Strain‐Release Pentafluorosulfanylation of Carbonyl‐Containing Disubstituted Bicyclobutanes: A Fortuitous Path to SF₅‐Containing Oxa[2.1.1]bicyclohexanes

ABSTRACT

The recent merger of pentafluorosulfanylation with strain‐release functionalization logic—largely enabled by an increase in SF₅Cl accessibility—has begun to reshape the ways we can imagine incorporating the SF₅ group into organic molecules. In this study, we first explore how both radical polarity matching and substituent effects, when employed synergistically, can be used to effect regio‐ and diastereoselective SF₅Cl addition across 1,3‐disubstituted (viz. 3‐aryl‐1‐carbonyl‐substituted) bicyclobutanes (BCBs) to access a class of congested α‐SF₅‐carbonyl‐containing tetrasubstituted cyclobutanes (CBs). Synthetic, computational, and structural studies provide insight on the observed selectivity and reveal unusually close intramolecular SF₅⋯C═X (X = C, N, or O) contacts in several SF₅‐CBs by SC‐XRD. Beyond accessing compounds of fundamental interest, we describe a complementary telescoped protocol to facilitate isolation of SF₅‐CBs as alcohols, as well as a fortuitous observation that led to the synthesis of the first suite of SF₅‐substituted oxa[2.1.1]bicyclohexanes (OBHs). Finally, we examine the compatibility of SF₅‐CBs and SF₅‐OBHs with various reaction conditions to expand the scope of accessible building blocks and report the synthesis of an SF₅‐OBH‐containing drug derivative. Results from comparative in vitro ADME profiling indicate the SF₅‐OBH motif may serve as an attractive “hybrid bioisostere” for the meta‐CF₃‐Ph motif.

A regio‐ and diastereoselective method for strain‐release pentafluorosulfanylation of carbonyl‐containing 1,3‐disubstituted [1.1.0]bicyclobutanes (BCBs) was developed that led to several fortuitous discoveries. For instance, the resultant SF₅‐cyclobutanes (SF₅‐CBs) exhibit close SF₅⋯C═X contacts (X = C, N, O) and provide synthetic access to a new class of m‐CF₃‐Ph “hybrid bioisosteres:” SF₅‐oxa[2.1.1]bicyclohexanes (SF₅‐OBHs).

1. Introduction

Strain‐release pentafluorosulfanylation has quickly graduated from a physical organic curiosity to a viable strategy for accessing new suites of “hybrid bioisosteres” [1, 2, 3, 4] in the kaleidoscopic landscape of modern medicinal chemistry [5, 6, 7, 8]. The sudden, recent advancements in this field have been fueled primarily by an increase in accessibility of pentafluorosulfanyl chloride (SF₅Cl) [9, 10, 11, 12, 13]—an archetypal SF₅ radical transfer reagent [14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29] that can now be synthesized safely and easily under mild oxidative fluorination conditions [30, 31, 32, 33, 34, 35]. Primarily using this reagent, our laboratory and others have established methods for pentafluorosulfanylation of [1.1.1]propellane [3, 36, 37, 38], [1.1.0]bicyclobutanes (BCBs) [39, 40, 41, 42], and aza[1.1.0]bicyclobutanes (ABBs) [10, 40, 43] over the past three years. These methods effectively situate the SF₅ group—an established bioisostere for a CF₃ or t‐Bu group [44, 45, 46, 47, 48]—on aliphatic, three‐dimensional ring structures [49, 50, 51, 52] that are in and of themselves known bioisosteres, for example, for para‐substituted benzene [53, 54] or piperidine rings [55]. Beyond applications in medicinal chemistry [14, 56] and in peripheral fields [57, 58, 59, 60], the resultant SF₅‐containing ring structures have often been the first of their kind, creating opportunities to obtain foundational understanding of novel molecular motifs and the associated reaction mechanisms that ushered them into existence.

Throughout our studies, we have noted a developing theme that unusual reactivity/selectivity patterns tend to emerge in strain‐release functionalization reactions involving highly electrophilic SF₅ radicals [38, 61]. One particularly striking observation in our previous work pertains to the pentafluorosulfanylation of 1‐carbonyl‐substituted BCBs [38]. Whereas radicals tend to add exclusively to the 3‐position of 1‐substituted BCBs, we found that pentafluorosulfanylation at the 1‐position was becoming nearly competitive in some instances; ultimately, we attributed this to an extreme radical polarity matching effect [62, 63]. Thereafter, we surmised that substituent effects could likely be used in tandem with radical polarity matching logic to direct SF₅ radicals exclusively to the α‐position relative to a carbonyl in 1,3‐disubstituted BCBs. In the past year, this idea has been implemented successfully toward guiding other electrophilic radicals, such as trifluoromethyl [64] or sulfonyl radicals [65], to the α‐position of 3‐aryl‐1‐carbonyl‐substituted BCBs to make tetrasubstituted cyclobutanes (CBs). Recently, Tlili and co‐workers also demonstrated that regioselective pentafluorosulfanylation of similar substrates is possible in the specific context of iminopentafluorosulfanylation, using an imine‐SF₅ transfer reagent developed in their group [40, 66].

In this article, we disclose a method for SF₅Cl addition across 1,3‐disubstituted (viz. 3‐aryl‐1‐carbonyl‐substituted) BCBs that is highly regioselective, can proceed with diastereoselectivities >5:1, and that inspired the development of complementary synthetic methods (Figure 1, top). Additionally, we found that reaction conditions translate well to tetrafluoro(trifluoromethyl)sulfanylation using CF₃SF₄Cl. Over the course of our systematic reactivity, selectivity, and mechanistic studies, we made additional discoveries that bolster the value of this transformation from both fundamental and practical perspectives. From a fundamental standpoint, the α‐SF₅ products reported herein boast the closest SF₅⋯C═X (X = C, N, O) contacts observed to date, providing an opportunity to characterize such an interaction through X‐ray crystallographic and computational analyses (Figure 1, bottom‐left). From a practical standpoint, we found that the SF₅Cl addition adducts can be converted to a variety of other SF₅‐containing building blocks, including heretofore unknown SF₅‐substituted oxa[2.1.1]bicyclohexanes (OBHs). As OBHs have garnered attention as bioisosteres for meta‐substituted benzene rings [67, 68, 69, 70], we synthesized the first SF₅‐OBH‐containing drug derivative (a PROT inhibitor) and subjected it to comparative in vitro ADME profiling to evaluate this motif as a meta‐CF₃‐Ph hybrid bioisostere [71, 72, 73, 74] (Figure 1, bottom‐right).

FIGURE 1.

(Top) Regio‐ and diastereoselective SF₅Cl addition across 1,3‐disubstituted BCBs and related methodological advancements in this work. (Bottom‐Left) Study of close SF₅⋯C═O noncovalent interactions enabled by these methods. (Bottom‐Right) Proposed significance of SF₅‐OBHs reported herein as potential “hybrid bioisosteres” for meta‐CF₃‐Ph motifs.

2. Results and Discussion

2.1. Initial Reactivity and Selectivity Studies

At the outset, our previously reported conditions for SF₅Cl addition across 1‐substituted BCBs did not directly translate to 1,3‐disubstituted BCB substrates; however, this provided a useful starting point for reaction optimization [39]. Ultimately, we found that stirring BCB 1 with SF₅Cl (1.2 equiv.; ∼0.1 M solution in hexanes) in Et₂O under irradiation with white LEDs for 1 h provided product 2 in 84% yield and 5.3:1 diastereomeric ratio (d.r.) by ¹⁹F NMR analysis (METHOD A; Figure 2, top). Although complete details for reaction optimization can be found in the Supporting Information, a few observations are worth mentioning: 1) the reaction proceeds well in ethereal solvents [75], toluene, and EtOAc, though best results were obtained using Et₂O, 2) the reaction is sensitive to concentration effects and benefits from dilution (i.e., concentration of 1 under optimized conditions is 0.03 M), and 3) the transformation can also be achieved in the absence of light using 10 mol % BEt₃/O₂ in a similar yield (consistent with the notion that strain‐release pentafluorosulfanylation reactions often proceed through radical chain propagation mechanisms [3, 9, 36, 38]), though results may vary based on reagent quality [76]. Our initial isolation attempt provided several milligrams of the major diastereomer of 2 as a clear oil. Analysis of the ¹³C{¹H} NMR spectrum revealed the carbonyl carbon atom as a pentet with a ³J_C–F  = 2.1 Hz coupling constant, indicating the α‐SF₅‐regioisomer was the major product formed.

FIGURE 2.

(Top) Optimized conditions and regioselectivity for SF₅Cl addition across 1 (METHOD A). (Bottom) SF₅Cl addition across 3 and X‐ray crystal structure of the major diastereomer of 4 (displacement ellipsoids depicted at 50% probability). Yields and diastereomeric ratios determined by ¹⁹F NMR analysis.

To confirm the identity of the major diastereomer, we performed the reaction on another substrate (3) that would ultimately produce a solid product amenable to crystallization (Figure 2, bottom). Accordingly, 3 was subjected to identical reaction conditions, forming 4 in 83% and 3.1:1 d.r. Isolation of the major diastereomer via column chromatography provided 4 as a white solid in sufficient purity, enabling us to grow crystals suitable for single‐crystal X‐ray diffraction (SC‐XRD). Thus, we confirmed that the SF₅ and Cl substituents are positioned anti to one another in the major diastereomer. Note that we assigned the identity of the minor isomer as the syn diastereomer based on observed ³J_C–F coupling by ¹³C{¹H} NMR analysis of the mixture, as opposed to a regioisomer (note that chemical modification of the ester later enabled complete isolation and characterization of the minor isomer to provide more conclusive evidence; see Supporting Information).

The discrepancy in observed d.r. between products 2 and 4 prompted a brief study of the effect of ester substituent size on selectivity. Under optimized conditions, the SF₅‐containing methyl (2), ethyl (5), isopropyl (6), benzyl (7), and phenyl (4) esters were all accessible from their corresponding BCBs in generally good yields ranging from 76%–84% by ¹⁹F NMR analysis (Figure 3). We observed an inverse correlation between substituent size and d.r. that is roughly linear, determined by plotting Taft steric parameters (E _s) against observed diastereoselectivities for Me, Et, i‐Pr, and t‐Bu ester substrates (see Supporting Information for details) [77, 78]. Note that effectively bulkier ester substituents, that is, tert‐butyl and 1‐adamantyl, led to further erosion of the d.r. (to <2:1) and of product yields (esters 8 and 9 were only formed in 30% and 37%, respectively). However, increasing ester substituent size did not appear to alter the regioselectivity. We also examined 1,3‐disubstituted BCBs with other carbonyl derivatives in the 1‐position, that is, to access phenyl ketone (10), methyl ketone (11), and Weinreb amide (12); however, the best results were obtained from SF₅Cl addition across 3‐aryl‐1‐ester‐substituted BCBs. The extra rotational degree of freedom in the ester substituents (e.g., compared to the ketones investigated) likely plays an important role in reducing steric hindrance during the approach of the SF₅ radical.

FIGURE 3.

Examining impact of carbonyl substituent identity on yield and selectivity. ^(a)Determined by ¹⁹F NMR analysis in acetone‐d₆ .

2.2. Computational Insight on Regio‐ and Diastereoselectivity

For additional insight on the observed regio‐ and diastereoselectivity of the transformation, we computed the free energy profile of the putative radical chain propagation steps associated with the conversion of 1 to 2 at the PWPB95‐D4/def2‐QZVPP//PCM(Et₂O)//ωB97X‐D/def2‐TZVPP level of theory (Figure 4, top) [79, 80, 81, 82, 83, 84, 85, 86]. First, we found that SF₅ radical addition to the 1‐position (α to the carbonyl) to access IM1₁ via TS1₁ is both thermodynamically and kinetically favored over addition to the 3‐position to access IM1₃ via TS1₃ (ΔΔG = 3.5 kcal mol⁻¹; ΔΔG ^‡ = 2.1 kcal mol⁻¹), consistent with the observed regioselectivity. The resultant carbon‐centered radical intermediate IM1₁ is then poised for chlorine atom abstraction from SF₅Cl from either the anti‐face (via TS2_1,anti ) or the syn‐face (via TS2_1,syn ) relative to the SF₅ group on the cyclobutane ring. While the calculations indicate there is virtually no thermodynamic preference for 2‐anti over 2‐syn , TS2_1,anti is predicted to be 1.6 kcal mol⁻¹ lower in energy than TS2_{1, syn} , which comports with our observed anti selectivity.

FIGURE 4.

(Top) Free energy profile for SF₅ radical addition to 1 computed at the PWPB95‐D4/def2‐QZVPP//PCM(Et₂O)//ωB97X‐D/def2‐TZVPP level of theory. (Middle) Free energy profile for SF₅ radical addition to 3 computed at the same level of theory. (Bottom‐Left) Condensed Fukui functions (f ⁰) comparing site‐susceptibilities to radical attack and Hirshfeld partial charges (q), both in units of elementary charge (e). (Bottom‐Right) Noncovalent Interaction (NCI) plots highlighting weakly attractive (green) and repulsive (red) interactions in TS1₁ and TS1₃ .

To cross‐check our analysis, we also computed the free energy profile for the conversion of 3 to 4 (Figure 4, middle). Despite the apparent increase in size of the ester substituent, SF₅ radical addition to the 1‐position via TS3₁ to access IM2₁ is still thermodynamically and kinetically favored over addition to the 3‐position via TS3₃ to access IM2₃ (ΔΔG = 3.1 kcal mol⁻¹; ΔΔG ^‡ = 2.4 kcal mol⁻¹). Complementary to our computed energetics, analysis of condensed radical Fukui functions (f ⁰) computed at the PCM(Et₂O)‐ωB97X‐D/def2‐SVP level of theory indicate that the 1‐position of both BCBs 1 and 3 would be more susceptible to radical attack than the 3‐position, and computed Hirshfeld partial charges (q) also support the notion that the 1‐position represents the polarity‐matched site for SF₅ radical addition (Figure 4, bottom‐left) [87, 88, 89]. Furthermore, the difference in activation energies (ΔΔG^‡) between TS4_1,anti to make 4‐anti and TS4_1,syn to make 4‐syn is only 1.1 kcal mol⁻¹; this is marginally smaller than the ΔΔG ^‡ in the diastereoselectivity‐determining step highlighted above in the computed energy profile for the conversion of 1 to 2. Accordingly, it is also consonant with the slightly diminished relative d.r. observed in the synthesis of 4 vs. 2.

The major factors contributing to the regioselectivity of SF₅ radical addition are likely a composite of a radical polarity matching effect, relative steric hindrance of the arene vs. ester substituent, and thermodynamic stability of the resultant tertiary benzylic radical intermediate (reflected in the free energy profiles in Figure 4). For another perspective, we also examined noncovalent interaction (NCI) plots of TS1₁ and TS1₃ (Figure 4, bottom‐right) and found evidence for a weakly attractive interaction between the fluorine lone‐pair and π* orbital on the ester substituent (note that the F_eq⋯C═O distance in TS1₁ is 2.98 Å) [90, 91, 92, 93, 94, 95, 96, 97]. Despite the verisimilitude of a “directing effect,” natural bond orbital (NBO) calculations reveal that the stabilization energy of this donor–acceptor orbital interaction (ΔE_ij) is effectively <1 kcal mol⁻¹ [98, 99, 100]. Furthermore, the NCI plot of TS1₃ suggests a similar weak interaction exists between the fluorine lone‐pair and π* orbital of the arene substituent. Thus, not only are these extremely subtle (if not negligible) interactions in the transition structures but they are present in both TS1₁ and TS1₃ and can likely be ruled out as a differentiating feature of the regioselectivity‐determining step.

2.3. Telescoped Reduction Protocol and Arene Substrate Scope

On a practical note, although ester‐containing BCBs were suitable substrates for regioselective SF₅Cl additions, the corresponding tetrasubstituted SF₅‐CB products were somewhat sensitive toward chromatographic purification. That is, we discovered that compounds 2 and 4–7 were partially degrading on silica gel among other stationary phases (e.g., alumina and Florisil) during isolation attempts; this was confirmed through 2D TLC analyses on the isolated products. Prior to examining arene substituent effects on reactivity/selectivity, these observations prompted us to better understand the decomposition pathway and entertain ways to convert these compounds into more chemically stable and useful SF₅‐CB building blocks.

Despite our best efforts, isolation and characterization of discrete decomposition products proved extremely challenging; however, we noted that several new signals materialized in the alkene region of the ¹H NMR spectrum upon extended exposure of ester‐containing SF₅‐CBs to silica, concomitant with a decrease in intensity of the SF₅ signals in the ¹⁹F NMR spectrum. This suggests that ester‐containing SF₅‐CBs are likely prone to SF₅ group elimination. Although it is exceedingly rare, the SF₅ group may act as a leaving group on an sp³ carbon atom under certain circumstances. For instance, we observed this phenomenon previously on a trisubstituted SF₅‐CB during an attempted lithiation reaction [38]. As for the ester‐containing products in Figure 3, we surmised that elimination of the SF₅ group may be facilitated by 1) the increased acidity of the cyclobutane protons adjacent to three strong electron‐withdrawing groups and 2) formation of a conjugated π‐system. In this light, we envisioned that one solution to mitigate this decomposition pathway is to reduce the ester to the corresponding alcohol prior to isolation.

An isodesmic analysis at the PWPB95‐D4/def2‐QZVPP//PCM(Et₂O)//ωB97X‐D/def2‐TZVPP level of theory lends credence to this idea, indicating that SF₅ elimination from ester 2 (i.e., to make the corresponding conjugated cyclobutene) is thermodynamically favored over SF₅ elimination from the corresponding alcohol 13 by 2.9 kcal mol⁻¹ (Figure 5, top). Accordingly, we developed a telescoped SF₅Cl addition–LiAlH₄ reduction protocol that converted BCB 1 into alcohol 13 without any intermediate purification [101]. To our satisfaction, the alcohol was isolated in 68% over 2 steps (5.0:1 anti:syn) via straightforward column chromatography on silica gel; we found no evidence that SF₅ elimination had occurred (Figure 5, bottom).

FIGURE 5.

(Top) Isodesmic equation computed at the PWPB95‐D4/def2‐QZVPP//PCM(Et₂O)//ωB97X‐D/def2‐TZVPP level of theory reflecting the thermodynamic preference for SF₅ elimination to occur from an ester‐containing CB over an alcohol‐containing CB. (Bottom) Telescoped SF₅Cl addition–LiAlH₄ reduction sequence developed to provide access to more readily isolable alcohol‐containing tetrasubstituted SF₅‐CBs. Isolated yield reported. Diastereomeric ratio determined by ¹⁹F and ¹H NMR analyses of the crude reaction mixture.

For ease of isolation, we implemented this telescoped SF₅Cl addition–LiAlH₄ reduction protocol throughout our remaining exploration of the scope of arene substituent effects (METHOD B; Figure 6, top). Under these conditions, the reaction tolerates H, F, Cl, Br, OCF₃, and CF₃ substituents in the para position on the arene, providing compounds 13–18 in isolated yields ranging from 53–68% over 2 steps. When performed on a gram scale, the isolated yield for compound 16 was comparable (i.e., 51% over 2 steps). In addition, we found that m‐Me, o‐F, m‐OMe, and di‐m‐OMe substituents are tolerated, providing compounds 19–22 in isolated yields ranging from 40–52% over 2 steps. Note that the methyl ester was used as the alcohol precursor in each case in attempt to boost the diastereoselectivity based on our results in Figure 3; in practice, we found that the d.r. was consistently >4.5:1 (up to 9.3:1) for these transformations. This also often enabled chromatographic separation and characterization of the major anti diastereomer. Beyond aryl‐substituted BCBs, we found that the reaction does not proceed when the arene at the 3‐position is replaced with a methyl or silyl group (as well as when the carbonyl at the 1‐position is replaced with a sulfone; see SI for details). However, we were able to obtain the alkyne‐substituted SF₅‐CB 23, albeit in a lower yield. This suggests the arene plays an important role, e.g., in resonance stabilization of the putative benzylic radical intermediate. Lastly, we found that SF₅Cl addition logic translates to CF₃SF₄Cl addition [102], enabling the formation of CF₃SF₄‐containing alcohols 24–27 in 47%–70% yields over 2 steps and in similar d.r. (i.e., ranging from 4.0:1 to 6.2:1).

FIGURE 6.

(Top) Arene substrate scope employing the telescoped SF₅Cl addition–LiAlH₄ reduction and extension to CF₃SF₄Cl addition (METHOD B). ^aIsolated yield. ^bDiastereomeric ratios determined by ¹⁹F and ¹H NMR analyses of the crude reaction mixture. ^cDetermined by ¹⁹F NMR. (Bottom) Syntheses of various tetrasubstituted α‐SF₅ CB derivatives. Isolated yields reported. X‐ray crystal structures of 18, 31, 32, and 35 determined by SC‐XRD (displacement ellipsoids depicted at 50% probability). The minor occupancy positions of the CF₃ group in 31, 32, and 35 are omitted for clarity. ^dDetermined by ¹⁹F NMR.

With these products in hand, we investigated whether the α‐SF₅ substituent on a 1,1,3,3‐tetrasubstituted cyclobutane would be compatible with several alcohol/carbonyl group manipulations (Figure 6, bottom). First, we isolated the major (anti) diastereomer of alcohol 18 and subjected it to Appel conditions to access alkyl bromide 28 in 70% [103]. Next, we established that 18‐anti can be converted to benzyl ester 29 in virtually quantitative yield under standard EDC coupling conditions [104]. Compound 29 was then subjected to radical dechlorination conditions using supersilane and substoichiometric AIBN to provide the corresponding 1,1,3‐trisubstituted SF₅‐CB 30 in 54% and 2.9:1 d.r. (major isomer assigned as anti from NOESY and HOESY experiments) [105]. Furthermore, 18‐anti was oxidized using Dess‐Martin periodinane (DMP) to provide aldehyde 31 in 70% [106]. Note that while 18‐anti proved compatible with oxidation, esterification, and radical dechlorination conditions, attempts at nucleophilic displacement from either 28 or the corresponding tosylate of 18 using azide and thiol‐based reagents were unsuccessful (see SI), likely due to pronounced β‐branching at the reaction site.

Starting from aldehyde 31, we were also able to demonstrate its conversion to enone 32 (85%) under Wittig‐type olefination conditions using a stabilized phosphorus ylide [107], to alkyne 33 (41%, determined by ¹⁹F NMR due to product volatility) under Seyferth–Gilbert homologation conditions [108], and to imines 34 (66%) and 35 (65%) via condensation with benzylamine and 2‐bromo‐benzylamine [109]. Thereafter, imine 34 was also subjected to standard Staudinger conditions to provide β‐lactam 36 in 31% [110]. The results from additional unsuccessful transformations (including attempted Wittig reactions employing alkali metal bases) are detailed in the Supporting Information. Nonetheless, this series of transformations demonstrated tolerance of the α‐SF₅ substituent to several reaction conditions, thus generating access to a broader scope of SF₅‐CBs.

2.4. Observation of Close SF₅⋯C═O and CF₃SF₄⋯C═O Interactions

Upon characterization of 4, 31, 32, & 35 by SC‐XRD, we found that the SF₅ group is situated on the conformationally rigid CB ring such that one of the equatorial fluorine atoms (F_eq) is forced into close proximity of the geminal substituent through a Thorpe‐Ingold‐like effect [111, 112, 113], lending way to notable intramolecular F_eq⋯C═X (X = C, N, or O) interactions. In fact, the measured intramolecular F_eq⋯C═O distance in the X‐ray crystal structure of aldehyde 31 is 2.460(3) Å, which is substantially less than the sum of the van der Waals radii of carbon and fluorine (∑r _vdW ≈ 3.17 Å) [114] and represents the shortest SF₅⋯C═O contact reported to date (Figure 7, left). A search for other short intramolecular SF₅⋯C═O contacts in the CSD reveals only 31 reported examples, with the next shortest SF₅⋯C═O contact being 2.478(3) Å [22]. It is also exceptionally short for a fluorine atom–carbonyl interaction in a broader sense. For perspective, the shortest F⋯C═O contacts in the CSD [115] were found to be 2.35(3) Å (associated with the interaction of a fluorine atom in a CF₃ group and the carbonyl carbon of an ester [116]) and 2.448(15) Å (associated with the interaction of a fluorine atom in a C(sp²)─F bond with the carbonyl carbon of an amide [117]).

FIGURE 7.

X‐ray crystal structures of 31 and 37 determined by SC‐XRD (displacement ellipsoids depicted at 50% and 30% probabilities, respectively; minor occupancies of the CF₃ groups are omitted for clarity), comparisons of distances/angles measured in the solid state to those in computed structures optimized at the ωB97X‐D/def2‐TZVPP level of theory, and isosurfaces (isovalue = 0.07) of Natural Bonding Orbitals contributing to donor‐acceptor interactions involving F_eq and C = X carbon atoms. NBOs were generated at ωB97X‐D/def2‐TZVPP level of theory.

Previously, we have drawn structural comparisons between SF₅ and SF₄CF₃ groups on strained rings [9, 36, 38, 39, 118], and thus posited that CF₃SF₄‐CBs would exhibit a similar type of CF₃SF₄⋯C═O interaction. Thus, we synthesized the aldehyde‐containing CF₃SF₄‐CB 37 from its corresponding alcohol 26 under DMP oxidation conditions (50% yield) and grew crystals suitable for SC‐XRD (Figure 7, right). The S─F_eq bond of interest is ∼2% longer in 37 than in 31 (1.608(2) Å vs. 1.570(2) Å), and the observed CF₃SF₄⋯C═O distance (2.457(4) Å) is on par with the distance observed in 31. Furthermore, 37 boasts the shortest CF₃SF₄⋯C═O contact involving the equatorial fluorine atoms observed thus far.  A CSD search revealed only 5 other examples, with the next shortest distance measured at 2.527(6) Å. In 31 and 37, computed donor–acceptor orbital interaction stabilization energies (ΔE _ij) and Wiberg bond indices (WBI) indicate that the n_F → π*_C═O interactions are fairly weak and, therefore, attraction is likely primarily electrostatic [119]. However, it is noteworthy that the SF₅ and CF₃SF₄ groups still exhibit attractive interactions when constrained into proximity of π bonds like the carbonyl, even if only circumstantial. One can envision this type of interaction could be useful in optimizing the trajectory of the SF₅ group when bound in an enzyme pocket [96].

Lastly, beyond SF₅⋯C═O and CF₃SF₄⋯C═O interactions, the measured intramolecular F_eq⋯C═C distance in 32 (2.526(2) Å) and the measured intramolecular F_eq⋯C═N distance in 35 (2.507(2) Å) represent the shortest SF₅⋯C═C and SF₅⋯C═C contacts reported to date, respectively (see Supporting Information for details and expanded discussion).

2.5. Cyclization Substrate Scope

During initial purification of alcohol 13, we serendipitously isolated a minor SF₅‐containing side‐product in ≤5% and were pleased to identify it as SF₅‐containing oxa[2.1.1]bicyclohexane (OBH) 38 (Figure 8). Ostensibly, this formed via sequential chloride ionization and intramolecular S_N1, reminiscent of the iodonium‐induced cyclization strategy to make OBH rings implemented by Mykhailiuk and co‐workers [65, 68]. In the same study, the authors also demonstrated the value of the OBH ring as a bioisostere for a meta‐substituted benzene ring. This prompted us to develop a method to intentionally construct SF₅‐containing OBH rings from their corresponding alcohols. After a brief round of reaction optimization (see Supporting Information), we found that chloride abstraction using a silver(I) cation, that is, AgSbF₆ (METHOD C) [120], effectively promotes formation of the OBH ring in the presence of the tertiary SF₅ group.

FIGURE 8.

(Top) Substrate scope for intramolecular cyclization of the corresponding alcohols to access SF₅‐OBHs and CF₃SF₄‐OBHs (METHOD C). Isolated yields reported. X‐ray crystal structure of 38 determined by SC‐XRD (displacement ellipsoids depicted at 50% probability). (Bottom) Syntheses of various SF₅‐OBH derivatives demonstrating compatibility with remote arene functionalization reactions. Isolated yields reported.

We successfully converted all SF₅‐CBs in the top panel of Figure 6 to their corresponding OBH products 38–52 using METHOD C (Figure 8, top). We obtained SF₅‐OBH products 38–48 in isolated yields ranging from 39%–79%, demonstrating similar functional group tolerance to METHOD B. We also found that the cyclization can be performed on a gram scale without detriment, providing 41 in 76% yield. In addition, all CF₃SF₄‐containing alcohols were successfully converted to their corresponding CF₃SF₄‐OBH products (49–52) in yields ranging from 64%–82%. Note that the structures of both SF₅‐OBH 38 and CF₃SF₄‐OBH 49 were confirmed definitively by SC‐XRD (see SI for structural comparisons) [121].

Subsequently, we investigated the chemical stability of the SF₅‐OBH motif under various reaction conditions (Figure 8, bottom). Gratifyingly, we found that SF₅‐OBH 41 tolerates lithium–halogen exchange and electrophile trapping with CO₂ [122] to access carboxylic acid 53 in 52% and with i‐PrO‐Bpin [123] to access 54 in 88%. Sonogashira cross‐coupling [124] and Pd‐catalyzed azidation using Liu's conditions [125] gave 55 in 82% and 56 in 80%, respectively. We then carried azide 56 through a telescoped metal boride reduction–Boc protection sequence [126] to provide Boc‐protected aniline 57 in 72% over 2 steps, as well as through a metal boride reduction–amide bond formation sequence [127] followed by electrophilic aromatic bromination [128] to provide 58 in 49% over 3 steps. Furthermore, we demonstrated compatibility of the SF₅‐OBH motif with two nitrene‐based ring expansions, providing azepines 59 and 60 in 70% and 52% [129, 130]. Overall, this exercise not only highlights that SF₅‐OBHs may be incorporated into multi‐step syntheses but provides a synthetic roadmap to expand the scope of our transformation to other valuable SF₅‐OBH building blocks, including heterocycles.

Lastly, from a pragmatic standpoint, we obtained proof of concept that it is possible to telescope BCB 1 through the entire three‐step SF₅Cl addition–LiAlH₄ reduction–intramolecular cyclization sequence to access OBH 38 in 60% isolated yield over 3 steps, without any intermediate chromatography (Figure 9).

FIGURE 9.

Telescoped three‐step SF₅Cl addition–LiAlH₄ reduction–intramolecular cyclization sequence to access OBH 38 from BCB 1 (0.4 mmol) without intermediate chromatographic purification. Isolated yield reported.

2.6. SF₅‐OBH as a m‐CF₃‐Ph Hybrid Bioisostere

Previous work on SF₅‐containing hybrid bioisosteres has focused primarily on replacements for para‐CF₃‐substituted benzene rings (i.e., SF₅‐BCPs and SF₅‐CBs), while other substitution patterns have remained largely unexplored [3, 35, 36, 37, 38, 39, 40, 41]. In parallel, the OBH motif has been validated as a saturated bioisostere of meta‐substituted phenyl rings, with prior studies emphasizing improvements in physicochemical properties associated with lower lipophilicity [65, 68]. The SF₅‐OBHs in Figure 8 represent the first class of aliphatic SF₅‐containing “hybrid bioisostere” motifs that afford an exit vector for the SF₅ group that maps onto a meta‐substituted benzene ring. However, because the SF₅ group itself is highly lipophilic, it was not a priori evident that incorporation of an SF₅‑substituted OBH would preserve these favorable property trends. Accordingly, with synthetic access to SF₅‑OBHs now in hand, we sought to evaluate whether this m‑CF₃‑Ph → SF₅‑OBH hybrid bioisosteric replacement could be implemented without incurring substantial liabilities in the in vitro ADME profile relative to a parent drug‑like molecule (Figure 10).

FIGURE 10.

Evaluation of the m‐CF₃‐Ph → SF₅‐OBH hybrid bioisosteric replacement. Synthesis of a derivative (62) of a previously reported PROT inhibitor (61) and summary of in vitro ADME properties for side‐by‐side comparisons. Isolated yields reported.

We identified 61 as a synthetically accessible m‐CF₃‐Ph‐containing bioactive comparator, which has been reported as an inhibitor of the proline transporter (PROT), a target class of interest for the treatment of cognitive disorders (e.g., Alzheimer's disease) [131]. We synthesized this reference compound according to literature [129] and incorporated SF₅‐OBH 41 into an analogous synthesis route. That is, we converted 41 to carboxylic acid 53 via lithiation/CO₂ trapping (vide supra), arylated piperazine with 2‐chloropyrimidine under S_NAr conditions (93%), and subsequently coupled the fragments via amide bond formation using EDC and DMAP to afford SF₅‐OBH derivative 62 (86%). Ultimately, compounds 61 and 62 were obtained to enable side‐by‐side in vitro ADME profiling for comparative assessment of physicochemical properties.

The data revealed several noteworthy features (Figure 10). Most prominently, the m‐CF₃‐Ph → SF₅‐OBH replacement resulted in a substantial reduction in lipophilicity, with measured LogD decreasing by more than one logarithmic unit from 4.40 in 61 to 3.30 in 62, despite incorporation of the SF₅ group. Across the remaining assays, including high‐throughput (HT) solubility, liver microsomal clearance, reversible inhibition of CYP isoforms, membrane permeability (LE‐MDCK A–B), and cytotoxicity (HepG2), the SF₅‐OBH derivative 62 displayed a profile that was broadly comparable or marginally improved, relative to the parent compound 61. Notably, 62 also exhibited reduced liability with respect to both hERG and CYP2C9 inhibition.

Taken together, these data indicate that the SF₅‑OBH motif can serve as a viable m‐CF₃‐Ph hybrid bioisosteric replacement from a physicochemical and ADME perspective, while offering potential advantages associated with decreased lipophilicity and select safety‑related readouts in medicinal chemistry programs.

3. Conclusion

This study originated as a means to evaluate whether substituent effects and radical polarity matching logic could be used in tandem to enable regio‐ and diastereoselective SF₅Cl addition across 3‐aryl‐1‐carbonyl‐substituted BCBs (METHOD A). Beyond our initial reactivity and selectivity studies demonstrating this point, we made additional discoveries along the way that stimulated evolution of this work in fortuitous and fruitful ways. From a fundamental perspective, X‐ray crystallographic analyses of the SF₅Cl addition adducts and derivatives thereof led to characterization of (and computational insight on) the shortest SF₅⋯C═X (X = C, N, or O) interactions and the first CF₃SF₄⋯C═O interaction observed to date. From a practical perspective, evaluation of substituent effects on the relative chemical stabilities of certain tetrasubstituted SF₅‐CBs led to the development of a telescoped addition‐reduction sequence that affords readily isolable alcohol‐containing tetrasubstituted SF₅‐ and CF₃SF₄‐CBs (METHOD B), as well as an intramolecular cyclization strategy to access heretofore unknown SF₅‐ and CF₃SF₄‐OBHs (METHOD C). After examining the chemical stability of SF₅‐CBs and SF₅‐OBHs under various reaction conditions and expanding the accessible building block space, our efforts culminated in the synthesis/evaluation of an SF₅‐OBH drug derivative. Comparative in vitro ADME profiling revealed that the m‐CF₃‐Ph → SF₅‐OBH double replacement afforded similar or improved physicochemical properties, establishing SF₅‐OBH as a reasonable m‐CF₃‐Ph hybrid bioisostere.

Ultimately, we envision the three synthetic methodologies presented herein—alongside increased understanding of SF₅⋯C═X interactions and introduction of SF₅‐OBHs as “hybrid bioisosteres” for m‐CF₃‐Ph motifs—will inform and inspire additional advancements in pentafluorosulfanylation chemistry, as well as provide opportunities for synthetic chemists to explore new ways to incorporate the SF₅ group into organic molecules.

Conflicts of Interest

None of the authors have a conflicts of interest to disclose.

Supporting information

The authors have cited additional references within the Supporting Information [133–178].

Supporting File 1: anie72858‐sup‐0001‐Data.zip.

Supporting File 2: anie72858‐sup‐0002‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are openly available in iochem‐bd at https://doi.org/10.19061/iochem‐bd‐6‐604.