shorter T1 values although the T1 for the cluster itself is
not very different than that found for the negative
clusters. Considering that there are probably many
clusters bound to the albumin at any one time the ap-
parently weaker interaction with the protein would place
tively charged clusters [Os3(CO)9(l-g2-Bz)(l-H)(P(C6H4SO3)3Na3)]a;b
Albumin/cluster
T1 (s)
T2 (ms)
(A) HBz ¼ 2-me-benzimidazole (8)
0
1.90/2.07
0.77/0.68
0.57
310/324
53/48
ffi32
0.04
0.1
(B) HBz ¼ quinoline-4-carboxaldehyde (9)
0
0.02
1.93/2.02
1.40/1.58
312/216
147/98
a The two values refer to the two components of the doublet.
b Values reported with an error of ꢁ10%.
fewer clusters on the albumin at any one instant leading
to more rapid tumbling of the protein and placing this
cluster protein system on the negative sloping part of the
T1 curve (Fig. 8). The trend for T2 is the same found for
the negative cluster as expected. The same behavior was
observed for 9 as expected (Table 4).
4. Conclusions
The reversible electrochemistry observed for com-
pounds 1 and 2 is the first example of this type of be-
havior for organometallic cluster complexes in water
and holds out the possibility that electrochemical mon-
itoring of cluster binding interactions with biomacro-
molecules is a realizable goal. The overall trends in the
electrochemical data suggest that complexes that show
reversible electrochemical behavior in non-aqueous sol-
vents can be carried into water even when modified with
relatively highly charged phosphine ligands. Although
the electrochemistry for the remaining compounds is
irreversible and complex the overall trends in the redox
data reveal that the nature of the heterocycle rather than
the overall charge on the complex determines the re-
duction potential of the water soluble cluster species.
The significance of this result is that redox active cluster
containing heterocyclic complexes can be designed as
electrochemical probes with either positively charged
water solubilizing ligands for binding to DNA [8] or
with negatively charged ligands for binding to proteins
with basic amino acids on their surface.
1,8
1,6
1,4
1,2
T
1
T
2
1,0
0,8
0,6
0,4
0,2
0,0
0,5
1,0
1,5
2,0
PEG /cluster
Fig. 9. Plot of the variation in T1 and T2 of 6 with added polyethylene
glycol (PEG).
The significance of the reported changes in T1 of the
hydride resonance on binding to albumin demonstrates
that this NMR resonance which is completely separated
from those associated with biomacromolecules can be
used as a qualitative screen for cluster bio-macromole-
cule interactions and that the direction of this change is
understandable in terms of the expected protein – cluster
electrostatic interactions. Thus, that the negatively
charged clusters appear to bind more tightly than their
positive analogues is not surprising in light of the fact
that albumin is rich in positively charged amino acids
Fig. 10. The 1H NMR spectrum of 8 in the presence of various al-
bumin to cluster ratios measured in water at 600 MHz.

C. Nervi et al. / Journal of Organometallic Chemistry 689 (2004) 1796–1805
[2] C.J. Burrows, J.G. Muller, Chem. Rev. 98 (1998) 1109.
1805
[16]. The combination of reversible redox behavior, di-
rect visualization with electron microscopy and the
availability of a readily observable NMR probe (the
chemically inert hydride) promises to make the class of
clusters reported here unique tools for probing the
structure and behavior of a wide range of bio-macro-
molecules. What needs to be done now is to develop
electrochemically active clusters that have high binding
constants to specific amino acid or nucleic acid residues.
This will require developing cluster functional groups
and tethers that have specific and perhaps covalent in-
teractions with given nucleic acid or peptide sequences.
These studies are underway in our laboratories.
[3] M.J. Abedin, B. Bergman, R. Holmquist, R. Smith, E. Rosenberg,
J. Ciurash, K.I. Hardcastle, J. Roe, V. Vazquez, C. Roe, S.E.
Kabir, B. Roy, S. Alam, K.A. Azam, Coord. Chem. Rev. 190–192
(1999) 975.
[4] E. Rosenberg, M.J. Abedin, D. Rokhsana, D. Osella, L. Milone,
N. Nervi, J. Fiedler, Inorg. Chim. Acta 300–302 (2000) 769.
[5] N. Nervi, R. Gobetto, L. Milone, A. Viale, E. Rosenberg, D.
Rokhsana, J. Fiedler, Chem. Eur. J. 9 (2003) 5749.
[6] E. Rosenberg, D. Rokhsana, N. Nervi, R. Gobetto, L. Milone, A.
Viale, J. Fiedler, M.A. Botavina, Organometallics 23 (2004)
215.
[7] M.P. Brown, P.A.S. Dolby, M.M. Harding, A.J. Mathews, A.K.
Smith, D. Osella, M. Arbrun, R. Gobetto, J. Chem. Soc., Dalton
Trans. (1993) 827.
[8] E. Rosenberg, F. Spada, K. Sugden, B. Martin, L. Milone, R.
Gobetto, A. Viale, J. Fiedler, J. Organomet. Chem. 668 (1–2)
(2003) 51.
[9] E. Rosenberg, F. Spada, K. Sugden, B. Martin (to be published).
[10] J. Heyrovsky, J. Kuta, Principles of Polarography, Czech Acad.
Sci., Prague, 1965, p. 161.
Acknowledgements
We gratefully acknowledge the support of the DOE
for an EPSCOR Implementation Grant for support of
this research (E.R., F.S. and D.R.), the NSF EPSCOR
program for visiting scholar grants (C.N. and R.G.) and
the NIH BRIN program for faculty travel award (E.R.).
We also acknowledge the Italian MURST for a COFIN
2000 Grant.
[11] A.A. Lcek, in: A. Cotton (Ed.), Polarographic Behavior of
Coordination Compounds, Progress in Inorg. Chem., vol. 5,
Interscience, 1963, p. 274.
[12] H. Lund, M.M. Braizer, Organic Electrochemistry, third ed.,
Mercel Decker, New York, 1991, p. 732.
[13] R.K. Haris, Nuclear Magnetic Resonance Spectroscopy, Pitman,
London, 1983.
[14] J. March, Advance Organic Chemistry: Reaction Mechanisms and
Structure, Wiely, New York, 1985.
[15] S. Aime, W. Dastru’, R. Gobetto, A. Viale, in: M. Peruzzini, R.
Poli (Eds.), Recent Advances in Hydride Chemistry, Elsevier,
Amsterdam, 2001, p. 351.
References
[1] K.E. Erkkila, D.T. Odom, J.K. Barton, Chem. Rev. 99 (1999)
2777.
[16] T. Peters, All about Albumin: Biochemistry, Genetics and Medical
Applications, Academic Press, New York, 1995.