Regioselective Crossed Aldol Reactions under Mild Conditions via Synergistic Gold-Iron Catalysis

SUMMARY

A synergistic gold/iron catalytic system was developed for sequential alkyne hydration and vinyl gold addition to aldehydes or ketones. Fe(acac)₃ was identified as an essential co-catalyst in preventing vinyl gold protodeauration and facilitating nucleophilic additions. Effective C-C bond formation was achieved under mild conditions (r.t.) with excellent regioselectivity and high efficiency (1% [Au], up to 95% yields). Extending reaction scope to intramolecular fashion achieved successful macrocyclization (16-31 ring sizes) with excellent yields (up to 90%, gram-scale) without extended dilution (0.2 M), which highlighted the great potential of this new crossed-aldol strategy in challenging target molecule synthesis.

INTRODUCTION

Aldol reaction is a classic C-C bond-forming transformation in organic synthesis.^1–5 Conventional aldol reactions often require strong bases to access the needed enolates, which limit the substrate scope. Moreover, when two different carbonyl groups involved, the reactions often suffer from unsatisfactory chemo- (self vs crossed aldol) and regioselectivity (thermodynamic vs kinetic enolate). As an improved modification, Mukaiyama aldol used silyl enol ethers as nucleophiles.^6–11 However, the requirements of stoichiometric silyl reagents and harsh reaction conditions greatly reduced the atom economy of this method as practical synthesis especially for challenging substrates (Scheme 1A). Although, efforts have been made to improve reaction performance, such as applications of Lewis acids^2–12–15 and amine organo-catalysts^16–23, challenges associated with regioselectivity and reaction efficiency remain for this fundamental but important transformation. Novel strategies with high efficiency are highly desirable.

Scheme 1.

Formal crossed aldol reaction with gold/iron dual catalysis

Over the past two decades, homogeneous gold catalysis has drawn great attentions due to its unique ability of activating alkynes and allenes under mild conditions.^24–32 As shown in Scheme 1B, one important intermediate involved is vinyl gold A upon nucleophilic addition towards gold-alkyne π-complexes. In most cases, a rapid protodeauration takes place, converting the C-Au bond into the C-H bond.^33–41 Up to date, alternative reactivity of vinyl gold intermediate A has been rarely explored.^42–51 Notably, incorporation of electrophilic carbon atoms, such as carbonyl or carbonyl equivalents, in gold catalysis has been explored.^52–56 The pioneering work of Ito, Sawamur, and Hyashi has demonstrated the first gold-catalyzed asymmetric aldol reaction.^57–58 Herein, we report the first successful example of vinyl-gold addition towards carbonyl, giving formal crossed aldol products with high efficiency (1% gold loading, up to 95% yields) and excellent regioselectivity (exclusively kinetic enolate addition). Moreover, reaction under intramolecular fashion gives successful synthesis of macrocyclic compounds, with ring size between 16-31 under mild conditions (r.t., up to 90% yields) with no need of extended dilution (Scheme 1C).

RESULTS and DISCUSSION

Our interest in pursuing vinyl gold intermediate as nucleophile was originated from the recent discovery of successful oxazolyl aldehydes synthesis via gold/iron dual catalysis.^59–62 As shown in Figure 1A, treating alkyne B with the mixture of gold and iron catalysts gave oxazolyl aldehydes D. Monitoring the reaction revealed interesting kinetic profiles: alkyne B under [Au] and [Fe] combined catalysis conditions gave a much faster reaction rate (formation of D) than reaction of alkene C with [Fe] (k₁/k₂>10). Since A is the intermediate upon cyclization, these kinetic studies clearly suggested that vinyl gold A is more reactive than enol ether C toward iron activated oxygen radical. Considering that the oxygen radical is electron deficient, it is reasonable to conclude that vinyl gold A is more electron rich than vinyl ether. Therefore, we wondered if intermediate A could be used to react with carbonyl as a new approach for C-C bond synthesis. To verify this hypothesis, reactions between aldehyde 2a and phenylacetylene were conducted. Unfortunately, only Telle hydration was obtained with no desired aldol products observed under various conditions (see detailed screening conditions in SI, Figure 1B).

Figure 1.

Electron-rich vinyl gold as potential nucleophile

Screening of alkynes was also performed, including propargyl amides B, hydroxyl-alkynes, and keto-alkynes 1a. No desired crossed aldol products were observed in all tested cases under gold catalytic conditions (See SI for details). Notably, a faster hydration reaction rate for 1a was observed compared with simple alkyne. This is likely due to carbonyl group neighboring group participation toward gold activated alkyne (5-exo-dig).⁶³ Alkyne hydration through vinyl gold protodeauration was the critical challenge for the attempt to apply vinyl gold as nucleophiles. To compete with undesired hydration, we applied various metal salts as co-catalysts for aldehyde 2a activation, including Sc(OTf)₃, Ga(OTf)₃, FeCl₃ etc. Still, no desired product was observed. Interestingly, when Fe(acac)₃ was used as co-catalyst, desired aldol product 3a was received in 52% yield along with 45% hydration product 4 (Figure 1C). Notably, slower hydration was observed while treating alkyne with mixture of [Au] and Fe(acac)₃ (no aldehyde presented), suggesting the unique role of Fe(acac)₃ in preventing alkyne hydration side reaction. To improve reaction yields, comprehensive condition screenings were performed, including different gold catalysts, solvents, co-catalyst loading, and the amount of water. Finally, the reaction reached 100% 1a conversion with desired aldol product 3a in 91% isolated yield under optimal conditions: 1% CyJohnPhosAu(TA-Me)OTf, 2% Fe(acac)₃, and 10% LiClO₄ in EtOAc with 5 eq. of water. Results of some representative conditions are summarized in Table 1.

Table 1.

Reaction optimization^a

Entry	Variation from standard conditions	3a (%)	4 (%)
1	None	95(91b)	4
2	[Au]=PPh3AuNTf2	55	40
3	[Au]=IPrAuNTf2	60	33
4	[Au]=JohnPhosAuNTf2	68	30
5	[Au]=CyJohnPhosAuNTf2	70	25
6	[Au]= CyJohnPhosAu(TA-H)OTf	88	10
7	No Fe(acac)3	0	>95%
8	1% Fe(acac)3	75	20
9	4% Fe(acac)3	90	6
10	2% Fe(dbm)3c	28	70
11	2% Fe(hfaa)3d	0	>95
12	1 equiv. H2O	60	35
13	10 equiv. H2O	88	8
14	No LiClO4	91	8
15	Mn+(acac)n instead of Fe(acac)3. M=Co, Ni, Ga, Al, Sn, In, Sc, etc.	<42	>50
16	Other solvents	<88	>10

^aConversions and yields were determined by ¹H NMR using 1,3,5-trimethoxybenzene as the internal standard.

^bIsolated yields.

^cdbm = 1,3-diphenyl-1,3-propanedionate.

^dhfaa = hexafluoroacetylacetone

Based on condition screening, the primary ligand on gold is essential (entries 1-6). CyJohnPhos was revealed to give the best result. Triazoles, especially N-Methyl benzotriazole (TA-Me), were proved to be suitable dynamic ligands to suppress hydration.^64–70 This could be explained by its good binding ability toward [L-Au]⁺, preventing the formation of [L-Au-H₂O]⁺ that leads to alkyne hydration under neutral conditions.⁷¹ Fe(acac)₃ was crucial for optimal reactivity. Other iron salts, such as FeCl₂, FeCl₃, Fe(OTf)₃, have been tested and gave exclusive hydration product 4. Fe(III) salts with different acac-type ligands were also tested. With dbm ligand, reduced yield of 3a was observed, likely due to increased acidity. The more acidic hfaa ligand gave exclusive hydration product 4 (entry 10-11). Other metal M(acac)_n complexes, such as Co²⁺, Ga³⁺, and Al³⁺, led to reduced yields of 3a, which highlighted the unique role of Fe(acac)₃ in this dual metal catalytic system. Notably, Li⁺ could promote carbonyl activation with slightly improved yield (entry 14). Clearly, the amount of water is critical. While water promotes the undesired hydration, it is necessary to use water to trigger hemiacetal formation after carbonyl cyclization (formation of vinyl gold, Figure 2B). Detailed screening revealed 5 eq. of water in ethyl acetate gave the optimal result with 3a obtained in excellent yield (95%). With the optimal conditions in hand, we evaluated substrate scope. The results are summarized in Table 2.

Figure 2.

Au/Fe catalyzed cross aldol and macrocyclization

Table 2.

Reaction scope of aldehydes and ketones^a

^aStandard conditions: 1% CyJohnPhosAu(TA-Me)OTf, 2% Fe(acac)₃, 10% LiClO₄, 5 eq. water was added to an ethyl acetate (0.45 mL) solution of aldehyde 1 (0.6 mmol) and alkyne 2a (0.3 mmol), and reaction was run at room temperature for 12 h, isolated yield.

^b5% CyJohnPhosAu(TA-Me)OTf, 10% Fe(acac)₃, 10% LiClO₄, 0.9 mmol aldehyde/ketone was used.

Various aldehydes and ketones were tested to react with alkyne 1a. The reaction generally worked well for various benzaldehydes, giving crossed aldol products 3 in good to excellent yields in almost all cases (3a-3r). Notably, with the unique vinyl gold mechanism, only kinetic enolate addition products were received in all cases. The structure of the resulting β–hydroxy-ketone was confirmed by X-ray crystallography (3c). Heteroaromatic aldehydes were also tolerated (3aa-3al). Particularly, substrates containing pyridine (3ac, 3ad) and triazole (3ah, 3ai) are all feasible despite of their strong coordination effect with metal catalysts. According to literature, with enone or enal electrophiles, Mukaiyama aldol dominantly goes through 1,4-addition.^72–76 With this gold/iron system, exclusive 1,2-addition products were obtained (3s-3u), suggesting an orthogonal selectivity for synthetic applications. Moderate yield was achieved with aliphatic aldehyde (3ba) due to reduced reactivity. Ketones are not good electrophiles under neutral conditions, good to modest yields were obtained with various ketones (3bb-3bj). Notably, LiClO₄ is essential for this transformation. Only trace amount of desired product (<5%) was obtained in the absence of LiClO₄, suggested the important role Li⁺ for carbonyl activation. To the best of our knowledge, this was the first successful example of catalytic crossed aldol reaction with ketone under such mild and neutral conditions. The scope of alkyne was also evaluated by reacting with 4-bromo-benzylaldehyde 2a under optimal conditions. A series of keto-substituted alkynes were prepared. The results are summarized in Table 3.

Table 3.

Reaction substrate of alkynes^a

^aStandard conditions: 1% CyJohnPhosAu(TA-Me)OTf, 2% Fe(acac)₃, 10% LiClO₄, 5 eq. water was added to an ethyl acetate (0.45 mL) solution of aldehyde 1 (0.6 mmol) and alkyne 2a (0.3 mmol), and reaction was run at room temperature for 12 h, isolated yield.

^b45 °C for 6 h.

^c50 °C.

^d35 °C for 24 h.

Substituents on carbonyl group showed dramatic impact on this transformation. In general, electron-deficient aryl keto-alkynes (5a-5d) gave crossed aldol products in excellent yields. Slightly reduced yields were obtained with electron-rich aryl substrates (5e-5i, 5n). Aliphatic substituted keto-alkynes were also suitable for this reaction (5j-5m, 5r). Substituents such as cyclopropyl, carbon-carbon triple bond and allyl group could remain intact in the reaction, which indicated good function group tolerance and mild conditions of this new gold/iron catalytic system. Internal alkynes also worked well for this reaction, giving crossed aldol products in moderate yields and low d.r. (5o, 5p). Comparing with terminal alkynes, activation of less reactive internal alkynes requires higher temperature (50 °C), which adversely accelerates hydration. Substrates with different linkages between alkyne and carbonyl groups, including β-ester (5q) and α,β-aryl substituents (5s and 5t) were tested. All these substrates delivered the desired products in good yields. Overall, this new dual gold/iron catalytic system showed superior regioselective and significantly improved reactivity for crossed aldol reaction under mild conditions.

To rationalize the role of Fe(acac)₃ in this transformation, several experiments were performed. First, conducting the reaction with Na(acac) instead of Fe(acac)₃ slowed down the reaction significantly (<15% 1a conversion, 4 h, Figure 2A) with no desired product 3a formed. This is likely due to a strong coordination of acac- anion toward [L-Au]⁺.^77–80 As precedence, an X-ray crystal structure of corresponding C-Au(I) bond in acetone complex has been reported.⁸¹ Interestingly, the addition of a catalytic amount of Fe(OTf)₃ retriggered the system, giving 3a in 30% yield. Monitoring the reaction with nESI-MS gave a diagnostic signal with m/z = 812.1266, corresponding with the formation of gold/iron complex (confirmed by CID, see SI). Based on these results, a plausible reaction mechanism is proposed as shown in Figure 2B.

The important step of this reaction is the addition of vinyl gold F toward carbonyl electrophiles. To conform vinly gold F as the key intermediate, we treat hydration product 4 with aldehyde under the optimal reaction condition (Au/Fe). No reaction was observed over extended reaction time (48 h). This result clearly indicated the importance of combing Au and Fe for this transformation. Moreover, we monitored the reaction with reactive MS. While treating vinyl gold intermediate with Fe complexes, no corresponding vinyl-iron compelxes were observed, suggesting that vinyl gold transmetalation with iron complexes is highly unlikely and vinyl gold is likely the nucleophile for the observed aldehyde addition (see SI for details).⁸² Due to the low concentration of H⁺ under neutral conditions, the most reactive proton source is [L-Au-H₂O]⁺, formed from water addition to [L-Au]⁺. Therefore, reducing the overall concentration of free [L-Au]⁺ is crucial for preventing protodeauration. As we have demonstrated previously, gold complexes with dynamic L-ligand (DLL, such as 1,2,3-triazole) could activate alkyne through dynamic concerted coordination-dissociation without the formation of [L-Au]⁺. ⁷⁰ Although further evidences are needed, it is reasonable to rationalize that Fe(acac)₃ serves as another special DLL and reduces the amount of [L-Au]⁺ in the system. Upon alkyne substitution, gold-alkyne π-complexes were formed for rapid nucleophilic addition (cyclization). As the result, combination of [L-Au]⁺ and Fe(acac)₃ achieved the needed slow protodeauration and allowed us to access vinyl gold reactivity (as Nu) without severe hydration.

Encouraged by the success of achieving cross aldol reaction under such mild conditions, we extend our attention into another challenging transformation: macrocyclization via C-C bond formation. Macrocycles are a group of important and valuable compounds in chemical, material, and biological fields.^83–86 However, protocols of macrocyclization using irreversible C-C bond forming approach are rarely reported.^87–93 In addition, to avoid undesired intermolecular polymerization, macrocyclization reactions often performed under diluted conditions, which obstacle their application in the large-scale synthesis. Thus, conducting macrocyclization via irreversible C-C bond formation under conventional concentration (no extended dilution) is highly desirable.

One important factor to improve macrocyclization performance, especially at high concentration, is to effectively reduce overall structure flexibility: having reaction units to be pre-organized under a favored conformation. As shown in Figure 2C, the formation of vinyl gold via a cyclic hemi-acetal moiety greatly increases overall structural rigidity, aligning C=O with vinyl gold at favorable position. We postulated that this unique activation mechanism could be applied into challenging macrocyclization. To testify this idea, substrates bearing alkyne and aldehyde were prepared. To our delight, macrocyclization were successfully achieved under room temperature (25 °C). Impressively, no polymerization was observed even at relative high reaction concentration (0.2 M). Macrocyclization products with ring sizes from 16-24 atoms were prepared with yields between 65% and 90% in gram scale (Figure 3A). Notably, besides intramolecular reaction, intermolecular condensation between diyne and di-aldehyde could also be achieved, giving the formation of 31-membered ring structure with 45% yield (Figure 3B).

Figure 3.

Macrocyclization via C-C bond forming reactions

In summary, we disclosed herein a novel gold/iron dual catalytic system to promote alkyne hydration and sequential aldol addition under mild conditions. The key of this design is to access vinyl gold nucleophilicity by avoiding undesired protodeauration. The overall transformation is highly efficient with low catalyst loading, mild condition, large substrate scope and excellent aldol regioselectivity. Moreover, based on the pre-cyclic conformational control, this strategy was further extended as a new macrocyclization method through irreversible C-C bond construction at high concentration. Clearly, this work greatly enriched cationic gold catalysis by allowing vinyl-gold as active intermediate for sequential transformations without undesired protodeauration. Other new transformations using this concepts and applications of this method for the preparation of complex molecules are expected and currently undergoing in our lab.

EXPERIMENTAL PROCEDURES

Full experimental procedures are provided in the Supplemental Information.

The Bigger Picture.

A novel method for facile C-C bond construction is always attractive to the community. Cross aldol reaction is one of the fundamental but essential transformations to achieve β-hydroxyl ketone moieties, which are valuable functional groups in the chemical, material, and biological fields. Herein, we describe a synergistic gold/iron catalytic system, which employs alkynes as ketone surrogates under sequential alkyne hydration and vinyl gold addition to aldehydes or ketones. A broad substrate scope was obtained under mild conditions (r. t.) with excellent regioselectivity and high efficiency (1% [Au], up to 95% yields). Notably, this protocol not only provides a practical solution to access vinyl-gold reactivity by avoiding frequently occurring protodeauration, but also achieves successful macrocyclization (16-31 ring sizes, up to 90%, gram-scale) without extended dilution (0.2 M operating concentration). It is expected that this strategy will greatly extend general reaction mode of gold catalysis for challenging molecule synthesis toward biological and medicinal research.

DATA AND SOFTWARE AVAILABILITY

The structure of 3a, 3d, 5g, 5s, 7a’ and 7e reported in this article has been deposited in the Cambridge Crystallographic Data Centre.