Cytotoxic titanium salan complexes: Surprising interaction of salan and alkoxy ligands

Article abstract of DOI:10.1002/chem.200902312

The synthesis, biochemical evaluation, and hydrolysis studies of a wide selection of alkyl- and halogen-substituted titanium salan alkoxides are presented herein. A systematic change in the employed alkoxides revealed that both the bulk of the salan ligands and the steric demand of the labile ligands are of great importance for the obtained biological activity. Surprisingly, these two factors are not independent from each other; lowering the steric demand of the alkoxide of a hitherto nontoxic complex renders it cytotoxic. Therefore, our data suggest that the overall size of the complex exerts a strong influence on its biological activity. To decide whether the correlation between the cytotoxicity and the steric demand of the whole complex is merely based on an altered hydrolysis or on the interaction with biomolecules, the behavior of selected complexes under hydrolytic conditions and the influence of transferrin were investigated. Complexes differing only in their labile alkoxy ligands gave the same hydrolysis products with similar hydrolysis rates but displayed cytotoxicities that differed in the range of one order of magnitude. Thus, it seems that the hydrolysis product is not the active species but rather that the unhydrolysed complex is important for the first interaction with a biomolecule. This promoted the idea of hydrolysis being a detoxification pathway. In accordance with the above conclusion, chloro-substituted complex [Ti(Ph ^ClN^Me)₂(O^iPr)₂] displayed a high cytotoxicity (IC₅₀ ≈ 5 μM) and surprisingly high hydrolytic stability (t_1/2=108 h). These findings, together with the observed cytotoxicity in a cisplatin-resistant cell line, make halo-substituted salan complexes an interesting target for further studies.

Full text of DOI:10.1002/chem.200902312

FULL PAPER
DOI: 10.1002/chem.200902312
Cytotoxic Titanium Salan Complexes: Surprising Interaction of Salan and
Alkoxy Ligands
Timo A. Immel, Ulrich Groth, and Thomas Huhn*[a]
Abstract: The synthesis, biochemical
evaluation, and hydrolysis studies of a
wide selection of alkyl- and halogen-
substituted titanium salan alkoxides are
presented herein. A systematic change
in the employed alkoxides revealed
that both the bulk of the salan ligands
and the steric demand of the labile lig-
between the cytotoxicity and the steric
demand of the whole complex is
merely based on an altered hydrolysis
or on the interaction with biomole-
cules, the behavior of selected com-
plexes under hydrolytic conditions and
the influence of transferrin were inves-
tigated. Complexes differing only in
their labile alkoxy ligands gave the
same hydrolysis products with similar
hydrolysis rates but displayed cytotox-
icities that differed in the range of one
order of magnitude. Thus, it seems that
the hydrolysis product is not the active
species but rather that the unhydro-
lysed complex is important for the first
interaction with a biomolecule. This
promoted the idea of hydrolysis being
a detoxification pathway. In accordance
with the above conclusion, chloro-sub-
A
stituted complex [Ti
displayed high cytotoxicity (IC50
(PhClNMe (OiPr)2]
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
obtained biological activity. Surprising-
ly, these two factors are not indepen-
ACHTUNGTRENNUNGdent from each other; lowering the
steric demand of the alkoxide of a hith-
erto nontoxic complex renders it cyto-
toxic. Therefore, our data suggest that
the overall size of the complex exerts a
strong influence on its biological activi-
ty. To decide whether the correlation
a
ꢀ5 mm) and surprisingly high hydrolytic
stability (t1/2 =108 h). These findings,
together with the observed cytotoxicity
in a cisplatin-resistant cell line, make
halo-substituted salan complexes an in-
teresting target for further studies.
Keywords: antitumor agents · cyto-
toxicity · hydrolysis · structure–ac-
tivity relationships · titanium salan
complexes
Introduction
fortunately the spectrum of activity of these platinum com-
plexes is limited and they are associated with significant
dose-limiting toxicities.[1–3] Therefore, much research has
been devoted to the identification of new metal complexes
with lower toxicity, similar antitumor activity, and reduced
tumor resistance.[4–14] Titanium(IV) complexes showed en-
couraging antitumor activity in various cell lines and little
cross resistance to cisplatin was observed.[15–17] Although
much effort was undertaken to identify completely new tita-
nium complexes that show antitumor properties, nearly all
investigated complexes were derivatives of either titanocene
Today, metal complexes, such as cisplatin, carboplatin, and
oxaliplatin, play an important role in cancer therapy, but un-
[a] T. A. Immel, Prof. Dr. U. Groth, Dr. T. Huhn
Fachbereich Chemie and
Konstanz Research School Chemical Biology
Universitꢀt Konstanz
Universitꢀtsstrasse 10
Fach 720, 78457 Konstanz (Germany)
Fax : (+49)7531-884424
E-mail: thomas.huhn@uni-konstanz.de
dichloride ([Ti
(OEt)2]; Hbzac=phenylbutane-1,3-dione).[23–25]
One characteristic of the above-mentioned complexes is
ACHTUGNTRNEU(NGN Cl2)ACHTUNGTRENNUNG ACHTUNGTRENNUNG(bzac)2-
(Cp2)])[18–22] or budotitane ([Ti
AHCTUNGTRENNUNG
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200902312, and contains synthet-
ic procedures for the preparation of ligands and titanium complexes;
detailed spectroscopic and analytical data for all new compounds;
their very fast rate of hydrolysis that leads to unidentified
products. In an aqueous environment, the loss of the labile
groups occurs within seconds, whereas the more inert lig-
1H NMR spectra of complexes [Ti
(OiPr)2], [Ti(PhFNMe (OiPr)2], [Ti(PhClNMe
2A
(OiPr)2] at their respective t1/2 values with overlaid spectra of the cor-
T
)2ACTHUNGTRENNUG ACTHUNGTRENNUNG
(OiPr)2], [Ti
AHCTUNGTRENNUNG
ands hydrolyze on a timescale of hours or days.[26,27] On the
A
E
)
U
A
)
G
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
one hand, this hampers the identification of the active spe-
cies as well as the biological target and the mechanism of
action. It was found that titanocene dichloride is enriched in
responding salans; details of the cytotoxicity assays and viability plots
in Hep G2 for all compounds; color graphics for Figures 2–8 and Fig-
ures 11–12.
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2775

areas near the nuclear chromatin and covalently binds to
DNA and inhibits DNA synthesis.[28–31] Titanocene dichlor-
moieties. Finally, employing four different titanium alkox-
ides ([Ti(OEt)4], [Ti(OiPr)4], [Ti(OnBu)4], [Ti(OtBu)4]) in
the complex formation allowed us to fine-tune the steric
demand of the titanium-bound labile ligands as well.
G
E
R
ACHTUNGTRENNUNG
ide was also reported to inhibit human topo
A
On the other hand, it is assumed that some ligand lability is
necessary for the biological effect and that ligand hydrolysis
leads to formation of the active species.
The o,p-disubstituted salans were accessible by a Mannich
condensation. Simple reflux of the appropriate substituted
phenol, N,N’-dialkylethylenediamine, and formaldehyde in
Concerning the cellular uptake, the role of ligand lability
is controversial. Although it is necessary to strip off the lig-
methanol gave ligands (Phb–fNa–c
) in moderate yield as
2
ands when TiIV is trafficked via transferrin,[33–37] a route via
shown in Scheme 1.[43–45] In the case of aryl unsubstituted
ligand (PhHNMe)2, salicylaldehyde was reacted with ethylene-
albumin could leave the complexes intact. An adduct of the
complex stabilized by albumin might enter the cell, which
would promote a more active role for these drugs in con-
trast to the prodrug role proposed for the transferrin deliv-
ery mechanism.[38]
Consistent with these findings, in 2007 two methyl substi-
tuted titanium salan complexes were reported to be cytotox-
ic independent of transferrin.[39] By using a small collection
of six alkyl-substituted titanium salan complexes, Tshuva
and co-workers showed an influence due to the steric
demand of the ligand and proposed a connection between
the hydrolytic behavior of these complexes and their biolog-
ical activity.[40–42] Exploring this third class of titanium com-
plexes with antitumor activity, we could demonstrate that
complexes of halogen-substituted salans in particular reveal
promising biological properties. Their IC50 values are com-
parable to cisplatin and, in contrast to their alkyl-substituted
congeners, they almost exclusively induce apoptotic cell
death.[43]
Scheme 1. Synthesis of salan ligands by Mannich condensation.[43–45]
[*] Salans with R1 =H were not accessible by this synthesis and were in-
stead synthesized according to a published procedure.[46]
diamine, then the resulting imine was reduced with sodium
borohydride and subsequently methylated by reductive ami-
nation to give the desired salan.[46] Finally, metalation of
Herein we describe the synthesis, biological evaluation,
and hydrolytic behavior of a library of more than 40 alkyl-
and halogen-substituted titanium salan complexes. The ob-
tained data are discussed with
these salans with different titanium alkoxides Ti
shown in Scheme 2, gave titanium(IV) salan complexes
(Oa–d)4, as
ACHTUNGTRENNUNG
respect to structure–activity re-
lations and the influence of hy-
drolysis on cytotoxicity.
Results and Discussion
To elucidate the impact of both
steric and electronic properties
of the ligands on the antitumor
activity and also the hydrolysis
Scheme 2. Synthesis of titanium salan complexes by metalation with different titanium alkoxides.[43, 46,47]
behavior, the synthesis of
a
small library of more than 40
different salan complexes was envisaged. For a systematic
study, the influence of substituents on the aromatic rings
and at the bridging nitrogen atoms of the salan as well as
the role of the labile alkoxy ligands had to be addressed in-
dependently. To differentiate between electronic and steric
influences, we decided to utilize a series of halogen-substi-
tuted phenols in the ligand preparation alongside different
alkyl-substituted ones. Coupling these phenols with three
different amino linkers allowed the systematic construction
of amino-phenolato-ligands (salans). Thus, those ligands fea-
ture different steric bulk at the bridging nitrogen atoms and
cover a broad range of electronic influences at the phenol
[Ti(Pha–fNa–c (Oa–d)2] in nearly quantitative yield as single
)2ACHTUGNRTENNNUG
(racemic) isomers.[43,47] To remove trace impurities, all com-
pounds were recrystallized at least once prior to cytotoxicity
studies. Purities were proven by combustion analysis.
All of the complexes described above were tested for an-
titumor behavior by using the well-established AlamarBlue
assay, which is known to be highly reproducible and more
sensitive than the MTT assay.[48] The cytotoxicity was studied
in two different cell lines: the human cervix carcinoma cell
line HeLa S3 and the Hep G2 cell line, an established
human hepatocarcinoma cell line with epithelial morpholo-
gy.[43] Cisplatin was used as a reference substance in all
2776
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
Table 1. Selected bond lengths [ꢂ] and angles [8] for a series of complexes with different alkoxides.
(PhMeNMe (OEt)2] (PhMeNMe (OnBu)2] (PhMeNMe (OtBu)2]
[Ti 2A [Ti 2A [Ti )2ACHTUNGTRENNUNG
assays. It showed an IC50 of
(1.2ꢁ0.4) mm in HeLa S3 and
(3.0ꢁ1.3) mm in Hep G2 cells.
G
)
N
G
)
E
G
ꢂ
O1 Ti1
1.8964(18)
1.8977(18)
1.8300(19)
1.8328(19)
2.342(2)
2.336(2)
104.60(9)
94.96(9)
92.50(8)
92.71(8)
95.97(8)
166.78(8)
164.54(8)
90.29(8)
88.41(8)
81.43(8)
90.20(8)
164.27(9)
80.72(8)
88.51(8)
75.43(8)
1.8937(13)
1.9086(13)
1.8209(13)
1.8226(13)
2.3392(16)
2.3214(14)
105.95(6)
97.03(6)
1.923(3)
1.903(3)
ꢂ
All IC50 values are given as O2 Ti1
ꢂ
O3 Ti1
1.780(2)
1.779(3)
2.371(3)
2.365(3)
mean values from at least three
independent experiments each
done in four replicates.
Selected complexes were sub-
ject to time-resolved hydrolysis
studies monitored by NMR
spectroscopy to assess their sta-
bility under biologically rele-
ꢂ
O4 Ti1
ꢂ
N1 Ti1
ꢂ
N2 Ti1
O3-Ti1-O4
O3-Ti1-O1
O4-Ti1-O1
O3-Ti1-O2
O4-Ti1-O2
O1-Ti1-O2
108.78(12)
95.90(11)
92.34(12)
93.18(11)
96.99(12)
164.16(11)
162.30(13)
88.50(12)
86.94(11)
80.57(11)
89.31(11)
161.01(12)
79.70(11)
87.48(11)
73.98(11)
93.58(6)
89.45(5)
96.06(6)
166.46(6)
164.25(6)
88.13(6)
88.92(5)
81.93(5)
90.36(6)
163.21(6)
80.21(5)
87.90(5)
76.24(5)
vant conditions. We were espe- O3-Ti1-N2
O4-Ti1-N2
cially interested in the influence
O1-Ti1-N2
of electronic features at the
O2-Ti1-N2
salan ligand (halogen vs. alkyl)
O3-Ti1-N1
on the speed and product for-
mation of hydrolysis. Further-
more, the impact of steric
crowding around the metal
center and its effect on hydroly-
O4-Ti1-N1
O1-Ti1-N1
O2-Ti1-N1
N2-Ti1-N1
sis was paid special attention by utilizing alkoxides with dif-
fering steric demands.
As revealed by the H NMR spectra, the geometrical fea-
were refined as described above, which resulted in a higher
R-factor of 6.97. A summary of experimental crystal data
for all determined structures is given in the Experimental
Section (Table 9).
1
tures of the tetradentate [ONNO]-type ligands in all of our
complexes are in general very similar. The N-CH2CH2-N
methylene and the benzylic protons showed the familiar AB
pattern consistent with C2 symmetry and a fac–fac wrapping
mode.[49] To compare the influence of the two monodentate
ligands on complex geometry, a series of complexes with dif-
Titanium-bound alkoxides
Influence on cytotoxicity in alkylsalans: Cytotoxicity studies
done in the late 1980s by Kçpf-Maier et al. with titanocenes
with different labile ligand substitutions showed the negligi-
ble influence on cytotoxicity.[50] The question as to whether
the influence of labile ligands is more pronounced in the
case of titanium salan complexes still remains open. Recent
studies by Tshuva and co-workers that utilized catechol as a
bidentate substitute for the isopropoxy groups showed a
two-fold decrease in the cytotoxicity of the resulting com-
plex.[40] Although the observed effect is significant, an ex-
planation is not evident because not only is the steric influ-
ence altered, but also the exchange of the two monodentate
isopropoxy groups for one bidentate catecholato ligand
leads to a different complex geometry. In the starting com-
plex the isopropoxy groups are oriented in a cis fashion, but
the two phenolato substituents of the salan are now occupy-
ing their position, which results in a loss of symmetry. We
decided to tackle the question of the labile ligandsꢃ influ-
ence on cytotoxicity by a more systematic approach. There-
fore, in the first series of experiments the salan backbone
was kept unchanged and instead the labile alkoxy ligands
were permutated, in the hopes of keeping the complex sym-
metry unchanged.
ferent labile alkoxides ([TiACTHNUTRGENNUG )2ACHTUNGTRENNUNG
(PhMeNMe (Oa,c,d)2]) was closely
investigated. Suitable crystals for X-ray structure determina-
tion were either grown by slow diffusion of hexane into a sa-
turated solution in toluene ([Ti
ly from hexane solutions at room temperature by slow evap-
oration ([Ti 2A 2A
(PhMeNMe (OEt)2], [Ti(PhMeNMe (OtBu)2]). The
(PhMeNMe (OnBu)2]) or direct-
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
A
)
E
N
) CHTUNGTRENNUNG
X-ray structures of this complex series again revealed the
typical geometry of C2-symmetrical complexes with very
similar structural features. With the labile alkoxy ligands
bound at the equatorial plane in a cis fashion and both phe-
nolates occupying the bis-trans-axial positions, the com-
plexes exhibit only minor differences in bond length and
ꢂ
angles. With a slightly shorter Ti alkoxide bond (1.78 ꢂ)
compared with complexes with unbranched alkoxides, [Ti-
ACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(PhMeNMe (OtBu)2] is different to the others members of this
ꢂ
series (1.82–1.83 ꢂ). This is compensated by a longer Ti N
bond of 2.37 ꢂ as compared with 2.32–2.34 ꢂ for the other
series members, which results in a smaller phenolate-Ti-phe-
nolate (O-Ti-O) angle of 164.168 compared with 166.46-
166.788 for the other members of this class of complexes
(Table 1, Figure 1). For [TiACTHNUTRGENNUG )2ACHTUNGTRENNUNG
(PhMeNMe (OEt)2], the H atoms
of all methyl groups were found to be disordered by a rota-
To keep the probable steric influence as low as possible,
the first set of salan complexes evaluated in the AlamarBlue
assay consisted of the aryl-unsubstituted N-methylated lig-
tion of 608 and were, therefore, refined to fix the occupan-
cies at 0.5 each. In [TiACTHNURGTENNUG )2ACHTUNGTRENNUNG
(PhMeNMe (OtBu)2], one of the tert-
butoxy groups was found to be disordered over two sites
with occupancies of 0.675 and 0.325. The methyl groups of
the second tert-butoxy group showed rotational disorder and
A
)
metalated with titanium alkoxides
(Oa–c)4. The dose-response curves obtained with the
HeLa S3 cell line for these three complexes ([Ti
(PhHNMe)2-
2
TiACHTUGNTRENNNUG
AHCTUNGTRENNUNG
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2777

T. Huhn et al.
Figure 2. Viability of Hela S3 cells after incubation for 48 h with top: [Ti-
(PhHNMe
)
T
N
)
G
G
(OnBu)2] (
)2ACHTUNRGTNEGNU )
!
*
*
(PhMeNMe
)
U
), [Ti
G
)
E
(
*
), [Ti-
)
U
were measured by using an AlamarBlue assay and show the dependence
on the steric bulk of labile ligands in complexes with an unchanged
[ONNO]-backbone.
Figure 1. ORTEP diagram of the molecular structures of complexes [Ti-
A
)
)
2A
2A
(OtBu)2] (bottom). Displacement ellipsoids are drawn at the
E
CHTUNGTRENNUNG
In the third set of the alkyl series, the 2,4-di-tert-butyl-sub-
50% probability level. Main occupancy is shown for disordered groups;
hydrogen atoms are omitted for clarity.
stituted salan was employed. Its titanium isopropoxy com-
plex, [Ti
(PhtBuNMe (OiPr)2], is known to be nontoxic. Based
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
on the above results of the [TiACHTNUGTRENNNUG )2ACHTUNGTRENNUNG
(PhMeNMe (Oa–d)2] complexes,
A
we initially hoped to regain cytotoxicity by reducing the
steric demand of the employed alkoxide. But even when the
smaller, non-branched ethoxide was used as the labile
E
) CHTUNGTRENNUNG
ligand, the resulting [Ti
nontoxic. An attempt to synthesize [Ti
failed, possibly due to steric interference of the labile lig-
ands and the salan backbone.
Table 2 summarizes the IC50 values of these titanium salan
(PhtBuNMe
ACHUTGTNRENNUG )2ACHTUNGTRENNUNG
(OEt)2] complex remained
G
)
G
A
)2ACHTUNGTRENNUNG
Unexpectedly, we did observe a clear influence due to the
labile ligands when testing the second set of four complexes
AHCTUNGTRENNUNG
([TiACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(PhMeNMe (Oa–d)2]). With methyl groups at positions two
and four of the phenolato substituents of the salan back-
bone, they feature slightly increased steric crowding around
the metal center and an enlarged overall steric demand.
This influence on cytotoxicity is especially pronounced when
comparing the bulky tert-butoxy-substituted [Ti
ACHTUNGTRENNUNG
complexes in both HeLa S3 and Hep G2 cells.
Influence on cytotoxicity in halogen-substituted salans: It is
known that halogen-aryl-substituted salan complexes of tita-
nium show a favorable biological profile.[43] For this reason
as well as the gradual change in electronic and steric influ-
ences, a group of complexes that have fluoro, chloro and
bromo substituents at the aromatic moiety was examined.
(PhMeNMe)2-
ACHTUNGTRENNUNG
thy 10-fold decrease in cytotoxicity was observed (Fig-
ure 2b).
2778
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
Table 2. IC50 [mm] values in Hela S3 and Hep G2 cells after incubation
for 48 h with complexes [Ti
AlamarBlue assay.
(Pha–cNMe (Oa–d)2], as measured by using an
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
Hela S3
R1 =Me R1 =tBu R1 =H
Hep G2
R1 =H
R1 =Me R1 =tBu
R3 =Et
2.5ꢁ0.4 2.0ꢁ0.3 nontoxic 5.2ꢁ1.3 1.6ꢁ0.3 nontoxic
R3 =iPr 4.5ꢁ1.3 2.3ꢁ0.1 nontoxic 5.5ꢁ1.9 2.1ꢁ0.1 nontoxic
R3 =nBu 2.7ꢁ0.7 5.6ꢁ0.5 nontoxic 4.4ꢁ2.6 5.2ꢁ0.5 nontoxic
R3 =tBu
–
19.8ꢁ2.4
–
–
26.9ꢁ8.6
–
Again, changing the alkoxide while keeping the backbone
unchanged should reveal the influence of steric interference
on the IC50. This may offer the possibility of fine-tuning the
biological properties of such complexes.
In the case of the fluoro-substituted salan [Ti
(Oa–c)2] complexes, changing the alkoxy ligands had virtually
no effect on the activity (Figure 3a) and resembles the [Ti-
(PhHNMe (Oa–c)2] case in this respect. All three complexes
2A
(PhFNMe
2
ACHTUNGTRENNUNG
A
) CHTUNGTRENNUNG
had their IC50 value within a very narrow margin of 1.3 to
1.6 mm.
In the chloro-substituted salan set of complexes, a strong
influence on the observed cytotoxicity due to the bulkiness
of the labile alkoxy ligand was observed (Figure 3b). The
complex with an ethoxy group had a more than 40-fold
higher cytotoxicity in HeLa S3 cells as compared with its
tert-butoxy congener, which shows that the effect was even
more pronounced than in the case of complexes [Ti-
ACHTUNGTRENNUNG ) CHTUNGTRENNUNG
(PhMeNMe (Oa–d)2].
2A
For the most sterically demanding bromo-substituted
salan backbone (PhBrNMe)2, a tremendous decrease in cyto-
toxicity was observed for complexes with the more sterically
demanding alkoxides (Figure 3c), with the tert-butoxide
complex being completely nontoxic.
The IC50 values of all halogen-bearing titanium salan com-
plexes [TiACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(Phd–fNMe (Oa–d)2] are summarized in Table 3.
Table 3. IC50 [mm] values obtained for the [TiACTHNUGTRENNGU )2ACHTUNGTRENNNUG
(Phd–fNMe (Oa–
d)2]complexes by using the AlamarBlue assay with Hela S3 and Hep G2
cells. The cells were incubated with the complexes for 48 h prior mea-
surement.
Figure 3. Viability of Hela S3 cells after incubation for 48 h with top: [Ti-
(PhFNMe (OEt)2] ( ), [Ti(PhFNMe (OiPr)2] ( ), [Ti(PhFNMe
*
ACTHNUGTERNNUNG )2AHCTNUGTRENNUGN AHCTNUGRTENNGNU ) CTHUGNTNERNUNG ACHTUTGNRENNUNG ) CTHUNGTRENNNUG
2A
2A
(OiPr)2] ( ), [Ti
(OtBu)2] ( ); and bottom: [Ti(PhBrNMe
) CHTUNGTRENNUNG
!
*
*
*
middle: [Ti
(PhClNMe
)
G
(PhClNMe
)
E
ACHTUNGTRENNUNG
(OnBu)2] ( ), [Ti
(PhClNMe
A
!
(
(PhBrNMe (OiPr)2] ( ), [Ti(PhBrNMe
*
), [TiACHUTNRGENNGU )2AHCTNUGTRENUNNG ACHTUNEGRTNNGNU )2ACTHUNGTRENNNUG
!
*
curves were obtained by using an AlamarBlue assay.
To elucidate the influence of transferrin on the cytotoxici-
ty of the halogen-substituted salan complexes, human apo-
transferrin was added to the medium prior to incubation of
selected complexes. For alkyl-substituted salans, it has al-
ready been shown that the cell penetration mechanism is in-
dependent of transferrin.[40] Remarkably, even though the
mechanism of action seems to be different for halogen-sub-
Hela S3
R1 =F R1 =Cl
Hep G2
R1 =F R1 =Cl
R1 =Br
R1 =Br
R3 =Et 1.3ꢁ0.4 1.0ꢁ0.2 1.3ꢁ0.4 1.8ꢁ0.3 1.9ꢁ0.3
1.5ꢁ0.6
40ꢁ6
R3 =iPr 1.6ꢁ0.1 5.3ꢁ0.2 13ꢁ1
2.2ꢁ0.1 4.0ꢁ0.2
R3 =nBu 1.4ꢁ0.6 8.4ꢁ2.0 45.5ꢁ5.8 1.5ꢁ0.4 14.6ꢁ6.0 83.6ꢁ33.5
R3 =tBu
–
41.4ꢁ6.6 nontoxic
–
53.2ꢁ14.5 nontoxic
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2779

T. Huhn et al.
stituted complexes,[43] their cytotoxicity was not altered upon
addition of human apo-transferrin (Figure 4).
*
Figure 5. Viability of cisplatin-resistant (MGH-U1; ) and cisplatin-hy-
(PhFNMe)2-
*
persensitive (833K; ) cells after incubation for 48 h with [Ti
AHCTUNGTRENNUNG
alkoxide is the most toxic one. Moreover, it seems that
these two factors are not independent from each other. The
cytotoxicity of the sterically more demanding complexes of
(PhCl,BrNMe
) is more strongly affected by an additional
2
bulky labile ligand than is the complex of (PhFNMe)2 with
only small fluoro substituents at the salan backbone. It is
evident that the influence of the steric demand of the labile
alkoxy ligands increases with the size of the halogen at the
backbone.
This result also gives an explanation as to why the cyto-
toxicity of [TiACTHNUTRGENNUG )2ACHTUNGTRENNUNG
(PhMeNMe (Oa–d)2] is dependent on the alkoxy
ligands in the earlier set of alkyl-substituted titanium salan
complexes, whereas it is not for the unsubstituted [Ti-
ACHUTNGRENNUG ) CHTUNGTRENNUNG
(PhHNMe (Oa–c)2] complexes.
It seems that cytotoxicity does not depend solely on
2A
Figure 4. Viability of Hela S3 cells (top) and Hep G2 cells (bottom) after
incubation for 48 h with [Ti
(PhFNMe
)
(OiPr)2] ( ) and [Ti
(PhClNMe
)
(OiPr)2]
*
!
(
) with (c) and without (b) added human apo-transferrin
either the steric demand of the salan backbone or the bulki-
ness of the alkoxide used. In fact, quite the contrary seems
to hold true. Complexes with small substituents at the aro-
(10 mgmLꢂ1).[51] Dose-response curves were obtained by using an Ala-
marBlue assay.
matic rings ([TiACTHNUTRGENNUG )2ACHTUNGTRENNUNG
(PhH,FNMe (Oa–d)2]) show no decreased IC50
Concerning the mechanism of action of salan complexes,
we were interested in whether they bear a resemblance to
the well-explored mechanism of cisplatin. Therefore, [Ti-
even with the bulky tert-butoxide as the labile ligand. The
smallest increase in steric demand on the side of the back-
bone leads to a stepwise decreased cytotoxicity of the com-
A
)2A
(OiPr)2] was incubated with two different cell lines,
plex ([Ti(PhMe,Cl,BrNMe (Oa–d)2]), which depends on the bulk-
)2ACHTUGNRTENNNUG
one of which was cisplatin resistant (human bladder carcino-
ma, MGH-U1) and one of which was cisplatin hypersensi-
tive (testicular germ-cell tumor, 833K). With sub-micromo-
lar IC50 values of 0.5ꢁ0.1 mm for MGH-U1 and 0.4ꢁ0.1 mm
for 833K, both cell lines showed a comparable sensitivity
(Figure 5). This is an indication that titanium salan com-
plexes and cisplatin differ in their mode of action.
iness of the alkoxide used. Being independent from the
nature of the employed substituent, alkyl or halogen, this
effect appears not to be of electronic origin but rather based
on the sheer size of the involved groups. Seemingly, there is
something like an upper threshold, from where on the
whole complex loses its cytotoxicity. To verify this conclu-
sion, we decided to synthesize another series of complexes
with additional steric hindrance at the bridging nitrogen
atoms of the salan backbone.
The results of the halogen-substituted [Ti
ACHTUNGTERNNUNG )
2
(Oa–d)2] complexes with N-methylated salans show that both
the bulk of the salan ligands and the size of the labile li-
gands are of great importance for the obtained biological ac-
tivity.
Steric influence at the bridging nitrogen atoms
In each series of complexes in which the salan is fixed and
the alkoxide changed, the complex featuring the smallest
Cytotoxicity of N-ethylated complexes: Recently it has been
reported that in the case of a salan ligand with two methyl
2780
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
groups at each aromatic ring, exchange of the methyl at the
bridging nitrogen atoms by an ethyl group resulted in a
complete loss of toxicity of the corresponding bis(isopro-
ACHTUNGTRENNUNG
poxy)titanium complex.[40] Therefore, we were interested in
restoring the cytotoxicity of similar complexes just by less-
ening the steric demand of the alkoxide groups.
A first set of N-ethylated complexes was, therefore, pre-
pared by metalating the dimethyl salan ligand (PhMeNEt)2
with four different titanium alkoxides Ti
ent steric demands. In accordance with results published for
a very similar complex,[40] we found [Ti(PhMeNEt) (OiPr)2] to
2A
be nontoxic. As expected, in the case of complexes [Ti-
2A 2A
(PhMeNEt) (OnBu)2] and [Ti(PhMeNEt) (OtBu)2] cytotoxicity was
(Oa–d)4 with differ-
ACHTUNGTRENNUNG
A
CHTUNGTRENNUNG
A
N
G
CHTUNGTRENNUNG
also not observed, presumably due to the even bulkier labile
alkoxides adding to the overall size. In the much more inter-
esting case of the [TiACTHUNTRGENNUG 2ACHTUNGTRENNUGN
(PhMeNEt) (OEt)2] complex with reduced
steric demand from the alkoxide, we could indeed restore
cytotoxicity in both cell lines used. With an IC50 value of
ꢀ25 mm in HeLa S3 and Hep G2 cells it showed a mediocre
biological activity, being one order of magnitude less than
the most toxic complexes (Table 4).
Table 4. IC50 [mm] values obtained for the [TiACTHNUGTRENNGU )2ACHTUNGTRENNNUG
(Phd–fNEt (Oa–d)2] complexes
in Hela S3 and Hep G2 cells. The cells were incubated with the complexes
for 48 h prior to measurement.
Figure 6. Viability of Hela S3 cells after incubation for 48 h with fluoro-
*
aryl N-ethyl-bridged complexes [Ti
(PhFNEt)
(Oa–c)2] (top;
: [Ti-
Hela S3
Hep G2
)2ACTHNGUTERNNUG
(OnBu)2]) and
R1 =Me R1 =F R1 =Cl
R1 =Me R1 =F R1 =Cl
!
*
: [Ti
G
)
R
G
)
U
R
G
) CHTUNGTRENNUNG
R3 =Et 25.0ꢁ6.7 2.7ꢁ0.6 2.9ꢁ0.8
24.4ꢁ8.7 4.9ꢁ0.9 3.4ꢁ0.9
nontoxic 2.5ꢁ1.1 14.3ꢁ4.0
(PhClNEt
)
E
(PhClNEt
)
N
A
CHTUNGTRENNUNG
!
*
: [Ti
*
R3 =iPr nontoxic 2.3ꢁ0.4 7.1ꢁ1.7
which demonstrate the dependence on the steric demand of the labile
ligand in complexes with an unchanged N-ethyl backbone. Dose-re-
sponse curves were obtained by using an AlamarBlue assay. The [Ti-
R3 =nBu nontoxic 4.9ꢁ0.9 239.0ꢁ41.1 nontoxic 5.7ꢁ1.6 256.6ꢁ82.7
R3 =tBu nontoxic
–
nontoxic
nontoxic
–
nontoxic
AHCTUNGTRENNUGN )2ACHTUNGTRENNUNG
(PhClNEt (OtBu)2] complex was nontoxic.
Encouraged by this finding, we turned our attention to-
wards the electron-deficient fluoro- and chloro-substituted
(OnBu)2] already showed a biological activity that was de-
ACHTUNGTRENNUNG
creased by more than two orders of magnitude. Finally, the
tert-butoxy complex is completely nontoxic. This effect is
even more pronounced than for the series of corresponding
complexes with N-ethyl groups [TiACHTNUTRGENNUG 2ACHTUNGTRENNUGN
(Phd,eNEt) (Oa–d)2]. As
shown in Figure 6a, all complexes resulting from metalation
of the sterically less demanding fluoro-substituted ligand
(PhFNEt)2 showed similar IC50 values in the range of 2.3 to
4.9 mm, even with the bulkiest titanium alkoxide. It seems
that the critical overall size for a significant decrease in tox-
icity cannot be reached within this specific set of ligands. A
completely different situation revealed a series of chloro-
N-methylated complexes [TiACTHNUTRGENNUG )2ACHTUNGTRENNUNG
(PhClNMe (Oa–d)2] (Table 3).
The observed dependence of cytotoxicity on the alkoxides
employed in the N-ethylated complexes [Ti(Phb,d,eNEt)2
AHCTUNGTRENNUNG
(Oa–d)2] and the intensification of this effect compared with
the corresponding N-methylated complexes [Ti(Phb,d,eNMe
)
2
AHCTUNGTRENNUNG
(Oa–d)2] supports our hypothesis: The combined steric hin-
substituted salan complexes, [Ti
G
N
drance of salan and labile ligands is crucial for the anti-
cancer activity of this class of titanium complexes.
dous effect on cytotoxicity that depended on the size of the
alkoxide used was observable (Figure 6b). With an IC50
On comparing both groups, we deduced the existence of a
lower steric threshold; below this threshold, alteration of
the overall steric demand either at the salan backbone or at
the labile ligands has only a minor influence on cytotoxicity.
This can be exemplified by varying the smallest of our com-
value of 2.9 mm, complex [TiACHTNUTRGENNUG 2ACHTUNGTRENNUGN
(PhClNEt) (OEt)2] with a labile
ethoxy ligand is highly toxic in HeLa S3 cells. With the in-
creasing size of the utilized alkoxide, the respective com-
plexes showed a stepwise decrease in cytotoxicity in both
cell lines (Table 4). Whereas the isopropoxy complex is
plexes ([Ti
backbone, either at the aromatic rings (F vs. Cl) or at the
(PhFNMe (OEt)2]). Minor variations of its salan
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
slightly less toxic with an IC50 value of 7.1 mm, [TiACTHNUTRGNEUNG
(PhClNEt)2-
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2781

T. Huhn et al.
bridging nitrogens (Me vs. Et) has virtually no influence on
measured cytotoxicity. The same is true for variations of the
labile ligands.
In all the above cases, a further reduction in size does not
enhance cytotoxicity. The minimum IC50 values for such
salan complexes seem to be in the region of 1 mm. When ex-
ceeding the lower steric threshold, even minor changes in
steric demand have a tremendous effect on biological activi-
ty. This can be impressively demonstrated by comparing
complexes [TiACHTUNGTRENNUNG 2ACHUTNGTNERNUG ACHTUNGERTNNUNG 2ACHTUNGTRENNGUN
(PhFNEt) (OnBu)2] and [Ti(PhClNEt) (OnBu)2].
Replacing the small fluoro by a larger chloro substituent led
to a 50-fold decrease in cytotoxicity. Exchanging methyl for
ethyl at the bridging nitrogens of the highly toxic [Ti-
A
)2ACHTUNGTRENNUNG
(OiPr)2] has an even more drastic effect because it
ACHUTNGRENNUG 2CAHTUNGTRENNUGN
(PhMeNEt) (OiPr)2].
Figure 7. Viability of Hela S3 cells after incubation for 48 h with [Ti-
Concerning the influence of the alkoxy ligands on cyto-
(PhFNMe (OEt)2] (PhFNEt (OEt)2] (PhFNBn (OEt)2]
*
ACHUTGNERTNUNG ) HCUTNTGRENUNG ( ), [TiACHUTGNRENTNUG )2ACHTNUGRTENNUG ( ), and [TiAHCUTNGERTNNUNG )2ACHTUNGTRENNUNG
2A
!
toxicity, a simple change from ethoxy to n-butoxy led to a
*
(
), showing the dependence on steric demand of the residue at the
nearly 100-fold reduced toxicity for complexes [Ti
(Oa,c)2].
Toxicity is completely lost when a certain overall size is
(PhClNEt)2-
ACHTUNGTRENNUNG
bridging nitrogens. Dose-response curves were obtained by using an Ala-
marBlue assay.
ACHTUNGTRENNUNG
exceeded. For the tert-butoxy-substituted salan complexes,
this upper steric threshold is already reached; even with
small labile ligands, such as ethoxy ([Ti
no toxicity was achieved.
An analysis of all the measured biological data and com-
parison with the corresponding chemical structures elucidat-
ed interesting structure–activity relationships, that is, the ob-
tained biological activity is not only dependent on the bulk
of the salan ligands alone, but also on the steric hindrance
of the labile ligands. Moreover, it seems that these two fac-
tors are not independent from each other because for bulky
salan ligands an additional sterically hindered labile ligand
is not tolerated. In fact, it leads to a strong decrease in or
even a complete loss of cytotoxicity, whereas the same ex-
change of labile ligands does not affect the cytotoxicity of
complexes with less sterically demanding salans.
(PhtBuNMe (OEt)2]),
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
Cytotoxicity of N-benzylated complexes: To explore the
utmost tolerable steric demand at the bridging nitrogen
atoms, we decided to introduce N-benzyl groups. For ease of
comparison, we used the same phenols as in the N-Et case
for the Mannich coupling. When tested for their cytotoxicity,
only the least sterically hindered complex [Ti
(OEt)2] was cytotoxic at all, with an IC50 value of ꢀ10 mm in
both cell lines (Table 5).
(PhFNBn)2-
ACHTUNGTRENNUNG
ACHTUNGTRENNUNG
The impact of this minute interplay of salan and labile
ligand on cytotoxicity pointed us to the conclusion that the
complex has to be intact at the particular moment when the
steric bulk determines the cytotoxicity.
Table 5. IC50 [mm] values obtained for N-benzylated complexes measured
in Hela S3 and Hep G2 cells after incubation for 48 h. Only the least
bulky complex, [TiACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(PhFNBn (OEt)2], displayed any cytotoxicity.
In contrast to titanocene complexes, not much is known
about the specific mechanism of action of titanium salan
complexes. Therefore, the question of at which point of in-
teraction with its biological target the complex size matters
still remains unanswered. In principle, the steric shielding
around the titanium center influences stability, either against
water alone or in conjunction with biomolecules, that is,
transferrin or albumin. Considering the reaction with water,
one might argue that the reduced or even absent toxicity of
the bulky complexes might be explained by the following
two scenarios: If the complex itself is toxic, then a higher
steric demand at the titanium center might lead to a faster
hydrolysis, thus generating nontoxic byproducts (detoxifica-
tion mechanism). If the starting complex acts like a pro-
drug and is activated by hydrolysis, slower activation should
lead to diminished or even absent toxicity (activation mech-
anism).
Hela S3
R1 =Me R1 =F
Hep G2
R1 =Me R1 =F
R1 =Cl
R1 =Cl
R3 =Et
nontoxic 10.8ꢁ3.4 nontoxic nontoxic 9.86ꢁ5.7 nontoxic
R3 =iPr nontoxic nontoxic nontoxic nontoxic nontoxic nontoxic
R3 =nBu nontoxic
R3 =tBu
–
–
nontoxic nontoxic
nontoxic
–
–
nontoxic
nontoxic
–
–
As demonstrated above, the overall steric demand is the
most important factor in determining biological activity. In a
comparison of [TiACHTUNGTRENNUNG 2ACHTUNGTRENNUGN
(PhFNBn) (OEt)2] (the only N-benzylated
complex to show any cytotoxicity) with its N-Me and N-Et
congeners, the dose-response chart shows a stepwise de-
crease in cytotoxicity that again depends on the overall size
of the respective complex (Figure 7).
The decreased cytotoxicity of sterically demanding com-
plexes might originate from inhibition of a specific interac-
tion with biomolecules. This may affect their transport into
2782
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
the cell, either by active transport via transferrin or by
simple passive diffusion, the transfer to or the interaction
with the cellular target. Experiments with added transferrin
showed no significant influence on cytotoxicity, making an
active cellular uptake via transferrin doubtable. If the
uptake is based on diffusion, one might expect a faster
transport of complexes featuring sterically demanding and,
therefore, lipophilic side groups (tBu, N-Bn, O-tBu). Be-
cause those complexes show a diminished cytotoxicity in
contrast to the ones with smaller and, therefore, more hy-
drophilic groups (F, Me, N-Me, O-Et), transport phenomena
do not explain the size-dependent alteration of cytotoxicity.
drolytic behavior of the halogen-substituted complexes [Ti-
AHCTUNGTRENNUGN )2ACHTUNGTRENNUNG
(Phd–fNMe (OiPr)2] because these show favorable biological
properties. In strong contrast to the alkylated complexes,
they induce cell death by a purely apoptotic pathway. This
might be due to a difference in hydrolysis, in respect to the
rate or obtained products. For comparison, the highly cyto-
toxic alkyl complex [Ti
N
(OiPr)2] and the nontoxic
) CHTUNGTRENNUNG
tert-butyl complex [Ti
A
(Figure 8).
Comparison of the hydrolytic half-life (t1/2) of the investi-
gated complexes (Table 6) revealed an interesting correla-
tion: Within the alkyl-containing complexes [TiACTHNUTRGNEUNG
(Phb,cNMe)2-
Hydrolysis—Activation or deactivation mechanism? To elu-
cidate whether the correlation of cytotoxicity and steric
demand is merely based on an altered hydrolysis or depends
on the interaction with biomolecules, we investigated the
stability of several complexes upon the addition of a certain
amount of water. We were interested in the influence the
complex structure exerts on the speed of hydrolysis because
this determines, besides the solubility, the applicability of
such complexes in aqueous media. A systematic investiga-
tion of the life-time of such complexes under hydrolytic con-
ditions with respect to their structure might, therefore, help
to find substances with an improved pharmacological pro-
file. Given that the fast formation of hydrolysis products has
hampered mechanistic research for all titanium complexes
found so far, stable and at the same time highly cytotoxic
complexes would be valuable tools.
On the other hand, investigation of the hydrolysis process
and the formed products might give insights into the role of
hydrolysis in biological systems. Here it is interesting to ask
whether hydrolysis is an activating or deactivating process.
To obtain not only the rate of hydrolysis but also structural
information about the products formed in aqueous media,
we decided to follow the hydrolysis of our complexes by
using time-resolved 1H NMR spectroscopy in [D8]THF/
D2O.[40]
The experiments were conducted in a mixture of [D8]THF
(95%), D2O (4.8%), and DMSO (0.2%) at 378C. To ana-
lyze the data, the decrease in at least two distinct signals of
the titanium-bound salan backbone and the increase in the
evolving signals of the free alkoxy ligands were monitored.
The integrals were then normalized by using the DMSO
signal as the internal standard and plotted against the hy-
drolysis time. Control measurements done in the absence of
DMSO showed no significant differences in the hydrolysis
rate and products formed. A complexation of the titanium
center by the added standard could therefore be excluded.
Keeping in mind that hydrolysis under in vivo conditions
follows different rules, our impetus was merely to gather rel-
ative rates of hydrolysis. This allows us to determine which
of our complexes is more and which is less stable in compar-
ison to each other under given conditions.
Table 6. Half-life of complexes [Ti
conditions determined by time-resolved 1H NMR spectroscopy at 378C.
(Phb–fNMe (OiPr)2] under hydrolytic
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
Complex
t1/2 [h]
Complex
t1/2 [h]
[Ti
[Ti
N
)
)
G
10
2
[Ti
[Ti
[Ti
G
)
)
)
G
6
108
>115
N
E
N
AHCTUNGTRENNUNG
A
) CHTUNGTRENNUNG
A
) CHTUNGTRENNUNG
as in the tert-butyl case half of the complex had already de-
composed 2 h after the initial water addition.
Under the chosen hydrolytic conditions, the bromo-substi-
tuted salan complex [Ti
sive t1/2 value of more than 115 h. It outruns its fluoro coun-
terpart [Ti 2A
(PhFNMe (OiPr)2], which has a t1/2 value of only
(PhBrNMe (OiPr)2] showed an impres-
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
A
) CHTUNGTRENNUNG
6 h, by a 20-fold increased hydrolytic stability. Unfortunate-
ly, this enormously stable complex displays only moderate
biological activity. Surprisingly, the chloro-substituted salan
complex showed a rather similar hydrolytic stability with a
t1/2 value of 108 h. This is remarkable because even though
the stability is in the same range as the bromo counterpart,
the biological activity is not. Compared with the most stable
cytotoxic complex of this class known so far ([Ti
(PhMeNMe)2-
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
ty that was more than 10 times higher. A rather high toxicity
(IC50 ꢀ5 mm) and the fact that this complex exclusively in-
AHCTUNGTRENNUNG
duces apoptosis[43] makes it one of the most interesting sub-
stances with respect to mechanistic studies, and a promising
candidate for further biological studies in the field of anti-
cancer titanium complexes.
It has been observed by Peri et al. that upon hydrolysis
[Ti
(PhtBuNMe (OiPr)2] releases its salan ligand as well as the
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
labile alkoxides. Hydrolysis of [TiACTHNUGTRENNUNG )2ACHTUNGTRENNUNG
(PhMeNMe (OiPr)2] results
in the formation of new species that possibly contain more
than one titanium center. Careful examination of our data
of the halogen-substituted complexes showed no liberation
1
of free salan ligand upon hydrolysis. Instead, the H NMR
In a first set of experiments, the influence of alterations at
the aromatic moieties on the hydrolysis of our complexes
was investigated. We were especially interested in the hy-
spectra showed the formation of new species with signals
that indicated lower symmetry compared with the starting
complex.[52]
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2783

T. Huhn et al.
(PhtBuNMe
)
(OiPr)2] ( ), [Ti
E
)
(OiPr)2] ( ), [Ti
E
)
G
(PhClNMe
)
G
~
&
^
^
Figure 8. Plots of hydrolysis vs. reaction time for [Ti
[Ti
(PhBrNMe
)
(OiPr)2] ( ). Data acquired by time-resolved H NMR spectroscopy at 378C. Data gathered by monitoring the decrease in isolated signals of
1
~
the titanium-bound salan backbone and the increase in the signals from the liberated alkoxy ligands. Integrals are normalized against the internal stan-
dard.
(PhFNMe
(PhClNMe
)
2A
(OiPr)2]
It is worth noting that X-ray structure determination re-
vealed remarkably little difference between the chloro and
Table 7. Selected bond lengths [ꢂ] and angles [8] for [Ti
and [TiACHTNUTRGENNUG )2ACHTUNGTRENNUNG
(PhClNMe
(OiPr)2].
the fluoro complexes ([Ti(PhF,ClNMe (OiPr)2]). These com-
)2AHCUTNGERNNUG
[Ti
H
)
C
[TiC
)
CHTUNGTRENNUNG
plexes have comparable angles and distances around the ti-
ꢂ
Ti1 O1
1.9081(13)
1.9315(13)
1.8204(12)
1.7989(12)
2.3336(14)
2.3425(14)
106.04(6)
97.17(5)
92.82(5)
92.45(5)
95.72(5)
164.87(5)
163.06(5)
90.36(5)
86.01(5)
81.47(5)
1.904(2)
1.930(2)
1.8031(19)
1.7945(18)
2.339(2)
2.336(2)
106.76(9)
97.49(9)
93.19(9)
91.80(9)
96.36(9)
164.19(8)
162.89(8)
89.58(8)
86.40(8)
81.11(8)
88.15(8)
164.46(8)
80.23(8)
87.34(8)
76.04(8)
ꢂ
tanium center (Table 7, Figure 9) and differ by only a d=
Ti1 O2
1
ꢂ
Ti1 O3
0.3 ppm shift of the aromatic protons in the H NMR spec-
ꢂ
Ti1 O4
tra, so the reason for the enormous difference in hydrolytic
stability remains unclear.
ꢂ
Ti1 N1
ꢂ
Ti1 N2
In a second set of experiments, we investigated the influ-
ence of different alkoxy ligands on the rate of hydrolysis.
We supposed the alkoxides to have a rather big effect on
the hydrolysis behavior compared with the substituents at
the aromatic rings because they are in the vicinity of the
O4-Ti1-O3
O4-Ti1-O1
O3-Ti1-O1
O4-Ti1-O2
O3-Ti1-O2
O1-Ti1-O2
O4-Ti1-N1
O3-Ti1-N1
O1-Ti1-N1
O2-Ti1-N1
O4-Ti1-N2
O3-Ti1-N2
O1-Ti1-N2
O2-Ti1-N2
N1-Ti1-N2
metal center. Therefore, complexes [TiACHTNURTGENNUG )2ACHTUNGTRENNUNG
(PhMeNMe (Oa–d)2]
were exposed to the same hydrolysis conditions described
above.
Keeping the different cytotoxicities of these complexes in
mind, we were surprised to find that all four complexes re-
sulted in the formation of the same degradation products
upon hydrolysis (Figure 10). Interestingly, virtually no free
ligand was observed (Figure 11).
87.87(5)
165.49(5)
81.13(5)
87.61(5)
76.14(5)
1
Time-resolved H NMR spectroscopy revealed that in the
process of hydrolysis the distribution of the evolving prod-
ucts does not change over time. Figure 11 shows NMR spec-
tra of the aromatic region of [Ti
at different times.
To estimate the influence of the different labile ligands on
the half-life of the complexes, the change in intensity of dis-
tinct signals was monitored. Signals were chosen so that
(PhMeNMe (OEt)2] recorded
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
2784
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
their integral ratios stayed constant relative to each other
during the process of hydrolysis. This measure effectively ex-
cluded a possible overlay with signals of newly formed spe-
cies.
The decreasing integrals of both the titanium-bound alk-
oxides and salan backbone and the increase in the evolving
signals of the free alkoxy ligands were then normalized by
using the DMSO signal as the internal standard and plotted
against the hydrolysis time. Results for complexes [Ti-
A
)2ACTHNGUTERNNUG
(Oa–d)2] are given in Figure 12.
dependence on the steric demand of the labile ligand was
observed (Table 8). In particular, complexes with alkoxides
branched at C1, such as isoproACTHNUTRGENUGNpanolate and tert-butanolate,
displayed significantly faster hydrolysis than complexes with
linear alkoxides, such as ethanolate or n-butanolate. When
the cytotoxicity of the two complexes with branched alkox-
ides was compared, a decrease of one order of magnitude
was observed despite very similar rates of hydrolysis. For
complexes with similar IC50 values, the difference in rate of
hydrolysis is, in the case of [Ti
(PhMeNMe (OiPr)2], quite pronounced.
2A
However, the difference in the rate of hydrolysis between
[Ti 2A
(PhMeNMe (OiPr)2] and its tert-butoxy counterpart is
(PhMeNMe (OEt)2] and [Ti-
ACHTUNGTRENNUGN )2ACHTUNGTRENNUNG
N
) CHTUNGTRENNUNG
Figure 9. ORTEP diagrams of the molecular structures of [TiACTHNUTRGNEUNG
(PhFNMe)2-
ACHTUNGTRENNUNG ACHTUNGTRENNUNG )2ACHTNUGTRENNUGN
(OiPr)2] (top) and [Ti(PhClNMe (OiPr)2] (bottom). Displacement ellipsoids
G
) CHTUNGTRENNUNG
are drawn at the 50% probability level. Hydrogen atoms are omitted for
clarity.
ACHTNUTRGNEUNG )2ACHTUNGRETNNUGN CAHTUTNGRENNUG )2CAHTUNGTNERNUGN AHCTUNGTRENNUNG )2ACHTUNGTRENNUNG
Figure 10. Superimposed 1H NMR spectra of the aromatic region of [Ti(PhMeNMe (OEt)2] (c), [Ti(PhMeNMe (OiPr)2] (c), [Ti(PhMeNMe (OnBu)2]
(c), and [Ti
ACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(PhMeNMe (OtBu)2] (c) at their respective t1/2 values in the hydrolysis studies. Large signals arise from unhydrolysed complexes, small sig-
nals arise from hydrolysis products. No difference in the obtained products was observed.
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2785

T. Huhn et al.
1
Figure 11. Multiple H NMR spectra of the aromatic region of [Ti
G
)
G
position shows the unchanged distribution of hydrolysis products over time. The spectrum of the (PhMeNMe)2 ligand (bold) is also shown for comparison.
Table 8. Comparison of half-life and IC50 values for complexes [Ti
[Ti
(PhMeNMe (OEt)2] [Ti(PhMeNMe (OiPr)2] [Ti
18 10
G
)2ACHTUNTGRENNUNG
(Oa–d)2].[a]
ACTHNGUTRENNUG )ACHTUNGTRENNUNG
(OnBu)2] [Ti(PhMeNMe (OtBu)2]
rather small at 3 h. This does
not reflect the rather drastic
change in biological activity. In-
G
)
G
E
)
G
E
)
G
t1/2 [h]
16
7
terestingly, the X-ray structure Hela S3 IC50 [mm] 2.0ꢁ0.3
2.3ꢁ0.1
2.1ꢁ0.1
1
5.6ꢁ0.5
19.8ꢁ2.4
ACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(PhMeNMe (OtBu)2] showed
Hep G2 IC50 [mm] 1.6ꢁ0.3
5.2ꢁ0.5
26.9ꢁ8.6
of [Ti
alkoxy–titanium bond of
[a] Hydrolysis was followed by time-resolved H NMR spectroscopy at 378C, and cytotoxicities in Hela S3 and
Hep G2 cells were determined by using an AlamarBlue assay after incubation for 48 h.
a
1.78 ꢂ, which is shorter than
complexes with unbranched
alkoxides ([Ti
G
2A
(OnBu)2] 1.82 ꢂ). Nevertheless, this
1.83 ꢂ, [Ti
N
)
U
biomolecule. This does not rule out the possibility of hydro-
lytic deactivation of very unstable complexes.
ꢂ
stronger O Ti bond does not result in enhanced hydrolytic
stability. Furthermore, the formation of the same degrada-
tion products indicates a similar mechanism of hydrolysis.
Taking into account the fact that there is a difference of one
order of magnitude in the cytotoxicity of these complexes, it
seems implausible that hydrolysis functions as an activation
mechanism. Activation via hydrolysis would require ex-
tremely different hydrolysis rates, which were not observed.
It seems that complexes with low IC50 values do not share
a common hydrolytic profile and neither do nontoxic com-
plexes. A certain rate of hydrolysis neither grants nor cir-
cumvents cytotoxicity in our experiments. Taken together
with the earlier observed cross-relevance of the steric
demand of salan and alkoxy ligand on cytotoxicity, our hy-
drolysis studies therefore strongly indicate that the unhydro-
lysed complex is important for the first interaction with a
Seemingly, the determination of cytotoxicity by the steric
demand of whole complexes and by inactivation via hydrol-
ysis interferes with each other, but hydrolysis only plays a
minor part within our complexes. This might be due to a
slow hydrolysis compared with a much faster cellular
uptake.
Conclusion
Herein, we present a library of more than 40 titanium salan
complexes in which the steric demand of both the salan and
the labile alkoxy ligands has been systematically altered. To
enlighten the interconnection between steric demand, cyto-
toxicity, and hydrolytic behavior, all complexes were tested
2786
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
(PhMeNMe
)
N
(PhMeNMe
)
E
(PhMeNMe
)
G
(PhMeNMe
)
N
~
^
*
Figure 12. Plots of hydrolysis vs. reaction time for [Ti
(&). Data acquired by time-resolved NMR spectroscopy at 378C.
on their biological activity and the hydrolysis of a selected
set was followed by time-resolved H NMR spectroscopy. It
same hydrolysis products and displayed only slightly varying
t1/2 values. Moreover, the distribution of the evolving prod-
ucts does not change over time. Therefore, we concluded
that the active species is not a hydrolysis product. This sup-
ports our hypothesis of an important role for the intact com-
plex and leads to the conclusion that hydrolysis is not an ac-
tivation but a deactivation mechanism. The unhydrolyzed
complex seems to be important for the first interaction with
a biomolecule. Whether this is essential to transport the
complex into the cell, for the portage to the as-yet unknown
cellular target inside the cell, or the interaction at the site of
action remains an open question for further studies.
Because our results promote the idea of hydrolysis being
a detoxification pathway, we also screened our library for
bioactive and highly stable complexes. Hydrolysis and the
fast formation of unidentifiable byproducts have hampered
the investigation of the mechanism of action of all titanium
complexes found so far. Thus, such complexes might not
only have an improved pharmacological profile, but also
would be valuable tools for biochemical research. Therefore,
we were happy to find a chloro-substituted complex that dis-
played rather high toxicity (IC50 ꢀ5 mm) and surprisingly
high hydrolytic stability. With a hydrolytic half-life of 108 h,
this complex showed a stability over 10 times higher than
the most stable cytotoxic complex of this class known so far.
When combined with its enhanced selectivity in inducing
apoptosis,[43] the stability of this complex in aqueous media
1
was already known that bulkier salan ligands result in less
toxic complexes. Interestingly, we found that, in contrast to
studies on titanocene derivatives, the exchange of the labile
monodentate ligands can also have an enormous effect on
cytotoxicity. However, not only did we find that the size of
the alkoxy ligands is crucial for the cytotoxicity, but we also
showed an interdependence of the steric demand of salan
and alkoxy ligands. Very small salan ligands tolerated even
the bulkiest of the alkoxides, tert-butoxide, without any loss
of cytotoxicity. Complexes of medium-sized salans were ex-
tremely sensitive to even minor changes in the steric
demand of the alkoxides. In this case, the respective com-
plexes showed a stepwise decrease in cytotoxicity as the size
of the alkoxide was increased. Finally, complexes of very
bulky salans were nontoxic, even with the smallest labile
ligand. We concluded that the overall size of the complexes
determines their biological activity. This ligand–ligand inter-
dependence also implies that the complex should be intact
at the particular moment when the steric bulk determines
the cytotoxicity of the complex. This result is especially un-
expected because partial hydrolysis is believed to be an acti-
vation mechanism for many metal complexes. To investigate
whether the overall steric demand itself alters the hydrolytic
behavior of our complexes and, therefore, the cytotoxicity,
hydrolysis experiments were conducted. We showed that
complexes that only differed in their labile alkoxy ligands
despite having very different cytotoxicities resulted in the
makes it one of the most promising sub
of anticancer titanium complexes.
ACHTUNGTRENsNUGN tances in the field
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2787

T. Huhn et al.
bridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_
request/cif. A summary of the relevant crystallographic data for this
paper is shown in Table 9.
Experimental Section
[43,44]
Materials and Methods: Syntheses of ligands (Pha–fNMe
)
and
2
[45]
(Phb,eNBn
)
were performed as described previously. Complexes [Ti-
Hydrolysis experiments followed by NMR spectroscopy were conducted
in a [D8]THF (95%)/D2O (4.8%)/DMSO (0.2%) mixture. The DMSO
signal was used as the internal standard for calibrating the integrals.
Spectra were recorded in 4 min intervals. Elemental analyses were per-
formed at the microanalytical laboratory of the University of Konstanz.
Cytotoxicity was measured on HeLa S3 and Hep G2 cells by using the
AlamarBlue assay. The cells were cultivated at 378C under humidified
5% CO2 in Dulbeccoꢃs modified Eagle medium (Invitrogen) that con-
tained 10% foetal calf serum, 1% penicillin, and 1% streptomycin, and
were tested for mycoplasma infections by using a mycoplasma detection
kit (Roche Applied Science, Mannheim, Germany) prior to use.
2
ACHTUNGTRENNUNG )2ACHTUNGTRENNUNG
(Pha–fNMe (OiPr)2] were synthesized according to literature proce-
dures.[43, 46,47]
Titanium tetra(isopropoxide) (97%) and N,N’-dibenzylethylenediamine
(97%) were purchased from Aldrich. Titanium tetraACHTNUTRGNEUNG(ethoxide) (99%), ti-
tanium tetra(n-butoxide) (98%) and N,N’-diethylethylenediamine (96%)
were purchased from ABCR (Karlsruhe, Germany); titanium tetra(tert-
butoxide) (99%) and N,N’-dimethylethylenediamine (85%) were pur-
chased from Acros Organics (Geel, Belgium). Deuterated solvents where
purchased from Euriso-Top (Saarbrꢄcken, Germany) in sealed ampoules
and dried if necessary; other solvents were purified according to standard
procedures.[53] All experiments requiring a dry atmosphere were carried
out under nitrogen by using Schlenk techniques. NMR spectra were mea-
sured by using JEOL Eclipse 400 and Bruker Avance DRX 600 spec-
trometers. Structure assignments were done based on 2D NMR experi-
ments (COSY, HMBC, HSQC).
Peri et al. showed recently that the toxicity of the uncomplexed alkyl-
substituted salans plays only a minor role in estimating complex toxici-
ty.[40] By using flow cytometry (PI/annexin-FITC staining), we demon-
strated the negligible cytotoxicity of the halo-substituted salans case as
well.[43]
General procedure for synthesis of the ligands: 2,4-disubstituted phenol
(10 mmol) was suspended or dissolved in methanol (5 mL), then formal-
dehyde (7.5 mL, 36% in water) and N,N’-disubstituted ethylenediamine
(5 mmol) were added immediately. The mixture was heated at reflux for
24 h. Upon cooling, the ligands crystallized as colorless solids (47–61%
yield).
The data collection for the X-ray structures was performed by using a
STOE IPDS-II diffractometer equipped with a graphite monochromated
radiation source (l=0.71073 ꢂ), an image plate detection system, and an
Oxford Cryostream 700 with nitrogen as the coolant gas. The selection,
integration, and averaging procedure of the measured reflection intensi-
ties, the determination of the unit cell by a least-squares fit of the 2q
values, data reduction, LP correction, and the space group determination
were performed by using the X-Area software package delivered with
the diffractometer. A semiempirical absorption correction method was
performed after indexing of the crystal faces. The structures were solved
by direct methods (SHELXS-97)[54] and refined by standard Fourier tech-
niques against F2 with a full-matrix least squares algorithm by using
SHELXL-97[54] and the WinGX (1.80.05)[55] software package. All nonhy-
drogen atoms were refined anisotropically. Hydrogen atoms were placed
in calculated positions and refined with a riding model. Graphical repre-
sentations were prepared with ORTEP-III.[56] CCDC-739047 ([Ti-
General procedure for synthesis of the complexes: The ligand (1.5 mmol)
was dissolved in toluene (10 mL), then titanium alkoxide (1.5 mmol) was
added over a 10 min period under a nitrogen atmosphere. The resultant
yellow reaction mixture was stirred overnight at RT. The solvent was re-
moved under reduced pressure and the complex was obtained as a yellow
solid (>90% yield). The crude product was recrystallized by using the
procedure given in the Supporting Information.
Characterization data for new compounds and details of cell cultivation
and cytotoxicity studies in HeLa S3 and Hep G2 cells are given in the
Supporting Information.
A
)
)
2A
G
G
)2ACTHNGUTERNNUG
(OtBu)2)], -739049 ([Ti-
R
(OnBu)2]),-739050 ([Ti
N
) CHTUNGTRENNUNG
H
)
this paper. These data can be obtained free of charge from The Cam-
Table 9. Crystallographic data for complexes [Ti
(PhMeNMe
)
E
R
)
G
G
)2ACHTUNTGRENNUNG
(OiPr)2].
[Ti
(PhFNMe
[Ti 2A
A
)
U
[Ti
A
)
E
G
)
N
U
)
G
[TiACHTGNUTRENNUG )2ACHTUNGTRENNUNG
(PhClNMe (OiPr)2]
formula
Mr
C26H40N2O4Ti
492.50
0.55ꢅ0.35ꢅ0.1
monoclinic
P21/c
11.7789(7)
16.9074(10)
13.1780(8)
90.00
C30H48N2O4Ti
548.60
0.55ꢅ0.4ꢅ0.05
monoclinic
P21/c
11.4262(7)
15.4592(6)
17.2942(10)
90.00
C30H48N2O4Ti
548.60
0.6ꢅ0.5ꢅ0.2
monoclinic
P21/c
17.5177(18)
9.9728(6)
18.3729(19)
90.00
C24H32F4N2O4Ti
536.39
0.5ꢅ0.35ꢅ0.15
monoclinic
P21/c
8.2933(4)
12.6602(7)
23.7332(10)
90.00
C24H32Cl4N2O4Ti
602.22
0.3ꢅ0.3ꢅ0.3
monoclinic
P21/c
9.270(4)
12.783(4)
23.673(6)
90.00
crystal size [mm3]
crystal system
space group
a [ꢂ]
b [ꢂ]
c [ꢂ]
a [8]
b [8]
105.661(5)
90.00
2527.0(3)
4
1.295
100(2)
107.064(5)
90.00
2920.4(3)
4
1.248
100(2)
109.273(8)
90.00
3029.9(5)
4
1.203
100(2)
95.308(3)
90.00
2481.2(2)
4
1.436
100(2)
95.31(2)
90.00
2793.3(16)
4
1.432
373(2)
g [8]
V [ꢂ3]
Z
1calc [gcmꢂ1
T [K]
]
m
N
]
0.372
0.329
0.317
0.408
0.721
F
G
1056
1184
1184
1120
1248
1.80–25.81
32883
1.3–28.4
37408
2.26–26.21
26501
1.72–26.88
35513
1.2–15.13
6483
)
4870 (0.0986)
4784/0/298
1.023
8771 (0.0719)
8688/0/342
0.975
6080 (0.1301)
6077/54/368
0.988
5363 (0.0661)
5273/0/316
1.063
6103 (0.0322)
6103/0/322
1.085
data/restraints/param
GOF on F2
R1, wR2 [I>2s(I)]
R1, wR2 (all data)
4.86, 10.88
7.63, 11.74
0.412/ꢂ0.365
5.25, 10.24
9.75, 11.45
0.444/ꢂ0.289
6.97, 13.42
14.17, 15.92
0.339/ꢂ0.388
3.46, 8.74
4.69, 9.61
0.274/ꢂ0.525
4.00, 9.05
6.34, 11.29
0.694/ꢂ0.332
largest diff. peak/hole [eꢂꢂ3
]
2788
www.chemeurj.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 2775 – 2789

Titanium Salan Complexes
FULL PAPER
[27] H. Kçpf, S. Grabowski, R. Voigtlꢀnder, J. Organomet. Chem. 1981,
216, 185–190.
Acknowledgements
[28] P. Kçpf-Maier, Eur. J. Clin. Pharmacol. 1994, 47, 1–16.
[29] C. V. Christodoulou, A. G. Eliopoulos, L. S. Young, L. Hodgkins,
D. R. Ferry, D. J. Kerr, Br. J. Cancer 1998, 77, 2088–2097.
[30] M. Guo, Z. Guo, P. J. Sadler, J. Biol. Inorg. Chem. 2001, 6, 698–707.
[31] G. Mokdsi, M. M. Harding, J. Organomet. Chem. 1998, 565, 29–35.
[32] G. Mokdsi, M. M. Harding, J. Inorg. Biochem. 2001, 83, 205–209.
[33] H. Sun, H. Li, R. A. Weir, P. J. Sadler, Angew. Chem. 1998, 110,
1622–1625; Angew. Chem. Int. Ed. 1998, 37, 1577–1579.
[34] M. Guo, H. Sun, H. J. McArdle, L. Gambling, P. J. Sadler, Biochem-
istry 2000, 39, 10023–10033.
The authors would like to thank the Konstanz Research School Chemical
Biology (KoRS-CB) for financial and scientific support. T.A.I. thanks the
KoRS-CB for a personal scholarship. The authors would also like to
thank Dr. Malgorzata Debiak for providing the cisplatin-resistant and
-hypersensitive cell lines, Dipl.-chem. Malin Bein for help with the bio-
logical assays, Anke Friemel for recording the time-resolved and 2D
NMR spectroscopy data, Martin Luka and Dominik Frçhlich for the syn-
thesis of several precursors, and Dr. Michael Burgert for solving the X-
ray structures.
[35] L. M. Gao, R. Hernandez, J. Matta, E. Melendez, J. Biol. Inorg.
Chem. 2007, 12, 959–967.
[36] A. D. Tinoco, A. M. Valentine, J. Am. Chem. Soc. 2005, 127, 11218–
11219.
[37] A. D. Tinoco, C. D. Incarvito, A. M. Valentine, J. Am. Chem. Soc.
2007, 129, 3444–3454.
[38] A. D. Tinoco, E. V. Eames, A. M. Valentine, J. Am. Chem. Soc.
2008, 130, 2262–2270.
[39] M. Shavit, D. Peri, C. M. Manna, J. S. Alexander, E. Y. Tshuva, J.
Am. Chem. Soc. 2007, 129, 12098–12099.
[40] D. Peri, S. Meker, M. Shavit, E. Y. Tshuva, Chem. Eur. J. 2009, 15,
2403–2415.
[41] E. Y. Tshuva, D. Peri, Coord. Chem. Rev. 2009, 253, 2098–2115.
[42] E. Y. Tshuva, J. A. Ashenhurst, Eur. J. Inorg. Chem. 2009, 2203–
2218.
[43] T. A. Immel, M. Debiak, U. Groth, A. Bꢄrkle, T. Huhn, ChemMed-
Chem. 2009, 4, 738–741.
[1] E. Wong, C. M. Giandomenico, Chem. Rev. 1999, 99, 2451–2466.
[2] M. A. Jakupec, M. Galanski, B. K. Keppler, Rev. Physiol. Biochem.
Pharmacol. 2003, 146, 1–54.
[3] M. Galanski, M. A. Jakupec, B. K. Keppler, Curr. Med. Chem. 2005,
12, 2075–2094.
[4] M. Coluccia, G. Natile, Anti-Cancer Agents Med. Chem. 2007, 7,
111–123.
[5] P. C. A. Bruijnincx, P. J. Sadler, Curr. Opin. Chem. Biol. 2008, 12,
197–206.
[6] M. A. Jakupec, M. Galanski, V. B. Arion, C. G. Hartinger, B. K.
Keppler, Dalton Trans. 2008, 183–194.
[7] M. J. Clarke, F. Zhu, D. R. Frasca, Chem. Rev. 1999, 99, 2511–2534.
[8] M. Galanski, V. B. Arion, M. A. Jakupec, B. K. Keppler, Curr.
Pharm. Des. 2003, 9, 2078–2089.
[9] B. Desoize, Anticancer Res. 2004, 24, 1529–1544.
[10] I. Ott, R. Gust, Arch. Pharm. Chem. Life Sci. 2007, 340, 117–126.
[11] N. Katsaros, A. Anagnostopoulou, Crit. Rev. Oncol. Hematol. 2002,
42, 297–308.
[44] E. Y. Tshuva, N. Gendeziuk, M. Kol, Tetrahedron Lett. 2001, 42,
6405–6407.
[45] P. Hormnirun, E. L. Marshall, V. C. Gibson, A. J. P. White, D. J. Wil-
liams, J. Am. Chem. Soc. 2004, 126, 2688–2689.
[46] J. Balsells, P. J. Carroll, P. Walsh, Eur. J. Inorg. Chem. 2001, 5568–
5574.
[47] S. Gendler, S. Segal, I. Goldberg, Z. Goldschmidt, M. Kol, Inorg.
Chem. 2006, 45, 4783–4790.
[48] R. Hamid, Y. Rotshteyn, L. Rabadi, R. Parikh, P. Bullock, In Vitro
Toxicol. 2004, 18, 703–710.
[49] S. N. MacMillan, C. F. Jung, T. Shalumova, J. M. Tanski, Inorg.
Chim. Acta 2009, 362, 3134–3146.
[50] P. Kçpf-Maier, B. Hesse, R. Voigtlꢀnder, H. Kçpf, J. Cancer Res.
Clin. Oncol. 1980, 97, 31–39.
[51] J. R. Boyles, M. C. Baird, B. G. Campling, N. Jain, J. Inorg. Biochem.
2001, 84, 159–162.
[52] NMR spectra are depicted in the Supporting Information.
[53] W. L. F. Armarego, C. L. L. Chai, Purification of Laboratory Chemi-
cals, 5th ed., Elsevier, Amsterdam, 2003.
[54] G. M. Sheldrick, SHELX-97, University of Gçttingen, Gçttingen,
1997.
[12] A. M. Evangelou, Crit. Rev. Oncol. Hematol. 2002, 42, 249–265.
[13] I. Kostova, Curr. Med. Chem. 2006, 13, 1085–1107.
[14] I. Kostova, Anticancer Agents Med. Chem. 2006, 6, 19–32.
[15] F. Caruso, M. Rossi, Mini-Rev. Med. Chem. 2004, 4, 49–60.
[16] E. Melendez, Crit. Rev. Oncol. Hematol. 2002, 42, 309–315.
[17] E. Y. Tshuva, D. Peri, Coord. Chem. Rev. 2009, 253, 2098–2115.
[18] H. Kçpf, P. Kçpf-Maier, Angew. Chem. 1979, 91, 509; Angew. Chem.
Int. Ed. Engl. 1979, 18, 477–478.
[19] P. Kçpf-Maier, F. Preiss, T. Marx, T. Klapçtke, H. Kçpf, Anticancer
Res. 1986, 6, 33–37.
[20] N. J. Sweeney, O. Mendoza, H. Mꢄller-Bunz, C. Pampillꢆn, F.-J. K.
Rehmann, K. Strohfeldt, M. Tacke, J. Organomet. Chem. 2005, 690,
4537–4544.
[21] P. M. Abeysinghe, M. M. Harding, Dalton Trans. 2007, 3474–3482.
[22] K. Strohfeldt, M. Tacke, Chem. Soc. Rev. 2008, 37, 1174–1187.
[23] F. Caruso, M. Rossi, J. Tanski, R. Sartori, R. Sariego, S. Moya, S.
Diez, E. Navarrete, A. Cingolani, F. Marchetti, C. Pettinari, J. Med.
Chem. 2000, 43, 3665–3670.
[24] T. Schilling, K. B. Keppler, M. E. Heim, G. Niebch, H. Dietzfelbing-
er, J. Rastetter, A. R. Hanauske, Invest. New Drugs 1996, 14, 327–
332.
[55] L. J. Farrugia, J. Appl. Crystallogr. 1999, 32, 837–838.
[56] ORTEP-III for Windows, L. J. Farrugia, J. Appl. Crystallogr. 1997,
30, 565.
[25] H. Bischoff, M. R. Berger, B. K. Keppler, D. Schmꢀhl, J. Cancer Res.
Clin. Oncol. 1987, 113, 446–450.
Received: August 20, 2009
[26] J. H. Toney, T. J. Marks, J. Am. Chem. Soc. 1985, 107, 947–953.
Published online: January 26, 2010
Chem. Eur. J. 2010, 16, 2775 – 2789
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2789