Intramolecular guanidine epoxide ring opening reactions

Article abstract of DOI:10.1016/S0040-4039(03)00519-7

The synthesis of a range of cyclic guanidines via intramolecular ring opening of epoxides or iodocyclisation is reported. A preliminary investigation of the glycosidase inhibitory activity of these substances is also discussed.

Full text of DOI:10.1016/S0040-4039(03)00519-7

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 44 (2003) 3075–3080
Intramolecular guanidine epoxide ring opening reactions
Mark Dennis,a Louise M. Hall,a Patrick J. Murphy,a,* Andrew J. Thornhill,a Robert Nash,b
Ana L. Winters,b Michael B. Hursthouse,c Mark E. Lightc and Peter Hortonc
aDepartment of Chemistry, University of Wales, Bangor, Gwynedd LL57 2UW, UK
bInstitute of Grassland and Environmental Research, Plas Gogerddan, Aberystwyth SY23 3EB, UK
cEPSRC National Crystallography Service, Department of Chemistry, University of Southampton, Highfield,
Southampton SO17 1BJ, UK
Received 20 December 2002; revised 21 February 2003; accepted 24 February 2003
Abstract—The synthesis of a range of cyclic guanidines via intramolecular ring opening of epoxides or iodocyclisation is reported.
A preliminary investigation of the glycosidase inhibitory activity of these substances is also discussed. © 2003 Published by
Elsevier Science Ltd.
As part of a project directed towards the synthesis of
the marine natural products ptilomycalin A and the
batzelladines we have reported the addition reaction of
guanidine to bis-a,b-unsaturated ketones 1. The inter-
mediates in this process can be directly converted, by
stereoselective reduction, to tricyclic structures 2 found
in the batzelladine alkaloids or alternately, by deprotec-
tion and spirocyclisation of pendent hydroxyl groups,
to the pentacyclic guanidine 3 structurally similar to
ptilomycalin A1 (Scheme 1). We were interested in
extending this methodology and as part of a project
directed towards the synthesis of the marine hepato-
toxin cylindrospermopsin 42 (Fig. 1) we wished to
investigate the reaction of guanidine with epoxides.
with guanidine leading to the two isomeric diols 6a and
6b.4 Le Merrer et al. then reported5 the reaction of
carbohydrate derived epoxides, for example 7, with
guanidine leading to the 9-membered cyclic guanidine 8
and also the formation of the furan 10, initiated by ring
opening of the epoxide 9 with guanidine. More recently
Taylor and co-workers reported that the cyclic
guanidine 11 readily reacts with epoxides to give ring
opened products, for example 12 (Scheme 2).6
Surprisingly, few examples of epoxide ring opening
processes utilising guanidines3 have been reported. The
first of these was the reaction of cis-benzene trioxide 5
Figure 1.
Scheme 1. Reagents and conditions: (a) (i) guanidine, DMF, 0°C, 5 h, (ii) 3:1:3 DMF, H2O, MeOH, then NaBH4, 16 h, (iii) HCl
(aq.), (iv) NaBF4 (satd aq.); (b) (i) guanidine, DMF, 3 h, (ii) MeOH, HCl, 0°C–rt, 24 h, (iii) NaBF4 (satd aq.).
Keywords: guanidines; epoxides; iodocyclisations; b-galactosidase inhibition.
* Corresponding author.
0040-4039/03/$ - see front matter © 2003 Published by Elsevier Science Ltd.
doi:10.1016/S0040-4039(03)00519-7

3076
M. Dennis et al. / Tetrahedron Letters 44 (2003) 3075–3080
Scheme 2. Reagents and conditions: (a) guanidine, t-BuOH, D, 4 h; (b) guanidine, EtOH, D, 1 h; (c) guanidine, t-BuOH, 60°C,
130 h; (d) EtOH, D.
We were interested in the intramolecular ring opening
of epoxides which would lead to the formation of 5-
and 6-membered guanidine heterocycles, with the ulti-
mate aim of applying this methodology to the synthesis
of the tricyclic ring system found in cylindrosper-
mopsin. We initially used the simple epoxides 13–167
and treated them with guanidine in t-BuOH at room
these reactions, purification of the product proved
problematic and repeated chromatographic separation
was necessary.
On reaction of epoxide 15 under these conditions a
different outcome was observed, in that the 7-endo
product 19 was formed in high yield, together with the
double addition product identified as 20. Analysis of
the NMR spectrum of the crude product indicated a
70:30 ratio of 19 to 20, from which 19 could be isolated
typically in 40–50% yield. Very little polymeric material
was found indicating that the primary epoxide position
is very susceptible to intramolecular ring opening. We
were able to increase the relative amount of 20 formed,
by varying the stoichiometry of the reaction. Where 2
equiv. of the epoxide 15 were used (conditions B) the
two products, 19 and 20, were isolated in a 35:65 ratio.
This could be increased to some extent if 3 equiv. were
employed (conditions C); however, considerable decom-
position and lower overall yields were observed in these
reactions.
temperature for 24
h to effect N-alkylation of
guanidine (which was presumed to be faster than the
epoxide ring opening process). At this point potassium
t-butoxide was added to regenerate the free guanidine
from its salt, following which the reaction was heated at
60°C for a further 24–48 h to effect cyclisation. The
results of these experiments are given in Table 1.
The reaction of epibromohydrin 13 under these condi-
tions, led to the formation of the heterocycle 178 in
modest yield together with what appeared to be a trace
amount of material arising for the reaction of guanidine
with two equivalents of 13. In addition to this a consid-
erable amount of what appeared to be a polymeric
guanidine containing material was also formed. We
attempted to increase the amount of the double addi-
tion product formed by varying the stoichiometry of
the reaction (conditions B and C), however this led only
to the formation of further polymeric material. Only
the 5-exo product was isolated from these reactions and
no evidence for the formation of a 6-endo product was
found. The reaction of epoxide 14 under identical con-
ditions was then investigated and it was found to give
essentially the same result. Typical yields of 30% were
obtained and the 5-exo product 18 was the major
product, together with trace amounts of a double addi-
tion product which was inseparable from the consider-
able quantity of polymeric material formed. In both
Finally we investigated the substrates 16 (X=Hal) and
were disappointed to find that the major products
formed in these reactions were alkenes produced by
dehydrohalogenation of the starting materials, proba-
bly arising from a base catalysed process. We attempted
to prevent this elimination by using the mesylate 16
(X=OMs) and found that this underwent cyclisation to
give the 6-exo product 21 in 58% yield together with
some alkene and very little double addition product or
polymeric material.
Our overall conclusions from these reactions are that in
the case of substrates 13 and 14 the 5-exo cyclisation is

M. Dennis et al. / Tetrahedron Letters 44 (2003) 3075–3080
3077
Table 1.
preferred over the 6-endo process, however the process
is complicated by the formation of high molecular
weight polymeric by-products. The situation is more
complex for substrates 15 and 16 where the 7-endo
product is favoured for substrate 15, but the 6-exo is
preferred in the case of 16. One possible explanation for
this difference is that the 7-endo cyclisation to give 19 is
a sterically more favourable ring opening at a primary
position and is thus preferred over the 6-exo
cyclisation.
found in substrate 17 (Boc2O, NaOH (aq.), rt, 24 h)
and were disappointed to find that the product
obtained was an inseparable mixture of several mono-
and bis-protected guanidines. We also attempted to silyl
protect the hydroxyl function of the monocyclic prod-
ucts 17–19 and 21 and found that although we could
prepare the corresponding silyl derivatives 22–25 the
yields were low and the products somewhat prone to
hydrolysis (Scheme 3). The reason for this instability
has not been investigated, however the presence of the
internal guanidinum species acting as a proton source
might be a contributing factor and Elliot and Long
have reported9 that in a compound structurally similar
to 22, a desilylation occurs as a result of anchimeric
assistance by the guanidine.
With these heterocyclic products in hand we hoped to
prepare derivatives of them in order that we might
investigate further synthetic modifications. We
attempted to bis-Boc protect the guanidine function

3078
M. Dennis et al. / Tetrahedron Letters 44 (2003) 3075–3080
The failure to discriminate effectively between the func-
tional groups in these heterocycles and the other com-
plications associated with polymer formation, together
with the associated purification problems are obvious
drawbacks of this methodology. Despite this, the work
does demonstrate a predictable mode of cyclisation for
v-halo-epoxides.
neutral conditions and on monitoring the reaction by
1H NMR it was apparent that after 24 h ca. 80% of the
substrate had been converted into the epoxide 27. On
continued stirring this intermediate was consumed to
give, on work up, a 62% yield of the cyclic product 28,8
the structure of which was confirmed by X-ray analysis
(Fig. 2)11 and by conversion to the previously prepared
compound 17 by treatment with CF3CO2H. We also
investigated the iodocyclisation of the guanidine 26 and
found that this was also an effective reaction leading to
the cyclic guanidine 298 in 85% yield. Again the struc-
ture was confirmed by X-ray crystallography (Fig. 3)11
(Scheme 4).
We wished to improve upon this initial work and to
direct it more towards a synthesis of cylindrosper-
mopsin and as the most predictable ring-opening mode
is the 5-exo cyclisation of substrates 13 and 14, we
studied these reactions further. We investigated an in
situ method for the formation of an intermediate simi-
lar to those proposed for the preparation of 17 and 18.
We thus took the known N-allyl-bis-Boc-guanidine10 26
and treated it with dimethyl dioxirane (DMDO) under
The synthesis of these two compounds in high yield
should enable us to prepare a range of related heterocy-
cles and also enable us to study a unique biomimetic2b
approach to cylindrospermopsin. In addition heterocy-
Figure 2.
Figure 3.
Scheme 3. Reagents and conditions: (a) (i) 3 equiv. TBSCl, imid., DMF, 16–24 h, (ii) NaBF4 (satd aq.).

M. Dennis et al. / Tetrahedron Letters 44 (2003) 3075–3080
3079
Scheme 4. Reagents and conditions: (a) 1.5 equiv. DMDO, acetone 0°C–rt, 48 h; (b) CF3COOH, 1 h; (c) 4 equiv. I2, 4 equiv.
K2CO3, MeCN, 48 h.
cles of this type are of general interest, as related
compounds such as 1-deoxynojirimycin12 and certain
amidines13 and guanidines14 are known to disrupt
biosynthesis of N-linked glycoproteins and glycolipids
and are very effective glycosidase inhibitors.
8315–8332; (e) Black, G. P.; Murphy, P. J.; Walshe, N.
D. A.; Hibbs, D. E.; Hursthouse, M. B.; Malik, K. M. A.
Tetrahedron Lett. 1996, 37, 6943–6946; (f) Black, G. P.;
Murphy, P. J.; Walshe, N. D. A. Tetrahedron 1998, 54,
9481–9488; (g) Black, G. P.; Murphy, P. J.; Thornhill, A.;
Walshe, N. D. A.; Zanetti, C. Tetrahedron 1999, 55,
6547–6554; (h) Moore, C. G.; Murphy, P. J.; Williams,
H. L.; McGown, A. T.; Smith, N. K. Tetrahedron Lett.
2003, 44, 251–254; (i) For a review on other synthetic
approaches to these metabolites, see: Heys, L.; Moore, C.
G.; Murphy, P. J. Chem. Soc. Rev. 2000, 29, 57–67.
2. (a) Ohtani, I.; Moore, R. E.; Runnegar, M. T. C. J. Am.
Chem. Soc. 1992, 114, 7941–7942; (b) For a review on the
biological activity and synthetic approaches to cylindros-
permopsin, see: Murphy, P. J.; Thomas, C. W. Chem.
Soc. Rev. 2001, 30, 300–312; Heintzelman, G. R.; Fang,
W. K.; Keen, S. P.; Wallace, G. A.; Weinreb, S. M. J.
Am. Chem. Soc. 2002, 124, 3939–3945.
We have performed a range of glycosidase assays15 on
compounds 17–19 and 21–25, and in general these
compounds appeared to be relatively weak but specific
inhibitors of b-galactosidase (Bovine liver) with com-
pound 25 being the most inhibitory (IC50=5.6 mg ml−1,
Ki=18 mM) and 23 (IC50=31 mg ml−1) and 19 (IC50=
49 mg ml−1) having less activity. There was an interest-
ing stimulation of b-galactosidase from E. coli by 19
(approximately threefold increase at 0.6 mM) and less
by 25 (twofold at 0.4 mM). Removal of the trifluoroac-
etate from 19 by anion exchange removed the stimula-
tion but not the inhibition whilst the trifluoroacetate
anion had no effect. Similarly, changing the counterion
of 25 to chloride removed the stimulation whilst not
changing the inhibition. Whilst the level of inhibitory
and stimulatory activity is still fairly low, the specificity
for b-galactosidase is of interest and we are directing
our synthetic efforts towards related compounds.
3. For an example using a deprotonated guanidine, see:
Chen, B.-C.; Quinlan, S. L.; Reid, J. G.; Jass, P. A.;
Robinson, T. P.; Early, W. A.; Delaney, E. J.; Humora,
M. J.; Madding, G. D.; Venit, J. J.; Winter, W. J.
Tetrahedron: Asymmetry 1998, 9, 1337–1340.
4. Fritsche-Lang, W.; Wilharm, P.; Hadicke, E.; Fritz, H.;
Prinzbach, H. Chem. Ber. 1985, 118, 2044–2078.
5. (a) Le Merrer, Y.; Gauzy, L.; Gravier-Pelletier, C.;
Depezay, J. C. Bioorg. Med. Chem. 2000, 8, 307–320; (b)
Gravier-Pelletier, C.; Bourissou, D.; Le Merrer, Y.;
Depezay, J. C. Synlett 1996, 275–277; (c) Gauzy, L.; Le
Merrer, Y.; Depezay, J. C. Synlett 1998, 402–404.
6. (a) Genski, T.; Macdonald, G.; Wei, X.; Lewis, N.;
Taylor, R. J. K. Synlett 1999, 795–797; (b) Genski, T.;
Macdonald, G.; Wei, X.; Lewis, N.; Taylor, R. J. K.
Arkivoc 2000, 1, 266–273.
Acknowledgements
Thanks are given to the EPSRC for a studentship for
A.J.T. (GR/K81966) and to the EPSRC Mass Spec-
trometry centre at Swansea.
7. Epibromohydrin 13 was purchased from Sigma–Aldrich;
epoxides 14 and 15 were prepared by epoxidation of the
corresponding bromides (mCPBA, 0°C–rt, CH2Cl2, 16 h;
for 14, 80%, for 15, 92%). Mesylated epoxide 16 was
prepared in two steps from E-1-hydroxyhex-3-ene [(i)
mCPBA, 0°C–rt, CH2Cl2, 16 h; 72%; (ii) MeSO2Cl, Et3N,
0°C–rt, CH2Cl2, 16 h; 73%].
8. Selected spectroscopic data. Compound 17; oil, lH
(CD3OD) 3.73, (1H, dd, J=6.1, 12.1 CHH), 3.77 (1H,
dd, J=5.2, 10.1 Hz, CHH), 3.84 (1H, dd, J=4.2, 12.1
References
1. (a) Murphy, P. J.; Williams, H. L.; Hursthouse, M. B.;
Malik, K. M. A. J. Chem. Soc., Chem. Commun. 1994,
119–120; (b) Murphy, P. J.; Williams, H. L. J. Chem.
Soc., Chem. Commun. 1994, 819–820; (c) Murphy, P. J.;
Williams, H. L.; Hibbs, D. E.; Hursthouse, M. B.; Malik,
K. M. A. J. Chem. Soc., Chem. Commun. 1996, 445–447;
(d) Murphy, P. J.; Williams, H. L.; Hibbs, D. E.; Hurst-
house, M. B.; Malik, K. M. A. Tetrahedron 1996, 52,

3080
M. Dennis et al. / Tetrahedron Letters 44 (2003) 3075–3080
Hz, CHH), 3.93 (1H, dd, J=10.1, 10.1 Hz, CHH), 4.30
R indices (all data): R1=0.1774. wR2=0.2612. Crystal
data for 29 (CCDC no. 190507). Colourless plates,
C14H24N3O4I, Mr=425.26, T=120(2) K, monoclinic,
(1H, m, CHH). FTIR wmax 3352, 1550. MS (CI): m/z 116
(40%, [M+H]+) HRMS (CI): C4H10N3O ([M+H]+)
requires 116.0824, found 116.0825. Compound 28; mp
130°C. lH (CDCl3) 1.49 (9H, s, t-Bu), 1.54 (9H, s, t-Bu),
3.72, (1H, dd, J=3.5, 11.5 Hz, CHH), 3.78–3.83 (2H, m,
2×CHH), 3.85 (1H, dd, J=9.1, 12.5 Hz, CHH), 4.2 (2H,
br s, NH, OH), 4.25 (1H, m, CH). FTIR: wmax 3346, 2983,
2975, 1753, 1743, 1602. MS (CI): m/z 316 (100%, [M+H]+)
HRMS (CI): C14H26N3O ([M+H]+) requires 316.1872,
found 316.1868. Compound 29; mp 60°C. lH (CDCl3)
1.50 (9H, s, t-Bu), 1.56 (9H, s, t-Bu), 3.29 (1H, t, J=9.2
Hz, CHH), 3.39 (1H, dd, J=2.7, 9.7, Hz, CHH), 3.63
(1H, dd, J=3.7, 13.4 Hz, CHH), 3.97 (1H, dd, J=9.5,
13.4 Hz, CHH), 4.22 (1H, br m, CH), 9.2 (1H, br s, NH).
FTIR: wmax 3316, 2981, 2934, 1758, 1706, 1530. MS (CI):
m/z 426 (30%, [M+H]+). HRMS (CI): C14H25N3O4I ([M+
H]+) requires 426.0890, found 426.0886.
space group P21/n, a=8.2871(3), b=9.5256(3), c=
3
,
,
22.8746(6) A, i=92.77(1), V=1803.61(10) A , zcalcd
=
1.566 g cm−3, v=1.795 mm−1, Z=4, reflections collected:
11437, independent reflections: 11441 (Rint=0.000), final
R indices [I>2I]: R1=0.0765, wR2=0.1879, R indices (all
data): R1=0.1289. wR2=0.2234.
12. Truscheit, E.; Frommer, W.; Junge, B.; Muller, L.;
Schmidt, D. D.; Wingender, W. Angew. Chem., Int. Ed.
Engl. 1981, 20, 744–761.
13. (a) Papandreou, G.; Tong, M. K.; Ganem, B. J. Am.
Chem. Soc. 1993, 115, 11682–11690; (b) Chan, A. W.-T.;
Ganem, B. Tetrahedron. Lett. 1995, 36, 811–814; (c)
Fotsch, C. H.; Wong, C.-H. Tetrahedron. Lett. 1994, 35,
3481–3484.
14. (a) Knapp, S.; Choe, Y. H.; Reilly, E. Tetrahedron Lett.
1993, 34, 4443–4446; (b) Bleriot, Y.; Genre-Grandpierre,
A.; Tellier, C. Tetrahedron. Lett. 1994, 35, 1867–1870.
15. The Sigma enzymes used were: b-galactosidase (bovine
liver), a-glucosidase (Bakers yeast), a-mannosidase (Jack
9. Elliott, M. C.; Long, M. S. Tetrahedron Lett. 2000, 43,
9191–9194.
10. Drake, B.; Patek, M. L.; Lebl, M. Synthesis 1994, 579–
581.
11. Cell dimensions and intensity data for 28 and 29 were
recorded using a Bruker Nonius KappaCCD area detec-
tor diffractometer mounted at the window of a molybde-
num rotating anode following standard procedures.
Crystal data for 28 (CCDC number=190508). Colourless
block, C14H24N3O5, Mr=314.36, T=293(2) K, mono-
bean), b-glucosidase (almond), a-L-fucosidase (human
placenta), N-acetyl-b-glucosaminidase (bovine kidney),
Naringinase (Penicillium decumbens) and a-galactosidase
(green coffee beans). The substrates were 5 mM p-
nitrophenylglycopyranosides and enzymes 1 unit/ml. The
method was as described previously.16
clinic, space group P21/c, a=19.380(2), b=13.2761(11),
3
16. Watson, A. A.; Nash, R. J.; Wormald, M. R.; Harvey, D.
J.; Dealler, S.; Lees, E.; Asano, N.; Kizu, H.; Kato, A.;
Griffiths, R. C.; Cairns, A. J.; Fleet, G. W. J. Phytochem-
istry 1997, 46, 255–259.
,
,
c=13.0097(11) A, i=91.600(3), V=3346.0(5) A ,
zcalcd=1.248 g cm−3, v=0.095 mm−1, Z=8, reflections
collected: 15984, independent reflections: 5259 (Rint
=
0.1193), final R indices [I>2I]: R1=0.0849, wR2=0.2158,