Thermal cyclisation reactions of methyl 2-( pyrrol-1-yl ) cinnamate and methyl 3-( 1-phenylpyrrol-2-yl ) propenoate

Flash vacuum pyrolysis (FVP) of the pyrrolylcinnamate derivative 3 gives the pyrrolo[1,2a]quinoline 10, the pyrrolo[2,1-a]isoindole 13 (R = Me) and the benzindoles 14 and 15, all in low yield, whose structures were determined by NMR methods. The isomeric precursor 4 cyclises efficiently under the same conditions to give the pyrrolizin-3-one 16.


Introduction
In previous work we have established a versatile elimination-cyclisation sequence for the formation of fused pyrrol-2-one units (e.g.pyrrolizin-3-one 1) under flash vacuum pyrolysis (FVP) conditions (Scheme 1).

Scheme 1
Although the general strategy was shown to be applicable to the synthesis of a variety of ring systems, attempted formation of the azepinone 2 by FVP of a vinylogous precursor was not † Hamish McNab was Secretary and Treasurer of the RSC Heterocyclic Group during the period 1989-1992 successful. 1 This disappointing result may be due to the rotational degrees of freedom in the diene precursor at the high pyrolysis temperatures (Scheme 1).In order to probe this issue further, we reasoned that the incorporation of a fused benzene ring as part of the diene system would reduce these degrees of freedom and so might favour cyclisation to the larger ring size.For the synthesis of benzannelated pyrrolizin-3-ones, we have shown that the corresponding 1arylpyrroles are useful starting materials, since they undergo a 1,5-aryl shift under FVP conditions prior to the elimination and cyclisation sequence (Scheme 2).The cinnamate ester 3 was therefore selected as the initial target for the work reported in this paper.In the event, this precursor did not lead to seven-membered rings but instead has resulted in the discovery of a new type of electrocyclisation-elimination process.The contrasting thermal behaviour of the isomeric precursor 4 is also reported.

Results and Discussion
The precursor 3 was made from 2-nitrocinnamic acid 5 by sequential formation of the ester 6 (64%), selective reduction to the amino compound 7 by zinc dust in the presence of ammonium chloride 3 (51%) and formation of the pyrrole ring by condensation with 2,5dimethoxytetrahydrofuran (54%).The crude precursor 3 was purified by dry-flash chromatography and characterised by standard methods.Its 1 H NMR spectrum confirms that the E-configuration of the alkene unit has been maintained throughout the reaction sequence ( 3 J 16.0 Hz).A phenyl-substituted analogue of 3, prepared by a different route, has been reported. 4The isomeric precursor 4 was made by formylation of 1-phenylpyrrole 8 and separation of the 2carboxaldehyde isomer 9 (91% of the mixture), followed by Wittig reaction.FVP of the cinnamate 3 was carried out over a range of furnace temperatures, and it was found that 850 °C was required for complete conversion to products.Four significant products were isolated in low yield by dry-flash chromatography of the pyrolysate (Scheme 3).The first product (8%) showed a molecular ion at m/z 167 and was identified as pyrrolo[1,2-a]quinoline 10 by comparison with literature data. 5,6These data serve to exclude the isomeric pyrrolo[2,1a]isoquinoline 11, 7 which might have been formed by loss of CO from the anticipated pyrroloazepinone product 12.The connectivity of 10 is the same as in the starting material and so it follows that a novel direct cyclisation mechanism must take place which is of lower (or at least comparable) activation energy than the initial 1,5-aryl shift of the expected cascade process.The second product (m/z 169) was obtained in low yield (<11%) as the major component of an impure fraction.The most significant features of its 1 H NMR spectrum were a mutually coupled doublet (δ H 1.68) and quartet (δ H 5.08) in 3:1 integral ratio.The 13 C NMR spectrum showed one methyl group and eight CH resonances, one of which was in the aliphatic region (δ C 56.9), apparently attached to a nitrogen atom.A NOESY spectrum indicated that the methyl group was close in space to an aromatic proton as well as the aliphatic CH to which it is coupled.The pyrrolo[2,1-a]isoindole structure 13 (R = Me) is consistent with these data.Although this is not a known compound, the methylene protons of the parent pyrrolo[2,1-a]isoindole 13 (R = H) resonate at δ H 4.92, 8,9 very close to the value reported above for 13 (R = Me), and the remainder of the spectra are very similar [13 (R = H), aromatic protons δ H 7.1-7.3,pyrrole-type protons δ H 6.3-7.0;8,9 13 (R = Me), aromatic protons δ H 7.2-7.5,pyrrole-type protons, δ H 6.3-7.0].Although present in low yield, this product shows skeletal connectivity derived from an initial 1,5-aryl shift and must therefore be formed in preference to products of the expected eliminationcyclisation sequence (e.g.11) shown in Scheme 3.
The final two products (14% in total) were closely related; although they co-eluted from the column, a pure sample of the component which was slightly more polar was obtained.Both products showed the presence of an NH signal in their 1 H NMR spectra, together with two other signals due to pyrrole-type protons and six signals in the aromatic region.The product which was not obtained in pure form, was unambiguously identified by comparison of its spectra with those reported for 1H-benz[g]indole 14 (see Experimental section). 10In particular the 13 C NMR spectra are closely comparable.The remaining isomer (m/z 167), which was obtained in a pure state, was identified as 3H-benz[e]indole 15 by the following NMR experiments.(The reported spectra of 15 are not sufficiently clear to allow unambiguous identification. 11) NOESY (and COSY) correlations between the aromatic doublets at δ H 7.57 and 7.64 define the 1,2-fusion of the naphthalene ring and a NOESY correlation between the pyrrole-type proton at δ H 7.14 and the aromatic doublet at δ H 8.27 defines the mode of fusion of the heterocycle.Further NOESY correlations allow a complete assignment of the 1 H NMR spectrum and subsequent 13 C assignments were gained as a result of an HSQC experiment (Figure 1).In view of the unexpected nature of the results of this pyrolysis experiment, it was of interest to study the thermal behaviour of the isomeric pyrrole 4. In this case FVP at 850 °C gave a single main product which was identified as the pyrrolizin-3-one 16 (38% after chromatography) by comparison of its spectra with those of an authentic sample (Scheme 4). 1 In contrast to the pyrrole derivative 3, therefore, in the case of 4 a 1,5-sigmatropic shift of the phenyl group takes place (regiospecifically to the 5-position) followed by elimination of methanol to generate the ketene 17 and electrocyclisation to the pyrrolizinone 16 (Scheme 4).
These results clearly demonstrate that the anticipated rearrangement-eliminationcyclisation cascade dominates the FVP reactions of 4, though no products are formed by this sequence in the FVP reactions of 3. Possible reasons for this are discussed below.

Scheme 4
In pyrolysis of 3, the product 10 is obtained without skeletal rearrangement but with loss of the entire ester function; such a cyclisation mechanism appears to be unprecedented.Our suggested route involves an initial 10-electron (disrotatory) electrocyclisation to provide the extended dipolar species 18 (Scheme 5).This may take place either with the E-cinnamate 3a or with the Z-isomer 3b; E-Z-isomerisation of cinnamate esters is known to be essentially equilibrated at 850 °C under FVP conditions. 12Aromaticity of both the pyrrole and the benzene ring may be restored prior to the elimination step, by a 1,8-hydrogen shift from 18 to provide the 4,5-dihydro compound 19 (Scheme 5).There is good evidence that this route is followed in related cases. 13

Scheme 5
The oxidative elimination step which leads from 19 to 10 appears to be a novel thermal behaviour of esters, though the high temperature elimination of sulfonyl groups to generate stilbene sub-units has some similarity. 14The study of model compounds, to be published elsewhere, suggest that at high temperatures the CO 2 Me group, when it is situated in a benzylictype position, is able to function as a radical leaving group leading (after subsequent loss of H . ) to the oxidative elimination reaction observed in the pyrolysis of 3. It is also of interest that the thermal elimination of sulfinate from sulfones to give stilbene-type products, takes place by a radical mechanism. 14The most likely mechanism for oxidative elimination step in the formation of 10 from 3, therefore involves radical cleavage of the ester group to form a radical stabilised by the adjacent pyrrole ring, which can lose a hydrogen atom to form 10 (Scheme 5).
Though a similar electrocyclisation mechanism to that in Scheme 5 can be drawn for the FVP of 4, in this case the 1,5-aryl shift (leading to the pyrrolizinone 16) is lower in energy.Electronic effects are unlikely to account for this result; electron withdrawing groups generally migrate better in 1,5-sigmatropic shifts (though the effect may be masked for aryl migrations 15 ) so, paradoxically, the 1,5-aryl shift might be expected to be relatively favoured for 3 rather than 4. Though the experimental results are clear, the reasons for the contrasting behaviour of 3 and 4 remain unknown.It was thought initially that the benzindoles 14 and 15 could be derived from the pyrroloquinoline 10 by a sequence of 1,5-shifts via 20 as the key intermediate.(Scheme 6).However, control pyrolyses of 10 at 850 °C and even at 950 °C showed no trace of these potential rearrangement products and so it is likely that they are formed by a different mechanism.Thus, initial 1,5-shift(s) of the aryl group can generate the 2-and 3-arylpyrroles 21 and 22 respectively.Compounds 14 and 15 can then be formed by electrocyclisation followed by 1,5-hydrogen shift and elimination of the ester moiety in a similar fashion to the formation of 10 (Scheme 7 -c.f.Scheme 5).Once the 1,5-aryl shift has taken place, therefore, we believe that the novel ester elimination route is favoured over the expected ketene formation by methanol elimination.These results suggest that such an elimination step, requiring an eight or ninemembered transition state, is simply not viable, allowing the other processes to intervene.

ISSN 1424-6376
Page 91 © ARKAT USA, Inc.The formation of the pyrroloisoindole 13 is unusual (formal loss of C 2 H 2 O 2 from the starting material) but a possible mechanism is shown in Scheme 8. We propose that the initial 1,5-aryl shift is followed by a 1,9-H shift leading to the intermediate 23 which can collapse to 2-(pyrrol-2-yl)styrene 24 by loss of CH 2 =O and CO.A further 1,9-H shift leads to the o-quinone dimethide 25, the precursor of the final product.It is again noteworthy that a route such as this can apparently compete successfully with the expected ketene formation by methanol elimination from 21 (c.f.previous paragraph).In conclusion, it is clear from the results of the pyrolyses of 3 that the sequence of 1,5-aryl shift, E/Z isomerisation, thermal elimination of methanol and electrocyclisation to provide a fused azepinone system, cannot compete with alternative thermal processes.Product 10 is formed from 3 even prior to the expected sigmatropic shift of the N-aryl group.Products 13-15 are obtained after sigmatropic shift, but are formed in preference to the azepinone 12 or its decomposition products (e.g.11).We conclude that the methanol elimination step (required for the formation of 12 from 21) is not feasible geometrically, even though the formation of fivemembered rings (via a six-or seven-membered transition state) is highly efficient.This efficiency is illustrated by the formation of 16 exclusively, by pyrolysis of 4.
The substrate 3 is however left with no clearly defined route to products and a number of different pathways take place.One set of products, 10, 14 and 15, has provided a prototype of a new thermal cyclisation method, in which a newly recognised oxidative elimination of an ester group, taking place by a free-radical mechanism, provides a key aromatisation step.Further applications of this methodology in the synthesis of polycyclic ring systems are under investigation and will be reported in the near future. 13perimental Section General Procedures. 1 H and 13 C NMR spectra are recorded at 250 (or 200) and 63 (or 50) MHz respectively for solutions in [ 2 H]chloroform unless otherwise stated.Coupling constants are quoted in Hz; 13 C NMR signals refer to CH resonances unless otherwise stated. 1 H/ 13 C HSQC experiments were performed at 360 MHz using a Bruker DPX360 NMR spectrometer with gradient selection using the INVIGPTP sequence as provided by Bruker.NOESY and COSY experiments were similarly carried out at 360 MHz.Mass spectra were obtained under electron impact conditions unless otherwise stated.Methyl 2-nitrocinnamate (6).A solution of 2-nitrocinnamic acid 5 (4.84 g, 0.025 mol) and anhydrous potassium carbonate (3.82 g) in dimethylformamide (30 cm 3 ) was stirred magnetically.Iodomethane (3.92 g, 0.028 mol) was added dropwise over 10 min, and the solution left to stir for 24 h.The yellow solution was diluted with water (50 cm 3 ) and extracted with diethyl ether (3 × 50 cm 3 ).The ether was washed with water (6 × 50 cm 3 ) and dried over MgSO 4 .The ether was removed under reduced pressure to give the solid product 6 (3.31 g, 64%), mp 71-73 °C (lit., 16 72-73 °C), δ H 8.03 (1H, d, 3 J 15.9), 7.98 (1H, d, 3 J 10.0), 7.62-7.43(3H, m), 6.30 (1H, d, 3 J 15.9) and 3.76 (3H, s).Methyl 2-aminocinnamate (7) 3 .Methyl 2-nitrocinnamate 6 (3.31 g, 0.016 mol) was dissolved in methanol (60 cm 3 ).Water (7 cm 3 ) and ammonium chloride (1.35 g, 0.025 mol) were added.The solution was stirred and zinc dust (9.00 g) was added over 15 min.After the initial reaction had subsided the solution was heated under reflux for 2 h and filtered hot through celite.The solution was diluted with water (100 cm 3 ) and cooled in ice to separate the solid.The solid was filtered off and the solution extracted with dichloromethane (3 × 75 cm 3 ).The dichloromethane layer was separated from the aqueous layer and concentrated to give 7 as a yellow solid 3

Flash vacuum pyrolysis (FVP) experiments
The precursor was volatilised under rotary pump vacuum (pumping speed 100 dm 3 min -1 ) through an empty, electrically heated silica tube (35 × 2.5 cm) and the products were collected in a liquid nitrogen trap situated at the exit point of the furnace.The pressure was measured by a Pirani gauge situated between the product trap and the pump.The precursor, pyrolysis