Rational Design and Synthesis of Androgen Receptor-Targeted Nonsteroidal Anti-Androgen Ligands for the Tumor-Specific Delivery of a Doxorubicin-Formaldehyde Conjugate

Article abstract of DOI:10.1021/jm0303305

The synthesis and preliminary evaluation of a doxorubicin-formaldehyde conjugate tethered to the nonsteroidal antiandrogen, cyanonilutamide (RU 56279), for the treatment of prostate cancer are reported. The relative ability of the targeting group to bind to the human androgen receptor was studied as a function of tether. The tether served to attach the antiandrogen to the doxorubicin-formaldehyde conjugate via an N-Mannich base of a salicylamide derivative. The salicylamide was selected to serve as a trigger release mechanism to separate the doxorubicin-formaldehyde conjugate from the targeting group after it has bound to the androgen receptor. The remaining part of the tether consisted of a linear group that spanned from the 5-position of the salicylamide to the 3′-position of cyanonilutamide. The structures explored for the linear region of the tether were derivatives of di(ethylene glycol), tri(ethylene glycol), N,N′-disubstituted-piperazine, and 2-butyne-1,4-diol. Relative binding affinity of the tethers bound to the targeting group for human androgen receptor were measured using a ³H-Mibolerone competition assay and varied from 18% of nilutamide binding for the butynediol-based linear region to less than 1% for one of the piperazine derivatives. The complete targeted drug with the butynediol-based linear region has a relative binding affinity of 10%. This relative binding affinity is encouraging in light of the cocrystal structure of human androgen receptor ligand binding domain bound to the steroid Metribolone which predicts very limited space for a tether connecting the antiandrogen on the inside to the cytotoxin on the outside.

Full text of DOI:10.1021/jm0303305

5258
J . Med. Chem. 2003, 46, 5258-5270
Ra tion a l Design a n d Syn th esis of An d r ogen Recep tor -Ta r geted Non ster oid a l
An ti-An d r ogen Liga n d s for th e Tu m or -Sp ecific Deliver y of a
Doxor u bicin -F or m a ld eh yd e Con ju ga te
Peter S. Cogan‡ and Tad H. Koch*,†,‡
Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-0215, and
Department of Pharmaceutical Sciences, University of Colorado Health Sciences Center, Denver, Colorado 80262
Received J uly 8, 2003
The synthesis and preliminary evaluation of a doxorubicin-formaldehyde conjugate tethered
to the nonsteroidal antiandrogen, cyanonilutamide (RU 56279), for the treatment of prostate
cancer are reported. The relative ability of the targeting group to bind to the human androgen
receptor was studied as a function of tether. The tether served to attach the antiandrogen to
the doxorubicin-formaldehyde conjugate via an N-Mannich base of a salicylamide derivative.
The salicylamide was selected to serve as a trigger release mechanism to separate the
doxorubicin-formaldehyde conjugate from the targeting group after it has bound to the
androgen receptor. The remaining part of the tether consisted of a linear group that spanned
from the 5-position of the salicylamide to the 3′-position of cyanonilutamide. The structures
explored for the linear region of the tether were derivatives of di(ethylene glycol), tri(ethylene
glycol), N,N′-disubstituted-piperazine, and 2-butyne-1,4-diol. Relative binding affinity of the
tethers bound to the targeting group for human androgen receptor were measured using a
3H-Mibolerone competition assay and varied from 18% of nilutamide binding for the butynediol-
based linear region to less than 1% for one of the piperazine derivatives. The complete targeted
drug with the butynediol-based linear region has a relative binding affinity of 10%. This relative
binding affinity is encouraging in light of the cocrystal structure of human androgen receptor
ligand binding domain bound to the steroid Metribolone which predicts very limited space for
a tether connecting the antiandrogen on the inside to the cytotoxin on the outside.
In tr od u ction
decades ago;4 yet, only in recent years has it received
vast attention in the literature. The initial attraction
to antiangiogenic drugs was the lack of resistance
mechanisms incurred by normal endothelial cells of the
tumor vasculature upon exposure to chemotherapeutic
agents. Unlike tumor cells, which can adjust to and
overcome a cytotoxic assault, the cells of the requisite
tumor vasculature are not prone to mutation and,
therefore, ultimately succumb to chemotherapy.5 Un-
fortunately, the tumors themselves have found ways to
develop artificial vasculature and grow in the absence
of authentic blood vessels.6 Despite such set backs,
antiangiogenic molecules are still of considerable inter-
est in the field and offer great promise for the future;
however, drugs of widespread clinical use have yet to
be identified.5
Although doxorubicin has been used extensively in
the clinic over the past 3 decades, its use is still limited
by severe acute and chronic systemic toxicities.1 Indeed,
the primary failure of the age of cytotoxic chemotherapy
has been the unacceptable side effects observed with the
agents employed.2 Extensive efforts in the field of
developmental therapeutics have focused on new para-
digms of chemotherapy for selective toxicity to develop-
ing tumors with minimal systemic side effects.
The strategies investigated to achieve the desired
tumor specificity are many and varied yet, to date, the
vast majority have been greatly unsuccessful. Immu-
notherapy is perhaps the most promising of these
approaches on account of the natural complexity and
perpetual evolution of the immune system, as well as
its inherent ability to recognize aberrant cell growth.
Unfortunately, the power of the immune system has yet
to be successfully harnessed for the conquest of cancer
in humans.3
While there has been limited success in the develop-
ment of nontoxic, protein-specific molecules, such as
Gleevec (Novartis, Switzerland), which selectively kill
cancerous cells, these agents are generally active against
only certain types of neoplasms.7 The majority of
tumors, which do not depend on the targeted proteins
for growth, do not succumb to these new drugs. Al-
though the development of nontoxic antitumor drugs is
still in its infancy and holds great promise for the design
of future therapeutics, there is an inherent shortcoming
in the mode of action of these drugs. The genetic
instability of cancerous cells leads to widespread muta-
tions in cellular proteins involved at all levels of
metabolism and growth.8 The selectivity required of
Other methods being explored include the inhibition
of developing intratumoral vasculature (tumor angio-
genesis) and the direct interference with tumor cell
metabolism via nontoxic ligands for receptors expressed
in cancerous cells. The concept of targeting tumor
angiogenesis was initially proposed by Folkman three
* Corresponding author. Phone 303-492-6193, fax 303-492-5894,
e-mail tad.koch@colorado.edu.
† University of Colorado.
‡ University of Colorado Health Sciences Center.
10.1021/jm0303305 CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/18/2003

Anti-Androgen Ligands for Tumor-Specific Delivery
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5259
Sch em e 1. Release of the Formaldehyde Schiff Base of Doxorubicin from Doxsaliform
these tumor-targeted drugs is dependent on binding to
a single, specific receptor expressed by malignant cells.
Mutation or overexpression of the targeted receptor or
circumvention of its participation in metabolic pathways
will ultimately lead to drug resistance. Indeed, this
phenomenon is already coming to light in the clinic.9
Still another approach to achieving antitumor speci-
ficity with concomitant reduction of systemic toxicity is
the selective delivery of cytotoxins. While the targeting
of cytotoxic agents to tumors via a carrier molecule is
relatively new to the clinic, much preclinical work has
been carried out in this promising field. Cytotoxins as
varied as nitrogen mustards,10 nitrosoureas,11,12 anthra-
cyclines,13-15 taxanes,16,17 mitomycin C,18 membrane
acting peptides,19 and assorted antibiotics20,21 have all
been employed in the search for tumor-selective thera-
peutics. Although these selective cytotoxins rely upon
the expression of specific protein targets and are,
therefore, prone to resistance mechanisms such as
mutation or changes in expression of the target, they
have several advantages over related nontoxic ligands.
While the efficacy of molecules which interfere with the
action of a specific cellular protein depend on expression
of the target in every cell of a tumor, targeting com-
pounds which release a nonspecific cytotoxin can po-
tentially act upon tissue surrounding the target-
expressing cell. Accumulation of the cytotoxin within
the tumor is the goal, as opposed to direct action of the
ligand on a cellular receptor. Ligands which act to
merely deliver a cytotoxin may even be expected to
exploit established resistance mechanisms such as
overexpression of the targeted receptor.9,22
Work in our laboratory, and others, has shown that
the concomitant delivery of formaldehyde with doxoru-
bicin, or other anthracyclines, to growing tumor cells
leads to a superior antiprolific response relative to the
delivery of doxorubicin alone.23-25 In an attempt to
capitalize on this observation, we have recently devel-
oped several unique prodrugs of a formaldehyde conju-
gate of the anthracycline doxorubicin. We propose that
the partial hydrolysis of these prodrugs renders the
doxorubicin-formaldehyde Schiff base, which subse-
quently serves to covalently modify genomic DNA, an
event proposed to ultimately be more toxic than the
mere intercalation of unmodified doxorubicin. From
among these novel prodrugs we have identified a
candidate for development as a tumor targeted source
of both doxorubicin and formaldehyde. The N-Mannich
base resulting from the condensation of doxorubicin
with salicylamide (2-hydroxybenzamide) and formalde-
hyde, nominally referred to as doxsaliform (Scheme 1),
has proven to be a superior cytotoxin, relative to the
parent drug, against both doxorubicin-sensitive and
-resistant cultured human tumor cells.26 The N-Mannich
base construct has been observed to be inactive when
intact, but upon time-dependent partial hydrolysis,
yields the superactive doxorubicin-formaldehyde Schiff
base. This partial hydrolysis occurs with a half-life of
57 min under physiological conditions, yet can be readily
increased via acyloxymethylation of the salicylamide
phenolic moiety. It is the salicylamide fragment of this
N-Mannich base which we have here modified via the
attachment of a tumor-targeting moiety.
The androgen receptor (AR) has been identified in a
wide array of human tumors in both male and female
patients. Carcinomas of the breast,27,28 ovary,29 esopha-
gus,30 lung,31 and prostate32 have all been shown to
express the androgen receptor. The AR exists primarily
as a cytosolic receptor33 in complex with several heat-
shock proteins (hsp70, hsp90, and hsp56-59). Ligand
binding leads to dissociation of the heat-shock proteins,
homodimerization, and translocation into the nucleus
where the dimeric receptor recognizes hormone-respon-
sive elements and various components of the transcrip-
tion machinery.34,35 The receptor is often overexpressed
in hormone refractory prostate cancer and is also known
to acquire mutations which lead to promiscuous binding
of various nonandrogen ligands.22,34 Our intention was
to exploit these characteristics of the androgen receptor
via the synthesis of a series of nonsteroidal anti-
androgens which may be used to deliver a doxorubicin-
formaldehyde conjugate to AR-expressing tumors.
Resu lts a n d Discu ssion
Ch em istr y. Doxorubicin has had only limited success
in the treatment of prostate cancer in the clinic.36
However, several groups have shown that the targeting
of DOX to prostate-derived neoplasms via prostate-
specific antigen (PSA) leads to not only preferential
accumulation of the drug in these tissues, but also leads
to a profound response, relative to untargeted drug, as
indicated by serum PSA levels and tumor mass in
treated mice.37,38 While individual PSA enzymes are
expected to activate several of the targeted prodrugs,
and thus allow for a continuous targeting effect, the
enzyme is extracellular and releases the targeting group
from the cytotoxin outside of the cell. We hope to utilize
the mechanism of AR nuclear translocation as a means

5260 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Cogan and Koch
and actually promotes growth of AR-expressing cells at
higher concentrations,33 nilutamide shows no growth
enhancing characteristics.47 Of considerable interest is
the observation that the 3′ nitrogen of the 1-cyano
derivative of nilutamide can be modified with a wide
variety of substituents (Figure 1) which lead to im-
proved binding over the parent drug. The binding pocket
of the AR apparently not only tolerates, but positively
interacts with substituents such as primary alcohols of
varying lengths, double and triple bonds, and aromatic
ring systems.44,45,48 While the direct attachment of
doxorubicin to nilutamide may not yield a viable ligand
for the androgen receptor, the accommodating nature
of the AR ligand binding domain is expected to allow
for the development of a suitable tether by which
nilutamide may be linked to salicylamide. A construct
of this type not only allows for the concomitant delivery
of doxorubicin and formaldehyde via preparation of an
N-Mannich base with the tethered salicylamide, but also
renders a generic targeting group which may be used
to deliver a variety of other compounds to AR-expressing
cells.
F igu r e 1. Various nonsteroidal antiandrogens derived from
flutamide (1a ) or nilutamide (2a ).
of not only localizing the prodrug in tumors but also as
a method of delivering a superactive doxorubicin-
formaldehyde conjugate to the nuclei of expressing cells,
thereby necessitating a smaller dose of the cytotoxin.
A variety of both steroidal and nonsteroidal ligands
for the AR have been described, providing many poten-
tial options to exploit as AR targeting molecules.33,39 Our
hope was to develop a targeting platform based on
endogenous ligands or known therapeutic agents so as
to take advantage of preestablished pharmacokinetics,
pharmacodynamics, toxicology, and metabolic mecha-
nisms. The obvious choice for an AR-specific ligand may
at first appear to be dihydrotestosterone (DHT). While
the endogenous steroid is enticing, it is characterized
by many drawbacks. The hydrophobic nature of steroid
drugs does not facilitate aqueous dissolution; conjuga-
tion of DHT to doxorubicin is not expected to yield a
molecule of acceptable solubility. Also of concern is the
fact that DHT is, by definition, an AR agonist. The use
of a growth-promoting agent to deliver a cytotoxin is
not expected to be an efficient method of inhibiting
tumor growth. Most disconcerting about the use of
derivatized DHT, or other steroids, as targeting groups
are the poor binding affinities of these molecules for the
AR. Several groups have synthesized various derivatives
of DHT, generally modified at the 17 position or by
etherification or esterification of the 17-hydroxy
group.40-42 With few exceptions, steroidal analogues
have been shown to exhibit less than 10% the AR
binding affinity of DHT.
The nonsteroidal antiandrogens (NSAs) bear little
resemblance to the endogenous steroids they antago-
nize. Most notably, they are smaller and are character-
ized by functional groups which lead to a considerably
more polarizable surface area relative to the steroidal
ligands. Although the clinically employed NSAs exhibit
decreased AR binding affinity relative to DHT, binding
can be readily improved through facile modifications of
the core structures.43-45 Because of these aspects, as
well as the general ease of synthesis, we chose to explore
the modification of NSAs through the introduction of
varying tethers for the attachment of the salicylamide
trigger.
Prompted by the superior AR binding affinity of the
alcohol 2d (Figure 1), relative to nilutamide and hy-
droxyflutamide, the active metabolite of flutamide,45,48
our initial efforts were aimed at the synthesis of a series
of ethylene glycol-derived tethers. Poly(ethylene glycol)s
are commonly used excipients for drug delivery. They
are well tolerated and relatively stable to metabolic
enzymes.49 Tethers consisting of di(ethylene glycol) and
tri(ethylene glycol) were explored based on their varying
lengths and steric similarities to the hydroxybutane arm
of 2d . The straight chain ethers were expected to occupy
the same cleft of the androgen receptor ligand binding
domain (AR-LBD) in which the hydroxybutyl chain of
2d resides. The ethylene glycols were also expected to
offer superior aqueous solubility relative to simple
homologous alkyl tethers. The ethylene glycol dimer and
trimer were both employed in an effort to identify a
tether of sufficient length to preclude interference of
ligand binding by the salicylamide and anthracycline
portions of the final drug.
The introduction of functional groups to salicylamide
to allow for tether attachment is no trivial task.
Candidate derivatives must have an electronic character
similar to the parent unsubstituted salicylamide. The
presence of electron-donating or -withdrawing groups
which are in conjugation with the aromatic salicylamide
core threatens to alter the known time constant for
partial hydrolysis of doxsaliform. Although a longer or
shorter time frame for release of the doxorubicin-
formaldehyde Schiff base may ultimately yield a su-
perior therapeutic agent, the current series of targeted
drugs was designed for direct comparison to the prede-
termined cytotoxicity of the untargeted doxsaliform.
Also of concern was the metabolic stability of the tether.
Attachment of the tethers to salicylamide via a benzylic
ether was ultimately explored in an attempt to address
these issues and yielded a stable construct.
Nilutamide 2a (Figure 1) is one of a small group of
clinically employed antiandrogens. Discovered in 1979,
nilutamide is classified as a pure anti-androgen.46
Unlike the most commonly employed clinical anti-
androgen, flutamide 1a , which acts as a partial agonist
Synthesis of the targeting group with di(ethylene
glycol) and tri(ethylene glycol) tethers was conducted
as shown in Scheme 2. The 1-cyano derivative of
nilutamide 2f was prepared in one step and 60% yield
from 4-fluoro-2-(trifluoromethyl)benzonitrile and 5,5-

Anti-Androgen Ligands for Tumor-Specific Delivery
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5261
Sch em e 2a
a
(a) K2CO3, DMF, 65 °C; (b) NaIO4, NaHCO3/H2O pH 8.0; (c1) B10H14, di(ethylene glycol), (c2) B10H14, tri(ethylene glycol), (c3) B10H14
,
2-butyne-1,4-diol; (d) 2,2-dimethoxypropane, p-TsOH, acetone; (e) MsCl, TEA, THF; f) LiBr (10 equiv); (g) sodium salt of 2f, 55 °C; (h)
p-TsOH, MeOH/H2O, reflux.
dimethylhydantoin in the presence of potassium carbon-
ate. The oxidation of Labetalol with sodium periodate
was accomplished using a modified literature procedure
to give 5-formylsalicylamide 3 in 70% yield.50 Introduc-
tion of the tethers, to generate the alcohols 4a and 4b,
was then carried out in good yields (up to 91%) via
decaborane mediated reductive etherification using the
respective ethylene glycol as solvent.51 Protection of the
amide and phenolic moieties of 4a and 4b to give the
dimethylbenzoxazines 5a and 5b was achieved in up to
88% yield by reflux in acetone and 2,2-dimethoxypro-
pane, containing a catalytic amount of p-toluenesulfonic
acid.52 The primary alcohol of each of the benzoxazine-
protected intermediates was then mesylated in 88-92%
yield by treatment with triethylamine in the presence
of pyridinum methanesulfonate, formed in situ, to give
compounds 6a and 6b. Coupling of 2f and tether bearing
salycilamide portions of the targeting group was ac-
complished by deprotonation of 2f with sodium hydride
followed by addition of either 6a or 6b. The resulting
benzoxazine-protected targeting groups 7a and 7b were
then deprotected by reflux in methanol containing 20%
water, in the presence of a catalytic amount of p-
toluenesulfonic acid, to yield the desired compounds 8a
and 8b.
Despite the solubilizing effect of the ethylene glycol
tethers, we feared that the presence of three hydropho-
bic aromatic ring systems in the final prodrug, namely
doxorubicin, salicylamide, and the NSA 2f, would lead
to problems with the formulation of this series of
compounds. There were two major concerns with the
ethylene glycol tethers. First was the possibility that
the inherent insolubility of the intact drugs would
prohibit dissolution in aqueous media. Of equal concern
was the scenario in which the drug would display some
degree of solubility but would fold upon itself to exploit
favorable π-stacking interactions between 2f and either
the doxorubicin or salicylamide portions of the pro-
drug.53 A second set of constructs was therefore devised
in order to introduce a solubilizing functionality and a
source of rigidity into the tether. The heterocyclic
diamine piperazine was chosen in an effort to address

5262 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Cogan and Koch
Sch em e 3a
Although little structural information about the AR
was available when the work described here was initi-
ated, crystallographic data on the ligand binding con-
formation of the AR were independently reported by
Miller et al.39 and Carrondo et al.54 while the syntheses
of our targeting groups were being conducted. While the
crystallographic data further indicated that an ap-
propriate tether could be accommodated by the AR, it
also suggested that the receptor would not tolerate
constructs of considerable steric bulk. Using homology
modeling techniques employing the X-ray crystal struc-
ture of the highly homologous progesterone receptor
(PR) LBD, Miller et al. proposed a 3-dimensional
conformation for nilutamide bound to the AR-LBD.39 In
this model, the 3′ nitrogen of nilutamide is shown to
occupy the same space in the binding pocket of PR as
does carbon 17 of dihydrotestosterone. This suggests
that tethers anchored to the 3′ nitrogen of 2f may be
expected to occupy the same cleft of the AR binding
pocket as Danishefsky’s testosterone conjugate tethers.
To further support the validity of Miller’s PR model,
Carrondo et al, have examined the crystal structure of
the LBDs of both human AR and PR bound to the
steroidal ligand metribolone.54 Comparison of the two
crystal structures shows not only remarkable overlap
of the amino acids defining the respective binding
pockets, but also a nearly identical orientation of the
bound ligand in both receptors. Figure 2 shows a space-
filled rendering of metribolone bound to the human
AR.54 Of note is the fact that the ligand is buried within
the receptor and is afforded little solvent exposure, a
fact which, upon initial examination, proved to be quite
discouraging. However, the ability of the receptor to
tolerate modifications of the ligand, such as those shown
in Figure 1, suggested that an appropriate tether may
be expected to protrude into the cytosol from the LBD.
On the basis of these observations, a fifth targeting
group was synthesized, incorporating the alkynyl moiety
employed by Danishefsky.
a
(a) B10H14, 2-bromoethanol; (b) NaH, DMF; (c) 1,4-dibromobu-
tane or bis(2-bromoethyl) ether, 60 °C; (d) piperazine, THF, reflux;
(e) 9, TEA, THF, reflux.
both concerns. The introduction of two ionizable amines
into the tether should afford additional solubility rela-
tive to the uncharged ethylene glycols. Also, the con-
formational constraints imposed by the six-membered
piperazine ring should serve to inhibit intramolecular
associations of the drug. The syntheses of two deriva-
tives incorporating piperazine into the tether are pre-
sented in Scheme 3. Deprotonation of 2f with sodium
hydride in DMF followed by addition of an excess of
either 1,4-dibromobutane or bis(2-bromoethyl) ether
yielded the brominated compounds 10a and 10b in 85%
and 82% yields, respectively. Subsequent displacement
of the bromide leaving group with excess piperazine in
tetrahydrofuran gave the diamino derivatives 11a and
11b. Finally, the target compounds 12a and 12b were
prepared by refluxing 11a or 11b in THF with the
2-bromoethoxy ether 9, which was prepared by the same
route as the ethylene glycol-derived benzylic ethers.
Although the four described compounds, 8a , 8b, 12a ,
and 12b, were expected to be sufficient to allow for
preliminary evaluation of our targeting strategy, a final
candidate was pursued in an attempt to capitalize on
noted attributes of previously characterized AR binding
molecules. Danishefsky et al. have recently described a
series of testosterone-geldanamycin conjugates which
show a wide range of efficacy, dependent solely upon
the length of an alkynyl tether employed to join the two
drugs.20 Danishefsky demonstrated that a â-propargylic
group at the 17-position of testosterone is necessary for
biological activity in the tested series. Presumably, the
triple bond serves to stringently direct the tether’s
protrusion from the binding pocket. The relevance of
this requirement for tether rigidity in testosterone-
geldanamycin conjugates to NSA derivative binding was
not immediately clear. There is no direct evidence to
suggest that the tethers of Danishefsky’s conjugates
reside in the same cleft in the AR binding pocket as do
the 3′ substituents of the series 2b-e. However, much
indirect evidence supports this very assertion.
Scheme 2 shows the stepwise synthesis of the target-
ing group incorporating 2-butyne-1,4-diol in place of the
ethylene glycol tethers. Reductive etherification of 3
with decaborane in the presence of molten 2-butyne-
1,4-diol yields the corresponding benzylic ether 4c. After
removal of excess butynediol by repeated extraction, the
crude product was dissolved in acetone where it was
refluxed with 2,2-dimethoxypropane and a catalytic
amount of p-toluenesulfonic acid to yield 70% of the
benzoxazine-protected intermediate 5c after two steps.
Attempts to mesylate the alcohol, as was done with the
ethylene glycol derivatives, gave a mixture of products
consisting of primarily the desired, yet unstable, me-
sylate and the corresponding chlorinated product in
varying ratios depending on the conditions used and the
reaction time. The chlorinated product apparently re-
sults from displacement of the successfully installed
methanesulfonate ester by the chloride ion liberated
from consumed methanesulfonyl chloride. In an attempt
to improve upon the yield and selectivity achieved in
the introduction of a leaving group to the propargylic
position of the tether, the mesylation reaction was
repeated in the presence of 10 equiv of LiBr. This served
to completely brominate the terminus of the tether, in
87% yield, which rendered 6c as a superior substrate

Anti-Androgen Ligands for Tumor-Specific Delivery
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5263
F igu r e 2. Space-filling model of the steroidal androgen metribolone (black) bound to the human androgen receptor (white). The
buried nature of the 17-hydroxy moiety (black) of Metribolone illustrates the tight steric demands of the AR binding pocket.
X-ray crystal structure reproduced from the Rutgers protein database entry 1E3G.54
for subsequent reaction with the 2f anion. Displacement
of the bromide with the sodium salt of 2f gave the
protected product 7c in 82% yield. Finally, removal of
the benzoxazine protecting group was carried out in 80%
yield to give the desired compound 8c.
release of the ligand serves to liberate receptors of spent
targeting molecules, which potentially allows for the
acquisition and delivery of multiple prodrugs by a single
receptor.35
A second concern may also be addressed by the rapid
release of nonsteroidal ligands from the AR. The genetic
instability and heterogeneous nature of neoplastic
growths allows for varying expression patterns of gene
products among cells within a single tumor.58 Resistant
tumors can, therefore, be seen to arise from selection
for clones which do not express a targeted protein. While
the rapid association of AR targeted prodrugs will serve
to concentrate the cytotoxin within AR-expressing cells
of a tumor, the rapid dissociation rate may allow for
distribution of the prodrug into nonexpressing cells as
well. Thus, the emergence of resistant colonies may be
retarded by localized diffusion within the tumor.
The competitive binding assays were run for 30 min
incubation periods to demonstrate the interaction of the
nonsteroidal antiandrogens with AR during a relevant
time frame for targeting. Tritiated Miboloerone (3H-
MIB) was chosen as the radioligand on account of its
availability and extensive use in this capacity in the
literature.56,59 All assays were run at 4 °C, to avoid
proteolytic degradation of the receptor, using a modified
protocol which employs hydroxyapatite to sequester and
wash the protein fraction of the assay solution.60 Hy-
droxyapatite was supplied as an insoluble calcium
phosphate coated agarose gel, which served to efficiently
remove proteins from solution. The gel was then col-
lected via filtration and washed extensively to remove
background radioactivity due to nonspecific interactions
Recep tor Bin d in g. The androgen receptor (AR) was
obtained from PC3 cells (donated by Dr. Kerry Burn-
stein, University of Miami; Miami, FL) which had been
stably transfected with the human androgen receptor
cDNA (PC3/AR).55 PC3/AR cells have been thoroughly
characterized and have been shown to express the AR
at ∼ 596 fmol/mg total cellular protein, which is
comparable to the expression of a mutant AR in the
established LNCaP cell line (∼816 fmol/mg).56 PC3/AR
cells were grown to near confluence, sonicated, and
centrifuged to consistently yield 5.0 mL of a lysate
containing approximately 1.9 mg/mL total cellular
protein. Division of the collected lysate into 100 µL
fractions yielded approximately 113 fmol of AR per
aliquot (∼ 1.1 nM). Crude lysate was used as the binding
reaction medium in order to account for undesirable yet
specific ligand-protein interactions. While purified AR
can be used for the binding assay, we felt it was
necessary to identify any unwanted binding events
which supersede the affinity of the targeting compounds
for the AR.
Work by Wakeling et al. suggests that nonsteroidal
antiandrogens are characterized by rapid association
and dissociation with the AR.57 The Wakeling group
found that while 30 min incubation in a steroidal
radioligand competition assay led to efficient nonste-
roidal ligand-receptor interactions, 18 h incubation
showed dramatically decreased competition for binding.
This phenomenon is proposed to reflect the rapid on/off
rates for the nonsteroidal ligands as compared to the
slow, tight binding affinity of steroids. This observation
bodes well for our targeting scheme, as rapid association
of our ligands with the AR is necessary in order to allow
for efficient localization of doxsaliform. Likewise, rapid
3
with H-MIB. Scintillation counting of the dry, washed
gel and filter was then employed to quantify total
binding of 1.0 nM 3H-MIB in the presence of various
concentrations of the test compounds. These numbers
were then compared to controls for nonspecific binding
and unchallenged total binding.59 Results are shown in
Table 1. All assays were performed in duplicate and

5264 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Cogan and Koch
Sch em e 4a
Ta ble 1. IC50 and Relative Binding Affinity (RBA) Values of
Test Ligandsa
test compound
IC50 (nM)
RBA
nilutamide
2f
8a
8b
8c
12a
12b
13
9
6
77
100
150
12
3
18
<1
3
10
6
,1
332
49
>1000
346
90
154
.1000
flutamide
salicylamide
a
IC50 and relative binding affinity values determined from
competitive binding for the human AR of the various test ligands
against 1.0 nM 3H-Mibolerone in PC3/AR cell lysate at 4 °C.
scintillation counting was repeated three times to
ensure reproducibility of the data.
A concern with the development of a modified method
is the validity relative to known protocols. There was
an initial concern that the small differences in specific
and nonspecific binding would not be accurately quanti-
fied by scintillation counting. To address this issue, a
positive control using unlabled Mibolerone as the test
ligand was performed. The cold Mibolerone was found
to compete off 50% of the radioligand at a concentration
of approximately 2.0 nM, suggesting that the developed
method is a valid measure of competitive binding.
Likewise, a negative control experiment was conducted
using salicylamide, which is expected to show no specific
binding to the AR. Table 1 shows that salicylamide was,
in fact, ineffective at competing for AR binding in the
presence of 3H-Mibolerone. Further control experiments
using the cytosolic fraction of PC3/neo cells, which do
not express the androgen receptor,56 also showed no
a
(a) Formalin, DMF, 55 °C, 15 min; (b) doxorubicin hydrochlo-
ride.
ring or the presence of a cationic amine in the tether
may account for this lack of activity.
The di(ethylene glycol) and butynediol derivatives, 8a
and 8c, respectively, exhibited the best RBA values,
although they were only 12% and 18% as efficient as
nilutamide, respectively, in competing for AR binding
against 3H-Mibolerone. Due to the short length of the
tethers in these compounds, it is possible that the
salicylamide moiety of each is responsible for beneficial
interactions which improve the binding affinity. Finally,
the alkynyl tether of 8c apparently serves to maintain
rigidity and direct the salicylamide portion of the
molecule out from the binding pocket, much as it is
proposed to do for Danishefsky’s geldanamycin conju-
gates.20
3
specific binding of H-Mibolerone, suggesting that the
differences measured in PC3/AR lysate are real and AR
specific.
Relative binding affinities (RBA) for the 10 analyzed
compounds are listed in Table 1. The clinically employed
3
Having identified compound 8c as a potential target-
ing molecule, we proceeded to prepare the N-Mannich
base which results from the condensation of 8c with
doxorubicin and formaldehyde as is shown in Scheme
4. The N-Mannich base was isolated in 60% yield and
was found to compete with H3-Mibolerone for AR
binding with an affinity (IC50 ) 90 nM) which was
comparable to that of unmodified 8c (Table 1). To
address the concern of hydrolysis of 13 in the binding
reaction, a solution of the targeted prodrug was pre-
pared in the reaction buffer and incubated at 4 °C for
30 min. Reverse phase HPLC analysis indicated no
appreciable hydrolysis of the N-Mannich base under
these conditions, suggesting that the specific binding
observed was attributed to 13 and not liberated 8c.
These results suggest that doxorubicin-formaldehyde
conjugates, and perhaps various other cytotoxins, may
be efficiently targeted to AR-expressing cells via attach-
ment to nonsteroidal antiandrogens by a suitable tether.
Future work will explore the AR interaction of these
constructs in whole cells and the efficacy of the targeted
N-Mannich base in a prostate tumor-expressing mouse
model.
nilutamide was found to inhibit H-Mibolerone with an
apparent IC50 of 10 nM and has been assigned an
arbitrary RBA of 100. The RBAs of the other test
compounds are expressed as fractions of nilutamide
binding, based on their respective IC50 values. The RBA
of 2f (150%) suggests that the use of this molecule as
the core of our targeting constructs was quite appropri-
ate. The majority of the compounds tested exhibit RBA
values between 1% and 20% of that observed for
nilutamide, indicating that the introduction of our
tethers has a detrimental effect on binding. However,
the IC50 value of the best of the targeting groups, 8c at
49 nM, is still on the same order of magnitude as the
unmodified 2f.
The tri(ethylene glycol) derivative 8b, having the
longest tether of the tested compounds, displayed only
3% of the binding affinity of nilutamide. This surprising
finding is likely the result of the excessive flexibility of
the tri(ethylene glycol) tether which is proposed to
facilitate folding of the molecule and subsequent in-
tramolecular interactions which preclude efficient re-
ceptor binding.53 Also of interest is the poor ability of
the piperazine analogues 12a and 12b to effectively
3
displace H-Mibolerone binding in the tested concentra-
Several experiments were carried out in attempts to
quantitate the cytotoxicity of 13, relative to doxorubicin
tion range. The added steric demands of the piperazine

Anti-Androgen Ligands for Tumor-Specific Delivery
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5265
gradients of acetonitrile and triethylammonium acetate (Et3-
NHOAc; TEAA), prepared as 20 mM triethylamine adjusted
to pH 6.0 with acetic acid. The method employed for all
analytical chromatography was as follows: A ) CH3CN, B )
pH 6.0 buffer; A:B, 0:100 to 70:30 at 10 min, isocratic until 12
min, 0:100 at 15 min. For preparative HPLC, a 5 µm spherical
particle C18 Ranin Dynamax semipreparative column was
employed, 10 mm × 25 cm with a 10 mm × 5 cm guard column,
eluting at 3.0 mL/min, monitoring at 260 and 310 nm.
Adequate preparative separation was achieved using the
following method: A ) CH3CN, B ) 20 mM triethylamine
adjusted to pH 3.5 or 4.0 as indicated with glacial acetic acid
(TEAA buffer); A:B, 0:100 to 70:30 at 20 min, isocratic until
30 min, 0:100 at 35 min. Water was distilled and purified with
a Millipore Q-UF Plus purification system to 18 Mohm-cm.
Flash silica gel (particle size: 32-63 µm, pore size: 60 Å, cat.
# 02826-25) was obtained from Scientific Adsorbants Inc.
(Atlanta, GA). Labetalol hydrochloride (97.5%) was purchased
from either Sigma (Milwaukee, WI) or Spectrum Chemicals
(New Brunswick, NJ ). Bis(2-bromoethyl) ether (96%) and
4-fluoro-2-(trifluoromethyl)benzonitrile (98%) were obtained
from Lancaster Research Chemicals, (Windham, NH). Sodium
periodate (reagent grade) and 5,5-dimethylhydantoin (97%)
were purchased from Acros Organics (Pittsburgh, PA). Meth-
anesulfonyl chloride (99.5%), p-toluenesulfonic acid (98.5%),
decaborane (white), di(ethylene glycol) (99%), tri(ethylene
glycol) (99%), 2-bromoethanol (95%), 2-butyne-1,4-diol (99%),
1,4-dibromobutane (99%), 2,2-dimethoxypropane (98%), and
piperazine (99%) were all purchased from Aldrich (Milwaukee,
WI). The BSA micro protein determination kit was purchased
from Sigma (Milwaukee, WI), and the Complete-mini protease
inhibitor cocktail was obtained from Roche. All other solvents
and reagents were purchased from Fisher as reagent grade
and used without further purification except for Fisher HPLC
grade acetonitrile and Aldrich Sure-Seal (anhydrous) N,N-
dimethylformamide (99.8%) and pyridine (99.8%).
and untargeted doxsaliform, against PC3/AR and PC3/
neo cells. Experiments were run in cell culture media
supplemented with either fetal bovine serum (FBS) or
dextran-charcoal stripped calf serum in an attempt to
account for the presence of testosterone in the unadul-
terated FBS. Unfortunately, an underlying problem
prevented the accurate analysis of the effect of prodrug
targeting. Extended treatment of cells with varying
concentrations of the two prodrugs leads to release of
the doxorubicin-formaldehyde conjugate by 13 and by
untargeted doxsaliform, both inside and outside the
cells, irrespective of receptor binding. Thus both pro-
drugs serve to bathe the cells in the doxorubicin-
formaldehyde conjugate via hydrolysis over the course
of the exposure period. Prodrug treatment times > 3 h
were required for extensive hydrolysis of the N-Mannich
base, but over this time period, both 13 and doxsaliform
release the same amount of the doxorubicin-formalde-
hyde conjugate. An in vivo system is expected to allow
for accumulation of the targeted prodrug in AR-express-
ing cells, where hydrolysis of the N-Mannich base will
lead to localized delivery of the active drug. This should
greatly contrast the deposition of the untargeted dox-
saliform, which is expected to experience no preferential
distribution. Simply stated, the targeted prodrug was
designed to exploit a dynamic system of circulation,
accumulation due to receptor binding, and release of the
cytotoxin from an inactive conjugate, while cell culture
only offers a static model for determining cytotoxicity.
While these IC50 studies indicated that the potency of
the targeted drug was not diminished relative to doxo-
rubicin or the untargeted N-Mannich base prodrug, the
effect of AR binding and subsequent release of the
doxorubicin-formaldehyde conjugate could not be as-
certained through cell culture experiments. However,
preliminary fluorescence microscopy has shown that
both 8c and 13 do, in fact, bind to the AR in live cultured
cells (work in progress). On the basis of the AR binding
affinity of 13, as determined here in cell lysate as well
as in whole cells (data not shown), we are currently
developing a mouse model, employing orthotopicly
implanted prostate tumors, which will serve as a
dynamic test system for assessment of the efficacy of
13.
All tissue culture materials were obtained from Gibco Life
Technologies (Grand Island, NY) unless otherwise noted. PC3/
AR and PC3/neo cells were a gift from Dr. Kerry L. Burnstein
(University of Miami, FL). Both cell lines were maintained in
vitro by serial culture in RPMI 1640 media supplemented with
either 10% fetal bovine serum (Gemini Bio-Products, Calbasas,
CA) or 10% dextran-charcoal stripped (delipidated) calf serum,
L-glutamine (2 mM), HEPES buffer (10 mM), penicillin (100
units/mL), and streptomycin (100 µg/mL). Cells were main-
tained at 37 °C in a humidified atmosphere of 5% CO2 and
95% air. Phenol red-free RPMI 1640 media supplemented with
L-glutamine was obtained from Sigma (Milwaukee, WI). Spin-X
centrifuge filters were purchased from Costar/Corning (Corn-
ing, NY).
Syn th eses. 4-(4,4-Dim eth yl-2,5-d ioxo-im id a zolid in -1-
yl)-2-tr iflu or om eth ylben zon itr ile (2f). To a stirring solu-
tion of 1.00 g (4.78 mmol) of 4-fluoro-2-(trifluoromethyl)-
benzonitrile in 15.0 mL of DMF was added 3.10 g (23.9 mmol)
of 5,5-dimethylhydantoin and 0.990 g (7.17 mmol) of K2CO3.
The resulting suspension was stirred under an argon atmos-
phere at 55 °C for 16 h and then at 65 °C for 48 h. The reaction
mixture was diluted to 300 mL with ethyl acetate, vacuum
filtered, and rotary evaporated at 40 °C followed by 50 °C and
50 µm Hg to yield a bright yellow paste. The paste was
dissolved in 25% hexanes/75% ethyl acetate and eluted from
a silica gel flash column (35 cm × 3 cm) with 50% hexanes/
50% ethyl acetate. The collected product was rotary evaporated
at 40 °C to give a white solid which was recrystallized from
ethyl acetate/hexanes to give 0.780 g (60%) of 2f as a white
crystalline solid (mp 208-210 °C): 1H NMR (500 MHz, (CD3)2-
CO) δ 1.53 (6H, s, 4-(CH3)2), 7.81 (1H, bs, NH), 8.13 (1H, dd,
J ) 8, 2 Hz, 5), 8.20 (1H, d, J ) 8 Hz, 6), 8.25 (1H, d, J ) 2
Hz, 3); m/z 297.0723 [M+] (calculated for 297.0725); anal.
(C13H10F3N3O2) C, H, N.
Exp er im en ta l Section
Gen er a l Rem a r k s. Melting points were determined in
open capillary tubes with a Thomas-Hoover capillary melting
point apparatus and are uncorrected. 1H NMR spectra were
acquired with a Varian Unity Inova 500 MHz spectrometer.
Unambiguous NMR assignments for the protons of the 2f,
salicylamide, and doxorubicin portions of the synthesized
compounds are designated by “nil”, “sal”, or “dox”, respectively.
Mass spectral data were acquired on a VG Instruments
AutospecM mass spectrometer by electron impact (EI) using
a perfluorokerosene internal standard for [M+] data or liquid
SIMS (LSIMS) ionization with a poly(ethylene glycol) (PEG)
internal standard for [MH+] data. Mass spectral data for
compound 13 were collected by Dr. Chris Hadad (Ohio State
University; Columbus, OH) with a 3-Tesla Finnigan FTMS-
2000 Fourier Transform mass spectrometer. HPLC analyses
were performed with a Hewlett-Packard 1090 liquid chro-
matograph equipped with a diode array UV-vis detector and
workstation; chromatography was performed with a Hewlett-
Packard 5 µm reverse phase C18 microbore column, 2.1 mm
i.d. x 100 mm, eluting at 0.5 mL/min, monitoring at 260 and
310 nm. Acceptable analytical resolution was achieved with
5-F or m yl-2-h yd r oxyben za m id e (3). A 600 mL stirring
aqueous solution of 2.00 g (5.48 mmol) of Labetalol hydrochlo-
ride in a 1.0 L round-bottom flask was neutralized with 4 mL
of saturated NaHCO3. The reaction flask was then fitted with

5266 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Cogan and Koch
a dropping funnel containing 1.17 g (5.48 mmol) of sodium
periodate in 50 mL of Millipore H2O. Dropwise addition of the
periodate solution over 15 min at room temperature gave a
pale pink solution which was stirred for an additional 20 min.
The solution was acidified with 3.0 mL of concentrated aqueous
HCl and stirred vigorously until a white precipitate was
formed (approximately 2 min). The resulting suspension was
stored for 12 h at 4 °C, to facilitate precipitation, at which
time it was filtered. The collected solid was recrystallized from
80 mL of boiling Millipore H2O and allowed to sit for 12 h at
4 °C. Vacuum filtration gave 0.634 g (70%) of 3 as white to
pale golden needles (mp 204-206 °C): 1H NMR (500 MHz,
(CD3)2CO) δ 7.06 (1H, d, J ) 8 Hz, 3), 7.47 (1H, bs, NH), 7.98
(1H, dd, J ) 8, 2 Hz, 4), 8.33 (1H, bs, NH), 8.40 (1H, d, J ) 2
Hz, 6), 9.84 (1H, s, HCO), 13.87 (1H, s, 2-OH); m/z 165.0421
[M+] (calculated for 165.0426).
2-Hyd r oxy-5-[2-(2-h yd r oxyeth oxy)eth oxym eth yl]ben z-
a m id e (4a ). A solution of 200 mg (1.21 mmol) of 3 was
prepared in 10 mL of di(ethylene glycol) heated under an argon
atmosphere to 70 °C in a mineral oil bath. After dissolution
was complete, the solution was removed from the oil bath and
allowed to cool for 5 min at which time 74 mg (0.61 mmol) of
decaborane was added. Strong effervescence was observed over
5 min but then subsided. The reaction was then placed back
in the oil bath and was stirred at 70 °C for 5 h. The solvent
was removed by rotary evaporation at 60 °C and 50 µm Hg.
After the bulk of the solvent was removed, the remaining oil
was transferred to a separatory funnel with 300 mL of ethyl
acetate. This solution was washed 4× with 50 mL portions of
saturated brine and the organic layer was collected, dried over
anhydrous magnesium sulfate, and rotary evaporated at 40
°C to a pale yellow oil. The desired product was then collected
from a silica gel flash column (35 cm × 3 cm diameter), eluting
with 10% hexanes/90% ethyl acetate. Removal of the solvent
by rotary evaporation at 40 °C gave 291 mg (91%) of 4a as a
clear, colorless oil: 1H NMR (500 MHz, (CD3)2CO δ 3.49-3.54
(2H, m, OCH2CH2OH), 3.55-3.59 (2H, m, 1/2(OCH2CH2O)),
3.59-3.66 (4H, m, CH2OH, 1/2(OCH2CH2O)), 3.89 (1H, bs,
CH2OH), 4.44 (2H, s, Bn), 6.86 (1H, d, J ) 9 Hz, 3), 7.19 (1H,
bs, NH), 7.4 (1H, dd, J ) 9, 2 Hz, 4), 7.81 (1H, d, J ) 2 Hz, 6),
8.01 (1H, bs, NH), 12.9 (1H, bs, 2-OH); m/z 255.1106 [M+]
(calculated for 255.1107).
6-[2-(2-H yd r oxyet h oxy)et h oxym et h yl]-2,2-d im et h yl-
2,3-d ih yd r o-ben zo[e][1,3]oxa zin -4-on e (5a ). A sample of
200 mg (0.78 mmol) of 4a was dissolved in 20 mL of acetone
and 10 mL of 2,2-dimethoxypropane. A catalytic amount of
p-toluenesulfonic acid was added, and the resulting solution
was refluxed under an argon atmosphere at 80 °C for 1.5 h.
The solvent was then removed by rotary evaporation at 40 °C.
The resulting brown residue was transferred to a separatory
funnel in 250 mL of ethyl acetate and was washed 3× with 50
mL portions of saturated brine containing 5% K2CO3. The
organic layer was collected, dried over anhydrous magnesium
sulfate, and rotary evaporated at 40 °C to give a yellow oil.
The washed product was then eluted from a silica gel flash
column (35 cm × 3 cm) in 5% hexanes/95% ethyl acetate.
Removal of solvent yielded 204 mg (88%) of pure 5a as a clear,
colorless oil: 1H NMR (500 MHz, (CD3)2CO) δ 1.61 (6H, s,
2-(CH3)2), 3.50-3.53 (2H, m, OCH2CH2OH), 3.59-3.65 (6H,
m, CH2OH, OCH2CH2O), 3.73 (1H, t, J ) 6 Hz, CH2OH), 4.52
(2H, s, Bn), 6.92 (1H, d, J ) 9 Hz, 8), 7.47 (1H, dd, J ) 9, 2
Hz, 7), 7.82 (1H, bs, NH), 7.85 (1H, d, J ) 2 Hz, 5); m/z
295.1412 [M+] (calculated for 295.1420).
6-{2-[2-(2-H yd r oxyet h oxy)et h oxy]et h oxym et h yl}-2,2-
d im et h yl-2,3-d ih yd r o-ben zo[e][1,3]oxa zin -4-on e (5b). A
solution of 200 mg (1.21 mmol) of 3 was prepared in 10 mL of
tri(ethylene glycol) heated to 70 °C under an argon atmosphere
in a mineral oil bath. After dissolution was complete, the
solution was removed from the oil bath and allowed to cool
for 5 min before 74 mg (0.60 mmol) of decaborane was added.
Strong effervescence was observed over 5 min but then
subsided. The reaction was then placed back in the oil bath
and was stirred at 70 °C for 5 h. The solvent was then removed
by heating to 125 °C in a Ku¨gelrohr oven at 150 µm Hg for 2
h. After the bulk of the solvent was removed, the remaining
viscous liquid was transferred to a separatory funnel with 100
mL of saturated brine. This solution was extracted into 4 ×
200 mL portions of ethyl acetate, which were collected, pooled,
dried over anhydrous magnesium sulfate, and rotary evapo-
rated at 40 °C to a pale yellow oil. The desired, semipure
product 4b was then collected from a silica gel flash column
(35 cm × 3 cm), eluting with 5% hexanes/95% ethyl acetate.
This semipure product was dissolved in 20 mL of acetone and
10 mL of 2,2-dimethoxypropane. A catalytic amount of p-
toluenesulfonic acid was added, and the resulting solution was
refluxed at 80 °C for 1.5 h. The solvent was then removed by
rotary evaporation at 40 °C. The resulting brown residue was
transferred to a separatory funnel in 300 mL of ethyl acetate
and was washed 3× with 50 mL portions of saturated brine
containing 5% K2CO3. The organic layer was collected, dried
over magnesium sulfate, and rotary evaporated to a pale
yellow oil at 40 °C. The washed product was then eluted from
a silica gel flash column (35 cm × 3 cm) in 5% methanol/95%
ethyl acetate. Removal of solvent yielded 231 mg (56% in two
steps) of pure 5b as a clear, colorless oil: 1H NMR (500 MHz,
(CD3)2CO) δ 1.61 (6H, s, 2-(CH3)2), 3.50-3.53 (2H, m, OCH2-
CH2OH), 3.56-3.65 (10H, m, OCH2CH2OH, 2(OCH2CH2O)),
3.74-3.77 (1H, m, CH2OH), 4.51 (2H, s, Bn), 6.92 (1H, d, J )
8 Hz, 8), 7.48 (1H, dd, J ) 8, 2 Hz, 7), 7.84 (1H, d, J ) 2 Hz,
5), 7.99 (1H, bs, NH); m/z 339.1681 [M+] (calculated for
339.1682).
6-(4-H yd r oxyb u t -2-yn yloxym et h yl)-2,2-d im et h yl-2,3-
d ih yd r o-ben zole[e][1,3]-oxa zin -4-on e (5c). A solution of
200 mg (1.21 mmol) of 3 was prepared in 10 g of 1,4-butyne-
2-diol by heating a mixture of the two solids at 70 °C in a
mineral oil bath for 15 min. The resulting solution was
removed from the oil bath and allowed to cool for 5 min before
74 mg (0.60 mmol) of decaborane was added. Strong ef-
fervescence was observed for approximately 5 min but then
subsided. The reaction was then stirred under an argon
atmosphere at 60 °C for 5 h and was subsequently diluted to
300 mL with ethyl acetate and transferred to a separatory
funnel. After 8 × 40 mL washes with saturated brine, the
organic layer was collected, dried over anhydrous magnesium
sulfate, and rotary evaporated at 40 °C to give approximately
1.5 mL of a clear amber oil. The crude product was diluted to
10 mL in ethyl acetate and was eluted from a silica gel flash
column (30 cm × 2 cm) in 100% ethyl acetate. Rotary
evaporation of the solvent gave 4c as a semipure clear, pale
yellow oil. The product from the flash column was dissolved
in 20 mL of acetone and 10 mL of 2,2-dimethoxypropane. A
catalytic amount of p-toluenesulfonic acid was added and the
resulting solution was refluxed at 85 °C for 1.5 h. The solvent
was removed by rotary evaporation and the resulting oil was
transferred to a separatory funnel in 300 mL of ethyl acetate.
This solution was washed 3x with 40 mL portions of saturated
brine containing 5% K2CO3 and the organic layer was collected,
dried over anhydrous magnesium sulfate and rotary evapo-
rated at 40 °C to give 233 mg (70% in two steps) of 5c as a
clear, colorless oil: 1H NMR (500 MHz, (CD3)2CO) δ 1.62 (6H,
s, 2-(CH3)2), 3.45 (1H, bs, CH2OH), 4.17 (2H, bm, CCH2OH),
4.3 (2H, bm, OCH2C), 4.53 (2H, s, Bn), 6.68 (1H, 2, J ) 8 Hz,
8), 7.43 (1H, dd, J ) 8, 2 Hz, 7), 7.90 (1H, d, J ) 2 Hz, 5), 8.21
(1H, bs, NH); m/z 275.1145 [M+] (calculated for 275.1158).
Meth a n esu lfon ic Acid 2-[2-(2,2-Dim eth yl-4-oxo-3,4-d i-
h yd r o-2H-ben zo[e][1,3]oxa zin -6-ylm eth oxy)eth oxy]eth -
yl Ester (6a ). To a stirring solution of 200 mg (0.68 mmol) of
5a in 4 mL of THF were added 55 µL (0.68 mmol) of dry
pyridine and 160 µL (2.1 mmol) of methanesulfonyl chloride.
This solution was stirred under an argon atmosphere at room
temperature for 30 min at which time 380 µL (2.8 mmol) of
triethylamine was added. A white precipitate was formed
immediately and the reaction was stirred for 2 h at room
temperature. The reaction mixture was then diluted to 300
mL with ethyl acetate and transferred to a separatory funnel.
After 3 × 40 mL washes with saturated brine containing 5%
NaH2PO4, the organic layer was collected, dried over anhy-
drous magnesium sulfate, and rotary evaporated at 40 °C to

Anti-Androgen Ligands for Tumor-Specific Delivery
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5267
a pale yellow oil. This crude oil was dissolved in 8 mL of ethyl
acetate and introduced to a silica gel flash column (35 cm × 3
cm) packed in 10% hexanes/90% ethyl acetate. Elution with
the same, followed by removal of solvent by rotary evaporation
at 40 °C, gave 233 mg (92%) of 6a as a clear, colorless oil: 1H
NMR (500 MHz, (CD3)2CO δ 1.62 (6H, s, 2-(CH3)2), 3.1 (3H, s,
SO2CH3), 3.62-3.66 (2H, m, 1/2(OCH2CH2O)), 3.67-3.70 (2H,
m, 1/2(OCH2CH2O)), 3.73-3.78 (2H, m, OCH2CH2OMs), 4.34
- 4.39 (2H, m, OCH2CH2OMs), 4.52 (2H, s, Bn), 6.92 (1H, d,
J ) 9 Hz, 8), 7.48 (1H, dd, J ) 9, 2 Hz, 7), 7.74 (1H, bs, NH),
7.81 (1H, d, J ) 2 Hz, 5); m/z 373.1188 [M+] (calculated for
373.1195).
°C, gave a pale yellow oil. Elution from a silica gel flash column
(30 cm × 2 cm) in 10% hexanes/90% ethyl acetate yielded 82
mg (42% in two steps) of 8a as a clear, colorless oil: 1H NMR
(500 MHz, (CDCl3) δ 1.54 (6H, s, nil-5-(CH3)2), 3.54-3.59 (2H,
m, OCH2CH2N), 3.61-3.65 (2H, m, 1/2(OCH2CH2O)), 3.66-
3.70 (2H, m, 1/2(OCH2CH2O)), 3.74-3.79 (2H, m, OCH2CH2N),
4.46 (2H, s, Bn), 5.91 (1H, bs, NH), 6.62 (1H, bs, NH), 6.95
(1H, d, J ) 8 Hz, sal-3), 7.33 (1H, dd, J ) 8, 2 Hz, sal-4), 7.41
(1H, d, J ) 2 Hz, sal-6), 7.88 (1H, d, J ) 9 Hz, nil-5), 7.92
(1H, dd, J ) 9, 2 Hz, nil-6), 8.09 (1H, d, J ) 2 Hz, nil-2), 12.22
(1H, bs, 2-OH); m/z 534.1711 [M+] (calculated for 534.1726);
anal. (C25H25F3N4O6) C, H, N.
Meth a n esu lfon ic Acid 2-{2-[2-(2,2-d im eth yl-4-oxo-3,4-
d ih yd r o-2H-ben zo [e][1,3]oxa zin -6-ylm eth oxy)eth oxy]-
eth oxy}eth yl Ester (6b). 6b was prepared as 6a in 90%
yield: 1H NMR (500 MHz, (CD3)2CO δ 1.61 (6H, s, 2-(CH3)2),
3.10 (3H, s, SO2CH3), 3.58-3.65 (8H, m, 2(OCH2CH2O)), 3.75
(2H, m, OCH2CH2OMs), 4.35 (2H, m, OCH2CH2OMs), 4.51
(2H, s, Bn), 6.92 (1H, d, J ) 9 Hz, 8), 7.48 (1H, dd, J ) 9, 2
Hz, 7), 7.83 (1H, d, J ) 2 Hz, 5), 7.92 (1H, bs, NH); m/z:
417.1452 [M+] (calculated for 417.1452).
5-[2-(2-{2-[3-(4-Cya n o-3-tr iflu or om eth ylp h en yl)-5,5-d i-
m et h yl-2,4-d ioxo-im id a zolid in -1-yl]et h oxy}et h oxy)et h -
oxym eth yl]-2-h yd r oxyben za m id e (8b). 8b was prepared as
8a in 58% yield: 1H NMR (500 MHz, (CD3)2CO) δ 1.57 (6H, s,
nil-5-(CH3)2), 3.58-3.61 (4H, m, OCH2CH2O), 3.61-3.67 (6H,
m, OCH2CH2N, OCH2CH2O), 3.73-3.76 (2H, m, OCH2CH2N),
4.46 (2H, s, Bn), 6.90 (1H, d, J ) 8 Hz, sal-3), 7.19 (1H, bs,
NH), 7.45 (1H, dd, J ) 8, 2 Hz, sal-4), 7.81 (1H, d, J ) 2 Hz,
sal-6), 7.97 (1H, bs, NH), 8.17 (1H, dd, J ) 9, 2 Hz, nil-6), 8.22
(1H, d, J ) 9 Hz, nil-5), 8.29 (1H, d, J ) 2 Hz, nil-2), 12.95
(1H, s, 2-OH); m/z 578.2000 [M+] (calculated for 578.1988);
anal. (C27H29F3N4O7) C, H, N.
6-(4-Br om ob u t -2-yn yloxym et h yl)-2,2-d im et h yl-2,3-d i-
h yd r o-ben zo[e][1,3]oxa zin -4-on e (6c). To a stirring solution
of 150 mg (0.55 mmol) of 5c in 2.0 mL of THF were added 45
µl (0.55 mmol) of pyridine and 170 µl (2.2 mmol) of methane-
sulfonyl chloride. After 30 min stirring under argon, 380 µL
(2.7 mmol) of triethylamine was added, and the resulting
solution was allowed to stir for 1 h at room temperature,
during which time a white precipitate gradually formed. To
this stirring suspension was added 480 mg (5.5 mmol) of LiBr
as a solution in 2 mL of THF. Once HPLC indicated completion
of the reaction, the resulting suspension was diluted to 200
mL with ethyl acetate and was washed 3× with 40 mL portions
of saturated brine containing 5% NaH2PO4. The organic layer
was then collected, dried over anhydrous magnesium sulfate,
and rotary evaporated at 40 °C to give a pale yellow oil. The
washed product was eluted from a silica gel flash column (30
cm × 2 cm) in 10% hexanes/90% ethyl acetate. Removal of
solvent by rotary evaporation at 40 °C yielded 162 mg (87%)
of 6c as a clear, colorless oil: 1H NMR (500 MHz, CDCl3) δ
1.67 (6H, s, 2-(CH3)2), 3.95-3.98 (2H, m, OCH2C), 4.21-4.25
(2H, m, CCH2Br), 4.55 (2H, s, Bn), 6.21 (1H, d, J ) 8 Hz, 8),
7.47 (1H, dd, J ) 8, 2 Hz, 7), 7.89 (1H, d, J ) 2 Hz, 5), 8.08
(1H, s, NH); m/z 337.0314 [M+] (calculated for 337.0314).
5-(2-{2-[3-(4-Cya n o-3-t r iflu or om et h ylp h en yl)-5,5-d i-
m eth yl-2,4-d ioxo-im id a zolid in -1-yl]eth oxy}eth oxym eth -
yl)-2-h yd r oxyben za m id e (8a ). A solution of 100 mg (0.34
mmol) of 2f was prepared in 2.0 mL of DMF, and to this was
added 13.5 mg (0.34 mmol) of sodium hydride as a 60%
emulsion in oil. The mixture was stirred under an argon
atmosphere at room temperature for 3 h. The resulting yellow
solution was then added to a stirring solution of 127 mg (0.34
mmol) of 6a in 2.0 mL of DMF, and the reaction flask was
heated to 60 °C in a mineral oil bath for 24 h. The product
was precipitated by dropwise addition of the reaction mixture
to 100 mL of saturated aqueous NaH2PO3. The pale yellow
precipitate was then extracted into 250 mL of ethyl acetate,
which was collected and rotary evaporated at 40 °C to yield a
yellow solution of crude product in DMF. Further rotary
evaporation at 50 °C and 100 µm Hg removed the DMF to give
a viscous yellow oil. This residue was eluted from a silica gel
flash column (30 cm × 2 cm) in 10% hexanes/90% ethyl acetate
to yield semipure 7a . The fractions containing 7a which were
collected from the column were pooled and the solvent removed
via rotary evaporation at 40 °C to give a clear, colorless oil
which was dissolved in 10 mL of 80% methanol/20% water. A
catalytic amount of p-toluenesulfonic acid was added, and the
resulting solution was refluxed at 90 °C for 30 h. The methanol
was removed and 50 mL of saturated brine was used to
transfer the resulting emulsion to a separatory funnel where
it was extracted into 250 mL of ethyl acetate and washed 2×
with 40 mL portions of saturated brine containing 5% NaH-
CO3. Collection of the organic layer, followed by drying over
anhydrous magnesium sulfate and rotary evaporation at 40
4-{3-[4-(2,2-Dim eth yl-4-oxo-3,4-dih ydr o-2H-ben zo[e][1,3]-
oxa zin -6-ylm eth oxy)bu t-2-yn yl]-4,4-d im eth yl-2,5-d ioxo-
im id a zolid in -1-yl}-2-tr iflu or om eth ylben zon itr ile (7c). A
solution of 130 mg (0.44 mmol) of 2f was prepared in 2.0 mL
of DMF, and to this was added 18 mg (0.45 mmol) of sodium
hydride as a 60% emulsion in oil. The mixture was stirred
under an argon atmosphere at room temperature for 3 h. The
resulting yellow solution was then added to a stirring solution
of 150 mg (0.44 mmol) of 6c in 2.0 mL of DMF. The reaction
was then stirred for 6 h at room temperature under argon.
The product was precipitated by dropwise addition of the
reaction mixture to 100 mL of saturated aqueous NaH2PO3.
The pale yellow precipitate was then extracted into 300 mL
of ethyl acetate which was collected and rotary evaporated at
40 °C to yield a yellow solution of crude product in DMF.
Further rotary evaporation at 50 °C and 100 µmHg removed
the DMF to give a viscous yellow oil. This residue was eluted
from a silica gel flash column (30 cm × 2 cm) in 10% hexanes/
90% ethyl acetate. Removal of solvent by rotary evaporation
gave 200 mg (82%) of pure 7c as a pale yellow oil: 1H NMR
(500 MHz, CDCl3) δ 1.64 (12H, s, sal-2-(CH3)2, nil-4-(CH3)2),
4.16-4.19 (2H, m, OCH2C), 4.29-4.33 (2H, m, CCH2N), 4.53
(2H, s, Bn), 6.91 (1H, d, J ) 8 Hz, sal-8), 7.44 (1H, dd, J ) 8,
2 Hz, sal-7), 7.53 (1H, bs, NH), 7.87 (1H, d, sal-5), 7.21 (1H, d,
J ) 8 Hz, nil-6), 8.00 (1H, dd, J ) 8, 2 Hz, nil-5), 8.15 (1H, d,
J ) 2 Hz, nil-3); m/z 554.1757 [M+] (calculated for 554.1777).
5-{4-[3-(4-Cyan o-3-tr iflu or om eth ylph en yl)-5,5-dim eth yl-
2,4-d ioxo-im id a zolid in -1-yl]bu t-2-yn yloxym eth yl}-2-h y-
d r oxyben za m id e (8c). A solution of 200 mg of 7c (0.36 mmol)
was prepared in 10 mL of 20% water/80% MeOH, and a
catalytic amount of p-toluenesulfonic acid was added. The
reaction was refluxed under an argon atmosphere for 24 h at
90 °C. The solvent was then removed by rotary evaporation
at 40 °C, and the residue was transferred to a separatory
funnel in 200 mL of ethyl acetate. This solution was washed
2× with 40 mL portions of saturated brine containing 5%
NaHCO3. Th organic layer was then collected, dried over
anhydrous magnesium sulfate, and rotary evaporated at 40
°C to give a pale yellow oil. Elution from a silica gel flash
column (30 cm × 2 cm) in 5% hexanes/95% ethyl acetate
followed by rotary evaporation yielded a product of ap-
proximately 95% purity. Further purification by preparatory
HPLC, eluting with 20 mM pH 4.0 TEAA buffer, yielded 148
mg (80%) of 8c as a clear colorless oil: 1H NMR (500 MHz,
CDCl3) δ 1.63 (6H, s, nil-5-(CH3)2), 4.15 (2H, s, OCH2C), 4.29
(2H, s, CCH2N), 4.50 (2H, s, Bn), 5.90 (1H, bs, NH), 6.62 (1H,
bs, NH), 6.96 (1H, d, J ) 9 Hz, sal-3), 7.37 (1H, dd, J ) 9, 2
Hz, sal-4), 7.46 (1H, d, J ) 2 Hz, sal-6), 7.92 (1H, d, J ) 8 Hz,
nil-5), 7.97 (1H, dd, J ) 8, 2 Hz, nil-6), 8.11 (1H, d, J ) 2 Hz,

5268 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Cogan and Koch
nil-2), 12.30 (1H, s, 2-OH); m/z 514.1469 [M+] (calculated for
separatory funnel where it was washed 3× with 40 mL
portions of saturated brine containing 5% NaH2PO4. The
organic layer was collected, dried over anhydrous magnesium
sulfate, and rotary evaporated free of solvent at 40 °C to give
a pale brown oil. This residue was dissolved in 10 mL of ethyl
acetate and introduced to a silica gel flash column (20 cm × 2
cm) packed in the 100% ethyl acetate. Elution with 78% ethyl
acetate/20% methanol/2% triethylamine followed by removal
of the solvent by rotary evaporation at 40 °C gave a pale golden
oil. Further purification by preparatory HPLC using 20 mM
pH 4.0 TEAA buffer was required and yielded 121 mg (71%)
of 12a as a clear, colorless oil: 1H NMR (500 MHz, CDCl3) δ
1.49-1.60 (2H, m, (CH2)2NCH2CH2), 1.52 (6H, s, 5-(CH3)2),
1.67-1.76 (2H, m, CH2CH2NCO), 2.37 (2H, t, J ) 7 Hz, (CH2)2-
NCH2CH2), 2.49 (8H, bs, N(CH2CH2)2N), 2.61 (2H, t, J ) 6
Hz, BnOCH2CH2N) 3.36 (2H, t, J ) 8 Hz, CH2NCO), 3.56 (2H,
t, J ) 6 Hz, BnOCH2), 4.42 (2H, s, Bn), 6.23 (1H, bs, NH),
6.87 (1H, bs, NH), 6.91 (1H, d, J ) 8 Hz, sal-3), 7.32 (1H, d, J
) 8 Hz, sal-4), 7.48 (1H, s, sal-6), 7.90 (1H, d, J ) 9 Hz, nil-5),
7.99 (1H, dd, J ) 9, 2 Hz, nil-6), 8.14 (1H, d, J ) 2 Hz, nil-2);
m/z 631.2833 [MH+] (calculated for 631.2856).
514.1464); anal. (C25H21F3N4O5) C, H, N.
5-(2-Br om oeth oxym eth yl)-2-h yd r oxyben za m id e (9). A
solution of 150 mg (0.91 mmol) of 3 was prepared in 12 mL of
2-bromoethanol by heating a stirring mixture of the two to 55
°C under an argon atmosphere. The solution was then allowed
to cool for 5 min at room temperature before 56 mg (0.46 mmol)
of decaborane was added. Excessive evolution of H2 was
observed for 5 min, after which time the reaction was again
heated to 55 °C. After being stirred for 4 h, the solvent was
removed by rotary evaporation at 40 °C and 100 µmHg. The
residue was dissolved in 8 mL of ethyl acetate and introduced
to a silica gel flash column (30 cm × 2 cm) packed in 25%
hexanes/75% ethyl acetate. Elution with the same solvent
system yielded 175 mg (70%) of 9 as a clear, colorless oil: 1H
NMR (500 MHz, (CD3)2CO) δ 3.55 (2H, t, J ) 6 Hz, OCH2-
CH2Br), 3.76 (2H, t, J ) 6 Hz, OCH2CH2Br), 4.47 (2H, s, Bn),
6.89 (1H, d, J ) 8 Hz, 3), 7.19 (1H, bs, NH), 7.44 (1H, dd, J )
8, 2 Hz, 4), 7.81 (1H, d, J ) 2 Hz, 6), 7.97 (1H, bs, NH), 12.93
(1H, s, 2-OH); m/z 273.007 [M+] (calculated for 273.001).
4-[3-(4-Br om o-bu tyl)-4,4-d im eth yl-2,5-d ioxo-im id a zoli-
d in -1-yl]-2-tr iflu or om eth ylben zon itr ile (10a ). To a stirring
solution of 306 mg (1.0 mmol) of 2f in 3.0 mL of DMF was
added 49 mg (1.2 mmol) of sodium hydride (60% in oil). The
resulting suspension was stirred at room temperature for 1.5
h at which time evolution of H2 had ceased and a yellow
solution persisted. To this solution was added 1.0 mL of 1,4-
dibromobutane, and the resulting reaction mixture was heated
under an argon atmosphere to 60 °C for 0.5 h. At this time,
the reaction mixture was added dropwise to 100 mL of
saturated brine containing 5% NaH2PO4. A pale yellow pre-
cipitate was formed which was extracted into 250 mL of ethyl
acetate. The organic layer was washed 2× with saturated
brine, collected, and rotary evaporated at 40 °C to a yellow
solution in DMF. Further rotary evaporation at 50 °C and 100
µmHg removed the DMF to yield a yellow oil. This crude
product was dissolved in 10 mL of 50% hexanes/50% ethyl
acetate and introduced to a silica gel flash column (30 cm × 2
cm) packed in 75% hexanes/25% ethyl acetate. Elution with
75% hexanes/25% ethyl acetate followed by removal of solvent
by rotary evaporation at 40 °C gave 378 mg (85%) of 10a as a
clear, colorless oil: 1H NMR (500 MHz, CDCl3) δ 1.54 (6H, s,
4-(CH3)2), 1.82-1.98 (2H, m, CH2CH2N), 1.82-1.98 (2H, m,
BrCH2CH2), 3.36-3.42 (2H, m, CH2CH2N), 3.43-3.49 (2H, m,
BrCH2CH2), 7.90 (1H, d, J ) 8 Hz, 6), 7.99 (1H, dd, J ) 8, 2
Hz, 5), 8.14 (1H, d, J ) 2 Hz, 3); m/z 431.0450 [M+] (calculated
for 431.0456).
4-{3-[2-(2-Br om oeth oxy)eth yl]-4,4-d im eth yl-2,5-d ioxo-
im id a zolid in -1-yl}-2-tr iflu or om eth ylben zon itr ile (10b).
10b was prepared as 10a in 82% yield: 1H NMR (500 MHz,
CDCl3) δ 1.55 (6H, s, 4-(CH3)2), 3.47 (2H, m, OCH2CH2N), 3.56
(2H, m, OCH2CH2N), 3.76, (2H, m, BrCH2CH2O), 3.81 (2H,
m, BrCH2CH2O), 7.91 (1H, d, J ) 9 Hz, 6), 8.00 (1H, dd, J )
9, 2 Hz, 5), 8.14 (1H, d, J ) 2 Hz, 3); m/z 447.0405 [M+]
(calculated for 447.0405).
4-{4,4-Dim e t h yl-2,5-d ioxo-3-[2-(2-p ip e r a zin -1-yle t h -
oxy)et h oxy]im id a zolid in -1-yl}-2-t r iflu or om et h ylben zo-
n itr ile (11b). 11b was prepared as 11a in 59% yield: 1H NMR
(500 MHz, CDCl3) δ 1.48 (6H, s, 4-(CH3)2), 2.45-2.60 (4H, bm,
(CH2)2NCH2), 2.51-2.56 (2H, m, (CH2)2NCH2), 2.86-2.98 (4H,
bm, HN(CH2)2), 3.46-3.53 (2H, m, CH2OCH2), 3.55 (2H, t, J
) 6 Hz, CH2NCO), 3.61-3.66 (2H, m, CH2OCH2), 5.20 (1H,
bs, NH), 7.87 (1H, d, J ) 8 Hz, 6), 7.95 (1H, dd, J ) 8, 2 Hz,
5), 8.09 (1H, d, J ) 2 Hz, 3); m/z 452.1897 [M-H] (calculated
for 452.1909).
5-{2-[4-(2-{2-[3-(4-Cya n o-3-tr iflu or om eth yl-p h en yl)-5,5-
d i m e t h y l -2 ,4 -d i o x o -i m i d a z o l i d i n -1 -y l ] e t h o x y }-
et h ylp ip er a zin e-1-yl]et h oxym et h yl}-2-h yd r oxyb en za -
m id e (12b). 12b was prepared as 12a in 83% yield: 1H NMR
(500 MHz, CDCl3) δ 1.53 (6H, s, 5-(CH3)2), 2.45-2.68 (8H, bm,
N(CH2CH2)2N), 2.54-2.61 (2H, m, (CH2)2NCH2CH2O), 2.63-
2.90 (2H, m, BnOCH2CH2N), 3.50-3.63 (6H, m, BnOCH2, CH2-
OCH2CH2NCO), 3.65-3.71 (2H, m, OCH2CH2NCO), 4.43 (2H,
s, Bn), 6.13 (1H, bs, NH), 6.92 (1H, d, J ) 8 Hz, sal-3), 7.31
(1H, dd, J ) 8, 2 Hz, sal-4), 7.57 (1H, d, J ) 2 Hz, sal-6), 7.90
(1H, d, J ) 8 Hz, nil-5), 8.00 (1H, dd, J ) 8, 2 Hz, nil-6), 8.14
(1H, d, J ) 2 Hz, nil-2); m/z 647.2816 [MH+] (calculated for
647.2805).
N-(5-{4-[3-(4-Cya n o-3-t r iflu or om et h ylp h en yl)-5,5-d i-
m et h yl-2,4-d ioxo-im id a zolid in -1-yl]bu t -2-yn yloxym et h -
yl)-2-h yd r oxyben za m id om eth yl)d oxor u bicin (13). To a
stirring solution of 20 mg of 8c (0.04 mmol) in 2.0 mL of DMF
was added 10 µL of a 37% formalin solution (0.13 mmol). The
reaction was stirred in a screw top vial for 15 min at 55 °C, at
which time 20 mg (0.03 mmol) of doxorubicin hydrochloride
was added to form a red suspension which was stirred at 55
°C. After 15 min, a clear red solution had formed and the
reaction was removed from the heat. Transfer of the solution
to a 250 mL round-bottom flask, followed by rotary evaporation
of the solvent at 55 °C and 50 µmHg gave a red film which
was readily dissolved in 20 mL of methanol containing 30% of
20 mM pH 1.9 1% TFA. After 10 min at room temperature,
the methanol was removed by rotary evaporation at 30 °C,
and the resulting aqueous suspension was diluted to 100 mL
with saturated brine and transferred to a separatory funnel.
Extraction into 50 mL of chloroform followed by addition of 1
mL of glacial acetic acid and rotary evaporation at 30 °C gave
a red film. The product was then dissolved in 1-2 mL of
methanol and filtered through a 0.45 µm Spin-X centrifuge
filter. Purification was achieved by preparative HPLC using
a pH 3.5 TEAA buffer as the aqueous eluent. Pure material
was collected into a test tube (100 mm × 10 mm) containing
0.5 mL of 1.0 M HCl. Acetonitrile was removed by rotary
evaporation at 30 °C to yield an aqueous suspension of the
pure product which was diluted to 50 mL with saturated brine
and transferred to a separatory funnel. Extraction into 50 mL
of chloroform followed by addition of 1 mL of glacial acetic acid
and rotary evaporation at 30 °C gave 23 mg (60%) of 13 as
the acetate salt: 1H NMR (500 MHz, (CD3)2CO) δ 1.32 (3H, d,
J ) 6 Hz, dox-5′-CH3), 1.62 (6H, s, 5-(CH3)2), 2.15-2.30 (3H,
m, 2(dox-2′), dox-8), 2.42 (1H, d, J ) 14 Hz, dox-8), 2.93 (1H,
d, J ) 18 Hz, dox-10), 3.12 (1H, J ) 18 Hz, dox-10), 3.85-4.0
(1H, bm, dox-3′), 4.06 (3H, s, dox-4-OCH3), 4.12 (3H, s, dox-
9OH, CCH2NCO), 4.38 (5H, s, dox-5′, Bn, BnOCH2C), 4.62-
4.78 (2H, m, dox-14), 4.91 (1H, bs, NCH2N), 5.03 (1H, bs,
NCH2N), 5.21 (1H, s, dox-7), 5.56 (1H, s, dox-1′), 6.71 (1H, d,
J ) 8 Hz, sal-3), 7.29 (1H, d, J ) 8 Hz, sal-4), 7.63 (1H, d, J
) 8 Hz, dox-3), 7.74 (1H, bs, sal-6), 7.90 (1H, t, J ) 8 Hz, dox-
5-[2-(4-{4-[3-(4-Cya n o-3-t r iflu or om et h yl-p h en yl)-5,5-
d im eth yl-2,4-d ioxo-im id a zolid in -1-yl]bu tyl}p ip er a zin -1-
yl)eth oxym eth yl]-2-h yd r oxyben za m id e (12a ). To a stir-
ring solution of 120 mg (0.27 mmol) of 11a and 80 mg (0.29
mmol) of 9 in 2.0 mL of THF was added 100 µL (0.72 mmol) of
triethylamine. The resulting solution was refluxed under an
argon atmosphere for 20 h, at which time the solvent was
removed by rotary evaporation at 40 °C. The residue was
dissolved in 250 mL of ethyl acetate and transferred to a

Anti-Androgen Ligands for Tumor-Specific Delivery
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24 5269
2), 7.96 (1H, d, J ) 8 Hz, dox-1), 8.14 (1H, d, J ) 8 Hz, nil-5),
8.20 (1H, d, J ) 9 Hz, nil-6), 8.25 (1H, s, nil-2), 10.32 (1H, bs,
NH), 11.82 (1H, bs, sal-2-OH), 13.26 (1H, s, dox-6/11-OH),
14.18 (1H, s, dox-6/11-OH); m/z 1092.3101 [MNa+] (calculated
for 1092.3097).
the total binding determined for each concentration. Compari-
son to the specific binding for the positive control, in which
no competing ligand was incubated with the 3H-Mibolerone,
yielded the percent of 3H-Mibolerone displaced by a given
concentration of test ligand. The IC50 values for each test
ligand were calculated by Logit-log(pseudo-Hill) analysis.
Ra d ioliga n d Com p etition AR Bin d in g Assa y. PC3/AR
or PC3/neo cells were grown in RPMI 1640 medium to
approximately 80% confluency in five Nunc T-175 flasks.
Growth medium in each flask was then replaced with 50 mL
of phenol red-free RPMI 1640 supplemented with 10% dextran-
coated charcoal-stripped FBS, and the cells were grown for
an additional 18-22 h. Two hours prior to harvesting, the
growth medium was again replaced with fresh phenol red-free,
charcoal-stripped RPMI. The cells were then washed with 10
mL of Hank’s balanced salt solution and dissociated with
trypsin. Trypsin was quenched with phenol red-free, charcoal
stripped RPMI, and the combined cells from each flask were
centrifuged in a 50 mL conical tube at 100 x g for 5 min. The
cells were then resuspended in 50 mL of phenol red-free,
charcoal-stripped RPMI and counted at this concentration.
Centrifugation at 100 x g gave approximately 1 mL of cells
which were resuspended in 5 mL of 4 °C lysis buffer (10 mM
Tris, 1.5 mM EDTA, 0.5 mM DTT, 10 mM NaMoO4, 1.0 mM
PMSF, 10% v/v glycerol) supplemented immediately before use
with Complete-mini protease inhibitor cocktail. Cells were
lysed via sonication at 4 °C with a microtip, set at maximum
power, for 10 cycles of 6 s on and 24 s off. The cytosolic fraction
of the lysate was isolated by ultracentrifugation at 4 °C and
225 000 x g for 45 min. The centrifuged samples were
dispensed into 100 µL aliquots and stored at -78 °C until used.
Total protein was quantified either in fresh or frozen aliquots
by the Sigma BSA micro protein determination method ac-
cording to the prescribed protocol.
Ack n ow led gm en t. We are grateful to the U.S.
Army Prostate Cancer Research Program (DAMD17-
01-1-0046) and the National Cancer Institute of the NIH
(CA-92107) for financial support and the National
Science Foundation for help with the purchase of NMR
equipment (CHE-0131003). T.H.K. thanks the Univer-
sity of Colorado Council for Research and Creative Work
for a Faculty Fellowship.
Su p p or tin g In for m a tion Ava ila ble: 1H NMR spectra for
all products and isolated intermediates; semilogarithmic IC50
determinations. This material is available free of charge via
the Internet at http://pubs.acs.org.
Refer en ces
(1) Minotti, G.; Cairo, G.; Monti, E. Role of iron in anthracycline
cardiotoxity: new tunes for an old song? FASEB J . 1999, 13,
199-212.
(2) Lowenthal, R. M.; Eaton, K. Toxicity of chemotherapy. Hematol.
Oncol. Clin. N. 1996, 10, 967.
(3) Acres, B.; Beverley, P.; Scholl, S. Tumor immunology and the
battle of Waterloo. Mol. Cancer. Ther. 2002, 1, 651-655.
(4) Folkman, J . Tumor angiogenesis: therapeutic implications. N.
Eng. J . Med. 1971, 285, 1182-1186.
(5) Madhusudan, S.; Harris, A. L. Drug inhibition of angiogenesis.
Curr. Opin. Pharm. 2002, 2, 403-414.
Aliquots of cell lysate were used fresh or thawed at 4 °C.
Stock solutions of 100x working concentration of the test
ligands, 3H-Mibolerone and unlabled Mibolerone were pre-
pared in DMSO and subsequently diluted to 10× in lysis
buffer. Concentrations of test compounds were determined
spectrophotometricly in DMSO by either absorbance at 310
nm for salicylamide containing molecules (ꢀ310 ) 3580 L/(mol
× cm); as determined from a Beer-Lambert plot described by
varying concentrations of 8a ), 264 nm for 2f (ꢀ264 ) 13000
L/(mol × cm)), or 276 nm for nilutamide (ꢀ276 ) 4620 L/(mol ×
cm)). Aliquots of cell lysate were complemented with 10 µL of
10× ligand solutions and 10 µL of the 10× 3H-Mibolerone
solution to yield concentrations of 1, 10, 100, and 1000 nM
test compound and 1 nM 3H-Mibolerone. Each reaction was
prepared in duplicate to yield eight total test assays. Duplicate
positive controls, consisting of 10 µL of lysis buffer in place of
a test ligand (total radioligand binding), and negative controls,
consisting of 1000 nM unlabled Mibolerone (nonspecific bind-
ing), each in the presence of 1 nM 3H-Mibolerone, were
prepared. The reactions were gently mixed and briefly cen-
trifuged before incubating at 4 °C for 30 min. After incubation
was complete, 100 µL of each reaction was introduced to 400
µL of ice cold hydroxyapatite (HA), as a 60% suspension in
pH 7.4 Tris buffer, on a 0.45 µm nylon filter in a Spin-X
centrifuge tube. Upon addition of the reaction solution, the
tubes were closed, briefly vortexed, and allowed to incubate
on ice for 12 min with vortexing every 3-5 min. The HA
suspensions were then centrifuged at 1200 x g for 10 min. The
filtrate was discarded and the dry pellet was resuspended in
400 µL of pH 7.3 20 mM Tris wash buffer containing 0.1%
Triton-X100. Following seven rounds of resuspension and
subsequent centrifugation, the final filtrate was discarded, and
the dry pellet was centrifuged for an additional 15 min. The
pellet and filter bucket for each sample were then transferred
to 20 mL scintillation vials and 4 mL of scintillation cocktail
was added to each. Vortexing for 30 s thoroughly mixed the
pellet with the scintillation liquid before counting. Each sample
was counted for five repetitions of 3 min counts. This counting
protocol was then repeated two additional times to ensure
precision. Specific binding for each test concentration was
determined by subtracting the nonspecific binding control from
(6) Folberg, R.; Hendrix, M. J . C.; Maniotis, A. J . Vasculogenic
mimicry and tumor angiogenesis. Am. J . Pathol. 2000, 156, 361-
381.
(7) Sawyers, C. L. Rational therapeutic intervention in cancer:
kinases as drug targets. Curr. Op. Genet. Dev. 2002, 12, 111-
115.
(8) Bignold, L. P. The mutator phenotype theory can explain the
complex morphology and behaviour of cancers. Cell Mol. Life Sci.
2002, 59, 950-958.
(9) Gorre, M. E.; Mohammed, M.; Ellwood, K.; Hsu, N.; Paquette,
R. et al. Clinical resistance to STI-571 cancer therapy caused
by BCR-ABL gene mutation or amplification. Science 2001, 293,
876-880.
(10) Carroll, F. I.; Philip, A.; Blackwell, J . T.; Taylor, D. J .; Wall, W.
E. Antitumor and antileukemic effects of some steroids and other
biologically interesting compounds containing an alkylating
agent. J . Med. Chem. 1972, 15, 1158-1161.
(11) Rank, P.; Peter, R.; Depenbrock, H.; Eisenbrand, G.; Schmid, P.
et al. Preclinical activity of 17â-[N-[N′-(2-chloroethyl)-N′-nitroso-
carbamoyl]-L-alanyl]-5R-dihydrotestosterone (E91) against tu-
mour colony forming units and haematopoietic progenitor cells.
Eur. J . Cancer. 1999, 35, 1009-1013.
(12) Kaleagasioglu, F.; Berger, M. R.; Schma¨hl, D.; Eisenbrand, G.
Chemosensitivity of human breast carcinoma cell lines to 1-(2-
chloroethyl)-1-nitroso-3-(2-hydroxyethyl) urea linked to 4-
acetoxybisdesmethyltamoxifen, estradiol or dihydrotestosterone.
Tr. J . Med. Sci. 1994, 21.
(13) Fernandez, A.; Van derpoorten, K.; Dasnois, L.; Lebtahi, K.;
Dubois, V. et al. N-Succinyl-(â-alanyl-L-leucyl-L-alanyl-L-
leucyl)doxorubicin: an extracellularly tumor-activated prodrug
devoid of intervenous acute toxicity. J . Med. Chem. 2001, 44,
3750-3753.
(14) Nagy, A.; Armatis, P.; Cai, R.; Szepeshazi, K.; Halmos, G. et al.
Design, synthesis, and in vitro evaluation of cytotoxic analogues
of bombesin-like petides containing doxorubicin or its intensely
potent derivative, 2-pyrrolinodoxorubicin. Proc. Natl. Acad. Sci.
U.S.A. 1997, 94, 652-656.
(15) de Groot, F. M. H.; Broxterman, H. J .; Adams, H. P. H. M.; van
Vliet, A.; Tesser, G. I. et al. Design, synthesis, and biological
evaluation of a dual tumor-specific motive containing integrin-
targeted plasmin-cleavable doxorubicin prodrug. Mol. Cancer
Ther. 2002, 1, 901-911.
(16) Guillemard, V.; Saragovi, H. U. Taxane-antibody conjugates
afford potent cytotoxicity, enhanced solubility, and tumor target
selectivity. Cancer Res. 2001, 61, 694-699.
(17) Bradley, M. O.; Swindell, C. S.; Anthony, A. C.; Donehower, R.
C. Tumor targeting by conjugation of DHA to paclitaxel. J .
Controlled Release 2001, 74, 233-236.

5270 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 24
Cogan and Koch
(18) Ishiki, N.; Onishi, H.; Machida, Y. Biological properties of
conjugates of mitomycin C with estradiol benzoate and estra-
diol: their stability characteristics in biological media and their
binding abilities to estrogen receptor. Biol. Pharm. Bull. 1997,
20, 1096-1102.
(19) Ellerby, H. M.; Arap, W.; Ellerby, L. M.; Kain, R.; Andrusiak,
R. et al. Anti-cancer activity of targeted pro-apoptotic peptides.
Nat. Med. 1999, 5, 1032-1038.
(20) Kuduk, S. D.; Harris, C. R.; Zheng, F. F.; Sepp-Lorenzino, L.;
Ouerfelli, Q. et al. Synthesis and evaluation of geldanamycin-
testosterone hybrids. Bioorg. Med. Chem. Lett. 2000, 10, 1303-
1306.
(21) McGavin, J . K.; Spencer, C. M. Gemtuzumab ozogamicin. Drugs
2001, 61, 1317-1322.
(22) Linja, M. J .; Savinainen, K. J .; Sarama¨ki, O. R.; Tammela, T.
L. J .; Vessella, R. L. et al. Amplification and overexpression of
androgen receptor gene in hormone-refractory prostate cancer.
Cancer Res. 2001, 61, 3550-3555.
(23) Fenick, D. J .; Taatjes, D. J .; Koch, T. H. Doxoform and Dauno-
form: Anthracycline-Formaldehyde Conjugates Toxic to Resis-
tant Tumor Cells. J . Med. Chem. 1997, 40, 2452-2461.
(24) Dernell, W. S.; Powers, B. E.; Taatjes, D. J .; Cogan, P.; Gaudiano,
G. et al. Evaluation of the epidoxorubicin-formaldehyde conju-
gate, epidoxoform, in a mouse mammary carcinoma model.
Cancer Invest. 2002, 20, 712-723.
(25) Cutts, S. M.; Rephaeli, A.; Nudelman, A.; Hmelnitsky, I.; Phillips,
D. R. Molecular basis for the synergistic interaction of adria-
mycin with the formaldehyde-releasing prodrug pivaloyloxy-
methyl butyrate (AN-9). Cancer Res. 2001, 61, 8194-8202.
(26) Cogan, P. S.; Fowler, C. R.; Post, G. C.; Koch, T. H. Doxsali-
form: a novel N-Mannich base prodrug of a doxorubicin form-
aldehyde conjugate. Lett. Drug Design Discovery, submitted.
(27) Brys, M. Androgens and androgen receptor: do they play a role
in breast cancer? Med. Sci. Monit. 2000, 6, 433-438.
(28) Hall, R.; Clements, J .; Birrell, S.; Tilley, W. Prostate-specific
antigen and gross cystic disease fluid protein-15 are coexpressed
in androgen receptor-positive breast tumors. Br. J . Cancer. 1998,
78, 360-365.
(29) Engehausen, D. G.; Tong, X.; Oehler, M. K.; Freund, C. T.;
Schrott, K. M. et al. Androgen receptor gene mutations do not
occur in ovarian cancer. Anticancer Res. 2000, 20, 815-819.
(30) Tihan, T.; Harmon, J . W.; Wan, X.; Younes, Z.; Nass, P. et al.
Evidience of androgen receptor expression in squamous and
adenocarcinoma of the esophagus. Anticancer Res. 2001, 21,
3107-3114.
for the androgen receptor: 7R-methyl-17R-(2′-(E)-iodovinyl)-19-
nortestosterone. J . Med. Chem. 1991, 34, 1019-1024.
(41) Stobaugh, M. E.; Blickenstaff, R. T. Synthesis and androgen
receptor binding of dihydrotestosterone hemisuccinate homo-
logues. Steroids 1990, 55, 259-262.
(42) Stobaugh, M. E.; Blickenstaff, R. T. Dihydrotestosterone deriva-
tives: relative binding affinity versus affinity purification.
Steroids 1991, 56, 581-585.
(43) van Dort, M. E.; Robins, D. M.; Wayburn, B. Design, synthesis,
and pharmacological characterization of 4-[4,4-dimethyl-3-(4-
hydroxybutyl)-5-oxo-2-thioxo-1-imidazolidinyl]-2-iodobenzoni-
trile as a high-affinity nonsteroidal androgen receptor ligand.
J . Med. Chem. 2000, 43, 3344-3347.
(44) van Dort, M. E.; J ung, Y. Synthesis and structure-activity
studies of side-chain derivitized arylhydantoins for investigation
as androgen receptor radioligands. Bioorg. Med. Chem. Lett.
2001, 11, 1045-1047.
(45) Battmann, T.; Bonfils, A.; Branche, C.; Humbert, J .; Goubet, F.
et al. RU 58841, anew specific topical antiandrogen: a candidate
of choice for the treatment of acne, androgenetic alopecia and
hirsutism. J . Steroid Biochem. Mol. Biol. 1994, 48, 55-60.
(46) Raynaud, J . P.; Bonne, C.; Bouton, M. M.; Lagace, L.; Labrie, F.
Action of a nonsteroidal anti-androgen, RU 23908, in peripheral
and cantral tissues. J . Steroid Biochem. 1979, 11, 93-99.
(47) Raynaud, J . P.; Bonne, C.; Moguilewsky, M.; Lefebvre, F. A.;
Belanger, A. et al. The pure antiandrogen RU23908 (Anandron),
a candidate of choice for the combined antihormonal treatment
of prostatic cancer: a review. Prostate 1984, 5, 299-311.
(48) Battmann, T.; Branche, C.; Bouchoux, F.; Cerede, E.; Philibert,
D. et al. Pharmacological profile of RU 58642, a potent systemic
antiandrogen for the treatment of androgen-dependent disorders.
J . Steroid. Biochem. Mol. Biol. 1998, 54, 103-111.
(49) Ukawa, K.; Imamiya, E.; Yamamoto, H.; Aono, T.; Kozai, O. et
al. Synthesis and antitumor activity of new amphiphilic alkyl-
glycerolipids substituted with a polar headgroup, 2, 2(trimethyl-
ammonioethoxy)ethyl or a congeneric oilgo(ethyleneoxy)ethyl
group. Chem. Pharm. Bull. 1989, 37, 3277-3285.
(50) Salomies, H.; Luukkansen, L.; Knuutila, R. Oxidation of â-block-
ing agents - VII. Periodate oxidation of Labetalol. J . Pharm.
Biomed. 1989, 7, 1447-1451.
(51) Lee, S. H.; Park, Y. J .; Yoon, C. M. Reductive etherification of
aromatic aldehydes with decaborane. Tetrahedron Lett. 1999,
40, 6049-6050.
(52) Fuhrer, W.; Ostermayer, F.; Zimmermann, M.; Meier, M.; Mu¨ller,
H. â-Adrenergic blocking agents: substituted phenylalkanol-
amines. Effects of side-chain length on â-blocking potency in
vitro. J . Med. Chem. 1984, 27, 831-836.
(53) Avasthi, K.; Farooq, S. M.; Sharon, A.; Maulik, P. R. Disappear-
ance of inter-and intramolecular stacking due to one-atom
addition in ‘propylene-linker’ in a pyrazolo[3, 4-d]pyrimidine-
based flexible molecule. Acta Crystallogr. E 2002, E58, o940-
o941.
(54) Matias, P. M.; Donner, P.; Coelho, R.; Thomaz, M.; Peixoto, C.
et al. Structural evidence for ligand specificity in the binding
domain of the human androgen receptor. J . Biol. Chem. 2000,
275, 26164-26171.
(55) Shen, R.; Sumitomo, M.; Dai, J .; Harris, A.; Kaminetzky, D. et
al. Androgen-induced growth inhibition of androgen receptor
expressing androgen-independent prostate cancer cells is medi-
ated by increased levels of neutral endopeptidases. Endocrinol-
ogy 2000, 141, 1699-1704.
(56) Dai, J . L.; Maiorino, C. A.; Gkonus, P. J .; Burnstein, K. L.
Androgenic up-regulation of androgen receptor cDNA expression
in androgen-independent prostate cancer cells. Steroids 1996,
61, 531-539.
(57) Wakeling, A. E.; Furr, B. J . A.; Glen, A. T.; Hughes, L. R.
Receptor binding and biological activity of steroidal and non-
steroidal antiandrogens. J . Steroid Biochem. 1981, 15, 355-359.
(58) Beheshti, B.; Vukovic, B.; Marrano, P.; Squire, J . A.; Park, P.
C. Resolution of genotypic heterogeneity in prostate tumors
using polymerase chain reaction and comparative genomic
hybridization on microdissected carcinoma and prostatic in-
traepithelial neoplasia foci. Cancer Genet. CytoGen. 2002, 137,
15-22.
(59) Steinsapir, J .; Mora, G.; Muldoon, T. G. Effects of steroidal and
nonsteroidal antiandrogens on the androgen binding properties
of the rat andogen receptor. BBA-Mol. Cell. Res. 1991, 1094,
103-112.
(60) Liao, S.; Witte, D.; Schilling, K.; Chang, C. The use of hydroxy-
lapatite-filter steroid receptor assay method in the study of the
modulation of androgen receptor interaction. J . Steroid Biochem.
1984, 20, 11-17.
(31) Kaiser, U.; Hofmann, J .; Schilli, M.; Wegmann, B.; Klotz, U. et
al. Steroid-hormone receptors in cell lines and tumor biopsies
of human lung cancer. Int. J . Cancer. 1996, 67, 357-364.
(32) Adiga, S. K.; Andritsch, I.; Rao, R. V.; Krishan, A. Androgen
receptor expression and DNA content of paraffin-Embedded
archival human prostate tumors. Cytometry 2002, 50, 25-30.
(33) Georget, V.; Lobaccaro, J . M.; Terouanne, B.; Mangeat, p.;
Nicolas, J . et al. Trafficking of the androgen receptor in living
cells with fused green fluorescent protein-androgen receptor.
Mol. Cell Endocrinol. 1997, 129, 17-26.
(34) Veldscholte, J .; A., B. C.; Brinkman, A. O.; Grootegoed, A.;
Mulder, E. Anti-androgens and the mutated androgen receptor
of LNCaP cells: differential effects on binding affinity, heat-
shock protein interaction, and transcription activation. Biochem-
istry 1992, 31, 2393-2399.
(35) Tyagi, R. K.; Lavorvsky, Y.; Ahn, S. C.; Song, C. S.; Chatterjee,
B. et al. Dynamics of intercellular movement and nucleocyto-
plasmic recycling of the ligand-activated androgen receptor in
living cells. Mol. Endocrinol. 2000, 14, 1162-1174.
(36) McMenemin, R.; Macdonald, G.; Moffat, L.; Bissett, D. A phase
II study of Caelyx (liposomal doxorubicin) in metastatic carci-
noma of the prostate: tolerability and efficacy modification by
liposomal encapsulation. Invest. New Drugs 2002, 20, 331-337.
(37) Dubois, V.; Dasnois, L.; Ledtahi, K.; Collot, F.; Heylen, N. et al.
CPI-0004Na, a new extracellularly tumor-activated prodrug of
doxorubicin: in vivo toxicity, activity, and tissue distribution
confirm tumor cell selectivity. Cancer Res. 2002, 62, 2327-2331.
(38) Garsky, V. M.; Lumma, P. K.; Feng, D.; Wai, J .; Ramjit, H. G.
et al. The synthesis of a prodrug of doxorubicin to provide
reduced systemic toxicity and greater target efficacy. J . Med.
Chem. 2001, 44, 4216-4224.
(39) Marhefka, C. A.; Moore, B. M.; Bishop, T. C.; Kirkovsky, L.;
Mukherjee, A. et al. Homology modeling using multiple molec-
ular dynamics simulations and docking studies of the human
androgen receptor ligand binding domain bound to testosterone
and nonsteroidal ligands. J . Med. Chem. 2001, 44, 1729-1740.
(40) Salman, M.; Chamness, G. C. A potential radioiodinated ligand
J M0303305