Novel fluorinated prodrugs for activation by carboxypeptidase G2 showing good in vivo antitumor activity in gene-directed enzyme prodrug therapy

Article abstract of DOI:10.1021/jm0502182

Sixteen novel polyfluorinated benzoic acid mustards have been synthesized for use in gene-directed enzyme prodrug therapy (GDEPT). Eight of these were benzoic acid L-glutamate mustards for evaluation as prodrugs and the other eight were the active drugs formed by the action of the bacterial enzyme carboxypeptidase G2 (CPG2). All of the di- and trifluorinated prodrugs were efficiently cleaved by the enzyme. In contrast, the tetrafluorinated prodrugs were found to be competitive inhibitors of CPG2, the first such inhibitors to have been described. The di- and trifluorinated prodrugs were differentially cytotoxic to human breast carcinoma cells (MDA MB 361) expressing CPG2, compared to control cells that did not express the enzyme. The difluorinated prodrug {4-[bis(2-bromoethyl)amino]-3,5-difluorobenzoyl}-L-glutamic acid and its iodoethylamino analogue were effective substrates for the enzyme and showed excellent therapeutic activity in CPG2-expressing MDA MB 361 xenografts, either curing or greatly inhibiting tumor growth and extending the life of the animals.

Full text of DOI:10.1021/jm0502182

J. Med. Chem. 2005, 48, 5321-5328
5321
Novel Fluorinated Prodrugs for Activation by Carboxypeptidase G2 Showing
Good in Vivo Antitumor Activity in Gene-Directed Enzyme Prodrug Therapy
Lawrence C. Davies, Frank Friedlos, Douglas Hedley, Jan Martin, Lesley M. Ogilvie, Ian J. Scanlon, and
Caroline J. Springer*
Cancer Research UK Centre for Cancer Therapeutics at the Institute of Cancer Research, 15 Cotswold Road,
Sutton, Surrey, SM2 5NG, United Kingdom
Received March 9, 2005
Sixteen novel polyfluorinated benzoic acid mustards have been synthesized for use in gene-
directed enzyme prodrug therapy (GDEPT). Eight of these were benzoic acid L-glutamate
mustards for evaluation as prodrugs and the other eight were the active drugs formed by the
action of the bacterial enzyme carboxypeptidase G2 (CPG2). All of the di- and trifluorinated
prodrugs were efficiently cleaved by the enzyme. In contrast, the tetrafluorinated prodrugs
were found to be competitive inhibitors of CPG2, the first such inhibitors to have been described.
The di- and trifluorinated prodrugs were differentially cytotoxic to human breast carcinoma
cells (MDA MB 361) expressing CPG2, compared to control cells that did not express the enzyme.
The difluorinated prodrug {4-[bis(2-bromoethyl)amino]-3,5-difluorobenzoyl}-L-glutamic acid and
its iodoethylamino analogue were effective substrates for the enzyme and showed excellent
therapeutic activity in CPG2-expressing MDA MB 361 xenografts, either curing or greatly
inhibiting tumor growth and extending the life of the animals.
Scheme 1. Hydrolysis of Prodrug 1 by
Carboxypeptidase G2
Introduction
Gene-directed enzyme prodrug therapy (GDEPT) is
a two-step approach to the therapy of cancer.1 The
foreign gene for a prodrug-activating enzyme is selec-
tively introduced into the tumor by a vector, such as a
virus or liposome, by either local or systemic delivery.
Time is allowed for the enzyme to be expressed in the
tumor, and then a nontoxic prodrug is administered.
This prodrug is converted by the enzyme within the
tumor to a toxic drug. This localized activation reduces
the normal tissue toxicity associated with conventional
cytotoxic chemotherapy. A number of enzyme/prodrug
systems have been evaluated for GDEPT.2 In our
laboratory, we have concentrated on the enzyme car-
boxypeptidase G2 (CPG2, glutamate carboxypeptidase,
EC 3.4.17.11) derived from Pseudomonas RS16. This
enzyme catalyses the cleavage of amide bonds between
a carboxyl-, phenol-, or aniline-substituted aromatic ring
and L-glutamic acid, an activity unknown in any mam-
malian enzyme. Our GDEPT system has significant
benefits in that prodrugs can be designed that release
several different types of effector molecule, including
aromatic mustards and antibiotics. Moreover, the active
cytotoxic compound is released from the prodrug in a
single step without requiring subsequent processing by
host enzymes that may be lacking in tumor cells. The
benzoic acid mustard-derived prodrug (2-chloroethyl)-
(2-(methylsulfonyloxyethyl)amino)aminobenzyl-L-glutam-
ic acid (1) is a substrate for CPG2 that releases the
cytotoxic alkylating agent 4-(2-chloroethyl)(2-(methyl-
sulfonyloxyethyl)amino)benzoic acid (1a) (Scheme 1).
Prodrug 1 has been shown to be effective both in vitro
and in vivo against tumor cells expressing CPG2,
especially when the enzyme had been engineered to be
expressed externally, tethered to the outer membrane
of the cell.3 When CPG2 was expressed in the cytoplasm,
the efficacy was much reduced (Table 1). This has been
shown to be due to poor transport of the prodrug
through the cell membrane.3 We have synthesized novel
benzoic acid prodrugs (21-28) for use in GDEPT with
di-, tri-, and tetrafluorinated benzene rings. These
prodrugs are designed to be more lipophilic than 1 with
the aim of improving intracellular access and efficacy.
The substitution of a single fluorine atom at the 3-posi-
tion of the aromatic ring has been shown previously by
our group to improve efficacy in ADEPT (antibody-
directed enzyme prodrug therapy).4 The prodrugs were
tested for their chemical half-life, as substrates of CPG2,
and for cytotoxicity against a human breast carcinoma
cell line (MDA MB 361) expressing CPG2, either cyto-
plasmically or tethered to the outer surface of the cell
membrane. Prodrugs active in the latter assay were
tested for therapeutic efficacy in xenografts of the same
cell line, growing in nude mice.
* To whom correspondence should be addressed. Phone: 44 20
73528133x4214. Fax: 44 20 87224046. E-mail: caroline.springer@
icr.ac.uk.
10.1021/jm0502182 CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/12/2005

5322 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16
Davies et al.
Table 1. Cytotoxicity of Prodrugs and Drugs to the Three MDA MB361 Cell Linesb
IC50, µM
differentiala
stCPG2(Q)3
no.
stCPG2(Q)3
CPG2*
LacZ
CPG2*
prodrugs
1
23
25
28
22
24
27
21
26
1653 ((13)
26 (18-38)
14 (7-29)
26 (16-42)
46 (30-70)
50 (31-82)
173 (104-288)
>4000
18803 ((128)
42 (25-69)
53 (26-109)
39 (22-66)
73 (35-153)
82 (28-239)
671 (384-1180)
>4000
31303 ((234)
>4000
19
1.7
>150
>280
127
>95
>75
86
>4000
3300 (2040-5330)
>4000
>87
>80
>23
N/A
N/A
>55
>49
> 6
N/A
N/A
>4000
>4000
>4000
>4000
>4000
>4000
drugs
48
50
53
47
49
52
46
51
25.9 (9.5-71)
21.1 (13-35)
12.7 (7.5-22)
57.8 (31-107)
30.0 (19-47)
24.7 (18-35)
378 (232-616)
60.5 (35-104)
22.0 (12-41)
20.6 (8.1-52)
11.9 (7-20)
57.9 (30-115)
35.1 (20-62)
34.4 (21-44)
533 (338-841)
66.7 (24-193)
23.8 (11-54)
23.4 (10-53)
6.0 (4.2-8.6)
47.6 (25-90)
44.9 (19-104)
29.6 (20-44)
447 (296-678)
57.7 (31-107)
1
1
1
1
1
1
1
1
1
1
0.5
1
1.5
1
1
1
a The differential is defined as (IC50 of the LacZ cell line)/(IC50 of the CPG2-expressing cell line). b The bracketed values indicate the
95% confidence intervals or the standard error of the mean.
Results and Discussion
size the bis-mesyl compounds using methane sulfonic
anhydride instead of the chloride. Following substitution
of the mesyl group by halide, we obtained excellent
yields of the bis-chloro, bis-bromo, and bis-iodo mustards
(11-18) without the need for intermediate chromatog-
raphy. In the final step the tert-butyl group was
removed by trifluoroacetic acid at room temperature to
give the desired prodrugs (21-28) as white crystalline
solids in yields of 80-100%. A preliminary report of the
synthesis of 28 has been published.6
The corresponding drugs (46-53) were made accord-
ing to Scheme 3. The commercially available polyflu-
orinated benzoyl chlorides (29-31) were reacted with
lithium tert-butoxide giving the tert-butyl esters (32-
34). The p-fluorine was then substituted with dietha-
nolamine to give the bis-hydroxyethylamino compounds
(35-37). The ester group is sufficiently electron with-
drawing to allow substitution of the p-fluorine, even in
the case of the 3,4,5-trifluorobenzoyl ester 34. The bis
(hydroxyethyl)amino compounds were then converted
to the bis-bromo, bis-chloro, and bis-iodo mustard esters
(38-45) as described above for the prodrugs. Acidic
hydrolysis with trifluoroacetic acid gave the drugs (46-
53) in 73-94% yield as white crystalline solids.
Biology and Biochemistry. We tested the eight new
prodrugs (21-28) in vitro for cytotoxicity against three
variants of the human breast carcinoma cell line MDA
MB 361 that had been engineered to express CPG2
either internally in the cytoplasm (CPG2*), externally
tethered to the outer cell membrane (stCPG2(Q)3), or
that had been engineered to express an irrelevant
enzyme (â-galactosidase) as a control (LacZ). All the
difluoro- and trifluorobenzoyl compounds (22-25, 27-
28) were selectively cytotoxic to both the CPG2*- and
stCPG2(Q)3-expressing cell lines, while the tetrafluo-
robenzoyl prodrugs (21, 26) showed no toxicity to any
of the cell lines even at 4 mM (Table 1). The compounds
most selective toward cells expressing CPG2 were the
three difluorobenzoyl compounds (23, 25, 28), with
differentials between the IC50 values in control â-gal-
expressing cells and in CPG2-expressing cells of >150×,
>280×, and 127× for stCPG2(Q)3-expressing cells and
Chemistry. Our initial strategy for the synthesis of
prodrugs (21-28) was to replace the p-fluorine of a
protected polyfluorinated benzoyl glutamate with di-
ethanolamine, followed by formation of the mustard
functionality and deprotection (Scheme 2). Reaction of
the polyfluorinated benzoyl chlorides (2-4) with di-tert-
butyl-L-glutamate gave the amides (5-7) in 88-98%
yield. Substitution of the p-fluorine of pentafluoroben-
zoyl-L-glutamate (5) by diethanolamine in DMA gave 8
in 82% yield. With 2,3,4,5-tetrafluorobenzoylglutamate
(6), substitution gave 9 in only 31% yield. With 3,4,5-
trifluorobenzoylglutamate (7), none of the desired prod-
uct 10 was obtained; even when the solution was
refluxed, only polar products were formed, probably due
to cleavage of the tert-butyl protecting groups under the
relatively harsh conditions. We reasoned that a strongly
electron withdrawing group would greatly increase the
reactivity of the p-fluorine to nucleophilic substitution.
The cyano group was chosen because of the relative ease
of its hydrolysis. Substitution of 3,4,5-trifluorobenzoni-
trile 19 with diethanolamine to form 4-(bis(2-hydroxy-
ethyl)amino)-3,5-difluorobenzonitrile readily occurred,
even at 50 °C. The cyano group was then hydrolyzed to
the carboxylic acid in 87% yield by sodium hydroxide
in aqueous ethanol, followed by acidic workup to form
4-(bis(2-hydroxyethyl)amino)-3,5-difluorobenzoic acid (20).
The benzoic acid 20 was then activated with diethyl
cyanophosphonate in DMF. Reaction with di-tert-butyl-
L-glutamate formed the amide bond to give the desired
di-tert-butyl-L-glutamate 10 in high yield. Protection of
the alcoholic hydroxy groups was found to be unneces-
sary.
In previous mustard syntheses,5 the mustard func-
tionality was introduced by reaction of the hydroxyl of
the hydroxyethyl group with methane sulfonyl chloride
to form a bis-mesyl compound, followed by nucleophilic
substitution with halide. This method works well in the
case of chloro mustards, but for bromo and iodo mus-
tards it is necessary to chromatographically purify the
bis-mesyl compounds in order to eliminate the possibil-
ity of contamination by chloride. We decided to synthe-

Prodrugs for Activation by Carboxypeptidase G2
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16 5323
Scheme 2
>95×, >75×, and 86× for CPG2*-expressing cells,
respectively. The bis(chloroethyl)amino- and bis(bromo-
ethyl)aminotrifluorobenzoyl compounds (22, 24) showed
cytotoxicity differentials of >87× and >80× for stCPG2-
(Q)3-expressing cells and >55× and > 49× for CPG2*-
expressing cells, respectively. The bis(iodoethyl)amino-
trifluorobenzoyl prodrug 27 was less cytotoxic toward
both types of CPG2-expressing cell, showing cytotoxicity
differentials between control â-gal-expressing cells and
stCPG2(Q)3-expressing cells or CPG2*-expressing cells
of >23× and >6×, respectively. All the new compounds
showed larger cytotoxicity differentials between control
cells and both CPG2*- and stCPG2(Q)3-expressing cells
than the literature compound 1. To determine the
reasons for the differences, we examined the kinetics
of the cleavage of these compounds by CPG2 and their
chemical half-lives.
The bis(chloroethyl)-, bis(bromoethyl)- and bis(iodo-
ethyl)amino difluoro prodrugs (23, 25, 28) and the
bis(chloroethyl)- and bis(bromoethyl)aminotrifluoro-
benzoyl prodrugs (22, 24) had Km values <3 µM and
Kcat values >700 and are thus better substrates than
the literature compound 1 (Table 2). The bis(iodoethyl)-
amino trifluorobenzoyl prodrug (27) has a Km of 2.8 µM
but a lower Kcat of 220. The poorer rate of turnover
probably explains the smaller cytotoxicity differential
between control and CPG2-expressing cells exhibited by
this compound. The two tetrafluorobenzoyl compounds
(21, 26) were not cleaved at all by CPG2, which explains
their lack of enhanced activity in CPG2-expressing cells.
As expected, all the drugs tested (46-53) were cyto-
toxic but showed no cytotoxicity differential between the
CPG2-expressing and â-gal-expressing lines (Table 1).
The cytotoxicity of the five prodrugs that are good
substrates (22-25, 28) toward the cell line expressing
CPG2 externally (stCPG2(Q)3) was comparable to that
of their corresponding drug, implying that complete
activation had occurred. In contrast, the cytotoxicity of
these same prodrugs toward the cell line expressing
CPG2 internally (CPG2*) was slightly lower, suggesting
that there is still, to some degree, a barrier to their
reaching the enzyme. Nevertheless, they are superior
in this respect to compound 1. The half-lives of all the
drugs and prodrugs, presumed to reflect the chemical
hydrolysis of the halogen in the mustard moiety, were
measured in the same buffer as that used for the
enzyme kinetics (Table 2). For a given extent of ring
fluorination, the comparative rate of hydrolysis between
different mustard halogens was bromo > iodo . chloro.
For a given mustard halogen series, the comparative
rate of hydrolysis between different degrees of ring
fluorination was difluoro > trifluoro > tetrafluoro.

5324 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16
Davies et al.
Scheme 3
Figure 1. Human breast carcinoma MDA MB 361 (stCPG2-
(Q)3) xenografts were treated with 22 (0), 23 (b), 24(2), 25
([), 27 (9), 28 (1), or nothing (/) according to the regime
detailed in Experimental Section. Relative tumor volumes are
plotted against time from establishment of tumors, and the
lines in each group are continued until the first death has
occurred. Two mice were tumor free for >150 days in the group
treated with prodrug 28.
gives a drug with a very long half-life (360 min) and
the latter is a poor substrate. The bis(chloroethyl)amino-
difluoro- and bis(bromoethyl)aminotrifluorobenzoyl com-
pounds (23, 24), both effective substrates but yielding
drugs with relatively long half-lives (90 and 9 min), gave
a substantial delay in the growth of the tumors. The
bis(bromoethyl)amino- and bis(iodoethyl)aminodifluo-
robenzoyl prodrugs (25, 28), both excellent substrates
for CPG2 and yielding drugs with short half-lives
(<5min), gave a sustained delay in the growth of the
tumors, and two mice treated with 28 were cured of
their tumors. The latter compound had no therapeutic
effect on a non-CPG2-expressing tumor (data not shown).
Table 2. Half-Lives of Prodrugs and Drugs and Enzyme
Kinetic Properties of Prodrugs
Km,
µM
Kcat
,
T1/2
,
T1/2,
prodrug
S-1
min
drug
min
Conclusions
1
23
25
28
22
24
27
21
26
3.41
1.71
1.18
1.74
2.6
5831
732
827
1150
1150
825
220
0
59
1a
48
50
53
47
49
52
46
51
30
91
2.5
4.2
We have synthesized eight novel polyfluorinated
prodrugs (21-28), along with their corresponding drugs
(46-53), for use in GDEPT. We have shown that our
fluorinated prodrugs are much more active than the
literature compound 1 toward cells in which CPG2 has
been engineered to be expressed in the cytoplasm and
speculate that this reflects improved access to the
enzyme. We have shown that several of our compounds
have a lower Km and higher Kcat than compound 1.
Within our series we find unsurprisingly that com-
pounds of less favorable kinetic properties are less
active. We also find that the tetrafluorobenzoyl prodrugs
(21, 26) were not substrates but were competitive
inhibitors, whereas the trifluorobenzoyl prodrugs (22,
24, 27) were moderate or good substrates. The only
difference between these two groups is the 6-fluorine
atom on the aromatic ring. Modeling of these compounds
in the active site of CPG2 will allow the design of more
potent inhibitors and substrates with improved kinetic
properties. These compounds could also serve as probes
for engineering improvements in CPG2. Two of the
difluorinated prodrugs (25, 28) demonstrated good
selectivity in vitro and were excellent substrates for
CPG2. They have been shown to have excellent thera-
peutic activity in a xenograft tumor model and will
make very good candidates for GDEPT.
294
5.5
9.0
1300
22.9
37.3
4940
120
360
1.1
8.9
13.4
2.8
N/A
N/A
1365
35
0
We tested both tetrafluorobenzoyl prodrugs (21, 26)
that were not substrates of CPG2 to see whether they
were inhibitors, and this proved to be the case, both
being competitive with methotrexate, the conventional
unit-definition substrate for this enzyme. The bis-
(chloroethyl)amino compound 21 was the more potent
inhibitor with a Ki of 127 µM, compared to 276 µM for
the bis(iodoethyl)amino compound, 26. This is the first
reported example of an inhibitor of CPG2 found to be
competitive with methotrexate. The only inhibitors of
CPG2 described in the literature7 are more potent but
are noncompetitive with respect to methotrexate. The
new compounds (21, 26) therefore have the potential to
be good models for designing a stable analogue to use
as a probe of the CPG2 substrate binding site.
The six active compounds were tested for therapeutic
efficacy against MDA MB 361 xenografts growing in
nude mice (Figure 1). The bis(chloroethyl)amino- and
bis(iodoethyl)aminotrifluorobenzoyl prodrugs (22, 27)
gave no delay in the growth of the tumors relative to
the untreated controls, probably because the former
Experimental Section
Chemistry. Preparative HPLC was performed on an Axxial
Chromatospac Prep 10 using Merck Kieselgel 60 (0.015-0.040)

Prodrugs for Activation by Carboxypeptidase G2
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16 5325
(Art. 15111). Melting points were determined on a Kofler
hotstage (Reichert Thermovar) melting point apparatus and
are uncorrected. EI, FAB, and LDI spectra were performed
by the School of Pharmacy, University of London. Electrospray
mass spectrum were carried out on a Waters/Micromass LCT
system. NMR spectra were determined in Me2SO-d6 on a
Bruker AC250 spectrometer at 25 °C.
-147.4 (2F, dd, F3, J2,3 ) 22 Hz, J3,6 ) 8 Hz); MS m/z (FAB)
669 (M + Na+), 647 (M+ + 1). Anal. (C16H16F4I2N2O5) C, H, N,
F, I.
Di-tert-butyl {4-[Bis(2-chloroethyl)amino]-2,3,5,6-tet-
rafluorobenzoyl}-L-glutamate (11). Hydroxy compound 8
was reacted with methane sulfonic anhydride, as in 16, and
then stirred with LiCl in DMF to give, after preparative HPLC
with CH2Cl2, 11 as white crystals: mp 73-74 °C; NMR δH 1.40
+ 1.42 (18H, 2s, t-Bu), 1.8 + 2.0 (2H, 2m, CH2CH), 2.31 (2H,
t, CH2CO2, J ) 8 Hz), 3.62 (4H, t, CH2N, J ) 6 Hz), 3.73 (4H,
t, CH2Cl, J ) 6 Hz), 4.3 (1H, m, CH), 9.08 (1H, d, NH, J ) 7.5
Hz); δF -143.3 (2F, dd, F2, J2,3 ) 24 Hz, J2,5 ) 8 Hz), -147.7
(2F, dd, F3, J2,3 ) 24 Hz, J3,6 ) 8 Hz); MS m/z (FAB) 597 (M
+ Na+). Anal. (C24H32Cl2F4N2O5) C, H, N, Cl, F.
{4-[Bis(2-chloroethyl)amino]-2,3,5,6-tetrafluoroben-
zoyl}-L-glutamic Acid (21). Ester 11 (0.188 g, 0.33 mmoL)
was stirred in trifluoroacetic acid (5 mL) for 55 min and worked
up as in 26 to give 21 (0.146 g, 96%) as white crystals: mp
114-118 °C; NMR δH 1.8 + 2.0 (2H, 2m, CH2CH), 2.33 (2H, t,
CH2CO2, J ) 7.5 Hz), 3.61 (4H, t, CH2N, J ) 6 Hz), 3.73 (4H,
t, CH2Cl, J ) 6 Hz), 4.40 (1H, ddd, CH, JCH2CH2 ) 9 Hz and
5.5 Hz), 9.03 (1H, d, NH, J ) 8 Hz), 12.5 (2H, bs, CO2H); δF
-143.2 (2F, dd, F2, J2,3 ) 23 Hz, J2,5 ) 8.5 Hz), -147.8 (2F,
dd, F3, J2,3 ) 23 Hz, J3,6 ) 8.5 Hz); MS m/z (FAB) 485 (M +
Na+). Anal. (C16H16Cl2F4N2O5) C, H, N, Cl, F.
Di-tert-butyl 2,3,4,5-Tetrafluorobenzoyl-L-glutamate
(6). 2,3,4,5-Tetrafluorobenzoyl chloride (3) (10 g, 47 mmol) was
reacted with di-tert-butyl-L-glutamate (13.3 g, 50 mmol), as
in 5, to give 6 (21.3 g, 98%) as an oil: NMR δH 1.40 + 1.42
(18H, 2s, t-Bu), 1.8 + 2.0 (2H, 2m, CH2CH), 2.33 (2H, t, CH2-
CO2, J ) 7.5 Hz), 4.3 (1H, m, CH), 7.55 (1H, m, H6), 8.80 (1H,
d, NH, J ) 7.5 Hz); δF -138.6 and - 139.0 (2F, 2m, F2 + 5),
-152.5 (1F, m, F4), -154.9 (1F, t, F3, J ) 22 Hz); MS m/z
(electrospray) 324 (M - 2tBu + 3), acc mass (C12H10NF4O5)
calcd 324.0495, found 324.0508.
Di-tert-butyl {4-[Bis(2-hydroxyethyl)amino]-2,3,5,-tri-
fluorobenzoyl}-L-glutamate (9). Amide 6 (21.3 g, 49 mmol)
was reacted with diethanolamine (15 mL, 0.15 mol), as in 8,
to give 9 (7.94 g, 31%) as an oil: NMR δH 1.39 + 1.42 (18H,
2s, t-Bu), 2.0 (2H, m, CH2CH), 2.32 (2H, t, CH2CO2, J ) 7.5
Hz), 3.33 (4H, t, CH2N, J ) 6 Hz), 3.48 (4H, q, CH2O, J ) 5.5
Hz), 4.30 (1H, m, CH), 4.52 (2H, t, OH, J ) 5.5 Hz), 7.18 (1H,
ddd, H6, JH,F2 ) 6 Hz, JH,F3 ) 1.5 Hz, JH,F5 ) 13 Hz), 8.49 (1H,
d, NH, J ) 7.5 Hz); δF -122.8 (1F, q, F5, JF5,H6 )13 Hz), -142.1
(2F, m, F2 + 3).
Di-tert-butyl {4-[Bis(2-iodoethyl)amino]-2,3,5,-trifluo-
robenzoyl}-L-glutamate (17). Hydroxy compound 9 (2.74 g,
5.25 mmol) was reacted with methane sulfonic anhydride and
then sodium iodide, as in 16, to give 2.65 g (68%) of 17: NMR
δH 1.39 + 1.42 (18H, 2s, t-Bu), 1.85 + 2.0 (2H, 2m, CH2CH),
2.32 (2H, t, CH2CO), 3.3 (4H, t, CH2I), 3.59 (4H, t, CH2N, J )
7 Hz), 4.3 (1H, m, CH), 7.25 (1H, dd, H6, JF5,H6 ) 12 Hz, JF2,H6
) 6 Hz), 8.57 (1H, d, NH, J ) 7.5 Hz); δF -122.5 (1F, t, F5, J
) 10 Hz), -141.3 (2F, m, F2 + 3); MS m/z (FAB) 763 (M +
Na+), 741 (M+ + 1).
{4-[Bis(2-chloroethyl)amino]-2,3,5,-trifluorobenzoyl}-
L-glutamic Acid (22). Hydroxy compound 9 (1.39 g, 2.65
mmol) was reacted with methane sulfonic anhydride and then
lithium chloride (0.5 g, 11.8 mmol) in DMA (25 mL), as in 11,
to give 12 (1.37 g). Amide 12 (1.27 g, 2.28 mmol) was stirred
in trifluoroacetic acid (30 mL) for 1 h and worked up as in 26
to give 22 (1.02 g, 98%) as white crystals: mp 133-138 °C;
NMR δH 1.9 + 2.05 (2H, 2m, CH2CH), 2.34 (2H, t, CH2CO, J
) 7.5 Hz), 3.60 (4H, CH2N, J ) 6 Hz), 3.71 (4H, t, CH2Cl, J )
6 Hz), 4.4 (1H, m, CH), 7.25 (1H, m, H6), 8.56 (1H, d, NH, J
) 8 Hz), 12.5 (2H, bs, CO2H); δF -122.7 (1F, t, F5, JF5,H6 )13.5
Hz), -141.3 (1F, m, F2), -141.5 (1F, d, F3, J2,3 ) 19 Hz); MS
m/z (FAB) 467 (M + Na+), 445 (M+ + 1). Anal. C16H17Cl2F3N2O5
C, H, N, Cl, F.
Di-tert-butyl Pentafluorobenzoyl-L-glutamate (5). To
an ice-cold solution of di-tert-butyl-L-glutamate hydrochloride
(4.7 g, 16 mmol) and Et3N (4.4 mL, 32 mmol) in dry CH2Cl2
(30 mL) was added pentafluorobenzoyl chloride (2.4 mL, 16
mmol) in dry CH2Cl2 (30 mL) over 5 min. After 1 h the ice
was removed and the next day CH2Cl2 (140 mL) was added.
The solution was washed with water, dried (MgSO4), and
concentrated to dryness. The residue crystallized from n-
hexane as white crystals (6.36 g, 88%): mp 68-69 °C; NMR
δH 1.40 + 1.43 (18H, 2s, tBu), 1.8-2.0 (2H, ddt, CH2CH), 2.32
(2H, t, CH2CO2, J ) 8 Hz), 4.35 (1H, ddd, CH), 9.24 (1H, d,
NH, J ) 8 Hz); δF -141.4 (2F, dd, F2, J2,3 ) 22 and 7 Hz),
-152.3 (1F, t, F4, J3,4 ) 22 Hz), -160.9 (2F, ddd, F3, J2,3+3,4
)22 Hz, J2,3) 7 Hz); MS (FAB) m/z 454 (M+ + 1), 398 (M+
tBu +2). Anal. (C20H24F5NO5) C, H, N, F.
-
Di-tert-butyl {4-[Bis(2-hydroxyethyl)amino]-2,3,5,6-tet-
rafluorobenzoyl}-L-glutamate (8). Ester 5 (3.63 g, 8 mmol)
and diethanolamine (1.5 mL, 16 mmol) were heated in DMA
(30 mL) at 85 °C for 6 days. The solvent was removed in vacuo
and the residue partitioned between CH2Cl2 and water. The
product was purified by preparative HPLC with CH2Cl2/EtOH
(19:1) as eluent to give 2.62 g (61%) of white crystals: mp 65-
67.5 °C; NMR δH 1.40 +1.42 (18H, 2s, t-Bu), 1.8 +2.0 (2H,
2m, CH2CH), 2.31 (2H, dt, 2H, CH2CO2, JCH2CH2 ) 8 Hz, JCHCH2
) 2.5 Hz), 3.33 (4H, t, CH2N, J ) 5.5 Hz), 3.50 (4H, q, CH2O,
J ) 6 Hz), 4.30 (1H, octet, CH, JCHCH2 ) 5 Hz), 4.54 (2H, t,
OH, J ) 5 Hz), 9.03(1H, d, NH, J ) 7.5 Hz); δF -144.2 (2F, q,
F2, J2,3 ) 23 Hz, J2,5 ) 8 Hz), -148.5 (2F, q, F3, J2,3 ) 23 Hz,
J3,6 ) 8 Hz); MS (FAB) m/z 561 (M + Na+), 449 (M - (CH2-
CH2OH)2 + 1), 280 (M - glutamate). Anal. (C24H34F4N2O7) C,
H, N, F.
Di-tert-butyl {4-[Bis(2-iodoethyl)amino]-2,3,5,6-tet-
rafluorobenzoyl}-L-glutamate (16). To an ice-cold solution
of ester 8 (0.94 g, 1.75 mmol), Et3N (1.23 mL, 8.75 mmol), and
DMAP (43 mg, 0.35 mmol) in dry CH2Cl2 (30 mL) was added
methanesulfonic anhydride (1.23 g, 7 mmol) in dry CH2Cl2 (10
mL). After 30 min the ice was removed. The next day CH2Cl2
(30 mL) was added, the solution washed with 10% aqueous
citric acid (70 mL), the solvent evaporated, and the residue
dried in vacuo (P2O5). This was dissolved in acetone (40 mL),
NaI (3.65 g) was added, and the solution was refluxed for 5 h.
The solvent was evaporated, the residue partitioned between
CH2Cl2 and water, the CH2Cl2 removed in vacuo, and the
residue purified by preparative HPLC eluting with a gradient
from CH2Cl2 to 10% EtOAc/CH2Cl2. The product crystallized
from CH2Cl2/hexane to give 0.94 g (71%) of white crystals: mp
115-117 °C; NMR δH 1.40 + 1.42 (18H, 2s, t-Bu), 1.8 +2.0
(2H, 2m, CH2CH), 2.34 (2H, t, CH2CO2, J ) 8 Hz), 3.33 (4H,
t, CH2I, J ) 7 Hz), 3.60 (4H, t, CH2N, J ) 7 Hz), 4.3 (1H,
sextet, CH, JCHCH2 ) 5 and 9 Hz, JCHNH ) 8 Hz), 9.07 (1H, d,
NH, J ) 8 Hz); δF -143.2 (2F, dd, F2, J2,3 ) 23 Hz, J2,5 ) 8.5
Hz), -147.4 (2F, dd, F3, J2,3 ) 23 Hz, J3,6 ) 9 Hz); MS m/z
(FAB) 781 (M + Na+), 759 (M+ + 1). Anal. (C24H32F4I2N2O5)
C, H, N, F, I.
{4-[Bis(2-iodoethyl)amino]-2,3,5,6-tetrafluorobenzoyl}-
L-glutamic Acid (26). Ester 16 (0.76 g, 1 mmol) was stirred
in trifluoroacetic acid (20 mL) for 50 min. The solvent was
evaporated off followed by five evaporations with EtOAc. The
residue was redissolved in EtOAc and filtered and then toluene
was added. The solution was concentrated until crystals began
to appear. After cooling at -30 °C, the white crystals were
filtered off (0.58 g, 82%): mp 142-144 °C; NMR δH 1.85 + 2.0
(2H, 2m, CH2CH), 2.33 (2H, t, CH2CO2, J ) 7.5 Hz), 3.33 (4H,
t, CH2I), 3.60 (4H, t, CH2N, J ) 7 Hz), 4.40 (1H, m, CH), 9.05
(1H, bs, NH); δF -142.9 (2F, dd, F2, J2,3 ) 22 Hz, J2,5 ) 8 Hz),
{4-[Bis(2-bromoethyl)amino]-2,3,5,-trifluorobenzoyl}-
L-glutamic Acid (24). Hydroxy compound 9 (1.39 g, 2.65
mmol) was reacted with methane sulfonic anhydride and then
lithium bromide (0.87 g, 10 mmol) in acetone (25 mL), as in
16, to give 14 (1.33 g). Amide 14 (1.25 g, 1.93 mmol) was stirred

5326 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16
Davies et al.
in trifluoroacetic acid (30 mL) for 1 h and worked up as in 26
to give 24 (1.03 g, 99%) as white crystals: mp 141-144 °C;
NMR δH 1.9 + 2.05 (2H, m, CH2CH), 2.35 (2H, t, CH2CO2, J
) 7.5 Hz), 3.58 (4H, t, CH2N, J ) 5.5 Hz), 3.66 (4H, t, CH2Br,
J ) 5.5 Hz), 4.4 (1H, m, CH), 7.25 (1H, dd, H6, JH6,F5 ) 12 Hz,
JH6,F3 ) 5 Hz), 8.56 (1H, d, NH, J ) 7.5 Hz), 12.5 (2H, bs,
CO2H); δF -122.4 (1F, t, F5, JF5,H6 ) 14 Hz), -141.1 (2F, m,
F2+3); MS m/z (FAB) 557 (M + Na+), 535 (M+ + 1). Anal.
C16H17Br2F3N2O5 C, H, N, Br, F.
{4-[Bis(2-bromoethyl)amino]-3,5-difluorobenzoyl}-L-
glutamic Acid (25). Ester 15 (0.388 g, 0.62 mmol) was stirred
in TFA (10 mL) for 1 h and worked up as in 26 to give 25
(0.317 g, 99%) as a white solid: NMR δH 1.95 + 2.05 (2H, 2m,
CH2CH), 2.35 (2H, t, CH2CO2, J ) 7.5 Hz), 3.55 (4H, t, CH2-
CO2, J ) 5.5 Hz), 3.61 (4H, t, CH2Br, J ) 5.5 Hz), 4.4 (1H, m,
CH), 7.60 (2H, d, 2 + 6-CH, J ) 10 Hz), 8.62 (1H, d, NH, J )
7.5 Hz), 12.4 (2H, bs, CO2H); δF -117.3 (2F, d, F3 + 5, J ) 10
Hz); MS m/z (FAB) 539 (M + Na+), 517 (M+ + 1). Anal. C16H18-
Br2F2N2O5 C, H, I, N, F.
tert-Butyl 3,4,5-Trifluorobenzoate (34). To an ice-cold
solution of 3,4,5-triflorobenzoyl chloride (5.0 g, 25.7 mmol) in
dry THF (25 mL) was added lithium tert-butoxide (30 mL of 1
M solution in THF) over 45 min. One hour later, the ice was
removed and after one further hour the solvent was evaporated
and the residue partitioned between CH2Cl2 and water.
Crystallization from pentane (5 mL) at -30 °C gave colorless
crystals: mp 20-21 °C (4.63 g, 78%); NMR δH 1.54 (9H, s,
t-Bu), 7.75 (2H, dd, 2+6CH, JH-F3,5 ) 8 Hz, JH,F4 ) 7 Hz); δF
-133.2 (2F, dd, F3, JFF ) 21 Hz, JHF ) 8 Hz), -154.2(1F, tt,
F4, JFF )21 Hz, JHF ) 7 Hz); MS m/z (EI) 176 (M+ - tBu + 1),
159 (M+ - OtBu). Anal. (C11H11F3O2) C, H, F.
{4-[Bis(2-iodoethyl)amino]-2,3,5,-trifluorobenzoyl}-L-
glutamic Acid (27). Amide 17 (0.67 g, 0.91 mmol) was stirred
in trifluoroacetic acid (17 mL) for 1 h and worked up as in 26
to give 27 (0.53 g, 93%) as white crystals: mp 111-114 °C;
NMR δH 1.9 + 2.05 (2H, 2m, CH2CH), 2.35 (2H, t, CH2CO2, J
) 7.5 Hz), 3.31 (4H, t, CH2I, J ) 7 Hz), 3.59 (4H, t, CH2N, J
) 7 Hz), 4.4 (1H, m, CH), 7.28 (1H, dd, H6, JH,F5 ) 11 Hz,
JH,F3 ) 2.5 Hz), 8.54 (1H, d, NH, J ) 7.5 Hz), 12.4 (2H, bs,
CO2H; δF -122.4 (1F, q, F5, JF,H6 ) 13 Hz), -141.1 (2F, d, F2
+ 3); MS m/z (FAB) 651 (M + Na+), 629 (M++1). Anal.
C16H17F3-
I2N2O5 C, H, N, F, I.
Di-tert-butyl 3,4,5-Trifluorobenzoyl-L-glutamate (7).
3,4,5-Trifluorobenzoyl chloride 4 (10 g, 51.4 mmol) was reacted
with tert-butyl-L-glutamate (14.8 g, 50 mmol), as in 5, to give
7 (19.9 g, 95%) as white crystals: mp 49-54 °C; NMR δH 1.39
+ 1.41 (18H, 2s, t-Bu), 1.9 + 2.0 (2H, 2m,CH2CH), 2.34 (2H, t,
CH2CO2, J ) 7 Hz), 4.3(1H, m, CH), 7.84 (2H, q, H2, J ) 7
and 9 Hz), 8.75 (1H, d, NH, J ) 7.5 Hz); δF -133.7 (2F, dd,
F3, J ) 8 and 22 Hz), -156.3 (1F, tt, F4, J ) 7 and 22 Hz);
MS m/z (FAB) 440 (M + Na+), 418 (M++1). Anal. C20H26F3-
NO5 C, H, N, F.
Di-tert-butyl {4-[Bis(2-chloroethyl)amino]-3,5-difluo-
robenzoyl}-L-glutamate (13). Di-tert-butyl-L-glutamate hy-
drochloride (1.28 g, 4.25 mmol), triethylamine (1.2 mL, 8.5
mmol), and hydroxy compound 20 (1.1 g, 4.25 mmol) were
dissolved in dry DMF (65 mL), diethylcyanophosphonate (0.7
mL, 4.7 mmol) was added and the solution was stirred under
argon for 3 days. The solvent was then evaporated and the
residue partitioned between EtOAc and water to give the bis
hydroxy compound 8. This was then dissolved in dry CH2Cl2
(80 mL), and DMAP (105 mg, 0.85 mmol) and triethylamine
(3.25 mL, 23 mmol) were added. The solution was cooled in
ice and methane sulfonic anhydride (2.95 g, 17 mmol) was
added in dry CH2Cl2 (25 mL). After 16 h CH2Cl2 (70 mL) was
added and the solution was extracted with 10% citric acid.
After evaporation the product was dissolved in DMA (60 mL)
containing LiCl (1.45 g, 34 mmol) and heated for 1 h at 90 °C.
The solvent was evaporated, and the residue was dissolved in
CH2Cl2, extracted with H2O, and purified by preparative HPLC
eluting with CH2Cl2 to give 0.55 g of pure oil: NMR δH 1.39 +
1.41 (18H, 2s, t-Bu), 1.81 +2.01 (2H, 2m, CH2CH), 2.33 (2H,
t, CH2CO2, J ) 8 Hz), 3.58 (4H, t, CH2N, J ) 7 Hz), 3.67 (4H,
t, CH2I, J ) 7 Hz), 4.3 (1H, sextet, CH, JCHCH2 ) 5 and 9 Hz,
JCHNH ) 8 Hz), 7.5 (2H, d HH2+H6, JHF )10 Hz), 8.6 (1H, d,
NH, J ) 8 Hz).
tert-Butyl 4-[bis(2-hydroxyethyl)amino]-3,5-difluoro-
benzoate (37). Diethanolamine (3.9 g. 37 mmol) was added
to ester 34 (4.33 g, 18.6 mmol), dissolved in DMA (50 mL),
and heated at 90 °C for 3.5 days. The solvent was evaporated
and the residue partitioned between CH2Cl2 and water, and
the organic layer was concentrated and purified by preparative
HPLC eluting with a gradient from CH2Cl2 to 5% EtOAc in
CH2Cl2 to give white crystals (2.21 g, 37%): mp 58-66 °C;
NMR δH 1.52 (9H, s t-Bu), 3.30 (4H, t, CH2N, J ) 6 Hz), 3.45
(4H, q, CH2O, JCH2CH2 ) 6 Hz, JCH2OH ) 5 Hz), 4.47 (2H, t,
OH, J ) 5 Hz), 7.40 (2H, d, 2+6-CH, J ) 10 Hz); δF -117.4
(2F, d, J ) 10 Hz); MS m/z (LDI) 340 (M + Na+), 318 (M+
1); Anal. (C15H21F2NO4) C, H, N, F.
+
tert-Butyl 4-(Bis(2-bromoethyl)amino)-3,5-difluoroben-
zoate (42). To an ice-cold solution of ester 37 (0.43 g, 1.5
mmol), Et3N (1.05 mL, 7.5 mmol), and DMAP (37 mg, 0.3
mmol) in dry CH2Cl2 (27.5 mL) was added methanesulfonic
anhydride (1.05 g, 6 mmol) in dry CH2Cl2 (9 mL) over a few
minutes. The ice was removed, the next day CH2Cl2 (25 mL)
was added, the solution was washed with 10% citric acid, dried,
and then evaporated to dryness. This was dissolved in acetone
(25 mL), LiBr (1.30 g, 15 mmol) added, and the mixture
refluxed for 5 h. The solvent was evaporated and the residue
dissolved in CH2Cl2, washed with water, and evaporated to
dryness. Preparative HPLC with CH2Cl2 as eluent gave a low-
melting solid (0.40 g, 60%): mp 22-26 °C; NMR δH 1.52 (9H,
s, t-Bu), 3.56 (4H, t, CH2N, J ) 5.5 Hz), 3.65 (4H, t, CH2Br, J
) 5.5 Hz), 7.48 (2H, d, 2+6CH, J ) 9 Hz); δF -117.0 (2F, d,
3+5F, J ) 9 Hz); MS m/z (FAB) 440/2/4 (M+ + 1), 385/7/9 (M+
- tBu + 1); Anal. (C15H19Br2F2NO2) C, H, N, F.
4-(Bis(2-bromoethyl)amino)-3,5-difluorobenzoic Acid
(50). Ester 42 (0.40 g, 0.91 mmol) was stirred in trifluoroacetic
acid (12 mL) for 55 min. The solvent was evaporated and then
EtOAc was added and evaporated five times. The residue was
redissolved in EtOAc and filtered, toluene added, and the
solution evaporated to dryness. It was redissolved in toluene,
hexane was slowly added to give a mass of white crystals (0.30
g, 85%): mp 111-113.5 °C; NMR δH 3.56 (4H, t, CH2N, J )
5.5 Hz), 3.63 (4H, t, CH2Br, J ) 5.5 Hz), 7.52 (2H, d, 2+6CH,
J ) 10 Hz), 13.2 (1H, bs, CO2H); δF -117.0 (2F, d, 3+5F, J )
10 Hz); MS m/z (FAB) 388 (M+ + 1), 345 (M+ - CO2 + 1), 307
(M+ - Br). Anal. (C11H11Br2F2NO2), C, H, N, Br, F.
Di-tert-butyl {4-[Bis(2-bromoethyl)amino]-3,5-difluo-
robenzoyl}-L-glutamate (15). Hydroxy compound 20 (1.1 g,
4.25 mmol) was reacted as in 13 above, using LiBr in acetone
instead of LiCl/DMF, to yield 0.87 g of oil: NMR δH 1.39 +
1.42 (18H, 2s, t-Bu), 1.87 +2.03 (2H, 2m, CH2CH), 2.33 (2H,
t, CH2CO2, J ) 8 Hz), 3.58 (4H, t, CH2N, J ) 7 Hz), 3.64 (4H,
t, CH2I, J ) 7 Hz), 4.3 (1H, sextet, CH, JCHCH2 ) 5 and 9 Hz,
JCHNH ) 8 Hz), 7.2 (1H, dd H, J ) 4.5 and 12.5 Hz,), 8.6 (1H,
d, NH, J ) 8 Hz).
{4-[Bis(2-chloroethyl)amino]-3,5-difluorobenzoyl}-L-
glutamic acid (23). Ester 13 (1.11 g, 2.05 mmol) was stirred
in TFA (28 mL) for 1 h and worked up as in 26 to give 23
(0.88 g, 100%) as white crystals: mp 114-115 °C; NMR δH
1.95 + 2.05 (2H, 2m, CH2CH), 2.35 (2H, t, CH2CO2, J ) 7.5
Hz), 3.57 (4H, t, CH2N, J ) 6 Hz), 3.68 (4H, t, CH2Cl, J ) 6
Hz), 4.4 (1H, m, CH), 7.60 (2H, d, 2 + 6-CH, J ) 10 Hz), 8.61
(1H, d, NH, 7.5 Hz), 12.4 (2H, bs, CO2H); δF -117.4 (2F, d, F3
+ 5, J ) 10 Hz); MS m/z (FAB) 449 (M + Na+), 427 (M+ + 1).
Anal. C16H18Cl2F2N2O5 C, H, N, Cl, F.
tert-Butyl 4-(Bis(2-chloroethyl)amino)-3,5-difluoroben-
zoate (40). Ester 37 (0.43 g, 1.5 mmoL) was reacted with
methane sulfonic anhydride as in 42 and then reacted with
LiCl (0.64 g, 15 mmol) in DMF (25 mL) at room temperature
to give, after preparative HPLC with CH2Cl2, 40 (0.32 g, 61%)
as white crystals: mp 59-60 °C; NMR δH 1.53 (9H, s, t-Bu),
3.59 (4H, t, CH2N, J ) 5.5 Hz), 3.69 (δH, t, CH2Cl, J ) 5.5
Hz), 7.48 (2H, d, 2+6CH, J ) 10 Hz); δF -117.1 (2F, d, F3 +
5, J ) 9.5 Hz); MS m/z (FAB) 354 (M+ + 1). Anal. C15H19Cl2F2-
NO2 C, H, N, Cl, F.

Prodrugs for Activation by Carboxypeptidase G2
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16 5327
4-(Bis(2-chloroethyl)amino)-3,5-difluorobenzoic Acid
(48). Ester 40 (0.25 g, 0.71 mmoL) was stirred in trifluoroacetic
acid (6 mL) for 55 min and worked up as in 50 to give 48 (0.177
g, 85%) as white crystals: mp78-79 °C; NMR δH 3.59 (4H, t,
CH2N, J ) 5.5 Hz), 3.69 (4H, t, CH2Cl, J ) 5.5 Hz), 7.51 (2H,
d, 2 + 6CH, J ) 9.5 Hz), 13.3 (1H, bs, CO2H); δF -117.1 (2F,d,
3 + 5F, J ) 9.5 Hz); MS m/z (FAB) 298 (M+ + 1). Anal. C11H11-
NCl2F2O2 C, H, N, Cl, F.
4-(Bis(2-iodoethyl)amino)-3,5-difluorobenzoic Acid (53).
Ester 45 (0.163 g, 0.30 mmoL) was stirred in TFA (5 mL) for
40 min and worked up as in 50 to give 53 (0.106 g, 73%) as
white crystals: mp 123-125 °C; NMR δH 3.29 (4H, t, CH2I, J
) 7 Hz), 3.58 (4H, t, CH2N, J ) 7 Hz), 7.52 (2H, d, 2 + 6CH,
J ) 10 Hz); δF -117.0 (2F, d, 3 + 5F, J ) 10 Hz); MS m/z
(FAB) 482 (M+ + 1), 481 M+. Anal. C11H11NF2I2O2 C, H, N, F,
I.
5.5 Hz), 3.73 (4H, t, CH2Cl, J ) 5.5 Hz), 7.45 (1H, m, H6),
13.5 (1H, bs, CO2H); δF -122.8 (1F, sextet, F5, J2,5 ) 13 Hz,
J3,5 ) 5 Hz), -138.5 (1F, octet, F2, J2,5 ) 13 Hz, J2,3 ) 20 Hz),
-144.5 (1F, q, F3, J2,3 ) 20 Hz, J3,5 ) 5 Hz); MS m/z (FAB)
338 (M + Na+), 316 (M+ + 1). Anal. C11H10Cl2F3NO2 C, H, N,
Cl, F.
4-(Bis(2-bromoethyl)amino)-2,3,5-trifluorobenzoic Acid
(49). Ester 41 (0.88 g, 1.9 mmol) was stirred in TFA (15 mL)
for 55 min and worked up as in 50 to give 49 (0.72 g, 94%) as
white crystals: mp 89-90 °C; NMR δH 3.60 (4H, t, CH2N, J )
5.5 Hz), 3.68 (4H, t, CH2Br, J ) 5.5 Hz), 7.45 (1H, m, H6),
13.5 (1H, bs, CO2H); δF -122.8 (1F, sextet, F5, J2,5 ) 13.5 Hz,
J3,5 ) 5 Hz), -138.5 (1F, octet, F2, J2,3 ) 20 Hz, J2,5 ) 13.5
Hz), -141.4 (1F, q, F3, J2,3 ) 20 Hz, J3,5 ) 5 Hz); MS m/z (FAB)
428 (M + Na+), 406 (M+ + 1). Anal. C11H10Br2F3NO2 C, H, N,
Br, F.
4-(Bis(2-iodoethyl)amino)-2,3,5-trifluorobenzoic Acid
(52). Ester 44 (0.82 g, 1.48 mmoL) was stirred in TFA (20 mL)
for 55 min and worked up as in 50 to give 52 (0.69 g, 93%) as
white crystals: mp 81-83 °C; NMR δH 3.33 (4H, t, CH2I, J )
7 Hz), 3.62 (4H, t, CH2N, J ) 7 Hz), 7.45 (1H, m, H6), 13.5
tert-Butyl 2,3,4,5-Tetrafluorobenzoate (33). 2,3,4,5-Tet-
rafluorobenzoyl chloride (10.6 g, 50 mmol) was reacted with
lithium tert-butoxide (4.8 g, 60 mmol) as in 34 to give 33 (5.35
g, 43%) of pale yellow crystals that melted at room tempera-
ture: NMR δH 1.55 (9H, s, t-Bu), 7.77 (1H, m, H6, JH,F5 ) 8.5
Hz); δF -136.25 (1F, m, F2), -138.4 (1F, quintet, F5, J ) 11
Hz), -148.9 (1F, m, F4), -154.0 (1F, t, F3, J ) 21.5 Hz); MS
m/z (EI) 177 (M+ - OtBu), 149 (M+ - CO2tBu). Anal. C11H10F4O2
C, H, F.
tert-Butyl 4-(Bis(2-hydroxyethyl)amino)-2,3,5-trifluo-
robenzoate (36). Ester 33 (3.75 g, 15 mmol) and diethano-
lamine (4.2 g, 40 mmol) were heated in DMA as in 37 to give
pure 36 (2.5 g, 40%): NMR δH 1.52 (9H, s, t-Bu), 3.35 (4H, t,
CH2N, J ) 5.5 Hz), 3.48 (4H, q, CH2O, J ) 5.5 Hz), 4.56 (2H,
(1H, bs, CO2H); δF -122.8 (1F, sextet, F5, J2,5 ) 13 Hz J3,5
)
5 Hz), -138.4 (1F, octet, F2, J2,3 ) 20 Hz, J2,5 ) 13 Hz), -141.3
(1F, q, F3, J2,3 ) 20 Hz, J3,5 ) 5 Hz); MS m/z (FAB) 522 (M +
Na+), 500 (M+ + 1). Anal. C11H10F3I2NO2 C, H, N, F, I.
tert-Butyl 2,3,4,5,6-Pentafluorobenzoate (32). Pentaflu-
orobenzoyl chloride (11.5 g, 50 mmol) was reacted with lithium
tert-butoxide (50 mmol) as in 34 to give, after distillation, 32
(9.8 g, 73%) as a solid: mp 9-11 °C; NMR δH 1.55 (9H, s, t-Bu);
δF -140.9 (2F, dd, F2, J2,3 ) 27 Hz, J2,4 ) 8.5 Hz), -149.9
(1F, t, F4, J3,4 ) 22 Hz), -160.7 (2F, m, F3); MS m/z (EI) 268
(M+), 195 (C6H5 CO). Anal. C11H9F5O2 C, H, F.
tert-Butyl 4-(Bis(2-hydroxyethyl)amino)-2,3,5,6-tet-
rafluorobenzoate (35). Ester 32 (5.36 g, 20 mmol) and
diethanolamine (4.2 g, 40 mmol) were heated in DMA as in
37 to give, after preparative HPLC with CH2Cl2/EtOAc (4:1)
as eluent, pure 35 (3.68 g, 52%) as pale yellow crystals: mp
45-49 °C; NMR δH 1.52 (9H, s, t-Bu), 3.37 (4H, t, CH2N, J )
5.5 Hz), 3.50 (4H, q, CH2O, JCH2CH2 ) 5.5 Hz, JCH2OH ) 5 Hz),
4.55 (2H, t, OH, J ) 5 Hz); δF -143.0 (2F, dd, F2, J2,3 ) 20
Hz), -148.7 (2F, dd, F3, J2,3 ) 20 Hz); MS m/z (LDI) 354 (M+).
Anal. C15H19F4NO4 C, H, N, F.
tert-Butyl 4-(Bis(2-iodoethyl)amino)-2,3,5,6-tetrafluo-
robenzoate (43). Hydroxy compound 35 (0.35 g, 1 mmol) was
reacted with methane sulfonic anhydride, as in 42, and then
refluxed with NaI (1.5 g, 10 mmol) in acetone (25 mL) to give
43 (0.40 g, 70%) as white crystals: mp 48-49.5 °C; NMR δH
1.53 (9H, s, t-Bu), 3.34 (4H, t, CH2N, J ) 7 Hz), 3.62 (4H, CH2I,
J ) 7 Hz); δF -142.3 (2F, dd, F2, J2,3 ) 22 Hz, J2,5 ) 8 Hz),
-147.6 (2F, dd, F3, J2,3 ) 22 Hz, J3,6 ) 8 Hz); MS m/z (EI) 573
(M+). Anal. C15H17F4I2NO2 C, H, N, F, I.
t, OH, J ) 5 Hz), 7.32 (1H, ddd, H6, JF5,H6 ) 13 Hz, JF2,H6
)
6.5 Hz, JF3,H6 ) 2 Hz); δF -122.8 (1F, dt, F5, JF5,H6 ) 13 Hz,
JF2,F5 ) 12.5 Hz, JF3,F5 ) 7 Hz), -139.3 (1F, ddd, F2, JF2,F3
20 Hz, JF2,F5 ) 12.5 Hz, JF2,H6 ) 6.5 Hz), -142.3 (1F, dd, F3,
JF2,F3 ) 20 Hz, JF3,F5 ) 7 Hz).
)
tert-Butyl 4-(Bis(2-chloroethyl)amino)-2,3,5-trifluo-
robenzoate (39). Hydroxy compound 36 (0.67 g, 2 mmol) was
reacted with methane sulfonic anhydride as in 42 and then
reacted with LiCl (0.85 g, 20 mmol) in DMF (33 mL) at room
temperature to give, after preparative HPLC with CH2Cl2, 39
(0.68 g, 91%) as white crystals: mp 52-54 °C; NMR δH 1.53
(9H, s, t-Bu), 3.63 (4H, t, CH2N, J ) 5.5 Hz), 3.72 (4H, t, CH2-
Cl, J ) 5.5 Hz), 7.42 (1H, q, H6, JF5,H6 ) 13 Hz, JF2,H6 ) 6.5
Hz); δF -122.7 (1F, q, F5, JH2,F5 ) 13 Hz, JF5,H6 ) 5 Hz), -138.8
(1F, m, F2, JF2,F3 ) 20 Hz, JF2,F5 ) 13 Hz), -141.3 (1F, q, F3,
JF2,F3 )20 Hz, JF3,F5 ) 5 Hz); MS m/z (FAB) 394 (M + Na+),
371 (M+). Anal. C15H18Cl2F3NO2 C, H, N, Cl, F.
tert-Butyl 4-(bis(2-bromoethyl)amino)-2,3,5-trifluo-
robenzoate (41). Hydroxy compound 36 (0.67 g, 2 mmol) was
reacted with methane sulfonic anhydride as in 42 and then
refluxed with LiBr (1.7 g, 20 mmol) in acetone (35 mL) to give,
after preparative HPLC with CH2Cl2, 41 (0.93 g, 100%) as
white crystals: mp 27-29 °C; NMR δH 1.52 (9H, s, t-Bu), 3.60
(4H, t, CH2N, J ) 5.5 Hz), 3.69 (4H, t, CH2Br, J ) 5.5 Hz),
7.42 (1H, m, H6); δF -122.7 (1F, sextet, F5, JF2,F5 ) 13 Hz,
4-(Bis(2-chloroethyl)amino)-2,3,5,6-tetrafluorobenzo-
ic Acid (46). Ester 38 (0.365 g, 0.94 mmoL) was stirred in
trifluoroacetic acid (10 mL) for 1 h and worked up as in 50 to
give 46 (0.288 g, 92%) as white crystals: mp 98-100 °C; NMR
δH 3.64 (4H, t, CH2N, J ) 6 Hz), 3.74 (4H, CH2Cl, J ) 6 Hz);
δF -141.6 (2F, dd, F2, J2,3 ) 22 Hz, J2,5 ) 8 Hz), -148.0 (2F,
dd, F3, J2,3 ) 22 Hz, J3,6 ) 8 Hz); MS m/z (FAB) 356 (M +
Na+), 334 (M+ + 1). Anal. C11H9Cl2F4NO2 C, H, N, Cl, F.
4-(Bis(2-iodoethyl)amino)-2,3,5,6-tetrafluorobenzoic
Acid (51). Ester 43 (0.287 g, 0.50 mmoL) was stirred in
trifluoroacetic acid (5 mL) for 40 min and worked up as in 50
to give 51 (0.22 g, 85%) as white crystals: mp 107-108 °C;
JF3,F5 ) 5 Hz), -138.7 (1F, octet, F2, JF2,F3 ) 20 Hz, JF2 F5 )
,
13 Hz), -141.2 (1F, q, F3, JF2,F3 ) 20 Hz, JF3,F5 ) 5 Hz); MS
m/z (FAB) 461 (M+), 406 (M - tBu + 1), acc mass (C11H11-
NBr2F3O2) calcd 403.9109, found 403.9142.
tert-Butyl 4-(Bis(2-iodoethyl)amino)-2,3,5-trifluoroben-
zoate (44). Hydroxy compound 36 (0.67 g, 2 mmol) was reacted
with methane sulfonic anhydride, as in 42, and then refluxed
with NaI (3.0 g, 20 mmol) in acetone (35 mL) to give, after
preparative HPLC with CH2Cl2, 44 (0.86 g, 77%) as an oil:
NMR δH 1.54 (9H, s, t-Bu), 3.35 (4H, m, CH2I), 3.63 (4H, t,
NMR δH
3.34 (4H, t, CH2N, J ) 7 Hz), 3.62 (4H, CH2I, J ) 7
Hz); δF -141.5 (2F, dd, F2, J2,3 ) 21 Hz, J2,5 ) 7 Hz), -147.9
(2F, dd, F3, J2,3 ) 21 Hz, J3,6 ) 7 Hz). Anal. C11H9F4I2NO2 C,
H, N, F, I.
CH2N, J ) 7 Hz), 7.43 (1H, dd, H6, JF5,H6 ) 11.5 Hz, JF2,H6
)
5.5 Hz); δF -123.1 (1F, dt, F5, JF2,F5 ) 13 Hz), -139.1 (1F, m,
F2), -141.6 (1F, d, F3, JF2,F3 ) 21.5 Hz); MS m/z (electrospray)
555 (M+), 500 (M - tBu + 1), acc mass (C15H18F3I2NO2) calcd
555.9457, found 555.9464.
Biology. Determination of Aqueous Half-Life. Com-
pounds were prepared as 10 mM concentrates in DMSO and
diluted 100 fold in CPG2 assay buffer (100 mM Tris-HCl, pH
7.3; 260 µM ZnCl2; 1 mL) to give 100 µM solutions. Aliquots
(10 µL) were injected onto a Synergi Polar RP phenyl phase
column (150 × 4.6 mm, 4 µM, Phenomenex) and eluted
isocratically (1 mL/min) with 10 mM ammonium acetate (pH
4-(Bis(2-chloroethyl)amino)-2,3,5-trifluorobenzoic Acid
(47). Ester 39 (0.64 g, 1.72 mmol) was stirred in TFA (15 mL)
for 1 h and worked up as in 50 to give 47 (0.48 g, 88%) as
white crystals: mp 81-83 °C; NMR δH 3.63 (4H, t, CH2N, J )

5328 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 16
Davies et al.
5.0) containing percentages of methanol that gave retention
times of 3-4 min. The eluate was monitored at the previously
determined wavelength of maximum absorbance. The amount
of starting material remaining after various periods of incuba-
tion was determined by repeat injections. The fraction of
compound remaining was expressed as a function of time, and
the half-life was determined by nonlinear regression to a one-
phase exponential decay, constraining the maximum to 1 and
the minimum to 0 (Graphpad Prism, Graphpad Software Inc.,
San Diego, CA).
Enzyme Substrate Kinetic Determinations. Initially, a
differential spectrum between prodrug and drug (100 µM each)
was used to determine the wavelength of greatest difference
and the conversion factor for the amount of substrate lost per
unit change in absorbance. Reactions (37 °C) were then set
up containing CPG2 assay buffer (1 mL) and prodrug at
concentrations from 0.2 to 50 µM in 25 steps using increments
of 0.2 µM (0.2-1.0 µM), 1 µM (1-10 µM), 2 µM (10-20 µM),
and 5 µM (20-50 µM), with or without 50 mU CPG2. The
change in the previously established wavelength was moni-
tored spectrophotometrically for 1 min, and the enzyme-specific
change in the concentration of prodrug was determined after
subtraction of the chemical-only change, and application of the
conversion factor. The Km and Kcat were determined by
nonlinear regression to a Michalis-Menten plot (Graphpad
Prism, Graphpad Software Inc., San Diego, CA).
and United Kingdom Coordinating Committee on Cancer
Research Guidelines.10 MDA MB 361 human breast carcinoma
cells that had been transfected with the CPG2 gene and were
expressing the CPG2 enzyme displayed on the cell surface
(stCPG2(Q)3) were grown as xenografts by implanting cells
(107) subcutaneously in the right flank of nude (nu/nu) female
BALB/C mice (20-22 g). After 4 days the tumor volumes were
124.3 ( 4.4 mm3. At this point the mice were randomly
distributed into control and treatment groups, and the pro-
drugs, freshly prepared in 1.26% sodium bicarbonate/5%
DMSO, were administered to treatment groups of six mice as
three intraperitoneal doses of 500 (22-25, 27) or 200 mg/kg
(28) over 24 h, defined as one course of treatment. All treated
groups received one such course of treatment per week for 3
weeks followed by 1 week free of treatment and were redosed
according to this schedule if palpable tumors remained. Tumor
growth was assessed by periodic calliper measurement. The
maximum bodyweight loss produced by these doses (<20%)
was in accordance with United Kingdom Coordinating Com-
mittee on Cancer Research guidelines.10 No other toxicity was
observed.
Acknowledgment. This work was supported by
Cancer Research UK [CUK] grant number C309/A2187.
References
Enzyme Inhibitor Kinetic Determinations. Repeat Km
determinations were performed as above, using methotrexate
as substrate at concentrations of 2-100 µM in 17 steps using
increments of 2 µM (2-10 µM), 5 µM (10-50 µM), and 10 µM
(50-100 µM), plus 50 mU CPG2, in the presence of increasing
concentrations of inhibitor (0-1000 µM). The observed Km
values were plotted against inhibitor concentration, and the
Ki values were calculated as the positive value where the linear
regression line crosses the ordinate.
(1) Niculescu-Duvaz, I.; Friedlos, F.; Niculescu-Duvaz, D.; Davies,
L.; Springer, C. J. Prodrugs for antibody- and gene-directed
enzyme prodrug therapies (ADEPT and GDEPT). Anti-Cancer
Drug Des. 1999, 14, 517-538.
(2) Niculescu-Duvaz, I.; Spooner, R.; Marais, R.; Springer, C. J.
Gene-directed enzyme prodrug therapy. Bioconjugate Chem.
1998, 9, 4-22.
(3) Marais, R.; Spooner, R. A.; Stribbling, S. M.; Light, Y.; Martin,
J.; et al. A cell surface tethered enzyme improves efficiency in
gene-directed enzyme prodrug therapy. Nat. Biotechnol. 1997,
15, 1373-1377.
Cell Lines and Culture Conditions. The generation of
â-gal-, CPG2*-, and stCPG2(Q)3-expressing MDA MB 361
stable cell lines together with a detailed description of the
subcellular localization of the enzymes has been described
previously.3,8 The cells were cultured in DMEM supplemented
with 10% fetal bovine serum at 37 °C in a 5% CO2 atmosphere.
Cytotoxicity Determinations. The cytotoxicity of the
compounds was determined in a two-step protocol as described
before.9 Cells (5 × 105 per well) were seeded into 6-well plates,
producing confluent monolayers in 48 h. Compounds were
initially dissolved in DMSO at 100-fold the highest dose (1-4
mM) and, immediately prior to treatment, prepared as a
dilution series in full medium. The medium in the wells was
removed and replaced with prodrug-containing medium (225
µL). After an incubation of 1 h, a similar concentration of
freshly prepared drug solution (1275 µL) was added, and the
cells were incubated for an additional 20 h. The cells were
harvested and reseeded in 96-well plates at 2 × 103/well and
incubated until the control wells achieved confluence (5 days).
The plates were then fixed and stained with sulforhodamine-
B, the extinction at 540 nm was read, the results were
expressed as a percentage of control growth as a function of
log dose, and the IC50 values were determined by nonlinear
regression to a four-parameter logistic sigmoid, constraining
the minimum to be positive (Graphpad Prism, Graphpad
Software Inc., San Diego, CA).
(4) Springer, C. J.; Niculescu-Duvaz, I.; Pedley, R. B. Novel prodrugs
of alkylating-agents derived from 2-fluorobenzoic and 3-fluo-
robenzoic acids for antibody-directed enzyme prodrug therapy.
J. Med. Chem. 1994, 37, 2361-2370.
(5) Springer, C. J.; Dowell, R.; Burke, P. J.; Hadley, E.; Davies, D.
H.; et al. Optimization of alkylating agent prodrugs derived from
phenol and aniline mustards: A new clinical candidate prodrug
(ZD2767) for antibody-directed enzyme prodrug therapy (ADEPT).
J. Med. Chem. 1995, 38, 5051-5065.
(6) Friedlos, F.; Davies, L.; Scanlon, I.; Ogilvie, L. M.; Martin, J.;
et al. Three new prodrugs for suicide gene therapy using
carboxypeptidase G2 elicit bystander efficacy in two xenograft
models. Cancer Res. 2002, 62, 1724-1729.
(7) Khan, T. H.; Eno-Amooquaye, E. A.; Searle, F.; Browne, P. J.;
Osborn, H. M. I.; et al. Novel inhibitors of carboxypeptidase G(2)
(CPG(2)): Potential use in antibody-directed enzyme prodrug
therapy. J. Med. Chem. 1999, 42, 951-956.
(8) Spooner, R. A.; Martin, J.; Friedlos, F.; Marais, R.; Springer, C.
J. In suicide gene therapy, the site of subcellular localization of
the activating enzyme is more important than the rate at which
it activates prodrug. Cancer Gene Ther. 2000, 7, 1348-1356.
(9) Marais, R.; Spooner, R. A.; Light, Y.; Martin, J.; Springer, C. J.
Gene-directed enzyme prodrug therapy with a mustard prodrug/
carboxypeptidase G2 combination. Cancer Res. 1996, 56, 4735-
4742.
(10) Workman, P.; Twentyman, P.; Balkwill, F.; Balmain, A.; Chaplin,
D.; et al. United Kingdom co-ordinating committee on cancer
research (UKCCCR) guidelines for the welfare of animals in
experimental neoplasia (second edition). Br. J. Cancer 1998, 77,
1-10.
In Vivo Therapy Studies. All experiments were conducted
in accordance with United Kingdom Home Office regulations
JM0502182