Palladium‐Catalyzed Regiodivergent Telomerization of Isoprene With Oxindoles

ABSTRACT

Transition metal‐catalyzed nucleophilic telomerization of isoprene is an efficient strategy for the construction of monoterpene skeletons, yet its development has been impeded by inherent challenges of chemo‐ and regioselectivity control, especially in reactions with carbon nucleophiles. Herein, we developed a versatile palladium catalysis for the regiodivergent telomerization of isoprene with oxindoles through synergistic control of ligands, solvents, and additives. In the presence of aprotic solvent NMP, we revealed the electron‐rich and less bulky tri(2‐furyl)phosphine (TFP) ligand can facilitate the exclusive formation of tail‐to‐tail geranyl oxindoles, meanwhile the reactivity was severely improved by adding base as additive. On the other hand, a newly designed hemilabile phosphine ligand enables selective access to tail‐to‐head isomer, where the MeOH worked as an irreplaceable solvent choice. Mechanistic experiments and density functional theory (DFT) calculations were conducted, which indicate that protolysis of the η¹,η³‐diallyl–Pd(II) complex is speculated to be the rate‐determining step. Apart from oxindoles, this regiodivergent protocol could also be applied to the nucleophilic telomerization of isoprene with a broad scope of amines. Furthermore, the synthetic utility of current synthetic strategy was illustrated by late‐stage alkene isomerization, as well as various derivatizations that effectively expand the structural diversity of monoterpenoids and complement alkaloid frameworks.

A palladium‐catalyzed regiodivergent telomerization of isoprene with oxindoles is achieved by synergistically tuning ligands, solvents, and additives. In aprotic solvent, the electron‐rich tri(2‐furyl)phosphine promotes tail‐to‐tail products under basic conditions, whereas a hemilabile phosphine ligand selectively affords tail‐to‐head isomers with methanol as the critical solvent.

1. Introduction

Monoterpenoids are a large and diverse group of naturally occurring compounds whose basic structure consists of two isoprene units [1, 2], playing an irreplaceable role in pharmaceuticals or agrochemicals due to their diverse bioactivities [3, 4, 5]. Besides, owing to their characteristic odor and taste, many monoterpenoids or their derivatives have also been applied in cosmetic materials [6], food flavors [7], or insect repellents [8, 9]. Although acyclic monoterpenoids can be structurally classified into four types—head‐to‐tail (H‐T), tail‐to‐tail (T‐T), tail‐to‐head (T‐H) and head‐to‐head (H‐H) coupling products—based on the different coupling sites of isoprene unit [10, 11], most natural monoterpenoids are found only with H‐T coupling backbone [12, 13]. This is mainly because the biosynthesis of monoterpenes relies on the transformation of geranyl diphosphate (GPP) intermediate, which is generated through enzymatic condensation of isopentenyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP) (Figure 1A) [14, 15, 16]. To overcome such limitation and expand the structural diversity of terpenoids beyond, sustained efforts have been devoted to artificial catalysis in past decades to enable the construction of monoterpenoid skeleton liberating from naturally existing H‐T selectivity [17, 18, 19, 20, 21, 22, 23, 24, 25].

FIGURE 1.

Catalytic constructions of acyclic monoterpenoids.

Isoprene is not only an important bulk fossil chemical but also the most abundant hydrocarbon in the atmosphere after methane, emitted via plant biogenic processes. In theory, the catalytic telomerization of renewable isoprene can serve as a transformative step for the construction of monoterpenoid backbones [26, 27, 28, 29]. However, isoprene contains an internal methyl group and represents the smallest non‐symmetrically substituted 1,3‐diene. The telomerization of isoprene has been challenging as it can generate up to four regioisomers (head‐to‐tail, tail‐to‐tail, and tail‐to‐head telomers). In addition, achieving chemoselectivity over direct X–H addition reactions, as well as suppressing its self‐dimerization, further complicates the reaction outcomes. In contrast to the success achieved with 1,3‐butadiene [30], very few catalytic systems have been developed to enable good selectivity in isoprene telomerization [31, 32, 33, 34, 35, 36, 37, 38]. Successful examples reported by Finn, Réau, and Castillon et al. mainly focus on C─N bond construction with diethylamine (Figure 1B) [32, 34, 39]. However, selectivity control using more challenging carbon nucleophiles remains in its infancy.

Oxindoles and their derivatives are prevalent structural motifs in numerous natural products and synthetic pharmaceutical molecules [40, 41]. In 2022, we achieved excellent head‐to‐head selectivity in the Pd‐catalyzed telomerization of isoprene with oxindoles by employing a bulky phosphine ligand [35]. Very recently, this strategy was successfully extended to nickel‐catalyzed arylative head‐to‐head telomerization using boronic acids [38]. Given that monoterpenes with head‐to‐tail selectivity can be readily accessed via nucleophilic substitution with bio‐derived geraniol, the controlled synthesis of the remaining two telomers (tail‐to‐tail and tail‐to‐head) is of great significance for enriching monoterpene structural diversity. In continuation of our long‐standing interest in the catalytic transformation of isoprene [36, 37, 42, 43, 44, 45, 46], we herein report the development of a ligand‐regulated regiodivergent telomerization of isoprene with oxindoles under Pd catalysis. Using the less sterically hindered tri(2‐furyl)phosphine (TFP) ligand, geranyl oxindoles with tail‐to‐tail selectivity were obtained with high chemo‐ and regioselectivity, while the tail‐to‐head telomer could be accessed by regulation with a newly designed and synthesized hemilabile phosphine ligand (Figure 1C). This work provides a novel paradigm for the synthesis of skeleton‐divergent terpenoids beyond naturally occurring cores (Figures 2, 3, 4, 5).

FIGURE 2.

Reaction optimization for tail‐to‐head (T‐H) telomerization of isoprene with oxindole.

FIGURE 3.

Substrate scope of tail‐to‐tail telomerization of isoprene with oxindoles and amines.

FIGURE 4.

Substrate scope of tail‐to‐head telomerization or isoprene with oxindoles and amines.

FIGURE 5.

Mechanistic investigations.

2. Results and Discussion

Initially, 3‐benzyloxindole (1a) and isoprene (2a) were selected as model substrates under palladium catalysis in N,N‐dimethylformamide (DMF) with Et₃N as additive (Tables 1 and S1–S5). Under the catalysis of Pd(OAc)₂ and PPh₃, the telomerization reaction delivered 60% yield of T‐T coupling isomer (3a), as well as 25% yield of tail‐to‐head product 5a and 3% yield of C5‐product 7a (Tables S1, entry 1). The assessment of palladium catalyst precursors showed that Pd(PPh₃)₄ exhibited higher yield of 3a than Pd(OAc)₂/PPh₃, but Pd₂dba₃ showed better selectivity (Table S1, entries 2–5). Therefore, we chose Pd₂dba₃ for further evaluation of ligands (Tables 1 and S2). The optimization of ligands revealed that phosphine ligands substituted by alkyl group would diminish the reaction yields (Table 1, entry 2 and see Table S2 for details). Among all tested triarylphosphine ligands, P(2‐furyl)₃ displayed the highest reactivity and regioselectivity, affording 3a and 4a in 59% and 6% yields, respectively (Table 1, entries 1–8). And the yield of 3a can be further increased to 61% yield when 20 mol% of P(2‐furyl)₃ was loaded (Table 1, entry 9). Then solvents were screened, disclosing that this reaction was generally facilitated by aprotic polar solvents, as a considerable loss of selectivity could be observed when MeOH employed (Table 1, entries 9–12 and Table S3). The N‐methyl‐2‐pyrrolidone (NMP) proved to be the best choice, giving product 3a in 69% yield, accompanied by 6% yield of 4a (Table 1, entry 10), Apart from Et₃N, other additives including organic or inorganic bases, acids, and water, were also surveyed (Table 1, entries 12–15 and Table S4). To our delight, the T‐T telomer 3a was produced in 84% yield with excellent regioselectivity by adding DABCO (entry 14). The regioisomer distribution (3a: 84% yield, 4a: 3% yield, 5a: not detected, 6a: not detected) was determined as 97:3:0:0, which matched the theoretical calculations excellently (94:5:<1:<1; see detailed discussion for Figure 6 and Page S76). In comparison, the telomerization reaction exhibited a significant loss of reactivity in the absence of additive (Table 1, entry 16). Controlled experiments showed that the reactions catalyzed by Pd₂(dba)₃ still performed better than those with Pd(PPh₃)₄ and Pd(OAc)₂ in NMP (Table S5).

TABLE 1.

Reaction optimization for tail‐to‐tail (T‐T) telomerization of isoprene with oxindole.

Entry	Ligand (x mol%)	Additive	Solvent	Yield (%)
				3a	4a		5a	6a	7a	8a
1	PPh3 (15)	Et3N	DMF	53	6	2		1	—	—
2	PPh2Cy (15)	Et3N	DMF	12	5	—		1	—	—
3	P(4‐MeC6H4)3(15)	Et3N	DMF	4	2	—		—	—	—
4	P(4‐ClC6H4)3(15)	Et3N	DMF	—	—	—		—	—	—
5	P(3‐MeC6H4)3(15)	Et3N	DMF	15	2	3		—	—	—
6	P(2‐MeOC6H4)3 (15)	Et3N	DMF	37	21	—		—	—	—
7	Ph‐JohnPhos (15)	Et3N	DMF	17	9	—		—	—	—
8	P(2‐furyl)3 (15)	Et3N	DMF	59	6	—		—	—	—
9	P(2‐furyl)3 (20)	Et3N	DMF	61	8	—		—	—	—
10	P(2‐furyl)3 (20)	Et3N	NMP	69	6	—		—	—	—
11	P(2‐furyl)3 (20)	Et3N	MeOH	31	2	17		21	5	8
12	P(2‐furyl)3 (20)	Et3N	iPrOH	52	11	—		4	2	2
13	P(2‐furyl)3 (20)	i Pr3N	NMP	60	8	—		—	—	—
14	P(2‐furyl)3 (20)	DABCO	NMP	84	3	—		—	—	—
15	P(2‐furyl)3 (20)	DBU	NMP	45	5	—		—	—	—
16	P(2‐furyl)3 (20)	—	NMP	26	—	—		—	—	—

Note: Condition: 1a (0.10 mmol), 2a (0.80 mmol), Pd₂dba₃ (5 mol%), ligand (15–20 mol%), additive (2.0 equiv.), solvent (0.2 mL), 80°C, 12 h. Yields were determined by GC‐FID analysis of the crude product mixture using mesitylene as internal standard.

Abbreviations: Ph‐JohnPhos, 2‐(diphenylphosphino)biphenyl; DBU, 1,8‐diazabicyclo[5.4.0]undec‐7‐ene; DABCO, 1,4‐diazabicyclo[2.2.2]octane.

FIGURE 6.

DFT calculations for the tail‐to‐tail telomerization with P(2‐furyl)₃ as ligand in NMP. B3LYP‐D3BJ/SDD (Pd), 6‐311+G(d,p)/SMD(NMP)//B3LYP/lanl2dz (Pd), and 6‐31G(d,p) 80°C (353.15 K).

Previously, we have demonstrated that excellent head‐to‐head selectivity (H‐H, 6a) could be achieved under palladium catalysis with bulky phosphine ligand P(2,6‐MeOC₆H₃)₃ when using MeOH as solvent [35]. It was also noted that the tail‐to‐head product 5a was also slightly facilitated under current condition in MeOH (Table 1, entry 11). Inspired by these results, further ligand screening was conducted with MeOH as solvent to improve the tail‐to‐head (5a) selectivity (Figure 2a and Table S6). The results showed that triaryl phosphine ligands bearing a biphenyl skeleton moderately switched the selectivity toward the tail‐to‐head product 5a (Figure 2a, entries 1–8). When 2‐(diphenylphosphino)biphenyl (Ph‐JohnPhos, L1) was employed, 5a was detected as the dominant product in 66% yield, along with other isomers (entry 1). Ligands with a substituent at the ortho‐position of the biphenyl skeleton (L2 and L3) successfully improved the yield of 5a to 69% and 76%, respectively. Replacing the phenyl groups with cyclohexyl groups (L4–L5) impeded current telomerization reactions, likely due to the increased electron density or steric hindrance (entries 4 and 5). This trend became even more pronounced with tert‐butyl‐substituted phosphine ligands (L6–L7), for which the corresponding telomerization reactions were almost completely suppressed (entries 6 and 7). Ligand L8 showed good reactivity in previous tail‐to‐head telomerization of isoprene with diethylamine [32], but only afforded 8% yield of 3a in current reaction (entry 8). Compared with diethylamine, the oxindole is a kind of weaker nucleophile with larger steric hindrance, thus calling for a more delicate decoration on Ph‐JohnPhos skeleton. The investigation of other solvent indicated that the success of current tail‐to‐head telomerization highly depended on MeOH (see Table S7). Replacing MeOH with other protic solvents would cause a thorough loss of reactivity, probably by influencing the protolysis process. Precedented work by Castillon et al. has revealed that the alcohol solvents controlled the regioselectivity as a function of its pK _a, leading to a continuous shift of the selectivity‐determining step [32]. The examination of commonly used palladium precursors disclosed that Pd₂dba₃ exhibited higher yield and regioselectivity of product 5a (Table S8).

Motivated by the promising “side arm” effect on selectivity improvement, we set out to optimize the yield and selectivity of 5a by decorating the ortho‐position of the Ph‐JohnPhos ligand via a P(III)‐directed C–H activation strategy (Table S9). Adopting a protocol developed by Shi et al. [47, 48, 49, 50], hemilabile phosphine ligands L9‐L15 were synthesized through Rh‐catalyzed C–H alkylation of the commercially available ligand L1 with activated alkenes (Figure 2b). As expected, telomerization reactions with these hemilabile ligands all showed a preference for the T–H coupling product 5a (Figure 2c, L9‐L15). Comparatively, ligands bearing a propionic ester group (L9, L11, and L12) significantly enhanced both reactivity and selectivity. Among them, ligand L11 with n‐butyl propionate exhibited best performance, affording the T‐H product 5a in 77% yield and 85% rr. By increasing the loading of L11 to 7.5 mol%, the yield of 5a was further improved to 82%, accompanied by a small portion of regioisomers (3a: 4% yield, 4a: 1% yield, 6a: 8% yield, see Table S19, entry 4). The product ratio 3a/4a/5a/6a was calculated to be 4:1:87:8, in good agreement with the theoretical isomer distribution (see detailed discussion for Figure 7 and Page S78). The mixed solvents of different alcohols with MeOH (1:1) were also examined, nevertheless, led to a considerable loss of selectivity (Table S20). Increasing the steric hindrance of the ligands (L10 and L13) proved detrimental to the reaction outcomes. In addition, replacing the ester group with a phosphorus unit—affording bisphosphine monoxide (BPMO)‐type hemilabile ligands—also enabled the reaction, albeit with relatively lower efficiency (L14 and L15).

FIGURE 7.

DFT calculations for the tail‐to‐head telomerization with L11 as ligand in MeOH. B3LYP‐D3BJ/SDD (Pd), 6‐311+G(d,p)/SMD(MeOH)//B3LYP/lanl2dz (Pd), and 6‐31G(d,p) 60°C (333.15 K).

The above promising results inspired us to further modify the ligand skeleton by incorporating an allene unit (Figure S1) [51, 52, 53]. After extensive attempts, a modified Rh‐catalyzed protocol was successfully developed via P(III)‐directed C–H alkenylation with allenyl phosphine oxide (Figure 2a and Tables S10–S15). Using Rh(COD)₂BF as pre‐catalyst, BPMO ligands L16‐L23 were afforded in 21%–96% yields as either (E)‐β,γ‐adducts or (E)‐α,β‐adducts, depending on the γ‐substituent of allenyl phosphites. Remarkably, the stereoselectivity could be inverted by employing a catalytic amount of tetrabutylammonium bromide (TBAB), affording the Z‐configured β,γ‐adduct L24 in good yield. The structures of ligands L16, L20, and L24 were further confirmed by single‐crystal x‐ray crystallography (CCDC 2425139, 2425232, and 2425141) [54]. With these new BPMO ligands in hand, we subsequently evaluated their performance in the T‐H telomerization of isoprene with oxindole (Figure 2b, L16‐24). Gratifyingly, although ligands with a tetrasubstituted alkene motif (L16‐18, L24) exhibited considerable telomerization reactivity and regioselectivity comparable to L11, a remarkable improvement in yield and T‐H selectivity was observed when using the (E)‐α,β‐adducts L20–23 as ligands. Variation of the substituents revealed that the ethyl‐substituted BPMO ligand L22 was optimal in terms of regioselectivity, delivering the T‐H product 5a in 83% yield.

With the optimized conditions in hand, the substrate generality of Pd‐catalyzed tail‐to‐tail telomerization was subsequently tested in the presence of P(2‐furyl)₃ and DABCO (Figure 3a). Subjecting 3‐Bn substituted oxindole 1a to standard conditions furnished geranyl oxindole 3a in 85% yield and 96:4 rr. The benzyl groups, regardless of the electron‐donating groups (Me, ^t Bu, OMe) or electron‐withdrawing groups (F, Ph) at its para position, were all well‐tolerated to afford corresponding products in 81%–98% yields with regioselectivity maintained at high levels (3b‐3e, 3 g). While 3‐(p‐CF₃‐Bn) or 3‐(o‐F‐Bn) substituted oxindoles gave products (3 h, 3j) in slightly inferior yields (58%–69%), 3‐(2‐naphthylmethyl) and 3‐(m‐Me‐Bn) substituted oxindoles efficiently generated products (3f, 3i) in 84%–99% yields. It was worth mentioning that when varying the alkyl substituents at the C3‐position of oxindole skeleton, the reactions were also compatible with the catalytic system and led to products 3k‐3m in decent 58%–78% yields. Additionally, 3‐cyanomethyl and heterocyclic substituents proved viable as well, providing the related products (3n‐3r) in 65%–99% yields. Oxindoles with different substituents on the benzene ring also participated in the reactions smoothly. For example, substrates with methyl, methoxy, fluoro, or ester groups on the oxindole framework could also deliver geranyl products (3s‐3x) in good yields (82%–95%).

Although the selective telomerization of isoprene with diethylamine has been extensively studied [32, 34, 39], the tolerance of substrate scope has still remained as an intractable task [32]. Delightfully, our established protocol can also be applied to a broad type of amine nucleophiles. As shown in Figure 3b, both cyclic and acyclic alkyl amines readily underwent the reactions, affording the corresponding products in 96%–99% yields (9a‐c) with excellent regioselectivity. Indoline was a feasible substrate as well, delivering the tail‐to‐tail (T‐T) geranylated product 9d in good 86% yield and 96:4 rr. Notably, the C3‐disubstituted oxindole could also undergo telomerization with isoprene smoothly at the N1‐position, although the regioselectivity slightly diminished (9e and 10e).

The substrate scope for Pd‐catalyzed tail‐to‐head (T‐H) telomerization was then evaluated in MeOH at 60°C using either L11 or L22 as the ligand (Figure 4a). In general, ligand L22 exhibited superior regioselective control compared to L11. However, the latter showed enhanced telomerization reactivity in some cases. Both telomerization protocols displayed a broad scope for C3‐benzyl‐substituted oxindoles: electron‐donating or electron‐withdrawing groups (e.g., Me, ^t Bu, Ph, F) on the benzyl moiety were well tolerated, leading to T‐H products (5a–5h) in good yields (up to 98%) and high regioselectivities (up to 93:7 rr). Other C3‐alkyl‐substituted oxindoles were also suitable substrates for the T‐H geranylation reactions (5i–5j). However, a decrease in regioselectivity was observed for the cyanomethyl‐substituted oxindole (5k), presumably due to its strong coordinating ability. To our delight, N‐protected oxindole exhibited better performance in both yield and regioselectivity in the reactions with L22 as the ligand (5a vs. 5l, 5k vs. 5o). And heterocyclic substituents, such as pyridinyl and furyl at the C3 position were also compatible, affording the corresponding products 5m and 5n in good yields and acceptable regioselectivities. Modifications of the oxindole framework with electron‐donating (Me, OMe) or electron‐withdrawing (F, CF₃, CO₂Me) groups at different positions were all feasible, providing T‐H telomers (5p–5t) in high yields (up to 99%), with regioselectivity remaining at a decent level throughout these transformations. Notably, the C3‐substituent of oxindole is crucial to ensure the reactivity and selectivity. Subjecting indolin‐2‐ones to the standard conditions, the telomerization reactions were completely impeded or accompanied by complicated side products (Figure S5).

The Pd₂dba₃/L22 catalysis also effectively promoted the geranylation of various amines in T‐H selectivity, even when the Pd loading was decreased to 0.5 mol% (Figure 4b). Secondary amines, whether cyclic or acyclic, all afforded products 11a–d in excellent yields with up to 95:5 rr. The T‐H telomerization of benzylamines was also effective, although the regioselectivity was slightly eroded (11e‐g). Remarkably, the geranylation of amino alcohols under the current conditions showed exclusive chemoselectivity for the amino group over the hydroxy group, giving products 11 h and 11i in 90% and 75% yields, respectively. Sterically hindered amines with acetal or adamantyl groups were tolerated as well (11j–k). Apart from aliphatic amines, anilines were also viable substrates. Among them, anilines bearing electron‐rich groups (p‐Me, p‐ ^t Bu, 3,5‐Me) smoothly delivered T‐H isomers 11l–n. Interestingly, primary alkyl amines with small substituents led to the formation of bis‐geranylated products 11o–11q in moderate yields of 34%–44%.

To gain mechanistic insights into the tail‐to‐tail telomerization of isoprene with oxindole, we prepared several Pd catalytic precursors with different phosphine ligand, such as [(acac)Pd(PPh₃)₂]BF₄ [Pd‐PPh₃], [(acac)Pd(TFP)₂]BF₄ [Pd‐TFP] and [(acac)Pd(dppp)]BF₄ [Pd‐dppp] (see Supporting Information for details) [55, 56]. Then the reactivity of these [Pd‐L] precursors was evaluated under standard condition in the coupling of isoprene with oxindole (Table S23, entries 1–3). Only telomerized products were produced under [Pd‐PPh₃ ] and [Pd‐TFP] catalysis and the tail‐to‐tail product 3a was generated in 88% yield (> 20:1 rr) with the aid of [Pd‐TFP], while [Pd‐dppp] catalysis mainly led to 1:1‐adducts (7a, 8a). Reaction of [Pd‐TFP] with isoprene in the presence of reductant zinc powder and acetone successfully delivered a mixture of bis(allyl)‐Pd intermediate B and E with a low‐field signal at −15.7 and −15.2 ppm observed via in situ ³¹P NMR analysis (Figures 5a and S23). Moreover, subjecting such intermediates to couple with oxindole 2a through an in situ procedure could produce telomeric products in 59%, 11%, and 5% yield respectively (T‐T, T‐H, and H‐T), indicating that the bis(allyl)‐Pd complex is probably the active intermediate for this telomerization (Figure 5a). To clarify the ligation state of palladium metal with hemilabile phosphine ligands (L9), the reactions of phosphine ligands (Ph‐JohnPhos L1 and L9), [Pd(allyl)Cl]₂ and AgBF₄ were conducted to synthesize [Pd‐L] complexes (Figure 5b) [57]. While a single signal at 22.6 ppm was found for [Pd‐Ph‐JohnPhos] complex via ³¹P NMR analysis (Figure S30), two signals at 21.3 and 20.7 ppm (1.6:1) were found for complexes arise from L9 (Figure S32), possibly attributes to the existence of Pd‐carbonyl coordination (Figure 5b, [Pd‐9] and [Pd‐L9]′). To verify this, infrared radiation absorption spectra were conducted with L9 and Pd‐L9 complex respectively. It showed that the intensity of IR absorption peak for unsymmetric stretching vibration (ν_C‐O‐C ) for Pd‐L9 complex was enhanced and underwent redshift compared with spectra for L9 (Figure S33). These results indicate the ester group in phosphine ligand L9 probably might act as hemilabile coordinated site with Pd^II center.

Kinetic experiments showed an approximately linear correlation between product yield and reaction time for both 3a and 5a under standard conditions, indicative of a steady catalytic process (Figure 5c). Deuterium‐labeling experiments were then conducted to get more mechanistic insights (Figure 5d). In the tail‐to‐tail telomerization, the addition of CH₃OD led to product 3a‐d in 48% yield with 61% deuterium at C6 site of geranyl chain. In comparison, subjecting oxindole N‐Me‐1a‐ d with 90% D at C3‐position to the standard reaction condition also delivered product 3o‐d with 82% deuteration rate at C6 site of geranyl chai. For the tail‐to‐head telomerization, the use of CH₃OD as solvent gave the product 5a‐d (64% yield, 71:8:17:4 rr) with 94% deuterium at C6 site of geranyl chain. When N‐Me‐1a‐d (90% D at C3‐position) was added in CH₃OD, product 5l‐d was obtained in 71% yield with 100% deuterium at C6 site. These results indicate that the protonolysis process takes place at C6 site of the bis(allyl)‐Pd intermediate. When the tail‐to‐tail telomerization proceeds in an aprotic solvent (NMP), the proton source involved in the protonolysis process should be directly derived from the C(3)‐H of oxindole (see Figure 6). Conversely, the alcohol may participate in the protonolysis when MeOH added, as illustrated by the DFT calculation (see Figure 7). In addition, our supplementary experiments confirmed that H–D exchange between the C(3)‐H of oxindole and MeOD was feasible under the standard conditions, with deuteration incorporations of 65% and 78%, respectively (Figure S24c). Therefore, it remains difficult to determine whether the proton source originates from the alcohol or the oxindole when reactions were conducted in protic solvent. Moreover, oxindole ( N‐Me‐1a) and deuterated oxindole ( N‐Me‐1a‐d ) were reacted in parallel under condition A and condition B, respectively (Figure 5e). The observation of a substantial kinetic isotope effect (KIE = 2.1) in the tail‐to‐tail telomerization pathway suggests the protonation of the bis(allyl)‐Pd intermediate might be the rate‐determining step. A measurable KIE (1.6) was also found for the tail‐to‐head telomerization pathway, implying that protonation step probably contributed to the rate‐limiting process as well.

To further elucidate the mechanism and understand the origin of its regioselectivity, density functional theory (DFT) calculations were performed for this Pd‐catalyzed telomerization on a model reaction with oxindole 1a and isoprene 2a. Based on our mechanistic experiments and related reports [31, 32, 33, 34, 35, 36, 37, 38], the reaction may mainly undergo three fundamental steps: oxidative cyclization between two isoprene molecules and Pd(0) precursor, protolysis and nucleophilic substitution. For the T‐T telomerization reaction (Figures 6 and S34), the Pd(0) initially coordinates with two isoprene molecules assisted by P(2‐furyl)₃ ligand. Due the nonsymmetric structure, the isoprene unit may coordinate with Pd(0) in the presence of P(2‐furyl)₃ ligand at either head‐ or tail‐site, leading to three Pd(0)‐alkene complexes (Int A^HH , A^HT/TH , and A^TT ). Then oxidative cyclization proceeds through transition state TS1, delivering four possible palladacycles (Int B^HH , B^HT, B^TH , and B^TT ). The transition state barrier of this oxidative cyclization step is relatively low (ΔG ^ǂ = 10.6–12.0 kcal/mol, TS1). The base additive, DABCO, may act as proton shuttle by assisting the deprotonation of oxindole and then promoting the protolysis of palladacycle B to generate allyl‐Pd(II) intermediate C (Figures S6 and S7). This transition state is calculated to be energetical climax and is speculated a rate‐limiting step. In this context, the Boltzmann distribution of the relative energies was calculated based on TS2 (pages S77). The predicted isomer distribution is 94%, 5%, <1%, and <1% for telomers 3a, 4a, 5a, and 6a, respectively. This distribution was in excellent agreement with the experimental values (97%, 7%, n. d., and n. d., Table 1, entry 14). Investigations by Jolly and our group all indicated that the formation of (η¹,η³‐Me₂C₈H₁₀)Pd complex may be reversible [38, 58]. The lowest energy barrier was found for tail‐tail (T‐T) dimer (ΔG ^ǂ = 20.4 kcal/mol, TS2^TT ), while the energy barrier for T‐H dimer was highest (ΔG ^ǂ = 24.1 kcal/mol, TS2^TH ). Finally, nucleophilic substitution of Int C^TT with oxindole anion via TS3 (ΔG ^ǂ = 12.8 kcal/mol) releases T‐T telomer 3a as preferential product and regenerates active Pd(0) catalyst.

For Pd‐catalyzed tail‐to‐head telomerization, the DFT calculations were performed using L11 as ligand for simplicity (Figures 7 and S35), which indicates that the reaction oxidative cyclization step of Pd(0)‐alkene complexes (Int D) may also proceed in an equilibrium prior to the next catalytic step (ΔG ^ǂ = 10.2–13.5 kcal/mol, TS4). For the protolysis, methanol may work as a proton shuttle under current condition, enabling a proton to migrate from the N‐atom of the oxindole to the allylic position of the isoprene unit (Int E). Among all isomers in this process, the transition state of T‐H dimer (TS5^TH ) was found to have the lowest relative energy (20.0 kcal/mol). The NH proton of 3‐benzyloxindole is proposed to activate methanol through hydrogen‐bonding interactions. Given that N‐protected oxindoles can also undergo these reactions, we performed additional DFT calculations on distinct protonolysis pathways starting from intermediate E^TH (Figure S36). The results indicate that direct protonolysis by MeOH—without hydrogen‐bond activation by the NH moiety of 3‐benzyloxindole—is also viable for reactions employing N‐protected oxindoles (TS5‐1^TH , ΔG^‡ = 25.0 kcal/mol, Figure S36). In contrast, the ligand‐to‐ligand hydrogen transfer (LLHT) pathway is less favorable due to its higher energy barrier (TS5‐2^TH , ΔG ^‡ = 28.8 kcal/mol, Figure S36). Besides, the calculations of the Boltzmann distribution based on TS5 showed that predicted isomer distribution for telomers 3a, 4a, 5a, and 6a is 6%, 1%, 80%, and 13%, respectively (pages S78). The isomer distribution was also in agreement with the experimental values (4%, 1%, 87%, and 8%, Table S19, entry 4). After protolysis step, the reaction preferentially delivers the T‐H telomeric allyl‐Pd(II) cation and oxindole nitrogen anion (Int F^TH and its de‐methanolization Int G^TH ). Subsequently, the ion migration occurs from nitrogen to benzylic carbon atom through the introduction of second oxindole molecule 1a (TS6, ΔG ^ǂ = 9.4 kcal/mol), leading to an ionic pair consisting of the allyl‐Pd(II) cation and the oxindole carbon anion (H^TH ). In the end, nucleophilic substitution via transition state TS7^TH (ΔG ^ǂ = 3.4 kcal/mol) gives the tail‐to‐head coupling product 5a and recycles the palladium catalysis.

The tail‐to‐tail products 3 can be further isomerized to internal skipped diene tail‐to‐tail II products 13 via Pd‐H catalysis (Figure 8a) [59]. After optimization (Table S24) [60], the product 13a could be obtained in 88% yield with the aid of bulky ligand P( ^t Bu)₃ and acid p‐TsOH in dichloroethane. The 2‐furyl heterocyclic substituted geranyl oxindole also went smoothly in the isomerization process to provide 13b in 90% yield. Oxindoles bearing electron‐withdrawing groups (F and CO₂Me) were all compatible and gave the corresponding products in good yields (13c and 13d). Additionally, geranylated oxindoles with alkyl substituents at C3‐position produced internal skipped diene products in 87% and 82% yields as well (13e and 13f).

FIGURE 8.

Scale‐up preparation and divergent transformation.

To further illustrate the practical utility of this telomerization protocol, scale‐up experiments were performed (Figure 8b). The tail‐to‐tail geranyl products 3a and 3r could both be obtained in gram scale with good yields and regioselectivity (5.0 mmol scale, 1.56 and 1.49 g, respectively). The tail‐to‐head geranyl product 5a was also isolated in 91% total yield in the presence of L11 (1.0 mmol scale, 0.33 g). The unnatural geranyl oxindoles could be transformed to various products by manipulating the geranyl olefin, cyano motif and oxindole skeleton (Figure 8c,d). For example, geranyl 3‐Bn oxindoles 3a and 5a could be further modified by nucleophilic substitution with prenyl bromide at N1 site to give prenylated products 14 and 15 in 97% and 88% yields. The terminal olefin unit at the geranyl group of telomeric products 3a and 5a could undergo the cobalt‐catalyzed Simmons‐Smith reaction to deliver cyclopropanation products 16 and 17 in 84% and 83% yields [61]. In the presence of LiAlH₄, the reduction of geranyl acetonitrile 3r was carried out to access a primary amine 18 in 65% yield [62]. Besides, 3r could be condensed with amino alcohol to form oxazoline 19 in 67% yield [63, 64]. The hexahydropyrrolo[2,3‐b]indole are very important skeleton in many natural products and synthetic pharmaceuticals [65, 66, 67]. To our delight, geranyl hexahydropyrrolo[2,3‐b]indole 22 could be prepared smoothly through stepwise hydrolysis, amidation and reductive annulation reactions from product 3r. Alternatively, the hydrolysis of product 3r in the presence of H₂O₂, followed by one‐pot reduction and ring‐closure by LiAlH₄ also delivered product 24 as hexahydropyrrolo[2,3‐b]indole skeleton in good efficiency.

3. Conclusion

In conclusion, a regiodivergent Pd‐catalyzed telomerization of isoprene with oxindoles has been achieved through synergistic manipulation of ligands, solvents and additives. With DABCO as additive, excellent tail‐to‐tail (T‐T) selectivity can be achieved with TFP as ligand in aprotic polar solvent. In contrast, the selectivity is switched to tail‐to‐head (T‐H) coupling products in the presence of hemilabile phosphine ligands when protic MeOH was employed as solvent. Through mechanistic experiments and DFT calculations, the oxidative cyclization is found to be reversible and rate‐determining step is speculated to the protolysis process. The synthetic utility is also demonstrated by Pd‐catalyzed alkene isomerization and a series of derivatizations of the geranyl olefin, cyano group, and oxindole skeletons. This strategy expands monoterpenoid diversity beyond biosynthetic H‐T limitations, advancing synthetic access to unnatural terpenoids for pharmaceutical exploration.

Author Contributions

Ying‐Ying Liu: conceptualization, investigation, methodology, validation, software, formal analysis, data curation, writing – original draft, and writing – review and editing. Shao‐Han Sun: conceptualization, investigation, writing – original draft, methodology, validation, writing – review and editing, software, formal analysis, and data curation. Xue‐Ting Li: validation, data curation, writing – original draft, and writing – review and editing. Zhi‐Hui Wang: data curation, validation, writing – review and editing, and writing – original draft. Ding‐Wei Ji: conceptualization, investigation, methodology, validation, software, formal analysis, data curation, supervision, writing – original draft, writing – review and editing, funding acquisition, visualization, project administration, and resources. Xiang‐Ping Hu: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation, supervision, and resources. Qing‐An Chen: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation, supervision, and resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: anie72888‐sup‐0001‐SuppMat.pdf.

Supporting File 2: anie72888‐sup‐0002‐cif.zip.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.