Introduction

The questions of how many carbonyl groups can be juxtaposed in a cyclic or acyclic compound and what are their unique properties and reactions are of continuing interest (Fig. 1a)1,2,3. Each adjacent carbonyl of a polycarbonyl compound modifies its neighbor and provides different reactivities and selectivities in its reactions. For example, pioneering work by Wasserman and coworkers4 demonstrated wide applications of vicinal tricarbonyl compounds (VTCs, 2) in the synthesis of natural products and synthetic intermediates5, and numerous examples have been reported that utilize the high electrophilicity of the central carbonyl of VTCs in diastereoselective and enantioselective nucleophilic addition reactions6,7,8,9. A close derivative of VTCs are diazodicarbonyl compounds, where the central carbonyl group is replaced by a diazo functional group. In this case, the central diazo functional group is stabilized by both adjacent electron-withdrawing groups, and their reactivity towards electrophiles (with diazodicarbonyl compounds) or nucleophiles (with metal carbenes) is diminished10. The formal introduction of another carbonyl group to VTCs gives a vicinal tetracarbonyl compound (TCC, 3), which represents an intriguing but highly underdeveloped class of organic compounds. An early study on their synthesis dates back to a report by Gray and Fuson from 1934 on the synthesis of dimesityl tetraketone 3a via acyloin condensation of mesityl glyoxal (Fig. 1b)11. Since then few efforts have been devoted to the synthesis of TCCs, yet even today only limited examples of symmetric TCCs have been described. The propensity of TCCs for nucleophilic addition reactions has been recognized1, 3, but applications are strictly limited;12 diazo analogs or examples with orthogonal masking of carbonyl groups are non-existent (Fig. 1c). As such, methods that would facilitate a more generalized synthesis of unsymmetric TCCs—ideally bearing complementary masking of each carbonyl group—are in high demand to access and to study the chemical properties of these building blocks.

Fig. 1: Synthesis of 1,2,3,4-tetracarbonyl compounds.
figure 1

a Polycarbonyl compounds. b Conventional synthesis of 1,2,3,4-tetracarbonyl compounds. c This work: reaction of enoldiazoacetates in the synthesis of unsymmetric 1,2,3,4-tetracarbonyl compounds.

Full size image

To achieve this goal, we hypothesized that an appropriately masked 1,2,3-tricarbonyl compound could be used to introduce the fourth carbonyl group, thereby achieving a modular synthesis of unsymmetric 1,2,3,4-tetracarbonyl compounds. Specifically, we considered 3-silyloxyvinyldiazoacetate as a suitable starting point due to its ease of accessibility and the presence of 3 carbonyl groups, which are present as an ester, diazo, and enolether functional group. The latter could then serve as an entry point to introduce the desired fourth carbonyl group by reaction with tert-butyl nitrite (TBN), a potent donor of the nitrosonium ion13, 14. At the same time, we anticipated the diazo functional group to remain intact thereby opening up a plethora of potential downstream transformations via carbene transfer reactions. As a net result, this strategy would allow the synthesis of a 1-ester-2-diazo-3-keto-4-oxime tetracarbonyl compound, which we considered to be readily convertible to 1-ester-2,3-diketo-4-oxime-containing tetracarbonyl compounds via metal-catalyzed carbene transfer reactions with water followed by oxidation.

Herein, we report a detailed study, on the reaction 3-silyloxyvinyldiazoacetate 6 with TBN, which gives facile access to the desired substrate class of tetracarbonyl compounds 7 (Fig. 1c) via nucleophilic addition and 1,2-silyl migration. Importantly, this approach now allows the synthesis of 1,2,3,4-tetracarbonyl compounds, where all 4 adjacent carbonyl groups bear different functionalities. We study this transformation with a range of different diazo compounds and show potential applications and limitations of the present method. Computational studies are conducted to gain an understanding in the exclusive formation of 1-ester-2-diazo-3-keto-4-oxime tetracarbonyl compounds and to rationalize the reaction mechanism. We conclude with applications of these compounds in carbene transfer reactions, which allows the introduction of protecting groups or the conversion into 1-ester-2,3-diketo-4-oxime functionalities.

Results and discussion

Synthesis of 1,2,3,4-tetracarbonyl compounds

To assess our hypothesis, we initially studied the reaction between Z-configured silyl-protected enoldiazoacetates Z−6 (R1 = Me, R2 = p-CF3C6H4CH2) and TBN in halocarbon solvents, but only unsatisfactory yields of the reaction product were obtained. However, when performed in acetonitrile, only a single product was formed in high yield for different ester functional groups and alkyl substituents R2 bound to the enolether with the diazo functional group remaining untouched (Fig. 2). Proton and carbon spectra, as well as HRMS, suggested the selective formation of either the E- or Z- stereoisomer of nitroso olefin (9) or a silyl-protected oxime (7). For an unambiguous structural assignment, we performed 15N-1H HMBC correlation spectroscopy that showed coupling at 390 ppm, which is in the range of an oxime (15N chemical shift: 300–400 ppm) but not a C-nitroso (15N chemical shift: 800–900 ppm) functional group15. HMBC data thus suggests the formation of silyl-protected oxime (7). To determine the scope of this reaction, we varied both the gamma substituent of the 3-silyloxyvinyldiazo ester and its ester alkyl group in reactions performed by the addition of excess TBN at 0 °C, then warmed to room temperature. Product yields range from 62–95% (Fig. 2), and reaction times showed a significant dependence on the ester alkyl group. Gamma ethyl and benzyl substituted vinyl diazoacetates 6 react at the same rates as the 3-silyloxyvinyldiazoacetates with a methyl substituent, but benzyl esters having an electron-donating substituent reacted faster. To further demonstrate the generality of this method, the tert-butyl(dimethyl)silyl (TBS) group as a protective group formed the oxime-OTBS-protected diazo compound 7i in high yield. A limitation lies with hydrogen as the gamma substituent; here only significantly reduced yields of 7j were obtained.

Fig. 2: Formation of vicinal 1-ester-2-diazo-3-keto-4-oximes from β-siloxy-α-diazovinylacetates with tert-butyl nitrite.
figure 2

a. Investigations on the applicability and structural confirmation, desilylation with TBAF and x-ray diffraction. Reaction conditions: tBuONO (3.5 equiv.) was added to a solution containing 6 (0.4 mmol) dropwise at 0 °C. The reaction was continued for the indicated time at room temperature. Another 1.0 equiv. of TBN was added after 24 h. bYield from reaction performed on a 1.00-g scale in parenthesis. c82% of desilylated benzyl-2-diazo-3-oxobutanoate was isolated.

Full size image

Further evidence for the formation of 7 was then found in a desilylation reaction with tetra-n-butylammonium fluoride (TBAF), which gave oxime 10 h. The structure of the latter was confirmed by x-ray analysis (Fig. 2) and now provides solid evidence for an intriguing example of a 1,2,3,4-tetracarbonyl compound, where all 4 carbonyl groups bear different functionalities; namely oxime, ketone, diazo, and ester functional groups.

Studies on the reaction mechanism

To better understand the formation of tetracarbonyl compound 7, we performed theoretical calculations at the (SMD = acetonitrile)-BP86-D3(BJ)/6–311 + +G(d,p)//BP86-D3(BJ)/6–31 G(d) level of theory, which has proven to be a suitable method in a previous study of a similar transformation, Fig. 3a16.

Fig. 3: Computational studies on the reaction mechanism.
figure 3

a Calculated reaction mechanism. b Reaction kinetics.

Full size image

We initiated calculation with the additional step of TBN onto the enol group of 6a. All attempts in locating a first transition state led us to a 1,2-addition transition state. This 1,2-addition step (TS1-1, ΔG = 29.9 kcal/mol, Supplementary Fig. 10) can be accelerated by the hydrogen bonding with additional tBuOH, which is consistent with experimental results and denoted as TS1 with an activation energy of 24.5 kcal/mol to give intermediate INT1. To verify the computed influence of tBuOH, we examined the reaction of enoldiazoacetate 6a with TBN with and without additional tBuOH. Indeed, a deceleration of the reaction was observed in the absence of tBuOH, which gave the reaction essentially a 40–60% yield after a 3–4 h reaction time, while in the presence of tBuOH, the reaction rapidly went to completion (80–90% yield) within the same time frame (Fig. 3b and Supplementary Figs. 3 and 4). Further calculation data concerned other possible pathways in the first step, such as [5 + 1] cycloaddition via TS1-2, metathesis-like process via TS1-3 and [5 + 2] cycloaddition via TS1-4, yet in all cases higher energy barriers (ΔG > 28 kcal/mol in Supplementary Fig. 12) were observed and are thus unlikely to account for the product formation.

From enoldiazoacetate-derived intermediate INT1, we analyzed a set of downstream pathways, which could account for the reaction outcome. We could identify a favorable proton transfer process through TS2 (ΔG‡ = 10.7 kcal/mol). We rationalize that the acetal group and NO group enhance the acidity of their α-hydrogen, which is required to promote this proton transfer step or direct tBuOH elimination via TS2 (ΔG‡ = 10.7 v.s. 12.5 kcal/mol for TS3)17. Importantly, in the case of enoldiazoacetate, a potential ring closing step through TS4 has a substantially higher energy barrier of 15.9 kcal/mol, which we reasoned to be due to reduced conformational flexibility caused by the methyl group. As a consequence, the formation of precursors that can lead to 1,2,3-triazine 1-oxides from 6a is less favored14. INT2 then undergoes elimination of tBuOH via TS5 and leads to the formation of a Z-nitroso silylenol ether INT3, which exhibits an ideally suited geometric orientation of the nitroso and the enolether group for subsequent silyl migration. Analysis of the structure of INT3 evidences a rather short N-Si distance of 2.57 Å, which indicates that the intermediate INT3 has a strong tetrel bonding interaction between N and Si atoms18.

We next examined the formation of the silyl-protected oxime 7a. Intramolecular silyl group transfer between oxygen atoms is a known process, but with limited examples11, and none of the kind proposed in Fig. 3a to explain the formation of 7. The tetrel interaction between N and Si in INT3 facilitates the intramolecular silyl migration step via the five-membered-ring transition state TS6. Finally, facile cis-migration of the silyl group via TS7 leads to the TIPS-protected (E)-oxime product 7a, which is the driving force that directs the initial vinylogous addition product from 3-silyloxyvinyldiazoacetates to irreversibly form 6. As part of these studies, we have also considered a one-step O-O 1,5-silyl migration process, which is unfavorable due to its high activation energy of 19.5 kcal/mol and leads to an undesired TIPS-protected (Z)-oxime product (Supplementary Fig. 11).

Applications of 1-ester-2-diazo-3-keto-4-oxime tetracarbonyl compounds

As anticipated from a diazo group between two carbonyls19, catalytic dinitrogen extrusion from 7 or 10 with either dirhodium tetraacetate or copper(I) catalysts does not occur at room temperature. However, treatment of 10c at a modestly higher temperature with rhodium acetate in acetone formed the adduct 11 expected from carbonyl ylide formation and carbonyl-induced ring closure in modest yield (Fig. 4) with 50% of reactant 10c recovered20. Interestingly, the same reaction performed with the TIPS-protected 7c under the same conditions did not produce the corresponding acetone adduct, and 90% of the reactant 7c was recovered.

Fig. 4: Carbene transfer reaction of oxime 10c.
figure 4

Rhodium-catalyzed reaction of unprotected oxime with acetone and X-ray crystal structure.

Full size image

Treatment of 7a with Rh2(esp)4 in 1,2-dichloroethane with added 5 equiv. of water under an atmosphere of air produced a surprising outcome. Instead of the expected alcohol by O-H insertion (12a/13a), the diazo carbon was converted to a carbonyl group in high yield; and the resulting product mixture was composed of both the keto form (8a) and its hydrate (14a) in approximately equal amounts (Fig. 5a). Since this process involves overall oxidation, we examined the catalyst by spectrophotometric analysis to determine if its oxidation state changed during the reaction, but we found no change in its oxidation state throughout the course of the reaction compared to the catalytic O-H insertion reaction with ethyl diazoacetate. However, when this reaction was performed under the same conditions but in an inert atmosphere, a new product was detected whose spectroscopic and HRMS analyses were consistent with enediol 12a which, depending on the solvent employed for analysis, is in equilibrium with its α-hydroxy carbonyl derivative 13a. Upon exposure to air, enediol 12a formed the 8a/14a mixture. Thus, the formation of 8a/14a is due to oxidation of 12a by molecular oxygen in a manner similar to, but faster than, ascorbic acid21, 22, and suggests that the oxime at the 4-position of 12a enhances this oxidation. Confirmation of this is found in DFT calculations (Fig. 5b). We found that dienol 12a undergoes a facile intramolecular proton transfer and leads to INT8, which is slightly endergonic. However, INT8 can readily undergo a proton-coupled electron transfer with triplet oxygen with an activation-free energy of 19.1 kcal/mol. Subsequent radical coupling leads to acetal intermediate INT10, which reacts via water-assisted H2O2 elimination to afford product 8a. This observation is in line with the electrochemical oxidation potential of enediol 12a (0.652 V in DMF), which suggests facile oxidation of enediol 12a.

Fig. 5: Reaction of a silyl-protected 1,2,3,4-tetracarbonyl compound in a rhodium-catalyzed carbene transfer reaction with water.
figure 5

a Experimental observation. b Analysis of tautomers. c Computational studies.

Full size image

The generality of this transformation can be seen in the results obtained with the diazo analogs in Fig. 6. Reactions performed in dichloroethane under an air atmosphere with 5.0 molar equivalents of water catalyzed by Rh2(esp)2 yielded the 1,2,3-tricarbonyl-4-oxime and its hydrate in modest to excellent yields usually within 24 h.

Fig. 6: Formation of 1,2,3,4-tetracarbonyl compounds.
figure 6

Rhodium-catalyzed reaction of oximes 7 with water and aerobic oxidation.

Full size image

In summary, we herein report on a streamlined and facile synthetic method of densely functionalized 1,2,3,4-tetracarbonyl compounds, wherein each of the four carbonyl groups bears different functionalities. The reaction of enoldiazoacetates with tert-butyl nitrite proceeds via C-alkylation of the enol functional group, without touching the diazo functional group, thereby installing an oxime onto the diazoacetate to give rise to 1-ester-2-diazo-3-keto-4-oximes in excellent yield. This approach not only solves the synthesis problem of 1,2,3,4-tetracarbonyl compounds, it also gives way to unprecedented vicinal tetracarbonyl compounds. Control experiments, and theoretical calculations provide important support of the reaction mechanism of this reaction. The work presented in this article will clearly stimulate further research into the reactivity of 1,2,3,4-tetracarbonyl compounds and allow their application in modern organic synthesis.

Methods

Formation of 1-ester-2-diazo-3-keto-4-protected oximes

tert-Butyl nitrite (3.5 equiv., 1.4 mmol) was added dropwise over 1 min to a round bottom flask containing a solution of enoldiazo compound 6 (0.4 mmol, 0.1 M in MeCN) at 0 °C under an N2 atmosphere. The reaction solution was slowly warmed to room temperature, and the progress of the reaction was followed by TLC until consumption of the enoldiazo compound was complete. The color of the solution went from orange/yellow to colorless. The solvent was then removed under reduced pressure, and the residue was purified by flash chromatography (hexane/EtOAc = 5/1) to give the desired diazo product 7.

Formation of 1-ester-2-diazo-3-keto-4-oximes

To a solution of the protected oxime 7 (0.1 mmol, 0.2 M in THF) in a dry 8-mL vial was added TBAF (1.5 equiv., 0.15 mmol) at 0 °C all at once. The progress of the reaction was followed by TLC until consumption of the protected oxime was complete, and the residue was purified by flash chromatography (DCM/MeOH = 9/1) to give the desired product 9.

Formation of 1-ester-2,3-diketo-4-oximes

In a dry 8-mL tube, to the solution of Rh2(esp)2 (1 mol%) in 1 mL of dichloroethane with H2O (5.0 equiv., 0.5 mmol), diazo-protected oxime 9 (0.1 mmol,) was added over 1 h with a syringe pump at room temperature. The color of the solution changed from light blue/green to light brown after 24 h. The residue was then purified by flash chromatography (hexane/ethyl acetate = 3/1) to give the desired product as an inseparable mixture of 1-ester-2,3-diketo-4-oximes 13 and their hydrate 14.