Prenyl Praxis: A Method for Direct Photocatalytic Defluoroprenylation

Abstract

The prenyl fragment is the quintessential constituent of terpenoid natural products, a diverse family which contains numerous members with diverse biological properties. In contrast, fluorinated and multifluorinated arenes make up an important class of anthropogenic molecules which are highly relevant to material, agricultural, and pharmaceutical industries. While allylation chemistry is well developed, effective prenylation strategies hav e been less forthcoming. Herein, we describe the photocatalytic defluoroprenylation, a powerful method that provides access to “hybrid molecules” that possess both the functionality of a prenyl group and fluorinated arenes. This approach involves direct prenyl group transfer under very mild conditions, displays excellent functional group tolerance, and relatively short reaction times (<4 h), which is the fastest photocatalytic C–F functionalization developed to date. Additionally, the strategy can be extended to include allyl and geranyl (10 carbon fragment) transfers. Another prominent finding is a reagent dependent switch in regioselectivity of the major product from para to ortho C–F functionalization.

Graphical Abstract

The prenyl fragment is ubiquitous in a multitude of natural products, and is the fundamental building block of the biodiverse and efficacious terpenoids which present impressive and diverse biological activities.¹ Consequently, the development of prenylation strategies within the context of natural products² has been widely studied.^2–3 Many prenylation strategies involve multistep procedures, and one-step prenylation has only been solved for a narrow subset of substrates.⁴ Even recently, Porco employed indirect allylation followed by cross-metathesis to affect prenylation to access a class of polyprenylated acylphloroglucinols.⁵

Meanwhile, fluorine substituents have been shown to impart a number of positive attributes, such as resistance to metabolic degradation, improved binding, enhanced lipophilicity, and passive diffusion of compounds across membranes.⁶ Not surprisingly, multifluorinated arenes make up an important class of molecules in pharmaceutical,⁷ material,⁸ and agricultural chemistry,⁹ but despite their importance, methods to build these molecules largely depends on just a handful of rather harsh reactions, such as the Balz-Schiemann decomposition of an aryl diazonium tetrafluoroborate salt,¹⁰ or the high temperature (ca 230 °C) halex process.¹¹ Arguably, this limitation has led to an overreliance on commercially available pre-fluorinated building blocks which can simply be incorporated into other molecules. Almost certainly, this leads to incomplete structure activity relationships studies.

Building on the strategy promoted by Braun¹², Richmond,¹³ Uneyama,¹⁴ and many others,¹⁵ we seek to address this limitation by developing reactions that start with inexpensive highly fluorinated arenes, and sculpt far more complex fluorinated arenes via C–F functionalization than have historically been synthetically accessible.¹⁶ We were particularly drawn to the prenyl motif due to its ubiquity within nature as well as its synthetic versatility. Accomplishing this goal would allow us to wed the natural prenyl group and the unnatural organofluorines to give hybrid molecules.

While allylation chemistry is relatively well developed, there are a number of serious challenges that arise as a consequence of the two additional methyl groups found within a prenyl group. As carbenium ions, they are prone to elimination,^{4a, 4c} and as anions regioselectivity issues arise (Scheme 1A).¹⁷ The use of a prenyl radical might lead to selectivity in the addition, but addition to an arene is expected to be a highly endergonic step.

Scheme 1.

Allylation and Prenylation of electron deficient perfluoroarenes

Several strategies have been explored to achieve related allylation and isoprenylation on perfluoroarenes¹⁸ via C–F functionalization both directly and indirectly (Scheme 1B–C). Sukbok^18c has shown that a Cu(NHC) catalyst is capable of C–H functionalization of highly fluorinated arenes (Scheme 1B), which has become increasingly relevant with recent improvements to hydrodefluorination technologies,^{15d, 19} but at the very least requires two steps. In an alternative approach, Kambe¹⁷ showed (Scheme 1C) that copper (II) chloride could facilitate formation of an prenyl magnesiate which would then undergo uncatalyzed addition to the perfluoroarene. The major product arises from linear addition of the most stable prenyl metal species, but gives rise to a total of 4 isomers in substantial quantities.

Inspired by the work of Zard who has popularized a number of allyl radicophiles,²⁰ we envisioned that such a reagent could be designed to facilitate C–F prenylation. Previously, we have shown that upon photocatalytic electron transfer to a perfluoroarene, an unstable radical anion results, which undergoes mesolytic cleavage to generate a fluoride and a perfluoroaryl radical.^{19e, 21} Our hope was that the radicophile would intercept the perfluoroaryl radical to regioselectively generate the key C–C bond. Addition would result in an alkyl radical intermediate that would undergo homolytic fragmentation of the β-leaving group, unmasking the prenyl group. We anticipated that the nature of the leaving group would be key to preventing simple hydrogen atom transfer (HAT) to the radical.^21e Vital to the rate of the β-fragmentation is the bond strength, and in this regard Zard has provided insight, demonstrating that the C–O bond of allyl alcohol (80.1 kcal/mol BDE)²² can be weakened by converting the hydroxy to an aryloxy group; rendering it susceptible to homolytic fragmentation.^{20, 23} This stabilization is mirrored by the phenoxy radical (PhO–H BDE = 87.3 kcal/mol) which is more stable than hydroxyl radical (HO–H BDE = 119.3 kcal/mol).²⁴

Thus, we began our investigation with the hope of finding a group capable of activating the C–O bond towards homolytic fragmentation (Table 1). As expected, reaction with isoprenyl alcohol (2a) itself simply results in hydroarylation of the alkene,^21b highlighting the need for an activating group that can increase the rate of fragmentation. Next, we assessed the 6-halopyridine motif (2b, 2c) previously explored by Zard²³ in lauroyl peroxide mediated transfer of xanthates to olefins. The pyridyl group is conveniently introduced via SNAr addition of isoprenyl alcohol. We were delighted to see the desired prenylated product (3a) in 54% and 43% respectively. The mass balance was primarily comprised of hydrodefluorination (HDF) as well as an oxidized version of the desired product. However, reagents 2b and 2c both underwent competitive [3,3]-sigmatropic rearrangement, which further complicated the situation and necessitate higher reagent loading to compensate for this background reaction. By blocking the ortho position with an ester group and increasing the steric demand (2d), we hoped to curtail both the rearrangement and the oxidation. Indeed, we halted the rearrangement, but unfortunately this reagent did not deliver desired prenylated product, and instead yielded only hydrodefluorination (HDF) product. This behavior is perhaps due to competitive and unproductive electron transfer to 2d rather than the perfluoroarene.

Table 1.

Screening of Prenyl sources

Observed: 2a - HDF, amine addition, C-O coupling, 2b and 2c - HDF, oxidized prenylation, [3,3]-rearrangement, 2d - >90% HDF, 2e - HDF, C-O coupling, multiple side products, 2f - mostly HDF, multiple side products, 2g - HDF, incomplete conversion, 2h - HDF, multiple side products, 2i - HDF, 2j - HDF, 2k - HDF. No further conversion to product over extended time.

The importance of a weak C–O bond can be observed in low yields of 2e and 2f, which would be expected to work better if the C–O bond were to break heterolytically. However, substrates 2g and 2h were expected to have a weaker C–O bond and still failed to give improved yields. We next evaluated aryl ethers formed from the perfluoroarene (2i and 2j). We were pleased to see that these prenyl transfer reagents afforded the desired product, with HDF as the only by-product.

It is possible that the fluorines on the reagent serve to prevent rearrangement and sterically reduce the activity of the resulting aryloxy radical, potentially involved in the formation of the previously observed oxidized product. However, 2k displayed significantly decreased activity even though it contained two prenyl groups. Given the positive results and the simple nature of 2j we opted to use it for further reaction development.

Using conditions that had facilitated C–F reductive alkylation,^21b using Blue LEDs we irradiated methyl pentafluorobenzoate and diisopropylethyl amine (DIPEA) as the electron source, along with 6 equiv of 2j at 0 °C. Previously, we observed that lowering the temperature could reduce the amount of competitive HDF in the related C–F arylation reaction.^21c We were pleased to see that the desired C−C-coupled product was formed as the major product in decent yield, albeit along with a significant amount of HDF product (Table 2, entry 1).²⁵

Table 2.

Optimization of reactions

Entry	Modifications	Yield of 3a(%)a	3a:3a’d	Time (h)
1	none	31/62	1.6:1	5/17
2	no cataylst or in dark	0	na	17b
3	without amine	0	na	17b
4	−10 °C instead of 0°C	29/64	1.7:1	5/25
5	DIPEA (1.0 equiv)	25/57	2:1	5/23b
6	DIPEA (1.8 equiv)	36/64	1.9:1	5/18
7	DIPEA (2.5 equiv)	38/57	1.3:1	5/15
8	0.25 mol% catalyst, DIPEA (1.8 equiv)	31/58	1.8:1	5/24
9	0.025 mol% catalyst, DIPEA (1.8 equiv)	6/28	1.6:1	5/24b
10	H2O (10 equiv), DIPEA (1.8 equiv)	57/65	1.9:1	3/4
11	H2O (15 equiv), DIPEA (1.8 equiv)	59/65	1.9:1	3/4
12	Entry 10 with 0.4 equiv TEMPO	39/68	2.1:1	4/10

^a determined by ¹⁹F NMR analysis. Reaction complete unless otherwise noted.

^bReaction did not go to completion over extended time, observed 0%, 0%, 83% and 44% conversion of 1 in entries 2, 3, 5 and 9 respectively.

^dReported for the final time point.

Control studies demonstrated the necessity of light, catalyst, and amine (entry 2-3). Carrying out the reaction at lower temperature increased the reaction time with no significant improvement in 3a:3a’ ratio (entry 4). While decreasing the DIPEA loading slowed the reaction, increasing amine resulted in more HDF (entry 5–7); ultimately, 1.8 equivalents of DIPEA produced the optimal yield. There appears to be a rate dependency on catalyst concentration (entries 8 and 9). Next, we investigated the effect of water on the reaction. Unexpectedly, the presence of water significantly accelerated the reaction from 18 h to 4 h and improved product: HDF ratio (entry 10-11). One explanation could be the increase in exothermicity of the reaction due to hydration of fluoride.²⁶ An alternative explanation is that water acidifies the pentafluorophenol generated in situ.²⁷ By doping the reaction mixture with pentafluorophenol we observed its inhibitory effect on the reaction progress. This effect was diminished upon the addition of water.²⁷ Recently, Wu showed that the competitive reduction pathway could be minimized by the addition of TEMPO,²⁸ we found it initially helped but ultimately slowed the reaction.²⁹

Ultimately, using 0.1 M substrate, 1.8 equiv DIPEA, 0.3 mol% fac-Ir(ppy)₃ and 10 equiv H₂O, we began evaluating substrate scope. Under these conditions a number of perfluoroarenes smoothly underwent C–F defluoroprenylation in good to modest yields. The reaction tolerates a number of functional groups including esters (Table 3, 3a-3c), nitriles (3d), CF₃ (3e), and perfluoroheterocyclic arenes (3g). Hexafluorobenzene, devoid of any additional electron-withdrawing functional group also proceeded to form 3f, though it required the use of 2b as the prenyl source. 3h-l are noteworthy as they demonstrate the preference for chlorine fragmentation over that of fluorine despite the position on the ring. Given that many chlorofluoro-starting materials are commercially available, it provides a convenient strategy for accessing complementary regioisomers. When a chlorine was placed at the site of preferential fluorine fragmentation (4 position for pyridine), it displayed moderately shorter reaction time compared to pentafluoropyridine (i.e. 3h compared to 3g).

Table 3.

Scope of prenylation

^a19F NMR yield determined using monofluorobenzene as internal standard. ^bisolated yield. ^cComprises of both products 3f required 2b as the prenylating source.

This could be because fragmentation of chloride ion is more exothermic than fluoride, resulting in a faster fragmentation event. Furthermore, the presence of chlorine substituents create greater steric demand than fluorine substituents, and could result in a relative increase the ground state energy.³⁰ Likewise, steric repulsion between the substituents in the radical anion may also accelerate its breakdown, leading to faster fragmentation. Interestingly, prenylation at the meta position was also possible but required longer reaction time (3i and 3j vs 3g). Whereas the mesolytic fragmentation of fluoride is highly regioselective, chloride is generally less regioselective and is temperature dependent (3g vs 3k and 3l), but still results in useful selectivities. This method can be used to access heterocyclic substituted perfluoroarenes like oxazoles (3m), benzimidazoles (3n) and benzothiazoles (3o). In all these reactions, HDF made up the mass balance.

Looking to expand the utility of the method, we applied this strategy towards allyl transfer to perfluoroarenes. As expected, allylation with transposition of the double bond proceeded smoothly to yield 5a-5f (Table 4) in good yield. Substitution at both the β-position (5c and 5d), and the γ-position were well tolerated (5e and 5f). While the yield was acceptable, in the case of unsymmetric 5f, the E/Z-selectivity was very modest (1.3:1) and did not display significant temperature dependence (See SI for details), which may limit its use in cases where the olefin geometry is essential.

Table 4.

Scope of allylation

^a19F NMR yield determined using monofluorobenzene internal standard. ^bisolated yield.

Encouraged by our success with prenyl transfer, we hoped to extend our strategy to geranyl transfer, adding a 10 carbon unit. Using the same optimized conditions, we reacted perfluoroarenes with pentafluoroisogeranyl ether (6, Table 5). We were concerned that the alkyl radical formed upon addition to the alkene would undergo intramolecular addition to the additional olefin (i.e. 5-exo-trig cyclization) faster than fragmentation of the pentafluorophenoxy radical.

Table 5.

Scope of geranylation

Determined by ¹⁹F NMR. Reactions completed in 6–8 h.

Thus, we were pleased to obtain geranylated product in similar yields to the prenylation reaction. A striking difference, however, was that the reaction took place preferentially at the position ortho to the substituent on the perfluoroaryl ring (c.a. ortho: para 1.6:1). This is particularly noteworthy as it is the first time that we have observed an external reagent capable of influencing the C–F regioselectivity in a photocatalytic C–F functionalization.³¹

Furthermore, while, 7a-7d were produced as ortho-para isomers, it demonstrated perfect diastereoselectivity, giving only the E-alkene.³² This selectivity is in stark contrast to the unsymmetric allyl derivative, 5f, which was produced as a mixture.

In conclusion, we have developed a strategy and reagents that enable photocatalytic defluoro-allylation, -prenylation, and -geranylation of perfluoroarenes. Further, as we moved from prenylation to geranylation, we observed a change in the regioselectivity of the major product from para-C–F functionalization to ortho-C–F functionalization. Our approach allows direct allyl, prenyl and geranyl substitution of C–F bonds using very mild conditions and short reaction times. This strategy should facilitate investigations involving synthesis of hybrid fluorinated analogs of natural products. Additionally, this reaction presents a number of interesting mechanistic facets which are currently being studied, and the findings will be reported in due course.