Report

# A general alkyl-alkyl cross-coupling enabled by redox-active esters and alkylzinc reagents

See allHide authors and affiliations

Science  13 May 2016:
Vol. 352, Issue 6287, pp. 801-805
DOI: 10.1126/science.aaf6123

## Carbon links without helpful neighbors

It's an irony of modern organic chemistry that the simplest-looking carbon-carbon bonds are often the hardest to make. Most reactions owe their efficiency to neighboring double bonds or oxygen and nitrogen atoms that linger in the products. Qin et al. now present a broadly applicable protocol for making C-C bonds in the absence of such surrounding help. The nickel-catalyzed process couples a zinc-activated carbon center to an ester that's poised to lose CO2. The ready availability of numerous carboxylic acids (which are easily converted to esters) contributes to the reaction's versatility.

Science, this issue p. 801

## Abstract

Alkyl carboxylic acids are ubiquitous in all facets of chemical science, from natural products to polymers, and represent an ideal starting material with which to forge new connections. This study demonstrates how the same activating principles used for decades to make simple C–N (amide) bonds from carboxylic acids with loss of water can be used to make C–C bonds through coupling with dialkylzinc reagents and loss of carbon dioxide. This disconnection strategy benefits from the use of a simple, inexpensive nickel catalyst and exhibits a remarkably broad scope across a range of substrates (>70 examples).

The heart of chemical synthesis relies on forging new C–C bonds, with the evolution and advancement of the field being easily correlated to new developments on this front. For example, pioneering work on the cross-coupling of halogenated aromatic or vinylic (sp2) systems (Heck, Suzuki, Negishi, and Stille) has transformed the practice of organic synthesis (1). Similarly, a general and practical approach to C(sp3)–C(sp3) variants would have the potential to open up new vistas in retrosynthetic analysis. Indeed, such transformations have been on organic chemists’ wish list for well over a century (2, 3). Historically, alkyl-alkyl transition metal–catalyzed cross-coupling reactions have been difficult to accomplish, but examples can be traced to the early work of Kharasch in the 1950s (4), followed by Noller (5, 6) and Kochi and Tamura (7, 8) in the 1960s to more recent work from the groups of Suzuki (9), Fu (10), Knochel (11), Kambe (12), Oshima (13), and many others (14). Thus far, the vast majority of approaches to this problem have involved the coupling of alkyl halides (or related species) to organometallic reagents (1518). However, the limited availability, perceived instability, and frequent toxicity of alkyl halides has perhaps prevented the area of alkyl cross-coupling from blossoming. If one only considers convenience, stability, and availability as desired attributes in a functional group for such a coupling, the carboxylic acid reigns supreme (Fig. 1A). Alkyl carboxylic acids are ubiquitous in every aspect of chemistry and can be readily found in medicines, materials, and natural products and in the pages of commercial chemical supplier catalogs. They are a stable functional group, nontoxic, and eminently diversifiable owing to the field of combinatorial chemistry, in which they are the “workhorse” building block. Although certain carboxylic acids have already been demonstrated to engage in cross-coupling reactions (19), the use of alkyl carboxylic acids in alkyl-alkyl cross-coupling remains elusive.

Carboxylic acids can be primed for reaction through a process known as activation (such as formation of an active ester, –OA*), dating back to the classic work of Sheehan in the synthesis of penicillin (20). Once activated, a gateway opens to access a myriad of related functional groups such as amides, ketones, esters, or alcohols via addition of a nucleophile or alternative oxidation states by the formal addition of hydrogen. In this Report, we present a broadly useful transform that is able to forge C(sp3)–C(sp3) bonds via this age-old activation process.

We recently reported a Ni-catalyzed decarboxylative cross-coupling of alkyl carboxylic acids with arylzinc reagents to forge C(sp3)–C(sp2) bonds by repurposing activating methods more typically associated with amide-bond formation (21, 22). Certain active esters [such as HOAt (N-hydroxy-7-azabenzotriazole), HOBt (N-hydroxybenzotriazole), NHPI (N-hydroxyphthalimide), and TCNHPI (N-hydroxytetrachlorophthalimide)] can accept an electron to trigger an ensuing cascade of events that liberates CO2 from the parent alkyl group (Alk1); such esters (23) are termed redox-active (21, 22). The application of this chemistry to sp3–sp3 C–C bond formation poses a number of substantial challenges, with potentially unproductive pathways far outnumbering the desired reaction (Fig. 1B) (1518). For example, β-hydride elimination from the alkyl metal intermediates, dimerization of the organometallic reagent, reduction of the electrophile, and proto-demetallation are problems that also historically plague traditional C(sp3)–C(sp3) cross-coupling reactions. With a redox-active ester as an electrophile, oxidative addition of low-valent Ni into the activated C–O bond could result in the formation of an acyl-Ni complex, which could reductively eliminate and ultimately result in undesired ketone by-products. These fundamental challenges notwithstanding, we describe a straightforward solution to this problem.

Dialkylzinc reagents were chosen for the organometallic coupling partner because of their ease of preparation from the parent alkyl halide, high functional group tolerance, and their propensity for facile transmetallation. Over the years, mono and dialkylzinc reagents have been shown to be viable cross-coupling partners in Negishi cross-coupling through use of alkyl halides, and related species with Ni-based catalysts (16, 18). Thus, an exploration of the coupling of redox-active esters 1a and 1b (Fig. 1C) with diethylzinc (2) was undertaken. An exhaustive screen of ligands, including bipy (L1, entries 1 and 2) and di-tBubipy (L2, entries 3 and 4) showed that the electron-deficient TCNHPI ester 1b with L2 afforded the desired product in 84% isolated yield (entry 4). TCNHPI can be easily prepared from tetrachlorophthalic anhydride, an industrial nontoxic flame retardant [~$48/kg from VWR (Radnor, PA)] (24), and it has recently been commercialized by Aldrich (catalog no. ALD00564). Additional screening of ligands based on phenanthroline (L3 to L5, entries 5 to 7) and terpyridine (L6 and L7, entries 8 and 9) did not improve the yield and were in many cases detrimental. When the reaction was performed in the absence of NiCl2·glyme, the desired product was not formed (entry 10). With an optimized set of conditions in hand, we explored the scope of this new reaction and found it to be remarkably broad. First, we explored a range of 16 dialkylzinc reagents (Fig. 2A) with piperidine esters 1a and 1b. With the exception of dibenzylzinc reagents, all primary dialkylzinc reagents explored were viable in the cross-coupling. From dimethylzinc (4) and simple alkyl chains (3, 5, 6, 9, and 10) to derivatives harboring olefins (7 and 8), alkynes (11 and 12), acetals (16), ethers (14 and 15), and even alkyl halides (13) were tolerated. The reaction was easily run on a gram scale (as exemplified by 3) and could even be used to produce isotopically labeled piperidine 4-13C. In addition to 3, this example highlights that a lower loading of dialkylzinc reagent (0.5 equivalent) could be used when a highly valuable alkyl group is involved. Secondary alkylzinc reagents such as cyclopropylzinc could also be accommodated, as exemplified by 17. Next, we explored 15 secondary alkyl carboxylic acids (Fig. 2B). Cyclic (18, 19, and 21), heterocyclic (20, 22, 26 to 28, and 31), bridging (25), indane (24), acyclic (29), and even fluorinated (23 and 30) alkyl carboxylic acids were all viable coupling partners. Substrates 30 through 32 are particularly striking and open the door to access a vast array of fluorinated building blocks and optically pure tartrate-derived materials. Many of these products would be either inconvenient or chemically intractable to access from the corresponding alkyl halide starting materials (27, 28, 30, 31, and 32). Primary alkyl carboxylic acids, representing some of the most inexpensive organic materials available, could also be readily used. Of the eight examples depicted in Fig. 2C, most notable are the use of mono-methyladipic acid (34; 2.5 billion kg of adipic acid are produced per year) (25), a pyridine-containing substrate (36), and a polyfluorinated acid (40). Tertiary alkyl carboxylic acids (Fig. 2D) could also be used to generate quaternary centers. Although limited in scope, bridgehead systems such as adamantane (41 and 42) or bicyclo[2.2.2]octane (43) smoothly participated in the cross-coupling. The preparation of such bridged systems is traditionally performed via a three-step Wittig/hydrogenation sequence from the parent aldehyde (26). Although pivalic acid (44) did not couple in this context, its failure led to a distinct type of coupling (vide infra). Simple primary and secondary α-oxyacids could also be used (Fig. 2E, 45 to 49), representing a practical route to form ethers that would not be possible by using alkyl halides. The classic Williamson ether synthesis is still the staple transform for constructing ethers, but in many cases, the SN2 reaction is either sluggish or unworkable because of steric hindrance. This cross-coupling opens a distinct disconnection strategy for ether synthesis by using easily obtained α-oxyacids (either commercial or derived via alkylation with bromoacetic acid) as progenitors for a virtually limitless array of new ethers. Redox-active esters enable the cross-coupling of alkylzinc reagents with high chemoselectivity (Fig. 2F) even in complex contexts, reminiscent of that exhibited in classic amide-bond formation. For example, a variety of sensitive fatty acids reacted smoothly to give the corresponding alkyl chains (50, 57, and 58) without olefin isomerization or epoxide opening. Naturally occurring carboxylic acids such as biotin, cholic acid, and dehydrocholic acid were also amenable to alkylation (54 to 56). Pharmaceuticals and agrochemicals such as pregabalin, 2,4-D, cetirizine, and atorvastatin smoothly reacted with alkylzinc reagents to afford good yields of the alkyl coupling products (51 to 53 and 59). In perhaps the most impressive feat of chemoselectivity for this methodology, the cross-coupling could be conducted on solid phase in the context of peptide synthesis. On-resin coupling of dialkylzinc reagents to proteinogenic amino acids (both side-chain and α-carboxylic acids) facilitates the late-stage introduction of designer amino acids, which would otherwise require de novo synthesis. Valuable synthetic handles including alkenes (60 and 62) and alkyl ethers (63) could be readily incorporated into resin-bound substrates (Fig. 3), and the installation of an aliphatic side chain (61) provides a facile approach to the modulation of peptide lipophilicity (27). Double activation of a peptide substrate bearing both Asp and Glu residues also enables the simultaneous introduction of multiple nonproteinogenic side chains (64 and 65). This convenient approach to diversely functionalized peptides provides a tool for the construction of therapeutic peptide leads and stapled peptides (28, 29) and for applications in bioconjugation (30). Returning to the limitations posed by tertiary systems, mechanistic studies involving 5-exo-dig cyclization, racemization, and ring-strain opening experiments clearly point to the radical nature of this reaction (figs. S1 to S3). We reasoned that the steric limitations encountered could be overcome by engaging these reactive species with a radical trap that could subsequently combine with an organozinc reagent. Remarkably, such a conjunctive cross-coupling could be realized, as shown in Fig. 4 (31). By using benzylacrylate as a radical trap and phenylzinc as a coupling partner, 13 different tertiary alkyl carboxylic acids could be smoothly engaged in a three-component cross-coupling to generate useful building blocks that would be extremely difficult to access in any other way (66 to 78). This reaction is scalable (as exemplified by 76), generates two new C–C bonds, and is a rare example of a multicomponent cross-coupling reaction that forms quaternary centers in high yield. In addition to the broad functional group tolerance [heteroatom containing substrates, Boc (tert-butoxycarbonyl), TBS (tert-butyldimethylsilyl), and MOM (methoxymethyl) groups], this reaction features the use of electronically different radical precursors, such as α-heteroatom acids or electronically unbiased and neutral alkyl acids such as pivalic acid. As with any methodology, there are obvious drawbacks. For example, the atom-economy is low because of the use of an activating group. We do note, however, that such considerations are ignored in the enormous field of peptide chemistry, in which expensive coupling agents are regularly used to make a simple amide bond (32). The activating agents used here, NHPI ($19.5/mol) and TCNHPI, are both commercially available and derived from cheap, readily available materials and open a gateway of reactivity heretofore inaccessible. The use of two equivalents of the dialkylzinc reagent is another disadvantage in cases in which the zinc-derived fragment is valuable. However, as shown in 3 and 4-13C, the reaction proceeds with workable yields when simply equimolar amounts of alkylzinc reagents and 5 mole percent (mol %) catalyst were used.

The data presented here suggests that the advantages associated with this method far outweigh its limitations. Without exception, the carboxylic acids used represent the most inexpensive sources of these carbon frameworks; all of these were commercially available. In contrast, in the cases in which an alkyl halide would be chemically stable, only a handful are commercially available. Conceptually, carboxylic acids can perhaps be considered nature’s version of a boronic acid. For the past seven decades, these ubiquitous functional groups have usually been dehydrated with incorporation of a nucleophile (such as an amine to make an amide). This method extends the native diversification of this functional group (33) to allow for the addition of a new carbon framework via extrusion of CO2. As such, it is anticipated that this technique will greatly expand planning options in retrosynthetic analysis.

## Supplementary Materials

www.sciencemag.org/content/352/6287/801/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S3

Tables S1 to S6

References (3460)

## References and Notes

1. Data retrieved from https://us.vwr.com/store on 6 April 2016.
2. S. T. Waddell et al., PCT international patent application US 20040133011 (2004).
3. I. D. Linney, I. M. McDonald, PCT international patent application 2001085704 (2001).
4. A. Ali, et al., PCT international patent application 2007070173 (2007).
5. L. S. Bertram, M. C. T. Fyfe, M. J. Procter, G. M. Williams, PCT international patent application 2007116229 (2007).
6. U. K. Medivir, Peptimmune, patent application WO2006, A1 (2006).
7. S. F. Dowdy, B. R. Meade, K. Gogoi, patent application US2015/238516 A1 (2015).
8. R. Clark, et al., patent application WO2011/145747 A1 (2011).
9. Rohm and Haas Company, patent application E B1, 213718 (1991).
10. D. L. Flynn et al., patent application US2014/275016 A1 (2014).
Acknowledgments: Financial support for this work was provided by Bristol-Myers Squibb, the Catalan Government (postdoctoral fellowship to J.C.), the U.S. Department of Defense (National Defense Science and Engineering Graduate fellowship to J.T.E.), NIH (F32GM117816 postdoctoral fellowship to L.R.M.), the China Scholarship Council (postdoctoral fellowship to C.L.), and the National Institute of General Medical Sciences (GM106210). We thank D.-H. Huang and L. Pasternack for assistance with nuclear magnetic resonance spectroscopy; M. Collins for providing samples of atorvastatin, cetirizine, and pregabalin; R. Mills for experimental assistance; A. Rheingold, C. E. Moore, and M. A. Galella for x-ray crystallographic analysis; M. R. Ghadiri for providing access to preparative HPLC equipment; and K. Chen for helpful discussions. Metrical parameters for the structures of 11 and 73 are available free of charge from the Cambridge Crystallographic Data Center under reference nos. CCDC-1457710 and CCDC-1457711, respectively.
View Abstract