An intramolecular inverse electron demand Diels–Alder approach to annulated α-carbolines

Intramolecular inverse electron demand cycloadditions of isatin-derived 1,2,4-triazines with acetylenic dienophiles tethered by amidations or transesterifications proceed in excellent yields to produce lactam- or lactone-fused α-carbolines. Beginning with various isatins and alkynyl dienophiles, a pilot-scale library of eighty-eight α-carbolines was prepared by using this robust methodology for biological evaluation.

Given their isomeric relationship to β-carbolines, α-carbolines have, unsurprisingly, attracted attention from synthetic chemists for a long time [15], and more recently from medicinal chemists [16]. Synthetic members of this class were shown to have a wide range of activities, including topoisomerase II inhibition [17], and 4-amino-α-carbolines have also been shown to possess anxiolytic properties by stabilization of the open chloride channel [18,19]. Relatively few patents have been granted on the medicinal use of α-carbolines, particularly in comparison to β-carbolines, with recent applications including use as antiviral agents [20], inhibitors of ApoB-100-associated lipoprotein production for cholesterol lowering [21], and more recently, as inhibitors of CDK1 kinase as potential anticancer agents [22]. This later filing has triggered investigations into α-carbolines as potential multikinase inhibitors [23].
The first reported synthesis of an α-carboline by Robinson in 1924 [15,36] proceeded through the acid-catalyzed decomposition of 1-(2-pyridyl)benzotriazole, a modification of the Graebe-Ullmann carbazole synthesis, which closes the indole ring. This procedure was improved upon and exploited for decades [37][38][39][40]. Later, reversing the roles of the benzene and pyridine rings, with the former as the nucleophile and latter as electrophile through the diazonium salt, led to improved yields in the indole ring closure [41], as did microwave promotion of the original methodology [42]. Nitrene insertion chemistry of appropriately substituted 3-arylpyridines also found application [43,44], and likewise falls into the category of indole ring closure onto an existing substituted pyridine, though in this case with the formation of the C-N bond. More recently, the group of Cuny has reported two strategies that exploit cross-couplings to prepare anilinopyridines, with final indole ring closure occurring by either a second cross-coupling [45] or a photocyclization [46]. The former, sequential cross-coupling strategy closely follows the previous work by Queguiner [47]. A Fischer indole synthesis route through the 2-pyridylhydrazone of cyclo-hexanone, catalyzed by PPA, followed by dehydrogenation over Pd-C has also been reported for the preparation of the unsubstituted α-carboline [48], but no other successful applications of this strategy have appeared in the literature.
Though the existing synthetic routes can readily produce various α-carbolines, the intramolecular IEDDA approach outlined in Scheme 1 has several advantages from the perspec-Scheme 1: Retrosynthetic inverse electron Diels-Alder approach to α-carbolines. tive of library synthesis. First, easy access exists to a wide selection of commercially available or easily synthesized starting materials, i.e., isatins, propargylic and homopropargylic amines and alcohols. Second, three easily modifiable diversification sites, R 1 , R 2 and R 4 , can be built into the core α-carboline structure. Furthermore, construction of the annulated lactam or lactone as the fourth ring would be easily accomplished by the IEDDA reactions, thereby adding two more diversification points in the form of R 3 and the lactam/ lactone ring size. A particular goal in this work was to establish the reaction chemistry and scope in order to build a library of α-carbolines for biological screening. The targeted α-carbolines 6 (X = O) are similar to those prepared by Dodd by a rather lengthy route, but with a transposed carbonyl group on the lactone ring [59].
One consideration in this design was the electron donation from the indole nitrogen into the isatin-derived triazine ring of 9, which would result in an elevated LUMO of the triazinyl azadiene, and thereby inhibit the desired cycloaddition. Thus, it was anticipated that an electron-withdrawing group R 2 , which could also serve as a diversification point, would be needed on the indole nitrogen.

Results and Discussion
Feasibility studies began with isatin-derived 1,2,4-triazine 9a (9, R 1 = H) [73,74], which was easily prepared by the condensation of isatin (10, R 1 = H) with ethyl oxaloamidrazonate [75][76][77] (11) in quantitative yield (Scheme 2). The first step in the cyclocondensation was accomplished by stirring in ethanol at rt for 12 h and heating under reflux for 20 min, after which the solvent was removed and the cyclocondensation completed by heating under reflux in bromobenzene (bp 156 °C) for 24 h. The two-step condensation with different solvents was needed to optimize the triazine formation. Sulfonylation of the indole nitrogen also proceeded routinely to give triazine 8a (R 1 = H, R 2 = p-Tol), and served two purposes. As noted, the reduction of electron donation from this nitrogen into the triazine ring was thought to be important for the subsequent cycloaddition to proceed, as was shown to be correct in later studies. Further- Table 1: Preparation of N-protected isatin-derived 1,2,4-triazines 8. a a Isolated yield over two steps (Scheme 2).
more, sulfonylation greatly improved the solubility of the triazines 8 in organic solvents in comparison to 9, which showed only limited solubility in dichloromethane, chloroform, THF, toluene, methanol and acetone. Starting with other 5-substituted isatin derivatives, analogous triazines were similarly prepared in good yields (84-92%, Table 1). 5-Nitro-and 5-carboxamidoisatins also readily participated in the cyclocondensation with the oxaloamidrazonate 11 to form the corresponding triazines (9, R 1 = NO 2 , CONH 2 ), but due to the electron withdrawing nature of the isatin substituents, subsequent sulfonylations were not successful (not shown).
Lewis acid catalyzed amidation of 8a with methyl propargyl amine (12) gave the cycloaddition precursor 13a in excellent yield when Al(Me) 3 was used as catalyst (Scheme 3, Table 2) [81][82][83][84]. However, the most convenient procedure for library protocols employed 1.2 equiv of Zr(Ot-Bu) 4 [85]. Weinreb amidation with Al(Me) 3 also gave excellent yields, but required the extra step of first mixing Al(Me) 3 and the amine, then cannulating this amine-AlMe 3 complex into the triazine solution in order to avoid ketonization of the ester [86]. Other Lewis acids, MgCl 2 , Mg(OTf) 2 , Zn(OTf) 2 , Yb(OTf) 3 and Sc(OTf) 3 , were not successful in catalyzing the amidation. Stoichiometric amounts of catalyst were required for the amidation, presumably due to the product itself sequestering the catalyst and preventing efficient turnover. The intramolecular cycloaddition of 13a, the alkyne-tethered triazine, was studied under various conditions (Scheme 4, Table 3). Ultimately, it was found that the IEDDA reaction proceeded smoothly under microwave irradiation, in diglyme (120 °C, 20 min; Table 3, entry 3) to give the γ-lactam annulated α-carboline 14a in quantitative yield. The microwave reaction conditions were preferred over the more traditional heating ( Table 3, entry 1) due to the shorter reaction time. Attempts to lower the temperature and/or shorten the reaction time led to lower yields (Table 3, entries 2, 4 and 5). Little reaction occurred in toluene under microwave irradiation (Table 3, entry 6) unless silicon carbide chips were added as a microwave facilitator (Table 3, entry 7) [87].   4 in the one-pot two-step sequence had no effect on the cycloaddition of 13a. For the library synthesis, the two-step sequence was adopted in order to isolate the triazine intermediates 13, which can also serve as library members.
With the two-step sequence optimized for the preparation of 14a, the scope of this chemistry was then probed with other alkynyl amines. The amidations all proceeded in high yields under the optimized conditions with Zr(Ot-Bu) 4 as catalyst. The cycloaddition precursors were then subjected to the optimized microwave-promoted cycloadditions to give the final cycloadducts 14 (Scheme 6 and Table 4).
In contrast to the tertiary amides, the secondary amides 13b and 13c (Table 4, entry 2 and 3) showed no or very little reaction under the standard microwave conditions. The reason for this lack of reactivity was thought to be strong intramolecular hydrogen bonding, which inhibits the rotation of the alkyne group to the proper position for the cycloaddition to occur (Scheme 7).
The NMR spectra also supported this hypothesis: While all tertiary amides showed the presence of two rotamers in a 1:1 ratio in CDCl 3 , the secondary amides 13b and 13c showed only a single rotamer in the NMR spectra. In the 1 H NMR spectrum for 13b, a very slowly exchanging proton (48 h for complete exchange) at δ 8.254 supported the presence of a single conformation with strong intramolecular hydrogen bonding. When the cycloaddition of 13b was run in DMF instead of diglyme (160 °C, 20 min), the cycloadditions proceeded in good yield to give the desired cycloadduct 14b (83%; Table 4, entry 3).
Presumably the significantly greater capabilities of DMF to accept hydrogen bonds in comparison to diglyme help to disrupt the intramolecular hydrogen bonding and enable the cycloaddition to proceed. When 13c, with the gem dimethyl substituted propargylic carbon, was run in diglyme at a higher temperature (160 °C, 30 min), the cycloaddition proceeded in good yield to give the desired product 14c (91%; Table 4, entry 3), presumably aided by the Thorpe-Ingold effect [88][89][90].
To confirm the importance of the electron-withdrawing group on the original indole nitrogen position, the reactivity of amide 9a (R 1 = H) was also examined (Scheme 8). As expected, no cycloaddition was observed under any conditions, presumably due to the greater electron donation from the indole nitrogen, which elevates the LUMO of the triazine ring, thereby preventing the desired cycloaddition.
In addition to the amide linkage of the alkyne dienophiles, transesterification of 8a with alkynyl alcohols led to tethered alkynyl esters as cycloaddition precursors 17a and 17b in good yields (Scheme 9). Various catalysts for the transesterification reaction were screened. Boronic acid [91] and indium(III) iodide [92] yielded no transesterification product with propargyl alcohol, while Ti(OiPr) 4 [93] gave only a trace of the desired product, with mostly detosylation resulting. Otera's catalyst [94] proved to be optimal, giving cycloaddition precursors 17a and 17b in excellent yields (84% and 87%, respectively).
Cycloaddition precursors 17a and 17b showed a much lower reactivity in the cycloadditions in comparison to the amides 13.
greater entropy loss for the larger rings [99]. Indeed, it has been estimated that the effective molarity for 5-membered ring closures can be as high as 1,000-fold greater in comparison to 6-membered ring cyclizations [100].
Using this optimized two-step amidation-IEDDA reactionsequence methodology, an 88-membered pilot-scale library of α-carbolines was prepared by using triazine analogues 8 and alkynyl amines ( Table 5). The crude library was analyzed by UPLC. The average yield of the library was 94%, with 90% of the library members produced in yields greater than 85%. Purification by LC-MS produced the final library.

Conclusion
Isatin-derived 1,2,4-triazines have proven to be excellent heteroaromatic azadienes for intramolecular inverse electron demand Diels-Alder reactions with tethered alkynyl dienophiles. These cycloadditions led to lactam-annulated α-carbolines in excellent yields under microwave assistance. The scope of the chemistry was probed by using various alkynyl amines and alkynyl alcohols, and by variation of the tether length between the aminoalkyne and the triazine. The triazines with ester linkages showed significantly less reactivity in cycloadditions compared to those with amide linkages. The longer tether length also led to a decrease in cycloaddition reactivity. This chemistry was subsequently applied to a library synthesis, producing a focused library of eighty-eight members. Diversity was introduced by using a combination of various substituted isatin-derived triazines, with various sulfonylations of the indole nitrogen, and propargyl amine derivatives as dienophiles.

Experimental
General procedure A, preparation of isatin-derived triazines 9: Freshly prepared ethyl oxalamidrazonate (11) [76] was dissolved in anhydrous EtOH (0.1 M) and the isatin (1.0 equiv) was added at rt under stirring. The reaction mixture was stirred at rt for 12 h, and then heated under reflux for 20 min. After removal of the EtOH in vacuo, the residue was dissolved in anhydrous bromobenzene (0.2 M) and refluxed for 24 h. After removal of the solvent in vacuo, the residue was dried by the addition and evaporation of toluene three times, and used directly in the next step without any further purification.
General procedure B, preparation of sulfonamides 8: The isatin-derived triazine 9 was suspended in THF (0.25 M) and triethylamine (2.0 equiv) was added to the solution at rt. The reaction mixture was stirred at rt for 30 min until dissolution was completed, then the sulfonyl chloride (2.0 equiv) was added at rt, and the reaction mixture was stirred at rt for 7 h. After removal of the solvent in vacuo, the residue was purified by flash chromatography to yield the desired sulfonylated triazine 8. General procedure D, cycloaddition of 13 to 14: A solution of the amide 13 in diglyme (0.1 M) was placed in a thick-walled microwave tube, and then the reaction mixture was subjected to microwave irradiation at 160 °C for 20 min under stirring, unless otherwise noted. After the irradiation, the solvent was removed in vacuo and the residue was purified by flash chromatography to yield the cycloadducts 14.