7-N-(mercaptoalkyl)mitomycins: Implications of cyclization for drug function

Article abstract of DOI:10.1021/ja0124313

The Kyowa Hakko Kogyo and Bristol-Myers Squibb companies reported that select mitomycin C(7) aminoethylene disulfides displayed improved pharmacological profiles compared with mitomycin C (1). Mechanisms have been advanced for these mitomycins that differ from 1. Central to many of these hypotheses is the intermediate generation of 7-N-(2-mercaptoethyl)mitomycin C (5). Thiol 5 has been neither isolated nor characterized. Two efficient methods were developed for mitomycin (porfiromycin) C(7)-substituted thiols. In the first method, the thiol was produced by a thiol-mediated disulfide exchange process using an activated mixed mitomycin disulfide. In the second route, the thiol was generated by base-mediated cleavage of a porfiromycin C(7)-substituted thiol ester. We selected four thiols, 7-N-(2-mercaptoethyl)mitomycin C (5), 7-N-(2-mercaptoethyl)porfiromycin (12), 7-N-(2-mercapto-2-methylpropyl)mitomycin C (13), and 7-N-(3-mercaptopropyl)porfiromycin (14), for study. Thiols 5 and 12-14 differed in the composition of the alkyl linker that bridged the thiol with the mitomycin (porfiromycin) C(7) amino substituent. Thiol generation was documented by HPLC and spectroscopic studies and by thiol-trapping experiments. The linker affected the structure of the thiol species and the stability of the thiol. We observed that thiols 5 and 12 existed largely as their cyclic isomers. Evidence is presented that cyclization predominantly occurred at the mitomycin C(7) position. Correspondingly, alkyl linker substitution (13) or extension of the linker to three carbons (14) led to enhanced thiol stability and the predominant formation of the free thiol species. The dominant reaction of thiols 5 and 12-14 or their isomers was dimerization, and we found no evidence that thiol formation led to mitosene production and aziridine ring-opening. These findings indicated that thiol generation was not sufficient for mitomycin ring activation. The potential pharmacological advantages of mitomycin C(7) aminoethylene disulfides compared with 1 is discussed in light of the observed thiol cyclization pathway.

Full text of DOI:10.1021/ja0124313

Published on Web 04/05/2002
7-N-(Mercaptoalkyl)mitomycins: Implications of Cyclization
for Drug Function
Younghwa Na,‡ Shuang Wang,‡ and Harold Kohn*,‡,|
Contribution from the Department of Chemistry, UniVersity of Houston,
Houston, Texas 77204-5641, and DiVision of Medicinal Chemistry and Natural Products,
School of Pharmacy, UniVersity of North Carolina, Chapel Hill, North Carolina 27599-7360
Received October 25, 2001
Abstract: The Kyowa Hakko Kogyo and Bristol-Myers Squibb companies reported that select mitomycin
C(7) aminoethylene disulfides displayed improved pharmacological profiles compared with mitomycin C
(1). Mechanisms have been advanced for these mitomycins that differ from 1. Central to many of these
hypotheses is the intermediate generation of 7-N-(2-mercaptoethyl)mitomycin C (5). Thiol 5 has been neither
isolated nor characterized. Two efficient methods were developed for mitomycin (porfiromycin) C(7)-
substituted thiols. In the first method, the thiol was produced by a thiol-mediated disulfide exchange process
using an activated mixed mitomycin disulfide. In the second route, the thiol was generated by base-mediated
cleavage of a porfiromycin C(7)-substituted thiol ester. We selected four thiols, 7-N-(2-mercaptoethyl)-
mitomycin C (5), 7-N-(2-mercaptoethyl)porfiromycin (12), 7-N-(2-mercapto-2-methylpropyl)mitomycin C (13),
and 7-N-(3-mercaptopropyl)porfiromycin (14), for study. Thiols 5 and 12-14 differed in the composition of
the alkyl linker that bridged the thiol with the mitomycin (porfiromycin) C(7) amino substituent. Thiol generation
was documented by HPLC and spectroscopic studies and by thiol-trapping experiments. The linker affected
the structure of the thiol species and the stability of the thiol. We observed that thiols 5 and 12 existed
largely as their cyclic isomers. Evidence is presented that cyclization predominantly occurred at the mitomycin
C(7) position. Correspondingly, alkyl linker substitution (13) or extension of the linker to three carbons (14)
led to enhanced thiol stability and the predominant formation of the free thiol species. The dominant reaction
of thiols 5 and 12-14 or their isomers was dimerization, and we found no evidence that thiol formation led
to mitosene production and aziridine ring-opening. These findings indicated that thiol generation was not
sufficient for mitomycin ring activation. The potential pharmacological advantages of mitomycin C(7)
aminoethylene disulfides compared with 1 is discussed in light of the observed thiol cyclization pathway.
The discovery of mitomycin C (1) and mitomycin A (2) in
the late 1950s heralded a new era in cancer chemotherapy.1,2
Findings documenting that the mitomycins3 targeted oxygen-
deficient cells provided new opportunities for cancer treatment
through the use of bioreductive alkylating agents.4 The clinical
success of 1 led to intense efforts by medicinal chemists to
prepare new mitomycins with improved pharmacokinetic and
therapeutic properties. Over 1000 compounds have been syn-
* To whom correspondence should be sent. Fax: 919-843-7835.
E-mail: harold•kohn@unc.edu.
‡ University of Houston.
| University of North Carolina.
(1) Wakaki, S.; Marumo, H.; Tomoika, K.; Shimizu, E.; Kato, G.; Kamada,
H.; Kudo, S.; Fujimoto, Y. Antibiot. Chemother. 1958, 8, 228-240.
(2) Hata, T.; Sano, Y.; Sugawara, R.; Matsumae, A.; Kanamori, K.; Shima,
T.; Hoshi, T. J. Antibiot. (Tokyo) Japan, Ser. A 1956, 9, 141-146.
(3) (a) Carter, S. K.; Crooke, S. T. Mitomycin C. Current Status and New
DeVelopments; Academic Press: New York, 1979. (b) Remers, W. A. The
Chemistry of Antitumor Antibiotics; Wiley: New York, 1979; Vol. 1, pp
271-276. (c) Franck, R. W.; Tomasz, M. In The Chemistry of Antitumor
Agents; Wilman, D. F. V., Ed.; Blackie and Sons Ltd.: Glasgow, Scotland,
1990; pp 379-394. (d) Tomasz, M. In Topics in Molecular and Structural
Biology: Molecular Aspects of Anticancer Drug-DNA Interactions; Neidle,
S., Waring, M., Eds.; Macmillan: New York, 1994; Vol. 2, pp 312-347.
thesized and evaluated.5 In the mid-1980s, the Kyowa Hakko
(e) Tomasz, M. Chem. Biol. 1995, 2, 575-579.
(4) (a) Lin, A. J.; Cosby, L. A.; Shansky, C. W.; Sartorelli, A. C. J. Med.
Kogyo and the Bristol-Myers Squibb companies reported that
Chem. 1972, 15, 1247-1252. (b) Lin, A. J.; Cosby, L. A.; Sartorelli, A. C.
ACS. Symp. Ser. 1976, No. 30, 71-86.
mitomycin C(7) aminoethylene disulfides 36 and 47 displayed
9
4666
J. AM. CHEM. SOC. 2002, 124, 4666-4677
10.1021/ja0124313 CCC: $22.00 © 2002 American Chemical Society

7-N-(Mercaptoalkyl)mitomycins
A R T I C L E S
Scheme 1. Proposed Mechanism for the Mode of Action of Mitomycin C
improved pharmacological profiles compared with 1.8,9 Signifi-
cantly, 3 was active in both 1-resistant P388 and non-hypoxic
cells.10 Both 3 and 4 were entered into clinical trials, and 3
advanced to phase II testing.
Scheme 2. Suggested Pathways for 3 and 4 Function
Suggestions were made that the therapeutic advantage of 3
and 4 stemmed from the mitomycin C(7) disulfide unit, the
disulfide-cleaved product 5, or both11-16 and that the mode of
action of 3 and 4 differed from the accepted enzymatic reductive
activation mechanism for 13b,c,17,18 (Scheme 1). Studies showed
that administering 3 to tumor-bearing mice led to the serum
albumin conjugate 11 (R′S ) serum albumin) (Scheme 2, route
(5) (a) Arai, H.; Kanda, Y.; Ashizawa, T.; Morimoto, M.; Gomi, K.; Kono,
M.; Kasai, M. J. Med. Chem. 1994, 37, 1794-1804. (b) Bradner, W. T.;
Remers, W. A.; Vyas, D. M. Anticancer Res. 1989, 9, 1095-1099.
(6) Kono, M.; Saitoh, Y.; Kasai, M.; Sato, A.; Shirahata, K.; Morimoto, M.;
Ashizawa, T. Chem. Pharm. Bull. 1989, 37, 1128-1130.
(7) Vyas, D. M.; Chiang, Y.; Benigni, D.; Rose, W. C.; Brander, W. T. In
Recent AdVances in Chemotherapy; Tshigami, J., Ed.; Anticancer Section;
University of Tokyo Press: Tokyo, 1985; pp 485-486.
(8) Compound 3 (KW-2149): (a) Ohe, Y.; Nakagawa, K.; Fujiwara, Y.; Sasaki,
Y.; Minato, K.; Bungo, M.; Niimi, S.; Horichi, N.; Fukuda, M.; Saijo, N.
Cancer Res. 1989, 49, 4098-4102. (b) Tsuruo, T.; Sudo, Y.; Asami, N.;
Inaba, M.; Morimoto, M. Cancer Chemother. Pharmacol. 1990, 27, 89-
93. (c) Morimoto, M.; Ashizawa, T.; Ohno, H.; Azuma, M.; Kobayashi,
E.; Okabe, M.; Gomi, K.; Kono, M.; Saitoh, Y.; Kanda, Y.; Arai, H.; Sato,
A.; Kasai, M.; Tsuruo, T. Cancer Res. 1991, 51, 110-115. (d) Ashizawa,
T.; Okabe, M.; Gomi, K.; Hirata, T. Anti-Cancer Drugs 1993, 4, 181-
188. (e) Dirix, L.; Gheuens, E. E. O.; van der Heyden, S.; van Oosterom,
A. T.; De Bruijn, E. A. Anti-Cancer Drugs 1994, 5, 343-354. (f) Dirix,
L.; Catimel, G.; Koier, I.; Prove, A.; Schrijvers, D.; Joossens, E.; de Bruijn,
E.; Ardiet, C.; Evene, E.; Dumortier, A.; Clavel, M.; van Oosterom, A.
Anti-Cancer Drugs 1995, 6, 53-63.
A) and that 11 resided for longer periods in the plasma than
3.14 In addition, mechanistic investigations suggested that
glutathione (GSH)-induced cleavage of 3 and 4 led to 5 (Scheme
2, route B), and that 5 can initiate drug-DNA adduction
processes.15,16
Central to many theories concerning 3 and 4 function is thiol
5. 7-N-(2-Mercaptoethyl)mitomycin C (5) has been neither
isolated nor characterized.13 In this paper, we provide details
on the formation of 5 and related thiols 12-14 and document
the reactivity of mitomycin C(7)-substituted thiols. Our findings
permit us to speculate on the importance of the C(7) amino-
ethylene disulfide unit in mitomycins 3 and 4 for drug function.
(9) Compound 4 (BMS-181174): (a) Bradner, W. T.; Rose, W. C.; Schurig,
J. E.; Florczyk, A. P. InVest. New Drugs 1990, S1-S7. (b) Doyle, T. W.;
Vyas, D. M. Cancer Treat. ReV. 1990, 17, 127-131. (c) Dusre, L.;
Rajagopalan, S.; Eliot, H. M.; Covey, J. M.; Sinha, B. K. Cancer Res.
1990, 50, 648-652. (d) Xu, B. H.; Singh, S. V. Cancer Res. 1992, 52,
6666-6670. (e) Rockwell, S.; Kemple, B.; Kelley, M. Biochem. Pharmacol.
1995, 50, 1239-1243. (f) Xu, B. H.; Gupta, V.; Singh, S. V. Br. J. Cancer
1994, 69, 242-246.
(10) Kobayashi, E.; Okabe, M.; Kono, M.; Arai, H.; Kasai, M.; Gomi, K.; Lee,
J.-H.; Inaba, M.; Tsuruo, T. Cancer Chemother. Pharmacol. 1993, 32, 20-
24.
(11) He, Q.-Y.; Maruenda, H.; Tomasz, M. J. Am. Chem. Soc. 1994, 116, 9349-
9350.
(12) Kohn, H.; Wang, S. Tetrahedron Lett. 1996, 37, 2337-2340.
(13) Wang, S.; Kohn, H. J. Med. Chem. 1999, 42, 788-790.
(14) Kobayashi, S.; Ushiki, J.; Takai, K.; Okumura, S.; Kono, M.; Kasai, M.;
Gomi, K.; Morimoto, M.; Ueno, H.; Hirata, T. Cancer Chemother.
Pharmacol. 1993, 32, 143-150.
(15) Masters, J. R. W.; Know, R. J.; Hartley, J. A.; Kelland, L. R.; Hendricks,
H. R.; Conners, T. Biochem. Pharmacol. 1997, 53, 279-285.
(16) McAdam, S. R.; Knox, R. J.; Hartley, J. A.; Masters, J. R. W. Biochem.
Pharmacol. 1998, 55, 1777-1783.
(17) (a) Iyer, V. N.; Szybalski, W. Science 1964, 145, 45-58. (b) Szybalski,
W.; Iyer, V. N. In Antibiotics. Mechanism of Action; Gottlieb, D., Shaw,
P. D., Eds.; Springer-Verlag: New York, 1967; Vol. 1, pp 211-245.
(18) Moore, H. W.; Czerniak, R. Med. Res. ReV. 1981, 1, 249-280.
9
J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002 4667

A R T I C L E S
Na et al.
Scheme 3. Synthetic Approaches to Mitomycin (Porfiromycin)
Thiols
limitations. The DTT-mediated cleavage of 15 to give thiol 16
proceeded at near-neutral “pH” values in MeOH with the con-
comitant release of a substituted thiopyridone readily detected
by HPLC. In these reactions, however, complex product profiles
were sometimes observed due to the presence of thiols (DTT,
substituted thiopyridines) that reacted with the generated mi-
tomycin thiol 16. Correspondingly, NaOMe cleavage of thiol
ester 17 gave 18 at low temperatures without production of
competing thiols, but these reactions proceeded only under basic
conditions.24
Mitomycins (porfiromycins) 20-27 were prepared to provide
information concerning the structure and reactivity of mitomycin
(porfiromycin) C(7)-substituted thiols. These eight mitomycins
Results
1. Experimental Design and Choice of Substrates. We
designed mitomycins that upon chemical activation gave mi-
tomycin C(7)-substituted thiols. Two general pathways were
developed (Scheme 3, routes A and B). In one route, mitomycins
(porfiromycins) 15 containing an activated C(7) disulfide unit
were prepared and treated with an external thiol (R′′′SH) to
initiate disulfide exchange19 to provide thiol 16 (Scheme 3, route
A). This pathway is similar to the drug activation steps11-13
proposed for KW-2149 (3) and BMS-181174 (4) (see Scheme
2, route B). D,L-Dithiothreitol (D,L-DTT) was the external thiol
of choice because of its reducing power and because most D,L-
DTT reactions proceed to completion.20 We also employed
L-DTT, GSH, and N,N′-dimethyl-N,N′-bis(mercaptoacetyl)-
hydrazine21 (DMH) (data not shown) in the disulfide cleavage
experiments. In the second pathway, porfiromycin 17 containing
a thiol ester moiety within the C(7) unit were prepared and then
the thiol ester bond cleaved with base22 (NaOMe) to yield the
porfiromycin C(7)-substituted thiol 18 (Scheme 3, route B).
Porfiromycin3b (19) is the N(1a)-methyl derivative of 1. We
introduced a methyl group at N(1a) in 17 to prevent base-
mediated rearrangement to the isomeric albomitomycin ana-
logue.23 The thiol-generating step (15 f 16, 17 f 18) for the
two protocols distinguished and defined their utility and
(porfiromycins) can be classified by the C(7) terminal substituent
(disulfide, thiol ester) and the structural composition of the linker
that bridged the C(7) terminal group with the mitomycin
(porfiromycin) C(7) amino substituent. Compounds 20-25 each
contained a terminal disulfide unit while 26 and 27 each
possessed a terminal thiol ester moiety. In 20, 21, and 26, we
maintained the ethylene bridge found for 3 and 4, and in 22-
25 and 27, we either modified the ethylene bridge by methyl
substitution (22, 23) or by extension of the ethylene linker to a
propylene group (24, 25, 27). Alterations in the ethylene linker
were made to enhance the stability of the generated free thiol.
2. Syntheses. Compounds 20-23 were prepared from 2 using
a modified procedure reported by Vyas and co-workers25 for
mitomycin C(7)-substituted disulfides. This route (Scheme 4)
benefited from the rapid synthesis of the intermediate mixed
disulfides and proceeded in two steps to a 42-90% overall yield.
(19) (a) Singh, R.; Whitesides, G. M. In The Chemistry of Sulphur-containing
Functional Groups, Supplement S; Patai, S., Rappoport, Z., Eds.; John Wiley
and Sons: Chichester, 1993; Chapter 13, pp 633-658. (b) Rabenstein, D.
L.; Weaver, K. H. J. Org. Chem. 1996, 61, 7391-7397.
(20) Cleland, W. W. Biochemistry 1964, 3, 480-482.
(21) Singh, R.; Whitesides, G. M. J. Org. Chem. 1991, 56, 2332-2337.
(22) (a) Patai, S. The Chemistry of the Thiol Group; Wiley: New York, 1974;
p 677. (b) Maskill, H. The Physical Basis of Organic Chemistry; Oxford
University Press: Oxford, 1985; p 159.
(23) (a) Kono, M.; Saitoh, Y.; Kasai, M.; Shirahata, K. J. Antibiot. 1995, 48,
179-181. (b) Kono, M.; Saitoh, Y.; Shirahata, K. J. Am. Chem. Soc. 1987,
109, 7224-7225.
(24) Use of lower levels of NaOMe (e.g., “pH” 9) led to either no or low levels
of reaction. Similarly, use of NH2NH2 or NH2OH in place of NaOMe was
unsuccessful.
9
4668 J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002

7-N-(Mercaptoalkyl)mitomycins
A R T I C L E S
Scheme 5. Synthesis of Unsymmetrical Disulfides 37 and 38
Scheme 4. Preparative Pathways for Mitomycin Disulfides 20-25
Disulfides 32-35 were prepared by treating the aminoal-
kanethiol hydrochlorides 28 and 29 with either 4,4′-dithiodipy-
ridine (30) or 2,2′-dithiobis(5-nitropyridine) (31). Use of excess
Scheme 6. Preparative Pathway for Porfiromycin Thiol Esters 26
and 27
30 and 31 and the incremental addition of the aminoalkanethiol
hydrochloride (28, 29) to a dilute alcoholic solution of the
symmetrical disulfide (30, 31) under Ar ensured the quantitative
formation of the mixed disulfides (32-35) and the prevention
of the disulfide from the aminoalkanethiol hydrochloride. The
reaction mixture was not purified but was immediately treated
with either 2 or 47. Since 3-aminopropanethiol hydrochloride
(36) was not commercially available, we prepared mixed
disulfides 37 and 38 by an alternative route beginning with
3-aminopropanol (39) (Scheme 5). Treatment of mitomycin F28
(47) with 37 and 38 gave porfiromycin C(7)-substituted disul-
fides 24 and 25, respectively.
Porfiromycins 26 and 27 were prepared using the four-step
procedure outlined in Scheme 6. Mitomycin F28 (47) was con-
verted to the porfiromycin derivatives 4829 and 4930 with etha-
nolamine (50) and 3-aminopropanol (39), respectively. Treat-
ment of 48 and 49 in THF with methanesulfonyl chloride and
triethylamine yielded mesylates 51 and 52, respectively. Com-
pounds 51 and 52 proved too unstable to permit chromatographic
purification. Accordingly, crude 51 and 52 were treated with
thiolacetic acid and triethylamine to give 2629 and 27. The over-
all yield of 26 and 27 from 2 (the precursor of 47) was 40-
57%.
3. Mitomycin (Porfiromycin) C(7)-Substituted Thiols
That Cyclize. 3.1. Thiol-Mediated Disulfide Cleavage of 20
and 21: Attempts To Prepare Thiol 5. 3.1.1. The HPLC
Profile. We first prepared thiol 5. Accordingly, 20 and 21 were
treated with D,L-DTT. We varied the temperature (0 to -78
(25) (a) Vyas, D. M.; Chiang, Y.; Doyle, T. W. Amino Disulfide Thiol Exchange
Products. U.S. Patent 4,866,180, September 12, 1989; Chem. Abstr. 1998,
108, 150167x. (b) Vyas, D. M.; Chiang, Y.; Doyle, T. W. Amino Disulfides.
U.S. Patent 4,803,212, February 7, 1989; Chem. Abstr. 1985, 102, 131829z.
(26) (a) Kiellor, J. W.; Brown, R. S. J. Am. Chem. Soc. 1992, 114, 7983-7989.
(b) Negro, A.; Garzo´n, M. J.; Martin, J. F.; El Marini, A.; Roumestant, M.
L.; La´zaro, R. Synth. Commun. 1991, 21, 359-369.
(27) For related references, see: (a) Kitano, M.; Kojima, A.; Nakano, K.;
Miyagashi, A.; Noguchi, T.; Ohashi, N. Chem. Pharm. Bull. 1999, 47,
1538-1548. (b) Lemaire-Audoire, S.; Savignac, M.; Genet, J.-P. Synlett.
1996, 1, 75-78.
(29) Kasai, M.; Saito, Y.; Kono, M.; Sato, A.; Sano, H.; Shirahata, K.; Morimoto,
M.; Ashizawa, T. Pharmacologically Active Mitomycin Derivatives. Eur.
Pat. Appl. EP0197,799, October 15, 1986; Chem. Abstr. 1987, 106, 49881j.
(30) Iyengar, B. S.; Lin, H.-J.; Cheng, L.; Remers, W. A.; Bradner, W. T. J.
Med. Chem. 1981, 24, 975-981.
(28) Kasai, M.; Kono, M. Synlett. 1992, 778-790.
9
J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002 4669

A R T I C L E S
Na et al.
Scheme 7. Origin of Multiple HPLC Peaks
°C), the ratio of D,L-DTT to mitomycins 20 and 21 (1-11:1),
the sequence of addition of mitomycin C(7)-substituted disulfide
(20, 21) and D,L-DTT, the solvent (methanol, buffered methanol,
acetone), the “pH” of the methanolic solution (5.5, 6.5), and
the concentration of the mitomycin in solution (0.1-100 mM).
The experiments were monitored by HPLC (200-400 nm,
photodiode array detection), and select products were verified
by comparison with authentic samples. Treatment of a deaerated
(Ar) methanolic (-78 °C) solution containing 21 and D,L-DTT
(10 equiv) led to the complete consumption of 21 (tR 32.1 min)
and the appearance of three peaks in the HPLC profile (365
nm) between tR 22.1-23.9 min (designated 5 and 53-55,
Scheme 7; see figure in Supporting Information and ref 13).
These peaks accounted for 40-80% of the mitomycin profile,
and the reaction profile remained essentially unchanged at -78
°C for ∼3 days. The same peaks between tR 22.1-23.9 min
were observed when we used DMH in place of DTT, acetone
for methanol, and 20 in place of 21 (data not shown). All
reactions gave detectable amounts of the mitomycin C(7)
symmetrical disulfide 5631 (tR 28.4 min). In the D,L-DTT-
mediated reactions, additional peaks were observed at tR 26.1
(57), 26.6 (58), 28.0,32a and 30.332b min. The relative amounts
of these adducts in the product profile depended on the reaction
conditions. Use of more concentrated solutions and high molar
equivalents of D,L-DTT led to increased yields of the tR 26.1
and 26.6 min compounds, which approached 30% of the total
mitomycin HPLC profile. Interestingly, the relative amounts of
the tR 26.1 min (57) peak was consistently higher than the 26.6
min (58) peak, an observation that may reflect the relative
stabilities of these two products.
complished by reacting 2 with cysteamine hydrochloride (28)
or cystamine dihydrochloride (59).31 We characterized thiol 5
and/or cyclized adducts 53-55 using the derivatization reagents
30, 31, and N-ethylmaleimide (NEM) (60).
3.1.3. Thiol 5 Derivatization Studies Using 30, 31, and 60.
Addition of either 30 or 31 to the D,L-DTT-treated 20 and 21
reactions at room temperature and below led to the loss of the
tR 22.1-23.9 (5, 53-55), 26.1 (57), and 26.6 (58) (when
observed) min HPLC peaks and the generation of either 20 or
21 as the major mitomycin product (HPLC analyses; see figure
in Supporting Information and ref 13). These findings indicated
that the tR 22.1-23.9, 26.1, and 26.6 min HPLC peaks can be
converted to thiol 5 and then trapped as thiol adducts 20 and
21 under the derivatization conditions. Since the tR 26.1 and
26.6 min peaks were not observed when we used DMH in place
of D,L-DTT these compounds are likely to be D,L-DTT adducts33
that undergo further change prior to derivatization.
We attempted without success to isolate the tR 26.1 (57) and
26.6 (58) min HPLC compounds by treating either 20 or 21
with D,L-DTT on a semipreparative scale. Additional information
concerning the structure of these compounds was obtained by
carrying out a parallel experiment with L-DTT in place of D,L-
(31) (a) Kono, M.; Saito, Y.; Goto, J.; Shirahata, K.; Morimoto, M.; Ashizawa,
T. Mitomycin Analogs (Kyowa Hakko Kogyo Co., Ltd.) Eur. Pat. Appl.
EPO 116,208A1, August 22, 1984; Chem. Abstr. 1985, 102, 6059. (b)
Senter, P. D.; Langley, D. R.; Manger, W. E.; Vyas, D. M. J. Antibiot.
1988, 41, 199-201.
3.1.2. Analysis of the HPLC Profile: The tR 22.1-23.9,
26.1, and 26.6 min Peaks. The identities of the tR 22.1-23.9,
26.1, and 26.6 min HPLC peaks were puzzling. We expected a
single peak for thiol 5. We tentatively assigned the tR 22.1-
23.9 min HPLC peaks to thiol 5 and/or the corresponding C(6)
(53), C(7) (54), and C(8) (55) cyclic isomeric adducts (Scheme
7).
We sought additional evidence that the tR 22.1-23.9 min
HPLC peaks were associated with isomeric forms of 5. Attempts
to isolate these peaks gave 56. Verification of 56 was ac-
(32) (a) We have tentative information concerning the structure of the 28.0 min
HPLC peak. Repetition of this reaction under semipreparative conditions
led to the isolation of two compounds corresponding to this signal whose
HPLC retention times (instrument B) corresponding to 27.6 and 27.7 min.
Mass spectroscopic studies indicated that these adducts were the diaster-
eomeric DTT-bridged mitomycin disulfides. For the 27.6 min HPLC adduct,
MS (+ES) m/e 939.0 [M + H]+ (23%), 470.1 [M + 2H]2+ (100%); for
the 27.7 min HPLC adduct, MS (+ES) m/e 938.8 [M + H+]+ (8%), 469.9
[M + 2H]2+ (100%). (b) We have not identified this adduct.
(33) For structurally related adducts, see: Li, Y.-J.; Rothwarf, D. M.; Scheraga,
H. A. J. Am. Chem. Soc. 1998, 120, 2668-2669.
9
4670 J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002

7-N-(Mercaptoalkyl)mitomycins
A R T I C L E S
DTT. L-DTT gave an HPLC profile nearly identical to that of
D,L-DTT except that only the tR 26.1 min peak (57) was detected.
Co-injection of the D,L-DTT and L-DTT reactions suggested that
the tR 26.1 min adduct 57 was derived from L-DTT. These
cumulative results indicate that the tR 26.1 and 26.6 min peaks
are the L- and D-diastereomeric mixed disulfides 57 and 58,
respectively, and that both compounds can be converted to thiol
5 and then trapped with disulfides 30 and 31.
Use of GSH (10 equiv) for mitomycin disulfide 20 and 21
cleavage in MeOH (22 °C) led to the formation of a new HPLC
product at tR 18.0 min (65) (>85%). It was significant that we
We also tested whether 20 or 21 produced in the thiol deriva-
tization experiments originated from 56 present in the product
mixture. Treatment of 56 with D,L-DTT (1-2 equiv) followed
by either 30 (1.4 equiv) or 31 (3 equiv) at room temperature
yielded 20 and 21, respectively. However, the rate of conversion
of 56 to either 20 or 21 was measurably slower (room
temperature, 1-3 days) than the corresponding rate of 20 and
21 production after either 30 or 31 was added to the D,L-DTT-
treated 20 and 21 reactions (room temperature, <2 min). This
demonstrated that 56 was not the precursor to these adducts.
We obtained similar findings when we used 60 in place of
either 30 or 31 in the thiol derivatization experiments. Addition
of 60 to D,L-DTT-treated methanolic solutions of 20 and 21 led
to the disappearance of the tR 22.1-23.9 (5, 53-55), 26.1, (57),
and 26.6 (58) min HPLC peaks and the concomitant generation
of two HPLC peaks at tR 25.4 and 26.0 min corresponding to
NEM-substituted adducts 61 and 62, respectively.
observed only low amounts of dimer 56 (∼8%). By repeating
the reaction on the preparative scale using 20 followed by chro-
matographic purification, we isolated and characterized the tR
18.0 min compound as the mitomycin-GSH mixed disulfide 65.
3.3. NaOMe-Mediated Cleavage of 26: Attempts to
Prepare Thiol 12. Our assignment of the tR 22.1-23.9 min
HPLC peaks in the DTT-mediated cleavage reactions of 20 and
21 to thiol 5 and isomeric cyclized adducts 53-55 required
further support. We sought an independent route to a comparable
species and determined whether base cleavage of thiol ester 26
provided 12. This method would generate 12 in the absence of
competing thiols, such as DTT and thiopyridone byproducts.
Cleavage of the thiol ester bond in 26 was accomplished using
NaOMe at -78 °C under dilute conditions. The HPLC product
profile (see figure in Supporting Information and ref 13) was
cleaner than those recorded for the D,L-DTT-cleavage reactions
of 20 and 21. We observed three peaks between tR 23.9 and
26.6 min; these HPLC signals have been assigned to thiol 12
and/or the corresponding isomeric forms 66-68 (Scheme 7).
We attributed the longer retention times (2-3 min) of 12 and
66-68 compared with 5 and 53-55 to the increased hydro-
phobicity of the porfiromycin caused by N(1a)-methylation.34
Low levels (∼10%) of the dimeric porfiromycin disulfide 69
(tR 31.2 min) were observed along with an unidentified peak
(tR 29.1 min, 7-18%).35 Noticeably absent from the HPLC
profile for the base-mediated reactions were the corresponding
later retention time peaks (tR >26 min) assigned to thiol- D,L-
DTT adducts (57, 58). We monitored (HPLC) the base cleavage
of 26 against time. The reaction occurred rapidly under these
conditions (<1 h), and the product profile remained largely
unaltered for 3 days. We verified 69 in the HPLC profile by
preparing an authentic sample by treating 47 with cystamine
dihydrochloride (59).31a The major product isolated was 70.
3.2. GSH-Mediated Disulfide Cleavage of 20 and 21.
Compounds 57 and 58 are the expected initial products for D,L-
DTT-mediated cleavage of 20 and 21 along with the released
thiopyridinone. Few DTT-based disulfide adducts have been re-
ported33 since subsequent intramolecular cyclization and disul-
fide cleavage to give oxidized DTT (63) is energetically favor-
able. Our evidence that the tR 26.1 and 26.6 min HPLC peaks
are 57 and 58 is circumstantial and is based on the apparent
conversion of these products to 20, 21, 61, and 62 upon addition
of the thiol derivatization agent and the detection of a single
DTT-bound adduct 57 with L-DTT. Accordingly, we obtained
additional support for mixed disulfide 57 and 58 formation.
Mitomycin C(7) aminoethylene disulfides 20 and 21 were
treated with a thiol that would generate a mixed disulfide product
capable of isolation and characterization. We chose GSH (64).
This cellular thiol has special significance in light of the
mechanistic proposals, suggesting GSH initiates 3 and 4 function
(Scheme 2, route B).15,16
(34) A similar difference in HPLC retention times was observed for mitomycin
C (16.7 min) and porfiromycin (18.1 min).
(35) The 29.1 min HPLC peak has not been identified and may correspond to
a cyclic adduct. Addition of trapping reagents led to the disappearance of
this adduct.
9
J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002 4671

A R T I C L E S
Na et al.
Evidence supporting that the tR 23.9-26.6 min HPLC peaks
corresponded to thiol 12 and/or 66-68 was provided by thiol-
disulfide exchange reactions. Successive treatment of a metha-
nolic solution of 26 with NaOMe followed by 30 led to the
loss of the three peaks between tR 23.9-26.6 min and the tR
29.1 min signal and gave 71 (tR 30.2 min) as the sole
porfiromycin product (HPLC analysis, see figure in Supporting
Information). Similarly, 26 furnished 72 (tR 34.4 min) after
treatment with base and 31. The identities of 71 and 72 were
confirmed by co-injection with authentic samples prepared by
treatment of 4728 with mixed disulfides 32 and 33, respectively.
3.4. Correlation of the DTT- and NaOMe-Mediated
Routes to Mitomycin (Porfiromycin) Thiols 5 and 12.
Multiple HPLC peaks associated with thiols 5 (tR 22.1-23.9
min) and 12 (tR 23.9-26.6 min) were observed after DTT and
NaOMe treatment of mitomycins 20 and 21 and porfiromycin
26, respectively. We confirmed that these HPLC peaks were
comparable by treating methanolic solutions of porfiromycin
mixed disulfides 71 and 72 with D,L-DTT (-78 °C). Under these
conditions, we observed the same peaks (tR 23.9-26.6 min)
obtained with thiol ester 26 and NaOMe. Furthermore, addition
of either 30 or 31 to these D,L-DTT cleaved products led to the
disappearance of the multiple HPLC peaks (tR 23.9-26.6 min)
and the production of 71 and 72, respectively.
4.3.1. L-DTT-Mediated Disulfide Cleavage Reactions:
Preparation of Thiol 14. When we added a methanolic L-DTT
(2 equiv) solution to a deaerated (Ar) methanolic solution of
25, we saw a simple HPLC profile consisting of three principal
porfiromycin products: one corresponding to 14 (tR 27.2 min),
one to the porfiromycin dimer 74 (tR 33.1 min), and a third
compound whose HPLC retention time was tR 31.7 min (75)
(see figure in Supporting Information) along with an unidentified
minor adduct. We confirmed formation of 74 by co-injection
(HPLC) and cospot (TLC) of the reaction mixture with an
authentic sample. Repetition of the reaction with 24 in place of
25 provided similar results. We prepared an authentic sample
of symmetrical disulfide 74 by treating 47 with 3-amino-1-
propyl disulfide‚2HCl (76) in methanolic TEA solution that was
4. Mitomycin (Porfiromycin) C(7)-Substituted Thiols That
Do Not Cyclize. 4.1. Introduction. The DTT- and NaOMe-
generated HPLC products for mitomycin (porfiromycin) C(7)
aminoethanethiols 5 and 12 were assigned as a mixture of free
thiol and cyclic adducts. Attempts to isolate 5 and 12 gave only
symmetrical disulfides 56 and 69, respectively. We sought to
prepare thiols that do not cyclize and so targeted 13 and 14. In
13, we placed a gem-dimethyl unit adjacent to the thiol moiety,
and in 14, we incorporated a propylene bridge between the thiol
unit and the porfiromycin C(7) amino substituent. We reasoned
in 13 that the gem-dimethyl unit would sterically impede
quinone cyclization processes (C(6)-C(8)) and diminish the rate
of free thiol dimerization to symmetrical disulfide 73, while in
14 the propyl side chain would entropically diminish intramo-
lecular cyclization processes.
4.2. Mitomycin Thiol 13: Effect of the Gem-Dimethyl
Unit. We generated 13 in acetone at room temperature using
22 and D,L-DTT. Mixed disulfide cleavage proceeded at a rate
considerably slower than that observed for 20 and 21. Moreover,
the 22 HPLC product profile for this transformation (see figure
in Supporting Information) was considerably cleaner than
previously observed for mitomycin disulfides 20 and 21. A
single major mitomycin peak was observed at tR 25.1 min along
with two minor peaks at tR 23.2 and 27.0 min. We have
tentatively assigned the tR 25.1 min peak to thiol 13. No
mitomycin dimer 73 was produced, and little differences in the
HPLC profile were observed after 1 day of storage at room
temperature. Attempts to isolate thiol 13 by preparative TLC
were unsuccessful. We found evidence that the three HPLC
peaks at tR 23.2, 25.1 and 27.0 min were associated with free
thiol 13 when we sequentially treated 22 with D,L-DTT (acetone,
room temperature) and then with excess 31 to give quantitatively
23 (HPLC analysis). We were unable to cleave the mixed
disulfide bond in 23 with D,L-DTT (acetone, room temperature)
(HPLC and TLC analyses).
not deaerated. Compound 76 was prepared from 44 by t-Boc
deprotection with HCl.
Repeating the reaction of 25 with L-DTT on a semipreparative
scale followed by chromatographic purification of the reaction
mixture permitted isolation of the tR 31.7 min compound 75.
We have assigned this compound as the L-DTT-bridged bis-
1
disulfide 75 on the basis of H NMR, 13C NMR, UV-visible,
and mass spectral data. This compound is a structural analogue
of the suspected DTT-bridged adducts32 (tR 28.0 min) observed
in DTT-mediated cleavage reactions beginning with either 20
or 21.
We determined the effect of solution “pH” (buffered metha-
nolic solutions) on the L-DTT-mediated cleavage process for
25. The reaction proceeded faster at “pH” 7.4 than at “pH” 5.5.
Furthermore, we observed that at “pH” 5.5 (70 min), the relative
yield of the L-DTT-bridged bisporfiromycin disulfide 75 (61%)
increased and the yield of porfiromycin thiol 14 (19%)
decreased, compared with the “pH” 7.4 transformation (5 min)
(75, 35%; 14, 61%). These results are consistent with the notion
that L-DTT cleavage of 25 first generated L-DTT-porfiromycin
disulfide 77, which is similar to 57. We suspect that 77 is more
4.3. Porfiromycin Thiol 14: Effect of the Propylene Unit.
We sought to prepare free thiol 14 using the same strategies
implemented for 5 and 12.
stable at “pH” 5.5 than at “pH” 7.4, leading to higher levels of
the coupled product 75. Correspondingly, at “pH” value 7.4,
9
4672 J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002

7-N-(Mercaptoalkyl)mitomycins
A R T I C L E S
77 undergoes accelerated intramolecular cyclization to give 14
and oxidized DTT (63).
4.3.2. Thiol 14 Derivatization Studies Using 30 and 31.
Chemical derivatization experiments provided support for the
assignment of the tR 27.2 min HPLC peak as free thiol 14. HPLC
analysis of a methanolic solution (0 °C) of 25 sequentially
treated with L-DTT and then 30 showed two major peaks at tR
30.5 and 32.2 min. The latter peak has been assigned as 24 and
was confirmed by co-injection (HPLC) of the reaction solution
with an authentic sample. We identified the tR 30.5 min adduct
by repeating the reaction on a semipreparative scale. Spectro-
scopic analysis of the isolated product was consistent with the
mixed bisdisulfide 78.
These findings indicated that 77 generated after L-DTT
cleavage of 25 is, in part, converted to 14 and then trapped by
30 to give 24 and, in part, reacts with 30 to afford 78. We
repeated the thiol disulfide exchange experiment with 24 in place
of 25 and 31 in place of 30. We observed similar results and
the appearance of a new peak (tR 33.9 min) that has been
tentatively assigned as 79, on the basis of its retention time.
This peak eluted ∼3.4 min later than the 78 peak (tR 30.5 min).
The HPLC retention time difference between 78 and 79 is nearly
the same found for 24 (tR 32.2 min) and 25 (tR 35.2 min). We
did not further characterize the tR 33.9 min HPLC peak.
4.3.3. NaOMe-Mediated Cleavage of 27: Preparation of
Thiol 14. Porfiromycin thiol 14 was generated from methanolic
solutions of 27 with NaOMe at -78 °C. The HPLC product
profile for this reaction (see figure in Supporting Information)
differed from the corresponding reaction of 26. For 26, we
observed multiple peaks while 27 gave two major peaks at tR
28.1 and 33.8 min in an approximate 7:3 ratio, respectively,
along with a minor signal at tR 35.6 min. We assigned the tR
28.1 min peak to thiol 14 and the tR 33.8 min signal to the
symmetrical disulfide 74. Thiol ester bond cleavage in 27
occurred rapidly at -78 °C, and the product profile remained
largely unchanged after 1 day. Elevation of the reaction
temperature from -78 °C to room temperature converted thiol
14 to disulfide 74.
4.3.4. Thiol Derivatization Studies Using 30 and 31. Thiol
derivatization reactions provided support that the tR 28.1 min
peak corresponded to free thiol 14. First, treatment of methanolic
solution of 27 with NaOMe followed by 30 at -78 °C gave 24
and 74. The identities of 24 and 74 were confirmed by
co-injection of authentic samples with the reaction mixture.
HPLC analysis of the reaction both before and after 30 was
added indicated that the relative amount of dimer 74 was not
affected by 30 addition. An analogous reaction was conducted
using 31 in place of 30 to provide 25 and 74 (see HPLC profile
(figure) in Supporting Information).
Figure 1. (a) 1H NMR spectrum of 7-N-(3′-mercaptopropyl)porfiromycin
(14) in CD3OD. (b) 1H NMR spectrum of 7-N,7′-N-dithiobis(3,1-pro-
panediyl)bisporfiromycin (74) in CD3OD.
to -4 °C and monitored by 1H NMR spectroscopy. The initial
1H NMR spectrum after 0.5 h showed the consumption of 7-N-
(3-acetylthiopropyl)porfiromycin (27) and the production of 14
(Figure 1a). In particular, we observed the disappearance of the
characteristic thioacetyl signal (δ 2.28) in 27 and the appearance
of the C(2′) and the C(3′) methylene protons at δ 1.85 and 2.56
as multiplet and triplet signals, respectively. These values (δ
1.85, 2.56) are upfield from comparable resonances for the
symmetrical porfiromycin disulfide 74 (δ 1.99, 2.75). This
difference in chemical shift values is diagnostic of thiol and
disulfide structures.36 Maintenance of the basic NMR solution
for 9 h led to the gradual loss of the signals associated with 14
and the quantitative production of disulfide 74 (Figure 1b). The
conversion of 14 to 74 prevented us from acquiring the 13C
NMR spectrum of 14.
Discussion
Synthesis of Mitomycin (Porfiromycin) C(7)-Substituted
Thiols. We utilized two methods that enabled us to generate
mitomycin (porfiromycin) C(7)-substituted thiols, isomers, or
both. The first method was a thiol-mediated disulfide exchange
process19 using an activated mixed mitomycin (porfiromycin)
disulfide (Scheme 3, route A). Incorporation of either the
4-thiopyridine (20, 22, 24, 71) or the 5-nitro-2-thiopyridine (21,
23, 25, 72) unit in the mitomycin (porfiromycin) disulfide
permitted selective and rapid disulfide cleavage with external
thiols (e.g., DTT). The second route was the base-mediated
cleavage of thiol esters22 (Scheme 3, route B). Accordingly, we
incorporated a thiol acetyl unit at the C(7) terminal end in 26
and 27, which permitted efficient NaOMe cleavage.
4.3.5. Structural Characterization of Thiol 14. The sim-
plicity of the HPLC product profile for the NaOMe-mediated
cleavage of 27 permitted us to characterize thiol 14 by UV-
1
visible and H NMR spectroscopy. The UV-visible spectrum
obtained by HPLC photodiode array detection showed two major
absorptions at ∼220 and ∼370 nm in a near 1:1 ratio. This
spectrum is similar to that observed for porfiromycin (19).
We modified our experimental procedure to permit charac-
terization of 14 by 1H NMR spectroscopy. A NaOCD3-
CD3OD solution was added to a concentrated CD3OD solution
containing 27 at -78 °C and then the solution temperature raised
(36) Pretsch, E.; Clerc, T.; Seibl, J.; Simon, W. Tables of Spectral Data for
Structure Determination of Organic Compounds, 2nd ed. Springer-Verlag
Berlin, 1989.
9
J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002 4673

A R T I C L E S
Na et al.
In total, four different thiols were selected for preparation
and evaluation. The first was 7-N-(2-mercaptoethyl)mitomycin
C (5), which corresponded to the putative intermediate generated
in the KW-2149 (3) and BMS-181174 (4) activation process-
es11-13 (Scheme 2, path B). For this thiol, we used a thiol-
mediated disulfide exchange reaction using either 20 or 21. The
second was the porfiromycin analogue of 5, 7-N-(2-mercapto-
ethyl)porfiromycin (12). To produce this thiol, we employed
both a thiol-mediated disulfide exchange reaction using 71 and
72 and the base-mediated thiol ester cleavage route using 26.
The final two compounds were the gem-dimethyl analogue of
5, 7-N-(2-mercapto-2-methylpropyl)mitomycin C (13) and the
12 homologue 7-N-(3-mercaptopropyl)porfiromycin (14). Syn-
thesis of 13 was accomplished using a thiol-mediated disulfide
exchange process from 22 while both synthetic strategies (thiol-
mediated disulfide cleavage of 24 and 25, base-mediated thiol
ester cleavage of 27) were employed to generate 14.
We have assigned the multiple HPLC peaks obtained from
mitomycins 20 and 21 to the mitomycin free thiol 5 and/or
isomers 53-55. A similar assignment has been given to the
products (12, 66-68) from porfiromycins 26, 71, and 72. The
observation of multiple products for both 5 and 12 was
surprising. The corresponding alcohols, (2-hydroxyethyl)mito-
mycin C (tR 18.4 min, data not shown) and porfiromycin alcohol
48 (tR 19.8 min) showed only a single peak in the HPLC.
A qualitatively different HPLC profile was observed for
reactions that generated thiols 13 and 14. For 13, we observed
a major peak (tR 25.1 min) flanked by two minor peaks (tR 23.2,
27.0 min) while for 14 we detected only a single major peak
(tR 28.2 min). We have assigned the tR 25.1 min and the tR
28.2 min peaks to free thiols 13 and 14, respectively. Addition
of thiol-trapping reagents (30, 31, 60) led to the disappearance
of these major and minor HPLC peaks and the production of a
new signal associated with the thiol-modified derivative.
The UV-visible absorption spectra (photodiode array detec-
tion) for the multiple peaks assigned to 5 and 12 and/or isomers
were of interest. All of the adducts exhibited absorptions at 225-
245 and 365-376 nm in an approximate 1:1 ratio. This pattern
was comparable with 1, 19, and 48 (UV-vismax λ 222-225
(∼1), 360-366 nm (∼1)). We asked whether disruption of the
quinone unit by C(6)-C(8) cyclization would alter the UV-
visible absorption spectrum associated with mitomycin C and
porfiromycin derivatives. Only a few reports have appeared on
the UV-visible spectrum of quinone-modified mitomycin
(porfiromycin) analogues. Kanda and Kasai showed that the
C(6), C(7)-modified mitomycin 80 exhibited major absorptions
Each thiol preparative method had advantages and disadvan-
tages. The thiol-mediated disulfide exchange process (Scheme
3, route A) proceeded rapidly in methanolic and acetone
solutions at low temperatures (e0 °C) for 20, 21, 24, 25, 71,
and 72. Elevated temperatures (22 °C) were needed for gem-
dimethyl derivative 22. Moreover, when we replaced the
4-thiopyridone-leaving group in 22 with the 5-nitro-2-thiopy-
ridone to give 23, we observed no reaction upon DTT addition
at room temperature. DTT-mediated thiol cleavage occurred at
“pH” 5.5, 6.5, and 7.4 in MeOH. These reaction parameters
allowed us to monitor mitomycin (porfiromycin) thiol generation
and subsequent processes under mild conditions. Important
limitations of this protocol, however, were the need to use excess
external thiol (3-11 equiv) at low temperatures to generate the
mitomycin (porfiromycin) thiol and the introduction into the
reaction solution of multiple external thiols (e.g., DTT, 4-thio-
pyridine, 5-nitro-2-thiopyridine) that could react with the newly
generated mitomycin (porfiromycin) thiol. The NaOMe cleavage
route of porfiromycin thiol esters 26 and 27 (Scheme 3, route
B) eliminated the need for external thiols. This reaction
proceeded rapidly at both room temperature and low temper-
atures (0 to -78 °C). Despite this advantage, the method
required our using basic solutions (“pH” 10-11).24
at 237 and 359 nm in a l:1 ratio.37 Kono and co-workers reported
that albomitomycin C (81), a C(4a), C(8a)-modified isomer of
1, displayed absorptions at 242 and 344 nm in a 1:1 ratio.23
Recently, we determined the UV-visible spectra of porfiro-
mycins 82 and 8338 and mitomycin 84.39 The spectra of 82 and
HPLC and Spectroscopic Characterization of the DTT-
and NaOMe-Generated Mitomycin (Porfiromycin) Products.
DTT activation of methanolic and acetone solutions containing
either 20 or 21 generated distinctive chromatograms containing
three peaks (tR 22.1-23.9 min) that correlated with mitomycin
C(7)-substituted thiol 5 formation. We expected a single peak.
The same set of peaks was observed regardless of the starting
material (20, 21), method of activation (D,L-DTT, L-DTT,
DMH), solvent (acetone, methanol), “pH” of the methanol
solution (5.5, 6.5), or temperature (room temperature to -78
°C). A comparable pattern (tR 23.9-26.6 min) was observed
upon base cleavage of porfiromycin thiol ester 26. The same
profile was obtained using D,L-DTT and either disulfide 71 or
72. This finding demonstrated that both thiol synthetic strategies
(Scheme 3) generated the same intermediates. Addition of thiol-
trapping reagents (30, 31, 60) led to the rapid disappearance of
these three peaks and the concomitant appearance of peaks for
the mitomycin (porfiromycin) C(7)-substituted thiol-modified
products.
83 consisted of major absorptions at 235-238 and 353-362
nm in a 1.2:1 ratio while 84 exhibited absorptions at 240 and
410 nm in a 0.9:1 ratio.
(37) Kanda, Y.; Kasai, M. J. Org. Chem. 1990, 55, 2515-2518.
(38) Na, Y.; Kohn, H. Heterocycles 2001, 55, 1347-1363.
(39) Wang, S.; Kohn, H. J. Org. Chem. 1996, 61, 9202-9206.
9
4674 J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002

7-N-(Mercaptoalkyl)mitomycins
A R T I C L E S
The UV-visible absorption maximums observed for the
multiple peaks associated with thiols 5 and 12 indicate that these
adducts are likely to be C(7) or C(6) cyclized adducts. Since
no C(6) cyclized adducts have been reported, we suggest these
are the free thiols 5 and 12 and the corresponding isomeric C(7)
adducts 54 and 67, respectively. In principle, C(7) cyclization
can produce four isomers that differ in their C(7) and C(6)
stereochemistry.
The UV-visible spectra (photodiode array detection) for 13
and 14 were comparable with 1 and 19. For 13, we observed
major absorptions between 240-243 and 364-377 nm in an
approximate 1:1 ratio for both the major (25.1 min) and the
minor (23.2, 27.0 min) HPLC peaks. Correspondingly, porfiro-
mycin thiol 14 exhibited absorptions at 220 and 370 nm in an
approximate 1:1 ratio.
of the disulfide bond at low temperatures (e0 °C). Similarly,
treatment of 26 and 27 with NaOMe in MeOH at -78 °C led
to rapid cleavage of the thiol ester bond. Correspondingly, we
observed that incorporation of a gem-dimethyl unit within the
linker noticeably reduced the ease with which these mitomycins
(22, 23) underwent disulfide cleavage with D,L-DTT in com-
parison with 20 and 21. For example, for 22, D,L-DTT (DMH)-
mediated cleavage only occurred at room temperature (acetone),
and no reaction was observed for 23 under these conditions
(HPLC, TLC analyses). We suspect that the bulky gem-dimethyl
group adjacent to the disulfide unit hampered bond breakage.
Different methods (disulfide exchange, thiol ester cleavage)
and conditions (solvents, temperatures, concentrations) were
used to generate thiols or isomers 5 and 12-14 and/or isomers.
This variation does not permit us to gain definitive conclusions
concerning the relative solution stabilities of the generated thiol
intermediates. However, several observations appear valid. The
dominant reaction product for 5, 12-14, and/or isomers in either
basic or near-neutral methanolic solutions was dimerization. No
evidence was obtained for mitomycin ring activation. We
observed neither mitosene production (loss of MeOH at C(9)
and C(9a)) nor ring-opening of the aziridine unit. The ease with
which thiols 5, 12-14, and/or cyclized isomers underwent
dimerization depended on the size and the substitution pattern
of the C(7) amino substituent. We found that when the linker
was C2, dimerization to the symmetrical disulfides (5 f 56,
12 f 69) proceeded more rapidly at 0 °C than when the bridge
unit was C3 (14 f 74). For example, under basic conditions
(NaOMe-MeOH), the approximate half-life for thiol (thiolate)
14 was 4.5 h, but for 12, dimerization occurred within minutes
at 0 °C. We also observed that the gem-dimethyl unit in 13
substantially decreased the ease in which this thiol dimerized
to the symmetrical disulfide 73.
Efforts to obtain structural information concerning thiols 5
and 12-14 (or isomers) by NMR spectroscopy were only
partially successful. Use of a concentrated sample of 20 and
D,L-DTT led to a mixture of disulfides 56-58 along with
suspected thiol 5. Similarly, when we attempted to prepare a
concentrated sample (NaOCD3-CD3OD) of 12 from thiol ester
26 suitable for NMR spectroscopy, HPLC analysis indicated
that the predominant species in solution was 69 (data not
shown). We found that disulfide 22 was not a suitable candidate
to generate a NMR sample of the free thiol 13 since D,L-DTT-
mediated disulfide cleavage occurred slowly and required our
using a large excess of D,L-DTT. We were successful in
1
obtaining a H NMR spectrum for thiol 14. Preparation of a
concentrated solution of thiol ester 27 in NaOCD3-CD3OD
provided a spectrum (Figure 1a) consistent with free thiol 14.
Key to our assignment was the detection of a multiplet at δ
1.85 and a triplet at δ 2.56 for the C(2′) and C(3′) methylene
protons, respectively. It is significant that we observed no signals
associated with C(6)-C(8) cyclic adducts. Attempts to obtain
the corresponding 13C NMR spectrum were unsuccessful since
14 converted to disulfide 74 during the NMR acquisition time
(NMR and HPLC analysis).
Our finding that thiols 5 and 12 existed predominantly as
cyclized adducts while 14 existed as the uncyclized, free thiol
species can be rationalized on the basis of the linker size.
Cyclization of 5 (12) at C(7) led to a five-membered-ring adduct
while the corresponding cyclization for 14 gave a six-membered-
ring adduct. A comparable difference in ring size exists for 5
(12) and 14 if cyclization proceeds at either C(6) or C(8). This
difference provides an entropic advantage for cyclization for
thiols 5 and 12 compared with 14. Similar entropic factors likely
contributed to the facile formation of 82 and 83.38
The absence of mitosene adducts in the reaction profiles was
striking. A similar finding was observed by Kono and co-
workers.40 Mitomycin (porfiromycin) reduction to the hydro-
quinone perturbs the delocalization of the N(4a) electrons with
the C(5a)-C(8a)-C(8)-O(8) R,â-unsaturated carbonyl bond
system and promotes expulsion of the C(9a) methoxy group
and mitosene production leading to aziridine ring-opened
adducts. This reaction proceeds rapidly at neutral “pH” in
water41 and alcoholic solvents at 0 °C.42 Similarly, we have
demonstrated that conversion of the C(8) quinone ring in C(7)
amino-substituted mitomycins to the corresponding C(8) (i.e.,
85) imine accelerates mitosene formation and aziridine ring-
Generation and Stability of Thiols: Insight into Structure.
Our inventory of mitomycins (porfiromycins) consisted of 10
compounds (20-27, 71, 72) the substitution patterns (CH2CH2,
CH2C(CH3)2) and linker sizes (C2, C3) of which varied.
Together these compounds permit us to determine whether the
linker affects the chemical reactivity of mixed mitomycin
(porfiromycin) disulfides and porfiromycin thiol esters and the
stability of the generated thiol. In KW-2149 (3) and BMS-
181174 (4), an unsubstituted C2 (CH2CH2) bridge separates the
terminal disulfide unit and the mitomycin C(7) amino substitu-
ent. We found that the ease with which the mixed mitomycin
disulfides and porfiromycin thiol esters underwent thiol forma-
tion was independent of the length of the linker. Treatment of
20, 21, 24, 25, 71, and 72 with DTT led to the selective cleavage
opening processes presumably by diminishing N(4a) delocal-
ization with the adjacent R,â-unsaturated imine system.12,43 For
(40) Kono, M.; Saitoh, Y.; Kasai, M.; Shirahata, K.; Morimoto, M.; Ashizawa,
T. J. Antibiot. 1993, 46, 1428-1438.
(41) Boruah, R. C.; Skibo, E. B. J. Org. Chem. 1995, 60, 2232-2243.
(42) (a) Kohn, H.; Li, V.-S.; Tang, M.-s. J. Am. Chem. Soc. 1992, 114, 5501-
5509. (b) Li, V.-S.; Choi, D.; Tang, M.-s.; Kohn, H. Biochemistry 1995,
34, 7120-7126.
(43) Wang, S.; Kohn, H. J. Org. Chem. 1997, 62, 5404-5412.
9
J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002 4675

A R T I C L E S
Na et al.
Scheme 8. Proposed Pathway for KW-2149 (3) and BMS-181174
(4) Function
example, the half-life for conversion of 85 to the ring-opened
C(1)-substituted methoxymitosene adducts in buffered metha-
nolic solutions (Tris-HCl, “pH” 7.4, 25 °C) was 40 h while no
reaction was observed for 86 after 10 days.12 These findings
indicate that the structure of the cyclized mitomycin (porfiro-
mycin) thiol (53-55, 66-68) has a significant impact on the
ease with which the mitomycin (porfiromycin) is converted to
the mitosene. For 55 (68), C(8) cyclization leads to a hybridiza-
tion change at C(8) from sp2 to sp3 and the loss of delocalization
of the N(4a) electron pair with the adjacent R,â-unsaturated
carbonyl system in the starting material. For 53 (66) and 54
(67), cyclization does not significantly perturb N(4a) delocal-
ization. These observations predict that 55 (68) would rapidly
be converted to mitosene adducts at neutral “pH” values under
our reaction conditions, but 53 (66) and 54 (67) would not. To
date, only C(7)37,38 and C(8)39 cyclized mitomycin (porfiromy-
cin) intermediates have been characterized. No C(6) adducts
have been reported, a fact that may reflect the effect of the C(6)
methyl group on substitution reactions at this site. We concluded,
therefore, that the HPLC peaks observed for thiols 5 and 12
likely corresponded to 54 and 67 and the free thiol.
Pathway of Thiol-Mediated Cleavage Reactions of Mito-
mycin (Porfiromycin) C(7)-Substituted Disulfides. A major
preparative route employed in our studies for mitomycin (porfi-
romycin) C(7)-substituted thiols was thiol (e.g., DTT)-mediated
cleavage of unsymmetrical mitomycin (porfiromycin) disulfides
20-25, 71, and 72. Our studies documented that these reactions
proceeded along a thiol-disulfide exchange pathway.19 First,
we detected (HPLC) intermediates 57 and 58 when either 20
or 21 was treated with D,L-DTT. Second, substitution of GSH
for D,L-DTT gave 65 in high yields (>85%). Compound 65,
unlike 57 and 58, cannot undergo intramolecular disulfide ex-
change to give 5, thus explaining the stability of this adduct
and the yield of this transformation. Third, HPLC analysis of
the D,L-DTT- and GSH-mediated cleavage reactions of 20 and
21 showed that 57, 58, and 65 appeared rapidly with time (data
not shown), suggesting that these compounds represented initial
reaction products. Finally, we isolated the DTT-bound adducts
75 and 78 after treatment of 25 with L-DTT followed by 30.
Compounds 75 and 78 likely originated from L-DTT adduct
77.
for unsymmetrical mitomycin C(7) disulfides,44 symmetrical
mitomycin C(7) disulfides,40,44 and mitomycin C(7) thiol esters29
containing C2 bridging elements?
Reflections on the Mechanism of Action of KW-2149 and
BMS-181174. Our results allowed conclusions concerning the
proposed thiol 5 generation step for KW-2149 (3) and BMS-
181174 (4) (Scheme 2, route B) and the reactivity of thiol 5.
We observed that thiol generation readily occurred for 20-22,
24, 25, 71, and 72 under mild conditions, indicating that this is
a plausible event for 3 and 4. We further showed that thiol (thiol
intermediates) stability increased with alkyl substitution of the
mitomycin C(7) thiol linker (13 vs 5) and with the length of
the linker (14 vs 5 and 12). When the linker was held to an
ethylene (C2) unit (5, 12), isomeric intermediates predominated
and spectroscopic analysis suggested the formation of C(7)
spiro-adducts (54, 67). Finally, thiol 5 and 12-14 generation
in MeOH did not lead to mitomycin ring activation and mitosene
production, indicating that simple thiol formation, in and of
itself, is not sufficient to initiate either 3 or 4 activation.11-13
The absence of mitomycin ring activation upon thiol genera-
tion 5 asks whether other factors are responsible for the different
pharmacological properties observed for 3 and 4 compared with
1. Recall that both 3 and 4 exhibit significant activity in 1-resis-
tant tumor cell lines.8-10 Furthermore, cells that are resistant to
1 and express low levels of the reductase DT-diaphorase45 are
not cross-resistant to 3.8b,e Despite these pharmacological differ-
ences, DNA adduction studies indicate that the sequence speci-
ficity for DT-diaphorase-activated 1 and glutathione-treated 3
are nearly identical,16 suggesting that both transformations pro-
ceeded by comparable molecular mechanisms. How can these
findings be explained? Several possibilities exist. We have dia-
grammed one in Scheme 8, which focuses on the unique chemi-
(44) Shirahata, K.; Kasai, M.; Kono, M.; Morimoto, M.; Ashizawa, T.; Saito,
Y. Disulfide Compounds and Antineoplastic Agents. Japan Pat. Appl. JP86/
0085, February 21, 1986.
(45) Prakash, A. S.; Beall, H.; Ross, D.; Gibson, N. W. Biochemistry 1993, 32,
5518-5525.
What are the causative factors for the improved preclinical
pharmacological properties of KW-2149 and BMS-181174 over
conventional mitomycins, and why is maximal activity achieved
9
4676 J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002

7-N-(Mercaptoalkyl)mitomycins
A R T I C L E S
cal properties of the disulfide unit in KW-2149 and BMS-181174
and the properties and reactivities of the C(7) spiro-adduct 54.
We divide our discussion into the extracellular and intracel-
lular processes that may occur for 3 and 4, which are represented
by 87 in Scheme 8. In the first stage (Scheme 8A), the mito-
mycin C(7) amino-substituted disulfide 87 undergoes thiol-
disulfide exchange with biological thiol B-SH (e.g., serum
albumin, glutathione) in the plasma to generate 88.14-16 Support
for this step comes from the metabolic studies of Kobayashi
and co-workers, who showed that iv administration of 3 to mice
led to the rapid production of 56, 92, and 93 in the plasma and
of the drug by reductases bound to cellular membranes and
present in the cytosol, thereby preventing the drug from reaching
its nuclear DNA target.
There is support for key features in Scheme 8B. First,
generation of 5 (12) under a variety of conditions provided
HPLC profiles consistent with the formation of isomeric adducts.
Second, both the UV-visible properties of the cyclic adducts
and the absence of aziridine ring-opened mitosene production
suggested that 5 (12) cyclization occurred predominantly at C(7).
Third, we found that 5 (12) and/or isomers underwent dimer-
ization more rapidly than either 13 or 14. This finding suggested
that dimerization may have incurred, in part, by the heterolytic
pathway outlined in Scheme 8B (54a f 91) where C-SH is 5.
Fourth, Tomasz and co-workers demonstrated addition of large
excesses of GSH to both 3 and 4 led to mitosene production
(UV analysis) and DNA interstrand cross-link formation.11 Fifth,
a heterolytic intramolecular thiol-initiated cleavage pathway
comparable to 54a f 91 has been postulated for the facile
reduction of mitomcyin A (2) by D,L-DTT.48-50
Conclusions
that the plasma half-lives for 56, 92, and 93 were 12, 13, and
40 min, respectively.14 Further evidence for this process comes
from Masters, McAdam, and Hartley, who showed that 3 is
activated extracellularly by serum.15,16 They further suggested
that an unidentified species that passively enters the cells more
rapidly than 3 and can efficiently modify DNA is produced upon
activation. The recent investigation by Yasuzawa and Tomer
documented that a human serum albumin (HSA)-KW-2149
adduct (93)46 is produced upon incubating HSA with 3. Finally,
we observed that 20, 21, 24, 25, 71, and 72 all underwent rapid
thiol-mediated disulfide exchange under mild conditions. For-
mation of small mixed disulfides 88 by extracellular processes
then permits passage across the cell membrane.
In the second stage of our hypothesis (Scheme 8B), passage
of 88 into the cell permits drug activation by one of two
pathways. The first is enzymatic reduction. We envision that
88 reacts, in part, with intracellular thiols C-SH (e.g., GSH)
by a thiol-disulfide exchange process to give 89 and 5. All
three mitomycins (5, 88, 89) are capable of undergoing quinone
reduction and ring activation by a route comparable to that
proposed for 1 (Scheme 1). The second activation pathway is
unique for mitomycin C(7)-substituted thiols with C2 (ethylene)
linkers. We have proposed that 5 largely exists as the C(7) spiro
isomer 54a (drawn as the enol tautomer). Reaction of 54a with
an intracellular thiol (C-SH) at sulfur leads to C(7)-S
heterolytic cleavage and aromatization of the quinone ring,
providing reduced mitomycin 91.47 Leucomitomycin 91 is in
the same reduction state as 90 and 6 (Scheme 1) and is expected
to undergo ring activation. This pathway (54a + C-SH f 91)
is an unconventional mitomycin activation pathway and provides
a possible explanation why cell lines deficient in DT-diaphorase
may be susceptible to 3.
A series of mitomycin C(7)-substituted thiols were prepared
that examine the effect of structural composition of the unit
that bridges the thiol with the mitomycin core on thiol reactivity.
We learned that the linker affects thiol structure and thiol
stability. Special emphasis was given to mitomycins where the
linker is two carbons in light of the pharmacological properties
reported for KW-2149 (3) and BMS-181174 (4). Our studies
indicated that the thiol 5 generated from these drugs exist as
the C(7) cyclic species. C(7)-Spiromitomycins, such as 54, may
provide special advantages for drug function that include
alternative pathways for ring activation and the generation of
intermediates not prone to unwanted activation by reductases.
Finally, previous investigations indicate that the pharmacological
properties of 5-conjugates (e.g., 88) contribute to drug efficacy.
Acknowledgment. This paper is dedicated to Professors Roy
A. Olofson (The Pennsylvania State University) and Ronald
Breslow (Columbia University), mentors and friends, and the
two Universities they have continuously served. This study was
supported by grants from the National Institutes of Health
(CA29756) and the Robert A. Welch Foundation (E-607). We
thank Drs. Masaji Kasai and Hitoshi Arai (Kyowa Hakko Kogyo
Co., Ltd., Osaka, Japan) for generously supplying mitomycin
A.
Supporting Information Available: Full experimental details
for the synthesis and characterization of 20-27, 32-35, 37,
38, 40-46, 51, 52, 56, 61, 62, 65, 69-76, and 78). Full
experimental details and procedures for the generation and
trapping of thiol intermediates 5, and 12-14. HPLC chromato-
grams of mitomycin (porfiromycin) thiol and/or isomers and
their trapped products (PDF). This material is available free of
charge via the Internet at http://pubs.acs.org.
The proclivity of mitomycin C(7) disulfides with a C2 bridge
to undergo cyclization may provide another potential pharma-
cological advantage for 3 and 4. Formation of 54 upon drug
activation may prevent unwanted consumption and destruction
JA0124313
(48) Paz, M. M.; Tomasz, M. Org. Lett. 2001, 3, 2789-2792.
(49) Paz, M. M.; Das, A.; Palom, Y.; He, Q.-Y.; Tomasz, M. J. Med. Chem.
2001, 44, 2834-2842.
(50) A homolytic route has been proposed for the corresponding bimolecular
reaction where the rate-limiting step is an intramolecular redox reaction
involving the (homolytic) dissociation of the sulfur-bound quinone species
to give a reduced mitomycin A species.49
(46) Yasuzawa, T.; Tomer, K. B. Bioconjugate Chem. 1997, 8, 391-399.
(47) Intracellular thiols include glutathione, cysteine-containing proteins, and
5. Reaction of 54a with 5 provides a dimeric mitomycin adduct. We have
recently shown that dimeric mitomycins can cross-link DNA: Na, Y.; Li,
V.-S.; Nakanishi, Y.; Bastow, K. F.; Kohn, H. J. Med. Chem. 2001, 44,
3453-3462.
9
J. AM. CHEM. SOC. VOL. 124, NO. 17, 2002 4677