Acremolin from acremonium strictum is N 2,3-etheno-2′- isopropyl-1-methylguanine, not a 1 H -azirine. Synthesis and structural revision

Article abstract of DOI:10.1021/ol400752s

The first synthesis of the heterocyclic marine natural product, acremolin, is reported along with the revision of the structure from a 1H-azirine to a substituted N²,3-ethenoguanine (5-methyl-7-isopropyl-4,5- dihydroimidazo[2,1-b]purine). Additional properties of acremolin are also described including its ¹H-¹⁵N-HMBC and fluorescence spectra.

Full text of DOI:10.1021/ol400752s

ORGANIC
LETTERS
2013
Vol. 15, No. 10
2370–2373
Acremolin from Acremonium strictum
is N2,3-Etheno-20-isopropyl-1-
methylguanine, not a 1H‑Azirine.
Synthesis and Structural Revision
Lawrence A. Januar† and Tadeusz F. Molinski*,†,‡
Department of Chemistry and Biochemistry, and Skaggs School of Pharmacy and
Pharmaceutical Sciences, University of California San Diego, 9500 Gilman Drive
MC0358, La Jolla, California 92093, United States
tmolinski@ucsd.edu
Received March 20, 2013
ABSTRACT
The first synthesis of the heterocyclic marine natural product, acremolin, is reported along with the revision of the structure from a 1H-azirine to
a substituted N2,3-ethenoguanine (5-methyl-7-isopropyl-4,5-dihydroimidazo[2,1-b]purine). Additional properties of acremolin are also described
15
including its 1Hꢀ N-HMBC and fluorescence spectra.
Compounds containing aziridine (azacyclopropane) or
azirine (azacyclopropene)1 rings are rare in nature. Fused
aziridine rings can be found in the structures of
mitomycins,1b potent DNA alkylating antitumor antibi-
otics;however, 2H-azirine-containing natural productsare
exceptionally uncommon. We were intrigued by the recent
report of a modified nucleobase, acremolin, obtained by
fermentation of the marine-derived fungus Acremonium
strictum, and claimed as the “first 1H-azirine containing
natural product”.2 The structure 1 (Figure 1) assigned to
acremolin is highly surprising.
Figure 1. 1H-Azirine (2a), 2H-azirine (2b), and natural azirines 3
and 4 (see text). Structure 1 originally assigned to acremolin2a
and here revised to 5a.
† Department of Chemistry and Biochemistry.
‡ Skaggs School of Pharmacy and Pharmaceutical Sciences.
(1) (a) Nair, V. The Chemistry of Heterocyclic Compounds; Hassner,
A., Ed.; Wiley: New York, 1983; Vol. 42, part 1, Chapter 2, pp 215.
(b) Carter, S. K.; Crooke, S. T.; Alder, N. A. Mitomycin C: Current
Status and New Developments; Academic Press: New York, 1979.
(2) (a) Julianti, E.; Oh, H.; Lee, H.-S.; Oh, D.-C.; Oh, K.-B.; Shin, J.
Tetrahedron Lett. 2012, 53, 2885–2886. (b) Hill, R.; Sutherland, A. Nat.
Prod. Rep. 2012, 29, 829–833.
It is well-known that the antiaromatic parent 1H-azirine
(2a) is unstable and rapidly tautomerizes to 2H-azirine
(2b); an isomer more stable by at least 30 kcal molꢀ1 3
Synthetic efforts to prepare stable substituted 1H-azirines
.
3
have failed although their ephemeral appearance as inter-
mediates in high energy reactions (e.g., N-methyl-
1H-azirines from flash-vacuum pyrolysis of substituted
1H-2,3,4-triazoles) has been postulated.4 Although the total
(3) The free energy of the parent 2a has been calculated to be
140 kJ molꢀ1 (33.5 kcal molꢀ1) higher than that for 2b. Zavoruev,
3
3
S. M.; Rakauskas, R-I. I. Theor. Exp. Chem. 1990, 25, 445–450.
r
10.1021/ol400752s
Published on Web 05/01/2013
2013 American Chemical Society

ring-strain energy of the more stable 2H-azirine tautomer 2b
with 10, bromoketone 814 and 2-chloro-3-methylbutanal
(915), in attempts to prepare both regiosiomers 5a and 5b.
has been calculated to be 44.6ꢀ48 kcal molꢀ1 using DFT
3
methods,5 substituted 2H-azirines including azirinomycin
(3), the long-chain C18 base dysidazirine (4),6 and
its analogs7 occur in nature and are remarkably stable.
Tautomerism is not possible in the proposed structure 1
due to the tertiary nitrogen, yet skepticism remains that such
an unstable compound could exist in nature.
Scheme 1. Synthesis of Acremolin: Revised Structure, 5a
Upon examination of the spectroscopic evidence pre-
sented for 12 we hypothesized two plausible non-azirine
structures for acremolin based on the N2,3-ethenoguanine
skeleton, 5a or the regioisomeric 5b (Scheme 1), would be
more compatible with the natural product.2a Independently,
Banert proposed the same hypothesis and suggested8 5a
better fits the NMR data for acremolin based on analysis
of predicted 13C NMR chemical shift increments for 5a
and 5b, although without experimental evidence. Here we
describe the first synthesis of 5a and experimental proof
1
(HRMS, UVꢀvis, H, 13C, and 15N NMR data) that 5a
and acremolin are the same.
The concise preparation of 5a (Scheme 1) improves on
earlier variants of N2,3-ethenoguanosine synthesis.9,10
Kusmierek reported low yields (∼28%) of the latter com-
pound by reaction of 2-chloroacetaldehyde with guanosine.
In our hands, attemped reactions of N-methylguanosine
with R-haloketones or R-haloaldehydes were complicated
by mixtures resulting from overalkylation and depurination.
LCMS evidence suggested that a dialkylated product arose
by initial alkylation at N7 to give a quaternary adeninium
salt which rapidly eliminated ribose, followed by a second
alkylation of the liberated free base.
To avoid these difficulties, we chose to start with a
suitably protected nucleobase. 1-Methylguanine (6) was
prepared in good yield (69%, Scheme 1) by selective
methylation of guanosine (DMSO, CH3I, NaH)11 followed
by hydrolysis (2 M HCl, 61%). Protection of 6 (PMB-Cl,
K2CO3, DMF)12 gave a mixture of N7- and N9-PMB
derivatives 7a,b (∼1.6:1, 56%) from which 7a could be
isolated by fractional crystallization (EtOH).13 We ex-
plored two complementary electrophiles for condensation
Reaction of 7 with bromoketone 8 ((i-Pr)2NEt, CH3CN,
40 °C) returned the N2,3-ethenoguanine derivative 10
(27%) as a single isomer, along with the guaninium salt
10a (54%) resulting from monoalkylation of the imidazole
ring.16,17 Attempted removal of the PMB group of 10
by hydrogenolysis (H2, PdꢀC, MeOH, 1 atm, 16 h)
returned only starting material. FriedelꢀCrafts type trans-
benzylation, under the conditions reported by Fujii and
co-workers (90% aq H2SO4, toluene, 40 °C),18 smoothly
delivered samples of 5a of high purity, albeit in low yield
(39%).19 Gratifyingly, 5a was obtained in good yield
(76%) by simple treatment of 10 with neat CF3CO2H
(40ꢀ80 °C).
(4) Gilchrist, T. L.; Gymer, G. E.; Rees, C. W. J. Chem. Soc., Perkin
Trans. 1 1975, 1.
(5) (a) Heimgartner, H. Angew. Chem., Int. Ed. Engl. 1991, 30, 238.
ꢀ
(b) Calvo-Losada, S.; Quirante, J. J.; Suarez, D.; Sordo, T. L. J. Comput.
Chem. 1998, 19, 912.
(14) Gaudry, M.; Marquet, A. Org. Synth. 1976, 55, 24.
(15) Prepared by chlorination of isobutyraldehyde (NCS, L-proline,
CH2Cl2). Tessie Borg, T.; Danielsson, J.; Somfai, P. Chem. Commun.
2010, 46, 1281–1283.
(16) The structure of 10a was supported by HMBC data (DMSO-d6)
which included correlations from the CH2(CdO) 1H signal (δ 5.33, s,
H-100) to both C-8 (δ 138.9, d) and C-4 (δ 106.9, s)
(17) Attempts to suppress alkylation of the more nucleophilic imid-
azole ring by employing the corresponding N7-Ns and N7-Ts protected
guanines failed to give the product or gave intractable mixtures,
respectively.
(6) Molinski, T. F.; Ireland, C. M. J. Org. Chem. 1988, 53, 2103–5.
(7) (a) Salomon, C. E.; Williams, D. H.; Faulkner, D. J. J. Nat. Prod.
1995, 58, 1463–1466. (b) Skepper, C. K.; Molinski, T. F. J. Org. Chem.
2008, 73, 2592–2597.
(8) (a) Banert, K. Tetrahedron Lett. 2012, 53, 6443–6445. (b) Hill, R.;
Sutherland, A. Nat. Prod. Rep. 2013, 30, 213–217.
(9) Kusmierek, J. T.; Jensen, D. E.; Spengler, S. J.; Stolarski, R.;
Singer, B. J. Org. Chem. 1987, 52, 2374–2378.
(10) Sattsangi, P. D.; Leonard, N. J.; Frihart, C. R. J. Org. Chem.
1977, 42, 3292.
€
€
(11) Hobartner, C.; Kreutz, C.; Flecker, E.; Ottenschlager, E.; Pils,
W.; Grubmayr, K.; Micura, R. Monatsh. Chem. 2003, 134, 851–873.
(12) Montgomery, J. A.; Hewson, K.; Temple, C. J. Med. Chem.
1962, 5, 15–24.
(18) (a) Leonard, N. J.; Fujii, T.; Saito, T. Chem. Pharm. Bull. 1986,
34, 2037. (b) Nohara, F.; Nishii, M.; Ogawa, K.; Isono, K.; Ubukata,
M.; Fujii, T.; Itaya, T.; Saito, T. Tetrahedron Lett. 1987, 28, 1287. (c)
Ogawa, K.; Nishii, M.; Inagaki, J.-I.; Nohara, J.; Saito, T.; Itaya, T.;
Fujii, T. Chem. Pharm. Bull. 1992, 40, 1315.
(13) The location of the PMB group in 7a was confirmed at N-7 by
1Hꢀ13C HMBC (optimized for 1JCH = 8 Hz). Long-range correlations
(19) The balance of material was mostly a polar product, retained in
the H2O layer after extractive workup. By MS, this appeared to be
1
were observed from the H NMR signal of the benzylic CH2 of PMB
(δ 5.35, s) to C-5 (δ 107.0, s) and C-8 (δ 143.4, s), but not to C-4 (δ 153.5, s).
See Supporting Information for complete HMBC correlations of 7a.
10 SO3H (m/z 432, [M þ H]þ), the product of sulfonation of 10 by
3
H2SO4.
Org. Lett., Vol. 15, No. 10, 2013
2371

1
The H NMR (DMSO-d6, 500 MHz) and 13C NMR
Doubts regarding the structures of 5a and 5b were
1
15
spectra of 5a (DMSO-d6, 125 MHz; Table 1) and other
data (UVꢀvis, FTIR, HMBC, and HRMS; see Support-
ing Information) were identical in every respect with those
reported by Shin and co-workers for acremolin,2a with one
exception: in the 1H NMR spectrum of 5a (DMSO-d6), we
observed an additional allylic coupling from H-10 to H-30
removed upon analysis of the Hꢀ N HMBC spectrum
of synthetic 5a (Figure 2);26 six long-range correlations
were observed for 5a which served to assign the 15N NMR
chemical shifts. Two three-bond correlations (3JHN) were
observed from H-8 (δ 8.16, s) to the expected low-field
imidazole N-7 (δ 232.6 ppm) and N-9 (166.1 ppm) of the
3
(4J = 1.0 Hz). Importantly, the measured one-bond hetero-
purine ring. Most importantly, JHN correlations were
20
cyclic coupling constants of C-10ꢀH-10 in 5a (1JCH
=
observed from H-30 (δ 2.88, sept d, J = 6.9,1.0 Hz) and
H-10 (δ 7.38, d, J = 1.0 Hz) to the low-field imidazole N-11
(δ 221.9), and a two-bond correlation (2JHN) from H-10 to
N-3 (δ 156.1). These data support only regioisomer 5a and
exclude 5b; thus, acremolin is identical to synthetic 5a and
its structural revision from 1 to 5a is confirmed.
195.5 Hz)21 were consistent with N2,3-ethenoguanosine
(1JCH = 180 Hz),22,23 imidazole (1JCH = 189 Hz),24 and
predicted values,8 but incompatible with estimated values
for an N-substituted 1H-azirine (1JCH ≈ 230 Hz).8
Unlike structure 1, compound 5a raises questions
of regioisomerism. The location of the isopropyl group
in 5a follows from the mechanism of formation of N2,3-
ethenoguanine bases. Conceivably, each of the isomers 5a
and 5b can be formed by reaction of an R-halocarbonyl
compound with 6 through two pathways that differ only
by the ordering of the elementary steps of alkylation and
condensation. Under near-neutral or slightly acidic condi-
tions, the most likely order of reactions that gives 5a is
reversible imine formation between the C-2-NH2 group of
6 and the R-halocarbonyl compound followed by slower
irreversible intramolecular alkylation of the weakly nu-
cleophilic N3 giving 5a after tautomerization. Guengerich
and Persmark obtained support for the latter mechanism
in the formation of 11 with guanosine in aqueous ethanol
(pH = 6.4) through isotopic labeling with 13C-labeled
2-haloacetaldehyde and interpretation of the NMR
spectra.22 In our case, the formation of 5a is also observed
under initially basic conditions (i-Pr)2NEt, CH3CN,
45ꢀ100 °C), suggesting the order of reactions may be
reversed for formation of 5a, i.e. alkylation first, this time
at N3, followed by condensation at the C-2 NH2 group,
Table 1. Comparison of 1H and 13C NMR Data for Natural
Acremolin, Synthetic 5a ((CD3)2SO) and Estimated 13C δ
Values
acremolinb
synthetic 5ac
est.
δH,
m, J (Hz)
δH,
m, J (Hz)c,e
no.b
δC
δC
δC
a
c,e
d
2
142.5
141.6
108.3
152.6
140.3
142.3
141.6
108.9
152.8
140.5
141.7
141.0
108.9
153.4
140.0
4
5
6
8
8.16, s
8.16, s
9-NH
10
10
20
30
13.85, br s
3.57, s
13.87, br s
3.57, s
28.9
28.9
7.37, s
103.1
148.0
27.6
7.38, d (1.0 f)
103.2
148.0
27.7
105.7
147.2
2.88, sept
(6.8)
2.88, sept d
(6.9, 1.0 f)
40, 50
1.25 (d, 6.8)
22.0
1.25 d (6.9)
22.1
a See ref 20. b Data from ref 2a. c This work. d See ref 8. e Assignments
confirmed by HMBC, DEPT, and HSQC (500 MHz). f Allylic coupling,
possibly promoted by the byproduct salt (i-Pr)2NEt HBr,
3
4JHH
.
but with the same outcome. In principle, detection of long-
1
13
range Hꢀ C HMBC correlations should resolve this
equivocal assignment of 5a vs 5b. Unfortunately the key
3JCH cross peak, H-10 to C-4, was absent in the reported
HMBC spectrum of 5a2a and our own measurements.25
(20) For ease of comparison, we have retained the numbering
scheme of the isopropyl-substituted ring in acremolin used by Shin et al.
See ref 2. The systematic name for 5a, 5-methyl-7-isopropyl-4,5-
dihydroimidazo[2,1-b]purine, reflects differences between IUPAC and
conventional purine numbering.
1
(21) JCH values were conveniently measured in 1H NMR spectrum
of 5a (DMSO-d6) from the ‘13C satellites’ of the H-10 doublet unambi-
gously identified by allylic coupling to H-30 (4JHH = 1.0 Hz). The H-8
1JCH value is significantly larger (1JCH = 212.7 Hz) than H-10.
(22) Guengerich, F. P.; Persmark, M. Chem. Res. Toxicol. 1994, 7,
205–208.
Figure 2. 1Hꢀ N HMBC correlations of 5a (600MHz, DMSO-d6,
15
optimized for J = 8 Hz). Critical correlations are highlighted in red.
(23) Coincidently, the 1JCH value for C-10 in 5a is similar to that of the
sp3 carbon (C-3ꢀH-3) in 2H-azirines, e.g. 4 (1JCH = 189.5 Hz). See ref 6.
€
(24) Pretsch, E.; Buhlmann, P.; Badertscher, M. Structure Determina-
Attempted reactions of 6 with 9 to obtain the comple-
mentary ‘endo’ cyclized regioisomer 5b (CH3CN, 45ꢀ
100 °C, with or without added base [(i-Pr)2NEt] or acid
(NH4OAc, AcOH) or a nucleophilic catalyst (n-Bu4NBr))
were unsuccessful, presumably due to steric hindrance,
and returned only starting material. Similarly, attempts
to effect reductive alkylation of 6 or 7a (9, NaCNBH3,
tion of Organic Compounds, 4th ed.; Springer-Verlag: Berlin, 2009; p 108.
(25) Our 1Hꢀ13C HMBC measurements of 5a (optimized both for
3J = 4 and 8 Hz) also failed to detect this cross peak; see Supporting
Information, Table S2.
(26) The 15N NMR spectra of guanosine and related purine
bases have been assigned previously. (a) Glemarec, C.; Besidsky, Y.;
€
Chattopadhyaya, J.; Kusmierek, J.; Lahti, M.; Oivanen, M.; Lonnberg,
H. Tetrahedron 1991, 47, 6689–6704. (b) Cho, B. P.; Kadlubar, F. F.;
Culp, S. J.; Evans, F. E. Chem. Res. Toxicol. 1990, 3, 445–452.
2372
Org. Lett., Vol. 15, No. 10, 2013

MeOH, 23ꢀ80 °C), in anticipation of subsequent cyclizationꢀ
autoxidation, failed to deliver the expected secondary amine.
Additional spectroscopic data were measured for 5a.
The UV-spectrum of synthetic 5a [MeOH, pH 7, λmax
223 nm (4.49), 269 (4.05)] was essentially idenitical to that
reported for acremolin [MeOH, λmax 224 (4.27), 269 nm
(3.89)]. The chromophore of the proposed structure of 1
would be expected to show UV properties similar to those
of 6 (λmax 249, 273 nm), but acremolin shows the short-
wavelength band red-shifted to λ 269 nm (log ε 3.89).
In addition, we found dilute solutions of 5a exhibited
visible blue fluorescence (MeOH, λex 296 nm; λem 420 nm;
Stokes shift = 124 nm), a property shared by N2,3-
ethenoguanosine (11, H2O λex 264; λem 400, Stokes shift =
136 nm)9 but not 6 or isomers 7a and 7b.
Etheno-bridged guanine bases have a rich history.
In 1957, Dunn27 and Hoard28 independently reported
a ‘fluorescent compound’ formed as an artifact during
acid-promoted hydrolysis of DNA and condensation
of guanine with deoxyribose degradation products, but
earlier mistaken as a constituent of native DNA. A decade
later, Howard and Dekker provided full characteriza-
tion of the fluorescent compound as 10-acetonyl-N2,3-
ethenoguanine (12) (Figure 3).29 The chemistry and
biology of 1,N2- and N2,3-ethenoguanines have been
extensively studied,30 most importantly in the context of
exocyclic base modifications of DNA and RNA that
arise from environmental exposure to toxicants (e.g., vinyl
chloride), lipid peroxidation, and tobacco smoke.30a,b
Exocyclic bases, notably N2,3-ethenoguanine (11) and
its linearly fused regioisomer, 1,N2-ethenoguanine (13),
have been detected in the DNA isolated from livers of rats
subjected to inhalation exposure to high concentrations of
vinyl chloride,31 a carcinogenic compound that is trans-
formed by liver cytochrome P450 enzymes to the genotoxic
metabolite, chloroacetaldehyde. Compound11and itscor-
responding riboside and 50-O-phosphate ester have been
prepared by direct condensation of 2-bromoacetaldehye
with 6-O-methylguanosine, followed by depurination or
enzymatic 50-O-phosphorylation, albeit in lower yield9
than reported for 10 (Scheme 1).
Figure 3. Etheno-bridged guanines.
origin of 11ꢀ13, it appears more likely that 5a arises from
condensation of 1-methylguanine (or 1-methylguanosine
and accompanying hydrolysis) with an isovaleraldehyde
equivalent, possibly derived from 2-oxo-isocaproic acid
originating from transamination of leucine, followed by
oxidative cyclodehydrationꢀdecarboxylation.32
In summary, acremolin has been synthesized and
the structure revised from the substituted 1H-azirine 1 to
N2,3-etheno-20-isopropyl-1-methylguanine (5a); a struc-
ture fully consistent with the published spectroscopic
properties2a of the natural product, and the 1Hꢀ N
15
HMBC data, is reported here. The one-bond homonuclear
coupling constants (1JCH) and fluorescence properties
of 5a are described. In all likelihood, 1H-azirines remain
“unprecedented among natural products”.2a
Acknowledgment. We thank B. Duggan and A. Mrse
for assistance with NMR measurements. The 500 MHz
NMR spectrometer and the HPLC TOFMS were pur-
chased with funding from the NSF (Chemical Research
Instrument Fund, CHE0741968) and the NIH Shared
Instrument Grant (S10RR025636) programs, respectively.
We are grateful for generous research support from the
NIH (AI039987, AI100776).
Note Added after ASAP Publication. Table 1 contained
errors in the version published ASAP on May 1, 2013; the
correct version reposted on May 17, 2013.
Supporting Information Available. Experimental pro-
cedures and full spectroscopic data for all new compounds
and synthetic acremolin (5a). This material is available
free of charge via the Internet at http://pubs.acs.org.
Shin et al. inferred the involvement of an ‘isoprene unit’
in the biosynthesis of acremolin.2 In view of the known
(27) Dunn, D. B. Biochim. Biophys. Acta 1955, 18, 317.
(28) Hoard, D. E. Ph.D. Dissertation, 1957, University of California,
Berkeley.
(29) Howard, F. B.; Dekker, C. A. Biochemistry 1967, 6, 3581–3602.
(30) (a) Medeiros, M. H. G. Chem. Res. Toxicol. 2009, 22, 419–425.
(b) Marnett, L. J.; Burcham, P. C. Chem. Res. Toxicol. 1993, 6, 771–785.
(c) Leonard, N. J. CRC Crit. Rev. Biochem. 1984, 15, 125–199.
(31) Swenberg, J. A.; Fedtke, N.; Ciroussel, F.; Barbin, A.; Bartsch,
H. Carcinogenesis 1992, 13, 727–729.
(32) Ethenoguanine adducts have been detected during cell incuba-
tions of guanosine under various conditions; therefore, we cannot rule
out the possibility that acremolin is an artifact of the fermentation of
Acremonium strictum.2a
The authors declare no competing financial interest.
Org. Lett., Vol. 15, No. 10, 2013
2373