Atom-Economical Cross-Coupling of Internal and Terminal Alkynes to Access 1,3-Enynes

Abstract

Selective carbon–carbon (C–C) bond formation in chemical synthesis generally requires pre-functionalized building blocks. However, the requisite pre-functionalization steps undermine the efficiency of multi-step synthetic sequences, which is particularly problematic in large-scale applications, such as in the commercial production of pharmaceuticals. Herein, we describe a selective and catalytic method for synthesizing 1,3-enynes without pre-functionalized building blocks. This method is facilitated by a tailored P,N-ligand that enables regioselective dimerization and suppresses secondary E/Z-isomerization of the product. The transformation enables several classes of unactivated internal acceptor alkynes to be coupled with terminal donor alkynes to deliver 1,3-enynes in a highly regio- and stereoselective manner. The scope of compatible acceptor alkynes includes propargyl alcohols, (homo)propargyl amine derivatives, and (homo)propargyl carboxamides. The reaction is scalable and can operate effectively with as low as 0.5 mol% catalyst loading. The products are versatile intermediates that can participate in various downstream transformations. We also present preliminary mechanistic experiments that are consistent with a redox-neutral Pd(II) catalytic cycle.

Graphical Abstract

Introduction

Catalytic methods that couple two distinct carbogenic fragments in a selective fashion constitute a core technology in organic synthesis with important applications in the pharmaceutical industry.¹ Classical palladium- and nickel-catalyzed C–C cross-coupling reactions between organohalides and organometallic reagents (Scheme 1A) are widely used but require pre-functionalized coupling partners that must be prepared in advance via multiple non-strategic steps, detracting from the overall efficiency of the process.

Scheme 1. Comparison of different approaches to C–C cross-coupling.

Thus, the development of C–C cross-coupling alternatives that directly employ unfunctionalized substrates and enable access to high-value products is of vital importance.² To this end, integrating π-systems (e.g., alkenes and alkynes) as cross-coupling components (Scheme 1B) is an attractive approach given the ability of π-systems to provide potential energy to the reaction, the ambiphilic reactivity profiles of the coupling partners, and the widespread availability of the feedstock chemicals. Specifically, the cross-coupling of two different alkynes—a terminal donor alkyne capable of forming a metal–acetylide in situ and an internal acceptor alkyne capable of undergoing hydrofunctionalization—represents a promising strategy for C(sp)–C(sp²) bond formation. If fully developed, this transformation would provide direct access to a range of 1,3-enynes without relying on pre-functionalization events that are typically required in the state-of-art methods, such as Sonogashira coupling.^4,9,10 The ability of 1,3-enynes to participate in diverse downstream transformations makes them extremely valuable building blocks in organic synthesis,^3,4 In addition, the 1,3-enyne moiety is found in bioactive natural products,^5,6 clinical therapeutics,⁷ and supramolecular assemblies.⁸ Herein, we describe a ligand-promoted, palladium-catalyzed method to couple donor alkynes with a variety of acceptor alkynes that takes advantage of coordination of a native Lewis basic functional group on the acceptor to enhance reactivity and control selectivity.

Previous research has demonstrated the viability of the envisioned coupling, while also illustrating challenges to be anticipated in pursuing a general terminal donor/internal acceptor cross-coupling (Scheme 2A). Trost reported pioneering work on redox-neutral, palladium-catalyzed terminal alkyne homo-coupling in 1987,¹² with improved scope and selectivity subsequently being realized by Trost, Pfaltz, Gevorgyan, and others over the ensuing decades.^3,11–29 Cross-coupling between a terminal alkyne and an electronically activated (i.e., conjugated) alkyne has also been described.^11–29 Relevant to the approach described herein, Trost has also described examples in which silyl-substituted and terminal propargyl alcohols are coupled with terminal alkynes, leading to C(sp)–C(sp²) bond formation proximal to the alcohol.¹⁴ While useful in their own right, existing methods are limited in scope, regioselectivity, stereoselectivity/specificity, and efficiency. Therefore, widespread adoption in preparative syntheses has been hampered.

Scheme 2. Approaches to redox-neutral alkyne coupling and optimization of conditions.

^{a 1}H NMR yield with CH₂Br₂ as internal standard. n.r. = no reaction. ^b Determined by ¹H NMR analysis of crude reaction mixture. r.r. = regioisomeric ratio. ^c Isolated yield.

We envisioned that the challenges outlined above could be surmounted by adopting a substrate directivity approach, in which a Lewis basic site on the acceptor alkyne—ideally a native functional group, like a free alcohol, would coordinate to the metal catalyst to facilitate downstream elementary steps. In particular, the bidentate coordination between the catalyst and acceptor alkyne would serve to activate the π-system through induced π-Lewis acid activation. An alkynylpalladium(II) species is formed from the donor alkyne followed by directed 1,2-migratory insertion of the alkynylpalladium(II) species to the acceptor alkyne. This process would lead to the regioselective formation of the alkenyl-palladacycle intermediate through stabilization of one of regioisomeric transition states (Scheme 2B). The chelation-stabilized intermediate structure would then undergo protodepalladation to close the catalytic cycle. This hypothesis builds on our previous work using bidentate directing groups for alkyne hydrofunctionalizations^30–31 but would obviate auxiliary attachment and removal steps.³²

Results and Discussion

1. Reaction optimization.

To reduce this idea to practice, we selected 2-propyn-1-ol, an internal propargyl alcohol, as a model acceptor alkyne and TIPS-acetylene as the donor alkyne (Scheme 2C). Pilot experiments with various ligand/pre-catalyst combinations (data not shown) indicated that the initial cross-coupling is fast and quickly followed by E/Z-isomerization. Hence, we deliberately used an extended reaction time of 16 h in order to identify conditions and ancillary ligands that would allow selective coupling while suppressing secondary isomerization. In summary, we found that phosphine ligands are essential for the reaction, as shown in previous studies.³ In previous methodology reported by Trost, L5 was the optimal ligand,^12–14 but in this case barely provided any regioselectivity differentiation (Entry 5). In contrast, the reaction was highly regioselective when P,N-bidentate ligands were used (Entry 2–4). After examining L2–L4, we found that a bulky and rigid N-coordination arm is beneficial for favoring the syn-product by suppressing secondary isomerization. In particular, the phosphinoimidazoline ligand L1, which is derived from diphenylethlenediamine and was previously developed at Boehringer Ingelheim,^33–37 was identified as the best ligand in terms of regio- and stereoselectivity. Further screening showed that the palladium source and the additive are important for reactivity (Entry 6–12). The combination of Pd(dba)₂ and ammonium acetate provides the best yield. These general trends also held with a representative propargyl carboxyamide substrate that is more prone to secondary E/Z isomerization (see SI).

2. Substrate scope.

The substrate scope was tested with optimized conditions (Table 1). We observed excellent reactivity and selectivity retention for propargyl alcohols with different substitution patterns (1–7). Notably, the reaction proceeds with the opposite sense of regioselectivity to the previously reported system by Trost,¹⁴ consistent with the hypothesis that the hydroxyl group is serving as a directing group in this case. Heterocycles are well tolerated in this reaction with only slightly diminished E/Z selectivity (8–10).

Table 1. Substrate scope.

^aPercentages represent isolated yields; in cases where two or more isomers were formed, percentages represent combined yields of isolated samples of each of the different isomers. Stereoisomeric ratios are shown in parentheses (syn:anti) and reflect the mass ratio of isolated samples unless otherwise specified; these ratios are consistent with those observed via ¹H NMR analysis of the crude reaction mixtures. See Supporting Information for details.

^b0.5 mol% Pd(OAc)₂, 0.55 mol% L2, without NH₄OAc, 1.2 equiv donor alkyne, 20–48 h.

^cMeCN:t-AmylOH = 1:1.

^dPd(OAc)₂ catalyst, without NH₄OAc.

^e10 mol% Pd(dba)₂, 11 mol% L1.

^fStereoisomers were inseparable; ratio determined by ¹H NMR.

Furthermore, we examined the reactivities of other native directing groups—including amides, sulfonamides, and amines—none of which had been previously explored in alkyne cross-coupling. Propargyl benzamide (11), Boc-protected propargyl amine (12), and propargyl phenyl amine (14) were all converted to the corresponding 1,3-enyne products, with the same selectivity as propargyl alcohols. Homopropargyl benzamide (15) and tosyl-protected homopropargyl amine (16) were also compatible, albeit with lower regioselectivity than propargyl amine derivatives. We also conducted a systematic study on the reactivities of (homo)propargyl carboxamides (18–42), which are highly challenging acceptors because their activated α-position can cause side reactions and secondary E/Z isomerization. We demonstrated that amides derived from various anilines and aliphatic amines worked well in the reaction, providing the products in good to excellent yield and with high selectivity. However, amides with small substituent groups (37 and 38) were more prone to isomerization, resulting in eroded stereoselectivity (vide infra). We further observed good yield and excellent selectivity with the different alkyl-substituted alkynes (39–42). Introducing alkyl branching at the α-position (41) or introducing an additional methyl spacer between the alkyne and the directing group (42) showed comparable reactivity to the general substrates discussed above.

Different donor alkynes were also examined (43–50). We found that alkynes substituted with different steric hindrance (46–48) and functional groups (44 and 49) worked well as donor alkynes. Consequently, 1,3-enynes with a variety of substitution patterns can be accessed directly, without the need for TIPS deprotection and further functionalization.

To evaluate the viability of this method in more structurally intricate settings, we introduced bioactive molecules and natural products onto each of the alkyne coupling partners and tested whether reaction performance was impacted. Biotin (13), tryptophan (33), and citronellic ester (50) were all found to be compatible with this method. This suggests the potential of applying this method for late-stage modifications of complex targets.

We performed several representative examples on larger scale. First, we found that by directly applying the standard conditions, product 25 could be prepared on 1.5 mmol scale in slightly improved yield compared to the small-scale experiment (71% versus 63%). Next, we evaluated more demanding conditions, with an eye towards further scale-up. We employed Pd(OAc)₂ instead of Pd(dba)₂, given that the former is more widely used in process chemistry. In addition, we decreased the catalyst, ligand, and donor alkyne loadings to 0.5 mol%, 0.55 mol%, and 1.2 equiv, respectively. We also used structurally simplified ligand L2 in lieu of L1, noting that with this ligand it was necessary to monitor reaction progress more carefully to avoid secondary E/Z isomerization. Under these modified conditions, products 5 and 6 were prepared on 1 mmol scale. While the yield and selectivity for acceptor 5 was comparable to the small-scale reaction (32% versus 82%; >20:1 in both cases), sterically hindered acceptor 6 was not fully consumed, even after elongated reaction time, resulting in diminished yield compared to the small-scale trial (32% versus 82%). This decrease in this case could be due to the slower rate of donor–acceptor cross-coupling, which allows the donor alkyne to be competitively consumed via homodimerization. Last, using these modified conditions, we prepared product 1 on 10 mmol scale (gram-scale) in 94% yield.

3. Stereodivergent alkyne cross-coupling.

As mentioned above, during the course of this study we found that the propargyl carboxamide products are especially prone to isomerization, presumably due to the presence of acidic α-protons.³⁸ By taking advantage of the fact that the anti-isomer is thermodynamically favored (see SI), we identified conditions that allowed in situ isomerization to deliver the anti-isomer as the major product with synthetically useful levels of anti/syn selectivity (Figure 1).³⁹ In this way, the transformation could be rendered stereodivergent by simply changing the palladium source and the additive.

Figure 1. Representative examples of anti-selective cross-coupling.

^a Percentages represent combined yields of isolated samples of each of the different isomers. Stereoisomeric ratios are shown in parentheses (anti:syn) and reflect the mass ratio of isolated samples; these ratios are consistent with those observed via ¹H NMR analysis of the crude reaction mixtures. ^b 10 mol% Pd(OAc)₂, 11 mol% L1.

4. Product transformations.

Various transformations of the 1,3-enyne products were examined to underscore the preparative utility of this method (Figure 2). Epoxidation (55) on the alkene moiety and conversion of the alcohol to a bromide (56) were successfully carried out. Removing the TIPS protecting group on the alkyne moiety enabled other alkyne-functionalizing reactions such as Sonogashira arylation (58),⁴⁰ hydrozirconation iodination (59),⁴¹ and 1,2,3-triazole formation via click chemistry (60).⁴² Additional transformations were carried out on the 1,3-enyne derived from propargyl carboxamides. Upon TIPS deprotection of compound 24, the resulting product underwent in situ isomerization to the conjugated allenyl alkene (51), likely due to the acidity of the α-protons. Compound 51 could then be transformed to highly substituted benzenes 53 and 54 through Diels–Alder cycloaddition.⁴³

Figure 2. Product transformations and formal syntheses of natural products.

1,3-Enynes are important building blocks that are often involved in total syntheses of natural products.⁴ For instance, 62 is a common intermediate in the syntheses of Brevisamide⁴⁴ and Vitamin A.⁴⁵ Ghosh and coworkers synthesized 62 with 37% yield over four steps. López and coworkers followed a procedure reported by Mori and prepared 62 with 4% yield over two steps.⁴⁶ In contrast, by using the alkyne cross-coupling method, we obtained 62 with 61% yield over two steps from commercially available materials. This improvement of efficiency is significant given that 62 is an early-stage intermediate in both syntheses (Figure 2).

5. Mechanistic studies.

We next probed the mechanism of the reaction by first investigating the oxidation state of the catalytically active palladium species. At the outset we considered three potential scenarios: (1) Pd(II)-assisted metal–acetylide formation followed by migratory insertion and protodepalladation; (2) oxidative addition of Pd(0) into the C(alkynyl)–H bond followed by acetylide insertion and C(alkenyl)–H reductive elimination; (3) oxidative addition of Pd(0) into the C(alkynyl)–H bond followed by hydride insertion and C(sp²)–C(sp) reductive elimination. First, we monitored the reaction by in situ ¹H NMR and did not observe a signal consistent with metal–hydride, ruling that out as a possible catalyst resting state (Figure S14–S17). Reactions performed with Pd(OAc)₂ as pre-catalyst and Pd(dba)₂ as pre-catalyst led to formation of the same intermediate, as observed by in situ ³¹P NMR (29.64 ppm), suggesting the oxidation state of the pre-catalyst does not dictate catalyst speciation to an appreciable extent. The chemical shift of the catalyst resting state is consistent with a Pd(II)–phosphine complex.⁴⁷ An isotope labeling experiment was then carried out with deuterated TIPS-acetylene, and less than 30% of deuterium was incorporated at the alkenyl position (Scheme 3A). The mixture of H/D incorporation is consistent with a protodepalladation pathway but not with a metal hydride pathway.

Scheme 3. Mechanistic studies and proposed mechanism.

In the particular case of propargyl carboxamide substrates, a possible pathway is initial alkyne isomerization to allene (63) followed by migratory insertion.⁴⁸ To test whether this pathway is operative, 63 was prepared and subjected to syn- and anti-selective conditions (Scheme 3B). In both cases, 1,3-enynes with opposite regioselectivity were obtained (64 and 65). Thus, the desired products are not generated from allene intermediates.

Consistent with earlier literature reports,³⁷ by combining Pd(MeCN)₂Cl₂ and P,N-ligands, we were able to cleanly prepare L•PdCl₂ complexes ligated with P,N-ligands, which could be characterized by NMR (see SI). While PdCl₂•L1 could not be characterized by crystallography, the X-ray crystal structure of model complex 66 where the phenyl groups on the phosphorous atom are replaced with cyclohexyl groups demonstrates that the ligand adopts a bidentate coordination mode in the solid state (Scheme 3C). In this arrangement one of the Ph groups of the diphenylethlenediamine moiety is brought into close proximity to a chloride ligand. A possible explanation for L1’s unique ability to suppress secondary isomerization is thus that its steric bulk effectively crowds out reassociation of the trisubstituted alkene moiety of the product.^49,50

Conclusion

In conclusion, we developed a method to deliver 1,3-enynes selectively from two alkynes by taking advantage various synthetically useful native directing groups. Gram-scale synthesis with 0.5 mol% catalyst loading was demonstrated. 1,3-Enynes were successfully transformed by different approaches. Furthermore, we proposed a reaction mechanism based on preliminary mechanistic studies (Scheme 3C). Pd(II) is the active catalyst and the reaction is initiated by Brønsted-base-assisted palladium acetylide formation. Then the acetylide is inserted into the acceptor alkyne in the syn-addition manner. Finally, the 1,3-enyne product is formed and the catalyst is released by protodepalladation.