Radical Functionalization of Unsaturated Amino Acids: Synthesis of Side-Chain-Fluorinated, Azido-Substituted, and Hydroxylated Amino Acids

A range of enantiomerically pure protected side-chain-fluorinated amino acids has been prepared (13 examples) by treatment of protected amino acids containing unsaturated side chains with a combination of Fe(III)/NaBH4 and Selectfluor. The modification of the conditions by replacement of Selectfluor with NaN3 allowed the preparation of side-chain azido-substituted amino acids (five examples), which upon catalytic hydrogenation gave the corresponding amines, isolated as lactams (four examples). Radical hydration of the unsaturated side chains leading to side-chain-hydroxylated protected amino acids has also been demonstrated.


■ INTRODUCTION
The synthesis of enantiomerically pure amino acids continues to attract substantial attention, and the development of new methods for the preparation of side-chain-functionalized amino acids is the focus of much effort. 1 In particular, the synthesis of amino acids with side-chain fluorine substitution has been widely studied, 2 specifically in the context of the preparation of (S)-γ-fluoroleucine, a component of the Merck Cathepsin K inhibitor odanacatib. 3 An early approach to (S)-γ-fluoroleucine ethyl ester relied on the use of the Schollkopf bis-lactim ether. 4 The synthetic routes that have been developed more recently include the use of aspartic acid as a chiral pool starting material, 5 a titanium-catalyzed asymmetric ene-reaction, 6 use of glycine-Schiff base alkylation followed by enzymatic resolution, 7 and more recently direct photochemical fluorination of leucine itself. 8 This followed an earlier indirect method which employed photochemical bromination of leucine, followed by treatment with AgF. 9 The latter two approaches rely on the selective radical cleavage of the tertiary C−H bond of leucine for their success. A complementary method makes use of AgF/tetra-n-butylammonium fluoride as the fluorine source, in conjunction with a manganese porphyrin catalyst. 10 In previous work, we have reported a single example of the synthesis of the γ-fluorinated cyclopentylalanine derivative 2a using the ionic addition of hydrogen fluoride (HF) (using HFpyridine) to an unsaturated amino acid 1a (Scheme 1). 11 This compound was a key component of a Cathepsin S inhibitor. Although this synthetic approach was rather inefficient, largely due to the in situ tert-butoxycarbonyl (Boc)-deprotection of the substrate under the conditions of the reaction, it was nevertheless a very considerable improvement over the previous method used for making this protected amino acid. 12 The report by Barker and Boger on the new radical addition of HF to unactivated alkenes, using Fe(III)/NaBH 4 , with Selectfluor as the fluorine source, 13 prompted us to revisit our general strategy. Indeed, after we had completed the majority of the work reported in this paper, another group reported the application of the Barker and Boger method to the synthesis of Fmoc-(S)-γ-fluoroleucine 4 (in presumed 94% enantiomeric excess (ee), based on the ee of the starting material), using the 2-propenylalanine derivative 3. 14 We now report an in-depth study of this reaction, to assess its generality and scope for the preparation of side-chain-fluorinated amino acids as well as a possible extension by trapping the intermediate alkyl radicals to allow the introduction of other functional groups into the amino acid side chains, by exploring the use of subsequently reported variants of the Boger process. 15 ■ RESULTS AND DISCUSSION Preparation of Unsaturated Amino Acids. The initial goal was to prepare a representative selection of protected enantiomerically pure unsaturated amino acids, with double bonds at both the γand δ-positions. A range of cycloalkenylalanine derivatives was prepared by room-temperature Negishi cross-coupling of cycloalkenyl triflates (5a−d, 6, and 7) with the serine-derived zinc reagent 8, according to our previously reported methods from the protected iodoalanine derivative 9 (Scheme 2, Table 1). 16 It was established that extending the reaction times for the cross-coupling generally resulted in higher isolated yields of the desired products (1a− d) than we had previously observed. 11,16 In addition, two more functionalized vinyl triflates 6 and 7 were successfully coupled, giving the corresponding protected amino acids 10 and 11.
Although there is less precedent for the use of vinyl bromides in the Negishi cross-coupling with zinc reagent 8, a range of vinyl bromides (12−15) was successfully coupled under the same conditions that we had employed for coupling the vinyl triflates to give the expected unsaturated amino acids (16−19) (Scheme 3, Table 2). The cross-coupling reactions of (E)-and (Z)-2-bromo-2-butene with zinc reagent 8 were stereospecific, leading to the isomeric products (E)-18 and (Z)-18, respectively. Interestingly, the use of an excess of the commercially available isomeric mixture of 2-bromo-2-butene (E/Z, 1.9:1, as determined by 1 H NMR, with both isomers used in excess relative to zinc reagent 8) in the cross-coupling resulted in the formation of a mixture of the products 18 that was substantially enriched in the (E)-isomer (E/Z, 7:1) ( Table   2, entry 5), indicating that the Negishi cross-coupling of the zinc reagent 8 with (E)-2-bromo-2-butene (E)-14 was faster than with (Z)-isomer 2-bromo-2-butene (Z)-14 (by a factor of 3.5). This most likely reflects a faster rate of oxidative addition of the less hindered (E)-isomer of compound 14 to Pd. The most notable example was the successful cross-coupling of zinc reagent 8 with 1-bromocyclobutene 15 leading to the cyclobutenyl alanine derivative 19, which is a rare example of the use of 1-bromocyclobutene in the Negishi crosscoupling.
A small selection of unsaturated amino acid substrates 20− 22 was prepared by our previously reported method using copper-catalyzed allylation of the zinc reagent 8 (Scheme 4). 17 Again, we observed that extended reaction times at room temperature resulted, in most cases, in improved yields over those we had previously observed.
Radical Addition of HF to Unsaturated Amino Acids. Barker and Boger's original report contained examples of the radical HF-addition to terminal alkenes containing an ester or a protected amine as well as a dipeptide in which the terminal alkene was incorporated into a tyrosine residue. 13 Therefore, it was reasonable that unsaturated amino acid substrates would also undergo the reaction. In the event, the application of the conditions reported by Barker and Boger to the cyclopentenylalanine derivative 1a resulted in the formation of 2a, identical to the material we had previously prepared, 11 but in substantially higher yield (64%) and without unproductive protecting group removal and using less toxic reagents (Scheme 5). The structure of compound 2a was confirmed by X-ray analysis and showed a hydrogen bond from the carbamate N−H to the carbamate carbonyl of the neighboring molecule as well as a very distinct alignment of the C−F bond with the C−F bond of the closest neighbor.  We were pleased to find that with the exception of the 4,4difluorocyclohexen-1-yl alanine derivative 11 (vide infra), the procedure proved applicable to the other substrates that we had prepared, giving the desired HF-addition products in good to excellent yields (Table 3). To establish whether or not the reaction proceeded without racemization, both enantiomers of the 3-methyl-3-butenylglycine derivatives 16 and ent-16 were each separately subjected to the radical HF-addition conditions; the products 25 and ent-25 were obtained with high ee (99 and 98%, respectively, as determined by chiral phase high-performance liquid chromatography (HPLC)), thus establishing that the reaction did indeed proceed without racemization. When the two isomeric alkene substrates 17 and 21 were separately subjected to the radical HF-addition conditions, the same tert-fluoride 26 was obtained, with a slightly higher yield from the starting material 17 with the more substituted double bond (cf. entries 9 and 13, Table 3). When the radical HF-addition was attempted on the 4,4difluorocyclohexen-1-yl alanine derivative 11, the conversion was incomplete and purification difficult. We, therefore, modified the procedure by increasing the amounts of Fe(III) and Selectfluor and by the dropwise addition of a solution of NaBH 4 (in 0.1 M NaOH) over a period of 6 h, which resulted in the isolation of the desired product 24 in comparable yield to that obtained with the other substrates. This modified procedure may be successful with other more challenging substrates. The X-ray crystal structure of compound 24 was obtained, confirming the structure and showing similar intermolecular interactions in the solid state to those observed in the X-ray structure of 2a.
The addition of HF to the substrates 18 and 20 resulted in the formation of products 27 and 29 each with a new stereogenic center. In each case, the removal of the Boc-protecting group with trifluoroacetic acid (TFA) to give the corresponding trifluoroacetate salts, 31 and 32, followed by analysis by NMR methods established that, in each case, the new stereogenic center had been formed with very low diastereomeric excess (de) (specifically less than 5% de), entirely consistent with the results of Barker and Boger. 13 Radical Addition of HN 3 to Unsaturated Amino Acids. Given the efficiency of the radical addition of HF to unsaturated amino acids and the precedent from Leggans et al. that a combination of Fe(III)/NaBH 4 and NaN 3 as the azide source promoted the efficient radical addition of HN 3 to alkenes, 15 we have explored the application of this method to the functionalization of unsaturated amino acids. The formation of amino acids with side-chain nitrogen functionality is of interest, since the compounds are analogues of lysine. Direct application of the literature protocol for the hydroazidation of citronellol 15 to unsaturated amino acid 16 resulted in the incomplete conversion of the starting materials. However, a minor modification, in which the excess of reagents was increased, resulted in a very efficient conversion of 16 to the corresponding tertiary azide 33. The application of this modified method to a representative selection of unsaturated amino acids resulted, in each case, in the formation of the expected tertiary azide in good to excellent yields (Scheme 6, Table 4). The two isomeric alkenes 17 and 21 gave the same tertiary azide 36, with a slightly higher yield being observed when using the more substituted alkene 17 (Scheme 7), as already observed during the HF-addition to the same substrates.

ACS Omega
Article Each of the azides 33−36 was reduced by catalytic hydrogenation to give the corresponding free amines, which underwent spontaneous cyclization to give the corresponding lactams 38−41, which were each isolated in high yields (Scheme 8, Table 5). The structure of the lactam 40 was confirmed by single-crystal X-ray structure analysis.
Radical Alkene Hydration. The final transformation that we briefly investigated was the radical hydration of unsaturated amino acids using the combination of Fe(III)/NaBH 4 and air as a radical trap. 15 Such a process would lead to amino acids with a side-chain hydroxyl group. The optimization of this particular radical addition process proved challenging and appeared to be critically dependent on the means by which the air was introduced into the reaction mixture. Most reliable results were obtained when the air was introduced using a sintered gas inlet, and in two cases (using 16 and 17 as substrates), moderate yields of the corresponding hydroxylated amino acids were obtained, albeit isolated as the corresponding lactone 42 in one case (Scheme 9).

■ CONCLUSIONS
In conclusion, the generality and scope of the radical HFaddition to unsaturated amino acid substrates have been established, further demonstrating the functional group tolerance of the Boger radical hydrofluorination process. In combination with our previously reported methods for the preparation of unsaturated amino acids (extended to the Negishi cross-coupling with vinyl bromides in this paper), this constitutes an effective and direct method for the preparation of important side-chain-fluorinated amino acid derivatives. Furthermore, radical hydroazidation of unsaturated amino acid substrates has been demonstrated, allowing the synthesis of side-chain amino-substituted amino acids. Finally, radical hydration of unsaturated amino acid substrates has been demonstrated, but further work is required to establish the full scope of this transformation.

■ EXPERIMENTAL SECTION
High-resolution mass spectrometry (HRMS) measurements were carried out using electrospray (ES) ionization, with a time-of-flight mass analyzer. IR spectra were recorded as thin films or using attenuated total reflection. The synthesis of N-Boc-β-I-Ala-OMe was accomplished by the literature methods. 18 For the preparation of zinc reagent 8 and general procedures A, B, and F, flame-dried glassware was used. Cycloalkenyl triflates 5a−d were prepared by general literature methods 19 from the corresponding ketones and used without purification. Triflates 6, 20 7 21 and 1-cyclobutenyl bromide 15 22 were each prepared by literature methods.
Preparation of Zinc Reagent 8. Zinc powder (3 equiv) was suspended in dry dimethylformamide (DMF) (3 mL) under nitrogen, and iodine (nine crystals) was added immediately. A change in color from colorless to dark brown and colorless again was observed. N-Boc-β-I-Ala-OMe 9 (1 equiv) was added followed immediately by iodine (three crystals), the aforementioned color change was observed once more, and the insertion process was allowed to proceed for 30−60 min.
General Procedure A, Coupling of 8 to Vinyl Triflates and Halides. Immediately after the preparation of 8, Pd 2 (dba) 3 (2.5 mol %) and SPhos (5.0 mol %) were added to the reaction vessel along with an excess vinyl halide/triflate (≈1.6 equiv), as indicated in the individual procedures below. The reaction mixture was stirred under a gentle flow of nitrogen for 72 h. The reaction mixture was then filtered through a silica plug eluting with EtOAc. The organic phase was washed with water (2 × 50 mL) and brine (50 mL). The organic phase was dried with Na 2 SO 4 , filtered, and the solvent removed under reduced pressure. Purification was carried out by flash column chromatography (EtOAc in petroleum ether (40−60) mixtures over silica).