Stereospecific α-(hetero)arylation of sulfoximines and sulfonimidamides

Abstract

The occurrence of sulfoximines and sulfonimidoyl groups in biologically active molecules within pharmaceuticals and agrochemicals has notably increased in the past decade. This increase has prompted a wave of discovery of methods to install S(VI) functionality into complex organic molecules. Traditional synthetic methods to form α-substituted sulfonimidoyl motifs rely on S-C bond disconnections and typically require control of the stereogenic S-centre or late-stage modification at sulfur, and comprise multistep routes. Here, we report the development of a stereospecific, modular S_NAr approach for the introduction of sulfonimidoyl functional groups into heterocyclic cores. This strategy has been demonstrated across 85 examples, in good to excellent yield, of complex and diverse heterocycles. Sulfoximines, sulfonimidamides and sulfondiimines are all compatible nucleophiles in the S_NAr reaction and hence, the methodology was applied to the synthesis of four sulfoximine-containing pharmaceuticals. Of these synthetic applications, most notably ceralasertib, an ATR inhibitor currently in clinical trials, was synthesized in an eight-step procedure on a gram-scale.

Introduction

The ability of sulfur to adopt a range of oxidation states (II-VI) with defined molecular geometries has led to many advancements in the discovery sciences.^1–4 From materials to medicines, sulfur-containing functional groups are pervasive across disciplines. The more common S(VI) functional groups, such as sulfones and sulfonamides, have attracted the most attention – finding their way into many drug discovery programs and greater than 70 FDA approved drugs.⁵ More recently, the exploration of historically neglected sulfonimidoyl S(VI) functional groups, containing S=N and S-N bond(s), has provided novel clinical candidates for a variety of indications as well as important agrochemicals (Figure 1).^6–12

Figure 1: Importance, value and accessibility of sulfonimidoyl groups.

(A) Biologically active heterocycles containing α-substituted sulfonimidoyl functional groups. (B) Structural and physiochemical features of sulfonimidoyls. (C) Different disconnections to install sulfonimidoyl functional groups, a traditional approach (a) in 4–7 steps and a stereospecific S_NAr approach (b) in a single step (this work). The modular disconnection (b) can be used with bifunctional sulfoximines and widely available heterocycles. R¹ in approach (a) is limited by the nucleophilic sulfide and benzylic activated heterocycles, approach (b) provides variability for R¹ utilizing t-Bu substituted sulfoximines with widely available halogenated heterocycles. Het, heterocycle; S_NAr, nucleophilic aromatic substitution; PG, protecting group; LG, leaving group.

Sulfoximines, the mono-aza S=N variants of sulfones (found in 1–5),^13–18 have recently been accepted in medicinal chemistry as bioisosteres or viable replacement groups for carboxylic acids, alcohols, sulfones and sulfonamides (Figure 1B).^6,19 Additionally, sulfonimidamides can serve as bioisosteres for amines, sulfones and sulfonamides.^9,12 The unique H–bond donor and acceptor properties of the sulfonimidoyl groups allow them to mimic a wide range of other functionality while commonly providing other advantages, such as a chiral environment and increases in aqueous solubility.^9,20 Recent enthusiasm over the physiochemical properties of sulfoximines (and other sulfonimidoyl groups)^8,9,19 has led to an exponential increase in their use to improve pharmacokinetic (PK) and pharmacodynamic (PD) properties during lead optimization studies.^6–8,21

A specific example of this was demonstrated by AstraZeneca during the discovery and subsequent development of their ATR inhibitor, ceralasertib (1).¹³ Amid the final optimization stage of the drug discovery program, a sulfone was replaced for a sulfoximine. The resulting introduction of a sulfoximine led to an increased aqueous solubility while maintaining potency, which allowed for the advancement of 1 to the clinic where it is currently undergoing multiple phase II clinical trials.²² Other discovery programs at Bayer, ^3,14,18,23 Pfizer,²⁴ Genentech,^25,26 Hoffman-La Roche,²⁷ Novartis,²⁸ Nestlé Skin Health²⁹ and Corteva Agriscience,³⁰ have been actively researching neglected S(VI) functional groups with respect to methods for their installation and incorporation into lead scaffolds.

Furthermore, the novelty of sulfonimidoyl groups, with their inherent stereochemical and additional spatial vectors capable of modifications, provides ample opportunities for new intellectual property (IP) development. An increase in patent applications and issuances within the last ten years is a growing testament to the untapped potential of sulfoximines, sulfonimidamides, sulfondiimines and related functional groups. Pioneering work by Bolm,^31–33 Bull,^34–36 Johnson,³⁷ Luecking^38–40 Maruoka,^41,42 Sharpless,^43–45 Willis^46–48 and others for the creation and modifications of S(VI) functionality has given rise to new possibilities in the field.^49–51 However, despite the increase in methods to access neglected sulfonimidoyl-containing compounds, there has been relatively little advancement toward the modular installation of these groups to pharmaceutical scaffolds.

Owing to the growing attention of higher order sulfur-based functional groups as bioisosteres and PK modulators in the pharmaceutical sciences, the unmet need for their incorporation into medicinally relevant structures with an emphasis on asymmetric control must be addressed. Traditional methods to introduce α-substituted sulfonimidoyl units relies on S–C bond disconnection (a) (Figure 1C) involving a laborious 4–6 step synthetic sequence from carboxylic acids or esters.^{12–14,16–18} Limitations of disconnection (a) include the lengthy step count, challenging asymmetric control at the stereogenic S-center and the difficulty of late-stage modifications at sulfur – the last of which explains the high prevalence of methyl substituted sulfoximines and sulfondiimines. Other methods for the synthesis of α-arylated sulfoximines rely on exotic transition metal-catalyzed systems and conditions accompanied by limited scope with respect to both sulfoximines and aryl coupling partners – especially with regards to heterocycles.^38,52

To address the aforementioned limitations and provide a modular approach to aid in discovery efforts, a straightforward solution was sought that can be applied to readily available pharmaceutically relevant heterocyclic building blocks. Disconnection (b) (Figure 1C) outlines an S_NAr approach for the installation of α-substituted sulfonimidoyl functionality to widely available electrophilic heterocycles. While scattered reports of sulfones undergoing S_NAr exist,^53–55 this alternative disconnection will provide discovery chemists with a multifunctional method for the introduction of highly oxidized neglected sulfur moieties of varying complexity in a single step. The utilization of enantiopure sulfonimidoyl nucleophiles grants an enantiospecific entry into α-heteroarylated products that will transform the targeted synthesis process, relieving the pressure of relying on chiral separation techniques and loss of material.

Results and discussion

Development and scope.

The requisite sulfoximines were accessible from the direct oxidation/imination of sulfides or sulfoxides via Bull and Luisi’s sulfoximine synthesis,³⁶ followed by N–H protection (Figure 2). Although the modern method developed by Willis for the synthesis of sulfondiimines could be used to access dimethyl sulfondiimine 16,⁴⁷ the direct imination of dimethyl sulfide with t-BuOCl and NH₃ was chosen for this application.⁵⁶ Orthogonally protected sulfonimidamide 20 was made available from a 4-step procedure starting from disulfide 18. Chiral alkyl sulfoximines can be made readily available via Maruoka’s S-alkylation strategy from N-Piv protected sulfinamide 22.⁴¹ Enantiopure t-Bu methyl sulfoximine 23 provides a bench-stable chiral bifunctional sulfoximine linchpin, which will be highlighted in the synthetic applications below. To maintain the practical nature of an S_NAr approach, we sought a general, operationally simple and scalable procedure that could be carried out with minimal manipulations.

Figure 2: Synthetic routes to access S(VI) nucleophiles.

Sulfoximines were synthesized by oxidation/imination of readily available sulfides (12) followed by imino (N-H) protection. Sulfondiimine 17 was made available from oxidative bis-amination of sulfide 15 followed by bis-imino protection. Sulfonimidamide 20 was accessed via oxidative amination of disulfide 18 followed by oxidative imination and imino protection. Enantiopure bifunctional sulfoximines such as 23 arise from pivaloyl protection (decagram scale) of commercially available chiral t-Busulfinamides followed by enantiospecific S-alkylation – the t-Bu group can be cleaved, and the sulfur atom functionalized via S-alkylation at a later time (vide infra). MeOH, methanol; DCM, dichloromethane; ACN, acetonitrile; THF, tetrahydrofuran; PhI(OAc)₂, (diacetoxyliodo)benzene; DMAP, 4-(dimethylamino)pyridine; Piv₂O, pivalic anhydride; PG, protecting group; Me, methyl; Bz, benzoyl; Piv, pivaloyl; Boc, tert-butyloxycarbonyl; Cbz, benzyloxycarbonyl; Ts, para-toluenesulfonyl (tosyl); TBS, tert-butyldimethylsilyl; TMS, trimethylsilyl; PMP, para-methoxybenzyl; rt, room temperature.

A variety of bases and reaction conditions were screened (see Supplementary Table S1 for full reaction screen) with respect to the S(VI) nucleophile 24 and electrophilic heterocycles 11 (Table 1, entries 1–6). Two different procedures, cryogenic (NaHMDS, −78 °C, entries 1–4) and thermal (NaH/15-crown-5, 50–80 °C, entry 6), were found to be compatible with a wide range of electrophilic heterocycles. For heterocycles containing more than one electrophilic site, trace amounts of bis-S_NAr product were observed and can be circumvented by adjusting addition rate of NaHMDS (see Supplementary Troubleshooting sections). Both methods were designed for ease of use by premixing the nucleophile and electrophile followed by the addition of base. Due to the increased acidity of the α-H in the S_NAr product 25 relative to the S(VI) nucleophile 24, at least 2 equivalents of base (entry 2) are required for 1° and 2° nucleophiles (1.1–1.5 equivalent for 3°) to obtain full conversions and high yields. Thermal conditions were employed for less electron-deficient heterocycles (entry 6), such as pyridines, that were unable to undergo the S_NAr reaction at room temperature (heating the reaction mixtures with NaHMDS lead to decomposition and side product formation). In nearly all cases, a 1:1 stoichiometry of nucleophile and electrophile provides good to excellent yields under the optimized reaction conditions.

Table 1:

Optimization of sulfonimidoyl S_NAr conditions with heterocycles.

Entry	PG	Heterocycle	Base (eq.)	Temperature	% yielda
1	Bz	pyrimidineb	NaHMDS (1.1)	−78 ºC to rt	55%
2	Bz	pyrimidineb	NaHMDS (2.2)	−78 ºC to rt	91%
3	Bz	pyridinec,d	NaHMDS (2.2)	−78 ºC to rt	0%
4	Ts	pyridinec	NaHMDS (2.2)	−78 ºC to rt	32%
5	Ts	pyridinec	NaH (2.2), 15-crown-5 (2.2)	rt	0%
6	Ts	pyridinec	NaH (2.2), 15-crown-5 (2.2)	rt to 80 ºC	81%

All reactions were performed on 0.25 mmol scale in THF or dioxane (0.1M) under an argon atmosphere.

^aYield determined after column chromatography.

^b4,6-Dichloro-2-(methylthio)pyrimidine was used as the heterocycle.

^c2-Fluoropyridine was used as the heterocycle.

^dBz transfer to a-position of sulfoximine was observed as the major product – desired product was not isolated. THF was used with NaHMDS (2M in THF). Dioxane was used with NaH and 15-crown-5 ether. PG, protecting group; LG, leaving group; Ph, phenyl, t-Bu, tert-butyl; Me, methyl; Het, heterocycle; THF, tetrahydrofuran; Bz, benzoyl; Ts, para-toluenesulfonyl (tosyl); NaHMDS, sodium bis(trimethylsilyl)amide; rt, room temperature.

With two procedures in hand that provide reactivity with electron-deficient and (relatively) more electron-rich heterocycles, the S(VI) nucleophile scope was investigated (Table 2). Symmetrical pyrimidine 26 was chosen to be the model heterocyclic electrophile due to its electronic nature as well as a practical scaffold for further synthetic manipulation. Various protecting groups including hydrolytically cleavable groups (28 N-COp-tol, 29 N-Bz,), commonly used tosyl (N-Ts, 30), N-Boc (31), N-Cbz (32) and silyl groups (N-TBS, 33; N-TMS, 39) were all compatible to provide good to excellent yields (Table 2). When an N-TMS protecting group is employed, silyl group cleavage is observed under the work-up conditions to provide free sulfoximine 40 (N–H) in 77% yield, allowing for the introduction and deprotection of a sulfoximine unit to heterocycles in a single step. N-Cyano (N-CN, 41) sulfoximines, a commonly used imino N-substituent,⁵⁷ undergo the S_NAr smoothly in high yields (86%).

Table 2.

Nucleophile scope containing enolizable sulfoximine, sulfonimidamide, and other S(VI/IV) functional groups.

All reactions were performed on 0.25 mmol scale, unless otherwise stated, in THF or dioxane (0.1M) under an argon atmosphere. All yields were determined after column chromatography.

^aMethod A was used.

^bMethod B was used.

^cDecomposition observed after 14 days at room temperature – increased stability at −20 or −80 °C under argon (see Supplementary Handling of Reagents section).

^dReactions were performed on gram scales (6 examples). The stereospecific transfer of sulfoximines was determined by chiral HPLC. PG, protecting group; Het, heterocycle; Me, methyl; Ph, phenyl; Et, ethyl; p-tol, para-toluene; Bz, benzoyl; Ts, para-toluenesulfonyl (tosyl); Boc, tert-butyloxycarbonyl; Cbz, benzyloxycarbonyl; TBS, tert-butyldimethylsilyl; TMS, trimethylsilyl; Piv, pivaloyl; PMP, para-methoxybenzyl, THF, tetrahydrofuran; NaHMDS, sodium bis(trimethylsilyl)amide.

A myriad of N-Bz protected sulfoximines were screened to provide a wide nucleophile scope as seen in Table 2. Sulfonimidoyl S_NAr is not limited to the previously mentioned primary sulfoximines. Both secondary (43, 51) and tertiary dimethyl sulfoximine (44) examples resulted in the desired S_NAr product in high yields. Cyclic α-substituted nucleophiles, such as cyclopropyl (45, 46), cyclobutyl (47) oxetane (48) and azetidine (49) provided sterically congested heterocyclic sulfoximines with increased molecular complexity in a single step. Enantiopure sulfoximines (52-55) were investigated and determined to undergo an enantiospecific S_NAr reaction with pyrimidine 26 on gram-scale and in high yields (75–90%). The utility of this transformation is two-fold: 1) introduction of an asymmetric sulfoximine unit without erosion of enantiopurity (as determined by chiral HPLC, see Supplementary Chiral HPLC analysis section) and 2) capability of further modifications at sulfur upon t-Bu cleavage, providing a platform for late-stage diversifications.

Arene-substituted sulfoximines containing electron donating and withdrawing groups were well tolerated (56-58). As expected, the presence of other acidic functionality (O–H or N–H) and enolizable groups (MeCO-R) were not compatible. Protection of these reactive groups, such as acetonide 59, provides access to masked carbonyl groups that can be later manipulated. Heterocycle-containing sulfoximines including pyridyl substituents (60, 61), a saccharin analog (62), and benzothiazine oxide (63) can be appended to other heterocyclic moieties in good yields (65–82%). In the cases of 62 and 63, method B was required due to the poor solubility of the nucleophiles at low temperatures. Cyclic aliphatic sulfoximine 64 and those containing heteroatoms (65, 66) were also compatible and gave diastereomeric mixtures (ca. 4:1 to 1:1) in good yields (62–78%).

With an established sulfoximine scope, we turned our attention to neglected sulfonimidamides and sulfondiimines to determine their compatibility under our optimized S_NAr conditions. Orthogonally-protected sulfonimidamide 67 proved to be a suitable nucleophile that give the desired S_NAr product in 82% yield – the first example of the direct installation this functional group to a heterocycle. The bis N-Bz protected sulfondiimine 68 was also a suitable nucleophile using both method A (65% yield) and method B (92% yield). Classical oxidized sulfur groups, such as sulfonamides, sulfones and sulfoxides afforded the expected S_NAr products in 77–99% yields (69-73). The six different gram-scale examples found in Table 2 demonstrate the scalability of the method without a diminishment in yield.

Next, the electrophilic scope with regards to common place heterocyclic scaffolds in drug discovery was investigated. Three different sulfoximines were used to interrogate electrophilic reactivity, each with a different protecting group that proved critical (Table 3). For most electron deficient systems (e.g. triazines and pyrimidines), a benzoyl (N-Bz) protecting group sufficed. For less reactive substrates, benzoyl transfer to the sulfoximine α-carbon was observed. To eliminate protecting group transfer, more robust PGs (N-Piv, N-Ts, N-Boc) were used. The use of both methods A and B allowed for an extensive electrophile scope that delivered a large variety of sulfoximine-containing heterocycles.

Table 3.

Electrophile scope containing common heterocyclic pharmacophores and electrophilic arenes

All reactions were performed on 0.25 mmol scale, unless otherwise stated, in THF or dioxane (0.1M) under an argon atmosphere.

^aMethod A was used.

^bMethod B was used.

^cDecomposition observed after 14 days at room temperature – increased stability at −20 or −80 °C under argon (see Supplementary Handling of reagents section).

^dReactions were performed on gram scales (7 examples).

^ePerformed on >2 mmol scale.

^fPerformed on >1 mmol scale. Pharmaceutical scaffolds and intermediates are outlined in grey boxes. PG, protecting group; LG, leaving group; Het, heterocycle; Me, methyl; Ph, phenyl; t-Bu, tert-butyl; Bz, benzoyl; Ts, para-toluenesulfonyl (tosyl); Piv, pivaloyl; Bn, benzyl; NaHMDS, sodium bis(trimethylsilyl)amide; THF, tetrahydrofuran.

Electron-deficient ring systems known to readily undergo S_NAr chemistry were first examined. Substituted 1,3,5-triazine (76) served as an excellent substrate along with pyrimidines that were substituted with electron donating groups (80, 81). When 2,4,6-trichloropyrimidine was used as an electrophile, a mixture of regioisomers (1:2, 77:78) was observed. Regioselective nucleophilic substitution on pyrimidine ring systems was achieved by a leaving group (LG) switch from chloro to SO₂Me to provide C-2 selective S_NAr products 78 and 79 in high yields (93% and 89%). Trifunctional 4,6-dichloro-2-iodopyrimidine provided C-4 selective displacement of a chloro over iodo leaving group to give 82 and 83 as the major products on gram-scale. The utility of both S_NAr products, 82 and 83, will be further demonstrated in the forthcoming synthetic applications. Other 2-substituted pyrimidines decorated with naphthyl (84) and azaindole (85) substituents resulted in the desired S_NAr products in high yields (79–89%). The diazine scope is not limited to pyrimidines; 2-chloropyrazine also served as a suitable electrophile in good yield (86, 71%). Commercial and readily available pyridines were thoroughly explored. Initial attempts at affecting the sulfonimidoyl S_NAr with 2-fluoropyridine under both cryogenic and thermal conditions using N-Bz protected sulfoximines proved unfruitful, due to Bz transfer to the sulfoximine starting material. To our delight, switching to more robust PGs (N-Ts, 87; N-Boc, 88) attenuates their reactivity allowing the S_NAr to prevail.

An electron-deficient pyridine bearing a t-Bu ester (CO₂t-Bu) at the 3-position provided 89 in 72% yield. Trifluoromethyl-substituted pyridines proved troublesome under the optimized reaction conditions, reflective by a 22% isolated yield of 90. However, other halogenated pyridines underwent the S_NAr smoothly to afford an array of highly useful pyridine products (91-95). Preferential displacement of fluoro over chloro was demonstrated with 2-chloro-4-fluoropyridine granting site-selective 4-substitution product 92. When 2,4,6-trichloropyridine was used, C-2 selectivity was observed in a modest, but still serviceable, 6.6:1 ratio of 93 (isolated r.r.). Site selectivity can be reversed by the replacement of chloro with –SO₂Me at the 4-position, where 94 was obtained in high yield (79%) as the sole regioisomer. It should be noted that when SO₂Me is used as a LG in the less reactive pyridine series, dimerization of the electrophile via S_NAr with the sulfone is observed as a side-product (not observed with pyrimidines). Conversely, 2,4,6-trifluoropyridine was less selective for the 2-position (1.9:1 r.r.) to give 95 and 96 in 50% and 27% yields respectively. In the case where iodide could act as a LG, demonstrated by 2,6-dichloro-4-iodopyridine, a 19:1 regioselectivity was observed favoring substitution at the 2-position to provide 97. In addition, 2-chloro-3-iodopyridine was subjected to method B to give 3-iodo pyridyl sulfoximine 98 (44% yield) capable of further functionalization. The polyhalopyridine substrates examined provide unique opportunities for downstream modifications, via further S_NAr and/or cross-coupling chemistry, which may serve as important intermediates for future discovery efforts.

Other medicinally relevant heterocyclic scaffolds, including 4,7-dichloroquinoline of the malaria drug chloroquine, were found to be suitable electrophiles for sulfonimidoyl S_NAr. The first reported sulfoximine chloroquine analog 99 was made accessible in good yield (73%) and will be further investigated biologically. 6,8-Dibromoimidazo[1,2-a]pyrazine served as an excellent electrophile to afford 100 in high yield (84%). Appropriately halogenated and protected pyrrolo[2,3-d]pyrimidines, and purines were all compatible electrophiles under thermal sulfonimidoyl S_NAr conditions to provide 101-103 – enabling a route to novel sulfoximine nucleotide analogs. Sulfur containing heterocycles such as 2,4-dichlorothieno[2,3-d]pyrimidine, commonly found in PI3K inhibitors and non-nucleoside reverse transcriptase inhibitors,^58,59 along with 2-chlorobenzothiazole proved to be suitable S_NAr substrates that gave 104 and 105 in good yields respectively (71–84%).

Although this work focuses on providing a straightforward and robust method for the installation of methylene-linked sulfonimidoyl functional groups to pharmaceutically relevant heterocycles, a brief exploration of arene compatibility was warranted to understand the full scope of sulfonimidoyl S_NAr chemistry. As expected, hexafluorobenzene was reactive under cryogenic S_NAr conditions (110, 89% yield) while 1,3,5-trifluorobenzene was not – more forceful thermal conditions were required (111, 77% yield). Interestingly, and in contrast to the pyridine example 90, 2- and 4- fluorotrifluorobenzene underwent an S_NAr with method B to provide 112 and 113. To our surprise, and with modifications to the general procedure, fluoro-nitrobenzenes were able to serve as electrophiles to give sulfoximines 114 and 115 (for full reaction details and modifications see Supplementary Electrophile scope section). Based on a short arene screen, the scope of this transition metal-free method for the installation of S(VI) functionality is not limited to activated heterocycles but is also suitable for numerous electron deficient arene substrates.

Synthetic applications.

Current synthetic strategies to access α-substituted sulfoximines, and other neglected S(VI) groups, are typically arduous. Nucleophilic substitutions at activated benzylic positions by alkyl thiolates followed by oxidation to the desired sulfoximine are the most common routes, as demonstrated in the synthesis of BAY 1251152 (5, Figure 3A, right).¹⁸ The recent disclosure of S-alkylations of sulfinamides with benzylic halides developed by Maruoka can circumvent the oxidation steps (1–2 steps), while providing enantiopure sulfoximine products.⁴¹ In order to apply an S-alkylation strategy to access sulfoximines, the requisite benzylic halides (Br or I) are required via 2–3 step functionalizations of carboxylic acids or esters and is mainly limited to primary halides. An alternative S_NAr strategy provides increased modularity and a large selection of electrophilic partners while maintaining enantiospecificity.

Figure 3. Enantiospecific synthesis of BAY 1251152.

(A) Synthetic route analysis of BAY 1251152 using Bayer’s route (a) and a sulfonimidoyl S_NAr route (b). (B) Synthesis of BAY 1251152 (5) via stereospecific sulfonimidoyl S_NAr between commercially available pyridine 119 and readily available enantiopure sulfoximine (R)-23. Me, methyl; t-Bu, tert-butyl; Piv, pivaloyl; MeI, methyl iodide; TFA, trifluoroacetic acid; dba, dibenzylideneacetone; Xantphos, 4,5-Bis(diphenylphosphino)-9,9-dimethylxanthene; DCM, dichloromethane; MeOH, methanol.

Bayer relied on the traditional approach, a thiolation/oxidation sequence (4 steps), to access the sulfoximine used in the synthesis of BAY 1251152 (Figure 3A, disconnection a). Their route resulted in a racemic mixture that was separated by preparative chiral HPLC in order to access the desired target. A more modular approach utilizing a stereospecific sulfoximine installation would expedite the target synthesis and aid in future analog discovery. Sulfonimidoyl S_NAr was employed in the development of a concise synthesis of BAY 1251152 that is amenable for target and medicinal chemistry applications (Figure 3A, disconnection b).

Beginning with commercially available 2-chloro-4-fluoropyridine (119) and our chiral bifunctional sulfoximine (R)-23, S_NAr method B gave enantiomerically pure pyridyl sulfoximine 106 in high yield (87%, Figure 3B). The two-step procedure for S-alkylation developed by Maruoka was employed to install the desired methyl (Me) sulfoximine. A previously reported method for the Buchwald–Hartwig coupling of 2-aminopyridine 120¹⁸ was adopted allowing coupling to 2-chloropyridyl sulfoximine 121. Deprotection of the N-Piv with NaOH furnished BAY 1251152 in 78% yield over two steps from 121. With our disconnection, the desired target was made accessible in 5 steps and 50% overall yield of the desired enantiomer from known starting materials – a more than five-fold increase in overall yield of BAY 1251152.

Other drug discovery programs have a vested interest in evaluating the sulfoximine moiety as shown by the antibacterial candidate 4 developed by Zoetis (Figure 4).¹⁷ The pyridyl sulfoximine moiety 122 used in the synthesis of 4 was accessed in the traditional manner starting from benzylic alcohol 123. Bayer’s macrocyclic CDK9 inhibitor 2, structurally related to BAY 1251152 (5), shares a similar protected pyridyl sulfoximine building block 125.¹⁴ Both sulfoximine building blocks (122 and 125) were prepared as racemic mixtures in a 5-step sequence from either 123 or 126.^14,17 By utilizing the S_NAr approach, both pyridyl sulfoximines can be readily made as single enantiomers.

Figure 4. Formal syntheses of pharmaceutical intermediates.

(A) Synthetic route analysis of antibacterial 4 and CDK9 inhibitor 2 with the current approach and a sulfonimidoyl S_NAr approach. (B) Enantiospecific syntheses of the a-pyridyl sulfoximine units found in 4 and 2 via sulfonimidoyl S_NAr with a decrease in total step count and increase in overall yield. Me, methyl; t-Bu, tert-butyl; Piv, pivaloyl; MeI, methyl iodide; NaHMDS, sodium bis(trimethylsilyl)amide; TFA, trifluoroacetic acid; THF, tetrahydrofuran; DCM, dichloromethane; MeOH, methanol.

For the synthesis of Zoetis’ pyridyl sulfoximine, commercially available 5-bromo-2-fluoropyridine (124) and chiral sulfoximine (R)-23 underwent a smooth S_NAr using method A (107, 86% yield). The two-step S-functionalization sequence provided methyl pyridyl sulfoximine 128 in 70% yield. Deprotection of the pivaloyl group under basic hydrolysis conditions gave the desired free sulfoximine 122 in 53% overall yield as a single enantiomer. A reported Suzuki coupling of the requisite boronic ester with rac-122 provides access to 4.¹⁷ The improved route increased the overall yield of Zoetis’ chiral sulfoximine intermediate from 5% to 53% overall yield and decreased the step count while providing a diversifiable intermediate for analog development (N-Piv sulfinamide after t-Bu cleavage). Bayer’s macrocyclic CDK9 inhibitor 2 was made accessible from (rac)-2,6-dichloropyridyl sulfoximine 125 where the protecting group was Cbz or Boc – introduced from the imination step.¹⁴ A pivaloyl protected 2,6-dichloropyridyl sulfoximine should serve the same purpose and is expected to be compatible with the subsequent chemistry for the synthesis of macrocycle 2. To prepare the appropriately protected sulfoximine, pyridyl sulfone 127 and chiral sulfoximine (R)-23 were subjected to S_NAr method A giving enantiopure t-Bu sulfoximine 108. The two step S-alkylation sequence resulted in the desired methyl pyridyl sulfoximine intermediate 129 in 50% overall yield in 3 steps from readily available starting materials. The overall yield was improved from 11% to 50% while decreasing the step count by two. This expedient 3-step synthesis of the chiral sulfoximine building block for Bayer’s macrocycle CDK9 inhibitor aids in targeted synthetic efforts while providing another platform for analog development

One of the most noteworthy developments pertaining to the use of sulfonimidoyl functional groups in medicinal chemistry is the development of the AstraZeneca’s ATR inhibitor ceralasertib (1). In order to provide sufficient quantities of 1 for evaluation in clinical trials, the process chemistry team at AstraZeneca (AZ) had to revamp the medicinal chemistry route that was disclosed in 2018 (Figure 5A, top left).¹³ Owing to the traditional method of α-substituted sulfoximine installation, the medicinal chemistry group produced 1 (and analogs) as mixtures of diastereomers that were separated by column chromatography or iterative recrystallizations. In order to overcome the cumbersome purifications and to increase overall yield of the desired diastereomer, AZ’s process team subsequently (Figure 5A, top right) utilized an enantioselective enzymatic oxidation to introduce the chirality at sulfur (as a sulfoxide) that was later iminated to give the desired sulfoximine functionality.⁶⁰ One of the major drawbacks with their 13-step process synthesis was the installation of the cyclopropyl ring system found in 1.

Figure 5. Target- and diversity-oriented syntheses of ceralasertib.

(A) Various routes towards ceralasertib (1) by AstraZeneca along with two enantiospecific sulfonimidoyl S_NAr routes that focus on scalability and modularity. (B) Target-oriented gram-scale enantiospecific synthesis of 1 achieved in eight steps from 141. (C) Medicinal chemistry route to 1 via sulfonimidoyl S_NAr outlining four points of diversity for analog development. Please see “Enantiospecific synthesis of ceralasertib” in the Supplementary Information for experimental details and characterization data for compounds 139-148. Me, methyl; t-Bu, tert-butyl; BPin, boronic acid pinacol ester; Bn, benzyl; Ts, para-toluenesulfonyl (tosyl); Piv, pivaloyl; NaHMDS, sodium bis(trimethylsilyl)amide; Et₃N, triethylamine; TFA, trifluoroacetic acid; MeI, methyl iodide; PPh₃, triphenylphosphine; THF, tetrahydrofuran; DCM, dichloromethane; DMF, N,N-dimethylformamide; MeOH, methanol; DMSO, dimethylsulfoxide; EtOH, ethanol; rt, room temperature.

As the AZ process team described: “The manufacture of AZD6738 remains a challenge for the future of this medicine, due to the difficult nature of installing the dense functionality around the pyrimidine core. Longer term it would be beneficial to have a more convergent approach with a higher yielding route”.⁶⁰

During the development of our sulfonimidoyl S_NAr method, AZ’s process team disclosed an improved route (Figure 5A, middle left)⁶¹ and an attempt at a photocatalyzed flow approach via a Minisci reaction (Figure 5A, middle right)⁶². The newly developed process route to 1 (Figure 5A, middle left) addressed the issue of installing a cyclopropyl methyl sulfide moiety from 134 and 135 in a single step, while still relying on an enantioselective enzymatic oxidation/imination sequence to access the sulfoximine. To address the remaining drawbacks of the AZ’s process route, we developed an enantiospecific convergent approach utilizing our S_NAr method to generate the core structure of ceralasertib (1) from enantiopure t-butyl cyclopropyl sulfoximine 138 and pyrimidine 139 (Figure 5A, bottom left). Concurrently, an alternative medicinal chemistry-oriented synthesis of 1 was developed (Figure 5A, bottom right) to provide multiple points of diversity that would aid in analog development for related scaffolds of 1.

The target-oriented, gram-scale synthesis of 1 (Figure 5B) began with a Suzuki coupling between pyrimidine 140 and boronic ester 142 to deliver the azaindole pyrimidine core 139. Utilizing method A, sulfoximine 138 and pyrimidine 139 gratifyingly provided the congested pyrimidine core of 1 in a single step on multi-gram scale (109, 79% yield, >3 g prepared in a single flask). A one-pot t-butyl cleavage/S_NAr sequence with morpholine 131 gave rise to protected sulfinamide 143 in 88% yield. The desired methyl sulfoximine arose from an S-alkylation with methyl iodide to afford 144. Lastly, bis-deprotection of N-Piv and N-Bn was realized after extensive reaction screening – various reductive (Pd using H₂ sources with pressures up to 100 psi as well as Na° reductions) and acid mediated hydrolytic (AlCl₃, FeCl₃, FeBr₃) conditions were all explored (see Supplementary Enantiospecific synthesis of ceralasertib section). A two-step optimized deprotection via acid hydrolysis of N-Piv followed by oxidative N-Bn cleavage resulted in 1 (>1 g) with an 87% yield over two steps and one final purification. Further development of protecting group strategy may be required for a more ideal manufacturing route of ceralasertib.

Conversely, the diversity-oriented synthetic route highlights four distinct sites and advanced intermediates that can be exploited for analog development (Figure 5C). Enantiopure t-Bu pyrimidyl sulfoximine 82 was made accessible on gram-scale from trifunctional pyrimidine 140 in 86% yield using method A. At this stage, an alkylation at the benzylic position, a second S_NAr on the pyrimidine or a cross-coupling are all options. Functionalization of the benzylic position was chosen as the second step (see Scheme S7 in SI for alternative routes) via alkylation with 1,2-dibromoethane (145) to give 83 in good yield (72%). A second S_NAr was achieved using morpholine 131 with good regiocontrol (ca. 9:1) and high yield (91%) at low reaction temperatures – providing advanced intermediate 146 and leaving two diversification sites remaining.

The two step S-alkylation sequence using 2-iodopyrimidine 146 could be employed at this point, leaving the last diversification site at the 2-position of the pyrimidine capable of eastern region analog development (see SI). However, we decided to push the boundaries of the t-butyl sulfoximine to determine if it were stable under typical cross-coupling conditions. Gratifyingly, 146 underwent a Suzuki coupling with boronic ester 141 to give fully protected t-Bu sulfoximine 147 in 83% yield. An S-alkylation sequence provided methyl sulfoximine 148 that was subsequently deprotected to give ceralasertib 1, fully demonstrating the feasibility of the diversity-oriented route. The disclosed improvements made to incorporate congested sulfoximine moieties provides an alternative approach to 1 and related scaffolds. By utilizing the S_NAr approach with trifunctional pyrimidine 140, the points of diversity were increased by two, the overall yield increased by 20%, and the step count decreased by two.

Conclusion

A modular approach for the installation of sulfonimidoyl motifs via S_NAr chemistry has been extensively examined and its utility demonstrated, culminating with 85 examples and 4 synthetic applications. The diversity displayed with both nucleophile and electrophile scopes exemplifies the far-reaching implications of the developed method. We fully anticipate that the method described herein will be adopted by discovery chemists interested in the unique physiochemical properties of sulfonimidoyl functional groups, with an emphasis on pharmaceutically relevant heterocycles. By removing synthetic barriers for the introduction and late-stage modifications, neglected S(VI) functionality can be incorporated into more discovery campaigns and hopefully become as commonly used as their related sulfone and sulfonamide counterparts. While developing the sulfonimidoyl S_NAr method, we noted the remaining limitations in accessing enantiopure sulfoximines and sulfonimidamides. Currently, investigations into further solutions for the asymmetric introduction of sulfoximines and sulfonimidamides is ongoing in our laboratory and will be reported on in due course.

Methods

General procedure for sulfonimidoyl S_NAr using method A.

In a septum capped 2-dram reaction vial or culture vial (for 0.25 mmol scale) equipped with a stir bar and argon balloon was charged with an S(VI) or S(IV) nucleophile (1 eq.), heteroaryl electrophile (1 eq.) and THF (0.1 M) then cooled to −78 °C. NaHMDS (2M in THF, 2.2 eq. for 1° and 2° nucleophile or 1.1–1.5 eq. for 3° nucleophile) was added then stirred at −78 °C for 1 hour before gradually warming to room temperature (not all substrates require warming to room temperature, see Supplementary Information for specific examples). Once complete (monitored by TLC and LC–MS), reactions were quenched with saturated aqueous NH₄Cl and water, extracted with EtOAc. The combined organic layers were dried over Na₂SO₄, filtered and concentrated. Further purification by silica gel column chromatography provided the desired α-(hetero)arylated S(VI) or S(IV) products.

General procedure for sulfonimidoyl S_NAr using method B.

In a septum capped 2-dram reaction vial or culture vial (for 0.25 mmol scale) equipped with a stir bar and argon balloon was charged with an S(VI) nucleophile (1 eq.), heteroaryl electrophile (1 eq.), dioxane (0.1 M) and 15-crown-5 ether (2.2 eq.). NaH (2.2 eq.) was added then stirred at room temperature for 3 minutes before heating at 50–80 °C for approximately 15 hours (see Supplementary Information for specific examples). Once complete (monitored by TLC and LC–MS), reactions were quenched with saturated aqueous NH₄Cl and water, extracted with EtOAc. The combined organic layers were dried over Na₂SO₄, filtered and concentrated. Further purification by silica gel column chromatography provided the desired α - (hetero)arylated S(VI) products.

Data availability

The experimental data as well as the characterization data for all the compounds prepared during these studies are provided in the Supplementary Information. The crystallographic data for the structures reported in this article have been deposited at the Cambridge Crystallographic Data Centre, under deposition numbers CCDC 2087123 (52), CCDC 2087121 (108), CCDC 2087124 (112), CCDC 2087122 (138). Copies of the data can be obtained free of charge via https://www.ccdc.cam.ac.uk/structures/.