Sarah G. Scrivener and
Yi-Ming Wang*
Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, USA. E-mail: ym.wang@pitt.edu
First published on 1st May 2024
Cyclopentadienyliron(II) dicarbonyl complexes capable of coordinating to and enhancing the acidity of a range of unsaturated substrates have emerged as a new class of base-metal derived catalysts for C–H functionalization. In this manuscript, the iron-catalyzed C–H functionalization of allylic C(sp3)–H bonds using nitrogen containing α,β-unsaturated carbonyl compounds as coupling partners is reported. Employing a cationic cyclopentadienyliron dicarbonyl complex, this redox neutral process converts simple alkenes into allylic anion equivalents for 1,4-addition into maleimides, acyclic α,β-unsaturated imides, and vinylogous amides. The judicious pairing of pyridine and alkylamine bases with Lewis acid additives allowed each of these classes of substrates to be successfully employed, allowing for the formation of a diverse collection of cyclic and acyclic nitrogen-containing compounds featuring C–C unsaturation. The resulting Michael adducts can be further transformed into a variety of useful scaffolds such as allylated pyrroles, pyrrolidines, and carbocyclic acids.
Succinimide derivatives in particular not only have independent bioactivity1–4 but also serve as versatile intermediates toward the synthesis of other five-membered nitrogen-containing heterocycles, including γ-lactams, pyrrolidines, and pyrroles, that feature prominently in drug candidates.5–7 In addition to providing a useful functional group that allows for further derivatization, the installation of allylic groups on to these azacycles can lead to bioactive compounds of interest in their own right (Scheme 1).1,3 Conjugate addition of carbon nucleophiles (Michael addition) to the readily available maleimide scaffold constitutes a straightforward and attractive route for quickly accessing 3-alkylated succinimides.8–11 However, there is a distinct lack of suitable nucleophiles for the CC bond allylation of maleimide derivatives. Common organometallic allyl nucleophiles such as magnesium,12 silicon,13 boron,14 indium,15 and tin16 reagents all selectively provide the 1,2-adduct of direct carbonyl addition, rather than 3-allylated succinimide derivative resulting from 1,4-addition.17 The installation of an allylic substructure at the 3-position by conjugate addition has only been successful in the context of additions by extended enolate species (Scheme 2).18
![]() | ||
Scheme 1 Examples of biologically active 3-allylated succinimides, β-substituted β-amino ketones (top) and overview of the transformations reported in this manuscript (bottom). |
![]() | ||
Scheme 2 Allylation of maleimide derivatives as an unmet synthetic challenge and an approach to the synthesis of 3-allylated five-membered azacycles. |
Although 1,4-addition of allylic radicals to maleimides has been demonstrated using allylic sulfones as radical sources, the synthesis of the allylic sulfone precursors is often a nontrivial endeavour, and only the installation of the parent allyl group and a narrow range of 2-substituted derivatives have been demonstrated.19–21 Maleimides are also known to react with allylbenzene derivatives through the Alder-ene reaction22 to give 3-cinnamylated succinimides at very high reaction temperatures. Practically speaking, the current state-of-the-art for accessing allylated succinimide derivatives still relies on the reaction of succinimide-derived enolates with allylic bromides.23–25 However, highly reactive allylic bromides may be difficult to purify or handle and can give side products, including bis-allylated or regioisomeric adducts.23 In addition, they require additional steps to synthesize, creating hurdles when a functional group-rich allylic fragment is needed. A protocol for the regioselective addition of a nucleophilic allylic fragment to maleimide under catalytic conditions, particularly one employing simple hydrocarbon-based precursors (i.e., by C–H functionalization), would be a valuable tool for the synthesis of 3-allylated succinimides and other functionalized five-membered nitrogen-containing heterocycles.
Our group and others have shown that coordination of a cationic late transition metal complex to a C–C π-bond can greatly enhance the C–H acidity of the neighboring allylic, propargylic, or allenic protons.26,27 In particular, through the use of cyclopentadienyliron(II) dicarbonyl derivatives as catalysts,28 this acidity enhancement enables the use of weak, functional group-tolerant amine bases to effect the allylic deprotonation of olefins to generate nucleophilic allyliron intermediates that react with electrophiles with SE2′ selectivity.29–35 We further discovered that these nucleophilic organoiron species will undergo electrophilic functionalization with chalcone derivatives through a 1,4-addition process.36
In this article we report the development of a series of C–H functionalization methods for the coupling of alkenes with maleimides and other synthetically versatile nitrogen-containing α,β-unsaturated carbonyl compounds. These processes allow for the synthesis of high value nitrogen-containing building blocks using simple, readily available precursors. Moreover, these procedures can be used for the elaboration of complex scaffolds derived from known pharmaceuticals and bioactive compounds.
Entry | L.A. + additivea | Base | Temp. | Ratio (LA/base) | Yieldb (%) | dr |
---|---|---|---|---|---|---|
a Bistriflimide additive (0.35 equiv.) used for entries 4–11. b Yields determined by 1H NMR using 2,4-dinitrotoluene as the internal standard. c Isolated yield. Conditions: 1a (3.0 equiv.), 2a (0.2 mmol), [Cp*Fe(CO)2(thf)]+[BF4]−(20 mol%), BF3·Et2O (1.5 equiv.), 4-chlorolutidine (4.0 equiv.), AgNTf2 (0.35 equiv.), toluene [1.5 M], 40 °C, 24 h. | ||||||
1 | TIPSOTf | Collidine | 60 | 3/4 | 0 | — |
2 | TMSOTf | Collidine | 60 | 3/4 | 0 | — |
3 | BF3·Et2O | Collidine | 60 | 3/4 | 30 | — |
4 | BF3·Et2O + LiNTf2 | Collidine | 60 | 3/4 | 37 | — |
5 | BF3·Et2O + LiNTf2 | TMPH | 60 | 3/4 | 0 | — |
6 | BF3·Et2O + LiNTf2 | Lutidine | 60 | 3/4 | 40 | — |
7 | BF3·Et2O + LiNTf2 | 4-Cl-lutidine | 60 | 3/4 | 47 | — |
8 | BF3·Et2O + LiNTf2 | 4-Cl-lutidine | 40 | 3/4 | 64 | — |
9 | BF3·Et2O + LiNTf2 | 4-Cl-lutidine | 40 | 1.5/4 | 83 | 3.7![]() ![]() |
10 | BF3·Et2O + Mg(NTf2)2 | 4-Cl-lutidine | 40 | 1.5/4 | 75 | 6.4![]() ![]() |
11 | BF3·Et2O + AgNTf2 | 4-Cl-lutidine | 40 | 1.5/4 | 89c | 7.6![]() ![]() |
12 | BF3·Et2O | 4-Cl-lutidine | 40 | 1.5/4 | 9 | 6.3![]() ![]() |
With these reaction conditions in hand, we explored a variety of nitrogen substituents and alkene coupling partners. The scope of well tolerated N-substituents was broad: alkyl groups spanning a range of steric profiles (Scheme 3, 3ba and 3ca) were tolerated, as well as cycloalkyl groups (3da, 3ea), as was a maleimide derived from a pharmaceutical fragment (3ga). The reaction performed well with N-benzylic substituents (3aa and 3ha), giving good yields and good diastereoselectivity. Although N-aryl maleimides furnished products with somewhat reduced diastereoselectivity, succinimide products were nonetheless obtained in good to excellent yields for both electron-donating (3ka) and electron-withdrawing (3ia) aryl substituents. This reaction provided good to excellent yields for a range of alkene substrates, including electron-rich (3ab, 3ae) and electron-poor (3ad) allylbenzene derivatives. Unfortunately, unactivated olefins (without an additional aryl group at the α-position) were unsuccessful (<5% yield), presumably because the weakly basic 4-chlorolutidine (pKaH = 5.4) and lower reaction temperatures (40 °C) needed in this protocol are insufficient for deprotonation of these substrates.30
A slightly modified protocol could be performed on gram scale while maintaining synthetic efficiency, giving the aryl bromide 3ja in excellent yield. Notably, both aromatic and saturated five-membered azacycles can be obtained as derivatives of the succinimide products. For example, reduction of 3fc using LiAlH4 selectively provided either the 3-allylic pyrrole or the 3-allylic pyrrolidine depending on the level of vigor of the reaction conditions (Scheme 4).
This reaction was found to require elevated temperatures relative to the maleimide coupling partners, in addition to a larger amount of the Lewis acids BF3·Et2O and LiNTf2. These reaction conditions reflect the decreased reactivity of the acyclic system, as well as the presence of additional Lewis basic sites on the oxazolidinone substrate. Although diastereoselectivities remained modest after optimization, the identity and stoichiometry of the co-Lewis acid additive nonetheless proved important for this reaction (Table 2).
Entry | Ratio (LA/base/add.) | Additive | Yielda (%) |
---|---|---|---|
a Yields determined by 1H NMR using 2,4-dinitrotoluene as the internal standard. b Isolated yield. Conditions: 1a (3.0 equiv.), 4a (0.2 mmol), [Cp*Fe(CO)2(thf)]+[BF4]− (20 mol%), BF3·Et2O (5.0 equiv.), 4-chlorolutidine (3.0 equiv.), LiNTf2 (1.0 equiv.), DCE [1.5 M], 80 °C, 24 h. | |||
1 | 1.5/4/0.2 | AgNTf2 | Trace |
2 | 5/3/0.2 | AgNTf2 | 19 |
3 | 5/3/0.2 | Cu(NTf2)2 | 42 |
4 | 5/3/0.2 | Yb(OTf)3 | 54 |
5 | 5/3/0.2 | Zn(OTf)2 | 57 |
6 | 5/3/0.2 | LiNTf2 | 67 |
7 | 5/3/1.0 | LiNTf2 | 85 (88)b |
Optimized reaction conditions allowed for the coupling of α,β-unsaturated N-acyl oxazolidinone electrophiles with alkenes bearing synthetic handles such as aryl halides (5bg), and an alkene derived from a pharmaceutical fragment (5ah). Hydrolysis of the oxazolidinone with lithium hydroperoxide provided the δ,ε-unsaturated carboxylic acid 12 in excellent yield (Scheme 5). Moreover, ytterbium-catalyzed alcoholysis using propargyl alcohol afforded ester 13 bearing an additional terminal alkyne function in excellent yields.40
We began our investigation with the addition of allylbenzene to the phthalimido enaminone 6a. In contrast to the α,β-unsaturated imides explored earlier, BF3·Et2O was ineffective as the Lewis acid activator. Instead, we turned to silyl triflates as strong Lewis acids that irreversibly trap the incipient enolate ion intermediates as silyl enol ethers, which could be cleaved in a subsequent workup step to reveal the β-phthalimido ketone.36 We posited that this initial trapping as the silyl enol ether could also prevent the occurrence of deamination as a side reaction. In preliminary experiments, we found that an acidic workup (in contrast to fluoride-mediated or base-mediated desilylation) led to good yields of the ketone and minimal formation of the deaminated product.
Optimization of this reaction was carried out with careful attention to reagent stoichiometry. An increase of the reagents provided a corresponding increase in yield (Table 3, entry 2). The selection of TIPSOTf over TMSOTf was found to be highly beneficial to the reaction (entry 3). Other changes to the conditions, like decreasing the amount of LiNTf2 additive or the reaction temperature, led to decreased yields (entries 4 and 5).
Entry | Ratio (LA/base/LiNTf2) | Lewis acid | Temp. | Yielda (%) |
---|---|---|---|---|
a Yields determined by 1H NMR using 2,4-dinitrotoluene as the internal standard. b Isolated yield. Conditions: 1a (3.0 equiv.), 6a (0.2 mmol), [Cp*Fe(CO)2(thf)]+[BF4]− (20 mol%), TIPSOTf (2.5 equiv.), collidine (3.0 equiv.), LiNTf2 (0.6 equiv.), DCE [1.5 M], 80 °C, 16 h. Then 4 N HCl in dioxane. | ||||
1 | 1.5/2.0/0.6 | TMSOTf | 80 | 18 |
2 | 2.5/3.0/0.6 | TMSOTf | 80 | 37 |
3 | 2.5/3.0/0.6 | TIPSOTf | 80 | 90 (93)b |
4 | 2.5/3.0/0.3 | TIPSOTf | 80 | 73 |
5 | 2.5/3.0/0.6 | TIPSOTf | 60 | 85 |
Using the optimized protocol, we investigated the scope of the alkene nucleophile. Allylbenzene coupling partners with a variety of substitution patterns provided excellent yields under the optimized conditions, including safrole (Scheme 7, 7ae) and allylpentafluorobenzene (7aj). Changing the conditions by using TMPH (4 equiv.) in place of 2,4,6-collidine (3 equiv.) allowed a variety of electronically unactivated alkenes to be utilized, including cyclopentene (7al) and a terminal alkene incorporating a pharmaceutical fragment (7aq). The improved yields using the stronger base TMPH over weaker pyridine bases is consistent with the decreased C–H acidity of the electronically unactivated substrates in relation to allylbenzene derivatives, further validating the importance of correctly identifying suitable combinations of base and Lewis acid for each pairing of alkene and electrophile. Alkenes bearing functional groups such as an alkyl p-toluenesulfonate ester (7am), a boronic acid pinacol ester (7ao), and a five membered heterocycle (7an), were also good coupling partners in this reaction. Finally, with slightly modified conditions, coupling with a phthalimide-protected allylic amine (7ap) provided the protected allylic 1,2-diamine in good yield.
The synthetic value of this method was further demonstrated through the elaboration of the product into a new scaffold: under Wittig conditions, the phthalimido protected β-amino ketone provided the β-amino-δ,ε-unsaturated diene, which was then cyclized through ring closing olefin metathesis to give 3-aminocyclopentene 14.53
These modifications allowed us to prepare a range of β-substituted β-sulfonamido ketones, including products derived from allylbenzene (9aa), an ester-containing eugenol derivative (9af), and an unactivated alkene (9ak). The protocol was also found to tolerate a wide variety of sulfonamides, including a p-nitrobenzenesulfonamide (9ba) and a sulfonamide derived from celecoxib (9da) (Scheme 8).
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2314269. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4sc00655k |
This journal is © The Royal Society of Chemistry 2024 |