Diastereoselective deacylative aldol reaction of 3-acetyl-3-fluorooxindoles with aldehydes

The deacetylative aldol reaction of N -methyl-3-acetyl-3-fluoro-2-oxindole is optimized with benzaldehyde and the most appropriate conditions are used for the survey of the scope of this transformation with different aldehydes. The relative configuration of the resulting compounds is also confirmed. The reaction is diastereoselective and affords the syn-stereoisomer in good yields. DFT calculations are used for the explanation of this diastereoselectivity through a traditional chair-like transition state

In the context of the last example dealing with deacylative alkylations (DaA) 33 of carbon nucleophiles the generation of quaternary stereocenters is one of the main features.First, the acetyl or trifluoroacetyl group can be easily introduced and then, the corresponding carbon nucleophile is generated by a retro-Claisen condensation promoted by a base. 33,34Our studies in deacetylative nucleophilic addition onto alkyl halides and electrophilic olefins, 35,36 and onto π-allyl-palladium systems from allylic alcohols, 14 yielded 3,3-disubstituted oxindoles.This methodology was compared with the photochemical process to assess the synthesis of 3,3′bioxindoles. 37In addition, the 3-acetyl-3-fluorooxindole was generated and applied to the synthesis of 3substituted 3-fluorooxindoles by base-promoted alkylation or Michael-type additions and Palladium-catalyzed DaA, 34 but the aldol reaction was not explored yet.So, in this work (Scheme 1e), the optimization and full analysis of the scope of the reaction between 3-acetyl-3-fluorooxindoles and aldehydes as well as some mechanistic details are surveyed.

Results and Discussion
The preparation of the N-methyl-3-acetyl-3-fluorooxindole (2) was carried out following the procedure detailed in the literature, 34 starting from acylation of the commercially available N-methyloxindole by means of a Claisen condensation and subsequent isomerization of the carbon-carbon bond in basic media.The resulting compound 1, isolated in 89% yield, was next fluorinated at the 3-position using NFSI as electrophilic fluorinating agent and Mg(ClO4)2 as Lewis acid in ethyl acetate at room temperature.Despite of the formation of a small amount of the deacetylated fluorinated compound (N-methyl-3-fluorooxindol, 8%), the compound 2 was generated in 85% yield.Scheme 2. Synthesis of the starting fluorooxindole 2.
Once the substrate 2 was synthesized, the deacetylative alkylation reaction was optimized by using benzaldehyde as reagent in the bench reaction (Table 1).First, the effect of the base was studied.Benzyltrimethylammonium hydroxide (Triton B) and lithium ethoxide were tested at 0 °C giving total conversions but affording complex mixtures in the reaction crude (observed by 1 H NMR) with null diastereoselectivity (Table 1, entries 1 and 2).Cleaner reaction crude with low diastereoselectivity was achieved when the reaction was run in the presence of triethylamine (Table 1, entry 3).Any diastereoselectivity was detected in the reactions mediated by Triton-B at -20 °C with or without LiBr (3 equiv) as additive (Table 1, entries 3 and 5).This additive 25,26,27 was also employed in the reaction with Et3N at different temperatures (Table 1, entries 6-11) obtaining the best operational conditions, diastereoselectivity and conversion at -20 °C (Table 1, entry 8).The highest diasteresoselectivity determined at -78 °C took place at lower conversions even after 2 days stirring at this temperature (Table 1, entries 10 and 11).Another two additives as Mg(ClO4)2 and LiCl were tested furnishing lower diasteresoelectivities and conversions (Table 1, entries 12 and 13).Decreasing the amount of LiBr was not beneficial for the conversion obtaining the deacylated product (N-methyl-3-fluorooxindol) in notable quantities (from crude 1 H NMR) (Table 1, entries 14 and 15).Finally, the employment of anhydrous toluene or dichloromethane as solvents did not improve the result obtained in the reaction run in dry THF (Table 1, entries 16 and 17).The diastereomeric ratio of 3a (and for the other family of molecules 3) was calculated measuring in the crude 1 H NMR spectra the integrals corresponding to the benzylic hydrogen and compared with the analogous already published data.The reported coupling constants for compound 3a (JF-H) were shown to be 8 and 13.5 Hz for the syn-3a and anti-3a diastereoisomers, respectively. 32The reaction with assorted aldehydes was studied employing LiBr (3 equiv) in anhydrous THF, at -20 °C for 12 h (Table 2, entries 1-7).Ortho-substituents diminished the chemical yields, whilst meta-or parasubstituents afforded similar yields and in the range of the non-substituted aromatic aldehydes.Electrondonating and electron-withdrawing groups in the aromatic ring of the aldehyde are appropriate substrates for this reaction.In the particular example of 4-nitrobenzaldehyde, a reverse diastereoselectivity was observed obtaining the anti-3g aldol in a 70:30 ratio.This effect was also detected, but in not so large extension, in the published enantioselective addition (Table 2, entry 7). 32Heteroaromatic aldehydes were also good precursors affording good yields and notable syn-diastereoselectivity (Table 2, entries 8 and 9).Formaldehyde furnished a very high yield (90%) of the corresponding 3-hydroxymethyloxindole 3j (Table 2, entry 10).Ethyl glyoxylate and dihydrocinnamaldehyde afforded the highest syn-diastereoselectivities of this series of molecules in good yields (Table 2, entries 12 and 13).However, isovaleraldehyde gave a very low conversion of an equimolar amount of diastereosiomers, which could not be isolated after column chromatography (flash silica-gel) (Table 2, entry 11).Finally, the discrimination between aldol reaction vs Michael-type addition was assessed in the presence of cinnamaldehyde (Table 2, entry 14).The aldol reaction proved to be faster than the Michael-type addition, giving rise to aldol 3n in 85% yield and any traces of 1,4-addition were detected.In order to explain the generation of the major syn-diastereoselectivity, the transition states energies were calculated using DFT calculations.The lithium enolate (EN) is coordinated to several THF molecules, stabilizing it, and then, it coordinates with the benzaldehyde (BA) to form the intermediate complex (IC).IC immediately rearranges to give rise to the two possible diastereomeric transition states TS-RR and TS-SR (Scheme 3).The computational analysis revealed an energy gap between them of 2.4 kcal•mol -1 , transition state TS-RR being the most stable, which justifies the generation of the major product syn-3a (Figure 3).It was also found that the syn-diastereoisomer is more stable than the anti-diastereoisomer around 3.3 kcal•mol -1 (Figure 3).
In the TS-RR transition state, it is observed that the aromatic ring of the enolate (more deficient in electronic density than in the starting reagent 2) presents a key π-type electrostatic interaction with the arene group of the aldehyde with a certain charge density, such as occurred in the example of benzaldehyde.This effect would be favored by the presence of substituents that promote transfer of charge to the aromatic ring of the corresponding aldehyde (Me, OMe or even Br).However, the presence of a substituent that removes a high charge density (as occurred with the nitro group) possibly contributes to a greater repulsion between the aromatic resulting, in this case, that the TS-RS transition state is much more stable allowing the aldol reaction faster.

Conclusions
In this study it was demonstrated the efficiency of the non-asymmetric deacetylated aldol reaction of the N-methyl-3-acetyl-3-fluoro-2-oxindole under very mild conditions generating a quaternary center after this carbon-carbon bond Despite a non-asymmetric process, the titled reaction can be compared with the enantioselective version from the corresponding detrifluoroacetylated process published by Soloshonok et al.In general, chemical yields and diastereomeric ratios are in the same range in both studies.Also, the temperature (-20 °C) is a common feature, but, from the atom economy point of view, the methodology described here is more advantageous.Among them, it has been found that an intense π-type electrostatic interaction is the driving force that allowed the syn-diastereoselectivity when electron-rich aromatic aldehydes participated in the reaction.However, electron-withdrawing groups bonded to the aromatic rings did not favor this interaction increasing the anti-diastereoselectivity. At this moment, several representative examples of the reported molecules are being tested in parallel biological activity programs.

Experimental Section
General.All commercially available reagents (Acros, Aldrich, Fluka, Fluorochem and Merck) were used without prior purification.In all these examples, deoxygenated solvents (freezing pump procedure) were employed.Melting points were determined with a Reichert Thermovar hot plate apparatus and are uncorrected.IR spectra were collected with a FT-IR 4100LE (JASCO) (PIKE MIRacle ATR).The nuclear magnetic resonance experiments of proton and fluorine 1 H NMR and 19 F NMR (282 MHz, using trifluoroacetic acid as internal reference) and carbon 13 C NMR (75 MHz) were carried out in the Nuclear Magnetic Resonance units of the Research Technical Services of University of Alicante with the Bruker Avance AC-300 spectrophotometer and using deuterated chloroform and tetramethylsilane (TMS) as internal standard, unless otherwise indicated.Chemical shifts (δ) are given in ppm and coupling constants (J) in Hz.The following abbreviations are used to describe the multiplicity of signals: s (singlet), d (doublet), t (triplet), m (multiplet), bs (broad singlet).The rd value of the products is determined by the 1 H NMR spectrum.The low-resolution mass spectrometry analyzes by electron impact (IE) were carried out in the Mass Spectrometry units of the same service with a Shimadzu QP-5000 by DIP injection (Direct Introduction Probe), and high-resolution mass spectra were obtained with a Finnigan MAT 95S.Ions arising from the breaks are given as m/z with relative percentage intensities in parentheses.For the thin-layer chromatography (TLC) technique, prefabricated Schleicher & Schuell F1400/LS 254 silica chromatoplates were used and the results were visualized under UV light (λ = 254 nm).Flash chromatography was performed with silica gel 60, 0.04-0.06mm.For general computational details, please, see supporting information.

Figure 1 .
Figure 1.Representative examples of natural products incorporating the oxindole unit.

Scheme 3 .Figure 3 .
Scheme 3. Intermediate species and transition states defined in the stereochemical course of the aldol reaction to generate the reaction products syn-3a-RR and anti-3a-SR (racemates).

Table 1 .
Optimization of the deacetylative aldol reaction between fluorinated oxindole 2 and benzaldehyde a Total conversion determined by1H NMR of the crude reaction product.In all these examples, deoxygenated solvents (freezing pump procedure) were employed.bVerycomplexreactioncrudes were observed by1H NMR.cThe reaction was stirred at -78 °C for 2 days.d Starting material 2 and the corresponding deacylated compound were identified in the crude 1 H NMR. e Reaction performed in toluene.f Reaction performed in dichloromethane.

Table 2 .
Continued a Yield of the major diastereoisomer purified by flash chromatography.In all examples, deoxygenated solvents (freezing pump procedure) were employed.b Ratio determined by 1 H NMR of the crude product.c Conversion estimated by 1 H NMR.