Recent advances in the application of ring-closing metathesis for the synthesis of unsaturated nitrogen heterocycles

Abstract

This short review summarizes recent advances relating to the application of ring-closing olefin-olefin and carbonyl-olefin metathesis reactions towards the synthesis of unsaturated five- and six-membered nitrogen heterocycles. These developments include catalyst modifications and reaction designs that will enable access to more complex nitrogen heterocycles.

Graphical Abstract

1. Introduction

Nitrogen heterocycles are among the most prevalent structural motifs in both natural and unnatural biologically active compounds.¹ This ubiquitous scaffold (Figure 1) can be found in pharmaceutically relevant compounds such as iminosugars (1-2),² therapeutics (3-5),³ and alkaloids (6-8)⁴ as well as chiral ligands⁵ and organocatalysts for asymmetric synthesis.⁶ Due to the overall utility and abundance of nitrogen heterocycles, their syntheses continue to be an active area of interest.

Figure 1.

Therapeutics and natural products containing pyrrolidine and piperidine structural motifs.

For the purposes of this review, we will focus on the unsaturated five- and six-membered nitrogen heterocycles – specifically 3-pyrrolines and tetrahydropyridines (Figure 2). Such compounds serve as useful synthetic building blocks as the resulting olefin provides a functional handle for further diversification.⁹ There are a variety of strategies to directly access these aza-cycles^{7, 8} Several cyclization strategies have been developed to access 3-pyrrolines, including the cyclization of α-amino allenes mediated by transition metals¹⁰ and potassium carbonate.¹¹ Other cyclization strategies include Heck-aza-Michael reactions,¹² in situ formation of alkylidene carbenes from vinyl bromides resulting in 1,5-C–H insertion,¹³ the use of azomethine ylides,¹⁴ and Lewis acid-mediated cyclizations.¹⁵ Another interesting approach includes the ring-expansion of aziridines¹⁶ to the corresponding pyrroline, or alternatively the ring contraction of diazo-oxazepanes.¹⁷ Tetrahydropyridines present a greater challenge, however many versatile methods have been developed including 6π-cyclization strategies¹⁸ such as the aza-Diels-Alder reaction¹⁹ (Figure 2). Alternative strategies take advantage of traditional amine reactivity including nucleophilic additions to substituted olefins²⁰ and phosphine catalysis²¹ to promote the cyclization to generate the desired tetrahydropyridines. Finally, similar mechanistic strategies for accessing 3-pyrrolines and tetrahydropyridines have been reported, including allene annulation²² and the reduction of the aromatic pyrroles and pyridines.²³

Figure 2.

Summary of synthetic strategies to access unsaturated nitrogen heterocycles.

One strategy that has been proven effective for the synthesis of unsaturated nitrogen heterocycles is ring-closing olefin metathesis. Metal alkylidenes have been implemented in ring-closing metathesis reactions to access both aliphatic and heterocyclic rings,²⁴ however electron-rich amines have proven to be challenging substrates under metathesis conditions. The synthesis of aza-cycles via olefin-olefin metathesis was initially reported by Grubbs,²⁵ who found that subjecting allylamines to a metal alkylidene such as Grubb’s first generation catalyst G-I or Schrock’s catalyst provided the ring-closing metathesis product in good to excellent yields. As this approach was expanded to more complex amine-containing systems, it was revealed that the ring-closing metathesis worked well for substituted or electron-deficient amines. However, systems in which the amines maintained high electron densities shut down due to decomposition of the catalyst caused by coordination between the amines and the metal alkylidene complex.²⁶

While the presence of Lewis basic amines in metathesis substrates has created a dogma that amines will disrupt the desired reaction, several strategies have emerged that have successfully led to favorable outcomes including utilizing steric affects, attenuating amine basicity, and controlling the nature of the catalyst itself. There have been several excellent reviews on this topic covering substrate design strategies that have led to the application of ring-closing metathesis to access increasingly complex nitrogen heterocycles.²⁷ This review will instead focus on recent developments in both reaction and catalyst design that have further advanced the field.

2. Expansion of Ring-Closing Metathesis Methods

As previously mentioned, amines represent a challenging functional group that can coordinate to metal alkylidenes resulting in catalyst decomposition and ultimately inhibition of the desired olefin metathesis reactions.²⁶ A variety of strategies have been employed to overcome this inherent shortcoming, including deactivation of the amine by introducing steric bulk to the substrates or by reducing electron-density around the nitrogen atom. With these solutions available, ring-closing metathesis has been applied successfully towards the synthesis of complex nitrogen heterocycles with lower catalyst loadings and shorter reaction times.

Since olefin metathesis was first leveraged for the synthesis of heterocyclic amines, ruthenium catalysts and the understanding of the reaction mechanism have enabled the ring-closing metathesis reaction of unsubstituted bis-allylamine 9 in good yields with low catalyst loadings (Scheme 1a). Yields continue to rise as the ruthenium-phosphine ligands were replaced with N-heterocyclic carbenes as in HG-II.²⁸ However, while these homogeneous catalysts have been employed in a variety of systems, one challenge – particularly on an industrial scale – is the high catalyst cost and the removal of residual ruthenium after the reaction is complete. As such there has been a continued effort to realize lower catalyst loadings.^29,30 Pederson, et al.³⁰ recently employed high throughput robotic techniques using Symyx technology in order to identify the optimum reaction conditions for accessing five-, six-, and seven-membered nitrogen heterocycles (Scheme 1b). With this technology, Pederson found that by extending the reaction times to 8 h, the catalyst loading could be lowered to as little as 500 ppm to generate the cyclic amines (10 and 12) in as high as 99% yield (Scheme 1b). The approach also worked well for di-, tri-, and tetra-substituted olefins.

Scheme 1.

Recent advances in the application of catalytic ring-closing metathesis towards the synthesis of nitrogen heterocycles.

Another hurdle that has been surpassed in recent years was the application of ring closing metathesis to access cyclic alkenyl halides which represent valuable synthons. The first reported examples of utilizing ring-closing metathesis for the synthesis of vinyl halides came from Weinreb and co-workers,³¹ who were able to access cyclic chloroalkenes in good yields; however the method did not work for the synthesis of the more synthetically useful vinylbromides.³² Dorta and coworkers³³ recently reported the first example of the synthesis of cyclic alkenyl bromides by replacing the unsubstituted olefins with the styrene derivatives (Table 1). They propose that the terminal olefin 19a undergoes initiation with G-II to generate intermediate 20a. The bromoalkene can then react with the Ru-center leading to undesired catalyst decomposition. During examination of the bromoalkene, Dorta found that while the unsubstituted 19a and (E)-styrene 19b both led to complete decomposition of the catalyst, the (Z)-styrene provided the desired product in just 30 minutes with greater than 98% yield (Table 1, entry 3).

Table 1.

Catalytic ring-closing metathesis towards the synthesis of cyclic alkenyl halides.

Conditions:

^aReactions were performed using 0.16 mmol of substrate in 0.1 M benzene and 2 mol% of G-II. Reactions were run for 1.5–2 h.

^bReaction was run for 4 h with 5 mol% of G-II in DCM.

This method has also been applied towards tosyl-protected amines to access pyrroles and tetrahydropyridines (Table 1). The reaction worked well for both unsubstituted substrates 13 and 15. Higher yields (up to 97%) were observed upon incorporation of a phenyl substituent adjacent to the amine (14, 16, and 17). The method was also surprisingly facile and led to the synthesis of tetra-substituted cyclic chloroalkene 18.

While this strategy focused on substrate modification, many approaches have targeted ways to reduce the electron-density around the amine without the introduction of additional functional groups or steps. One such approach involves the in situ protection of amines via the formation of amine salts in order to prevent catalyst decomposition (Scheme 2). Similar deactivation strategies have been employed performing the in situ deactivation of amines via the addition of Brønsted³⁴ or Lewis acids.³⁵ Recently, this approach was successfully utilized by Woodward, et al.³⁶ to generate both cyclic and acyclic aminoalkenes. Initially, the approach was hindered by the solubility of the amine salts in organic solvents. After examining a variety of amine salts including halide, triflate, and tetrafluoroborate salts, the amine tosylate salts 22 and 24 were identified as superior substrates due to their solubility in dichloromethane at elevated temperatures. The amine tosylate salts were subjected to ring closing metathesis under both conventional and microwave heating conditions. While the ammonium salt 22 gave excellent yields of pyrroline 23 under both conditions, the quaternary amine 24 resulted in low conversion under conventional methods. However, upon subjection to microwave irradiation pyrroline 25 was obtained in excellent yield (Scheme 2). Furthermore, this approach is desirable since the resulting product 23 could be readily deprotected via subjection to basic conditions without further chemical modifications.

Scheme 2.

(a) Conventional Method: HG-II (5 mol%), 0.1 M DCM, 40 °C, 24 h. (b) Microwave Conditions (yield in parentheses): HG-II (5 mol%), 0.1 M DCM, 2 h, 100 W, under nitrogen. All yields determined by ¹H NMR spectroscopy in d₄-MeOH.

Other advances in the metathesis-mediated formation of nitrogen heterocycles is the strategic implementation of this synthetic tool to access highly desired chiral nitrogen heterocycles.³⁷ One such example came from Feringa and coworkers,³⁸ who developed a copper-catalyzed method for the asymmetric substitution of allylic bromides with Grignard reagents through the use of chiral ferrocene-based bisphosphine ligands L1. Recognizing that this method produced terminal olefins, Feringa utilized it to convert the allyl bromides 26 and 29 to the chiral metathesis substrates 27 and 30 which could then be subjected to metathesis conditions to obtain access to enantioenriched products 28 and 31 (Scheme 3). The allylic bromides could then be subjected to either olefin metathesis or ene-yne metathesis to provide chiral tetrahydropyridines in good yields and complete stereoretention. Additionally, this method was applicable for the synthesis of seven- and eight-membered rings.

Scheme 3.

Asymmetric allylic alkylation and subsequent RCM for the preparation of chiral tetrahydropyridines.

3. Evaluation of Catalyst Design

While the previous discussion focused on modifications in the reaction design and conditions to promote the desired ring-closing metathesis, another key strategy employs changes in catalyst design. Significant advances in metathesis reactions have been realized as the result of the development of more robust and highly reactive ruthenium catalysts that promote ring-closing metathesis without unfavorable side reactions. Specifically, modifications on the benzylidene ligands have resulted in significant changes in steric strain, chelate ring size, and electron density of the aromatic ring. Two such examples of major modifications in ligand design have been the replacement of the tricyclohexyl phosphine ligand from G-I with a more active N-heterocyclic carbene (G-II)²⁸ and the incorporation of benzylidene ligands (HG-II).³⁹

Continued studies of ligand development and design reveal that the electronic substitution of the ligand also plays a critical role in catalyst reactivity and stability (Figure 3).⁴⁰ For instance, the reactivity of HG-II can be significantly enhanced by the addition of an electron-withdrawing substituent such as the nitro group in HG-II-NO₂ which diminishes the donor activity of the oxygen chelate (Figure 3c). Grela and Lemcoff⁴¹ recently undertook a variety of studies exploring electronic effects in ruthenium catalysts containing oxygen, nitrogen, and sulfurchelated ligands. Changing the chelating heteroatom had a significant impact on the reactivity of the catalyst and its ability to undergo initiation. The electron-rich nitrogen (Ru-II, Ru-III, and Ru-IV) and sulfur derivatives required elevated temperatures and longer reaction times to give yields comparable to those obtained with HG-II and HG-II-NO₂ (Figure 3d). However, while these catalysts were relatively slow to initiate, addition of the nitro group greatly affected the overall stability of the catalyst and its ability to mediate the ring-closing metathesis of allyl amines 32 and 34. The reaction temperature could be lowered from 55 °C to room temperature and provided good yields of the product, albeit in longer reaction times. When the ligand was altered to contain an electron-rich benzylidene ring as in Ru-IV, the yield decreased to 61% (Figure 3c). A similar trend was observed in the synthesis of tetrahydropyridine 35. Use of electron-rich catalyst Ru-IV required elevated temperature and longer reaction times whereas the electron-deficient catalyst Ru-III gave similar yields at room temperature in only 7 hours.

Figure 3.

Role of electronics on benzylidene catalyst design.

The chelating heteroatom can also play a significant role in the catalyst stability. While most metal alkylidene catalysts perform well at room temperature or slightly elevated temperatures, some selected applications require that the catalyst have high thermal stability. For examples, the latent catalyst is activated, the methylidene species can rapidly form in solution. At higher temperatures, decomposition of the metal alkylidene is proven to lead to the formation of ruthenium hydride species that can result in the isomerization of olefinic bonds and the formation of byproducts such as 36 (Table 3).⁴² Ligands with a strongly coordinating heteroatom chelator such as oxygen,⁴³ nitrogen,⁴⁴ sulfur,⁴⁵ and selenium⁴⁶ provide greater thermal stability. For instance, Slugovc et al.⁴⁷ began exploring the use of ruthenium alkylidene catalysts bearing a chelating phosphine ligand Ru-V for both ring-opening metathesis polymerization and ring-closing metathesis. When this catalyst was used for the synthesis of pyrroline 33, it provided the desired product in >97% yield with reaction temperatures as high as 110 °C. Slugovc also found that the catalyst Ru-V could provide good yields of the product at lower temperatures when the solvent was switched to chloroform. At both elevated and lower temperatures, some traces of the isomerized product 36 were observed, but this could further be avoided by the introduction of a hydride scavenger such as benzoquinone (Table 3, entry 6).

Table 3.

Role of chelating ligands, solvents and additives in ring closing metathesis.

While the phosphine chelator provides increased stability of the active methylidene species at elevated temperatures, another continued goal is accessing a tunable, shelf-stable catalyst. Such a catalyst was serendipitously discovered during the synthesis of the macrocyclic backbone of the HCV therapeutic agent grazoprevir 37 (MK-5172, Figure 4).⁴⁸ The first quinoxaline Ru-VI was isolated as a byproduct of the metathesis reaction to obtain macrocycle 38. Quinoxaline Ru-VI was found to be stable in deuterated dichloromethane for up to 30 days, while the neat catalyst demonstrated shelf stability of up to 6 months. This stability is likely due the steric repulsion between the quinoxaline and the mesitylene rings preventing the rearrangement to the inactive cis isomer of the catalyst.⁴⁹

Figure 4.

Retrosynthetic strategy towards the synthesis of grazoprevir via an olefin ring-closing metathesis and subsequent reduction.

The effectiveness of the catalyst Ru-VI was tested against allyl amine 32 to access pyrroline 33. Catalyst Ru-VI gave complete conversion of sulfonamide 32 in 6 hours and 99% yield (Table 3, entry 7). Considering that electron-deficient catalysts are more active in ring-closing metathesis reactions, McLaughlin proposed⁴⁹ that the reaction rate could be increased through the protonation of the quinoxaline ligand with an acid catalyst. After examining both Brønsted and Lewis acids, AlCl₃ was found to give complete conversion and excellent yields of 33 in only 30 minutes (Table 3, entries 9 and 10).

As previously discussed, the decomposition of the active methylidene species in solution to ruthenium hydrides results in undesired isomerization of the olefins.⁴² While hydride scavengers such as quinones, acids, and chlorocatecholborane have proven effective in the prevention of undesired pathways, they do not necessarily prevent catalyst decomposition. Wozniak and coworkers⁵⁰ envisioned incorporating the quinone moiety into the catalyst structure, and synthesized the Hoveyda-Grubbs derivatives with benzoquinone HG-BQ, napthoquinone HG-NQ, and anthraquinone moieties HG-AQ. These catalysts were tested and compared to HG-II for the metathesis of tosylamide 39 and lactam 41 (Figure 5). The benzoquinone catalyst gave comparable yields to HG-II demonstrating that there is no loss in reactivity of the catalyst. These catalysts were also tested for their ability to prevent olefin isomerization by gauging their ability to selectively perform the homodimerisation of dodec-1-ene. While HG-II could perform the metathesis reaction in 69% conversion, it only had a selectivity of 70%. However, HG-BQ and HG-NQ were able to complete the reaction in 89% and 94% conversion, respectively. And not only did these catalysts outperform HG-II, but they both formed the desired product in 95% selectivity.

Figure 5.

(a) Application of quinone moiety as part of ligand design to function as a Ru-hydride scavenger; (b) Comparison of catalysts containing quinone motifs towards the synthesis of nitrogen heterocycles.

4. Indenylidene Complexes

One class of ruthenium catalysts that has become increasingly popular are ruthenium-pyridine adducts which have been referred to as the “third generation” of olefin metathesis catalysts.⁵¹ Such complexes are straightforward to access via ligand substitution by stirring in excess pyridine.⁵² One advantage of the pyridine ligands is that they are only weakly coordinating to the metal center, and initiation of the catalyst is much faster. Unfortunately, this means the catalysts show decreased stability overtime and are often outperformed by their tricyclohexylphosphine-containing analogues.⁵³ In an effort to increase the overall stability of these ruthenium-pyridine catalysts, indenylidene catalyst have been developed. This new class of catalyst is accessible from commercially available ruthenium precursors. These complexes exhibit enhanced stability in harsh reaction conditions, higher functional group tolerance, and greater stability on the bench.^53c,54,55

Nolan and coworkers⁵⁵ recently reported a series of indenylidene catalysts (Figure 6) to explore their reactivity and effects of sterics in ring-closing, enyne, and cross metathesis reactions. When tested against the diallylamine 32, the pyridine ligands outperformed the phosphine derivatives by up to 30% with catalyst loadings as low as 100 ppm (Figure 6b, entries 1, 2, and 3). The reaction also worked exceedingly well with the more substituted prenyl amine 43, however, higher catalyst loadings of 250 ppm were required due to the increase in steric bulk around the olefin.

Figure 6.

Studies into indenylidene catalysts: (a) catalyst design and synthesis; (b) analysis of new catalysts towards the synthesis of 3-pyrroline; (b) applications of indenylidene catalysts towards the synthesis of substituted pyrrolines and tetrahydropyridines.

Nolan, et al.⁵⁶ expanded studies on the indenylidene catalysts and synthesized catalysts Ru-XI, Ru-XII, and Ru-XIII. For the unsubstituted diallylamine 32, all the catalysts were able to provide the metathesis product 33 in excellent yields, however, a significant difference in reactivity between the more substituted catalysts was observed. The NHC ligands bearing increased steric bulk required longer reaction times of 3 hours (Figure 6b, entries 6 and 7). Comparing catalyst Ru-VII and Ru-VIII, the catalyst containing the pyridine ligand was more reactive than the phosphane. The reaction went to completion in 30 min with Ru-VII, while the reaction with Ru-VIII was completed in 15 min (Figure 6b, entries 4 and 5, respectively). When the amine bearing the substituted olefin 43 was subjected to metathesis conditions with catalyst Ru-X, the reaction required 3 hours and only resulted in 76% conversion of the starting material. This may be due to unfavorable steric interactions of the catalyst that slows down the reaction, but when the phosphane was replaced with the pyridine analogue Ru-XI, the reaction proceeded in >99% conversion in just 1 hour. The reaction also resulted in excellent yields of the pyrroline 44, while the more reactive analogue did give a slightly diminished yield.

Because less sterically encumbered NHC ligands such as Ru-VIII tend to be more reactive, albeit less stable overall, there have been significant efforts made to improve their general stability to make them more productive for catalysis. One factor that contributes to the decomposition of the less substituted (and more reactive) indenylidene catalysts is due to increased rotation around the aryl C–N bond.⁵⁷ In an effort to prevent this rotation, one strategy to increase stability is to add alkyl substituents to the backbone as illustrated in Ru-XI,⁵⁸ however bulky substituents are also known to lead to reaction inhibition. A viable alternative is to replace the benzylidene with the more stable indenylidene Ru-XIII.⁵⁹ When this catalyst was used to promote the ring-closing metathesis of substituted olefin 45 (Figure 6c), it out performed its benzylidene counterpart resulting in 98% of the desired product. The catalyst could also be applied towards the synthesis of substituted tetrahydropyridines 48 and 50, albeit with longer reaction times and in diminished yields.

Another new and exciting field is the conversion of the neutral NHC catalysts to their corresponding cationic derivatives. Only a handful of groups have developed systems that utilize cationic catalysts in ring-closing metathesis, but the catalysts can be a powerful tool for accessing challenging metathesis products.⁶⁰ A recent example comes from Cazin, et al.⁶¹ with the synthesis of cationic catalyst Caz-1⁺ with a phosphite ligand. Interestingly, similarly reported cationic ruthenium species which are 14 electron complexes tend to dimerize to the more stable 16 electron species. In the case of cationic species Caz-1⁺, no dimerization is observed due to the formation of an unusual sawhorse geometry. This makes the catalyst especially stable in solution. While studying ability of the catalyst Caz-1⁺ to convert tosylamine 45 to the pyrroline 46 compared to G-II and HG-II at 140 °C, G-II and HG-II both displayed rapid decomposition and only 40% conversion of the tosylamine. Interestingly, Caz-1⁺ displayed higher thermal stability and reached 90% conversion after only 10 min. This catalyst was used to access other challenging substrates including tetrahydropyridines 35 and 50 in 97% and 85% yield, respectively.

5. Unsymmetrical NHC Ligands

While NHC ligands have led to a new generation of metal alkylidene catalysts that enable access to new, more complex structures, an ongoing challenge in ring-closing metathesis reactions is performing them asymmetrically to access enantioenriched products.⁶² While asymmetric olefin metathesis reactions have been successfully employed in ring-opening cross metathesis, controlling the olefin geometry continues to be a challenge in asymmetric ring closing metathesis. Key strategies for applying ring-closing metathesis are through the kinetic resolution of dienes or the desymmetrization of meso-trienes, particularly when the unique olefin is less bulky than the enantiotopic olefins.⁶² Asymmetric ring closing metathesis has been successfully applied to a variety of systems using molybdenum alkylidene catalysts (Figure 7).⁶³ Specifically Hoveyda and coworkers were able to utilize molybdenum catalyst Mo-I to access both cyclic six-, seven-, and eight-membered rings and bicyclic amines in good yields with up to 98% ee.

Figure 7.

(a) Development of chiral molybdenum catalysts and (b) applications of chiral alkylidenes towards the synthesis of enantioenriched tetrahydropyridines.

Grubbs and coworkers⁶⁵ postulated that for prochiral trienes to proceed in an enantioselective metathesis reaction, the pathway can either consist of an irreversible alkylidene formation of one to the enantiotopic olefins, or the alkylidene can from the unique olefin which can then cyclize with one of the other enantiotopic olefins. Cavallow⁶⁶ performed computational studies on the origin of stereoselectivity and found that the non-reacting olefin is oriented in pseudo-equitorial and pseudo-axial positions in the respective diastereomeric transition states for cyclization. With larger substituents, higher selectivities are expected due to the large energy difference between the two configurations. Grubbs hypothesized that utilizing adamantyl catalyst Ru-XIV would promote the alkylidene formation with the unique olefin (Figure 8). The utility of this catalyst was first probed against tosylamine 54, and the reaction gave good yields of the pyrroline 55 with modest enantioselectivity. The reaction also works well in the synthesis of tetrahydropyridine 56 providing the product with similar yield and slightly higher ee. It is proposed that the stereoinduction is due to the possible transition state in which the vinyl groups are in the pseudo-equitorial position.

Figure 8.

Application of adamantyl-containing ruthenium catalyst towards the synthesis of chiral pyrrolines and tetrahydropyridines.

Unsymmetrical NHC catalysts are also of interest for addressing the challenge of achieving high selectivity in certain metathesis reactions. Such complexes as demonstrated in the previous example can significantly alter the stability of key intermediates thus introducing the ability to control the reaction outcome. In order to expand this class of catalyst, Grisi and Grela⁶⁷ set out to identify new ways to enhance the stability of the unsymmetrical NHC catalysts by introducing substituents to the ligand backbone. In particular, syn- and anti-complexes Ru-XV and Ru-XVI were synthesized, as well as catalysts containing the N-neopentyl backbone (Figure 9). After 1 week, the syn complexes Ru-XVa and Ru-XVb were almost completely decomposed, however, the anti-complexes Ru-XVIa and Ru-XVIb proved more resistant to decomposition and were stable for up to 10 days. Furthermore, bulkier N-alkyl groups tended to help stabilize the catalyst.

Figure 9.

(a) Examples of ruthenium alkylidene catalysts containing unsymmetrical NHC ligands, (b) effects of olefinic substitution on the kinetic profiles of ruthenium catalysts.

In terms of overall reactivity, the N-alkyl substituent did not play a major role in the catalyst reactivity (Figure 9b). When less substituted olefin 32 was subjected to a variety of catalysts with varying N-alkyl substituents, no significant difference in yield was observed, however, the bulkier catalysts required slightly longer reaction times. The backbone substituents played a more significant role in reactivity, as the anti-complexes Ru-XVIa and Ru-XVIb gave higher yields than their syn-counterparts. In the presence of bulkier olefins, a slight decrease in reactivity was observed which is attributed to unfavorable steric interactions.⁶⁸

6. Carbonyl-Olefin Metathesis

While traditional olefin-olefin metathesis has been a powerful tool in a wide range of applications including natural product synthesis, materials and polymers, medicines, and fine chemical synthesis, a continued challenge is finding inexpensive, sustainable catalysts. Furthermore, the application of olefin-olefin metathesis often requires the prerequisite synthesis of the olefin precursors through olefination reactions which generate stoichiometric quantities of waste. An intriguing alternative that has emerged in recent years is the application of carbonyl-olefin metathesis. In 1993 Grubbs and Fu reported the synthesis of five-, six-, and seven-membered cycloalkenes (Table 4) by subjecting various olefinic ketones to Schrock’s catalyst to perform the carbonyl-olefin metathesis sequence via the formation and fragmentation of intermediate oxametallacycles 59^.69 However, stoichiometric quantities of Schrock’s catalyst were required due to the formation of a metal-oxo species 61, which is difficult to reduce back to the active metal alkylidene.

Table 4.

Early examples of carbonyl-olefin metathesis applied towards the synthesis of cycloalkenes.

The first reported example of carbonyl olefin metathesis in the presence of amines came from Zhou and Rainier.⁷⁰ Their strategy focused on the formation of titanium alkylidenes. This method was used to synthesize five-, six-, and seven-membered rings through a carbonyl-olefin metathesis pathway (Scheme 4a). Mechanistic insights for this synthesis came when tosylamine 71 was subjected to the metathesis conditions resulting in a mixture of tetrahydropyridine 72A and acyclic enamide 72B. When the acyclic enamide was resubjected to the reaction conditions, none of the cyclic product was formed. This supports the hypothesis that the cyclic product proceeds via a carbonyl-olefin pathway as opposed to an olefin-olefin metathesis mechanism (Scheme 4b).

Scheme 4.

Titanium-mediated carbonyl-olefin metathesis for the synthesis of unsaturated nitrogen heterocycles (a) quinolines and (b) tetrahydropyridines.

Recently, our group reported the development of a Lewisacid catalyzed carbonyl-olefin metathesis reaction which proceeds via activation of the carbonyl to promote the formation of intermediate oxetanes. This approach has since been applied to a variety of systems, but amine-containing systems proved to be challenging. Under initial reaction conditions, we found that allylamines (Scheme 5c) only underwent the cyclization in about 37% yield with substoichiometric quantities of the Lewis acid catalyst.⁷¹ The Li group⁷² found that by replacing the prenyl fragment with a styrenyl olefin and adding superstoichiometric amounts of allyl-TMS they could isolate up to 99% yield of the pyrrolines (Scheme 5b); however, this method did not work as well in the synthesis of tetrahydropyridines. Li and coworkers hypothesized that the allyl-TMS plays a dual role under optimized reaction conditions by coordinating to the iron catalyst and promoting the formation and subsequent fragmentation of the oxetane intermediate as well as acting as a benzaldehyde scavenger.

Scheme 5.

(a) Initial development of an iron-mediated carbonyl-olefin metathesis reaction; (b) evaluation of allyl-TMS as an additive to promote the desired transformation; (c) recent developments towards the application of carbonyl-olefin metathesis towards the synthesis of chiral pyrrolines and tetrahydropyridines.

During the course of our studies, we hypothesized that the diminished yields for nitrogen heterocycles were likely due to the Lewis basic amine acting as a competitive binding site that prevents the Lewis acid catalyst from promoting the desired transformation.⁷³ In order to prevent the sequestration of the catalyst, we replaced the tosyl protecting group with the more electron-withdrawing 4-trifluoromethylbenzenesulfonyl (^FTs) derivative 75 (Figure 10a). When 75 subjected to 50 mol% of FeCl₃, we were able to generate pyrroline 77 in 99% yield. Computational analyses of the two derivatives also revealed that 75a was 1.9 kcal/mol higher in energy compared to tosylamine 75b resulting in preferential binding of the Lewis acid to the carbonyl oxygen.

Figure 10.

(a) Optimized reaction conditions with phenylalanine-derived 76; (b) computational studies into Lewis acid coordination to the Lewis basic sulfonamide versus the aryl ketone.

With this strategy in hand, we were able to access a wide range of pyrrolines (Table 5). The reaction worked well for both electron-rich and electron-deficient aryl ketones, but the addition of more Lewis basic sites could result in slightly lower yields. This reaction all worked well for a variety of steric motifs ranging from the unencumbered glycine derivative 78 to the disubstituted amino-isobutyric acid derivative 82. This reaction could even be applied to the sterically bulky napthyl 86 in good yields. It is also interesting to note that when compared to the corresponding tosyl-derivatives, the ^FTs-protected substrates consistently outperformed the tosyl-protected materials by up to 30% yield.⁷³

Table 5.

Application of the carbonyl-olefin metathesis reaction for the synthesis of chiral 3-pyrrolines.

Conditions: Reactions were performed with 0.20 mmol of substrate and 0.5 equiv. of FeCl₃ in dichloroethane [0.01 M]. The reaction was stirred at 0 °C for 1 hr, and then allowed to warm to room temperature and stirred for 24 h.

After applying the Lewis acid-mediated carbonyl-olefin metathesis reaction towards the synthesis of pyrrolines, we set out to apply this strategy to the more challenging tetrahydropyridines derivatives (Table 6).⁷⁴ Reaction optimization revealed that higher temperatures were required, but the reaction exhibited a similar broad substrate scope. It could be applied to sterically unencumbered substrates such as the glycine derivative 87 in good yields, as well as more sterically bulky systems. Furthermore, the reaction exhibited the same tolerance for electron poor systems 93 and 94 as well as in the systems of heteroaromatic substrates 91 and 95. The reactions did exhibit a similar decrease in yield in the presence of Lewis basic sites 92, however we found that longer reaction times can help overcome this limitation.

Table 6.

Evaluation of substrate scope for iron-catalyzed carbonyl-olefin metathesis.

Conditions: Reactions were performed with 0.20 mmol of substrate and 0.3 equiv. of FeCl₃ in dichloroethane [0.01 M]. The reaction was stirred at 84 °C for 24 h.

7. Conclusions

Electron-rich amines have caused significant challenges for the application of olefin metathesis. Strategies for attenuating their Lewis basicity in order to promote desired metathesis reactions have enabled access to more complex ring systems including the incorporation of protecting groups and electron-withdrawing groups as well as sterically bulky substituents to prevent undesired coordination to the metal alkylidene catalysts; however, these can add undesired steps to overall synthetic strategies. Recent developments in reaction and catalyst design have enabled the application of ring-closing metathesis to more challenging nitrogen heterocycles without the need for protection or substrate alteration.

Moving forward, advances in metathesis will enable access to more complex, chiral nitrogen heterocycles under greener reaction conditions. The newly developed carbonyl-olefin metathesis reaction eliminates the need for an additional olefination step and utilizes iron as an inexpensive, earth abundant catalyst. While current carbonyl-olefin metathesis methods have been successfully applied towards the synthesis of chiral pyrrolines and tetrahydropyridines, there are many areas where this approach could prove fruitful. The application of this method to larger ring systems would be a significant advancement that would enable rapid access to bioactive molecules. Furthermore, a protecting group-free strategy would allow for rapid access to the desired nitrogen heterocycles without requiring additional protection/deprotection steps. The application of Lewis acid-catalyzed carbonyl-olefin metathesis towards the synthesis of more complex, protecting group-free nitrogen heterocycles will mark a significant advancement in the metathesis toolkit.

Table 2.

Investigations into the olefin subunit for the synthesis of cyclic alkenyl bromides.

entry	substrate	R1	R2	time	%yield
1	19a	H	H	24 h	0
2	19b	Ph	H	24 h	0
3	19c	H	Ph	30 min	>98

Conditions: All reactions were performed in benzene with 0.1 M substrate concentration. Yields are based on NMR analysis

Biosketches

Corinna S. Schindler received her diploma in chemistry from the Technical University of Munich. After a research stay with K. C. Nicolaou at the Scripps Research Institute, she joined the group of Erick Carreira at ETH Zürich for her graduate studies. She then returned to the US to conduct postdoctoral studies with Eric Jacobsen at Harvard before starting her independent career at the University of Michigan in 2013.

Emilia J. Groso was born in Nevada in 1990 and received her BS in chemistry from the University of Nevada, Reno under the supervision of Professor Robert S. Sheridan in 2013. She is currently a graduate student at the University of Michigan working with Professor Corinna S. Schindler on the development of new methods towards the synthesis of nitrogen heterocycles.