Electrophilic Aromatic Substitution of a BN Indole

Abstract

We report the first examples of a “BN-fused” indole, and we demonstrate that this new family of unnatural indole derivatives undergoes electrophilic aromatic substitution (EAS) reactions with the same regioselectivity as its organic analogue. Competition experiments reveal that N-t-Bu-BN-indole is more nucleophilic in EAS reactions than its carbonaceous counterpart. X-Ray structural analysis between BN indole and classic indole highlights significant differences in bond distances, in particular for bonds associated with the boron atom.

Indole¹ is one of the most ubiquitous heterocyclic motifs in nature. Due to the abundance of biologically active indole derivatives,² the indole ring system has become an important structural component in drug discovery efforts. Consequently, the synthesis and functionalization of indoles has been a major focus in research, the expansion of the chemical space of accessible indole structures being one of the goals.³ An alternative approach to expand structural diversity is “elemental isosterism”. To this end, the BN/CC isosterism has recently emerged as a viable strategy to create biomimetic analogues of common structural units in organic molecules (e.g., olefin,⁴ benzene,⁵ and indene⁶). Despite the recent advances in this area, the elemental isosterism of the biologically important indole has remained virtually unexplored. To date, the only BN-substituted indoles are phenylenediamine-type heterocycles containing an external BN unit as illustrated in 1 (Scheme 1).^7,8 To the best of our knowledge, electrophilic aromatic substitution (EAS), a crucial reaction of the biochemistry of indoles⁹ has not been demonstrated with these phenylenediamine-type BN indoles. Herein we report the first example of a “BN-fused” indole (e.g., heterocycle 2 in Scheme 1), and we demonstrate that this new BN indole undergoes EAS reactions with the same regioselectivity as its organic analogue, N-t-Bu-indole 3.¹⁰

Scheme 1.

A novel BN indole.

Synthesis of N-t-Bu-BN-indole 2 begins with the condensation of N-t-Bu-N′-allylethylenediamine with in situ-generated allylboron dichloride (Scheme 2), affording heterocycle 4 in 50% yield. Ring-closing metathesis (RCM) with Grubbs first generation catalyst provides the bicyclic 5 in 51% yield.¹¹ The yield of the RCM step can be improved to 79% using Schrock’s catalyst. Dehydrogenation of precursor 5 to furnish the target compound 2 is accomplished in the presence of Pd/C in refluxing decane.¹²

Scheme 2.

Synthesis of N-t-Bu-BN-indole 2.

With N-t-Bu-BN-indole 2 in hand, we then investigated its reactivity toward EAS. Indole itself displays high EAS reactivity, which is estimated to be orders of magnitude greater than benzene.¹³ This is due to indole’s electron-rich nature, and the high electron density at its 3-position is responsible for indole’s regioselectivity toward EAS reactions. The bicyclic N-t-Bu-BN-indole 2 consists of a 6-membered 1,2-dihydro-1,2-azaborine heterocycle^14,15 and a 5-membered 2,3-dihydro-1H-1,3,2-diazaborole core (Scheme 3).¹⁶ Thus, we were particularly interested in whether unnatural N-t-Bu-BN-indole 2 would display the classical indole-type regioselectivity in EAS reactions (i.e., at the 3-position) or substitution at the 7- and 5- positions,¹⁷ which have been demonstrated to be most nucleophilic in monocyclic 1,2-azaborine structures.¹⁸

Scheme 3.

Regioselectivity in EAS reactions of N-t-Bu-BN-indole 2.

We surveyed the EAS reaction of N-t-Bu-BN-indole 2 with a variety of electrophiles. As can be seen from Table 1, bromination of 2 occurred at the 3-position, yielding the brominated product 6a (Table 1, entry 1). The Mannich reaction of 2 with dimethyliminium chloride provided our first BN indole alkaloid, N-t-Bu-BN-gramine 6b in 53% yield (Table 1, entry 2). Lewis acid-mediated Michael addition of 2 to cyclohexenone with zirconium tetrachloride¹⁹ furnished 6c in 57% yield (Table 1, entry 3). Deuterium exchange²⁰ occurred in degassed 1:1 CD₃OD/D₂O at 100 °C providing 6d in 39% isolated yield (Table 1, entry 4). Friedel-Crafts acylation of 2 with acetyl chloride was accomplished in the presence of Et₂AlCl²¹ to yield 6e (Table 1, entry 5). In each case, we observed substitution at the 3-position of 2. The H(2) and H(3) proton signals of 2 appear as one singlet in ¹H NMR (in CD₂Cl₂) integrating for two protons. Thus, we determined the substitution pattern of EAS products by either single crystal X-ray diffraction (e.g., for 6e), or NOESY experiments (e.g, for 6a and 6c).²²

Table 1.

EAS Reactions of 2

entry	electrophile (E+)	catalyst	3-substituent (E)	product	yield (%)
1	Br2	-		6a	39
2		-		6b	53
3		ZrCl4		6c	57
4	CD3OD/D2O	-		6d	39a
5		Et2AlCl		6e	23

^a1H NMR indicates ~80% deuterium enrichment.

The modest yields of these reactions are likely due to side reactions associated with the relatively acidic reaction conditions. Indoles are known to undergo self-condensation reactions in the presence of acids.²³ We determined that N-t-Bu-BN-indole 2 forms the B-methoxy-substituted trimer 7 (Figure 1) in acidic methanol solution among other byproducts.²⁴ In one instance, we were able to isolate a single crystal of 7 and determine its structure by X-ray crystallography.^25,26 Although substituted at the nitrogen and at the C(3) position, the structure of 7 provides us with the first glimpse into the bonding of BN indoles fused at the BN unit. Comparing the intra-ring bond distances of the 6-membered ring of 7 and a B-diphenylamino-substituted 1,2-azaborine A (Figure 1),²⁷ the most noticeable difference is observed in the B–N bond distances. In the fused bicyclic BN indole 7 the B(1)–N(2) distance (1.463(2) Å) is longer than the corresponding B–N bond in a monocyclic 1,2-azaborine (1.446(2) Å). This is likely due to the better π-overlap between the “exocyclic” nitrogen atom in 7 (i.e., N(1)) versus in A. It is also consistent with the shorter N(1)–B(1) distance in 7 (1.440(2)Å) compared to the B–N(exocyclic) bond length of 1.486(2)Å in A. We are also interested in comparing the BN indole structure 7 with its carbonaceous counterpart. The typical bond distances of a 1,3-dialkylsubstituted indole are illustrated in the bottom right corner of Figure 1 (structure B).²⁸ The most significant differences in bond distances are directly associated with the replacement of the C=C bond in B with a B–N unit. Due to the larger covalent radius of boron,²⁹ the B(1)–C(7) (1.509(2) Å) and B(1)–N(1) (1.440(2) Å) bond lengths are significantly longer (by ~0.1 Å) than those in indole B. On the other hand, presumably due to the slightly smaller covalent radius of the nitrogen (vs. carbon),²⁹ the N(2)–C(4) (1.370(2) Å) and N(2)–C(3) (1.404(2) Å) distances are shorter than the corresponding bonds in B, although this difference is less pronounced (i.e., shorter by only ~0.02 Å). Thus, while BN indoles are geometrically similar in shape compared to natural indoles, the differences in bonding (e.g., bond lengths and electronic structure) could potentially lead to significantly different biological activity.

Figure 1.

ORTEP illustrations, with thermal ellipsoids drawn at the 35% probability level, of 7 and 6 e.

We were also very pleased to discover that acylated BN indole 6e forms crystalline solids suitable for single crystal X-ray diffraction analysis. The structure of 6e is illustrated in Figure 1 (top right). Because of the electronic impact of the acetyl substituent at the C(3) position, significant changes in bond distances are observed compared to 7. The N(1)–C(2) (1.363(2)Å), C(2)–C(3) (1.372(2) Å), and the B(1)–N(2) bond distances in 6e are consistent with a significant contribution of structure 6e’.

In order to probe the effect of BN/CC isosterism of indoles on the reactivity in EAS reactions, we performed competition experiments between N-t-Bu-BN-indole 2 and its corresponding carbonaceous analogue 3. Protection of the pyrrolic nitrogen with an N-t-butyl group allows the reactivity of the 3-position to be evaluated without the influence of an indole N-H. Using one equivalent of each indole 2 and 3 and 0.5 equivalent of electrophile (E⁺ = dimethyliminium chloride) in CD₂Cl₂, we observed only EAS products associated with N-t-Bu-BN-indole 2, with indole 3 remaining intact (Scheme 4).³⁰ We hypothesize that N-t-Bu-BN-indole 2 exhibits greater enamine character in the pyrrolic ring and is therefore more nucleophilic than indole 3.

Scheme 4.

EAS competition experiment with dimethyliminium chloride as the electrophile.

In summary, we synthesized the first examples of BN-fused indole derivatives. We also demonstrated that this new family of BN indoles undergoes EAS reaction with classical indole-type regioselectivity at the 3-position. Competition experiments revealed that N-t-Bu-BN-indole 2 is more nucleophilic in EAS reactions than its carbonaceous counterpart. Single crystal X-ray structure analysis showed that while BN indoles are similar in shape compared to classical indoles, significant differences in bond distances, in particular those associated with the boron atom, are observed. Our work lays the synthetic foundation for BN-substituted unnatural products containing the indole motif and highlights the potential of BN/CC isosterism as a general strategy in expanding the chemical space of biologically active molecules.