Selective treatment of hypoxic tumor cells in vivo: Phosphate pre-prodrugs of nitro analogues of the duocarmycins

Article abstract of DOI:10.1002/anie.201004456

Hitting the hypoxic target: Combining a nitro prodrug with a water-soluble phosphate converts duocarmycin analogues from highly toxic DNA-alkylating agents to highly selective antitumor compounds. These prodrugs (see scheme) have outstanding activity agai

Full text of DOI:10.1002/anie.201004456

Communications
DOI: 10.1002/anie.201004456
Antitumor Agents
Selective Treatment of Hypoxic Tumor Cells In Vivo: Phosphate
Pre-Prodrugs of Nitro Analogues of the Duocarmycins**
Moana Tercel,* Graham J. Atwell, Shangjin Yang, Amir Ashoorzadeh, Ralph J. Stevenson,
K. Jane Botting, Yongchuan Gu, Sunali Y. Mehta, William A. Denny, William R. Wilson, and
Frederik B. Pruijn
In memory of Francis Arthur Tercel and Naumai Aroha Mary Tercel
Duocarmycin SA (1) is one member of a small group of
natural products that share a spirocyclopropylcyclohexadie-
none moiety and the following properties: binding in the
minor groove of DNA, sequence-selective alkylation at N3 of
adenine, and extreme cytotoxicity.[1] Several analogues have
shown highly promising anticancer activity in animal
models,[2] but all failed in clinical trial as humans proved to
be very sensitive to the associated myelotoxicity.[3] Attempts
to introduce a tumor-selective release or activation step have
been widely pursued, including conjugation to tumor-specific
antibodies[4] and the formation of glycosidic prodrugs.[5]
Our approach to introduce tumor selectivity is based on
an amino analogue of the seco form of the alkylating subunit
(3 in Scheme 1), which has the potential to ring-close to an
Scheme 1. Duocarmycin SA (1), and the proposed mechanism of
action of hypoxia-selective nitroCBI prodrugs. R is a side chain that
can bind in the minor groove of DNA.
imino version (4) of the spirocyclopropylcyclohexadienone.
AminoCBI (3) shares many of the same properties as the
phenol congener, including sequence-selective DNA alkyla-
tion and potent cytotoxicity,[6] and can be formed by
enzymatic reduction, in an oxygen-sensitive manner, of the
prodrug nitroCBI (2). With appropriate A-ring substituents
and side chains, 2 is up to several hundred times more toxic to
hypoxic than to well-oxygenated cells.[7] Hypoxia is much
more prevalent and severe in solid tumors than in normal
tissue, and hypoxic tumor cells contribute to treatment failure
as they tend to be chemoresistant, radioresistant, and highly
malignant.[8] Hypoxia-activated prodrugs thus offer great
promise for selective tumor therapy.[9] Application of 2 in vivo
however requires more-water-soluble versions than those
previously reported—the introduction of tertiary amino side
chains resulted in only a marginal increase in solubility, and
inactivity in vivo.[10] Herein we describe phosphate “pre-
prodrugs” of 2 and their activity against hypoxic tumor cells.
Previous studies have shown that for high hypoxic
selectivity, a primary or secondary sulfonamide or carbox-
amide at the 7-position of nitroCBI is favored.[7,10] Avariety of
alcohols 8–15 compatible with this design were prepared by
EDCI-mediated coupling with the acids R2CO2H (Scheme 2).
The indoline starting materials were either known (5) or
prepared in a single step (6, 7). For 7 an alternative and higher
yielding route was developed based on hydrolysis of a primary
carboxamide.[11] The nitroCBIs 10, 11, and 14 were reduced to
the corresponding aminoCBIs 16–18 by hydrogenation or Zn
reduction.
Phosphates were prepared via their tBu esters which were
cleaved in the final step of the synthesis to minimize handling
of these highly polar compounds. The phosphate esters were
prepared by reaction of an alcohol with di-tert-butyl-N,N-di-
isopropylphosphoramidite and subsequent oxidation, a strat-
egy used to introduce the phosphate into either the side chain
R2[11] or the A-ring substituent (Scheme 3). Reaction of
sulfonyl chloride 19 or acid 25 with amine 20 and cleavage of
the trifluoroacetamide provided the indolines 21 and 26,
respectively, which were coupled with the acids R2CO2H as
above, and the phosphate esters cleaved with trifluoroacetic
acid (TFA).
[*] Dr. M. Tercel, G. J. Atwell, Dr. S. Yang, Dr. A. Ashoorzadeh,
Dr. R. J. Stevenson, K. J. Botting, Dr. Y. Gu, S. Y. Mehta,
Prof. W. A. Denny, Prof. W. R. Wilson, Dr. F. B. Pruijn
Auckland Cancer Society Research Centre
Faculty of Medical and Health Sciences
The University of Auckland, Private Bag 92019
Auckland 1142 (New Zealand)
Fax: (+64)9373-7502
E-mail: m.tercel@auckland.ac.nz
[**] This research was funded by the Foundation for Research, Science
and Technology, NZ, and Proacta. We thank Sisira Kumara, Maruta
Boyd, Sally Bai, Wouter van Leeuwen, and Caroline McCulloch for
technical assistance.
Hypoxia-selective cytotoxicity in vitro was assessed with
the cell-permeable alcohols rather than the highly polar
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201004456.
2606
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 2606 –2609

Figure 1. Cytotoxicity of nitroCBI 11 (open symbols) and aminoCBI 17
(filled symbols) in the human cervical carcinoma cell line SiHa, as
determined by clonogenic cell killing. SF: surviving fraction; C:
concentration; circles: oxic; triangles: hypoxic. Stirred cell suspensions
were exposed to the compounds for 4 h under the specified gas phase
and then plated. After 14 days incubation the number of colonies
compared to untreated cells was used to determine SF. The interpo-
lated concentration of 11 to cause 90% of the cells to be killed
(SF=10À1, dotted line) was 7.5 mm under oxic conditions and 0.05 mm
under hypoxic conditions.
Scheme 2. Synthesis of nitroCBI and aminoCBI analogues bearing a
free hydroxy group: a) R2CO2H, 1-ethyl-3-(3-dimethylaminopropyl)car-
bodiimide (EDCI), toluenesulfonic acid (TsOH), dimethylacetamide
(DMA), 51–88%; b) H2, PtO2, THF, 64–74% or (for 18) Zn, NH4Cl,
acetone/water, 85%.
hypoxic cytotoxicity ratio [HCR = IC50(oxic)/IC50(hypoxic)].
Consistent with previous findings, HCR was sensitive to
structural changes (possibly influencing ease of enzymatic
reduction), and of the alcohols, 11 and 14, bearing the basic
side chain C, exhibited the highest selectivity.[11] NitroCBI 11
was further investigated in a panel of 14 human tumor cell
lines of various tissue origin, and in every case significant
HCRs were observed (Figure 2). Notably 17 again displayed
no hypoxic selectivity, but potent cytotoxicity with an average
IC50(oxic) of 28 nm across the cell line panel.
phosphates. NitroCBI 11 for example, when exposed to
suspensions of the human cervical carcinoma cell line SiHa,
eliminated colony-forming cells about 150 times more effec-
tively under hypoxic (< 20 ppm O2) than oxic (20% O2)
conditions (Figure 1). Under hypoxia 11 becomes as toxic as
the aminoCBI 17, which displays no oxygen dependence in its
toxicity. The alcohols 8–15 were compared by measuring IC50
values under oxic and hypoxic conditions and deriving a
Phosphates were prepared of nitroCBI alcohols that
exhibited both low and high HCRs. In all cases the phosphates
were much more water-soluble than the corresponding
Scheme 3. Synthesis of phosphate pre-prodrugs of nitroCBI: a) Et3N, THF then Cs2CO3, MeOH, 91%; b) R2CO2H, EDCI, TsOH, DMA, 66–90%;
c) TFA, CH2Cl2, 65–97%; d) benzotriazol-1-yl oxytripyrrolidinophosphonium hexafluorophosphate (PyBOP), iPr2NEt, 90%.
Angew. Chem. Int. Ed. 2011, 50, 2606 –2609
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 2607

Communications
Figure 2. Cytotoxicity of nitroCBI 11 and aminoCBI 17 in a panel of
human tumor cell lines, as determined by inhibition of proliferation.
IC50: concentration required to inhibit cell proliferation by 50%; black
and white bars: 11 under oxic and hypoxic conditions, respectively;
cross-hatched and hatched bars: 17 under oxic and hypoxic conditions,
respectively. Cell monolayers were exposed to the compounds for 4 h
under the specified gas phase and proliferation was measured 5 days
later in comparison to untreated control cells. The numbers above the
bars are the HCR values for 11 in each cell line.
Figure 3. Antitumor activity in vivo by excision assay. A) Experimental
design: Immunocompromised (CD1 nude) mice bearing SiHa xeno-
grafts were treated with a) a large single dose of g irradiation (15 Gy),
and 5 min later with b) an intravenous dose of a nitroCBI phosphate
(or aminoCBI). After 18 h c) the tumor was excised, dissociated
enzymatically, and plated d) to determine the number of surviving
colony-forming cells per gram of tumor tissue. B) Antitumor activity of
nitroCBI phosphate 23 (closed circles) and aminoCBI 17 (at its MTD;
open circle). LCK (hypoxic log cell kill): additional cell kill for the
combination treatment compared to radiation alone; D: dose; each
point represents the meanÆSEM for groups of 5 or 6 animals.
Dashed line: cell kill by radiotherapy alone; dotted line: approximate
limit of quantitation; upward arrows: the number of tumors/groups
for which no surviving clonogens could be detected.
alcohols. The solubility of 11 for example is only 4 mm in
culture medium, but 23 can be dissolved in phosphate-
buffered saline at > 4 mm, an increase in aqueous solubility of
more than 1000-fold. Administration of 23 to mice and
analysis of plasma samples showed that the phosphate is
rapidly hydrolyzed to the alcohol, and that the plasma
exposure to 11 at the dose used (42 mmolkgÀ1) is more than
enough to provide significant hypoxic cell kill, based on the in
vitro assays.[11]
Administration of higher doses of 23 (> 100 mmolkgÀ1)
caused sporadic acute toxicity—some animals died within a
few minutes, while others survived and exhibited no weight
loss or any other signs of ill health over 60 days of observation.
As a maximum tolerated dose (MTD) could not be defined
the phosphates were instead compared at a nontoxic equi-
molar dose of 42 mmolkgÀ1. Mice bearing a SiHa tumor were
treated with a large dose of radiation, sufficient to cause
about 99% of all cells to be killed (SF = 10À2), as determined
by excising the tumor 18 h after treatment and culturing the
surviving cells. In other words, this radiation dose eliminates
all but the 1 in 100 most hypoxic and radiation-resistant cells
and any further killing of the resistant population when
radiation is combined with a nitroCBI indicates activity
against hypoxic tumor cells in vivo (Figure 3A). Of the
phosphates the greatest (and highly significant) activity was
seen with 23 and 27, corresponding to the alcohols with the
highest hypoxic selectivity in vitro.[11,12] Using the same assay
phosphate 23 was also found to be highly active in five other
human xenografts representing tumors of the cervix, colon,
and lung.[11]
alkylating agent toxicity (reversible weight loss with a nadir
at 7–10 days) and an MTD of 4.2 mmolkgÀ1, but at this dose
was completely inactive against hypoxic tumor cells. In
contrast 23 caused dose-dependent hypoxic cell kill. At
56 mmolkgÀ1, a nontoxic dose, the combination of 23 and
radiation was so effective that in 3 of 5 animals the tumor was
completely sterilized—within the sample size and detection
limits of the assay (about 1 in 100000 surviving cells) no
colony-forming cells could be detected. Phosphate 23 was
further examined in a tumor growth delay assay (Figure 4).
While 23 alone proved to be inactive, it significantly enhanced
the effect of fractionated radiotherapy without apparent
additional toxicity.[11]
In conclusion, the combination of a phosphate pre-
prodrug strategy with our nitroCBI design converts non-
selective and toxic DNA-alkylating agents into well-tolerated
and water-soluble prodrugs that are highly selective for and
active against hypoxic tumor cells in vivo. This approach
offers great promise in tapping the anticancer potential of the
The activity of 23 was compared to that of aminoCBI 17
(Figure 3B). Unlike the prodrugs, 17 exhibited classical
2608 www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 2606 –2609

[1] D. L. Boger, D. S. Johnson, Angew. Chem. 1996, 108, 1542;
Angew. Chem. Int. Ed. Engl. 1996, 35, 1438.
[2] a) L. H. Li et al., Cancer Res. 1992, 52, 4904; b) E. Kobayashi, A.
Okamoto, M. Asada, M. Okabe, S. Nagamura, A. Asai, H. Saito,
K. Gomi, T. Hirata, Cancer Res. 1994, 54, 2404.
[3] a) N. Pavlidis et al., Cancer Chemother. Pharmacol. 2000, 46,
167; b) S. N. Markovic et al., Am. J. Clin. Oncol. 2002, 25, 308.
[4] L. Ducry, B. Stump, Bioconjugate Chem. 2010, 21, 5.
[5] a) L. F. Tietze, F. Major, I. Schuberth, Angew. Chem. 2006, 118,
6724; Angew. Chem. Int. Ed. 2006, 45, 6574; b) L. F. Tietze, T.
Feuerstein, A. Fecher, F. Haunert, O. Panknin, U. Borchers, I.
Schuberth, F. Alves, Angew. Chem. 2002, 114, 785; Angew.
Chem. Int. Ed. 2002, 41, 759.
[6] a) G. J. Atwell, M. Tercel, M. Boyd, W. R. Wilson, W. A. Denny,
J. Org. Chem. 1998, 63, 9414; b) M. A. Gieseg, J. Matejovic,
W. A. Denny, Anti-Cancer Drug Des. 1999, 14, 77.
[7] a) W. R. Wilson, S. M. Stribbling, F. B. Pruijn, S. P. Syddall, A. V.
Patterson, H. D. S. Liyanage, E. Smith, K. J. Botting, M. Tercel,
Mol. Cancer Ther. 2009, 8, 2903; b) M. Tercel et al., J. Med.
Chem. 2009, 52, 7258.
[8] J. A. Bertout, S. A. Patel, M. C. Simon, Nat. Rev. Cancer 2008, 8,
967.
[9] a) Y. Chen, L. Hu, Med. Res. Rev. 2008, 29, 29; b) J. M. Brown,
W. R. Wilson, Nat. Rev. Cancer 2004, 4, 437.
Figure 4. Antitumor activity in vivo by growth delay assay. Immuno-
compromised mice bearing SiHa xenografts were treated on days 1, 2,
and 3 with 23 (42 mmolkgÀ1) and/or g irradiation (5 Gy). V: tumor
volume of median mouse; t: time; circles: control; triangles: 23 alone;
squares: radiation alone; diamonds: 23 5 min after radiation. The time
to quadrupling of the initial treatment volume was 16, 24, 77, and
128 days respectively.
[10] M. Tercel, S. Yang, G. J. Atwell, E. Smith, Y. Gu, R. F. Anderson,
W. A. Denny, W. R. Wilson, F. B. Pruijn, Bioorg. Med. Chem.
2010, 18, 4997.
duocarmycins and directing this action against the therapeuti-
cally most resistant tumor cell population.
[11] See Supporting Information.
[12] Some activity was noted when phosphates were administered
without radiotherapy. This may represent killing of the hypoxic
fraction, or killing of oxic cells, either directly or by diffusion of
aminoCBI from hypoxic zones.
Received: July 21, 2010
Revised: January 2, 2011
Published online: February 17, 2011
Keywords: antitumor agents · duocarmycins · hypoxia ·
.
phosphates · prodrugs
Angew. Chem. Int. Ed. 2011, 50, 2606 –2609
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 2609