A Mild, Large-Scale Synthesis of 1,3-Cyclooctanedione: Expanding Access to Difluorinated Cyclooctyne for Copper-Free Click Chemistry

Abstract

We report the large-scale synthesis of 1,3-cyclooctanedione in five steps with 29% yield. This molecule is a synthetic precurser to difluorinated cyclooctyne, which participates in a bioorthogonal copper-free click reaction with azides. The final step demonstrates the first successful application of the Wacker-Tsuji oxidation to form a cyclic 1,3-dione.

In the past decade, bioorthogonal conjugation methods have emerged as powerful and increasingly indispensable tools for performing chemistry in biological contexts.^1–3 Notably, the strain-promoted azide-alkyne cycloaddition (SPAAC) has proven especially promising as a cytocompatible ligation method. The reaction boasts nearly all the advantages of the canonical copper(I)-catalyzed Huisgen [2+3] cycloaddition between an azide and a terminal alkyne, but requires no copper catalyst, which is cytotoxic at >0.1 mM concentrations.^4,5 Thus, the chemistry can be performed readily (k ~ 10⁻¹ M⁻¹ s⁻¹) in the presence of individual cells as well as full living organisms.⁶ Bertozzi and coworkers have pioneered the synthesis of several cyclooctyne reagents and have successfully applied their use as chemical probes to perform in vivo imaging to study glycosylation in zebrafish and mice.^7,8 In addition, other molecules for strain-promoted ractions with azides have recently been developed including (aza-)dibenzocyclooctynes,^9,10 biarylazacyclooctynone,¹¹ and oxanorbornadienes¹² which exhibit similar utility.

More recently, the use of Cu-free click chemistry has expanded from simple labeling procedures to include the formation of complex materials.^13–15 By end-functionalizing synthetic polypeptides with a difluorinated cyclooctyne reagent (DIFO3)¹⁶ and reacting with a four-arm poly(ethylene glycol) tetraazide in an aqueous medium, hydrogels are formed that allow for the direct encapsulation of cells.^13,15 Despite this preliminary success, the difficulty of the cyclooctyne synthetic preparation has ultimately hampered the widespread implementation of SPAAC in material formation to date. Additionally, utilizing these reagents in material fabrication requires gram-quantities of the reagents as opposed to the µg- to mg-scale required for labeling experiments. Thus, we sought to develop an improved synthetic route that enables DIFO3 to be formed on the multi-gram scale.

We found the main difficulty in preparing DIFO3 in significant quantity lies in producing large amounts of the key synthetic intermediate 1,3-cyclooctanedione. The published route of Pirrung and Webster¹⁷ (Scheme 1) for preparation of this compound has several undesirable characteristics, including the use of large molar equivalents of water-reactive sodium metal and diethylzinc.¹⁸ Pyrophoric organometallic reagents are not ideal for larger scale reactions and have in some cases been implicated in laboratory accidents.¹⁹ Moreover, the acyloin condensation (Scheme 1, first step) exhibited highly-variable yields in our hands. This is in part due to the nature of cyclizing larger rings: significant oligomerization/polymerization likely occured, as evidenced by the formation of a non-volatile, colored byproduct and reduced product recovery, even with high dilution and slow addition of diester. Thus, we sought to develop a mild, high-yielding synthesis of 1,3-cyclooctanedione starting from an inexpensive, readily-available substrate.

Scheme 1.

Traditional synthesis of DIFO3 from 1,3-cyclooctanedione intermediate.

Initially, we were inspired by a kg-scale preparation of 1,3-cycloheptanedione,^20,21 whose key step relies on the cycloaddition between dichloroacetyl chloride and 1-trimethylsilyloxycyclopentene. We hypothesized that this synthetic approach could be easily adapted to prepare 1,3-cyclooctanedione. However, we abandoned this route after experiencing difficulty obtaining significant coupling between 1-trimethylsilyloxycyclohexene and dichloroacetyl chloride. Ultimately, cyclooctanone was chosen as a starting point as it is inexpensive ($77 per 100 g via Aldrich), readily available, and several potential methods for installing the ketone functionality at the β-position were identified, including epoxidation and subsequent Pd-catalyzed rearrangement^22,23 as well as the direct Wacker-Tsuji oxidation²⁴ (Scheme 2).

Scheme 2.

Strategy for synthesis of 1,3-cyclooctanedione.

From the starting cyclooctanone (1), enone 5 was synthesized in four steps and 78% yield via α-bromination followed by elimination, an established method to introduce an α,β C=C double bond.²⁵ Bromination of ketone (1) was accomplished in excellent yield (90%, Scheme 3) with 1.5 equiv of Br₂ in ethanolic hydrochloric acid and the product was pure enough after standard aqueous workup to obviate chromatographic purification.

Scheme 3.

Complete synthesis of 1,3-cyclooctanedione (6) from cyclooctanone (1) with reaction conditions and obtained yields. Full reaction conditions, as well as ¹H, ¹³C, and COSY NMR and HR-MAS characterization for all compounds can be found in the supporting information.

Many attempts to effect direct dehydrobromination on 2 with various bases and solvent systems were made (KOH in isopropanol, DBU in toluene or dichloromethane,²⁶ and LiBr/Li₂CO₃ in DMF²⁷). All such attempts, however, failed to generate significant quantity of enone 5. Thus, a two-step protection/deprotection of the ketone moiety was adopted.²⁸ Formation of bromo ketal 3 under Dean-Stark conditions was achieved in near-quantitative yield and reasonable purity. Quantitative dehydrobromination of crude ketal bromide 3 was accomplished after reacting for 1 hr in neat 1,8-diazabicycloundec-7-ene (DBU) at 160 °C. We note that the direct conversion of cyclooctanone 1 to the bromo ketal 3 has been reported elsewhere^29,30. Crude product 4 was deprotected via trans-ketalization in PPTS/acetone/water, and resultant enone 5 was recovered in high yield after purification by flash chromatography. Note that while Plumet²³ reports a method to synthesize enone 5 in two steps from cyclooctanedione, we feel the extra two steps for the present scheme are justified: The overall yield is higher (78% vs 49%), both require just one chromatographic purification (for the final enone), and the palladium reagent used in the Plumet method (73 mol%) would add considerable cost to a large scale preparation.

Two potential methods to convert enone 5 to 1, 3-cyclooctanedione (6) were identified involving epoxidation then palladium-catalyzed rearrangement.^31,23 However, after an initial attempt to form the epoxide from enone (5) under Weitz–Scheffer conditions (aq. H₂O₂ or TBHP and cat. NaOH)³² failed, we found that Wacker-Tsuji oxidation conditions gave the desired dione in a single step (Scheme 3).²⁴ After briefly screening reaction conditions (Table 1), conversion of enone 5 to dione 6 was realized in 37% yield by treatment with 0.1 equiv Na₂PdCl₄ and 3 equiv tert-BuOOH in AcOH/H₂O (1:1) at 50 °C – undesired formation of unidentified side products resulted in the modest yield of this step. Given the high yields and simple purifications for the preceding steps, even with a yield <40% for the final step, this represents significant improvement over the standard methods for 1,3-cyclooctanedione preparation (Scheme 1)¹⁷. Furthermore, to the best of our knowledge, no successful application of the Wacker-Tsuji application to a cyclic substrate has ever been reported.

Table 1.

Isolated yields of dione (6) via the Wacker-Tsuji oxidation of enone (5) for a variety of catalyst and reactant amounts, as well as temperature.

Na2PdCl4 (mol equiv)	tert-BuOOH (mol equiv)	Temperature (°C)	Yield (%)
0.2	1.5	50	24
0.2	1.5	100	11
0.2	3	50	36
0.2	3	30	36
0.1	1.5	50	30
0.1	3	50	37
0.4	1.5	50	22

In summary, we have described a new method to synthesize 1,3-cyclooctanedione in five steps (Scheme 3) using an inexpensive starting material, robust reactions, and with only two steps requiring chromatographic purification. In addition, the first successful application of the Wacker oxidation in the direct synthesis of a cyclic 1,3-dione is shown. Using this method, we were able to generate 15+ g of dione 6 in 29% overall yield which can be used to further synthesize DIFO3 for SPAAC.