Synthesis of (±)-phloeodictine A1

Article abstract of DOI:10.1021/ol034042e

(Matrix presented) The antitumor antibiotic phloeodictine A1 (2) has been synthesized by a convergent seven-step route in 8% overall yield. The key step was the Eguchi aza-Wittig reaction of 6 to give 13 followed by a retro Diels-Alder reaction to liberat

Full text of DOI:10.1021/ol034042e

ORGANIC
LETTERS
2003
Vol. 5, No. 5
765-768
Synthesis of (±)-Phloeodictine A1
Bobbianna J. Neubert and Barry B. Snider*
Department of Chemistry MS 015, Brandeis UniVersity,
Waltham, Massachusetts 02454-9110
snider@brandeis.edu
Received January 9, 2003
ABSTRACT
The antitumor antibiotic phloeodictine A1 (2) has been synthesized by a convergent seven-step route in 8% overall yield. The key step was
the Eguchi aza-Wittig reaction of 6 to give 13 followed by a retro Diels−Alder reaction to liberate 5. Addition of 11-dodecenylmagnesium
bromide to 5 to give 4b, alkylation with 18b, and deprotection completed the first synthesis of 2.
Phloeodictines A (1), A1 (2), and A2 (3) were isolated from
the New Caledonian sponge Phloeodictyon sp. by Pa¨ıs and
Scheme 1
co-workers.1 They exhibit in vitro antibacterial activity
against Gram-positive and Gram-negative bacteria and are
moderately cytotoxic against KB cells. They possess a
6-hydroxy-1,2,3,4-tetrahydropyrrolo[1,2-a]pyrimidinium skel-
eton that presents a considerable synthetic challenge. We
envisaged that they could be prepared from 4 by alkylation
with the appropriate length guanidino alkyl side chain
(Scheme 1). Addition of the appropriate length side chain
to amide 5 should provide 4. We envisioned that 5 might be
available from azido maleimide 7 by an Eguchi aza-Wittig
reaction.2 However, we were concerned that the electron-
poor double bond of 7 would not be compatible with
introduction of the azide or the phosphine used for the aza-
Wittig reaction,3 so we developed an alternate route using
azide 6 in which the double bond of 7 was protected as the
Diels-Alder adduct with furan.
Reaction of the furan-maleic anhydride Diels-Alder
quantitatively to mesylate 10 with Et3N and MsCl (Scheme
adduct 8 with 3-aminopropanol in MeOH at 56 °C for 3
2). Reaction of 10 with NaN3 in DMF for 14 h at 25 °C
days provided 70% of imide 9,4 which was converted
afforded azide 6. To our surprise, azide 6 rapidly polymerized
on concentration, presumably by cycloaddition of the azide
(1) (a) Kourany-Lefoll, E.; Pa¨ıs, M.; Se´venet, T.; Guittet, E.; Montagnac,
A.; Fontaine, C.; Gue´nard, D.; Adeline, M. T. J. Org. Chem. 1992, 57,
3832-3835. (b) Kourany-Lefoll, E.; Lapre´vote, O.; Se´venet, T.; Montagnac,
A.; Pa¨ıs, M. Tetrahedron 1994, 50, 3415-3426.
(2) Eguchi, S.; Takeuchi, H. J. Chem. Soc., Chem. Commun. 1989, 602-
603.
(3) In fact, reaction of the mesylate precursor to 7 with azide destroyed
the double bond. For addition of azide to maleimides, see: Myers, J. K.;
Jacobsen, E. N. J. Am. Chem. Soc. 1999, 121, 8959-8960.
(4) Zhou, Z.-h.; Chen, R.-y. Synth. Commun. 2000, 30, 3527-3533.
10.1021/ol034042e CCC: $25.00 © 2003 American Chemical Society
Published on Web 02/07/2003

Although we were able to use 5 successfully for the
synthesis of phloeodictine A1, it would have been desirable
to keep the double bond protected as the Diels-Alder adduct
until later in the synthesis. The aza-Wittig reaction requires
toluene at reflux, so it was not possible to prepare 13 without
also converting most of it to 5. We therefore investigated
more stable Diels-Alder adducts. The anthracene-maleic
anhydride adduct was elaborated to the aza-Wittig product
analogous to 13. However, it now was impossible to effect
the retro Diels-Alder reaction. The aza-Wittig product
distilled at 300 °C under reduced pressure without loss of
anthracene.
Scheme 2
Addition of Grignard reagents to 5 proceeded poorly. The
best results were obtained by adding the appropriate Grignard
reagent10 to a 1:1 mixture of 5 and CeCl3 in THF at 0 °C
(Scheme 3). The reaction was quenched with aqueous
Scheme 3
and strained double bond.5 The DMF solution of azide 6,
was therefore diluted with toluene, washed with water to
remove DMF, dried, and immediately subjected to the aza-
Wittig reaction.
Reaction of 6 with Bu3P or (n-C8H17)3P6 in toluene at
reflux for 4 h gave bicyclic amidine 117 in 15-25% yield
from mesylate 10, rather than the desired amidine 5. A
plausible mechanism involves loss of nitrogen from azide 6
at 25 °C to give ylide 12, which should undergo an aza-
Wittig reaction on heating to provide 13. A thermal retro
Diels-Alder reaction will give 5, which might undergo a
base-catalyzed isomerization to give 11. We thought that this
isomerization might be prevented by the use of less basic
Ph3P.8
Reaction of 6 with Ph3P in toluene at reflux for 4 h gave
the desired amidine 5, which cannot be easily separated from
Ph3PO. Fortunately, polystyrene-supported Ph3P9 worked
equally well. Unreacted phosphine and phosphine oxide
byproducts were removed by filtration, giving pure 5 in
43% yield from 10 after Florisil chromatography. Heating
the reaction for only 2-3 h provided 8-13% of 13 and
20-25% of 5, indicating that the retro Diels-Alder reaction
occurs, at least primarily, after the aza-Wittig reaction.
Reaction of 5 with either Bu3P or Bu3PO in toluene at reflux
for 4 h gave only traces of 11, while heating 5 with DBU
gave 5-10% of 11, indicating that the mechanism for the
formation of 11 from 10 and trialkylphosphines is complex.
NH4Cl solution, and the mixture was extracted into CH2Cl2
and concentrated. The residue was triturated with pentane
to give 40-45% of 80-90% pure 14 as a brownish unstable
solid. Washing a CH2Cl2 solution of 14 with 1 M NaOH
solution afforded 4 as a brown oil that was used immediately
for the next step.
The electron-deficient ring double bond of 4 is too reactive
to permit the elaboration of the guanidine after addition of
the side chain.11 We therefore chose a convergent route using
iodide 18 containing a protected guanidine on the other end
of the chain. Reaction of 15 with the appropriate ω-amino-
1-alkanol in THF at 50 °C for 2 h gave 16 quantitatively.12
Mesylation and displacement with iodide afforded 18 as
shown in Scheme 3.
(10) Unsaturated Grignard reagents were prepared in THF from Mg and
the known ω-bromo-1-alkenes: Watson, M. D.; Wagener, K. B. Macro-
molecules 2000, 33, 5411-5417.
(11) Alkylation of 4c with 4-chlorobutyl triflate proceeded cleanly. Azide
added to the ring double bond during attempted SN2 reaction with the
chlorobutyl side chain.
(5) For related polymerizations, see: (a) Johnson, K. E.; Lovinger, J.
A.; Parker, C. O.; Baldwin, M. G. J. Polym. Sci., Polym. Lett. Ed. 1966, 4,
977-979. (b) Gilliams, Y.; Smets, G. Makromol. Chem. 1968, 117, 1-11.
(6) Separation of 11 from (n-C8H17)3PO was easier than from the more
polar Bu3PO.
(7) For related compounds, see: (a) Jokic´, M.; Sˇkaric´, V. J. Chem. Soc.,
Perkin Trans. 1 1989, 757-763. (b) Spiessens, L. I.; Anteunis, M. J. O.
Bull. Soc. Chim. Belg. 1984, 93, 191-203.
(8) Issleib, V. K.; Bruchlos, H. Z. Anorg. Allg. Chem. 1962, 316, 1-11.
(9) Bernard, M.; Ford, W. T. J. Org. Chem. 1983, 48, 326-332.
(12) (a) Botta, M.; Corelli, F.; Maga, G.; Manetti, F.; Renzulli, M.;
Spadari, S. Tetrahedron 2001, 57, 8357-8367. (b) Ishiwata, T.; Hino, T.;
Koshino, H.; Hashimoto, Y.; Nakata, T.; Nagasawa, K. Org. Lett. 2002, 4,
2921-2924. (c) Slassi, A.; Sumanas, R. PCT Int. Appl. Patent WO 9,514,-
027, 1995; Chem. Abstr. 1996, 124, 9337b.
766
Org. Lett., Vol. 5, No. 5, 2003

Nucleophilic substitution of a neutral substrate (18) with
a neutral nucleophile (4) should proceed best in a polar
solvent, since the transition state is much more polar than
the starting materials. Reaction of 4c with 18b in acetone
for 2 h at 25 °C provided 92% of acetonide 20 (Scheme 4).
to pH 2 as described in the isolation paper afforded
1
phloeodictine A1 (2) in 26% overall yield from 5. The H
and 13C NMR, HSQC, and FAB mass spectral data are
identical to those reported.1b
We then turned our attention to the preparation of
phloeodictine A (1) from 4a and 18a. Unfortunately, reaction
in DMSO-d6 as described above for the preparation of 21
yielded only 10% of the desired product 22 and 90% of
protonated amidinium salt 14a (Scheme 6).
Scheme 4
Scheme 6
The structure was confirmed by HSQC and HMBC experi-
ments. H7, H8a, and H8b absorb at δ 4.67 (dd, 1, J ) 5.2, <1
Hz), 3.85 (dd, 1, J ) 17.5, 5.2 Hz), 2.96 (dd, 1, J ) 17.5,
<1 Hz), respectively. The initial alkylation reaction occurred
as expected. The hydroxy group reacted with acetone to form
hemiacetal 19, which underwent an intramolecular conjugate
addition to give 20. Amidine 4 is insoluble in CH3CN, so
we then investigated the use of DMSO as a solvent. This
was particularly appealing since the progress of the reaction
could be monitored by the shifts of the absorptions of the
alkene hydrogens in DMSO-d6.
Amidine 4 is not only a nucleophile but also a strong base
that can reversibly deprotonate the protected guanidine of
18 to give anion 23. We thought that 23a should undergo
an intramolecular SN2 reaction to give five-membered ring
pyrrolidine 24a13 much more rapidly than 23b does to give
six-membered ring piperidine 24b14 (Scheme 7). This was
Reaction of 4b [δ 6.51 (d, 1, J ) 6.1) and 6.00 (d, 1, J )
6.1)] and 18b in DMSO-d6 for 1 day afforded a 2:1 mixture
of the desired alkylation product 21 [δ 7.39 (d, 1, J ) 6.1)
and 7.13 (d, 1, J ) 6.1)] and protonated amidinium salt 14b
[δ 7.30 (d, 1, J ) 6.1) and 6.52 (d, 1, J ) 6.1)] (Scheme 5).
Scheme 7
Scheme 5
confirmed by examination of the cyclization of 18a and 18b
with 2 equiv of collidine or (i-Pr)2EtN in DMSO-d6. After 1
day, 18a cyclized to give 80-90% of 24a, while 18b
afforded only 10-15% of 24b.
We then decided to prepare iodobutyl guanidines with the
internal nitrogen protected so that a pyrrolidine could not
be formed by deprotonation and intramolecular SN2 reaction.
Alkylation of N,N′,N′′-tri-Boc-guanidine (25) with 4-bromo-
1-butanol, DEAD, and Ph3P by the procedure of Goodman15
The mixture was diluted with water and extracted with
CH2Cl2, which was concentrated. Even though 21 is a salt,
it is more soluble in CH2Cl2 than in water, while 14b remains
in the water layer. Purification at this point was difficult, so
the mixture was deprotected by stirring in 1:1 TFA/CH2Cl2
for 2 h. Concentration and reverse-phase flash chromatog-
raphy using MeOH, 0.2 M aqueous NaCl, and THF adjusted
(13) Feichtinger, K.; Zapf, C.; Sings, H. L.; Goodman, M. J. Org. Chem.
1998, 63, 3804-3805.
(14) (a) Yong, Y. F.; Kowalski, J. A.; Lipton, M. A. J. Org. Chem. 1997,
62, 1540-1542. (b) Guo, Z.-X.; Cammidge, A. N.; Horwell, D. C. Synth.
Commun. 2000, 30, 2933-2943.
(15) Feichtinger, K.; Sings, H. L.; Baker, T. J.; Matthews, K.; Goodman,
M. J. Org. Chem. 1998, 63, 8432-8439.
Org. Lett., Vol. 5, No. 5, 2003
767

afforded 74% of bromide 26a, which was converted to 71%
of iodide 26b with NaI in acetone (Scheme 8). Unfortunately,
Scheme 9
Scheme 8
4, the Cbz groups cannot be removed in the presence of the
side chain double bond, and the bis Boc protected guanidine
analogous to 29a cannot be prepared by the same procedure.
In conclusion, we have developed a convergent seven-
step route to the antitumor antibiotic phloeodictine A1 (2)
that proceeds in 8% overall yield. We have developed a four-
step synthesis of the novel and synthetically versatile bicyclic
amidine 5 using the furan Diels-Alder adduct as a protecting
group for the double bond. The key step is the Eguchi aza-
Wittig reaction of 6 to give 13 followed by a retro Diels-
Alder reaction to liberate 5. Use of polystyrene-supported
Ph3P prevents isomerization of 5 and facilitates purification
of the polar product. Addition of 11-dodecenylmagnesium
bromide, alkylation with 18b, and deprotection completes
an efficient synthesis of 2.
attempted alkylation of 4a with 26b in DMSO-d6 gave only
protonated amidinium salt 14a. Deprotonation followed by
cyclization was still the major reaction. Treatment of 26b
with 2 equiv of collidine in DMSO-d6 provided 50% of 2816
after 1 day and 92% after 7 days. Even though the cyclization
of 26b was slower than the cyclization of 18a, the alkylation
was no more successful. The third Boc group will make the
remaining NH proton more acidic, so that the equilibrium
may be shifted to protonated amidinium salt 14a and anion
27.
We then prepared bis Cbz-protected guanidines 29a and
29b from the known alcohols.17 Alkylation of DBN (30) with
29b in DMSO-d6 provided 31b quantitatively, while alky-
lation with 29a affords only 40% of 31a and 15 and 23% of
cyclization products 32 and 33, respectively (Scheme 9). We
did not investigate this approach to 1 further because the
alkylation proceeds much more cleanly with DBN than with
Acknowledgment. We thank the National Institutes of
Health (GM-50151) for financial support. We thank Mr.
Zheng Chen for carrying out preliminary experiments.
(16) For similar cyclizations, see: (a) Ueda, T.; Oh, R.; Nagai, S.-i.;
Sakakibara, J. J. Heterocycl. Chem. 1998, 35, 135-139. (b) Mesza´rosova´,
K.; Holy´, A.; Masoj´ıdkova´, M. Collect. Czech. Chem. Commun. 2000, 65,
1109-1125. (c) Le Merrer, Y.; Gauzy, L.; Gravier-Pelletier, C.; Depezay,
J.-C. Bioorg. Med. Chem. 2000, 8, 307-320.
Supporting Information Available: Experimental pro-
cedures and copies of spectra. This material is available free
of charge via the Internet at http://pubs.acs.org.
(17) Snider, B. B.; Shi, Z. J. Org. Chem. 1993, 58, 3828-3839.
OL034042E
768
Org. Lett., Vol. 5, No. 5, 2003