A synthetic and in silico study on the highly regioselective Diels-Alder reaction of the polyenic antifungal antibiotics natamycin and flavofungin

Article abstract of DOI:10.1016/j.tetlet.2010.07.049

The tetraenic macrolide antibiotic natamycin and the pentaenic macrolide flavofungin gave monoadducts on Diels-Alder reaction with 4-phenyl-1,2,4- triazoline-3,5-dione. The regioselectivity of the reaction as well as the conformation of the products was studied using theoretical calculations.

Full text of DOI:10.1016/j.tetlet.2010.07.049

Tetrahedron Letters 51 (2010) 4968–4971
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
A synthetic and in silico study on the highly regioselective Diels–Alder
reaction of the polyenic antifungal antibiotics natamycin and flavofungin
e
c
a,
Zsolt Fejes a, Attila Mándi b,c, István Komáromi d, , László Majoros , Gyula Batta , Pál Herczegh
*
*
a Department of Pharmaceutical Chemistry, University of Debrecen, H-4032 Debrecen, Hungary
b Research Group for Carbohydrates of the Hungarian Academy of Sciences, University of Debrecen, H-4032 Debrecen, Hungary
c Department of Organic Chemistry, University of Debrecen, H-4032 Debrecen, Hungary
d Haemostasis, Thrombosis and Vascular Biology Research Group of the Hungarian Academy of Sciences, University of Debrecen, H-4032 Debrecen, Hungary
e Department of Medical Microbiology, University of Debrecen, H-4032 Debrecen, Hungary
a r t i c l e i n f o
a b s t r a c t
Article history:
The tetraenic macrolide antibiotic natamycin and the pentaenic macrolide flavofungin gave monoadducts
on Diels–Alder reaction with 4-phenyl-1,2,4-triazoline-3,5-dione. The regioselectivity of the reaction as
well as the conformation of the products was studied using theoretical calculations.
Ó 2010 Elsevier Ltd. All rights reserved.
Received 24 February 2010
Revised 29 June 2010
Accepted 9 July 2010
Available online 21 July 2010
Dedicated to Professor Károly Lempert on
the occasion of his 85th birthday
Polyene macrolide antibiotics such as amphotericin B, nystatin,
and natamycin are important therapeutics for the treatment of
fungal infections.1 Their structure consists of four to eight conju-
gated carbon–carbon double bonds with E-configuration. This
structural unit results in the rod-shaped form of these molecules.2
This conformation probably plays an important role in binding to
fungal sterol molecules resulting in the antifungal action of these
antibiotics. In this work two representatives of the polyene macro-
lide antibiotic family, natamycin (1) and flavofungin (2), have been
chosen to study a Diels–Alder (DA) reaction on their polyenic moi-
ety, as well as the influence of the newly formed pyridazine ring
annelation on the conformation of the macrocycle and on the anti-
fungal activity.
have been found, however, in the literature where geometrical
constraints exist for the polyene, for example, when a linear poly-
ene is assembled in a cyclic compound as in the case of polyene
macrolide antibiotics.
Natamycin4 (1) has a macrolactone structure consisting of 26
atoms. The conjugated tetraenic part of the molecule was the tar-
get for a DA reaction with the highly reactive diazadienophile 3.
The DA reaction of 1 with 3 at room temperature led to a single
product which was isolated by adsorption chromatography. On
the basis of NMR, MS, and UV–vis spectroscopic data (the configu-
ration of C-20 and C-23 was assigned on the basis of TS calcula-
tions) the structure 4 was assigned to this adduct (Scheme 1) as
Only a few examples have been published on Diels–Alder reac-
tions where the reacting diene is part of a polyene. In these cases
the polyene can be regarded as a substituted diene for which the
regio- and/or stereoselectivity is determined by its substituents
(e.g., the remainder of the polyene). A detailed experimental and
theoretical analysis of DA reactions of this type with polyenes pos-
sessing a carboxylate and hydroxyl group at the end was reported
by Turner et al.3 They found that the dienophile always reacts with
the outermost diene unit of the polyene, and the theoretical calcu-
lations carried out for a model compound supported their findings.
They also found that a simple model based on the frontier molec-
ular orbital coefficients for the prediction of regioselectivity of DA
reactions was not always applicable for such systems. No examples
* Corresponding authors. Tel.: +36 52 512 900x55178; fax: +36 52 340 011 (I.K.);
tel.: +36 52 512 900x22895; fax: +36 52 512 914 (P.H.).
E-mail addresses: komaromi@med.unideb.hu (I. Komáromi), herczeghp@gmail.
com (P. Herczegh).
Scheme 1. Diels–Alder reaction of natamycin (1) and 3.
0040-4039/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2010.07.049

Z. Fejes et al. / Tetrahedron Letters 51 (2010) 4968–4971
4969
Scheme 2. Natamycin theoretical model (1a) and its theoretical adduct model (4a).
Figure 1. Most stable ONIOM(B3LYP/6-31G(D):AM1) transition state for the Diels–
Alder reaction between 1a and 3.
detailed in the Supplementary data together with the observed
antifungal activity (Fig. S1, Table S1) . In an antifungal test against
Candida strains, product 4 exhibited no activity.
the trajectory and each of these was submitted to 10 ns constant
temperature (T = 400 K) simulations. The most stable geometries
from the corresponding trajectories were minimized at the HF/6-
31G and B3LYP/6-31G levels of theory.
Natamycin theoretical model (1a) (Scheme 2): Starting from a
plausible all-transoid structure,5 constant temperature (500 K)
molecular dynamics (200 ns total time, 1 fs time step, GAFF force
field6) simulations were carried out in order to explore the struc-
tures available with non-negligible populations for DA reactions.
A total of 200,000 structures were saved, and each was minimized
and classified into clusters regarding only the polyene chain con-
formation. The lowest energy representative from each cluster
was then submitted to semiempirical (AM1, PM3, PM6 and RM1)
and density functional (B3LYP/6-31G) calculations. Transition
states for cisoid structures which contain only a single cisoid con-
formation (for both possible orientation of the dienophile) were
calculated at AM1, PM3, PM6, RM1, and ONIOM(B3LYP/6-
31G(d):AM1) levels of theory.
Classifying the optimized structures from the conformational
search resulted in eight clusters within a narrow energy range
which means that any cisoid conformation suitable for a DA reac-
tion can be formed with sufficient probability. These clusters can
feature from zero to three cisoids and two to five transoids (i.e.,
the Ci + k–Ci + 1 + k–Ci + 2 + k–Ci + 3 + k torsion angles close to zero or
180 degrees, respectively, where i = 15 and k = 1 and/or 3 and/or
5, see Fig. S2). The description of these clusters as well as the en-
ergy values obtained for their lowest energy representatives at var-
ious levels of theory can be found in the Supplementarydata (Table
S2). The transition state (TS) calculations between 3 and 1a with
single cisoid geometries (at k = 1 or 3 or 5) at the AM1, RM1, and
PM6 as well as ONIOM(B3LYP/6-31G(d):AM1) levels of theory
show the TS of cisoid diene structures correspond to k = 5, being
the most stable in all types of computations. It is in perfect agree-
ment with the preparative findings where only this product was
formed. The graphical representations of the TSs calculated at ONI-
OM(B3LYP/6-31G(d):AM1 level are shown in Figures S3–S8 while
the relative TS energies for all the methods and for all the single-
cisoid diene subsystems are given in Table S3. The most stable
transition state obtained from ONIOM(B3LYP/6-31G(d):AM1)
calculations is shown in Figure 1.
Depiction of the energy and conformational changes during one
representative molecular dynamics simulation leading to the most
stable conformer is given in Figure 2 and Figure S9. Although the
macrocyclic ring in the most stable TS (Figure 1) obtained from
the single cisoid conformer and in the next local minimum
(Fig. S10) shows appreciable similarity in shape to those observed
at the global minimum of natamycin, it can be seen from these fig-
ures that product 4 does not prefer the corresponding elongated
rod-shaped geometry. Instead, it adopts a more compact confor-
mation (Fig. S12), which was reached through an intermediate
product (Fig. S11). Of the 20 short molecular dynamics simulations
carried out for the DA product 4a, 19 led to the same conformation
(Fig. S12) which was significantly different from the starting ‘TS-
like product’ (Fig. S10) and was lower in energy (Table S4) than
that was obtained for the geometry from the 20th simulation
(Fig. S13).
Flavofungin7 (2) (Scheme 3) was discovered 50 years ago at
Debrecen University. Its structure consists of a 32-membered mac-
rocycle with a conjugated pentaenic lactone moiety.8 In the reac-
tion with an equimolar quantity of 3 the major product isolated,
on the basis of spectroscopic data, proved to have the structure 5
Scheme 3. The numbering, details on the structural elucidation as
well as the observed antifungal activity of compound 5 are given
in the Supplementary data (Fig. S1, Table S1). The antifungal activ-
ity of 5 was evaluated against Candida yeast strains and this com-
pound proved to be inactive.
It should be mentioned that although in both cycloadditions
two new chiral centers were formed, by careful analysis of the
NMR spectra of 4 and 5, only a single diastereomer could be
observed.
Theoretical calculations on flavofungin (2): The computational
protocols outlined in the case of the natamycin theoretical model
were repeated with the exceptions that in this case there were more
clusters found due to the longer polyene chain and the transition
states were determined using AM1, PM3, PM6, RM1 as well
as ONIOM (B3LYP/6-31G:AM1:AMBER, B3LYP/6-31G:HF/6-31G:
AMBER, B3LYP/6-31G(d):HF/6-31G:AMBER, HF/6-31G(d):HF/6-
31G:AMBER, MP2/6-31G:HF/6-31G:AMBER, and MP2/6-31G(d):
HF/6-31G:AMBER) methods.
The structure of natamycin theoretical model DA adduct (4a)
(Scheme 2). The starting product geometry was obtained by mini-
mizing the adduct derived directly from the most stable transition
state (Fig. 1). Using this ‘transition state like product geometry’,
5 ns constant temperature molecular dynamics simulation was
performed at 900 K saving 20 geometries being equidistant from
This molecule has many rotatable dihedral angles and therefore
is too flexible to find the global and all the low energy local minima
with full certainty. Considering, however, only the polyene part of
the flavofungin (Fig. S2), it can be classified into 16 clusters
Figures and tables starting with ’S’ can be found in the Supplementary data.

4970
Z. Fejes et al. / Tetrahedron Letters 51 (2010) 4968–4971
Figure 2. Total energy of the Diels–Alder adduct 4a during 10 ns molecular dynamics simulation for one of the representative trajectories leading to the most frequently
found conformation.
Scheme 3. Diels–Alder reaction of flavofungin (2) and 3.
according to its pentaenic structure. These clusters can be charac-
terized from zero to four cisoids and three to seven transoids.
Based on the energy values (Table S5) obtained for each of the 16
geometries, the probability of the presence of any cisoid conforma-
Figure 3. ONIOM(B3LYP/6-31G(d):HF/6-31G:AMBER) transition state for the Diels–
Alder reaction between flavofungin (2) and 3 at k = 7.
tion suitable for a DA reaction is non-negligible.
effect of the carboxyl function by means of more sophisticated cal-
culations. Conjugated tri-, tetra-, and pentaenes with a carboxyl
group at the end of the polyene chain (with the restriction that
only one cisoid diene subsystem exists at the same time in the
chain) were optimized at AM1, HF/6-31G(d), B3LYP/6-31G(d),
and MP2/6-31G(d) levels of theory. The MO coefficients were
calculated at HF/STO-6G and B3LYP/6-31G(d) levels. The AM1,
HF/6-31G(d), B3LYP/6-31G(d), and MP2/6-31G(d) transition states
for DA reactions between these carboxylic acids (as dienes) and the
dienophile (3) were also calculated.
Regarding the DA transition state energies and the corresponding
single cisoid diene subsystem geometries in the polyene chain,
the PM3, as well as the B3LYP/6-31G:HF/6-31G:AMBER, MP2/6-
31G:HF/6-31G:AMBER, HF/6-31G(d):HF/6-31G:AMBER, B3LYP/6-
31G(d):HF/6-31G:AMBER, and MP2/6-31G(d):HF/6-31G:AMBER
ONIOM methods predict the most stable TS with a participating ci-
soid diene conformation at C8@C9–C10@C11. In contrast, the AM1
and PM6 methods prefer DA reaction at C6@C7–C8@C9, while the
RM1 and B3LYP/6-31G:AM1:AMBER ONIOM methods indicate reac-
tion at the C4@C5–C6@C7 atoms. Only the PM3 and ONIOM calcula-
tions with ab initio HF/6-31G ‘intermediate level’ of theory resulted
in a TS geometry in which the participation of the cisoid diene situ-
ated furthest from the lactone group is preferred in agreement with
the experimental finding. Energy values and geometries for the pos-
sible transition states for the single cisoid polyene can be found in
the Supplementary data (Table S6, Figs. S14–S21) while the most
stable ONIOM(B3LYP/6-31G(d):HF/6-31G:AMBER) TS is shown in
Figure 3.
The calculations outlined for model polyenoic acids were re-
peated for trienoic acid systems that also possess a –CH2–CH2–
OH group at the opposite end of polyene chain to the carboxyl
group. The TSs were determined for the
x-Me polyenoic acids at
ONIOM(MP2/6-31G(d):HF/6-31G) level of theory, as well. The ONI-
OM(MP2/6-31G(d):HF/6-31G) method, applying HF level only the
phenyl group, was validated against the pure MP2/6-31G(d) meth-
od at
x-Me-substituted and x-unsubstituted trienoic acids (Table
S10) and was proven to give qualitatively the same results.
Transition state calculations of the DA reactions between the
model polyenoic acids (as dienes) and 3 showed that the TSs
Polyenoic acids as flavofungin theoretical models: the DA reac-
tion of retinoic acid was studied experimentally and theoretically
by Yamada et al.9 Since flavofungin can be regarded as ‘substituted’
retinoic acid, in this work, we attempted to analyze the directing
involving the C
x-3–Cx-2–Cx-1–Cx (numbering started at the
carboxyl carbon) atoms were the most stable for all polyenoic acids

Z. Fejes et al. / Tetrahedron Letters 51 (2010) 4968–4971
4971
In contrast, carrying out theoretical calculations on flavofungin
itself shows a somewhat different picture since in this case the pre-
dicted regioselectivity showed method dependency. However,
applying the ab initio HF/6-31G method as an ‘intermediate’ level
of theory for the polyenoic acid part in ONIOM calculations was
sufficient to predict the experimentally observed regioselectivity.
In the case of natamycin the theoretical calculations demon-
strated that the rod-shaped conformation of the antibiotic was de-
formed as a consequence of the DA reaction, leading to a more
compact conformation of 4. The elongated shape is probably neces-
sary for a sterol complex formation, which might explain the loss
of the antifungal activity of 4 compared to the parent compound 1.
Software packages used in this work: The AMBER software
package6 was used for empirical force-field calculations while
quantum chemistry calculations were carried out using either MO-
PAC11 or GAUSSIAN’0312 suites. The conformers were classified into
clusters using in-house software.13 The MOLEKEL14 software package
was used for visualization of the results.
Figure 4. B3LYP/6-31G(D) transition state for the Diels–Alder reaction between a
pentaene carboxylic acid (flavofungin model) and 3.
and for all but the MP2/6-31G(d) levels of theory applied. At the
MP2/6-31G(d) level the TSs involving the Cx-5–Cx-4–Cx-3–Cx-
2 atoms were systematically the most stable one. Even though
the B3LYP method was shown to be remarkably successful for peri-
cyclic reactions,3,10 additional series of calculations were carried
out in order to resolve the discrepancy in the predicted (by means
MP2 versus other dependable methods) DA regioselectivity. Model
Acknowledgments
This work was supported by the Hungarian National Infrastruc-
ture Development Program (Grant: NIIF 1116) and the Hungarian
Scientific Research Funds (Grants: OTKA 68578, NKTH-OTKA CK
77515, TÁMOP-4.2.2-08/1/2008-0019, TÁMOP 4.2.1./B-09/1/
KONV-2010-0007 and OTKA 79126). We are grateful for the help-
ful discussions with Professor László Szilágyi in the field of NMR
assignments.
reactions between the Me-substituted (at the
x position relative to
the carboxyl group) polyenoic acids and 3 at ONIOM(MP2/6-
31G(d):HF/6-31G) level prefer again a DA reaction involving the
atoms of the farthest cisoid conformation from the carboxyl group.
Since the x-Me-substituted model is expected to mirror the proper-
ties of flavofungin in a greater extent, the results of calculations are
in a nice agreement with our experiments where the product corre-
sponding to this adduct (a ‘substituted pentaenoic acid’) was ob-
tained exclusively. It is also worth mentioning that this transition
state corresponds to a rather asynchronous reaction step since the
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.tetlet.2010.07.049.
bond between C
formation of the C
x
and the N of the dienophile precedes the
x
References and notes
-3–N bond, like those observed by Turner3
et al. from their theoretical calculations on symmetric polyenes
(Fig. 4). Attaching the –CH2–CH2–OH group to the other end of the
polyene (triene) chain resulted qualitatively in the same findings.
Energies for each of these transition states calculated at various
levels of theory can be found in Table S10 while the graphical rep-
resentations of the corresponding structures are given in Figures
S22–S39. It should be noted that using frontier MO coefficients
to predict the regioselectivity of the DA reaction seems to work
for trienoic and tetraenoic acids while it frequently fails (depend-
ing on the applied method) for the pentaenoic acid. The predicting
power of MO coefficients on the directing effect can be even more
doubtful in the trienoic acid case when the –CH2–CH2–OH substi-
1. Omura, S.; Tanaka, H. In Macrolide Antibiotics: Chemistry, Biology and Practice;
Omura, S., Ed.; Academic press: New York, 1984; pp 341–404.
2. Silberstein, A. J. Membr. Biol. 1998, 162, 117–126.
3. Turner, C. I.; Paddon-Row, M. N.; Willis, A. C.; Sherburn, M. S. J. Org. Chem. 2005,
70, 1154–1163.
4. Raab, W. Arzneim. Forsch. 1967, 17, 538–543.
5. Volpon, L.; Lancelin, J.-M. Eur. J. Biochem. 2002, 269, 4533–4541.
6. (a) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollamn, P. A.; Case, D. A. J. Comput.
Chem. 2004, 25, 1157–1174; (b) Case, D. A.; Cheatham, T.; Darden, T.; Gohlke,
H.; Luo, R.; Merz, K. M., Jr.; Onufriev, A.; Simmerling, C.; Wang, B.; Woods, R. J.
Comput. Chem. 2005, 26, 1668–1688.
7. Uri, J.; Békési, I. Nature 1958, 181, 908–909.
8. Szilágyi, L.; Sándor, P. Magn. Reson. Chem. 1990, 28, 963–972.
9. Yamada, S.; Hamano, K.; Shimizu, M. Tetrahedron Lett. 1991, 32, 2379–2382.
10. Wiest, O.; Montiel, D. C.; Houk, K. N. J. Phys. Chem. A 1997, 101, 8378–8388.
11. Stewart, J. P. P. MOPAC 2007, 2007, Stewart Computational Chemistry (SCC),
Paddington Circle Colorado Springs, CO 80921, USA.
tuent is at the
x position. Demonstrative MO coefficients and
eigenvalues are listed in Tables S8 and S9.
12. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;
Cheeseman, J. R.; Montgomery, Jr., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.;
Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.;
Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.;
Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao,
O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken,
V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A.
J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G.
A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.;
Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.;
Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.;
Fox, D. J. ; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe,
M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A.
GAUSSIAN 03, Revision C.02, Gaussian Inc., Wallingford CT, 2004.
In conclusion, the DA reactions at compounds 1 and 2 have been
carried out and the obtained products were characterized experi-
mentally. The theoretical calculations demonstrated that despite
the cyclic structures of 1 and 2, the flexibility of these molecules
allows the presence of the cisoid conformation at any position of
the polyene unit of the macrocycles.
In the case of the natamycin theoretical model, all the applied
quantum chemical methods supported the same regioselectivity
preference in perfect agreement with the experimental findings
despite the missing bulky sugar group. The applied semiempirical
and ab initio quantum chemical calculations on polyenoic acids (as
models of the flavofungin polyene chain) also predict a product in
agreement with the experiments. From these facts we favor a
mainly electronic nature of the observed DA regioselectivity.
13. Mándi, A. unpublished work.
14. Flükiger, P.; Lüthi, H. P.; Portmann, S.; Weber, J. MOLEKEL 5.4., 2000–2002, Swiss
Center for Scientific Computing, Manno, Switzerland.