5-Nitrofuran-2-ylmethyl group as a potential bioreductively activated pro-drug system

Article abstract of DOI:10.1039/a607202j

5-Substituted isoquinolin-1-ones have been synthesised by one-pot Curtius rearrangement of the corresponding substituted 3-phenylpropenoyl azides and cyclisation. Arylmethylation of the anions of the isoquinolinones with benzyl halides [4-methoxybenzyl chloride, 2-(chloromethyl)furan and 5-nitro-2-(tosyloxymethyl)furan] takes place exclusively at nitrogen. Nitration of 2-(furan-2-ylmethyl)isoquinolin-1-one in strongly acidic medium gives 2-(5-nitrofuran-2-ylmethyl)isoquinolin-1-one, whereas weaker acidic conditions lead to dinitration. Curtius rearrangement of 3-carboranylbutanoyl azide and trapping with 5-nitrofuran-2-ylmethanol gives 5-nitrofuran-2-ylmethyl N-(3-carboranylpropyl)carbamate. Biomimetic reduction of these nitrofuranylmethyl derivatives of anticancer drugs triggers release of the parent drugs. Thus, these nitrofurans have potential applications as pro-drugs for selective release of therapeutic drugs in hypoxic solid tumours.

Full text of DOI:10.1039/a607202j

View Article Online / Journal Homepage / Table of Contents for this issue
5-Nitrofuran-2-ylmethyl group as a potential bioreductively activated
pro-drug system
Jane M. Berry,a Corrine Y. Watson,a William J. D. Whishb and
Michael D. Threadgill*,b
a
School of Pharmacy and Pharmacology, University of Bath, Claverton Down,
Bath BA2 7AY, UK
b School of Biology and Biochemistry, University of Bath, Claverton Down, Bath BA2 7AY, UK
5-Substituted isoquinolin-1-ones have been synthesised by one-pot Curtius rearrangement of the
corresponding substituted 3-phenylpropenoyl azides and cyclisation. Arylmethylation of the anions of the
isoquinolinones with benzyl halides [4-methoxybenzyl chloride, 2-(chloromethyl)furan and 5-nitro-2-
(tosyloxymethyl)furan] takes place exclusively at nitrogen. N itration of 2-(furan-2-ylmethyl)isoquinolin-1-
one in strongly acidic medium gives 2-(5-nitrofuran-2-ylmethyl)isoquinolin-1-one, whereas weaker acidic
conditions lead to dinitration. Curtius rearrangement of 3-carboranylbutanoyl azide and trapping with
5-nitrofuran-2-ylmethanol gives 5-nitrofuran-2-ylmethyl N-(3-carboranylpropyl)carbamate. Biomimetic
reduction of these nitrofuranylmethyl derivatives of anticancer drugs triggers release of the parent drugs.
Thus, these nitrofurans have potential applications as pro-drugs for selective release of therapeutic drugs
in hypoxic solid tumours.
Regions of chronic and acute hypoxia are present in most solid
tumours owing to the primitive state of the tumour vascu-
lature.1 Viable cells in such tissue are relatively resistant to
radiotherapy and to many chemotherapeutic strategies.1 Much
effort has been expended 2 on development of radiosensitisers
with electron-affinity and bioreductively activated cytotoxins
for selective therapy of this tissue, and of a variety of pro-drugs
to deliver cytotoxins selectively to tumours. 1-Substituted
2-nitroimidazoles are known 3,4 to be selectively retained in
hypoxic tissue by reductive metabolism. However, relatively
little attention has hitherto been focussed on exploiting the
physiological difference in concentration of molecular oxygen
between normal and hypoxic tumour tissue by design of bio-
logically inactive pro-drug systems which, upon selective bio-
reduction in hypoxic tissue, would release known therapeutic
drugs only in that tissue. This would improve greatly the select-
ivity of biodistribution of such agents. Sykes et al. have
reported 5 early studies on a bioreductively triggered release
system based on 2-nitroarylamides, whereas 4-nitrobenzyl-
oxycarbonyl pro-drugs have been put forward 6 for use in
the Antibody-Directed Enzyme Prodrug Therapy (ADEPT)
strategy, using a bacterial nitroreductase attached to a tumour-
selective antibody. For the potential pro-drugs described here,
2-nitrofuran was selected as the redox-sensitive moiety. The
redox potential of this heterocycle is relatively high [E1
=
7
Ϫ325 mV for 2-methyl-5-nitro-N-(prop-2-enyl)furan-3-carbox-
amide],7 which would favour selective reductive metabolism in
hypoxic tumour tissue effected by endogenous enzymes such as
cytochrome P450 reductase.8 The general design of the pro-
drugs and the mechanisms of bioreductively triggered release
are shown in Scheme 1. For the simpler pro-drugs 1, the
5-nitrofuran-2-ylmethyl unit is attached to a heteroatom in the
drug. Reduction in hypoxic tumour tissue will give the corre-
sponding aminofuran 2 (or the analogous hydroxylamine). The
presence of an available electron pair will then promote frag-
mentation as shown to release the drug 3 in the tumour tissue.
This fragmentation is clearly not available to the nitrofuran 1.
Where appropriate, the nitrofuranylmethyl unit could be linked
to the drug by an additional readily fissile group, as in the car-
bamate pro-drugs 4. Bioreductively triggered fragmentation
will again afford 3 according to the mechanism shown. Here
Scheme 1 Proposed mechanisms of bioreductively triggered release of
drugs from general nitrofuranylmethyl pro-drugs 1 and 4
we report syntheses of examples of pro-drugs of each type
(1 and 4) and biomimetic studies on the release of parent drugs
3.
Isoquinolin-1-one 18d and several 5-substituted analogues
are potent inhibitors9 of poly(ADP-ribose)polymerase
(PARP), an enzyme with a central role in initiating excision
repair of DNA following damage by radiation or electrophilic
drugs. Thus inhibitors of PARP are potentiators of these
J. Chem. Soc., Perkin Trans. 1, 1997
1147

View Article Online
therapeutic strategies. Prodrugs which selectively release iso-
quinolinones in hypoxic tumour tissue would therefore act as
tumour-selective radiosensitisers. In the design of these pro-
drugs, it is necessary to conceal the arylcarboxamide motif,
with the N᎐H held syn to the carbonyl oxygen, which is essen-
tial for enzyme inhibitory activity.9,10 Within the context of the
nitrofuranylmethyl system, this could be achieved through
either 1-(5-nitrofuran-2-ylmethoxy)isoquinolines or 2-(5-nitro-
furan-2-ylmethyl)isoquinolin-1-ones, pro-drugs of general
type 1.
with tosyl chloride in the presence of powdered potassium
hydroxide afforded the tosylate 9 in modest isolated yield.
1-Alkoxyisoquinolines have been prepared by substitution in
the corresponding 1-chloroisoquinolines using sodium alkox-
ides under vigorous conditions, although the range of 1-alkoxy
substituents reported to have been introduced in this way has
been restricted to simple examples,20 such as methoxy and
ethoxy. 1-Benzyloxyisoquinoline has not been reported. Treat-
ment of 1-chloroisoquinoline21 with 5-nitrofuran-2-ylmethanol
7a and furan-2-ylmethanol 7b alkoxides gave only products of
furan degradation under a variety of reaction conditions,
making the corresponding pro-drugs unavailable by this
route.
The usual route for preparation of 2-substituted isoquino-
lin-1-ones has been oxidation of 2-substituted isoquinolinium
salts in the presence of hydroxide ion.22 However, this is
inappropriate for preparation of the proposed pro-drugs, owing
to the harsh reaction conditions required. Routes employing
N-alkylation of isoquinolin-1-ones were therefore investigated.
To form examples of pro-drugs of type 4, with the carbamate
link, 2-phenylethylamine was selected as a model and a car-
boranylalkylamine was selected as a drug for delivery. Boron
neutron capture therapy (BNCT) is under active investigation
for the treatment of various cancers, notably gliomas and
melanomas.11 When the 10B isotope is irradiated with slow
7
(‘thermal’) neutrons, an [n,α] reaction ensues, giving Li and
4He nuclei with kinetic energy (2.31 MeV). With this energy, the
α-particle has a range of ca. one cell diameter in biological
tissue and damage is limited to the cell containing the boron.
Early clinical failures of BNCT were attributed 12,13 to in-
adequate concentrations of 10B in the tumour tissue or to lack
of selectivity of disposition of 10B, leading to damage of nor-
mal tissue. Carboranes have been linked to nucleosides,14 to
porphyrins15 and to nitroimidazoles4,16,17 in attempts to target
boron selectively to tumours. Where a 5-nitrofuran-2-ylmethyl
N-(carboranylalkyl)carbamate forms pro-drug 4, selective bio-
reduction in a hypoxic tumour cell would release a carboranyl-
alkylamine, which, at the relatively acidic pH of a tumour cell,
would be protonated and hence unable to diffuse out through
the cell membrane. Boron would therefore accumulate in the
tumour tissue.
A
convenient synthesis of isoquinolin-1-ones by Curtius
rearrangement of cinnamyl azides and thermal cyclisation, in a
one-pot process, has been described by Eloy and Deryckere,23
although this is limited to isoquinolinones without electron-
withdrawing substituents. 2-Iodobenzyl alcohol 10 was oxidised
to the corresponding aldehyde 11a with pyridinium dichromate
and a direct condensation with malonic acid afforded the iodo-
cinnamic acid 13a (Scheme 3). 2-Methylbenzaldehyde 11b had
to be converted into its diethyl acetal 12 for efficient conden-
sation under the Knoevenagel–Doebner conditions24 to give
the methylcinnamic acid 13b. The corresponding bromocin-
namic acid 13c was prepared from 2-bromoiodobenzene in a
convenient modification of the iodine–selective Heck reaction
conditions used by Plevyak et al.25 Use of boiling propionitrile
as the reaction solvent obviated the need for the sealed tube
required to conduct the reaction at 100 ЊC in acetonitrile. Vari-
ous modifications of the original conditions of Eloy and Dery-
ckere23 were investigated for the Curtius rearrangement and
cyclisation sequence. The iodocinnamic acid 13a was converted
into its acid chloride 15a and treatment with sodium azide fur-
nished the acid azide 16a. For ease of isolation of the product
5-iodoisoquinolin-1-one 18a from the reaction mixture by
precipitation with water, the Curtius rearrangement and cycli-
sation were effected in boiling dry tetraglyme. The acid chloride
15b and acid azide 16b of the methylcinnamic acid were pre-
pared similarly, but the original boiling diphenyl ether was
found to be the optimum solvent for the high-yielding synthesis
and isolation of 5-methylisoquinolin-1-one 18b. The bromo-
isoquinolinone 18c was prepared similarly but in poor yield,
using the acid azide 16c formed from a mixed anhydride.
To provide nucleophilic and electrophilic reagents to intro-
duce the nitrofuranylmethyl group into the pro-drugs, the alco-
hol 7a and the nitrofuranylmethyl electrophiles 8a and 9 were
prepared (Scheme 2). Attempts to reduce 5-nitrofuran-2-
The relatively few reports26 of alkylation of isoquinolinones
indicate that alkylation of isoquinolin-1-ones with simple halo-
geno alkanes occurs predominantly at the nitrogen of the con-
jugate anion under a variety of conditions. Two sets of con-
ditions were used in a series of model experiments designed to
establish the site of reaction of isoquinolinone anions with halo-
genomethyl arenes and to optimise the reaction conditions. The
anion formed from 5-iodoisoquinolinone 18a and lithium hexa-
methyldisilazide was benzylated in high yield by benzyl chlor-
ide, giving 20a. The 5-bromo analogue 18c was also converted
into the N-(4-methoxybenzyl) derivative 20c under these con-
ditions. Benzylation of the sodium anion of isoquinolinone 18d
with benzyl bromide in DMF was also highly efficient in form-
ing 20d, although the yield in the corresponding benzylation of
5-methylisoquinolinone 18b to give 20b was poor. However,
treatment of the anion derived from reaction of isoquinolin-
1-one 18d with sodium hydride and with lithium hexamethyl-
disilazide with the chloromethylnitrofuran 8a failed to give the
required N-(nitrofuranylmethyl)isoquinolinone 19. Compound
19 was formed in poor isolated yield (24%) when the sodium
salt of 18d was treated with the more reactive tosylate 9 in
Scheme 2 Synthesis of electrophilic (nitro)furan-2-ylmethyl reagents
8a,b and 9. Reagents: i, Al(OPri)3, PriOH; ii, SOCl2, pyridine, CHCl3; iii,
TsCl, KOH, THF.
carbaldehyde 6 selectively at the aldehyde using sodium boro-
hydride were unsuccessful, leading only to products of degrad-
ation, in contrast to a previous report.18 However, Meerwein–
Ponndorf–Verley reduction gave the required nitrofuranyl-
methanol 7a almost quantitatively. Activation of the methylene
as an electrophile was attempted through conversion into the
corresponding chloromethylnitrofuran 8a and the tosylate 9,
respectively. Replacement of OH with Cl to give 8a was
achieved by treatment of 7a with thionyl chloride and pyridine
in a modification of the method of Wang et al.19 Reaction of 7a
1148
J. Chem. Soc., Perkin Trans. 1, 1997

View Article Online
Scheme 3 Synthesis of isoquinolin-1-ones 18a–c. Reagents: i, pyridinium dichromate, CH2Cl2; ii, CH2(CO2H)2, piperidine, pyridine; iii, HC(OEt)3,
SOCl2, EtOH; iv, H2C᎐CHCO2H, Pd(OAc)2, Et3N, EtCN; v, SOCl2, DMF; vi, EtO2CCl, Et3N, Me2CO; vii, NaN3, water, 1,4-dioxane or acetone;
᎐
viii, heat, Ph2O or (MeOCH2CH2OCH2CH2)2O.
scopy and, in the case of 20a, by IR spectroscopy. The CH2
groups resonated at δ 51.90, 51.68, 51.59, 44.33 and 44.16,
respectively, values which correspond closely to those typical
for ArCH2N but not ArCH2O. The IR spectrum of 20a con-
tained a band at 1650 cmϪ1, indicating a carbonyl group. The
structure of 20c was assigned by analogy. Since the target nitro-
furan 19 prepared by alkylation of 18d with 9 was identical with
material formed by selective nitration of 21d, the nitrofuranyl-
methylation must also have taken place at N, rather than at O.
As the combined yield over two steps 7a→9→19 was very low
(6%), an alternative longer route was developed. Reaction of
the lithium salt of isoquinolin-1-one 18d with freshly prepared
unstable chloromethylfuran 8b27 in THF gave a moderate
yield of the N-furanylmethylisoquinoline 21d. The 5-methyl
analogue 21b was formed similarly. Selective nitration at the
furan 5-position was then required to form 19. The usual con-
ditions for nitration of furans and related heterocycles are
relatively mild, e.g. acetyl nitrate or nitric acid in acetic acid.
However, all applications of these and other relatively weakly
acidic nitrating conditions gave only the dinitrated product 22
where the isoquinolinone 4-position has also reacted. In an
extensive study of substitution of isoquinolin-1-ones, Horning
et al.28 reported that the principal site of reaction of various
electrophiles was the 4-position. However, Kawazoe and
Yoshioka 29 noted that, on treatment of 18d with potassium
nitrate in concentrated sulfuric acid, nitration took place at the
5- and 7-positions, presumably owing to deactivation of the
heterocyclic ring by protonation. Adopting this approach to
selective deactivation of the nitrogen heterocycle, the reaction
of a solution of 21d in trifluoroacetic acid with nitric acid or,
preferably, copper() nitrate at low temperature effected select-
ive mononitration on the furan, giving the target N-(nitro-
furanylmethyl)isoquinolinone 19. Traces of the dinitro com-
pound 22 were also isolated but the major by-product was the
parent isoquinolinone 18d resulting from dealkylation under
the acidic conditions. The 5-methyl analogue 21b gave the
parent 5-methylisoquinolinone 18b as the only isolable product.
Thus the sequence 7b→8b→21d→19 proceeded in higher over-
all yield (16%) than the more direct sequence above.
Since it was planned to form the carbamate link in the tar-
get carborane-nitrofuran 38 by addition of 5-nitrofuran-2-
ylmethanol 7a to an appropriate isocyanate, model reactions
for the Curtius rearrangement and addition were investigated.
The model sequence, in which phenyl replaces carboranyl, also
Scheme 4 Synthesis of 1-(arylmethoxy)isoquinolines 20a–d, 2-(furan-
2-ylmethyl)isoquinolin-1-ones 21b,d and 2-(5-nitrofuran-2-ylmethyl)-
isoquinolin-1-one 19. Reagents: i, LiN(SiMe3)2, BnCl, THF or NaH,
BnBr, DMF; ii, LiN(SiMe3)2, 4-MeOC6H4CH2Cl, THF; iii, NaH, 9,
DMF; iv, LiN(SiMe3)2, 8b, THF; v, CF3CO2H, HNO3 or Cu(NO3)2; vi,
e.g. HNO3, HOAc.
DMF. Assignment of the structures of 20a,b,d and 21b,d as
being the N-(arylmethyl)isoquinolin-1-ones rather than the
1-(arylmethoxy)isoquinolines was made by 13C NMR spectro-
J. Chem. Soc., Perkin Trans. 1, 1997
1149

View Article Online
provides another target compound in which the biomimetic
reductively triggered release from a pro-drug 4 can be studied.
3-Phenylpropanoic acid 23 was converted into its acid chlor-
ide 24 and hence to the acid azide 25 (Scheme 5). Curtius
required carbon atoms were present as the carborane was
formed, was therefore developed. Hex-5-ynenitrile 30 was pre-
pared straightforwardly from 5-chloropentyne 29. Following
the standard method for synthesis of carboranes from alkynes
and decaborane(14) at elevated temperature in the presence of a
Lewis base,16,17,31 the cyanopropylcarborane 31 was prepared
in excellent yield. Acidic hydrolysis afforded the carborane-
butanoic acid 32. Formation of the acid chloride 33, substi-
tution with sodium azide and Curtius rearrangement of 34 in
warm chloroform afforded the isocyanate 35. This was not iso-
lated but was treated with benzyl alcohol under basic con-
ditions to give the Z-protected carboranylpropylamine 36.
This sequence served both as a model for the reaction of 35
with arylmethanols and as an entry into the synthesis of the
hitherto unreported carboranylpropylamine. Interestingly,
treatment of 36 with hydrogen in the presence of palladium did
not effect deprotection and it was necessary to remove the Z
group with hydrogen bromide to give the salt 37. With the car-
boranylpropylamine ‘drug’ 37 now available, the corresponding
pro-drug 38 was prepared by addition of nitrofuranylmethanol
7a to the isocyanate 35. As with the phenylethyl series above,
quantities of the analogous nitrofuranylmethyl ester 39 were
obtained from some runs, again indicating incomplete Curtius
rearrangement.
A mild method for selective chemical reduction of the nitro
group was needed to test release of ‘drugs’ from the two types
of pro-drug 1 and 4. In particular, the conditions must not
permit hydrogenolysis of the ‘benzylic’ CH2᎐O or CH2᎐N
bonds, which would not be biomimetic for the nitroreductases
and the cytochrome P450 reductase enzymes. Sodium boro-
hydride in the presence of palladium fulfils these criteria,17,32
although the usual solvent, methanol, was replaced by propan-
2-ol in these studies to minimise any alcoholysis. Firstly, iso-
quinolinone 18d was released in 67% yield by this method from
pro-drug 19, through the intermediacy of the aminofuran 40
(Scheme 7). The failure of the furanylmethyl analogue 21d to
release 18d under the same conditions serves to validate the
selectivity of the reduction by excluding a benzylic hydro-
genolysis mechanism. An analogous selective reduction of the
nitro group in the nitrofuranylmethyl carbamate 27 caused
release of 2-phenylethylamine 42 in satisfactory yield, via the
aminofuran 41. The physical properties of the carboranyl-
propylamine were not conducive to easy isolation from boron-
containing by-products in this reaction mixture. Therefore,
after reductively triggered cleavage, the amine 43 was trapped as
its Z derivative 36 by treatment with benzyl chloroformate.
The Z-trapped drug was obtained in 26% yield, the relatively
low yield being probably due to the isolation procedure. As a
final positive control experiment, the pro-drug 38 was subjected
to selective reduction of the nitro group by tin() chloride.33
After the tin complex had been decomposed with sodium
hydroxide, the carboranylpropylamine 43 was isolated in 40%
yield.
Scheme 5 Synthesis of nitrofuranylmethyl N-(2-phenylethyl)carba-
mate 27. Reagents: i, (COCl)2; ii, Me3SiN3, PhMe; iii, heat, PhMe; iv,
7a.
rearrangement in boiling toluene gave the isocyanate 26 which
was not isolated but was trapped by reaction with 7a under
basic conditions, giving the nitrofuranylmethyl carbamate 27.
From some runs of this reaction, significant yields of the ester
28 were also isolated, indicating incomplete Curtius rearrange-
ment.
Scheme 6 shows the application of this sequence to the syn-
A preliminary evaluation of the biological activities of the
pro-drug 19 and the corresponding delivered drug 18d was
made to check that the pro-drug is indeed a less potent inhibitor
of the target enzyme, PARP, than is the 2-unsubstituted
isoquinolin-1-one drug. PARP was extracted with aqueous
sodium chloride (0.4 ) from nuclei isolated from L929 murine
areolar cells. The enzyme activity was measured in the presence
and absence of test compounds by the rate of incorporation of
radioactivity from NADϩ labelled with 3H in the adenosine into
acid-insoluble material. At the test concentration, 10 µ, the
pro-drug 19 inhibited the enzyme by 60% whereas isoquinolin-
1-one 18d inhibited the activity by >95% at the same concen-
tration.
Scheme 6 Synthesis of nitrofuranylmethyl N-(3-carboranylpropyl)-
carbamate 38. Reagents: i, KCN, EtOH, water; ii, B10H14, MeCN; iii,
H2SO4, water; iv, SOCl2, DMF; v, NaN3, Me2CO, water; vi, heat,
CHCl3; vii, BnOH, Et3N, CHCl3; viii, HBr, HOAc; ix, 7a, Et3N, CHCl3.
thesis of the nitrofuranylmethyl N-(carboranylalkyl)carba-
mate 38. Although the carboranebutanoic acid 32 has
been reported 30 to be formed by carboxylation of the Grig-
nard reagent derived from 1-(3-bromopropyl)-1,2-dicarba-
closo-dodecaborane(12), the yield is low, owing to a com-
peting cyclisation to give cyclopentano[1,2]-1,2-dicarba-closo-
dodecaborane(12). An alternative method, in which all the
In conclusion, it can be seen that a potential bioreductively
triggered pro-drug system has been developed, based on reduc-
tion of 5-nitrofuran-2-ylmethyl derivatives. The nitrofuranyl-
methyl group has been linked directly to the 2-position of iso-
1150
J. Chem. Soc., Perkin Trans. 1, 1997