Iterative Catalyst-Controlled Diastereoselective Matteson Homologations Enables the Selective Synthesis of Benzestrol Isomers

Abstract

We report the development of an iterative Matteson homologation reaction with catalyst-controlled diastereoselectivity through the design of a new catalyst. This reaction was applied to the selective synthesis of each stereoisomer of benzestrol, a bioactive compound with estrogenic activity featuring three contiguous stereocenters. The different stereoisomers were assayed to determine their binding affinity for the Estrogen Receptor α (ERα) and the absolute configuration of the compound having uniquely high activity was determined. This research lays a framework for the catalytic synthesis and study of complete stereoisomeric sets of other bioactive molecules and chemical probes containing contiguous stereocenters.

Graphical Abstract:

The construction of compounds bearing contiguous carbon stereocenters represents one of the longest standing challenges and intensively pursued endeavors in the field of organic synthesis.^1–4 Methods that allow for the efficient preparation of all stereochemical permutations of such targets are of particular interest,^5–7 especially in light of the emerging utilization of complete stereoisomeric sets of chemical probes in medicinal chemistry and chemical biology.^8,9 The Matteson homologation reaction stands as a particularly versatile approach to the iterative construction of compounds bearing contiguous stereocenters. The one-carbon extension of boronic esters generates α-chloro boronic esters that can be stereospecifically elaborated to secondary boronic esters, and those can in turn serve as substrates for subsequent homologation reactions.^3,10–12

Matteson successfully demonstrated the principle of iterative homologation through the use of chiral diol-derived auxiliaries on boron (Fig. 1A).^13–16 In these reactions, the configuration of each new stereocenter on the growing chain is dictated by the absolute stereochemistry of the diol. While the auxiliary approach enables the highly selective synthesis of stereochemically complex compounds, the auxiliary must be replaced with its enantiomer to access all stereoisomers.^3,17 The need for repeated auxiliary swaps is circumvented if the chiral controlling element does not reside on the boronate. In that context, the Aggarwal group has leveraged enantioenriched organolithium stoichiometric reagents in iterative Matteson-type homologation sequences (Fig. 1B).^4,7,18–21 The chiral lithium reagent approach has been applied in many remarkable synthetic contexts, with construction of products bearing up to ten contiguous stereocenters set iteratively with absolute and independent control.⁴

Figure 1.

Strategies for synthesizing contiguous stereocenters through controlling the selectivity of 1,2-boronate rearrangements. ^asp = sparteine, ^bCb = carbamate, es = enantiospecificity

Our group recently uncovered an enantioselective catalytic variant of the Matteson homologation reaction that relies on a new class of chiral lithium-isothiourea-boronate complexes as catalysts (Fig. 1C).^22–23 The method was found to display broad scope with primary alkylboronic esters. We envisioned that the α-chloro boronic ester products could be elaborated stereospecifically and reintroduced to a suitably adapted catalytic reaction to generate new contiguous α-chloro boronic ester stereocenters under catalyst control (Fig 1D). Given the wide range of elaborations available to the α-chloro boronic ester produced in each iteration, the catalytic reaction could enable the highly modular synthesis of an extraordinarily wide variety of products bearing contiguous stereocenters. Herein, we report the development of an iterative catalytic stereoselective Matteson homologation, and its application towards the synthesis and biological studies of the estrogen-receptor agonost benzestrol and each of its stereoisomers.

Benzestrol was first described as part of an extensive effort in the 1940’s to prepare highly potent non-steroidal estrogens.²⁴ Its structure contains three contiguous tertiary stereocenters, and can thus potentially exist as eight stereoisomers in four diastereomeric enantiomeric pairs. The synthetic methods of the time did not allow selective synthesis of the individual isomers, but the four sets of racemates were isolated from a stereorandom synthetic product mixture by laborious fractional crystallizations. In vivo potency assays in rats revealed that the potency of one racemate, designated B-2, exceeded that of the other three by 6–44 fold.²⁵ This component was named benzestrol, but nothing was known about its absolute stereochemistry. As a potent synthetic estrogen, benzestrol was briefly used clinically to prevent premature childbirth, but was withdrawn for safety concerns.^26–28

To our knowledge, the individual bioactivity of each of the eight potential stereoisomers of benzestrol has not been described,^24,29 nor has a stereoselective route to this class of molecules been reported. This imposes obvious practical challenges to performing an analysis of the role that stereochemistry plays in estrogenic activity.^24,30,31 The development of a high-yielding, stereomodular synthesis of all eight benzestrol isomers could, in principle, enable elucidation of the structure-activity relationship of this family of therapeutically relevant chemical probes, potentially enabling the design and synthesis of more potent synthetic estrogens. More broadly, such an effort could provide a framework for the synthesis and study of other chemical probes bearing multiple contiguous stereocenters.⁸

We envisioned that the carbon framework of benzestrol could be constructed via three iterative sequences each involving catalyst-controlled homologation and stereospecific chloride elaboration, followed by a stereospecific C–B arylation to deliver the final product (Fig. 1D). Through this approach, the stereochemistry of each tertiary center might be dictated by the choice of catalyst enantiomer in the homologation step, thereby enabling the stereoselective synthesis of each benzestrol stereoisomer. The foremost consideration in such a strategy was that high levels of catalyst-induced stereocontrol would be required in homologations of boronic esters bearing α-stereocenters. However, in the original study of lithium-isothiourea-boronate–catalyzed rearrangement reactions, the scope of highly effective substrates was found to be limited to substrates bearing primary alkyl and small cyclic secondary alkyl migrating groups.

We therefore initiated our efforts with a systematic examination of the performance of secondary boronic esters in the catalytic rearrangement reaction. As expected, given the previous observations, achiral secondary-alkyl boronic esters were found to undergo 1a-catalyzed homologation with decreasing levels of enantioselectivity as a function of increasing steric demand of the migrating alkyl group (Fig. 2A). Analysis of the computed transition-state model for the reaction of a primary alkyl boronate that was included in the original study suggested a possible explanation for the deleterious effect of sterically demanding migrating groups on catalyst performance (Fig. 2B). In the model for the lowest energy transition state leading to the major product enantiomer, the methyl substituent on the aryl pyrrolidine component of the catalyst appears to project toward the migrating group, in a manner that could potentially result in a steric clash with bulkier secondary alkyl substrates. To test this hypothesis, the des-methyl analog (1b) of the original catalyst was prepared and compared to 1a in homologations across a set of alkyl boronic ester substrates (Fig. 2A). Although the effects are subtle, a clear trend can be discerned in the formation of products 3a-3f, with smaller substrates performing better with the original catalyst, but hindered alkylboronic ester substrates 2d-f undergoing reaction with slightly improved enantioselectivity with catalyst 1b.³² A similar subtle improvement in catalyst-induced stereoselectivity was observed in the homologation of 2g, a chiral substrate that was selected to model the second iteration step in the planned synthesis of benzestrol. Catalyst 1b outperformed 1a, with both enantiomers inducing excellent levels of catalyst control over product diastereoselectivity (Fig. 1C).

Figure 2.

Catalyst optimization studies. ^aDifferent enantiomers of each catalyst were used for different results in part A. The absolute value of the ee is listed. ^bYield was determined by ¹H NMR analysis with 1,3,5-trimethoxybenzene or dibromomethane as an internal standard. ^cIntrinsic d.r. is a calculated value to normalize entries with different ee starting material (see SI for full discussion). DFT is reproduced with permission from ref. 22. Copyright [2024] [The American Association for the Advancement of Science].

The anticipated third iterative homologation in the proposed synthesis of the benzestrol stereoisomers was modeled initially using compound 2h, which bears an ethyl-substituted migrating group as required in the ultimate target structures. However, poor reactivity resulting in only 9–17% yield of the desired homologation product was observed with these highly sterically demanding substrates, even under the conditions optimized for the second homologation step using catalyst 1b (Fig 3). After an extensive reaction optimization effort, the yield was only improved to 27% by decreasing the reaction concentration (see SI for a full discussion of optimization efforts). We suspected that the increased steric crowding near the boron center was responsible for the low reactivity, given that large migrating groups are known either to reduce the overall rate of boronate formation relative to decomposition via dichlorocarbene formation, or to favor reversion to starting material instead of productive 1,2-migration.^3,17,5,33 Consistent with that hypothesis, the analogous substrate 2i bearing a methyl substituent on the migrating group underwent reaction with a moderate increase in yield relative to 2h, from 27% to 44%. A more substantial improvement in reactivity was obtained by replacing the substituent on the migrating group with a vinyl group, affording the desired product in 71% yield (3j). The vinyl group could be introduced stereospecifically in the appropriate elaboration step and easily transformed to the ultimately desired ethyl group in a late-stage hydrogenation, or could potentially serve as a synthetic handle for future efforts in analogue synthesis.

Figure 3.

Optimization of yield in the third diastereoselective Matteson homologation. ^aYield of the combined diastereomers determined by ¹H NMR analysis with 1,3,5-trimethoxybenzene or dibromomethane as an internal standard. ^bThe d.r. represents the ratio of the major diastereomer (shown) to the sum of all the other diastereomers. The yield represents the sum of the four diastereomeric products. ^cA mixture of syn and anti diastereomers was used.

With the results of the model studies in hand, the synthesis of the benzestrol stereoisomers was initiated with the highly enantioselective homologation of EtBpin with catalyst 1a, followed by stereospecific elaboration with an aryl Grignard reagent (Fig. 4). A brief survey of arene substituent effects on the diastereoselectivity of the second homologation revealed that superior results were obtained with substrates bearing electron-withdrawing groups, so the p-chlorophenyl derivative was employed as a phenol surrogate. The second homologation proceeded with excellent catalyst control on boronic ester 4, with both enantiomers of catalyst 1b inducing similar levels of diastereocontrol. Elaboration of the a-chloroboronic ester intermediates with (vinyl)MgBr afforded the diastereomic products 5 and 6 poised for the third elaboration step. The intrinsic substrate bias was modest for the homologation of 5 in the absence of catalyst (8:7 = 63:37), and negligible for the conversion of 6 (9:10 = 50:50). Both enantiomers of catalyst 1b exerted high levels of diastereoselectivity over boronic ester 5 (d.r. = 7:8:9:10 = 95:2:1:2 or 9:88:2:1) (Fig 4).

Figure 4.

Optimization of the three iterative diastereoselective Matteson homologations. ^aThe d.r. from the Matteson homologation decreased from 97:3 to 96:4 during purification. ^bThe d.r. from the Matteson homologation increased from 5:95 to 4:96 d.r. during purification. ^cYield of the combined diastereomers determined by ¹H NMR analysis with dibromomethane as an internal standard.

In contrast boronic ester 6 underwent homologation to 9 and 10 with more variable levels of catalyst control (4:1:59:36 or 1:2:16:81) (Fig 4). The especially poor control in the formation of 9 may be ascribable to a particularly unfavorable steric interaction between the anti-configured R groups (ethyl and vinyl) on 6 and the catalytic pocket (Fig 2B and Fig. 4). However, we recognized that by taking advantage of the modularity of the Matteson homologation, even the benzestrol stereoisomer that would arise from elaboration of 9 could be accessed with high levels of selectivity (Fig 5).

Figure 5.

Synthesis of all four benzestrol diastereomers from two common intermediates. The structures of tBuBrettPhos and the Pd precatalyst are provided in footnote 37.

Thus, using the product of the more selective third iteration (7 or 10) and either 1) elaborating first with an aryl Grignard reagent followed by conversion of the boronic ester to a methyl substituent or 2) elaborating first with methyllithium followed by conversion of the boronic ester to an aryl substituent, both relevant stereoisomers of benzestrol could be accessed with high d.r. (Fig 5). This approach is similar to the one applied in the iterative synthesis of the stereotriad Tatanan A by the Aggarwal group, where undesired dominant substrate control in the third iteration was addressed by replacing a stereoinvertive process with a streoretentive one.⁷

In this manner, completion of the synthesis could be accomplished by a common protocol. Intermediate ent-7 was elaborated by displacement of the chloride leaving group with p-chlorophenyl Grignard reagent followed by homologation of the resulting boronic ester 11 with chloromethyllithium, followed by protodeborylation to afford 12.^22,34–36, Palladium-catalyzed hydroxylation of both aryl chlorides gave a bis-phenol, which was subjected to hydrogenation to afford benzestrol isomer 1 (13) with a 4.9% overall yield, >99% ee, and 97:<1:<1:3 d.r. (d.r. = 13:16:19:22).³⁷ Inverting the displacement-replacement sequence gave access to isomer 2 (16). Thus ent-7 was first treated with methyllithium, then subjected to a stereospecific conversion of the boronic ester to a phenol to access 15.³⁸ As with isomer 1, hydroxylation and hydrogenation proceeded smoothly to afford isomer 2 (16) in an 8.0% overall yield, >99% ee, and <1:98:1:1 d.r.. The final two diastereomers (19 and 22) were prepared via the same synthetic steps starting from intermediate 10.³⁹

The enantiomers of each diastereomer were synthesized using the opposite enantiomers of catalysts 1a and 1b to ultimately afford all eight stereoisomers of benzestrol with excellent diastereo- and enantioselectivity.

With all eight stereoisomers of benzestrol available—each with known absolute configurations—we had the opportunity to explore the relationship between ligand stereochemistry and biological activity. A competitive radiometric binding assay with ERα was performed, revealing that the benzestrol stereoisomers exhibited distinctive binding activities (Fig. 6). Notably, Isomer 19 emerged as the only ligand with very high binding affinity for ERα, with any change in its configuration significantly diminishing its binding affinity. In fact, Isomer 19 binds more strongly to ERα than estradiol (E2), the native ligand. The absolute stereochemistry of 19 was confirmed by X-ray crystallography (See SI), Our ongoing studies aim to further investigate the relationships between stereochemistry and biological activity for these isomers, particularly in the context of other conformationally flexible bisphenolic non-steroidal estrogens.

Figure 6:

Relative Binding Affinities (RBAs) and K_i values for all eight benzestrol isomers and estradiol (E2) with ERα. RBA values are referenced to that of E2 with the RBA of E2 set to be 100. RBA values are from two or more assays. K_i values are calculated from RBAs through the equation 0.2*(100/RBA) = K_i. ^acontains 2% of 19 and ent-19. ^bContains <1% of 19 and ent-19.

In conclusion, we have developed a catalytic, iterative diastereoselective Matteson homologation and applied it to the synthesis of benzestrol and each of its stereoisomers. This method offers a versatile approach to the synthesis and investigation of other chemical probes with multiple contiguous stereocenters.