Synthesis, structures and hypoxia-selective cytotoxicity of cobalt(III) complexes containing tridentate amine and nitrogen mustard ligands

Article abstract of DOI:10.1039/a909447d

Cobalt(in) complexes containing the tridentate nitrogen mustard ligand Ar, Ar-bis(2-chloroethyl)diethylenetriamine (DCD) and the non-alkylating analogs AW-diethyldiethylenetriamine (DED) and diethylenetriamine (dien) have been synthesised and the mustard complexes evaluated as potential hypoxia-selective anticancer drugs. The complexes were prepared from the precursor /ra/w-K.2[Co(acac)(CO3)(NO2)2]'H2O in a charcoal-catalyzed substitution reaction. Reaction of this precursor with dien-3HCl in water resulted in the isolation of two isomeric products, mer- and s-fac[Co(dien)(acac)(NO2)]f. Reaction in methanol with DED-3HC1 gave one product, ma complex containing DCD coordinated in a bidentate fashion, fra;w-Co(r|2-DCD)(acac)(NO2)2, were isolated. The two DCD complexes were characterized by X-ray structure determinations, which confirm the tridentate coordination with a long cobalt-tertiary nitrogen bond distance in the former, and bidentate coordination with the mustard nitrogen as a pendant arm in the latter. The bidentate complex, which contains a free nitrogen mustard, had cytotoxicity similar to that of the free ligand, but cytotoxicity was successfully masked in the tridentate complex, which showed modest hypoxic selectivity (five-fold) in a clonogenic assay. The Royal Society of Chemistry 2000.

Full text of DOI:10.1039/a909447d

View Article Online / Journal Homepage / Table of Contents for this issue
Synthesis, structures and hypoxia-selective cytotoxicity of
cobalt(III) complexes containing tridentate amine and nitrogen
mustard ligands
David C. Ware,*a,b Penelope J. Brothers,a George R. Clark,a William A. Denny,b
Brian D. Palmerb and William R. Wilsonb
a Department of Chemistry, Faculty of Science, The University of Auckland, Private Bag 92019,
Auckland, New Zealand. E-mail: d.ware@auckland.ac.nz
b Auckland Cancer Society Research Centre, Faculty of Medicine and Health Sciences,
The University of Auckland, Private Bag 92019, Auckland, New Zealand
Received 30th November 1999, Accepted 31st January 2000
Cobalt() complexes containing the tridentate nitrogen mustard ligand N,N-bis(2-chloroethyl)diethylenetriamine
(DCD) and the non-alkylating analogs N,N-diethyldiethylenetriamine (DED) and diethylenetriamine (dien) have
been synthesised and the mustard complexes evaluated as potential hypoxia-selective anticancer drugs. The complexes
were prepared from the precursor trans-K2[Co(acac)(CO3)(NO2)2]ؒH2O in a charcoal-catalyzed substitution reaction.
Reaction of this precursor with dienؒ3HCl in water resulted in the isolation of two isomeric products, mer- and s-fac-
[Co(dien)(acac)(NO2)]ϩ. Reaction in methanol with DEDؒ3HCl gave one product, mer-[Co(DED)(acac)(NO2)]ClO4,
while from the reaction with DCDؒ3HCl both the tridentate mustard complex mer-[Co(DCD)(acac)(NO2)]ClO4
and, in low yield, a complex containing DCD coordinated in a bidentate fashion, trans-Co(η2-DCD)(acac)(NO2)2,
were isolated. The two DCD complexes were characterized by X-ray structure determinations, which confirm the
tridentate coordination with a long cobalt–tertiary nitrogen bond distance in the former, and bidentate coordination
with the mustard nitrogen as a pendant arm in the latter. The bidentate complex, which contains a free nitrogen
mustard, had cytotoxicity similar to that of the free ligand, but cytotoxicity was successfully masked in the tridentate
complex, which showed modest hypoxic selectivity (five-fold) in a clonogenic assay.
ate),6–8 [Co(DCE)(S2CNEt2)2]BPh4 9 and [Co(DCE)(trop)2]ClO4
Introduction
(trop = tropolonate).10 Coordination of the nitrogen lone
One class of bioreductive drugs, the hypoxia-selective cyto-
toxins, are designed to exploit the hypoxic sub-population of
cells that are present in many solid tumors, and which are
resistant to radio- and chemo-therapy.1 The basic approach is
the selective conversion of a non-toxic prodrug to an active
form.2–4 The prodrug undergoes reduction by the endogenous
reducing enzymes present in all cells, resulting in the formation
of a transient intermediate, usually a one-electron reduction
product. This can then undergo further chemical or metabolic
transformation to produce the cytotoxic species in its active
form. In order to be selective for hypoxic cells this reductive
activation must be inhibited by O2. In normal, well oxygenated
cells, this occurs through a redox-cycling mechanism in which
the one-electron reduced intermediate is reoxidised by molec-
ular oxygen. In hypoxic cells, in which oxygen is limited, net
reduction can take place and the reaction sequence leading to
the active cytotoxin can proceed. There is strong evidence that
redox cycling is the major mechanism by which O2 inhibits net
reduction of hypoxia selective prodrugs such as nitroaromatic
compounds in cells.5
One of our approaches to the design of such compounds is to
utilize net one-electron reduction at a transition metal center as
the activation step. Nitrogen mustards (which contain the bis(2-
chloroethylamine) moiety) are potent cytotoxins, utilizing the
lone pair on the amine nitrogen to initiate a reaction sequence
in which DNA is cross-linked by double alkylation. The nor-
mally cytotoxic nitrogen mustard can be deactivated by the
formation of a coordinate bond from the mustard nitrogen to a
transition metal. We have previously discussed the design and
evaluation of nitrogen mustard cobalt complexes such as
[Co(DCE)(Meacac)2]ClO4 1 (Meacac = 3-methylacetylaceton-
pair of the bidentate mustard ligand DCE (DCE = N,N-bis(2-
chloroethyl)ethylenediamine) to cobalt() substantially sup-
presses toxicity, since the electron pair is no longer available to
act as an intramolecular nucleophile. Low-spin, d6 octahedral
Co() complexes are kinetically very inert, whereas high spin,
d7 Co() complexes are much more labile.11 It was anticipated
that, provided the reduced Co() complex was sufficiently stable
to allow reoxidation12 in oxygenated cells to compete effectively
with ligand loss, such compounds would be hypoxia-selective.7
We showed recently that the bidentate mustard complex 1
indeed has substantial hypoxic selectivity in disperse cell
culture,7 and outstanding activity against hypoxic cells in the
interior of multicellular spheroids which have been widely used
as a model for solid tumors.8 However, selectivity does not
appear to be due to redox cycling (reoxidation of the initial one-
electron reduced Co() species by molecular oxygen). Pulse
radiolysis studies13 of 1 have allowed the kinetics of ligand
substitution to be investigated, and have showed that initial
reduction to the Co() species is followed by two other pro-
cesses. The first of these, which has a first-order rate constant of
120 sϪ1, is assigned to release of the cytotoxic DCE ligand but is
not inhibited by oxygen. This suggests that reversible reoxid-
ation of the initially-formed Co() species by oxygen is too slow
to compete with irreversible ligand loss. It was suggested13 that
the hypoxic selectivity of 1 may be due to competition between
the Co() species and oxygen for cellular reductants. The
cobalt complexes may represent an instance where kinetic com-
petition with O2 for reducing species is the basis for selectivity.
The lack of importance of redox cycling in the hypoxic select-
ivity of 1 would suggest that the lifetime of the initial Co()
reduction product is unlikely to be an important determinant
DOI: 10.1039/a909447d
J. Chem. Soc., Dalton Trans., 2000, 925–932
This journal is © The Royal Society of Chemistry 2000
925

View Article Online
of hypoxic selectivity. However, the apparently greater hypoxic
selectivity of the bidentate mustard complex 17 than a related
monodentate mustard14 or aziridine15 complex leaves open the
possibility that chelate effects might be important. In order to
examine this further, we have investigated incorporating the
mustard functionality into a potentially tridentate ligand.
We report here, the synthesis and evaluation of some Co()
complexes containing a new tridentate nitrogen mustard, N,N-
bis(2-chloroethyl)diethylenetriamine (DCD, 5). We have also
investigated cobalt() complexes containing the unsubstituted
ligand diethylenetriamine (dien) and the non-alkylating analog
N,N-diethyldiethylenetriamine (DED).
gradient of MeCN (80% v/v) in ammonium formate (0.45 M,
pH 4.5). Decomposition of compounds in culture media was
determined similarly using a Waters µ-Bondapak C18 column
and a diode array detector.
Preparations
N,N-Diethyldiethylenetriamine trihydrochloride, DEDؒ3HCl.
DED (3.0 g) was dissolved in MeOH (20 mL) and acidified with
a solution of dry HCl in MeOH and then Et2O, added slowly
until a precipitate persisted. The solution was cooled and the
hygroscopic, microcrystalline product was filtered off, washed
with MeOH–Et2O (1:2 v/v) then Et2O, and dried over silica gel
(4.52 g, 89%). 1H NMR (D2O): δ 3.63 (m, 2 H, CH2CH2NH2),
3
3.57 (m, 2 H, CH2CH2NH2), 3.54 (t, 2 H, J = 7.4 Hz, CH2-
CH2NR2), 3.44 (t, 2 H, 3J = 7.4 Hz, CH2CH2NR2), 3.33 (q, 4 H,
3
3J = 7.3 Hz, CH2CH3), 1.33 (t, 6 H, J = 7.3 Hz, CH2CH3).
13C{1H} NMR (D2O): δ 49.86 (2 C, CH2CH3), 48.26 (CH2NR2),
46.41 (CH2CH2NR2), 43.51 (CH2CH2NH2), 37.04 (CH2NH2),
9.86 (2 C, CH2CH3) (Found: C, 35.50; H, 9.12; N, 15.36; Cl,
39.75. Calc. for C8H24N3Cl3: C, 35.77; H, 9.00; N, 15.64; Cl,
39.59%).
N-{[NЈ,NЈ-Bis(2-hydroxyethyl)]-2-aminoethyl}-NЉ-trityl-
glycyl carboxamide 2. 1,1Ј-Carbonyldiimidazole (26.0 g, 0.16
mol) was added to a solution of N-tritylglycine (40.0 g, 0.13
mol) in dry DMF (200 mL). After CO2 evolution ceased, the
solution was warmed to 40 ЊC for 10 min, and a solution of
N,N-bis(2-hydroxyethyl)ethylenediamine (19.3 g, 0.13 mol) in
DMF (50 mL) was added in one portion. After 30 min, the
solvent was evaporated under reduced pressure, and the residue
was partitioned between ethyl acetate and saturated aqueous
NaHCO3. The organic layer was washed well with brine, and
worked up to give an oil which was chromatographed on
silica gel. Ethyl acetate eluted unidentified material, while ethyl
acetate–MeOH (19:1) gave N-{[NЈ,NЈ-bis(2-hydroxyethyl)]-2-
aminoethyl}-NЉ-tritylglycyl carboxamide 2 as a viscous oil
Experimental
General procedures
Elemental analyses were carried out by the Microanalytical
Laboratory, University of Otago. Melting points were deter-
mined on an Electrothermal apparatus using the supplied stem-
corrected thermometer. NMR spectra were determined on a
Bruker AM-400 spectrometer. For spectra recorded using D2O
as the solvent, the sodium salt of 3-(trimethylsilyl)propionic-
2,2,3,3-d4-acid (TSP-d4) was used as the internal reference, and
spectra recorded using CDCl3 or CD3CN were referenced either
to internal TMS or to the residual solvent peak. FAB mass
spectra were determined on a VG 7070 spectrometer. Cyclic
voltammograms were recorded with a BAS 100W electro-
chemical analyser using a three-electrode cell under N2 with iR
compensation applied. The BAS-supplied electrodes used were
glassy carbon (working electrode), platinum wire (counter elec-
trode) and Ag–AgCl gel (reference electrode). The Ag–AgCl
gel reference electrode was determined to be 40 mV negative
of SCE. The solvent was acetonitrile containing 0.15 mol LϪ1
1
3
(41.9 g, 72%). H NMR (CDCl3): δ 7.73 (t, 1 H, J = 5.6 Hz,
CONH), 7.39 (d, 6 H, 3J = 7.3 Hz, Ar-H), 7.25 (t, 6 H, 3J = 7.3
3
Hz, Ar-H), 7.17 (t, 3 H, J = 7.3 Hz, Ar-H), 3.92 (br, 3 H, OH
and NH), 3.51 (t, 4 H, 3J = 5.0 Hz, NCH2CH2OH), 3.33
3
(dt, 2 H, J = 5.8, 5.6 Hz, CONHCH2), 2.94 (s, 2 H, TrNH-
CH2CO), 2.60 (overlapping t ϩ t, 6 H, CH2NR2 ϩ CH2-
CH2OH). 13C{1H} NMR (CDCl3): δ 172.54 (CONH), 145.30
(3 C, Ar-Cipso), 128.52 (6 C, Ar-C), 128.03 (6 C, Ar-C), 126.66
(3 C, Ar-C), 70.91 (CPh3), 59.62 (2 C, NCH2CH2OH), 56.89
(2 C, NCH2CH2OH), 54.84 (TrNHCH2CO), 48.07 (NCH2),
37.67 (NCH2).
N-{[NЈ,NЈ-Bis(2-hydroxyethyl)]-2-aminoethyl}glycyl carbox-
amide 3. The identity of 2 was further established by character-
isation of the product following detritylation. A solution of 2
(10.0 g, 0.022 mol) in MeOH (100 mL) was treated with concen-
trated HCl (40 mL), and the mixture was warmed at 50 ЊC for
30 min then evaporated to dryness under reduced pressure. The
residue was dissolved in water, and the aqueous layer was
washed well with ethyl acetate and again evaporated to dryness
under reduced pressure to give the dihydrochloride salt of N-
{[NЈ,NЈ-bis(2-hydroxyethyl)]-2-aminoethyl}glycyl carboxamide
3 (5.8 g, 95%). Crystallisation from MeOH by the addition of
[NBun ]ClO4 as the electrolyte. The ferrocenium–ferrocene
4
couple (Fcϩ–Fc), which was observed at ca. 0.43 V vs. Ag–
AgCl, was used as an internal reference in each measurement.
Potentials given in the text are reported vs. internal Fcϩ–Fc
(0.400 V vs. NHE).16 Acetonitrile for electrochemistry was dried
by distillation from CaH2 under N2. [NBun ]ClO4 was twice
4
recrystallised from dry ethyl acetate and then dried in vacuo at
80 ЊC. N,N-bis(2-hydroxyethyl)ethylenediamine7 was prepared
as previously reported. The salt dienؒ3HCl was prepared by
a method analogous to that described below for DEDؒ3HCl
(Found: C, 22.69; H, 7.73; N, 19.48; Cl, 50.24. Calc. for
C4H16N3Cl3: C, 22.60; H, 7.59; N, 19.77; Cl, 50.04%). DED was
purchased from Carbolabs, Inc. (New Haven, Conn.). All other
chemicals and reagents were purchased from Aldrich.
HPLC analyses were performed using a Waters 600 quater-
nary pump with WISP injector and Gilson 202 fraction col-
lector controlled by an HP chemstation and using an HP 1040A
diode array spectrometer as detector. The purity of new com-
pounds was checked by analytical HPLC with absorption
detection at 254 nm on a Bondclone C18 column using a linear
1
Et2O gave glistening plates, mp 158–160 ЊC. H NMR (D2O):
δ 3.96 (t, 4 H, 3J = 5.0 Hz, CH2OH), 3.86 (s, 2 H, H2NCH2CO),
3.72 (t, 2 H, 3J = 6.2 Hz, CONHCH2), 3.53 (t, 2 H, 3J = 6.2 Hz,
3
CH2NR2), 3.48 (br t, 4 H, J = 5.0 Hz, CH2CH2OH). 13C{1H}
NMR (D2O): δ 170.89 (CONH), 58.08 and 57.80 (each 2 C,
CH2CH2OH), 55.76 (H2NCH2CO), 43.12 (CONHCH2),
37.20 (CH2NR2) (Found: C, 34.5; H, 7.6; N, 15.1; Cl, 25.6. Calc.
for C8H19N3O3ؒ2HCl: C, 34.5; H, 7.6; N, 15.1; Cl, 25.5%).
NЈ,NЈ-Bis(2-hydroxyethyl)-Nٞ-trityldiethylenetriamine 4. A
solution of the trityl carboxamide (2) (10.0 g, 0.022 mol) in
926
J. Chem. Soc., Dalton Trans., 2000, 925–932

View Article Online
THF (200 mL) was cooled to 0 ЊC under N2 and borane–methyl
sulfide complex (6.81 mL of 10.5 M solution in THF, 0.071
mol) was added dropwise. After gas evolution ceased, the solu-
tion was heated under reflux for 4 h, then cooled to 0 ЊC and
the excess reagent was destroyed by the dropwise addition of
MeOH. The mixture was partitioned between ethyl acetate
and water, the ethyl acetate was evaporated from the organic
layer and the residue was chromatographed on silica gel. Ethyl
acetate eluted unidentified material, while ethyl acetate–MeOH
(19:1) gave NЈ,NЈ-bis(2-hydroxyethyl)-Nٞ-trityldiethylenetri-
amine 4 (6.14 g, 64%), which crystallised from acetone as cubes,
in water (25 mL). The mixture was stirred for 1 h then filtered
through Celite and the filtrate diluted to 300 mL with water.
This solution was loaded onto a Sephedex SP-C25 cation
exchange column (4 × 12 cm) and the column washed with
water. An aqueous solution of LiClO4 (0.05 mol LϪ1) eluted a
red–orange band which broadened as it was eluted, but did not
resolve into two bands. The first, major portion of the band was
collected until Kϩ could be detected by precipitation using an
aqueous solution of Na3[Co(NO2)6]. The remaining portion of
the red–orange band was collected separately. The first fraction
was concentrated to a volume of ca. 10 mL under reduced
pressure, and cooled at 4 ЊC overnight. Orange crystals of
mer-[Co(dien)(acac)(NO2)]ClO4 (mer-7) (0.35 g, 21.6%) were
collected and washed with a small amount of ice-cold water
and then extensively with Et2O, and dried in a desiccator. A
sample for analysis was recrystallised from hot water. The
second fraction was collected and treated as above, to give a
crystalline solid which was comprised of KClO4, mer-7 and
s-fac-7. Successive recystallisation of this solid from water
allowed removal of the less soluble mer-7 and KClO4. Filtration
and concentration of the filtrate gave pure s-fac-[Co(dien)(acac)-
(NO2)]ClO4 (s-fac-7) as large red crystals (0.088 g, 5.8%).
mer-[Co(dien)(acac)(NO2)]ClO4 (mer-7): 1H NMR (D2O):
δ 5.75 (s, 1 H, CH), 3.20 (m, 4 H, CH2NHCH2), 3.04 (m, 2 H,
CH2NH2), 2.64 (m, 2 H, CH2NH2), 2.17 (s, 3H, CH3CO), 2.132
(s, 3H, CH3CO). 13C{1H} NMR (D2O): δ 195.10, 194.09 (CO),
101.91 (CH), 51.99 (CH2NHCH2), 49.04 (CH2NH2), 29.20,
28.91 (CH3CO) (Found: C, 26.55; H, 4.96; N, 13.76; Cl, 8.73.
Calc. for C9H20N4O8ClCo: C, 26.58; H, 4.96; N, 13.78; Cl,
8.72%).
1
3
mp 154 ЊC. H NMR (CDCl3): δ 7.45 (br d, 6 H, J = 7.3 Hz,
3
Ar-H), 7.25 (br t, 6 H, J = 7.3 Hz, Ar-H), 7.16 (br t, 3 H,
3J = 7.3 Hz, Ar-H), 4.44 (br, 4 H, OH and NH), 3.52 (t, 4 H,
3J = 4.9 Hz, N(CH2CH2OH)2), 2.84 (t, 2 H, 3J = 5.8 Hz,
3
NHCH2CH2NR2), 2.75, 2.66 (2 × br t, each 2 H, J = 4.8 Hz,
CH2N), 2.55 (t, 4 H, 3J = 4.9 Hz, N(CH2CH2OH)2), 2.42 (t, 2 H,
3J = 5.8 Hz, CH2NR2). 13C{1H} NMR (CDCl3): δ 145.7 (3 C,
Ar-C), 128.67 (6 C, Ar-C), 128.86 (6 C, Ar-C), 126.37 (3 C,
Ar-C), 71.01 (CPh3), 59.50 (2 C, N(CH2CH2OH)2), 57.31
(2 C, N(CH2CH2OH)2), 52.26 (NCH2), 49.08 (NCH2), 46.58
(NCH2), 41.43 (NCH2).
N,N-Bis(2-chloroethyl)diethylenetriamine
trihydrochloride,
DCDؒ3HCl (5ؒ3HCl). A solution of the N3-trityltriamine 4
(6.81 g, 0.016 mol) in SOCl2 (200 mL) was stirred at room
temperature for 48 h. Excess SOCl2 was then removed under
reduced pressure, and the residue was dissolved in water and
washed several times with ethyl acetate. The aqueous layer
was evaporated to dryness under reduced pressure, and the
residue was recrystallised several times from MeOH–Et2O to
give N,N-bis(2-chloroethyl)diethylenetriamine trihydrochloride
(DCDؒ3HCl, 5ؒ3HCl) as hygroscopic plates, mp 138–140 ЊC
1
s-fac-[Co(dien)(acac)(NO2)]ClO4 (s-fac-7): H NMR (D2O)
2
δ 5.75 (s, 1 H, CH), 3.20 (m, J = 13.0 Hz, 4 H, CH2NHCH2),
3.04 (d, 3J = 7.2 Hz, 2 H, CH2NH2), 2.64 (d, 3J = 7.2 Hz,
2 H, CH2NH2), 2.16 (s, 6 H, CH3CO). 13C{1H} NMR (D2O):
δ 193.73 (CO), 100.82 (CH), 54.36 (CH2NHCH2), 45.37
(CH2NH2), 28.91 (CH3CO) (Found: C, 26.68; H, 5.15; N, 13.66;
Cl, 8.72. Calc. for C9H20ClCoN4O8: C, 26.59; H, 4.96; N, 13.78;
Cl, 8.72%).
1
3
(4.12 g, 86%). H NMR (D2O): δ 4.03 (t, 4 H, J = 5.5 Hz,
3
CH2CH2Cl), 3.79 (t, 4 H, J = 5.5 Hz, CH2CH2Cl), 3.77 (m,
2 H, CH2NR2), 3.71 (m, 2 H, CH2CH2NR2), 3.57 (t, 2 H,
3J = 7.0 Hz, CH2CH2NH2), 3.46 (t, 2 H, 3J = 7.0 Hz, CH2NH2).
13C{1H} NMR (D2O): δ 57.67 (2 C, CH2CH2Cl), 51.46 (CH2N),
47.36 (CH2N), 44.44 (CH2N), 40.00 (2 C, CH2CH2Cl), 37.98
(CH2NH2) (Found: C, 28.2; H, 6.8; N, 12.2. Calc. for C8H19-
Cl2N3ؒ3HCl: C, 28.5; H, 6.6; N, 12.5%).
mer-(N,N-Diethyldiethylenetriamine)(acetylacetonato)nitro-
cobaltate(III) perchlorate, mer-[Co(DED)(acac)(NO2)]ClO4 8.
To an intimate mixture of ground, recrystallised trans-K2[Co-
(acac)(CO3)(NO2)2]ؒH2O 6 (0.30 g, 0.738 mmol) and activated
charcoal (0.12 g) was added a solution of DEDؒ3HCl (0.20 g,
0.744 mmol) in MeOH (13 mL). The mixture was stirred for 2 h,
NaClO4ؒH2O (210 mg) was added to precipitate most of the
Kϩ, then filtered through Celite and the filtrate diluted to 300
mL with water. This solution was loaded onto a Sephedex
SP-C25 cation exchange column and the column washed with
water. An aqueous solution of NaClO4 (0.05 M) eluted a
magenta colored band which was collected, concentrated to a
volume of ca. 10 mL under reduced pressure, and cooled at 4 ЊC
overnight. Magenta colored crystals of mer-[Co(DED)(acac)-
(NO2)]ClO4 8 were collected and washed with a small amount
of ice-cold water and then extensively with Et2O, and dried in
a desiccator (0.14 g, 41%). 1H NMR (D2O): δ 5.91 (s, 1H, CH),
3.43 (dt, 2J = 13.0, 3J = 3.8 Hz, 1H, CH2NR2), 3.08 (dt,
Potassium trans-(acetylacetonato)(carbonato)dinitrocobalt-
ate(III) monohydrate, trans-K2[Co(acac)(CO3)(NO2)2]ؒH2O 6.
This complex was prepared essentially as described.17 These
authors assigned cis stereochemistry to the complex, but it
appears to actually be the trans isomer on the basis of NMR
and reactivity data. In the absence of excess nitrite ion the
complex rapidly decomposes in aqueous solution so in order to
purify it the complex is rapidly recrystallised from water by the
addition of an excess of KNO2. Crude K2[Co(acac)(CO3)-
(NO2)2]ؒH2O (10 g) was quickly dissolved in ice-cold water
(140 mL) and then filtered rapidly (ca. 800 mg recovered undis-
solved) into a solution of KNO2 (100 g) in water (20 mL). The
filtrate was cooled in ice for 45 min then filtered and the crystals
washed with cold H2O–MeOH (1:2 v/v), MeOH and EtOH
1
and dried over silica gel to give red needles (3.0 g). H NMR
(D2O, 1.0 mol LϪ1 NaNO2): δ 5.70 (s, 1H, CH), 2.16 (s, 6H,
CH3). 13C{1H} NMR (D2O, 1.0 mol LϪ1 NaNO2): δ 193.73
(CO), 100.91 (CH), 28.54 (CH3) (Found: C, 17.75; H, 2.34;
N, 6.70. Calc. for C6H7CoK2N2O9ؒH2O: C, 17.74; H, 2.23; N,
6.90%).
2J = 13.5, J = 3.9 Hz, 1H, CH2CH2NR2), 3.02 (dt, J = 12.5,
3J = 3.7 Hz, 1H, CH2NR2), 2.92 (dd, 2J = 13.5, 3J = 3.8 Hz, 1H,
CH2CH2NR2), 2.69 (m, 4H, CH2CH2NH2), 2.59 (dd, 2J = 12.8,
3J = 7.1 Hz, 1H, CH2CH3), 2.52 (m, 2H, CH2CH3), 2.39 (dq,
2J = 14.0, 3J = 7.2 Hz, 1H, CH2CH3), 2.28 (s, 3H, CH3CO), 2.23
(s, 3H, CH3CO), 1.06 (t, 3J = 7.2 Hz, 3H, CH3), 0.92 (t, 3J = 7.1
Hz, 3H, CH3). 13C{1H} NMR (D2O): δ 195.96, 193.39 (CO),
102.85 (CH), 63.64 (CH2NR2), 55.85, 52.04 (CH2NHR); 50.12
(CH2NH2); 48.79, 48.74 (CH2CH3); 29.14, 29.07 (CH3CO);
11.11, 9.32 (CH3) (Found: C, 33.57; H, 6.02; N, 11.82; Cl, 7.96.
Calc. for C13H28ClCoN4O8: C, 33.74; H, 6.10; N, 12.11; Cl,
7.66%). The compound was determined by analytical HPLC to
be 99.5% pure.
3
2
mer-(Diethylenetriamine)(acetylacetonato)nitrocobaltate(III)
perchlorate, mer-[Co(dien)(acac)(NO2)]ClO4 (mer-7) and s-fac-
(diethylenetriamine)(acetylacetonato)nitrocobaltate(III)
per-
chlorate, [Co(dien)(acac)(NO2)]ClO4 (s-fac-7). To an intimate
mixture of finely ground, recrystallised trans-K2[Co(acac)-
(CO3)(NO2)2]ؒH2O 6 (2.0 g, 4.92 mmol) and activated charcoal
(0.75 g) was added a solution of dienؒ3HCl (0.80 g, 3.76 mmol)
J. Chem. Soc., Dalton Trans., 2000, 925–932
927

View Article Online
Table 1 Crystal data and structure refinement for mer-[Co(DCD)(acac)(NO2)]ClO4 10 and trans-[Co(η2-DCD)(acac)(NO2)2] 9
mer-[Co(DCD)(acac)(NO2)]ClO4 10
trans-[Co(η2-DCD)(acac)(NO2)2] 9
Formula
M
C13H26Cl3CoN4O8
531.66
C13H26Cl2CoN5O6
478.22
Crystal system
Space group
Monoclinic
P21/c
Monoclinic
P21/c
T/K
λ/Å
293(2)
0.71069
293(2)
0.71069
a/Å
b/Å
15.7714(3)
11.4969(2)
19.242(14)
7.475(3)
c/Å
α/Њ
12.75700(10)
90
14.545(6)
90
β/Њ
108.74(1)
90
2191.35(6)
103.42(5)
90
2034.9(19)
γ/Њ
V/Å3
Z
4
4
Dc/g cmϪ3
1.611
1.195
1.561
1.145
µ/mmϪ1
Measured reflections
Unique reflections (Rint
Final R indices [I > 2σ(I)]
R indices (all data)
13351
4817 (0.026)
R1 = 0.0618, wR2 = 0.1545
R1 = 0.0852, wR2 = 0.1726
3592
3572 (0.026)
R1 = 0.0790, wR2 = 0.1912
R1 = 0.1915, wR2 = 0.2353
)
trans-[ꢀ2-N,N-Bis(2-chloroethyl)diethylenetriamine](acetyl-
ther 25 mg (6.3%). Total yield = 16.4%. 1H NMR (acetone-
d6): δ 6.65 (br t, 1H, NH), 5.95 (s, 1H, CH), 4.62, 4.28 (br,
1H, NH2); (4.10, 3.85), (3.95, 3.70) (m, 1H, CH2Cl); 3.74,
3.05 (m, 1H, CH2CH2NR2); 3.38, 2.70 (m, 1H, CH2NR2);
3.20, 2.85 (m, 1H, CH2CH2NH2); (3.12, 2.85), (3.10, 2.75) (m,
1H, CH2CH2Cl), 2.80, 2.78 (m, 1H, CH2NH2); 2.312, 2.307
(s, 3H, CH3). 13C NMR (acetone-d6): δ 194.24, 191.72 (CO);
101.02 (CH), 62.66 (CH2NR2), 60.27, 54.01 (CH2CH2Cl);
50.62 (CH2NH2), 48.83, 48.11 (CH2NHCH2); 38.00, 37.35
(CH2Cl); 27.55, 27.47 (CH3CO). FABϩ mass spectrum (m-
NBA) m/z (%): 435, 433, 431 (Mϩ, 15, 65, 100; 37Cl2, 37Cl35Cl,
35Cl2); 389, 387, 385 (Mϩ Ϫ NO2, 13, 67, 100; 37Cl2, 37Cl35Cl,
35Cl2). High resolution FABϩ mass spectrum (m-NBA) m/z:
Mϩ(37Cl2): calc. 435.0604; found 435.0596; Mϩ(35Cl37Cl): calc.
433.0634; found 433.0651; Mϩ(35Cl2): calc. 431.0663; found
431.0666.
acetonato)dinitrocobaltate(III),
(NO2)2] 9. To stirred suspension of recrystallised
trans-[Co(ꢀ2-DCD)(acac)-
a
trans-K2[Co(acac)(CO3)(NO2)2]ؒH2O 6 (0.30 g, 0.746 mmol)
in dry MeOH (16 mL) was added DCDؒ3HCl (0.25 g, 0.744
mmol) and activated charcoal (0.11 g). The mixture was stirred
for 45 min at 20 ЊC then NaClO4ؒH2O (0.21 g.) was added (to
precipitate most of the Kϩ as KClO4) and then filtered through
Celite. The charcoal and Celite were washed with small
amounts of MeOH and water, and the washings were added to
the filtrate. The filtrate was diluted to 250 mL with water
and cooled whereupon a yellow–orange microcrystalline solid
slowly formed. Filtration gave trans-[Co(η2-DCD)(acac)-
(NO2)2] 9 (15 mg, 4.2%) which was washed with cold water and
1
then repeatedly with dry Et2O and dried over silica gel. H
NMR (CDCl3): δ 5.7 (br s, 1H, NH2), 5.49 (s, 1H, CH), 5.2 (br s,
1H, NH2), 3.68 (m, 4H, CH2Cl), 3.38 (br m, 2H, CH2CH2NH2),
3.04 (m, 4H, CH2CH2Cl), 2.86 (m, 2H, CH2NH2), 2.55 (br m,
2H, CH2NR2), 2.21 (s, 3H, CH3), 2.2 (br m, 2H, CH2CH2NR2),
2.06 (s, 3H, CH3). 13C NMR (CDCl3): δ 98.17 (CH), 55.73
(CH2CH2Cl), 52.09 (CH2NR2), 50.37 (CH2CH2NH2), 47.93
(CH2CH2NR2), 42.66 (CH2NH2), 41.91 (CH2Cl), 26.98, 26.30
(CH3CO). FABϩ mass spectrum (m-NBA) m/z (%): 435, 433,
431 (Mϩ Ϫ NO2, 15, 65, 100; 37Cl2, 37Cl35Cl, 35Cl2); 389, 387, 385
(Mϩ Ϫ 2NO2, 15, 60, 100; 37Cl2, 37Cl35Cl, 35Cl2) (Found: C,
32.23; H, 5.57; N, 14.61. Calc. for C13H26Cl2CoN5O6: C, 32.65;
H, 5.48; N, 14.64%). The compound was determined by
analytical HPLC to be 95.6% pure.
X-Ray crystal structure determinations
A summary of crystal data, structure solutions and refinements
is given in Table 1. After initial isotropic refinement, anisotropic
thermal parameters were refined for all non-hydrogen atoms.
Hydrogen atoms were refined using the riding model, except for
the positions of the hydrogens on N(1) and N(2) in 10 which
were refined individually.
CCDC reference number 186/1832.
See http://www.rsc.org/suppdata/dt/a9/a909447d/ for crystal-
lographic files in .cif format.
mer-[N,N-Bis(2-chloroethyl)diethylenetriamine](acetylaceton-
ato)nitrocobaltate(III) perchlorate mer-[Co(DCD)(acac)(NO2)]-
ClO4 10. The tridentate DCD complex was isolated from the
filtrate in the above preparation of 9. The filtrate was loaded
onto a Sephedex column (2 × 16 cm) and washed with water.
Elution was begun with 0.025 mol LϪ1 NaClO4 (ca. 200 mL)
and then with 0.05 mol LϪ1 NaClO4 until the first band was
collected. In this way, the residual Kϩ remaining after the
precipitation of KClO4 was not eluted (no precipitate with
aqueous Na3[Co(NO2)6] solution) with the product which
precluded difficulties with cocrystallisation of KClO4. The
magenta coloured solution (ca. 100 mL) of the first band
eluted from the column was evaporated to small volume (ca.
8 ml) under reduced pressure and cooled at 4 ЊC for a week.
The crystals that formed were filtered off and washed with a
little ice-cold water then repeatedly with Et2O and dried over
silica gel to give mer-[Co(DCD)(acac)(NO2)]ClO4 10 (0.040 g,
10.1%). From the filtrate and water wash was isolated a fur-
Cytotoxicity testing
Cell lines were maintained and tested in αMEM with 5% fetal
bovine serum. Stock solutions of compounds were prepared in
acetone and stored at Ϫ80 ЊC (9,10), or were dissolved in cul-
ture medium immediately before use. Inhibition of cell prolifer-
ation was tested using log phase cultures in 96 well trays, using
18 h exposure to drugs and growth for a further 4 days before
staining with methylene blue as previously.18 The IC50 was
defined as the drug concentration required to lower cell density
to 50% of controls on the same plate. Kinetics of cell killing, as
assessed by clonogenic assay, was compared under aerobic and
anoxic (<10 ppm O2) conditions by gassing stirred cell suspen-
sions with 5% CO2 in air or N2, respectively, as described previ-
ously.19 Drug concentrations were chosen that gave similar
rates of killing, and the drug concentration × time to reduce
cell survival to 10% of controls (CT10) was determined as an
inverse measure of cytotoxic potency.
928
J. Chem. Soc., Dalton Trans., 2000, 925–932

View Article Online
ventional synthesis of cobalt() complexes, oxidation of
Results and discussion
Syntheses
cobalt() in the presence of the ligands, can lead to oxidative
dealkylation of the susbstituents on nitrogen, and in any case
can give rise to mixtures when mixed-ligand complexes are
desired. We investigated the reaction of [Co(NO2)6]3Ϫ with dien,
but the rate of substitution was too slow and, when the mustard
ligand was used, the reaction could not compete with decom-
position of the mustard.
The preparation of cobalt() complexes containing multi-
dentate nitrogen mustard ligands presents several synthetic
challenges. Firstly, the inert nature of the cobalt() center,
crucial to the design of the complexes for cancer chemotherapy,
results in slow substitution reactions and consequent difficulties
when the incoming ligand is reactive. This is the case for the
mustard ligands which are susceptible to hydrolysis or solvoly-
sis in their free base forms, and are therefore stored and handled
as their hydrochloride salts. Deprotonation of these salts is an
integral part of the coordination process. The mustard ligands
themselves are not trivial to prepare, and ideally should be
introduced into the coordination sphere at the last step of the
preparation procedure. The unsubstituted parent tridentate tri-
amine dien and the non-alkylating analog DED were used to
develop the synthetic routes and refine the preparative condi-
tions prior to synthesizing the mustard complexes containing
the DCD ligand. Although DED and dien do not undergo
corresponding decomposition reactions to DCD in their free
base forms, the ligands were used as the trihydrochloride salts in
the syntheses in order to model the reaction conditions for the
much more reactive mustard DCD. Complexes of the non-toxic
analogs are also useful as controls in the biological evaluation.
DED and dien are commercially available, and were con-
verted to hydrochloride salts prior to their use in complexation
reactions. The novel mustard DCD was prepared as outlined
in Scheme 1. CDI-induced coupling of N-tritylglycine and
The precursor complex we have previously utilized for the
preparation of cobalt complexes containing bidentate mustard
ligands was trans-[Co(Racac)2(NO2)2]ϩ, which undergoes
relatively fast substitution reactions for a cobalt() center.6,7
Reactions with the neutralized hydrochloride salts of bidentate
mustards (L) gave products [Co(Racac)2(L)]ϩ, including the
lead compound [Co(DCE)(Meacac)2]ϩ 1. A closely related pre-
cursor complex is K2[Co(acac)(CO3)(NO2)2]ؒH2O 6 which had
been reported in the literature and assigned as the isomer con-
taining cis-NO2 ligands.17 In the original preparation the prod-
uct was isolated from the reaction mixture and characterized
without further purification. Like Na[Co(acac)2(NO2)2]ؒH2O,
the complex is subject to decomposition in aqueous solution
but is stabilized by the presence of excess NO2Ϫ ions, and can be
recrystallised under these conditions. We observed that the acac
ligand exhibits two equivalent CH3 groups in the NMR spec-
trum, consistent with trans stereochemistry rather than the cis
geometry proposed in the original report.
trans-K2[Co(acac)(CO3)(NO2)2]ؒH2O 6 also exhibits rela-
tively rapid substitution chemistry, and as such proved to be
useful for the preparation of the target tridentate amine and
nitrogen mustard complexes. The amines are utilized as the
trihydrochloride salts, and the reactions proceed without the
requirement for added base since the carbonato ligand acts as a
base, neutralising two equivalents of acid per mustard ligand,
and is itself labilised in the process. The precursor complex is
somewhat soluble in MeOH, allowing the substitution reactions
to proceed in this medium, so long as the amine hydrochloride
salts are also sufficiently soluble which is the case for DEDؒ
3HCl and DCDؒ3HCl, but not dienؒ3HCl. Another factor
which results in improved yields of the mustard complexes is
that solvolysis of the free mustard ligand is slower in MeOH
than in aqueous solution. In addition, the substitution reac-
tions are expected to proceed by a dissociative mechanism, and
the MeOH ligand in an intermediate solvato complex would be
more readily displaced by the incoming mustard nitrogen than
an aqua ligand.
trans-K2[Co(acac)(CO3)(NO2)2]ؒH2O 6 reacts with dienؒ3HCl
in the presence of a charcoal catalyst to produce the cationic
complex [Co(dien)(acac)(NO2)]ϩ 7. Purification by cation
exchange chromatography serves to separate the monocations
from other products, but does not allow complete separation of
Scheme 1
1
the one mer and two fac isomers possible for this complex. H
N,N-bis(2-hydroxyethyl)ethylenediamine gave a good yield of
the carboxamide 2 (characterized in part as the deprotected
carboxamide 3) which was reduced with borane–methyl sulfide
complex to give the N-trityltriaminediol 4. Reaction of this
with thionyl chloride hydrolyzed the trityl group and converted
the diol to the mustard DCD 5, isolated as the trihydrochloride
salt DCDؒ3HCl.
Stereochemistry is a further important consideration in
the preparation of mixed ligand cobalt() complexes. An
octahedral complex containing a flexible tridentate triamine
ligand such as diethylenetriamine (dien) may adopt fac or mer
geometry, ∆ or Λ configuration at the cobalt center and, for
the mer isomer, (R) or (S) configuration at the secondary amine.
In addition, the unsymmetrical nature of the substituted
triamines DCD and DED means that additional diastereomers
are potentially possible, especially when the other three ligands
on cobalt are not identical.
NMR spectroscopy of the product after chromatography shows
a mixture of isomers, which could be assigned (after subsequent
isolation of the pure diastereomers) to the mer isomer (the pre-
dominant product), the s-fac isomer and a very small amount
of a third compound, presumably the u-fac isomer. The
perchlorate salts of the mer and s-fac isomers can be separated
by fractional crystallization, with the less soluble isomer
mer-[Co(dien)(acac)(NO2)]ClO4 (mer-7) crystallizing first. Pure
s-fac-[Co(dien)(acac)(NO2)]ClO4 (s-fac-7) can be isolated as the
second fraction to crystallize. The mer, s-fac and u-fac isomers
There are a few reported examples of cobalt() complexes
containing N-substituted dien derivatives as ligands. The con-
J. Chem. Soc., Dalton Trans., 2000, 925–932
929

View Article Online
are clearly distinguished by 1H and 13C NMR spectroscopy on
the basis of the symmetry differences. The mer isomer contains
the acac, NO2 and secondary nitrogen of the dien ligand in a
plane of symmetry, and has inequivalent acac methyl groups. In
the s-fac isomer the plane of symmetry contains the secondary
nitrogen but bisects the acac group, and thus the two acac
methyl groups are equivalent. The u-fac isomer has no sym-
metry planes or axes and thus has no equivalent resonances in
the NMR spectra.
The reaction of DEDؒ3HCl with trans-K2[Co(acac)-
(CO3)(NO2)2]ؒH2O 6 proceeds under very mild conditions, in
MeOH for 2 h at room temperature in the presence of charcoal.
The product comprises a single stereoisomer, mer-[Co(DED)-
(acac)(NO2)]ϩ 8, isolated in 41% yield as the ClO4Ϫ salt follow-
ing cation exchange chromatography. The assignment of the
mer stereochemistry was made by comparison of its NMR data
with that of the corresponding DCD complex (vide infra). DED
contains a tertiary amine donor and is a more sterically
demanding ligand than dien, and presumably this results in
significant destabilisation of the fac isomers.
DCD)(acac)(NO2)2] 9 the trans-NO2 ligands and the bidentate
nature of the DCD ligand with the tertiary mustard nitrogen
not coordinated to cobalt (Fig. 1), and for mer-[Co(DCD)-
(acac)(NO2)]ClO4 10 the mer arrangement and stereochemistry
at the secondary nitrogen of the DCD ligand (Fig. 2). The
crystal structure of the tridentate DCD complex (10) shows a
hydrogen bonding interaction between the hydrogen on the
secondary nitrogen of the DCD ligand and one of the oxygen
atoms of the NO2 ligand, which may explain why only this
diastereomer is observed.
X-Ray crystal structure determinations
The Co–N bond lengths to the primary and secondary nitro-
gens in 9 and 10 are similar, averaging 1.937 Å for mer-
[Co(DCD)(acac)(NO2)]ClO4 10 and 1.948 Å for trans-[Co(η2-
DCD)(acac)(NO2)2] 9. However the tertiary amine Co–N(3)
distance of 2.114(4) Å in 10 is exceptionally long. Related
complexes which contain a coordinated terminal tertiary amine
(i.e. not constrained as part of a macrocycle) are [Co(DCE)-
(Meacac)2]ClO4 (1)20 and [Co(L)]PF6,21 which contains an
acylic hexadentate ligand (L) incorporating a diethylamine
donor. The tertiary amine Co–N distances in these molecules are
2.092(4) and 2.082(4) Å, respectively. The long Co–N distances
arise as a consequence of the steric bulk of the coordinated
tertiary amine. This is also evident from the N(3)–Co–O(2)
angle in 10 which is 97.3(1)Њ.
The plane of the NO2 ligand is not constrained to a partic-
ular orientation for electronic reasons. However, in mer-
[Co(DCD)(acac)(NO2)]ClO4 10 it is oriented such that it is
aligned with the O(2)–Co–N(2) plane. The hydrogen atom
H(2NA) on N(2), the orientation of which is determined by
the configuration at N(2), was located crystallographically
and points towards O(4) of the NO2 ligand, indicative of an
N(2)–H(2NA)–O(4) hydrogen bond. The N(2)–H(2NA) and
H(2NA) ؒ ؒ ؒ O(4) distances of 0.85 and 2.16 Å, respectively,
are consistent with this. A similar situation occurs for the
orientation of one NO2 ligand in 9, with an N(1)–H(1NA)–O(3)
hydrogen bond (corresponding distances are 0.90 and 2.17 Å).
The corresponding reaction of DCDؒ3HCl with trans-
K2[Co(acac)(CO3)(NO2)2]ؒH2O 6 was carried out in a shorter
reaction time (45 min). Dilution of the reaction mixture with
water prior to cation exchange chromatography resulted in
crystallization of a neutral complex in low yield, identified as
the bidentate DCD complex trans-Co(η2-DCD)(acac)(NO2)2 9,
which contains the triamine ligand coordinated through only
the primary and secondary nitrogen donor atoms. Cation
exchange chromatography of the material remaining in solution
and elution with 0.05 mol LϪ1 NaClO4 solution gave the tri-
dentate complex mer-[Co(DCD)(acac)(NO2)]ClO4 10 contain-
ing the η3-DCD ligand coordinated through all three nitrogens.
The DED complex (8) and the bidentate (9) and tridentate
1
(10) DCD complexes were characterised by H and 13C NMR
spectroscopy and FAB mass spectrometry. The identities of the
two DCD complexes were further established by X-ray crystal
structure determinations. These confirmed for trans-[Co(η2-
Fig. 1 ORTEP view and atomic labeling scheme of the trans-[Co(η2-
DCD)(acac)(NO2)2] cation (9). Selected bond lengths (Å): Co–O(1)
1.869(7), Co–O(2) 1.887(6), Co–N(1) 1.932(8), Co–N(5) 1.946(8), Co–
N(4) 1.954(8), Co–N(2) 1.965(7).
Fig. 2 ORTEP view and atomic labeling scheme of the mer-[Co-
(DCD)(acac)(NO2)]ClO4 cation (10). Selected bond lengths (Å): Co–
O(2) 1.885(3), Co–O(1) 1.913(3), Co–N(4) 1.923(3), Co–N(2) 1.927(4),
Co–N(1) 1.946(4), Co–N(3) 2.114(4).
930
J. Chem. Soc., Dalton Trans., 2000, 925–932

View Article Online
Table 2 Biological activity of cobalt mustard complexes and the corresponding free mustard ligands
Growth inhibition assay
Clonogenic assay
IC50 a/µM
CT10(air)b/
No.
Compound
AA8
UV4
µM hϪ1
Air/N2 ratioc
Mustard ligands
DCE
1.44 0.26
0.055 0.003
2.67 0.02
5
DCD
50.0
1
Co complexes
8
10
9
[Co(DED)(acac)(NO2)]ClO4
3710 410
750 100
27
3150 20
44800
262
1.0
5.2
[Co(DCD)(acac)(NO2)]ClO4
[Co(η2-DCD)(acac)(NO2)2]
[Co(DCE)(Meacac)2]ClO4
35.4
1.5
9
1
4.6 0.6
0.113 0.019
2.34
20.3
a IC50 values determined against aerobic cells, using an exposure time of 18 h. Values are means SEM (standard error of the mean). Values without
SEM are for a single determination only. b CT10: drug concentration (µM) × time (h) required to reduce cell survival to 10% of controls under
hypoxic conditions, using UV4 cells in a clonogenic assay (see text). c Ratio of CT10 values in air and N2 [CT10(air)/CT10(nitrogen)].
Electrochemistry
Biological evaluation
The substitutional lability of cobalt() often leads to irrevers-
ible electrochemical behavior of cobalt() complexes due to
ligand loss from the reduced Co() form. Non-aqueous solvents
are often employed in an attempt to slow solvolysis reactions of
the reduced Co() complex in order to enhance chemical revers-
ibility for the Co()–Co() couple. However, even in MeCN
solution, irreversible behavior is observed for [Co(DCE)-
(Meacac)2]ClO4 1 and the related tropolonate complexes.7,10
The chelate effect can also improve electrochemical revers-
ibility, with multidentate ligands slowing down the rate of
ligand loss from Co(). However, when the electrochemistry of
mer-[Co(dien)(acac)(NO2)]ClO4 (mer-7), mer-[Co(DED)(acac)-
(NO2)]ClO4 8 and mer-[Co(DCD)(acac)(NO2)]ClO4 10 were
investigated in MeCN, both cyclic voltammetry (CV) and
Osteryoung square wave voltammetry (OSWV, step E = 4 mV,
SW amplitude = 25 mV, SW frequency = 15 Hz) showed com-
pletely irreversible behavior (ia ≈ 0, i.e. no reoxidation wave
observed by CV), even though 7, 8 and 10 contain tridentate
rather than bidentate ligands. For each complex the cathodic
peak potentials (Epc) were similar in both the CV and OSWV
experiments, using both glassy carbon and platinum working
electrodes. OSWV at glassy carbon gave somewhat more repro-
ducible values for Epc and are reported here: for mer-7,
Epc = Ϫ1.11 V (versus ferrocenium–ferrocene, Fc), for 8,
Epc = Ϫ810 mV (Fc), and for 10, Epc = Ϫ670 mV (Fc). We have
found for irreversible couples that while Epc is not thermo-
dynamically meaningful, it does nonetheless have predictive
value and can be used to compare similar complexes.7,9,10 The
shift of Epc from Ϫ1.11 V for the dien complex mer-7 to Ϫ810
mV for the DED complex (8) can be attributed to the weaker
ligand field of the tertiary nitrogen donor in DED. The shift
of 140 mV to a more positive potential on going from
the diethyl-substituted complex 8 to the bis(chloroethyl)-
substituted complex 10 can be attributed to both electronic and
steric effects, probably with the former predominating. The
shift to more positive potential is consistent with the shifts
observed for the bidentate diamine complexes containing a
number of different ancillary ligand sets, including those based
on dithiocarbamates, examples of which display nearly revers-
ible electrochemistry. For complexes in which L = DCE or DEE
(N,N-diethylethylenediamine), ∆Ep = 175 mV for [Co(L)-
The free mustard DCD 5, the DED complex 8, and the biden-
tate and tridentate DCD complexes 9 and 10 were evaluated for
growth inhibitory properties (IC50 values) in two cell lines in
culture, and the results are summarized in Table 2. AA8 is a
Chinese hamster ovary fibroblast line, and the related mutant
subline UV4 is deficient in nucleotide excision repair, and it is
therefore hypersensitive to agents which form bulky DNA
monoadducts or crosslinks.22 The ratio of IC50 values in these
two lines, (IC50(AA8)/IC50(UV4)) therefore provides some
information about the mechanism of cytotoxicity, with ratios
significantly > 10 implying DNA-crosslinking as the primary
cytotoxic event.
The tridentate mustard DCD proved to be much less cyto-
toxic than the corresponding bidentate analogue DCE used
previously (see Table 2). The reason for this is not known; while
both would be expected to have similar chemical reactivity,
the (tris-cationic) tridentate analogue DCD may have slower
kinetics of cell uptake. However, both had similar AA8/UV4
IC50 ratios (19- vs. 26-fold), suggesting that DNA crosslinking is
the major mode of cytotoxicity in both cases. As expected, the
model non-mustard complex 8 had very low cytotoxicity
(IC50 > 3 mM), and showed little differential between AA8 and
UV4. The tridentate DCD complex 10 was significantly less
toxic than the free mustard DCD in both cell lines, showing that
complexation of the mustard nitrogen results in significant
deactivation. Complex 10 also had a high AA8/UV4 IC50 ratio
(21-fold), indicating that aerobic cytotoxicity was due to a
DNA cross-linking event; probably release of the free mustard
to a small extent. The bidentate DCD complex 9, where the
mustard nitrogen is not complexed to the metal, retained high
cytotoxicity (comparable to that of the free mustard DCD).
The hypoxic selectivities of 1, 8 and 10 (sufficient quantities
of 9 were not available) were evaluated in stirred suspension
cultures of UV4 cells, using a clonogenic assay following a 4 h
drug exposure under either aerobic (5% CO2 in air) or hypoxic
(5% CO2 in N2) conditions23 (Table 2). Under these conditions,
the bidentate mustard complex [Co(DCE)(Meacac)2]ϩ
1
showed an air/N2 ratio of ca. 20 (Table 2).6–8 As expected, the
tridentate non-mustard complex 8 was very non-potent, and
had a ratio of unity. In contrast, the tridentate mustard
complex 10 showed a 5-fold selectivity for hypoxic conditions
(Table 2).
ϩ
ϩ
₁
(acac)2] , ∆Ep = 162 mV for [Co(L)(trop)2] , ∆E = 155 mV
₂

for [Co(L)(S2CNMe2)2]ϩ and ∆E = 180 mV for [Co(L)-
All of the bidentate mustard complexes containing substi-
tuted acac ligands show hypoxia selectivity, with the maximum
value observed for the Meacac derivative 1. The tridentate mus-
tard complex 10 continues this trend, and although the hypoxia
selectivity is not as high as that measured for 1, it is better than
the series of bidentate mustard complexes containing dithio-
₁
₂

(S2CNEt2)2]ϩ.7,9,10
The Ep value for [Co(DCE)(Meacac)2]ϩ 1 under these condi-
tions was found to be Ϫ780 mV (vs. Fc) similar to the reported
value in CH2Cl2 (Ϫ853 mV)8 and much more negative than the
value observed for the DCD complex 10 (Ϫ670 mV).
J. Chem. Soc., Dalton Trans., 2000, 925–932
931

View Article Online
5 W. R. Wilson, in Cancer Biology and Medicine. Vol. 3. The Search for
New Anticancer Drugs, ed. M. J. Waring and B. A. J. Ponder,
Kluwer, Lancaster, 1992, pp. 87–131.
6 D. C. Ware, W. R. Wilson, W. A. Denny and C. E. F. Rickard,
J. Chem. Soc., Chem. Commun., 1991, 1171.
carbamate or tropolonate ligands (for example [Co(DCE)-
(S2CNEt2)2]BPh4 and [Co(DCE)(trop)2]ClO4) for which no
hypoxia selectivity was observed.9,10
7 D. C. Ware, B. D. Palmer, W. R. Wilson and W. A. Denny, J. Med.
Chem., 1993, 36, 1839.
8 W. R. Wilson, J. M. Moselen, S. Cliffe, W. A. Denny and D. C. Ware,
Int. J. Radiat. Biol. Oncol. Phys., 1994, 29, 323.
9 D. C. Ware, H. R. Palmer, F. B. Pruijn, R. A. Anderson, P. J.
Brothers, W. A. Denny and W. R. Wilson, Anti-cancer Drug Des.,
1998, 13, 81.
10 D. C. Ware, H. R. Palmer, P. J. Brothers, C. E. F. Rickard, W. R.
Wilson and W. A. Denny, J. Inorg. Biochem., 1997, 68, 215.
11 M. Simic and J. Lilie, J. Am. Chem. Soc., 1974, 96, 291.
12 D. R. Eaton and A. O’Reilly, Inorg. Chem., 1987, 26, 4185.
13 R. F. Anderson, W. A. Denny, D. C. Ware and W. R. Wilson,
Br. J. Cancer, 1996, 74 (Suppl. XXVII), S48.
14 B. A. Teicher, M. J. Abrams, K. W. Rosbe and T. S. Herman, Cancer
Res., 1990, 50, 6971.
15 D. C. Ware, B. G. Siim, K. J. Robinson, W. A. Denny, P. J. Brothers
and G. R. Clark, Inorg. Chem., 1991, 30, 3750.
16 R. R. Gagne, C. A. Koval and G. C. Lisensky, Inorg. Chem., 1980,
19, 2854.
17 H. Ichikawa and M. Shibata, Bull. Chem. Soc. Jpn, 1970, 43, 3789.
18 W. R. Wilson, W. A. Denny, S. M. Pullen, K. M. Thompson,
A. E. Li, L. H. Patterson and H. H. Lee, Br. J. Cancer, 1996, 74
(suppl. XXVII), S43.
Conclusions
A cobalt() complex containing the tridentate mustard ligand
DCD was successfully prepared, although the difficulty in
coordinating the bulky tertiary amine moiety in the ligand is
demonstrated by the isolation, in low yield, of a complex con-
taining the DCD ligand coordinated in a bidentate fashion,
bearing the mustard nitrogen atom as a pendant arm. Cell cul-
ture studies show that the cytotoxicity of the mustard is greatly
diminished in the tridentate complex, where the mustard is
deactivated through coordination to the metal, but not in the
bidentate complex where the mustard nitrogen remains uncom-
plexed. The cytotoxicity of the tridentate mustard DCD is more
effectively masked upon complexation to cobalt() than is
observed for the bidentate mustard DCE. However the tri-
dentate mustard complex 10 is less selective under hypoxic con-
ditions. A possible reason for this is that the higher potential of
10 relative to 1 might reduce the ability of O2 to compete
against 10 for reducing equivalents.13
19 B. G. Siim, G. J. Atwell and W. R. Wilson, Br. J. Cancer, 1994, 70,
Acknowledgements
596.
20 D. C. Ware, W. A. Denny and G. R. Clark, Acta Crystallogr., Sect.
C, 1997, 53, 1058.
21 P. J. Brothers, G. R. Clark, H. R. Palmer and D. C. Ware, Inorg.
Chem., 1997, 37, 5470.
D. C. W. is grateful to the Cancer Society of New Zealand for
financial support of this work.
22 C. A. Hoy, E. P. Salaza and L. H. Thompson, Mutat. Res., 1984,
130, 321.
23 W. R. Wilson, W. A. Denny, S. J. Twigden, B. C. Baguley and
J. C. Probert, Br. J. Cancer, 1984, 49, 215.
References
1 J. M. Brown and A. J. Giaccia, Cancer Res., 1998, 58, 1408.
2 W. A. Denny, W. R. Wilson and M. P. Hay, Br. J. Cancer, 1996, 74
(Suppl. XXVII), S32.
3 A. M. Rauth, T. Melo and V. Mistra, Int. J. Radiat. Oncol. Biol.
Phys., 1998, 42, 755.
4 I. J. Stratford and P. Workman, Anti-cancer Drug Des., 1998, 13,
519.
Paper a909447d
932
J. Chem. Soc., Dalton Trans., 2000, 925–932