Counterion Control of t-BuO-Mediated Single Electron Transfer to Nitrostilbenes Constructs N-Hydroxyindoles or Oxindoles

Abstract

tert-Butoxide unlocks new reactivity patterns embedded in nitroarenes. Exposure of nitrostilbenes to sodium tert-butoxide was found to produce N-hydroxyindoles at room temperature without an additive. Changing the counterion to potassium changed the reaction outcome to yield solely oxindoles through an unprecedented dioxygen-transfer reaction followed by a 1,2-phenyl migration. Mechanistic experiments established that these reactions proceed via radical intermediates and suggest that counterion coordination controls whether an oxindole or N-hydroxyindole product is formed.

Graphical Abstract

tert-Butoxide unlocks divergent reactivity embedded in nitrostilbenes to construct N-hydroxyindoles or oxindoles depending on whether sodium or potassium is the counterion.

Introduction

The ubiquitous nature of N-heterocycles in pharmaceuticals and organic materials has inspired the discovery of new reactivity patterns to facilitate their construction.^[1],[2] The development of reductive methods to access these privileged scaffolds by constructing C–N bonds using nitroarenes has received significant attention because of the ready availability and robust nature of nitroarenes. While traditional C–N bond formation methods were developed using a superstoichiometric quantity of a reductant to deoxygenate the nitro-group,^[3] recent efforts have focused on the development of catalytic processes that use the combination of a transition metal-^[4] a base-metal-^[5] or a phosphine^[6] catalyst and a stoichiometric reductant. In contrast to these approaches, 175 years ago,^[7] sodium methoxide was reported to reduce nitrobenzene to azoxybenzene in boiling methanol (Scheme 1). Subsequent mechanistic investigations in 1962 by Ogata and Mibae suggested that electron-transfer from alkoxide triggered deoxygenation to produce an ArNO intermediate which reacted with an N-hydroxybenzene to produce the azoxy product.^[8] After these mechanistic studies, interest in exploiting electron-transfer from alkoxides waned, until a recent resurgence in the use of potassium tert-butoxide as a single-electron reductant.^[9] In 2010, the Hayashi and Shi groups reported that tert-butoxide mediated the transition metal-free couplings of iodoarenes.^{[9a, 9b]} Investigations by Murphy and co-workers established that electron-transfer to the iodoarene occurred from the in situ-generated phenanthroline dianion 1.^[9d] Inspired by these reports, we were curious if exposure of a nitrostilbene to an alkoxide would generate a nitrosoarene intermediate that might be intercepted intramolecularly to afford an N-heterocycle (Scheme 1). In contrast to our expectations, we found that nitrostilbene 2 could be converted to N-hydroxyindole 5 using simply NaOt-Bu at room temperature without an additive. Strikingly, changing the counterion to potassium triggered an unprecedented dioxygen transfer and [1,2] aryl migration tandem reaction to afford oxindole 6 as the only N-heterocycle from nitrostilbene 2.

Scheme 1.

tert-Butoxide-mediated N-heterocycle formation from nitroarenes.

Results and Discussion

Investigation of the scope and limitations of N-heterocycle formation.

To test our hypothesis that nitrosoarene intermediates could be accessed using an alkoxide and trapped to afford an N-heterocycle, the reactivity of nitrostilbene 2a was examined (Table 1). While no reaction was observed upon exposure of 2a to sodium methoxide in boiling methanol, changing the solvent to tetrahydrofuran resulted in the formation of N-hydroxyindole 5a at room temperature (entries 1 and 2). The yield of 5a was improved by changing the identity of the alkoxide with the best outcome obtained when tert-butoxide was used (entries 2 – 4). Increasing the temperature or the reaction time, however, did not further increase the yield (entries 5 and 6). The reaction was not light-dependent:^[10] performing the reaction in a foil-covered vessel provided N-hydroxyindole 5a in 55% yield (entry 7). A solvent screen revealed that N-hydroxyindole formation occurred smoothly in ethereal solvents, but was attenuated in tert-butanol (entries 7 – 10).^[11] To our surprise, the identity of the counterion was critical to the reaction outcome: while no reaction was observed when lithium- or magnesium tert-butoxide was employed, 3-phenyl-3-hydroxy-2-oxindole 6a was obtained as the only product using potassium tert-butoxide irrespective of exposure to light (entries 11 – 13). The formation of 6a, whose structure confirmed by X-ray crystallography,^[12] involves not only oxygen transfer to the ortho-alkenyl substituent but also a [1,2]-phenyl shift. While oxygen transfer from nitroarenes to pendant acetylenes has been reported to afford N-heterocycles,^[13] this oxidation-migration tandem reaction of ortho-alkenyl substituents is unprecedented.

Table 1.

Development of reductive and divergent N-heterocycle formation.

entry	MOR (2 equiv)	solvent	h	T (°C)	yield, %a	5a:6a
1	NaOMe	MeOH	16	70	n.r.	...
2	NaOMe	THF	16	25	14	>20:1
3	NaOEt	THF	16	25	26	>20:1
4	NaOt-Bu	THF	16	25	55	>20:1
5	NaOt-Bu	THF	16	60	57	>20:1
6	NaOt-Bu	THF	40	25	60	>20:1
7b	NaOt-Bu	THF	40	25	55	>20:1
8	NaOt-Bu	dioxane	40	25	53	>20:1
9	NaOt-Bu	2-MeTHF	40	25	55	>20:1
10	NaOt-Bu	t-BuOH	40	25	38	>20:1
11	LiOt-Bu	THF	16	25	0	...
12	Mg(Ot-Bu)2	THF	16	25	0	...
13c,d	KOt-Bu	THF	16	25	62	<1:20

^aIsolated after silica gel chromatography.

^bReaction performed in a foil-covered vessel.

^cThe outcome of the reaction did not change if light was excluded.

^dThe X-Ray structure of 6a was deposited in the Cambridge Crystallographic Database (CCDC 198003).

THF = tetrahydrofuran.

The scope and limitations of the NaOt-Bu-mediated reductive cyclization was explored by developing conditions to alkylate the N-hydroxyl group as well examining the effect of changing the electronic- and steric environment of the nitrostilbenes (Table 2). The substrates for this study were either commercially available or readily prepared in one-step from 2-bromonitroarenes and styrene using a Heck reaction.^[14] Because N-alkoxy- or N-acetoxyheterocycles are more stable,^[15] we explored telescoping the reaction by adding an electrophile to determine if the sensitive N-hydroxy functionality could be alkylated or acetylated. We found that a methyl-, benzyl-, or benzoate group could be easily added to the initially formed N-hydroxyindole to improve the reproducibility of the reaction sequence and ease purification. Our investigation of a series of nitrostilbenes that varied their electronic- and steric nature revealed that higher yields were obtained in the presence of electron-donating groups positioned para to the styryl group (entries 2 and 5). In contrast, the presence of electron-withdrawing groups severely attenuated the reaction yield. Gratifyingly, the yield of N-hydroxyindole could be rescued if the reaction was heated to 100 °C (entries 4 and 7). Changing the electronic nature of the β-aryl group had less of an overt effect on the reaction outcome: while higher yields were observed for electron-rich aryl groups, the presence of a fluorine did not diminish the yield as substitution on the nitroarene moiety (entries 8 – 11). While adding a stronger electron-withdrawing Cl did reduce the yield, increasing the temperature of the reaction increased the yield of 5o. The reactivity of naphthalene-derived 2q illustrated that N-hydroxyindole formation was relatively insensitive to steric constraints (entry 12).

Table 2.

Scope of NaOt-Bu-mediated N-hydroxindole formation.

entry	#	nitroarene 2	N-heterocycle 5	R3	yield, %a
1	a			H	60
				Me	67 (63)b
				Bn	60
				Bz	59
2	b			H	58
				Me	62
3	c			H	53
				Me	42
4	e			H	24 (38)c
				Me	22 (44)c
5	g			Me	63
6	h			Me	44
7	i			H	25 (60)c
				Me	17 (63)c
8	l			H	46
				Me	60
9	m			H	68
				Me	75
10	n			H	67
				Me	59
11	o			H	25 (53)c
				Me	28 (55)c
12	q			Me	52

^aIsolated after silica gel chromatography.

^bReaction performed on 2 mmol scale.

^cThe tert-BuO-mediated reduction was performed at 100 °C for 16 h.

After our initial investigation into the scope of using NaOt-Bu as the reductant, we turned our attention to exploring the reactivity of nitrostilbenes towards KOt-Bu (Table 3). Analysis of the reactivity patterns revealed several differences in comparison to N-hydroxyindole formation. First, the success of the oxygen-transfer-migration reaction was less dependent on the electronic nature of the nitrostilbene with chloro-, bromo- and even trifluoromethyl groups tolerated (entries 2 – 10). Further, the opposite electronic trend was observed for β-aryl groups: more electron-deficient groups led to higher oxindole yields (entries 11 – 15). Naphthalene 2q demonstrated that oxindole formation tolerated an increased steric environment around the nitro-group to afford 6q albeit in diminished yield relative to 6a (entry 16). In contrast to N-hydroxyindole formation, the yield was not improved when the reaction temperature was increased.

Table 3.

Scope of KOt-Bu-mediated oxindole formation.

entry	#	R1	R2	Ar	yield, %a
1	a	H	H	Ph	62 (55)b
2	b	MeO	H	Ph	43
3	c	F	H	Ph	43
4	d	Cl	H	Ph	48
5	e	F3CO	H	Ph	57
6	f	F3C	H	Ph	38
7	g	O–CH2–O		Ph	41
8	h	H	MeO	Ph	49
9	j	H	F	Ph	74
10	k	H	Br	Ph	49
11	l	H	H	4-MeOC6H4	41
12	m	H	H	p-Tol	38
13	n	H	H	4-FC6H4	56
14	o	H	H	4-ClC6H4	50
15	p	H	H	4-CF3C6H4	66
16	q				41

^aIsolated after silica gel chromatography.

^bReaction performed on 2 mmol scale.

To assess the impact of the ortho-styryl substituent on the reaction outcome, substrates 2s, Z-2a and 2r were investigated (Scheme 2). The position of the phenyl substituent was found to control which N-heterocycle was formed: exposure of α-phenylnitrostyrene 2s to tert-butoxide afforded only N-methoxyindole 5sa using either KOt-Bu or NaOt-Bu after telescoping the reaction with MeI, although an increased reaction temperature was required using NaOt-Bu. The stereochemistry of the 2-nitrostilbene was also critical to the reaction outcome. While E-nitrostilbene formed either N-hydroxyindole 5a or oxindole 6a depending on identity of the tert-butoxide counterion, the reaction of Z-2a afforded only oxindole 6a irrespective of whether sodium- or potassium tert-butoxide was used. Similar reactivity was observed when nitrostilbene 2r was exposed to NaOt-Bu, which produced only oxindole 6r where one of the oxygen atoms was transferred to the ortho-alkenyl substituent.^[16] These results suggest that steric interactions between the β-phenyl substituent and a reactive intermediate or stabilization of a radical- or charged intermediate by the α-phenyl substituent can override the counterion effect.

Scheme 2.

Effect of phenyl substituent position on the reaction outcome.

Mechanistic EPR investigations.

To investigate if the tert-butoxide-mediated N-heterocycle formation involved the formation of radical intermediates, the reaction was analyzed using X-band, solution-phase EPR spectroscopy (Figures 1 and 2). At room temperature, reaction mixtures from addition of KOtBu and NaOtBu exhibit EPR spectra near g = 2.00. The additional most-common feature of the spectra is the existence of a three-line pattern, typical for a radical interacting with a ¹⁴N (I = 1) nucleus. We attribute these peaks to radical intermediates of an electron transfer from tert-butoxide (E_ox = +0.10 V vs SCE)^[17] to nitrostilbene (E_red = −1.13 V vs SCE).^{[4g, 18]} The high potential gap of ΔE = 1.03 V between the oxidation peak potential of t-BuO⁻ and the reduction peak potential of nitrostilbene suggests that coordination of the counterion occurs to facilitate the transfer.^[19] This hypothesis is supported by the simulation of the X-band EPR spectrum from the NaOt-Bu-mediated reaction, which afforded the best match with experiment when a nitro radical anion was assumed coordinated to the sodium counterion (Figure 1a).^[20]

Figure 1.

Experimental (black) and simulated (red) X-band EPR spectra of tert-butoxide-mediated reduction of nitrostilbene 2a (THF, 298 K, 1 h) (a, result of NaOt-Bu reaction, b, result of KOt-Bu reaction) Data collection parameters: Frequency = 9.4306 GHz, Power = 0.2 mW, Modulation = 0.1 G; a. Fitting parameters for 7 are g_iso = 2.005, A_iso = 29 MHz, and an line width for isotropic broadening, lw =[0.81 0.53] in mT (the first Gaussian and the latter Lorentzian broadening); b. Fitting parameters for 8 are g_iso = 2.009, A = [22.1 18.2] MHz, lw = 1.09 mT, τ_corr = 5 ns, and weight = 0.85; and fitting parameters for 9 are g_iso = [2.007], A = [21.2 14.7 11.9 8.7], and lw = 0.2 mT, relative contributions to spectral intensity are 93% 8 and 7% 9.

The foregoing outcome suggests that counterion identity is critical to directing the radical formation pathway. For example, the EPR spectrum obtained from the NaOt-Bu-mediated reaction displays only three well-resolved peaks and no further couplings to ¹H in Figure 1a. The absence of additional peaks from the latter couplings strongly suggests that the spin density is localized only in the nitro group, not delocalized onto the aromatic ring. We propose that the small size as well as strong Lewis acidity of Na⁺ allows for effective binding to the nitro group and prevention of spin density delocalizing on the aryl ring. In contrast, a spectrum displaying significantly more complicated hyperfine interactions was observed when KOt-Bu was employed (Figure 1b).^[21],[22] Simulation of the spectra with a variety of possible radicals revealed a best match when assuming a mixture of 93% of 8 and 7% of 9. We interpret these data to suggest that the nitro radical is less effective in coordinating the larger potassium counterion. As a result of the weaker coordination, the formation of two spin isomers is allowed, one featuring an NO₂ radical and one with spin density delocalized onto the aryl ring. We note that this preliminary interpretation requires that the radical is not delocalized across both moieties. This outcome could occur if interactions with K⁺ drive the nitro-groups away from planarity with the stilbene.^[23] Extension of this interpretation could explain the less complex EPR spectrum for the Na⁺ system: stronger binding of the nitro radical anion to Na⁺ produces only one, NO₂-localized isomer. We tentatively interpret the difference in coordination in these simulated spectra to account for the difference in reaction outcome: the lack of coordination to the counterion could enable oxygen-atom transfer from 8 to the ortho-styryl substituent to afford oxindole 6a, which would not be possible when it is bound to the sodium ion.^[24]

These EPR experiments spurred us to examine the effect of a crown ether on the reaction outcome (Scheme 3). When 15-crown-5 was added in combination with NaOt-Bu, we found that N-hydroxyindole 5a was no longer formed, and oxindole 6a was produced as the only N-heterocyclic product. This phenomenon proved to be general: exposure of 2-nitrostilbenes 2c, 2g, or 2l produced only oxindoles 6c, 6g, or 6l albeit in lower yield than using KOt-Bu.^[25] In contrast, the addition of 18-crown-6 to the KOt-Bu-mediated reaction resulted in no change to the outcome—oxindole 6a was the only N-heterocycle formed. Analysis of the reaction using EPR spectroscopy reveals a substantial change in spectral shape when the crown ether is present, potentially suggesting changes in spin identity via crown-ether binding to the counterion.^[26] Together, these experiments suggest that coordination of the counterion controls whether N-hydroxyindole 5 or oxindole 6 is produced.

Scheme 3.

Effect of 15-crown-5 on the reaction outcome.

Reactivity of ¹⁸O-labeled reagents.

To determine if the C2- and C3 oxygens in oxindole 6 originated from an intra- or intermolecular reaction, the reactivity of ¹⁸O-labeled potassium tert-butoxide and ¹⁸O-labeled nitrostilbene 2g was investigated (Scheme 4). When K¹⁸Ot-Bu was used, no ¹⁸O-incorporation into 6a was observed to suggest that both oxygens were transferred from the nitro-group. To test if oxygen transfer occurred intermolecularly, a mixture of 2a and 2g-¹⁸O were submitted to reaction conditions and double crossover of the labeled- and unlabeled oxygens to the oxindole product was observed to afford 16% of 6a and 6a-¹⁸O and 64% of 6g-¹⁸O₂ and 6g-¹⁸O. Analysis of the mass spectrum showed an increase of the [M + 2]⁺ signal for 6a and [M − 2]⁺ signal for 6g-¹⁸O to suggest that only one of the oxygen atoms is transferred, and the parent ion [(M) − OH]⁺ revealed that intermolecular O-transfer occurred only to the C2-position of the oxindole.

Scheme 4.

Investigation of an intra- or intermolecular mechanism for oxygen transfer.

Potential mechanisms for N-heterocycle formation.

Our data suggests a possible mechanism that accounts for the dependence of the reaction outcome on the identity of the counterion (Scheme 5). Electron transfer from tert-butoxide to nitrostilbene produces radical anion 8 and tert-butoxy radical,^[27] which fragments to produce acetone and methyl radical. Mesomer 10 traps sodium ion to produce 7, which accepts a second electron to form 11. Fragmentation of 11 produces nitrosostilbene 12, which undergoes a 6π-electron-five atom electrocyclization to form 13 that isomerizes to produce N-hydroxyindole 5a.^[28]

Scheme 5.

Potential mechanism for N-hydroxyindole formation.

Our data suggests that oxindole 6 is formed through an intermolecular mechanism when the counterion is not coordinated to the nitrostilbene radical anion (Scheme 6).^[29] The intermolecular oxygen-atom transfer could occur by attack of radical anion 8 onto the nitrostilbene 2 at the β-position to afford nitrite 14,^[30] which undergoes a 3-exo-tet radical cyclization to produce epoxide 15.^{[31],[32],[33]} Ring-opening by the proximal nitro group could produce 16.^{[34],[35],[36]} tert-Butoxide-mediated single electron reduction followed by fragmentation produces 18. Hydrogen-atom abstraction by tert-butoxy radical or methyl radical produces ketone 19.^[37],[38] Isomerization forms enol 20,^[39] which attacks the nitroso group to produce N-heterocycle 21. A subsequent 1,2 phenyl shift affords N-hydroxyoxindole 22,^[40] which is reduced to oxindole 6a. To test if oxygen-transfer occurred via an epoxide intermediate, 2-nitrostilbene oxide was examined as a substrate. In line with our mechanistic hypothesis, treatment of 15 with KOt-Bu resulted in the formation of oxindole 6a. Conversion of 2-nitrostilbene oxide to oxindole supports that the intermolecular reaction occurs between a nitroarene radical anion and a neutral nitroarene.

Scheme 6.

Potential intermolecular mechanism for oxindole formation.

Conclusion

In conclusion, we have discovered a novel tert-butoxide-mediated reaction of 2-nitrostilbenes that produces either a N-hydroxyindole or an oxindole depending on the identity of the counterion. The reactivity patterns exhibited by the 2-nitrostilbenes suggest that the N-heterocyclic products are formed by either an intra- or intermolecular reaction, where the degree of coordination of the counterion to the radical anion dictate which mechanism occurs. Our findings illustrate that nitrosoarene reactive intermediates can not only be generated at room temperature from nitroarenes but that unprecedented reactivity patterns can be triggered divergently to produce functionalized N-heterocycles. In addition to leveraging this novel reactivity to construct different sized N-heterocycles, future studies are also aimed at obtaining a deeper understanding of the spin dynamics and spin Hamiltonian parameters of the observed radical intermediates to conclusively assign their identity.