Enantioselective Hydrodifluoroalkylation of Alkenes with Conformationally Tuned Peptidyl Hydrogen Atom Transfer Catalysts

Abstract

We report the enantioselective hydrodifluoroalkylation of alkenes proceeding via an asymmetric hydrogen atom transfer (HAT) event catalyzed by thiol-containing tetrapeptides. Photocatalytic generation of a difluoroacetyl radical followed by carbon–carbon bond formation results in a prochiral carbon-centered radical that engages with the chiral catalyst. A trialkylamine reductant is proposed to turn over the catalyst in this net-reductive transformation. Notably, incorporating an (S)-β-methyl-substituted cysteine as the N-terminal residue improved selectivity relative to the native N-terminal cysteine (Cys) residue, and X-ray crystallographic analysis supports the conformational underpinning of this effect. A range of enantioenriched γ–substituted amides were synthesized in up to 96:4 enantiomeric ratio, demonstrating the broad functional group tolerance of this method. Models accounting for asymmetric induction are proposed with supporting DFT calculations.

Graphical Abstract:

INTRODUCTION

Enantioselective catalysis through radical intermediates is a nascent yet prospering field in which asymmetric hydrogen atom transfer (HAT) is a particularly exciting development.¹ Despite inherent challenges with rendering HAT enantioselective, early successes in the field have been observed with chiral Lewis acid and transition metal catalysis,² mostly with achiral H-atom donors. Foundational studies on asymmetric HAT via chiral H atom sources were reported by Roberts, which utilize a glycosyl thiol scaffold (S1, Figure 1A).³ Enantioselective HAT catalysis has more recently been coupled with photochemically driven processes, recording initial successes in photocatalytic radical-based coupling reactions mediated by enzymes.⁴

Figure 1.

(A) Thiol-based organocatalysts for asymmetric HAT. (B) This work: enantioselective olefin hydroalkylation via asymmetric HAT from a peptide thiol catalyst.

The potential for similar achievements with small molecule catalysts was highlighted by the discovery of a photochemically driven deracemization protocol, where a cysteinyl peptide-based thiol catalyst (P1) proved effective for the terminal, enantiodetermining HAT step.⁵ This tetrameric peptide has a β-turn-biased structure that is highly effective in asymmetric catalysis by facilitating key non-covalent interactions (NCIs) between the substrate and catalyst.⁶ Tetrapeptide (P1) was also demonstrated by Ye to have high enantioselectivity in various photocatalytic reductive couplings of specified exo-cyclic alkenes and in the deracemization of δ- and γ-lactams.⁷ Another tetrapeptide (P2) has been employed by our groups in a enantioselective bimolecular hydroamination,⁸ and more recently, a chiral C2-symmetric thiophenol-based HAT catalyst (Ar1) was developed by Dong for a intramolecular hydroamination protocol.⁹

Inspired by these precedents, we embarked on a study to explore the generality of cysteine-based catalysis to achieve asymmetric HAT in the context of a C–C bond-forming reaction (Figure 1B). While significant advances in the coupling of C(sp²) fragments have been made over the last few decades,¹⁰ the coupling of C(sp³) fragments remains challenging.¹¹ The development of such methods coupled with stereochemical control can provide access to complex molecular scaffolds of importance to synthetic chemistry and materials science.¹² Notably, the addition of difluoroacetyl groups to bioactive molecules is a process of significant interest in medicinal chemistry. There is also a fundamental interest in remote asymmetric induction (γ-functionalization).¹³ To this end, we targeted a reductive hydrodifluoroalkylation reaction where difluoroacetyl radicals are added to 1,1-disubstituted alkene in an anti-Markovnikov fashion (Figure 1B). This photoredox-catalyzed reaction is facilitated by a terminal reductant that regenerates both the HAT and photocatalytic cycles.

Reported herein are synthetic, mechanistic, and computational studies that unveil new catalyst requirements and mechanistic determinants for the enantioselective hydroalkylation of alkenes. A striking interplay of catalyst conformational dynamics, photoredox tuning, and subtle catalyst-substrate non-covalent interactions combine to deliver an efficient and highly enantioselective transformation. These findings highlight significant advances in our understanding of the factors that govern photochemical asymmetric organocatalysis and establish catalysis design principles with broad applicability.

RESULTS AND DISCUSSION

We commenced this study by exploring previously reported chiral thiol catalysts in the context of photocatalytic reductive olefin hydrodifluoroalkylation. Our initial efforts focused on the hydroalkylation of isopropenyl acetate (2a) with N-benzyl-2-bromo-2,2-difluoroacetamide (1a) mediated by a binary catalyst system comprised of [Ir(ppy)₂(dtbbpy)]PF₆ as the photocatalyst and a tetrameric peptide thiol as the organocatalyst (Table 1). Using triethylamine as a terminal reductant to facilitate catalyst turnover, product 3a was formed in 53% yield and 43:57 er with peptide Boc-Cys-d-Pro-Acpc-Phg-NMe₂ (P1) (entry 1). When Hantzsch ester (HEH) was used as the reductant, 3a was obtained in 61% yield and 41:59 er (entry 3). Decreasing the reaction temperature to −20 °C led to an increase in yield and enantioselectivity regardless of the reductant used (entries 1–4). Having identified preliminary reaction conditions using HEH as the reductant, P2 and S1 were also examined as HAT catalysts, but reduced enantioselectivities were observed (entries 5–6). To achieve higher levels of enantioinduction, systematic modification of each position of P1 were evaluated.

Table 1.

Reaction Optimization for the hydroalkylation of 2a^a

entry	thiol	reductant	temperature	yield (%)b	erc
1	P1	Et3N	25°C	53%	43:57
2	P1	Et3N	−20°C	75%	41:59
3	P1	HEH	25°C	61%	41:59
4	P1	HEH	−20°C	72%	34:66
5	S1	HEH	−20°C	63%	52:48
6	P2	HEH	−20°C	70%	36:64
7	P3	HEH	−20°C	77%	26:74
8	P4	HEH	−20°C	73%	29:71
9	P5	HEH	−20°C	72%	25:75
10	P6	HEH	−20°C	76%	21:79
11	P7	HEH	−20°C	77%	52:48

^aReactions were conducted with 0.05 mmol of 1a and 1.5 equiv. of 2a.

^bIsolated Yields.

^cDetermined via HPLC analysis using a chiral stationary phase. The er values are reported in order of elution and may not represent absolute stereochemistry.

The thiol peptides studied herein possess the catalytic residue at the i position, a proline residue at i + 1, an α,α-disubstituted amino acid at the i + 2 position, and a chiral i + 3 residue to achieve the β-turn-biased structure. To optimize our catalyst, we first evaluated the effect of varied i + 2 residues. When Acpc was replaced by Aib, an enhancement in enantioselectivity was observed (entries 4 and 7), indicating a preference for a smaller τ angle which results in a closer distance between the i + 3 and the i residues. We also sought to assess the influence of the i + 3 residue. With Acpc in the i + 2 position, changing Phg for Phe increased the selectivity (entries 4 and 8), and additive effects were observed with Aib and Phe at the i + 2 position and i + 3 positions, respectively (entry 9). The effect of the C-terminus was also studied, and the incorporation of a bulkier cyclic amide improved the selectivity (entry 10). The stereochemistry of the Pro residue was crucial for high selectivity. A peptide containing l-Pro furnished a nearly racemic mixture (52:48 er, entry 11.) We evaluated analogous peptides with further variations of the i + 2 and i + 3 residues; however, these structural modifications resulted in either diminished or insignificant changes in the selectivity compared to P6.

Substrate Studies.

Having identified P6 as a selective HAT catalyst, we decided to evaluate the efficiency of this species with other substrates. When phenylamide 1b was used as the radical precursor for the hydroalkylation of 2a, the desired product 3b was formed in 79% yield and 66:34 er. While the yield of 3b is comparable to that of 3a under the same reaction conditions, the enantioselectivity is significantly lower (66:34 vs. 21:79). To investigate this observed difference in selectivity, a Hammett study was conducted using five para-substituted phenylamide substrates (Figure 2). We observed a negative linear correlation (ρ = −0.22) between σ_ρ and log(er), suggesting that the amide moiety likely engages in an electrostatic interaction during HAT. An X-ray crystal structure of P6 reveals a type II′ β-hairpin secondary structure that places the Aib amide N–H bond on the same face of the catalyst as the thiol group (Figure 3a).¹⁴ With this in mind, we hypothesize that the C-centered radical species is anchored to the peptide catalyst via a hydrogen-bond interaction between the substrate carbonyl group of the amide and the Aib N–H bond in the peptide backbone. Based on this hypothetical binding mode, we revisited the thiol catalyst P6 and further explored catalyst modifications to enhance enantioselectivity.

Figure 2.

Hammett study for para-substituted phenylamides. Reactions were conducted with 0.05 mmol of 1b-f and 1.5 equiv of 2a and [Ir(ppy)₂(dtbbpy)]PF₆ as the photocatalyst.

Figure 3.

β-substituent effect in the hydrodifluoroalkylation of 2a with 1a. Reactions were conducted with 0.05 mmol of 1a and 1.5 equiv of 2a. The X-ray crystal structures of P6, P9, and P10 were rendered using CYLview20.

The β -Me effect.

Upon further analysis of the X-ray crystal structure for P6 (Figure 3a), we hypothesized that N-terminal native cysteine residues could potentially be involved in non-selective HAT pathways due to populated catalyst conformations resulting from rotations of the cysteine C–C bond that might position the thiol group away from the chiral environment of the catalyst backbone (Figure 3b). We proposed the addition of a methyl group at the β position of cysteine to introduce a conformational bias.¹⁵ Evaluating the expected lowest energy Newman projections of the two possible epimers, we hypothesized that the (S)-β-methyl substitution would lead to a greater population of the catalyst with the ideal positioning of the thiol. In contrast, the (R)-β-methyl-Cys could result in reduced selectivity by positioning the thiol away from the chiral environment and thereby increasing the likelihood of non-selective HAT. To study this, we synthesized both diastereomers of a β-methyl-Cys-containing tetrapeptide via the corresponding cystines, which are derived from Boc-threonine and Boc-allo-threonine (See Scheme S2). P8 and P9 were tested in the hydro-difluoroalkylation of 2a with 1a, and we were pleased to observe a small improvement in enantioselectivity with the (S)-β-methyl-Cys-bearing catalyst (P8) over P6 from 21:79 er to 18:82 er. The diastereomeric catalyst (P9) proved to be nearly non-selective with 3a formed in 46:54 er. Notably, the differences between the performance of P8 and P6 were even greater when evaluated under optimized reaction conditions (vide infra). X-ray crystallographic analysis of P8 and P9 (Figure 3c) supports our structural hypotheses, highlighting the retention of the type II′ β-hairpin conformation throughout this catalyst series while also displaying the bond orientations as expected by the Newman projection analysis.

Furthermore, in our studies with the N-terminal native cysteine-containing peptide catalysts, we observed catalyst degradation by the end of the reaction, as the peptide was eventually consumed through addition of the thiyl radical to the alkene to form an inactive thioether-containing peptide (P6-Alk). An additional benefit of the β-methyl group is its ability to inhibit catalyst deactivation, presumably by slowing the reaction between the thiyl radical and the alkene starting material (Figure S2).

While a small increase in enantioselectivity was observed with the (S)-β-methyl containing catalyst, we suspected that non-selective HAT between the prochiral radical and the terminal reductant may be limiting the selectivity of this transformation. Strikingly, in the absence of a peptide thiol, racemic 3a was obtained in yields comparable to the thiol-catalyzed reaction. (Table 2, entries 1–2.). This prompted us to investigate alternative terminal reductants using the improved (S)-β-methyl-Cys-bearing catalysts.

Table 2.

Terminal Reductant Studies for the hydroalkylation^a

entry	thiol	reductant	yield (%)	er
1	P6	HEH	88	20:80
2	-	HEH	85	50:50
3	P8	HEH	62	18:82
4	P6	HEH2	52	18:82
5	P8	HEH2	20	15:85
6	P6	HEH3	21	18:82
7	P8	HEH3	20	13:87
8	P8	triethylamine	64	28:72
9	P8	triisobutylamine	59	14:86
10	P8	triisopentylamine	57	17:83
11	P8	tribenzylamine	74	18:82
12	P8	1-ethylpiperidine	63	34:66
13	P8	N-cyclohexyl-N-methylcyclohexanamine	25	40:60
14	P8	tris(3,6-dioxaheptyl)amine	93	12:88

^aReactions were conducted with 0.05 mmol of 1a and 1.5 equiv of 2a. The X-ray crystal structure of HEH2, previously reported by Isomura et al. (J. Am. Chem. Soc. 2019, 141, 4738–4748; CCDC 1902779)¹⁸ and HEH3, determined in this study, were rendered using CYLview20.

Reductant Studies.

Upon exploring different substituents at the 2- and 6-positions of HEH, we observed improved enantioselectivity as the steric demand of the substituent was increased (Table 2, entries 1–7). In all cases, the (S)-β-methyl-Cys-containing catalyst (P8) outperformed P6. However, regardless of the peptide used, the yields were greatly diminished as the steric bulk of the substituent was increased. This is likely due to the decreased overall reactivity of the HEH: while racemic background reactivity was lowered, it also decreased the efficiency of HAT between the reductant and thiyl radical. Interestingly, increasing the steric bulk at the 2-position is known to reduce the planarity of the HEH motif.¹⁶ This structural change renders the HEH hydrogen more hydridic.¹⁶ This would result in a polarity mismatch between the nucleophilic prochiral radical and the HEH H-atom, thereby reducing the rate of background reactivity.¹⁷ This structural analysis aligns with the observed increase in enantiocontrol when using the bulkier substituents, as the reduced polarity match for the radicals would result in a lower reaction rate.¹⁷

Although the increase in enantioselectivity using HEH3 and P8 (entry 7) was promising, the low yield of hydroalkylation prompted us to reconsider tertiary amine reductants (entries 8–14). When triethylamine was employed as the terminal reductant, 1c was obtained in only 28:72 er. However, as the steric bulk of the amine was increased, an increase in enantioselectivity was observed with no noteworthy decrease in yield (entries 8–11). A broad scope of unique tertiary amines were tested (entries 12–13; see Table S12), and tris(3,6-dioxaheptyl)amine was identified as an efficient terminal reductant, resulting in both improved yield and enantioselectivity in this transformation (entry 14).

Photocatalyst Studies.

Upon identifying an appropriate reductant for this reaction, a range of photocatalysts with varied redox properties were evaluated. Interestingly, a negative correlation between the enantioselectivity in the hydrodifluoroalkylation of 2a with 1a and the excited state reduction potentials (*E_1/2) of the various iridium photocatalysts was observed (Table 3). This suggests that an increase in the reducing power of the photocatalyst corresponds to an increase in the enantiomeric ratio of 1c. While the reaction yields exhibited slight variations with different photocatalysts, no clear correlation between yield and photocatalyst redox properties was observed. We identified Ir(dFppy)₃ as the lead photocatalyst for this transformation, as it offers comparable selectivity to Ir(ppy)₃ while achieving higher yields of 3a.

Table 3.

Photocatalyst Studies^a

Photocatalyst		yield (%)	er
Ir(ppy)3		52%	91.5:8.5
Ir(dFppy)3		83%	90:10
Ir(p-CF3-ppy)3		83%	84.5:15.5
[Ir(dtbbpy)(ppy)2)PF6		82%	86:14
Ir(p-F-ppy)3		77%	77.5:22.5
(Ir[dF(CF3)ppy]2(dtbbpy))PF6		49%	67.5: 32.5

^aEffect of photocatalyst on the enantioselectivity of the hydrodifluoroalkylation of 2a with 1a. Reactions were conducted with 0.05 mmol of 1a and 1.5 equiv of 2a.¹⁹

Catalyst Re-optimization.

Upon determining the optimal conditions for this transformation, we initiated a comprehensive re-evaluation of structural modifications on P8. Transitioning from a cyclic amide cap to an NMe₂ C-terminal cap and substituting Phe with cyclohexylglycine (Chg) at the i+3 position resulted in a notable enhancement in selectivity from 12:88 to 5:95 er (Table 4, entry 1). Upon systematically examining a series of peptides bearing unique i+2 residues (entries 1–7), we observed that peptides bearing residues with smaller τ angles yielded higher selectivity, with Aib, Aic, and Cle each achieving an enantiomeric ratio of 5:95.

Table 4.

Further Catalyst optimization^a

entry	thiol	yield (%)	er
1	P10 - Boc-((S)-β-Me-Cys)-DPro-Aib-Chg-NMe2	82	5:95
2	P11 - Boc-((S)-β-Me-Cys)-DPro-Aic-Chg-NMe2	87	5:95
3	P12 - Boc-((S)-β-Me-Cys)-DPro-Ac6c-Chg-NMe2	90	8:92
4	P13 - Boc-((S)-β-Me-Cys)-DPro-Cle-Chg-NMe2	85	5:95
5	P14 - Boc-((S)-β-Me-Cys)-DPro-Acbc-Chg-NMe2	87	7:93
6	P15 - Boc-((S)-β-Me-Cys)-DPro-Gly-Chg-NMe2	78	17:83
7	P16 - Boc-((S)-β-Me-Cys)-DPro-Acpc-Chg-NMe2	83	14:86
8	P17 - Boc-Cys-DPro-Aic-Chg-NMe2	89	18:82
9	P18 - Boc-(S)-β-Me-Cys-OMe	91	48:52

^aReactions were conducted with 0.05 mmol of 1a and 1.5 equiv of 2a.

Following these findings, we designated P11 as the lead catalyst. Notably, P11 confers much higher selectivity over the analogous peptide bearing a native cysteine as the catalytic residue (entry 8), as evidenced by a significant increase in selectivity from 18:82 er to 5:95 er, which highlights the critical role of the β-substituent. Furthermore, a fully protected monomeric (S)-β-Me-Cys derivative was also evaluated as a catalyst in this transformation, yielding product in 48:52 er (entry 9). This outcome underscores the necessity of the peptide’s secondary structure for effective enantiocontrol.

Reaction Scope Studies.

Having identified a highly selective peptide catalyst for asymmetric hydrodifluoroalkylation, we explored the substrate scope of this method (Table 5). The model product 3a was synthesized on a 1mmol scale with comparable yield although with slightly lower selectivity compared to the 0.15 mmol scale, which is likely due to less efficient cooling on larger scales. Crystallization of 3a revealed its absolute configuration to be (R) through single-crystal X-ray diffraction analysis. Even under optimized conditions, 3b was synthesized in lower selectivity than 3a, prompting us to re-evaluate the series of para-substituted phenylamides (3b–3g). Electron-rich substrates (3c–e) exhibited enhanced selectivity compared to their electron-poor counterparts (3f–g). Homobenzylic amide (3h) was also obtained with high levels of selectivity. The N-methylated analog of 3a (3i) demonstrates the compatibility of this system with tertiary amides, and benzhydryl amide(3j) resulted in a slight decrease in enantioselectivity, potentially due to the increased steric bulk of the system. Additionally, we evaluated the impact of a chiral center adjacent to the hypothesized hydrogen-bonding site of the starting material on the selectivity of this transformation. Employing an enantiopure amide, we obtained the (S, R) diastereomer 3k in a 9:91 dr. When the enantiomer of the amide was used with the same olefin, the (R, R) diastereomer 3l emerged as the major stereoisomer with a 9:1 dr. The hydrodifluoroalkylation of isopropenylacetate (2a) with 2-bromo-N-(2,4-dimethoxyphenyl)-2,2-difluoroacetamide (1m) afforded product in high-yield and 95:5 er showing that ortho/para-disubstituted phenylamides are also tolerated (3m).

Table 5.

Substrate Scope of Asymmetric hydroalkylation^a

^aYields and enantioselectivities are from isolated material and are the average of two trials. Reactions were conducted on 0.15 mmol scale. The X-ray crystal structure of 3d was rendered using CYLview20.

¹1 mmol scale.

²At −41 °C with an acetonitrile and dry ice cooling bath for 4h.

³At −64 °C with a chloroform and dry ice cooling bath for 4h.

⁴With Ir(ppy)₃ as the photocatalyst.

Using 1m as a representative amide, we found that enol esters, enol ethers, and an enol phosphate were effective olefin substrates (Table 5). Enol esters bearing a quinolinyl (3n), adamantyl (3o), or benzyl (3p) moiety underwent highly selective hydroalkylation. 3n is particularly notable due to the presence of a basic nitrogen atom that could potentially interfere with the hydrogen-bonding network crucial for enantioinduction. An enol acetate bearing a cyclohexyl group yielded 3q in high yields but only 60:40 er, highlighting the role of methyl substitution in the enol esters for highly selective hydroalkylation. A methyl-substituted enol phosphate (3r) and trimethylsilyl enol ether (3s) were also tolerated, and cyclohexyl- and tert-butyl-substituted silyl enol ethers result in greater or equal enantioselectivity despite increased steric bulk (3t and 3u, respectively). This introduces a complementary route to access aliphatic enantioenriched secondary alcohols. Deprotection of 3n and 3s, followed by chiral HPLC analysis revealed that the absolute configuration of 3s was also (R). The hydrodifluoroalkylation of a derivative of a xanthine oxidase inhibitor, Febuxostat, occurred readily to afford the corresponding product with high selectivity, underscoring the high functional group compatibility of the current method (3w). A cyclic urea derivative underwent hydroalkylation, yielding product 3v in good selectivity. An exo-cyclic alkene inspired by Roberts’ work was evaluated, but low selectivity was observed (3x). Additionally, a dehydroalanine derivative was assessed; however, no reactivity was observed under the optimized conditions (3y).

With an established substrate scope, we sought to further investigate the compatibility of the reaction conditions with non-fluorinated substrates. Amide (4a) was explored as a radical precursor for hydroalkylation. When 2a was employed as the coupling partner, only moderate yield and selectivity were observed, suggesting that the geminal fluorine atoms in the fluorinated amide may contribute to the NCIs in the HAT transition state, which is supported by DFT calculations (vide infra). When a silyl enol ether was used as the coupling partner, 5b is obtained in 63% yield and 14:86 er. This suggests a different enantiodetermining transition state for the silyl enol ethers. Sulfone (4b) was also tested in the reaction, and the same trends were observed. Remarkably, despite the significantly different scaffold, the reaction of 4b with 2s proceeds efficiently, delivering the product 5d with both good yield and great enantioselectivity.

Lower temperature studies were performed at −41 °C for 4 selected substrates (1a, 1c, 1k, and 1m). Due to challenges associated with low-temperature photochemical reactions, these reactions were only run for 4 hours (see SI for experimental set-up). Even at reduced reaction times, 3a, 3c, 3k, and 3m were all obtained in comparable yields and with enhanced selectivity, achieving enantiomeric ratios above 95:5 for all four substrates. This highlights that even under the optimized conditions at −25 °C, racemic pathways are still present, however these substrate-dependent reaction rates are suppressed by additional cooling. To further explore this, 1i and 1m were subjected to reaction condition at −64 °C, but both yields and selectivity significantly decreased.

Finally, a derivative of the antibiotic penicillin was prepared and studied in the hydroalkylation with 1m (Table 6). When an achiral HAT catalyst was used, the hydroalkylation product was obtained in 45:55 dr, showcasing a slight preference towards one diastereomer. Employing P11 as the catalyst formed 3zb in 91:9 dr, reversing the inherent diastereomeric preference of the system. The high level of selectivity obtained highlights the catalyst’s control in diastereomeric induction. The enantiomer of P11 in this reaction produced the alternative diastereomer of 3zb (3za) with a 5:95 er, now matched with the inherent selectivity of the system.

Table 6.

Hydroalkylation of Penicillin derivative^a

^aYields and enantioselectivities are for isolated material and are the average of two experiments. Reactions were conducted on 0.15 mmol scale.

Mechanistic Studies.

We initially used [Ir(ppy)₂(dtbbpy)][PF₆] as the photocatalyst; however, switching to Ir(dF-ppy)₃ resulted in increased enantioselectivity (vide supra). To better understand this shift, we considered the contrasting electronic properties of heteroleptic and homoleptic iridium complexes. Heteroleptic iridium complexes are typically stronger oxidants in the excited state, while homoleptic complexes are generally more effective reductants. Based on these differences, we proposed two potential mechanistic pathways: reductive quenching and oxidative quenching of the photocatalyst. Stern–Volmer quenching (SVQ) studies were then conducted to further explore these mechanisms (See Table S13–S24).

SVQ studies conducted in chloroform revealed efficient electron transfer to [Ir(ppy)₂(dtbbpy)][PF₆] (*E^red(*Ir^III/Ir^II) values of 0.26 V vs Fc⁺/Fc in MeCN)¹⁹ from tertiary amines (E^ox = 0.49 and 0.59 V vs Fc⁺/Fc in CH₂Cl₂ for Et₃N and tris(2-(2-methoxyethoxy)ethyl)amine, respectively, See Figure S25). In comparing Et₃N and tris(2-(2-methoxyethoxy)ethyl)amine, Et₃N had the higher quenching rate constant, by a factor of 2.4 based on the SVQ constant (K_SV) (Figure 4A). This is consistent with Et₃N being easier to oxidize, as indicated by cyclic voltammograms (CVs) showing that Et₃N has a 100 mV less anodic peak potential (Figure 4A).

Figure 4.

(A) Stern—Volmer experiments, electrochemical studies for amine reductant, and proposed mechanism for [Ir(ppy)₂(dtbbpy)][PF₆]. (B) Stern–Volmer experiments and proposed mechanism for Ir(dFppy)₃

Building on this foundation, we proposed a catalytic cycle depicted in Figure 4A. Here, photoinduced electron transfer between *Ir(III) and the tertiary amine generates Ir(II) and an aminium radical cation, which subsequently undergoes deprotonation to form an α-amino carbon-centered radical. We hypothesize that this radical abstracts a bromine atom from the bromodifluoroamide to form the the ^•CF₂C(O)NR₂ radical, as is common.²⁰ This radical then adds to the alkene, forming a carbon-carbon bond and producing a carbon-centered radical that undergoes HAT from the thiol cocatalyst, yielding the difluoro-hydroalkylation product. The thiyl radical can then undergo single electron transfer with the reduced iridium photocatalyst and protonation, regenerating the active forms of both catalysts.

In contrast, SVQ studies with Ir(dFppy)₃ (*E^red(*Ir^III/Ir^II) values of −0.04 V vs Fc⁺/Fc in MeCN)¹⁹ in chloroform revealed that the tertiary amines could no longer quench the photocatalyst (Figure 4B). Instead, it was found that the bromodifluoroamides could quench this excited species (E^red = −2.01 V vs Fc⁺/Fc in CH₂Cl₂ for 1b). This signifies a transition from a reductive quenching of the excited state, *Ir(III)/Ir(II) to an oxidative quenching by electron transfer to the bromodifluoroamides. This is consistent with the Ir(dFppy)₃ excited state (*E^ox(*Ir^III/Ir^IV) values of −1.68 V vs Fc⁺/Fc in MeCN) being a less potent oxidant and a stronger reductant than the cationic [Ir(ppy)₂(dtbbpy)]⁺ excited species (*E^ox(*Ir^III/Ir^IV) values of −1.36 V vs Fc⁺/Fc in MeCN).^{19, 21}

The shift in the quenching mechanism leads to a different mechanism, with an Ir(III)/Ir(IV) cycle as illustrated in Figure 4B. In this mechanism, the excited photocatalyst transfers an electron to the bromodifluoroamide, cleaving the C–Br bond. The following steps of addition to the alkene and HAT from the peptide catalyst are similar to those in Figure 4A. The tertiary amine serves as a reductant to restore the photocatalytic cycle, also regenerating the RSH species to complete the HAT cycle.²² Radical trapping experiments with TEMPO successfully captured the key radical intermediates involved in the proposed mechanism (See Scheme S6).

Interestingly, in this mechanism, the formation of the prochiral C-centered radical is independent of the aminium radical cation, while the rate of formation of the aminium radical cation depends on the rate of formation of the difluoroalkyl radical. We hypothesize that this interplay modulates the relative concentrations of the aminium radical cation and the C-centered radical. Consequently, when comparing the two mechanistic pathways, the oxidative quenching mechanism in Figure 4B is expected to exhibit a lower background reaction rate, leading to enhanced enantioselectivity.

Computational Studies.

To better understand the factors responsible for the enantioselective HAT, we performed DFT calculations. Geometry optimizations were conducted in chloroform using the dispersion-inclusive ωB97XD²³ functional with def2SVP²⁴ basis set. Single point energies were calculated with the ωB97XD functional and the def2TZVPP basis set. Additionally, solvation effects were included through performing geometry optimization and single point energy calculations with the implicit CPCM²⁵ solvation model in chloroform²⁶ (ε=4.81). Finally, intrinsic reaction coordinate (IRC) calculations were performed to validate the nature of the transition states.

We conducted detailed conformational searches with CREST²⁷ on tetrapeptides and key intermediates, applying the quasiharmonic approximation from Grimme to compute the thermal corrections with a cut-off frequency of 50 cm⁻¹.²⁸ These quasiharmonic approximations were calculated using GoodVibes²⁹ and were employed to obtain more accurate Gibbs free energies. All DFT calculations were performed with Gaussian 16³⁰ on UCLA Hoffman2 and XSEDE³¹ supercomputers.

Initial studies were focused on peptide P6, and xTB metadynamics conformational analysis with CREST predicts that P6 favors a stable β-turn/β-hairpin secondary conformation (See Table S25 and Figure S26). The lowest energy conformation of the carbon center radical intermediate 6 is stabilized by two hydrogen bonds between the acetate group with the amide N–H (2.0 Å) and with the CH₂ group adjacent to the CF₂ group (2.3 Å).

In the lowest energy pro-chiral intermediate IM1, the substrate amide C=O hydrogen bonds with the N–H of the Aib residue. The subsequent HAT step with catalytic peptide P6 kinetically favors the formation of the R product (Pro-R) (black) via TS1-R by 1.5 kcal/mol, as compared to TS1-S (red) (Figure 5A). Our computed ΔΔG‡ reflects that the experimentally observed R product is formed about 12.6 times faster than the enantiomeric S product.³² Given that the product Pro-S is thermodynamically favored over Pro-R by 1.1 kcal/mol, the enantioselectivity of this reaction is kinetically controlled.

Figure 5.

(A) Computed free energy profile for hydrogen atom transfer from tetrapeptide P6 Boc-Cys-d-Pro-Aib-Phe-N(piperidine) to 6.(B) Distortion/interaction activation strain (D/IAS) analysis to determine the origin of enantioselectivity. (C, D) Effects of β-Me substitution on HAT transition states with P8 and P9.

We then studied DFT optimized enantioselectivity-determining transition states TS1-R and TS1-S. A distortion/interaction activation strain (D/IAS) analysis³³ revealed that the smaller substrate distortion in TS1-R was the major contributor to the difference in electronic energy (ΔΔE‡) and consequently promotes R-enantioselectivity (Figure 5B). Newman projections from the acetate O to the adjacent carbon show that the hydrogen bond between acetate group and the CH₂ group adjacent to the CF₂ group is preserved in TS1-R. The inductive effect of CF₂ makes this C–H⋯O a stronger hydrogen bond, as compared to the C–H⋯O hydrogen bond with a terminal methyl group in TS1-S (See Figures S30–S31).

Overall, our DFT calculation, along with D/IAS analysis indicates that the dissociation of C–H⋯O hydrogen bond in TS1-S leads to significantly larger substrate distortion energy, and thus kinetically favors TS1-R by 1.5 kcal/mol. Consistent with the experimental observation, Pro-R is the kinetically favored major product.

We subsequently investigated the effect of β-Me substitution on the HAT enantioselectivity. Our DFT calculations successfully reproduced the experimentally observed enantioselectivities, where β-(S)-Me substitution in P8 led to comparable selectivity to P6, and β-(R)-Me substitution in P9 led to diminished enantioselectivity (Figure 5 C, D).

We have also conducted DFT mechanistic studies with the best-performing tetrapeptide P11. Our calculations suggest that the experimentally observed R product is formed about 41 times faster than the enantiomeric S product (ΔΔG‡ = 2.2 kcal/mol), which is quite consistent with experiment, especially when the background rate is considered (Figure 6, See Table S28 and Figures S35).

Figure 6.

Computed free energy profile for hydrogen atom transfer from tetrapeptide P11 Boc-Cys-d-Pro-Aic-Chg-NMe₂ to 6.

CONCLUSION

In summary, we disclose the development of a photocatalytic reductive hydrodifluoroalkylation of terminal olefins facilitated by thiol-containing peptide catalysts. This process combines carbon-carbon bond formation with enantioselective hydrogen atom transfer (HAT), achieving enantiomeric ratios up to 96:4 and excellent yields under optimized conditions. Key to this success was the identification of peptides with N-terminal cysteine residues possessing (S)-β-methyl substitution, which exhibited the highest selectivities. X-ray crystallographic analysis provided structural insights into their enhanced performance, while mechanistic studies revealed that the choice of terminal reductant significantly influenced the reaction outcome by modulating a thiol-independent background reaction. Nevertheless, the highly selective cysteine-containing catalysts were able to overcome this deleterious process, and we were able to demonstrate the protocols versatility with a range of bromoamido amides, diverse alkene acceptors, and a penicillin derivative, where double diastereocontrol was achieved.

Computational investigations, including DFT calculations, aided in the elucidation of probable mechanisms for asymmetric induction and highlighted the critical influence of the diastereomeric configurations in β-methyl-Cys-containing catalysts. These findings underscore the importance of catalyst design in achieving high enantioselectivity.

This work establishes a foundation for designing enantioselective peptide catalysts for photochemical transformations, with significant potential for applications in synthetic and medicinal chemistry. Future efforts in our laboratories will focus on expanding this methodology to other systems amenable to asymmetric HAT coupled with computational studies to better elucidate the peptide-substrate interactions.