Regio- and stereospecific assembly of dispiro[indoline-3,3′-pyrrolizine-1′,5′′-thiazolidines] from simple precursors using a one-pot procedure: synthesis, spectroscopic and structural characterization, and a proposed mechanism of formation

Dispiro[indoline-3,3′-pyrrolizine-1′,5′′-thiazolidine]s containing four contiguous stereogenic centres have been synthesized with high regio- and stereoselectivity in a one-pot procedure using simple starting materials. The various modes of supramolecular assembly depend upon different combinations of N—H⋯N, N—H⋯O, N—H⋯S=C and C—H⋯S=C hydrogen bonds.


Refinement
Crystal data, data collection and structure refinement details are summarized in Table 1. The atom labelling for the central dispiro unit is based on the systematic chemical numbering, following the convention used previously (Quiroga et al., 2017); thus, the atoms with chemical locants N1, C2 and so on are labelled here as N11, C12, etc.; those atoms with chemical locants such as C1 0 , C2 0 and so on are labelled here as C21, C22, etc.; and those atoms with chemical locants such as S1 00 , C2 00 and so on are labelled here as S31, C32, etc. All other chemical fragments are treated as substituents on the central dispiro unit. All H atoms were located in difference maps. H atoms bonded to C atoms were subsequently treated as riding atoms in geometrically idealized positions, with C-H = 0.95 (aromatic), 0.98 (CH 3 ), 0.99 (CH 2 ) or 1.00 Å (aliphatic C-H), and with U iso (H) = kU eq (C), where k = 1.5 for the methyl groups, which were permitted to rotate but not to tilt, and 1.2 for all other H atoms bonded to C atoms. For the H atoms bonded to N atoms, the atomic coordinates were refined with U iso (H) = 1.2U eq (N), giving an N-H distance of 0.862 (17) Å in (I) and 0.77 (3) Å in (II). For compound (II), the correct orientation of the structure with Table 1 Experimental details.

Results and discussion
Compounds (I)-(III) (Scheme 1) were each isolated as a single stereoisomer in yields of 53% for (I), 49% for (II) and 50% for (III). For all three products, the compositions were established by elemental analysis, complemented by high-resolution mass spectrometry in the case of (I) (x2.1). The 1 H and 13 C NMR spectra contained all the signals expected for the proposed formulations, and the regioselectivity of the reactions leading to the products was established by analysis of the 1 H spectra; it is necessary here to discuss only the analysis for (I), as those for (II) and (III) follow entirely similar lines. For (I), the signal from the proton H2 0 bonded to atom C2 0 (atom C22 in the crystallographic labeling scheme; see Fig. 1 and x2.2) was observed as a singlet at 4.64, while the signal for H7a 0 bonded to C7a 0 (C27A) was observed as a triplet (J = 7.22 Hz) at 4.42. These two signals indicate the formation of the pyrrolizine in (I), singly substituted at position C2 0 and doubly substituted at positions C1 0 and C3 0 , so confirming the identity of regioisomer (I) (Scheme 1 and Fig. 1). Had the alternative regioisomer (Ia) been formed, the appearance of these two pyrrolizine signals would have been different; that for atom H7a 0 would have been a doublet of triplets and, crucially, that for atom H1 0 would have appeared as a doublet, rather than the singlet actually observed. Entirely similar remarks apply to the spectra of compounds (II) and (III) but, in addition, five of the signals in the 13 C NMR spectrum of (III) exhibit coupling to the 19 F nucleus at position 5, namely, those at 159.0 for C5, 131.7 for C7, 116.2 and 112.2 for C4 and C6, and 111.2 for C3A; the four-bond coupling to atom C7A is too small to be resolved.
Although the regiochemistry of the synthetic reactions can be deduced from the NMR data, it is not possible to establish from these data the relative stereochemistry of all four stereogenic centres, but this is readily achieved by crystal structure analysis. The space groups for compounds (I) and (II) ( Table 1) show that they have both crystallized as racemic mixtures and, for each compound, the reference molecule was selected as that having the R configuration at atom C13 (Figs. 1 and 2); on this basis, the configuration at each of atoms C21, C22 and C27A is S, with these atoms corresponding, respectively, to locants C3, C1 0 , C2 0 and C7a 0 in the chemical numbering scheme, so that the overall configuration of these compounds is (3RS,1 0 SR,2 0 SR,3 0 SR).
Based on earlier work (Pardasani et al., 2003;Quiroga et al., 2017), a reaction sequence can be proposed which commences with nucleophilic addition of the proline component (B) to the isatin (A) (Scheme 1) to form the intermediate (D) (Scheme 2), followed by sequential cyclodehydration to give (E) and decarboxylation to form the key azomethine intermediate (F). This intermediate then undergoes a 1,3-dipolar cycloaddition with the electron-deficient alkene (C) to form the products (I)-(III). The alternative orientation of the alkene relative to the azomethine in the addition reaction would give the products (Ia)-(IIIa) with transposed chlorophenyl and rhodanine units, but these have not been detected. Thus, the negative pole of intermediate (F) has coupled to the heterocyclic end of the alkenic double bond, adjacent to the carbonyl group, rather than to the chlorophenyl end. Neither of the components in the cycloaddition reaction step contains any stereogenic centres, and there are no reagents present which could induce enantioselectivity; hence the products are formed as racemic mixtures. These each contain four contiguous stereogenic centres, so that whichever of these centres is formed first, it appears to exert strong control over the formation of all the others. In the transition state leading to the formation of the products (I)-(III), the reactants can  The molecular structure of the (3R,1 0 S,2 0 S,7a 0 S) enantiomer of compound (I), showing the atom-labelling scheme. Displacement ellipsoids are drawn at the 30% probability level.

Figure 2
The molecular structure of the (3R,1 0 S,2 0 S,7a 0 S) enantiomer of compound (II), showing the atom-labelling scheme. Displacement ellipsoids are drawn at the 30% probability level. approach one another in two orientations: the endo transition state, in which the Cl atom is remote from the aryl ring of the isatin unit, leads to the observed (3RS,1 0 SR,2 0 SR,3 0 SR) stereochemistry, whereas the alternative exo transition state, with the Cl atom close to the aryl ring of the isatin, would lead to the alternative (3RS,1 0 RS,2 0 RS,3 0 RS) stereochemistry, which is not observed. The choice of the transition state in this step is presumably determined by the minimization of steric hindrance. Hence this proposed reaction mechanism can account for both the regiochemistry and for the relative stereochemistry at the four stereogenic centres.
Within the molecules of (I) and (II), the rhodanine rings are almost planar, with r.m.s. deviations from the mean planes of the five ring atoms of 0.0573 Å in (I) and 0.0210 Å in (II). However, the rings containing atoms C22 and C25 (Figs. 1 and 2) both adopt half-chair conformations, as indicated by the ring-puckering parameters (Cremer & Pople, 1975) shown in Table 2. For an idealized half-chair conformation, the value of ' 2 is (36k + 18) , where k represents an integer (Boeyens, 1978). Here the rings denoted A (Table 2) are twisted about a line joining atom C27A to the mid-point of the C13-C22 bond, while the rings denoted B are twisted about a line joining atom N24 to the mid-point of the C26-C27 bond.
Overall, therefore, the composition of compounds (I)-(III) has been determined; the constitutions, including the regiochemistry, have been established from the NMR spectra and the relative configurations of the stereogenic centres, and the conformations of the nonplanar rings in (I) and (II) have been established from the X-ray structure analyses. We have also investigated the crystal structure of compound (III). Despite repeated attempts at crystallization, this compound consistently formed tightly-packed clusters of very thin lath-like crystals, and the resulting diffraction data and the structure deduced from it is of somewhat indifferent quality (see supporting information). After conventional refinement of (III), the resulting difference map contained several significant electron-density maxima, but no chemically plausible solvent model could be developed from these peaks. Accordingly, SQUEEZE (Spek, 2015)  Part of the crystal structure of compound (I), showing the formation of a hydrogen-bonded C(6) chain running parallel to [010]. Hydrogen bonds are drawn as dashed lines and, for the sake of clarity, H atoms bonded to C atoms have all been omitted. Table 2 Ring-puckering parameters (Å , ).
Parameters for rings A and B are calculated for the atom sequences N24-C13-C22-C21-C27A and N24-C25-C26-C27-C27A, respectively. Data for (III) are available in the supporting information. refinement using this modified data set established that the constitution and configuration of (III) are the same as those for compounds (I) and (II), and that the conformations of the type A and B rings are also very similar to those in (I) and (II) ( Table 2). However, the identity of the included solvent species remains undetermined. The supramolecular assembly of both (I) and (II) is very simple. In compound (I), a single N-HÁ Á ÁN hydrogen bond (Table 3) links molecules which are related by a 2 1 screw axis to form simple C(6) chains (Etter, 1990;Etter et al., 1990;Bernstein et al., 1995) running parallel to the [010] direction ( Fig. 3), but there are no direction-specific interactions between adjacent chains. In compound (II), molecules which are related by a 2 1 screw axis are linked by a combination of one N-HÁ Á ÁO hydrogen bond and one C-HÁ Á ÁS C hydrogen bond (Table 3). These two interactions, acting singly, give rise to C(8) and C(12) chains, respectively, while in combination they generate a C(8)C(12)[R 2 2 (11)] chain of rings (Fig. 4). There are no direction-specific interactions between adjacent chains. The crystal structure of (III) contains N-HÁ Á ÁN and N-HÁ Á ÁS hydrogen bonds (Table 3), which individually generate C(6) and C(9) chains, both running parallel to the [010] direction, and in combination these interactions generate a sheet of R 4 4 (24) rings lying parallel to (001) (Fig. 5). A number of structures have been reported for spiro[indoline-3,3 0 -pyrrolizine] derivatives (Sarrafi & Alimohammadi, 2008a,b;Sathya et al., 2012;Fathimunnisa et al., 2015;Corres et al., 2016), but often without any mention of either the relative or the absolute stereochemistry, despite the presence of multiple stereogenic centres.  Part of the crystal structure of compound (III), showing the formation of a hydrogen-bonded sheet of R 4 4 (24) rings; see supporting information for full details. Hydrogen bonds are drawn as dashed line lines and, for the sake of clarity, H atoms bonded to C atoms have all been omitted. Table 3 Hydrogen-bond parameters (Å , ).