Formal synthesis of (−)-agelastatin A: an iron(II)-mediated cyclization strategy

Summary An iron(II)-mediated aminohalogenation of a cyclopentenyl N-tosyloxycarbamate provided new access to the key intermediate for the synthesis of (−)-agelastatin A (AA, 1), a potent antiproliferative alkaloid. The present synthetic endeavour offered an insight into the mechanism underlying the iron(II)-mediated aminohalogenation of N-tosyloxycarbamate, in which the radical properties of the N–iron intermediates in the redox states were operative.

(Scheme 2). Moreover, a plausible mechanism of the present iron(II)-mediated aminohalogenation, which is inferred from the unique reactivity of N-tosyloxycarbamate 8 with the reagents, is discussed.

Results and Discussion
N-Tosyloxycarbamate 8 was prepared from alcohol 6, which was obtained by a previously reported protocol (Scheme 2) [26,27]. Alcohol 6 was first treated with CDI (N,N'-carbonyldiimidazole) and then with hydroxylamine hydrochloric acid salt to afford N-hydroxycarbamate 7 in 67% yield [36]. Thereafter, N-hydroxycarbamate 7 was reacted with TsCl and triethylamine in THF to furnish N-tosyloxycarbamate 8 in 92% yield.
With this carbamate 8, we examined the iron(II)-mediated cyclization under various conditions ( Table 1).
The present study on the aminohalogenation reaction of carbamate 8 has inspired mechanistic insights that deserve discussion (Scheme 4). We hypothesize that cyclized material 5a/5b, reduced material 9, and enone 10 are generated from an N-iron complex (i) that has free-radical character, as previously proposed in the catalytic cyclization of azidoformates [30,[38][39][40]. The contrasting yields obtained from N-tosyloxycarbamate 8 and azidoformate 3 under FeBr 2 /Bu 4 NBr in EtOH conditions (see Table 1, entry 1 versus Scheme 3) likely originate from the distinct chemical property of the N-iron species (i) generated from each substrate. The possible coordination of tosylate anion to the N-iron after the N-O bond cleavage with FeX 2 may have affected the electronic and steric characters of intermediate (i), leading to retardation of the subsequent cyclization. Because of the low cyclization rate, the production of reduced carbamate 9 and enone 10 became pronounced. This is consistent with the observation that the relatively efficient production of cyclized material 5a was observed for azidoformate 3, where N-iron intermediate (i) was free from such interactions. One of the other characteristics found in the present transformations was the incomplete consumption of substrate 8 by lowering FeBr 2 / Bu 4 NBr loading (e.g., Table 1, entry 5), which, in turn, enabled the efficient conversion of structurally simple N-tosyloxycarbamates into the corresponding cyclic aminobromides [28]. This poor conversion under conditions of less FeX 2 /Bu 4 NX loading may be attributable to the decrease of the concentration of reactive FeX 2 through capture with the polar amide functionality of 8. It is speculated that product 9 may be produced by trapping N-iron complex (i) with another FeX 2 (i→vii→9), whereas enone 10 is likely to be generated via intramolecular allylic hydrogen abstraction followed by halogen transfer to regenerate iron(II) species (i→iv→v→10) and/or by directly releasing FeX 2 (i→iv→vi→10) [41]. However, it is worth discussing the process for yielding 9, which theoretically gener-ates two equivalents of iron(III) species per one equivalent of vii. Given the observation that FeCl 3 /Bu 4 NCl gave none of the products shown in Table 1, an iron(III) species possibly generated via the halogen exchange of vii with Bu 4 NX, if any, no longer has catalytic activity and thus the catalytic cycle is terminated. Therefore, active FeX 2 species should somehow be regenerated to maintain the catalysis. One possible pathway that may account for the production of carbamate 9 through the regeneration of FeX 2 species is the intermolecular hydrogen abstraction from substrate 8 by N-iron species (i) (Scheme 5).
The intermediacy of the intermolecular hydrogen abstraction of N-iron species (i) is supported by the fact that the production of 9 was more pronounced in EtOH having a C-H bond α to the oxygen, which likely served as a hydrogen donor ( Table 1, entries 1 and 2). It should be mentioned that reduced material 9 may also be produced by Bu 4 NBr alone as observed in our previous study [28]. To elucidate the contribution of this pathway, compound 8 was treated with Bu 4 NX in t-BuOH. However, no reduced material was obtained within the reaction times depicted in , indicating that the non-iron-mediated process is not significant [42]. Various yields of 9 obtained by loading consistent amounts (1.2-1.5 equiv) of Bu 4 NX salts also indicated the poor contribution of the pathway. Chan and co-workers demonstrated that an iron-imido complex generated from FeCl 2 /PhI=NTs underwent radical hydrogen abstraction from a formyl group, and combined the resultant radicals (hydrogen atom abstraction/radical rebound pathway) to provide amides [43,44]. The involvement of such an iron complex (shown in brackets in Scheme 4) that features radical/ metal-nitrenoid properties can be considered in our reactions. A recent study by Betley and co-workers on high-spin iron-imido complexes generated by the reactions of alkyl azides with FeCl 2 bearing dipyrromethene ligands revealed the radical character of the complex [39,40], harmonizing well with our result, which implies the intermediacy of the nitrogen radical species.

Scheme 5:
Plausible reaction pathway to produce compounds 9 and 10.

Conclusion
We have developed a new approach to key compounds 5a/5b for (−)-agelastatin A (1) synthesis, which features the iron(II)mediated radical cyclization of N-tosyloxycarbamate, a safe azidoformate surrogate. Although somewhat moderate chemical yields of the compounds were obtained in this study, the elimination of hazardous synthetic processes enables the establishment of more robust strategies to access 1. Furthermore, the present study has allowed us to obtain mechanistic insights suggesting that N-iron species (i) has a metal-radical character. Much work is currently being undertaken to comprehend fully the unique properties of the present reactions.

Supporting Information
Supporting Information File 1 Experimental procedures, characterization data of new compounds, and 1 H/ 13 C NMR spectra.