An Unexpected Access to a New Sphingoid Base Containing a Vinyl Sulfide Unit

Abstract

An unexpected access to a new sphingoid base containing a vinyl sulfide unit is described. The process involves the ‘one-pot’ regioselective opening of an epoxide with a thiolate, followed by intramolecular acyl transfer, base-promoted elimination, and final hydrolysis. Excess sodium hydride and high dilution conditions are required for optimal yields. The process is amenable to a variety of thiolates. The resulting compounds can be regarded as new hybrid sphingoid bases that combine some structural motifs present in other reported sphingolipid analogues.

In the course of our current research on new modulators of sphingolipid metabolism, we required the preparation of several batches of the novel dihydroceramide desaturase inhibitor XM462. [¹] The synthesis relies on the regiocontrolled nucleophilic opening of epoxide 1, in DMF, with n-tridecylthiolate (A) to afford the required β-alkyl­thio alcohol 2A for its subsequent transformation into XM462 (Scheme  [¹] ). Compound 2A was obtained in acceptable yields in all cases. However, a careful TLC analysis of the crude reaction mixtures revealed the presence of a hitherto unnoticed, more polar compound in variable yields, which could be isolated and characterized as the vinyl sulfide 3A (around a 40:1 E/Z mixture, based on NMR data; Scheme  [¹] and Table  [¹] , entry 1). [²] Although, at first glance, formation of 3A seemed detrimental for our purposes, we recognized it as an interesting new hybrid sphingoid base that combines the presence of a sulfur atom at C5 position (as in XM462) with the C3 unsaturation present in a series of related ceramidase inhibitors containing a 2-aminoethanol amide framework. [³]

A plausible mechanism to account for the formation of 3A from epoxide 1 is depicted in Scheme  [²] .

Thus, intramolecular acyl transfer [4] of the initially formed alkoxide 4A leads to the bicyclic oxazolidinone 5A. Subsequent E2 elimination (with loss of CO₂) and final hydrolysis of the intermediate hemiaminal 6A during the aqueous workup affords vinyl sulfide 3A. The E stereo­chemistry observed for the major isomer is in agreement with the anti disposition between the acidic OC(H)SR proton and the carbamate leaving group in the most stable conformation, as illustrated in the ball and stick representation shown in Scheme  [²] . [5] The stereochemistry of the starting epoxide seems crucial for the outcome of the process, since no elimination products were observed from the corresponding epoxide epimeric at the C-O carbon. [6] Although the base-promoted formation of oxazolidinones structurally related to 5A has been reported in the literature, [7] no ensuing elimination products are usually found, as illustrated in our previous work involving a similar process from an aryloxy derivative structurally related to 2A. [8] We rationalize the unexpected reaction outcome from 2A as a result of the higher acidity of the vicinal methylene protons α to the sulfur atom in intermediate 5A, thus triggering the observed elimination process.

In an attempt to exploit the synthetic usefulness of this unexpected process, we carried out a systematic survey of the optimal reaction conditions leading to 3A, as shown in Table  [¹] . Interestingly, the stoichiometry of NaH seemed crucial, since an, at first sight irrelevant, slight increase in the excess NaH (2.0 equiv/mol epoxide vs. our previously reported [¹] 1.6 equiv/mol) led to a roughly 1:1 mixture of 2A and 3A (entry 1). Furthermore, by increasing the NaH/1 molar ratio to 5:1, no trace of 2A was observed in the crude reaction mixture and compound 3A (E/Z isomeric ratio around 40:1) was obtained in 49% yield (entry 2). In agreement with these observations and the above mechanistic proposal, the independent treatment of 2A under the same conditions cleanly afforded 3A in comparable yields (entry 5). In all cases, the use of THF as solvent led to slow conversion rates (entries 4 and 6). Moreover, since the process leading to 2A is formally catalytic in base, [9] the use of substoichiometric NaH led exclusively to 2A in excellent yield (Table  [¹] , entry 3). [¹0]

In an attempt to widen the scope of this process, thiols B-F were also evaluated as reaction partners for epoxide 1 (Table  [²] ). However, the reaction conditions shown above for 3A (Table  [¹] , entry 2) failed to give the expected vinyl sulfides 3, since epoxide-opening adducts 2B-E were obtained as the major reaction products (Table  [²] , entries 1-4). The only exception was 2-naphthylthiol F (entry 5), which afforded olefin 3F (9:1 E/Z mixture) together with the opening adduct 2F. After much experimentation with cyclohexanethiol B, which was chosen as model, optimized reaction conditions for the exclusive formation of olefin 3B were found. Thus, an excess base (NaH/1 molar ratio 15:1) and high dilution conditions (0.02 M relative to 1) turned out to be crucial for this unprecedented addition-elimination reaction (Table  [²] , entry 7). [¹¹] These optimized conditions led also to 3A in a slightly higher yield (entry 6) than that reported above and, most importantly, they also proved successful for the remaining thiols C-F used in this study (entries 8-11), although formation of minor amounts of epoxide-opening adducts 2E and 2F was still observed in some cases (entries 10 and 11). [¹²]

Table 2 Optimized Reaction of Epoxide 1 with Thiols A-F

Entrya	Thiol	R	Baseb	Concnc	Product, yield (%)	Other, yield (%)
1	B	cyclohexyl	5	0.09	2B 58d
2	C	t-Bu	5	0.09	2C 61d	3C 6d
3	D	Bn	5	0.09	2D 74d
4	E	Ph	5	0.08	2E 47d
5	F	2-naphthyl	5	0.08	3F 42d (9:1)e	2F 18d
6	A	n-tridecyl	15	0.02	3A 61f (40:1)e
7	B	cyclohexyl	15	0.02	3B 79f (10:1)e
8	C	t-Bu	15	0.02	3C 78f (6:1)e
9	D	Bn	15	0.02	3D 76f (13:1)e
10	E	Ph	15	0.02	3E 67f (9:1)e	2E 9f
11	F	2-naphthyl	15	0.02	3F 59f (9:1)e	2F 18f
a Reactions were carried out in DMF at 40 ˚C for 4 h (see ref. 12 for details). b Molar ratio of NaH relative to starting epoxide 1. c Molar concentration of epoxide 1 in DMF. d Calculated by NMR with dimethyl terephthalate as external standard. e Aprox E/Z ratio (based on ¹H NMR). f Isolated yield.

The vinyl sulfides herein reported were obtained as mixtures of E/Z isomers in ratios ranging from 6:1 (for the bulky tert-butyl derivative 3C, entry 8) to 40:1 (for 3A). In general, the acceptable overall yields obtained with the selected thiols (entries 6-11), together with their straightforward chromatographic separation from the corresponding adducts 2, whenever formed, and the possibility to carry out the separation of the minor Z olefin stereoisomer at a later synthetic stage, [¹³] makes this process synthetically useful. As far as vinyl sulfide 3A is concerned, acceptable yields are also obtained using the more practical reaction conditions shown in Table  [¹] (entry 2), in which the use of high dilution conditions and a large excess NaH can be avoided.

In summary, a new protocol for the synthesis of β-amino alcohols containing a vinyl sulfide unit is reported. This represents an addition to other methods based on radical chemistry, [¹4] as well a potential new entry into the vinyl sulfone framework, a well-established motif found in several families of Cys-protesase inhibitors. [¹5] In addition, our interest in vinyl sulfides 3 as a new, hybrid sphingoid backbone opens new possibilities for the design of sphingolipid analogues with potential biological properties. Efforts along this line are currently in progress in our group, and results will be reported in due course.

Supporting Information for this article is available online at http://www.thieme-connect.com.accesdistant.sorbonne-universite.fr/ejournals/toc/synlett.

Acknowledgment

Partial financial support from the ‘Ministerio de Ciencia e Innovación’, Spain (Project SAF2008-00706), CSIC (PIE 200880I034) and ‘Generalitat de Catalunya’ (Grant 2009SGR-1072) is acknowl­edged. MG is grateful to CSIC for predoctoral research training support within the JAE-Predoc program. The authors are grateful to Mrs. Eva Dalmau for technical assistance with HRMS.

References and Notes

• 1 Munoz-Olaya JM. Matabosch X. Bedia C. Egido-Gabas M. Casas J. Llebaria A. Delgado A. Fabrias G. ChemMedChem  2008,  3:  946 
  CrossrefSearch in Google Scholar
• 3 Bedia C. Canals D. Matabosch X. Harrak Y. Casas J. Llebaria A. Delgado A. Fabrias G. Chem. Phys. Lipids  2008,  156:  33 
  CrossrefSearch in Google Scholar
• 4 Ager DJ. Prakash I. Schaad DR. Chem. Rev.  1996,  96:  835 
  CrossrefPubMedSearch in Google Scholar
• 6 Koviach JL. Chappell MD. Halcomb RL. J. Org. Chem.  2001,  66:  2318 
  CrossrefSearch in Google Scholar
• 7 Triola G. Fabrias G. Casas J. Llebaria A. J. Org. Chem.  2003,  68:  9924 
  CrossrefSearch in Google Scholar
• 8 Grijalvo S. Matabosch X. Llebaria A. Delgado A. Eur. J. Org. Chem.  2008,  150 
  Crossref
• 14 Friestad GK. Jiang T. Fioroni GM. Tetrahedron  2008,  64:  11549 
  CrossrefSearch in Google Scholar
• 15 Santos MA. Moreira R. Mini-Rev. Med. Chem.  2007,  7:  1040 
  Search in Google Scholar

The higher ^³ J value found for C4-H in the major isomer was indicative of the E-stereochemistry. Compound 3A (major isomer): δ = 6.32, (d, J = 14.9 Hz, 1 H), minor isomer: δ = 6.11, (d, J = 9.5 Hz, 1 H).

Minimizations were carried out with the MM2 package found in Chem3D ultra (version 9.0) from Cambridge Soft.

The nucleophilic thiolate required for the opening of epoxide 1 can arise from deprotonation of the starting thiol by the transient alkoxide 4A (see Scheme  [²] ).

In light of these results, we cannot rule out at this point the possibility that the excess NaH reported in our previous work (ref. 1) was lower than assumed, probably due to adventitious reagent hydrolysis over prolonged storage.

Reactions carried out at higher concentration (0.05 M relative to epoxide 1) failed to give 3B, even in the presence of excess base (NaH/1 molar ratio 15:1). Concentrations between 0.05 and 0.02 M afforded variable mixtures of 2B and 3B.

Typical Procedure
A solution of 0.60 mmol of the required thiol in DMF (5 mL) was added dropwise over an ice-cooled suspension of NaH (250 mg of a 60% dispersion in mineral oil, 6.0 mmol) in DMF (10 mL) under Ar. Once the addition was complete, the reaction mixture was allowed to warm to r.t. and stirred for an additional 30 min until a suspension formed. A solution of epoxide 1 (100 mg, 0.4 mmol) in DMF (5 mL) was next added dropwise to the above suspension, and the reaction mixture was heated to 40 ˚C. After stirring for 4 h, the reaction was cooled to r.t. and quenched with H₂O (2.5 mL), dried with anhyd MgSO₄, and filtered. The solids were washed with Et₂O (3 × 5 mL) and the combined filtrates were evaporated to dryness to afford a crude residue, which was purified by flash chromatography. Data for compounds reported in Table  [²] are collected in the Supporting Information.

For example, the major E-isomers of a series of amides of 3A could be easily purified by conventional chromatog-raphic methods. Unpublished results.

References and Notes

• 1 Munoz-Olaya JM. Matabosch X. Bedia C. Egido-Gabas M. Casas J. Llebaria A. Delgado A. Fabrias G. ChemMedChem  2008,  3:  946 
  CrossrefSearch in Google Scholar
• 3 Bedia C. Canals D. Matabosch X. Harrak Y. Casas J. Llebaria A. Delgado A. Fabrias G. Chem. Phys. Lipids  2008,  156:  33 
  CrossrefSearch in Google Scholar
• 4 Ager DJ. Prakash I. Schaad DR. Chem. Rev.  1996,  96:  835 
  CrossrefPubMedSearch in Google Scholar
• 6 Koviach JL. Chappell MD. Halcomb RL. J. Org. Chem.  2001,  66:  2318 
  CrossrefSearch in Google Scholar
• 7 Triola G. Fabrias G. Casas J. Llebaria A. J. Org. Chem.  2003,  68:  9924 
  CrossrefSearch in Google Scholar
• 8 Grijalvo S. Matabosch X. Llebaria A. Delgado A. Eur. J. Org. Chem.  2008,  150 
  Crossref
• 14 Friestad GK. Jiang T. Fioroni GM. Tetrahedron  2008,  64:  11549 
  CrossrefSearch in Google Scholar
• 15 Santos MA. Moreira R. Mini-Rev. Med. Chem.  2007,  7:  1040 
  Search in Google Scholar

The higher ^³ J value found for C4-H in the major isomer was indicative of the E-stereochemistry. Compound 3A (major isomer): δ = 6.32, (d, J = 14.9 Hz, 1 H), minor isomer: δ = 6.11, (d, J = 9.5 Hz, 1 H).

Minimizations were carried out with the MM2 package found in Chem3D ultra (version 9.0) from Cambridge Soft.

The nucleophilic thiolate required for the opening of epoxide 1 can arise from deprotonation of the starting thiol by the transient alkoxide 4A (see Scheme  [²] ).

In light of these results, we cannot rule out at this point the possibility that the excess NaH reported in our previous work (ref. 1) was lower than assumed, probably due to adventitious reagent hydrolysis over prolonged storage.

Reactions carried out at higher concentration (0.05 M relative to epoxide 1) failed to give 3B, even in the presence of excess base (NaH/1 molar ratio 15:1). Concentrations between 0.05 and 0.02 M afforded variable mixtures of 2B and 3B.

Typical Procedure
A solution of 0.60 mmol of the required thiol in DMF (5 mL) was added dropwise over an ice-cooled suspension of NaH (250 mg of a 60% dispersion in mineral oil, 6.0 mmol) in DMF (10 mL) under Ar. Once the addition was complete, the reaction mixture was allowed to warm to r.t. and stirred for an additional 30 min until a suspension formed. A solution of epoxide 1 (100 mg, 0.4 mmol) in DMF (5 mL) was next added dropwise to the above suspension, and the reaction mixture was heated to 40 ˚C. After stirring for 4 h, the reaction was cooled to r.t. and quenched with H₂O (2.5 mL), dried with anhyd MgSO₄, and filtered. The solids were washed with Et₂O (3 × 5 mL) and the combined filtrates were evaporated to dryness to afford a crude residue, which was purified by flash chromatography. Data for compounds reported in Table  [²] are collected in the Supporting Information.

For example, the major E-isomers of a series of amides of 3A could be easily purified by conventional chromatog-raphic methods. Unpublished results.

• Supplementary Material