Regioselectivity in the multi-component synthesis of indolizinoquinoline-5, 12-dione derivatives

Article abstract of DOI:10.1002/ejoc.200600317

The one-pot cyclisation to form the indolizinoquinoline-5,12-dione ring system has been investigated starting from 6,7-dichloroquinoline-5,8-dione by reaction with pyridine and ethyl acetoacetate. Both the N,N-syn and the N,N-anti products have been fully characterised by mass spectrometry and NMR analysis, and the regioselectivity of their formation is discussed in terms of solvent polarity and/or the nature of the metal ion. We demonstrate here that, among a wide series of polar and apolar solvents, tert-butyl alcohol is the best choice to enhance the yield of the N,N-syn regioisomer, while the N,N-anti regioisomer can be obtained with a very high selectivity when the reaction is carried out with scandium triflate. In the presence of metal chelation, semi-empirical ZINDO calculations offer an appropriate tool to explain the regioselectivity observed when different metal ions are used. The study of the regioselectivity is also supported by the data for the corresponding two-step reactions with reversed addition of the nucleophilic reagents, and it has been extended to the cases involving 4-methylpyridine and 6,7-dichloroisoquinoline-5,8-dione as reagents. Wiley-VCH Verlag GmbH & Co. KGaA, 2006.

Full text of DOI:10.1002/ejoc.200600317

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
FULL PAPER
DOI: 10.1002/ejoc.200600317
Regioselectivity in the Multi-Component Synthesis of Indolizinoquinoline-5,12-
dione Derivatives
Andrea Defant,[a] Graziano Guella,[a] and Ines Mancini*[a]
Keywords: Cyclisation / Nitrogen heterocycles / N,O ligands / Regioselectivity / Solvent effects
The one-pot cyclisation to form the indolizinoquinoline-5,12-
dione ring system has been investigated starting from 6,7-
dichloroquinoline-5,8-dione by reaction with pyridine and
ethyl acetoacetate. Both the N,N-syn and the N,N-anti prod-
ucts have been fully characterised by mass spectrometry and
NMR analysis, and the regioselectivity of their formation is
discussed in terms of solvent polarity and/or the nature of the
metal ion. We demonstrate here that, among a wide series of
polar and apolar solvents, tert-butyl alcohol is the best choice
to enhance the yield of the N,N-syn regioisomer, while the
N,N-anti regioisomer can be obtained with a very high selec-
tivity when the reaction is carried out with scandium triflate.
In the presence of metal chelation, semi-empirical ZINDO
calculations offer an appropriate tool to explain the regiose-
lectivity observed when different metal ions are used. The
study of the regioselectivity is also supported by the data for
the corresponding two-step reactions with reversed addition
of the nucleophilic reagents, and it has been extended to the
cases involving 4-methylpyridine and 6,7-dichloroisoquino-
line-5,8-dione as reagents.
(© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2006)
asymmetric chlorine atoms at the C-6 and C-7 positions in
2 (Scheme 1). This reactivity finds examples in the nucleo-
philic attack both of amines[5] and active methylene rea-
gents.[6] Surprisingly, the single-step reaction of 2 with
active methylene compounds and pyridine derivatives has
been reported to give indolizino[2,3-g]quinoline-5,12-dione
as the only product,[7] which roused our interest in investi-
gating the key parameters that control the selectivity of the
production of the two regioisomers. We chose the effect of
solvent and the coordination of metal ions as variables for
both the one-pot and two-step reactions, and the outcome
of this study is here reported.
Introduction
One interesting group of chemotherapeutic agents used
in recent cancer therapy comprises molecules that interact
with DNA. In particular, polycyclic aromatic compounds
have been successfully studied as DNA-intercalating topo-
isomerase inhibitors. In accordance with Moore and Pind-
ur’s theory,[1] the structural requirements for active agents
are that the molecules contain both a planar tricyclic or
tetracyclic ring system and a conjugated p-quinone moiety
for the generation of reactive oxygen species, which are ulti-
mately considered to be the DNA fragmentation effectors.
The presence of one or more nitrogen atoms in this skeleton
is therefore vital for the formation of hydrogen bonds with
DNA.
In our recent study on the design and synthesis of new
compounds and on the evaluation of their bioactivity as
antitumour agents,[2] we have focused our attention on suit-
able derivatives bearing a naphthindolizinedione ring sys-
tem. The preparation of the central skeleton is based on a
single-step, multi-component reaction between 2,3-
dichloro-1,4-naphthoquinone (1), pyridine and an active
methylene compound.[3] Aza analogues bearing the quino-
line-5,8-dione system can be obtained by using 6,7-dichlo-
roquinoline-5,8-dione (2),[4] which gives the possibility of
producing two regioisomers due to the presence of the
[a] Laboratorio di Chimica Bioorganica, Dipartimento di Fisica,
Università di Trento,
via Sommarive14, 38050 Povo-Trento, Italy
Fax: +39-046-188-2009
E-mail: mancini@science.unitn.it
Scheme 1. Sequence for the cyclisation. Reagents: a) pyridine, ethyl
acetoacetate under different conditions.
Eur. J. Org. Chem. 2006, 4201–4210
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4201

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
A. Defant, G. Guella, I. Mancini
FULL PAPER
Results and Discussion
The one-pot cyclisation giving product 3[8] has been ex-
tended here to the preparation of the aza analogues. Re-
fluxing 6,7-dichloroquinoline-5,8-dione (2) with ethyl ace-
toacetate and an excess of pyridine in EtOH for 4 h gave a
deep-reddish mixture consisting of the N,N-syn (4) and
N,N-anti (5) regioisomers in a 64:36 ratio. The relative
1
amounts of the two products were deduced by H NMR
analysis of the reaction mixture. Each isomer could be iso-
lated in pure form by chromatography and fully character-
ised by mass spectrometry and NMR analysis.
The same molecular composition C18H12N2O4 was de-
duced for both regioisomers by HR-EIMS and further
structural indications came from tandem fragmentation ex-
periments in ESI(+)-MS analysis (see Experimental Sec-
tion). NMR spectroscopic data, including NOESY and
HMBC experiments, indicated that : (i) in the regioisomer
4, the resonance at δ = 179.17 ppm is attributable to C-10
by long-range hetero-correlation with 9-H at δ = 8.56 ppm;
as a consequence, the resonance at δ = 173.14 ppm must be
assigned to C-5; (ii) in the regioisomer 5, the resonance at
δ = 173.71 ppm is attributable to C-5 by long-range hetero-
correlation with 6-H at δ = 8.57 ppm, hence the resonance
at δ = 178.26 ppm must be assigned to C-10 (Scheme 2). In
the ester 3, the 13C NMR signals at δ = 180.30 and
175.20 ppm can easily be assigned to the carbon atoms C-
5 and C-10, respectively,[2] and this assignment is also in
good agreement with empirically calculated chemical shift
values from shielding effects (see Experimental Section). It
must be emphasised that the NMR spectroscopic data for
regioisomers 4 and 5 do not allow us to establish which
isomer is which. Empirical calculations, used for ester 3, are
not helpful here since they lead to the same values for both
regioisomers. A decisive hint for the unambiguous regioch-
Scheme 2. Structural assignments of the N,N-syn and N,N-anti re-
gioisomers.
verified by carrying out the reaction in two steps. In fact, a
stoichiometric amount of sodium acetate was required
when ethyl acetoacetate was added as the first reagent[10]
(Table 1, Entry 2). In this case, a slightly better yield could
be obtained than in the one-pot reaction, although signifi-
cant amounts of the dipolar by-products 7 and 8 (Figure 1)
were generated. On the other hand, when pyridine was
added as the first reagent to the substrate 2, a lower yield
was obtained due to a major formation of the same by-
products (Table 1, Entry 3). These were characterised by
NMR spectroscopy and ESI-MS analysis and the structures
were established in analogy with that of by-product 9, al-
ready reported from the reaction of 2,3-dichloronaphtho-
quinone (1),[8,11] by the assumption that the N,N-anti prod-
uct 7 would be the major isomer (see Experimental Sec-
tion).
1
emical assignment, based on H-13C long-range couplings,
requires the availability of a compound with protons lo-
cated at the right and left side of the quinone ring which
could correlate with one or both of the carbonyl groups. We
reasoned that the product obtainable by decarboxylation of
4 or 5 could be a good choice. Therefore, the pure re-
gioisomer showing the 4-H signal at δ = 9.94 ppm was sub-
jected to decarboxylation[9] to give a compound for which
both the doublet for 9-H and the singlet for 11-H were
found to be hetero-correlated to the same signal (δ =
181.06 ppm; Scheme 2). This is clear-cut evidence for the
N,N-syn structure of the regioisomer 6; more importantly,
the starting regioisomer must be 4, with the same N,N ge-
ometry.
Among all the active methylene compounds previously
used in the cyclisation involving substrate 1,[9] ethyl aceto-
acetate was selected here in order to introduce a carboxy-
ethyl unit for further functionalisation towards molecules as
potential antitumour agents.[2] This choice was also dictated
by its better yield when compared with other nucleophiles
such as diethyl malonate[8] and ethyl cyanoacetate.[9] In the
one-pot cyclisation, pyridine performs the double role of
Table 1. Synthesis of the regioisomers 4 and 5 from 6,7-dichloro-
quinoline-5,8-dione (2) upon changing the solvent.
Entry
Solvent
Conditions[a]
Ratio[b] 4/5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
EtOH
EtOH
EtOH
A
B
C
64:36
75:25
12:88
64:36
63:37
61:39
76:24
85:15
93:7
83:17
80:20
75:25
78:22
80:20
78:22
74:26
78:22
80:20
EtOH
D
nPrOH
nBuOH
iPrOH
A[c]
A[c]
A
tBuOH
tBuOH
2-methyl-2-butanol
benzene
toluene
C6H5Cl
THF
A
B
A[c]
A
A[c]
A[c]
A
DME
A
dioxane
DMSO
DMF
A[c]
A[c]
A[c]
[a] Reaction conditions: see Experimental Section. [b] Ratio evalu-
1
ated by integration of H NMR signals for 4-H of products 4 and
nucleophile and base, leading to the acetoacetate anion, as 5. [c] Reaction temperature fixed at 80 °C, for comparison.
4202
www.eurjoc.org
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2006, 4201–4210

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Multi-Component Synthesis of Indolizinoquinoline-5,12-dione Derivatives
FULL PAPER
tion of product 3,[8,11] we propose here a comprehensive
mechanism for the formation of 4 and 5 (Scheme 3).
After a first nucleophilic attack (possibly by a Michael
addition in the case of a C nucleophile) and displacement
of one chlorine atom, the second nucleophilic attack in-
duces the displacement of the second chlorine atom, fol-
lowed by cleavage of the acetyl group to give ring closure
and final aromatisation to the products.[8,9] In such a
mechanism the inversion in the regioselectivity ratios ob-
tained by changing the order of the reagents might be ex-
plained by a preferential attack at C-6 of both the nucleo-
philes. Our results are in fair agreement with the data re-
ported for nucleophilic attacks on compound 2 by ethyl ace-
toacetate ion,[12] whereas amines have been reported to give
substitution at C-6,[5a] at both C-6 and C-7[13] or only at C-
7.[5b] The slight enhancement of regioselectivity (12:88 vs.
25:75) observed when pyridine is the first reagent is ex-
plained by a major role of route (d) with respect to route
(b), possibly due to a mass-law effect of pyridine, which is
present in excess.
Surprisingly, Yanni[7] has reported that the same reaction
gives a single product, to which the N,N-syn structure of
compound 4 was ascribed mainly on mechanistic grounds.
However, the reported structural details cannot unambigu-
ously establish the structure of such a regioisomer and,
more importantly, the reaction carried out under the same
experimental conditions as reported by Yanni[7] affords two
regioisomers in the expected ratio (Table 1, Entry 4).
Changing from EtOH to other alcohols as solvent gave
Figure 1. Structures of by-products 7–9. Adopted numbering is for
convenience.
The regioselectivity, as observed in the one-pot sequence
(4/5 = 64:36), derives from a combination of the contri-
butions of the nucleophilic attacks at the C-6 and C-7 posi-
tions of the substrate 2. As shown in Scheme 3, the N,N-
syn isomer 4 can be obtained either by route (a), where the
first attack is by ethyl acetoacetate ion at the C-6 position,
or by route (c), where the first attack is by pyridine at the
C-7 position. Likewise, the N,N-anti isomer 5 is obtained
by routes (b) and (d), which suggests that the major role in
the selectivity is played by the nucleophilicity of the first
attacking reagent.
When such a reaction is carried out in two steps, the
regioselective ratio of the products changes depending upon
the addition sequence of the reagents. In particular, we ob-
tained a 75:25 ratio when acetoacetate ion was added as the practically the same selectivity as was obtained with pri-
first nucleophile and a 12:88 ratio when pyridine was the mary alcohols, irrespective of their chain length, as ob-
first one (Table 1, Entries 2 and 3). Following arguments served with EtOH, 1-propanol and 1-butanol (Table 1, En-
similar to those previously suggested to explain the forma- tries 1, 5 and 6, respectively). However, a slightly increased
Scheme 3. Possible routes to products 4 and 5.
Eur. J. Org. Chem. 2006, 4201–4210
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
4203

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
A. Defant, G. Guella, I. Mancini
FULL PAPER
N,N-syn/N,N-anti ratio was obtained for the reaction con-
ducted in 2-propanol (iPrOH) and tert-butyl alcohol
(tBuOH) or 2-methyl-2-butanol (Table 1, Entries 7–10).
These results indicate that the selectivity can be modulated
by solvents with similar polarities. Moreover, they are in
agreement with the lower solvent acidity of the tertiary
alcohol such that when the latter is used as solvent, an in-
creased availability of the free ethyl acetoacetate anion oc-
curs. This explanation finds support in the higher regiose-
lectivity (93:7) observed in tBuOH for the two-step reaction
by route (a) (Table 1, Entry 9).
In the presence of aprotic solvents, such as benzene, tolu-
ene, chlorobenzene, THF, DME, 1,4-dioxane, DMSO and
DMF, similar high regioselectivity was found (Table 1, En-
tries 11–18). This behaviour has been interpreted as deriv-
ing from an increased C-6 attack of ethyl acetoacetate
anion, which is a better nucleophile in aprotic media than
in EtOH and other primary alcohols as solvents.
Figure 2. Metal ion coordination to 6,7-dichloroquinoline-5,8-di-
one (2) and the minimised structure of silver complex 10.
The effect of some metal ions was also investigated for cluding the ScIII complex, as discussed below. The molecu-
the chelation of compounds like 5,8-quinolinequinone[14] lar composition of the complex was determined from the
and 6,7-dichloroquinoline-5,8-dione (2),[15] where, in par- ESI-MS data recorded in positive-ion mode, where the most
ticular, the regioselectivity of amination[5a,16] is affected by intense peak at m/z = 336 is attributable to the pseudo-
the formation of a metal chelate complex. The complex molecular ion [2·Ag]+; the isotopic distribution of the latter
where the N-1 atom and the carbonyl oxygen atom at C-8 was in perfect agreement with the simulated spectrum for a
in 2 are coordinated to the metal ion was proposed to be molecular composition C9H3AgCl2NO2 (Figure 3).
an activated intermediate that increases the electrophilicity
In the one-pot cyclisation of 2, when assisted by metal
at C-6 (Figure 2). We reasoned that such a metal chelation chelation, a preferential C-6 attack carried out at the same
effect would be of great help in the investigation of the re- time by ethyl acetoacetate anion and pyridine was involved.
gioselectivity and we focused our attention only on salts Thus, in the presence of 0.3 mol-equiv. of scandium triflate
soluble in a given solvent.
in EtOH, a 21:79 ratio of the regioisomers 4 and 5 was
Complex 10, prepared by treating a solution of 2 with an obtained due to preferential attack at C-6 by pyridine
equimolar amount of silver perchlorate in CD3OD, con- through enhancement of route (d) with respect to route (a)
firmed the metal–ligand binding by showing downfield shits in Scheme 3 (Table 2, Entry 1).
in the NMR spectra (Figure 2). In particular, the stronger
The regioselectivity was also affected by varying the
electrophilic nature of the C-6 centre in the chelate was indi- amount of the metal salt; in particular, the effect was more
cated by a major deshielding effect observed for this carbon marked in the case of scandium triflate than silver perchlo-
atom (∆δ = 1.50 ppm) than for C-7 (∆δ = 0.02 ppm). Sim- rate (Table 2, Entries 1, 4, 5 and 9 and 10, respectively).
ilar effects were also detected with other metal chelates, in- Again, this evidence suggests that a competitive formation
Figure 3. Experimental (left) and simulated (right) ESI(+) mass spectra for the silver complex 10, showing the cluster with the most
intense signal at m/z = 336 [C9H3AgCl2NO2+].
4204
www.eurjoc.org
© 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2006, 4201–4210