Diversity-oriented synthesis has facilitated drug discovery by efficiently generating compound collections with high structural complexity and diversity13,14. Stereoisomeric compounds, with their different topographical features, usually result in distinct interactions with targeted proteins. Diverse molecular scaffolds based on carbon stereogenic centres have provided a wide range of chemical space for drug discovery15. Sulfur, with its multiple oxidation states, is widely present in biologically active compounds16. However, sulfur stereogenic centres are often overlooked as pharmacophores1,2,3, apart from the marketed chiral sulfoxides esomeprazole and armodafinil (Fig. 1a).

Fig. 1: Diverse chiral sulfur pharmacophores for drug discovery and their synthesis.
figure 1

a, Examples of biologically active compounds containing S(IV) and S(VI) stereogenic centres. b, Examples of diverse chiral sulfur pharmacophores for drug design and discovery. c, Synthesis of chiral sulfinate esters through dynamic kinetic resolution with chiral amine catalysts. d, Synthesis of chiral sulfinate esters through asymmetric condensation of sulfinates and alcohols with pentanidium (this work). ASNS, ʟ-asparagine synthase; ATR, ataxia telangiectasia and rad3-related; CDK, cyclin-dependent kinase; Et, ethyl; Me, methyl; Nu, nucleophile; Ph, phenyl.

Sulfoximine, a moiety with a S(VI) stereocentre, has become increasingly important in drug discovery owing to its unique physicochemical and pharmacokinetic properties4,5. Sulfoximine is tetrahedral and has been designed as a stable transition-state analogue to inhibit ʟ-asparagine synthase6. Although no candidate containing sulfoximine has been approved as a drug, several compounds such as AZD6738 and BAY 1000394 have entered clinical trials (Fig. 1a)8,9. Other sulfur stereogenic centres such as sulfinate ester, sulfinamide17, sulfonimidate ester and sulfonimidamide18 have started to attract attention owing to the advances made by sulfoximine (Fig. 1b). Although some methodologies have been developed for the racemic synthesis of these stereogenic centres19,20,21, the preparation of enantiopure sulfur stereocentres is still a formidable challenge22. Established methods mainly rely on stoichiometric amounts of chiral reagents23,24,25 or kinetic resolution of racemic substrates26,27. Only a handful of catalytic approaches have been reported and structural diversity is limited28,29,30,31,32.

Among the sulfur stereogenic centres, sulfinate ester holds the linchpin position for two reasons. First, several enantiopure sulfinate esters can be reliably and affordably derived from chiral alcohols. Next, a variety of approaches have been developed to convert sulfinate esters to other sulfur stereogenic centres33,34,35. Reports on the catalytic synthesis of enantioenriched sulfinate esters are scarce and all are based on dynamic kinetic resolution of sulfinyl chlorides with alcohols using peptides or Cinchona alkaloids as catalysts (Fig. 1c)36,37,38. The community is still yearning for a general and efficient method for the catalytic synthesis of enantiopure sulfinate esters with broad substrate compatibility. Considering the increasing interest in using novel chiral sulfur stereogenic centres as pharmacophores, a catalytic method suitable for the late-stage manipulation of drugs with diverse sulfur stereocentres is imperatively required.

Here we report the desymmetrization of pro-chiral sulfinate to afford enantioenriched sulfinate esters using pentanidium (PN)39,40 as a catalyst (Fig. 1d). Sulfinate, a stable and easily accessible reagent, is well known as a carbon-radical source for coupling via desulfitation41,42 or as a sulfur-centred nucleophile43. It is less known that sulfinate is an ambident nucleophile, and that the enantiotopic oxygen atoms are also potential nucleophilic sites. We realized this pathway through the use of ethyl chloroformate as the oxophilic electrophile. In the presence of pentanidium as a catalyst, sulfinate and ethyl chloroformate form a mixed anhydride intermediate, which in turn is converted to enantioenriched sulfinate ester through a replacement reaction with an alcohol. Sulfinate can also be easily derived from sulfur functional groups in drugs such as sulfonamide in celecoxib44 or methylsulfone in etoricoxib45. Thus, this methodology is suitable for late-stage diversification of existing drugs containing sulfur functional groups. In addition, drugs and drug intermediates containing an alcohol group, for example, the intermediate of remdesivir, an antiviral drug approved for the treatment of coronavirus disease 2019 (COVID-19), can be manipulated into novel analogues by replacing its phosphorus stereocentre (phosphoramidate) into a sulfur stereocentre. Phosphoramidate prodrugs, including remdesivir, are part of pronucleotide (ProTide) therapies for viral disease and cancer46,47,48. Similar to phosphorus, sulfur is also available in multiple oxidation states and a diverse range of structures; its adoption in place of phosphorus may lead to new therapies.

Optimization of reaction conditions

We started our investigation using potassium 4-methylbenzenesulfinate 1 as a model for sulfinate (Fig. 2). Several acyl chlorides (2a2g) and sulfonyl chlorides (2h, 2i) were selected, and the respective mixed anhydrides were generated as intermediates, which were immediately replaced by ethanol at the sulfur stereocentre to afford sulfinate ester 4 (Fig. 2, entries 1–9). Ethyl chloroformate 2a was found to give the most consistent and favourable results. Most of our earlier investigations were performed using pentanidium PN2 (entry 10). Serendipitously, we discovered that pentanidium PN1, containing a phenol substituent, provided a high level of stereocontrol. We speculate that this may be due to the selective hydrogen bonding between the phenol group on PN1 and sulfinate 1. When the phenol group was methylated to form pentanidium PN3, enantioselectivity decreased substantially (entry 11). We also detected the formation of acylated pentanidium PN4 during the reaction process when ethyl chloroformate 2a was used. When we prepared pentanidium PN4 separately and subjected it to the same reaction conditions, only low enantioselectivity was obtained (entry 12). It is likely that formation of pentanidium PN4 was an undesirable pathway, which additives such as thiolates (3a3d) mitigated to improve the reaction (entries 13–16; see Supplementary Information for details). Under the optimized conditions, we were able to perform the reaction at gram scale with high yield and enantioselectivity (entry 15).

Fig. 2: Optimization of reaction conditions.
figure 2

Reaction conditions: potassium sulfinate 1 (0.1 mmol), catalyst (5 mol%), 2a2i (1.6 equiv.), EtOH (1.2 equiv.), K2CO3 (1.1 equiv.), additive 3a3d (0.1 equiv.), Et2O (0.5 ml), −20 °C, 24 h. Isolated yields are reported, and e.e. values were determined by chiral high-performance liquid chromatography (HPLC) analysis. aReaction was performed on a 12.0 mmol scale and 1.94 g sulfinate ester 4 was isolated. Ar, aryl; tBu, tert-butyl.

Reaction scope

On the basis of these results, we proceeded to investigate the scope of sulfinates suitable for our methodology (Fig. 3). Electron-rich phenyl sulfinates with different substitution patterns gave the desired sulfinate esters with high stereoselectivity. Phenyl sulfinate esters with alkoxy substitution (57), alkyl substitution (8, 9), bulky mesityl group (10) and para-acetamido substitution (11) were obtained with high enantiomeric excess (e.e.) values. This reaction was also efficient to obtain a variety of phenyl sulfinate esters 1318 substituted with halogen atoms. Phenyl substitution at the para position gave sulfinate ester 19 and 2-trifluoromethoxybenzenesulfinate gave sulfinate ester 20, both with good levels of enantioselectivity. 4-Cyanobenzenesulfinate, which contained a strong electron-withdrawing cyano group, gave sulfinate ester 21 in moderate yield and with a moderate e.e. value. In general, strong electron-withdrawing aryl sulfinates gave moderate results. Several naphthyl sulfinates with different substitutions gave the corresponding sulfinate esters 2224 with high enantioselectivities. Thiophene and benzothiophene sulfinate esters 2529 were also obtained with excellent results. This methodology also worked well for alkyl sulfinates and enantioenriched products (3033) were efficiently generated. During these investigations, we found that the catalyst PN1 was quickly acylated to form PN4 in reactions with electron-rich sulfinates, which resulted in decreased yields and enantioselectivity. This was solved by using dipotassium phosphate (K2HPO4) as a base and increasing the amount of catalyst or additive.

Fig. 3: Reaction scope.
figure 3

Reaction conditions: potassium sulfinate (0.1 mmol), PN1 (5–10 mol%), 2a (1.3–1.6 equiv.), alcohol (1.0–1.2 equiv.), K2CO3 (1.1 equiv.), 3c (0.1–0.2 equiv.), Et2O (0.5–1.0 ml), −20 °C, 24 h. Isolated yields are reported, e.e. values were determined by chiral HPLC analysis, and d.r. values were determined by chiral HPLC or NMR analysis. aK2HPO4 (2.0 equiv.) instead of K2CO3. b3d (0.1–0.2 equiv.) as additive. cSodium sulfinate was used. d2a (2.0 equiv.), 3d (0.5 equiv.). eK2HPO4 (2.0 equiv.), 3d (0.2 equiv.), additional H2O (10 μl). fAlcohol (0.1 mmol), potassium sulfinate (0.15 mmol), 2a (0.2 mmol), K2CO3 (0.15 mmol). gMTBE (1.0–2.0 ml) as solvent. h2.0 ml of mixed solvent Et2O/EA (1:1). i2.0 ml of mixed solvent MTBE/EA (2:1). jAlcohol (0.1 mmol), potassium sulfinate (0.2 mmol), 2a (0.4 mmol), K2HPO4 (0.4 mmol), 3d (0.04 mmol), H2O (20 μl), Et2O (2.0 ml). See Supplementary Information for details. Boc, tert-butoxycarbonyl; EA, ethyl acetate; MTBE, methyl tert-butyl ether; TBS, tert-butyldimethylsilyl.

Next, we found that this methodology efficiently installed sulfur stereogenic centres to various alcohols with high functional group compatibility (Fig. 3). (S)-Glycidol was successfully functionalized, without affecting the epoxide moiety, to sulfinate ester 34 with a diastereomeric ratio (d.r.) of 98:2. With (R)-1,3-butanediol, primary alcohol was preferred over secondary alcohol with mono-sulfinylated product 35 obtained with d.r. of 97:3. To investigate the potential of using this methodology to complement the ProTide strategy, we investigated the functionalization of several nucleosides. The desired nucleoside sulfinate esters 3642 were obtained with moderate to high yields and excellent stereoselectivity. Sulfur stereogenic centres were successfully installed on the corresponding alcoholic intermediates of several marketed antiviral drugs such as zidovudine, sofosbuvir and remdesivir. We also demonstrated stereoselective sulfinylation of several bioactive cyclic alcohols, including cholecalciferol, cholesterol, epi-androsterone and menthol, to their corresponding sulfinate esters 4348. With cholesterol and menthol, we also showed that when ent-PN1 was used as the catalyst, the diastereomeric ratio is inverted, indicating catalyst control rather than substrate control of this reaction. Our methodology is suitable for primary and secondary alcohols including isopropanol; however, bulky tert-butanol, phenols and amines were not viable nucleophiles (Supplementary Information).

Modification of drugs

To demonstrate the generality and efficiency of our methodology, we prepared several complex sulfinate salts from drugs or drug intermediates (Fig. 4). Using sildenafil as an example, chlorosulfonation of an electron-rich arene led to its sulfonyl chloride intermediate, which can be easily converted to sulfinate 49 (Fig. 4a). Using our asymmetric condensation condition with ethanol, sildenafil sulfinate ester 50 was obtained with high enantioselectivity. Next, we converted methylsulfone on etoricoxib to sulfinate 51 through alkylation and in situ elimination of styrene (Fig. 4b)45. Subsequently, enantioenriched etoricoxib sulfinate ester 52 was obtained efficiently through our method. Recently, a group from Merck reported the preparation of sulfinates from primary sulfonamides through carbene-catalysed deamination44. Using this approach, we transformed several bioactive primary sulfonamides into their corresponding sulfinates (Fig. 4c). Likewise, the respective (S)-sulpiride, glibenclamide and valdecoxib sulfinate esters (5355) were afforded with high stereoselectivities.

Fig. 4: Functionalization and diversification of drugs.
figure 4

a, Synthesis of sildenafil sulfinate ester. b, Synthesis of etoricoxib sulfinate ester. c, Functionalization of sulfonamide drugs into sulfinate esters. d, Synthesis of celecoxib sulfinate esters using different alcohols. e, Late-stage diversification of celecoxib into a plethora of derivatives with sulfur stereocentres. Reaction conditions: aPotassium sulfinate (0.1 mmol), EtOH (1.0 equiv.), PN1 (20 mol%), 2a (2.1 equiv.), K2HPO4 (2.0 equiv.), 3a or 3d (1.0 equiv.), Et2O or toluene (1 ml), 0 °C or −20 °C, 24 h. b56 (0.1 mmol), ROH (1.0 equiv.), PN1 (5 mol%), 2a (1.6 equiv.), K2CO3 (1.1 equiv.), 3c (0.2 equiv.), H2O (10 μl), MTBE (1.0 ml), −20 °C, 24 h. See Supplementary Information for details. CuAAC, copper-catalysed azide–alkyne cycloaddition; LiHMDS, lithium bis(trimethylsilyl)amide; nPr, n-propyl; THF, tetrahydrofuran.

As mentioned, sulfinate ester is the ideal linchpin intermediate for late-stage diversification of drugs into a plethora of sulfur stereogenic centres. Therefore, we utilized celecoxib as a model to justify that our methodology is a valuable addition to the toolkit of drug discovery programmes (Fig. 4d, e). Primary sulfonamide on celecoxib was converted smoothly to celecoxib sulfinate 56. Asymmetric condensation of sulfinate 56 with cholesterol gave celecoxib–cholesterol sulfinate ester conjugate 57 with a high diastereomeric ratio (95:5). Through condensation of celecoxib sulfinate 56 with 2-propyn-1-ol, we obtained enantioenriched propargyl sulfinate ester 59. This nicely set it up for ‘click reaction’ with the azide group on zidovudine, generating celecoxib–zidovudine conjugate 60. Celecoxib sulfinate ester 58 was obtained with a high e.e. value as a versatile precursor of other S(IV)/S(VI) stereogenic centres and able to be substituted by various nucleophiles at the sulfur centre with inverted configuration. Methyl Grignard reagent and lithium enolate are useful nucleophiles, providing respective enantioenriched sulfoxides (61, 62). With lithium bis(trimethylsilyl)amide, we obtained directly unprotected sulfinamide 63. Both primary and secondary amines are effective nucleophiles through formation of lithium amide or activation with Grignard reagents. Inversion at the sulfur stereocentre provided respective enantioenriched sulfinamides 6466. Further imidations49,50 of celecoxib sulfinate ester 58, celecoxib sulfoxide 61 and celecoxib sulfinamide 66 gave the corresponding sulfonimidate ester 67, sulfoximine 68 and sulfonimidamide 69 in high yields and without erosion of e.e. values. Many of these enantioenriched S(IV)/S(VI) stereogenic centres have been previously deemed as synthetically challenging1,22.


We have presented a viable and unified synthetic strategy for the stereoselective preparation of sulfinate esters and related sulfur stereogenic centres. This methodology is mild and tolerates a wide range of functional groups, allowing it to be compatible with late-stage diversification of celecoxib and other marketed drugs. In addition, several marketed antiviral drugs, for example, zidovudine, sofosbuvir and remdesivir, can be redecorated with sulfur stereogenic centres through sulfinylation of their alcoholic intermediates. This approach complements the ProTide strategy through replacement of the phosphorus stereogenic centre with sulfur stereogenic centres. In view of the increasing use of sulfur stereogenic centres as pharmacophores, we believe that this methodology will ameliorate the toolkits of drug discovery programmes for the exploration of these pharmacophores.