Sequential EDA/HAT Enables C(sp³)─C(sp³) Cross-Dehydrogenative Coupling: Accessing Non-Canonical Amino Acids with Quaternary Carbon Centers from Activated and Non-Activated Alkanes

Abstract

Current strategies for the alkylation of amino acids via C(sp³)─H functionalization require the use of transition metals, often in the presence of excess peroxides at high temperatures. Herein, we developed a unique electron-donor─acceptor (EDA) system using aryl halides and N-phenylglycine derivatives to achieve a C(sp³)─C(sp³) cross-dehydrogenative coupling that gives access to noncanonical amino acids. This metal- and photocatalyst-free protocol enables functionalization of ethers and unactivated alkanes via a hydrogen-atom transfer (HAT) process performed by an in situ generated aryl radical. The reaction proceeds via radical─radical coupling, which enables the formation of sterically demanding quaternary carbon centers and delivers site-selective α-functionalized N-phenylglycine moieties even when appended to complex bioactive molecules or small peptides.

Graphical Abstract

Noncanonical amino acids (ncAAs) have become a quintessential motif in drug development, showcasing improved pharmaceutical properties when compared to their naturally occurring counterparts.¹ They have been utilized in small -molecule and peptidomimetic drug discovery, as well as affinity labeling-based techniques.² Recent examples from pharmaceutical companies include the development of small peptides containing ncAAs that enabled the inhibition of rat sarcoma virus (RAS). This was attributed to greater membrane permeability enabled by ncAA-derived peptides.³ In 2023, Merck showcased another example of an oral macromolecule embedded with ncAAs that exhibits effective reduction of LDL cholesterol in patients with hypercholesterolemia.⁴

Glycine is an ideal choice for a direct C(sp³)─H functionalization to generate ncAAs, as it is the sole achiral proteinogenic amino acid with two α-C(sp³)─H bonds.⁵ Current strategies to alkylate glycine derivatives require the use of prefunctionalized precursors such as N-hydroxyphthalimide esters⁶ or Katritzky salts (Scheme 1A).⁷ In spite of the elegance of these methods, a direct functionalization of C(sp³)─H bonds via cross-dehydrogenative coupling (CDC) between glycine derivatives and alkyl scaffolds would bring simplicity when compared to atom-intensive prefunctionalization strategies.⁸

Scheme 1. Traditional Ways to Alkylate Glycine Derivatives and Proposed Dual EDA/HAT Cross-Dehydrogenative Coupling.

Traditional CDC approaches to form ncAAs require either (i) the use of excess peroxides at high temperature,⁹ (ii) a mixture of peroxides or strong oxidants with transition-metal catalysts at high temperature,¹⁰ or (iii) a photocatalyst in the presence of peroxides or metal catalysts (Scheme 1B).¹¹ Beyond the safety concerns of using excess peroxides under harsh reaction conditions, most of these methods suffer from poor functional group compatibility due to the nature of the reagents employed. Additionally, these methods are often limited to the activation of ethereal or benzylic C─H bonds, with only one example enabling the functionalization of unactivated C(sp³)─H bonds.^9a Thus, a mild and versatile strategy that alkylates glycine derivatives via the functionalization of both activated and unactivated C(sp³)─H bonds remains to be developed.

Inspired by previous work in electron donor─acceptor (EDA) complexes of aryl halides (Scheme 1C),¹² we hypothesized that fine-tuning the electronic parameters between anionic electron-donor species and aryl halide electron-acceptors would generate EDA complexes capable of producing persistent C(sp³)-radicals alongside highly reactive aryl radicals. We surmised that electron-deficient aryl radicals would perform hydrogen-atom transfer (HAT) in the presence of both activated and unactivated C(sp³)─H bonds, thereby generating a second set of alkyl radicals. The obtained alkyl radical could then couple with the persistent radicals and form the desired C(sp³)─C(sp³) cross-dehydrogenative coupling product (Scheme 1D).¹³ Given that radical─radical recombination mainly occurs via barrierless processes, we expected that this method would enable the formation of sterically demanding quaternary carbon centers.

Herein, we present a direct photoinduced synthesis of ncAAs via an EDA complex that enables C(sp³)─C(sp³) CDC between N-arylglycine derivatives and alkyl groups (Scheme 1E). The reaction proceeds via the coupling between a persistent C(sp³)-α-amino radical¹⁴ and an alkyl radical generated by a HAT process. The method enables coupling of activated and unactivated C(sp³)─H bonds to N-arylglycine derivatives to access sterically demanding quaternary carbon centers.

We started our optimization procedure with N-phenyl-2-(phenylamino) acetamide (1) as the glycine derivative and THF (2) as the alkane substrate to generate N-phenyl-2-(phenylamino)-2-(tetrahydrofuran-2-yl)acetamide (3) as the desired ncAA product. Using 4-iodoacetophenone (Ar-1) (Table 1, entry 1), the reaction generated product 3 in 87% yield. While electron deficient 4-iodobenzonitrile (Ar-2) (entry 2) led to a small reduction in yield (82%), 1-iodo-4-(trifluoromethyl)benzene (Ar-3) (entry 3) afforded product 3 in an excellent 95% yield. Switching to electron donating 4-iodoanisole (Ar-4) (entry 4) led to an erosion in the yield to 46%. These results demonstrate the importance of using an electron-deficient aryl iodide for both EDA complex formation and subsequent HAT process. Reducing the amount of Cs₂CO₃ (1.5 equiv) (entry 5) did not affect product formation (95%). When the reaction was performed in the dark (entry 6), no product was observed, and illuminating the reaction using 427 nm (entry 7) lamps also had a deleterious effect on yields (59%). Interestingly, in the absence of ethereal solvents, low yields (<20%) of coupling between 1 and the aryl radical (Ar-3) was observed. For full optimization, see SI pages S5-S9.

Table 1.

Optimization of the Reaction and Its Conditions^a

^aOptimal conditions: 1 (0.1 mmol, 1 equiv), 2 (0.5 mL), base (0.15 mmol, 1.5 equiv), MeCN (0.5 mL), 390 nm LED (40 W), room temperature was 35─40 °C (caused by the LED lamp), reaction flask capped under argon, 24 h.

^bDark.

^c427 nm LED lamp.

^d1H NMR yields using dibromomethane as the internal standard.

To further understand the mechanism of this reaction, we initiated radical trapping experiments (Scheme 2) using agents such as 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO), 1,4-dinitrobenzene (1,4-DNB), butylated hydroxytoluene (BHT), and 1,1-diphenylethylene (1,1-DPE) as additives. All of these additives showed complete quenching of the transformation. Additionally, GC-MS analysis of reaction crudes showed trapping of the benzotrifluoride radical with both TEMPO and 1,1-DPE, supporting the formation of the aryl radical. The THF radical was also trapped using 1,1-DPE, indicating that a HAT process had taken place.¹⁵ Lastly, when using ethyl phenylglycinate as an amino acid derivative in the absence of radical traps (Scheme 2, entry 5), we observed the formation of the N-phenylglycine dimer, which supports the formation of the α-amino radical, thus hinting at the possibility of a radical─radical coupling step for the formation of the desired product.

Scheme 2. Mechanistic Experimentsa^a.

^a(A) Radical trapping. Reaction yields were obtained via ¹H-NMR using CH₂Br₂ as the internal standard. Entries 1─4 were performed using 1 as the amino acid derivative. Entry 5 used ethyl phenylglycinate (25a) as the amino acid derivative in the absence of trapping agents to generate product 25; the amino acid dimer was observed as a byproduct of the reaction. (B) UV-Vis experiments for EDA complex validation.

UV─vis spectroscopy revealed the possibility of an EDA complex between 1 and the electron-deficient aryl iodide (Ar-3) in MeCN (Scheme 2B).¹⁶ Indeed, there is increased absorption when both components (1 and Ar-3) are mixed together when compared to solutions containing only the glycine derivative or aryl iodide. Importantly, the absorption spectra shifted further into the visible range when Cs₂CO₃ (base) was added, presumably through the partial deprotonation of the phenyl N─H group, which could also involve a ternary EDA complex involving a base as previously proposed by Xu.^6a As shown in the SI (page S67), ¹H NMR experiments show that addition of base to 1 stabilizes the dynamic scrambling of the aniline proton,¹⁷ which presumably enhances the electron donation of the lone pair on nitrogen to the electron-deficient arene.

Based on these mechanistic experiments and previous literature reports,¹⁸ we propose that N-arylglycine derivatives A and electron-deficient aryl iodides B form a photoactive EDA complex, which upon visible light irradiation undergoes a single electron transfer (SET), thereby forming the corresponding amino radical cation C and aryl radical anion D. Deprotonation of C generates the persistent α-amino radical E. Simultaneously, loss of iodide from intermediate D generates aryl radical F, which in turn performs a HAT process with alkane reagents G. Using THF-d₈ provided the deuterated arene byproduct (SI page S64-65). The resulting alkyl radical H undergoes radical─radical coupling with species E to generate the desired C(sp³)─C(sp³) cross-coupled product (Scheme 3).

Scheme 3. Proposed Mechanism.

Finally, we turned our attention to the substrate scope of the reaction (Scheme 4). A wide array of ether substrates were coupled to glycine derivative 1 to form the desired ncAAs. Cyclic ethers including THF, 1,4-dioxane, tetrahydro-2H-pyran, and 1,3-dioxolane were successfully coupled to form products 3, 4, 5, and 6 in exceptional yields of 95% (d.r. = 1:1.6), 84% (d.r. = 1:1.5), 89% (d.r. = 1:1.7), and 60%, respectively. Linear ethers like ethoxyethane, 1-butoxybutane, 1,2-dimethoxyethane, and 2-methoxy-2-methyIpropane also successfully coupled to obtain products 7, 8, 9, and 10, respectively, in good to excellent yields ranging from 81% to 97%. As shown, the protocol was able to activate primary and secondary C(sp³)─H bonds with these ether substrates and, as expected, two regioisomers were obtained for compound 9 (90% 4:1 ratio), favoring the formation of the most stable secondary radical.

Scheme 4. Reaction Scope^b.

^aAdamantane was used in 10 equiv, in CH₃CN (1 mL). ^bAll yields are isolated.

Quaternary ethers are highly valuable in the field of medicinal chemistry due to the reduced in vivo metabolic degradation conferred by the increased steric hindrance near the ethereal bond.¹⁹ Although such structures are in high demand for pharmaceutical applications, strategies to generate such versatile motifs remain limited.²⁰ Hence, we focused our attention toward the C(sp³)─H functionalization of tertiary ethers to generate quaternary carbon centers. Diisopropyl ether and 2,5-dimethyl-THF were successfully cross-coupled with a variety of glycine derivatives. Our model amino acid 1 generated compounds 11 and 12 in good to excellent yields (72% and 90%, respectively). While electron-withdrawing groups were tolerated in the aniline moiety, affording product 13 in 70% yield, electron-donating groups surprisingly provided product 14 in only 36% yield. Increasing the steric hindrance and electron donation using 2,4,6-trimethyl aniline derivatives was also deleterious and gave product 15 in only 30% yield. meta-Substitutions with ─CF₃ and ─Br afforded products 16 and 17 in moderate but synthetically useful yields (48% and 56%, respectively). Amide substitutions were also investigated with compounds 18─22. Both electron-donating and electron-withdrawing groups were tolerated (57─68% yields, 18─21), but increased steric hindrance afforded 22 in only 33% yield. Secondary amides and esters, including phenol esters, were also compatible with our reaction conditions and afforded products 23─25 in 35%, 53%, and 66% yields, respectively.

The versatility of this reaction was further extended to the functionalization of unactivated alkanes. Cyclopentane, cyclohexane, and cyclooctane were successfully coupled to glycine derivative 1 to form the corresponding ncAAs 26, 27, and 28 in excellent yields (91%, 78%, and 91%, respectively). This was particularly gratifying given that the difference in BDE between ethereal (~92.1 kcal/mol) and unactivated C─H bonds (~100 kcal/mol) is significant enough that most current methods choose harsh reaction conditions to achieve these transformations^9-11 or instead use preactivated intermediates.^6,7 Using adamantane as a coupling partner, we were also able to access all-carbon quaternary centers, albeit in lower yields (29, 34% yield). It is worth noting that adamantane, a solid, was only used in 10 equiv, which opens the door to the functionalization of other nonliquid complex alkanes. Alkane functionalization was also achieved in good yields with different amide moieties (30─33, 61%─72% yields).

The reactivity of dipeptides and bioactive molecules embedded with an N-phenylglycine moiety was also explored. Selective functionalization of the phenylglycine motif in Gly─Gly (34) and Gly─Ala (35) was achieved in 34% and 35% yields, respectively. The remainder of the mass balance was starting dipeptide material. Borneol, pregnenolone, and menthol derivatives were also coupled, generating products 36, 37, and 38 in 51%, 46%, and 44% yields, respectively. It is worth highlighting that in these substrates, the desired C(sp³)─C(sp³) bonds were generated regioselectively despite the presence of other reactive C─H bonds present in the substrates. For example, a pregnenolone derivative led to the formation of the desired quaternary carbon-center despite the proximity of allylic C─H bonds.

Unfortunately, the α-position of thioethers and benzylic C(sp³)─H substrates failed to provide the desired products in good yields using the presented reaction conditions (see SI page S10 for failed substrates). We are currently exploring alternatives to enable these and other failed couplings.

Lastly, to demonstrate the applicability of these derivatives, we performed a 1.0 mmol functionalization, affording compound 39 in 34% yield (Scheme 5). It is worth highlighting that when compared to aniline product 14 (35%), this larger-scale attempt had minimal erosion in yield (34%). The importance of the methoxyphenyl group comes from its ability to uncover the free-amine using ceric ammonium nitrate (CAN).^6a As a consequence, compound 40 (45% yield), for example, possesses three vectors for further functionalization or coupling: (i) removal of the methoxyphenyl group to obtain the ─NH₂ group, (ii) saponification of the ester to afford the carboxylic acid moiety,²¹ and (iii) acetal deprotection^10a to uncover an aldehyde (Scheme 5).

Scheme 5. Exemplification of Possible Vector Handles for Further Couplings and Transformations.

In summary, we report a metal- and photocatalyst-free cross-dehydrogenative coupling between N-aryl glycine derivatives and alkanes. This photoinduced strategy proceeds via an electron donor─acceptor complex and a subsequent HAT process, thereby enabling a barrierless radical─radical coupling that generates quaternary carbon centers. In addition, this system showed moderate compatibility with dipeptides, which can be directly applied to ncAA-derived peptide drug development studies. Further functionalizations and applications of this strategy are currently being explored in our lab.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.orglett.5c01349.

¹H and ¹³C{1H} NMR spectra for all products, GC─MS spectra, and substrate scope and its limitations (PDF)

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.