Studies directed toward the exploitation of vicinal diols in the synthesis of (+)-nebivolol intermediates

While the exploitation of the Sharpless asymmetric dihydroxylation as the source of chirality in the synthesis of acyclic molecules and saturated heterocycles has been tremendous, its synthetic utility toward chiral benzo-annulated heterocycles is relatively limited. Thus, in the search for wider applications of Sharpless asymmetric dihydroxylation-derived diols for the synthesis of benzo-annulated heterocycles, we report herein our studies in the asymmetric synthesis of (R)-1-((R)-6-fluorochroman-2-yl)ethane-1,2-diol, (R)-1-((S)-6-fluorochroman-2-yl)ethane-1,2-diol and (S)-6-fluoro-2-((R)-oxiran-2-yl)chroman, which have been used as late-stage intermediates for the asymmetric synthesis of the antihypertensive drug (S,R,R,R)-nebivolol. Noteworthy is that a large number of racemic and asymmetric syntheses of nebivolol and their intermediates have been described in the literature, however, the Sharpless asymmetric dihydroxylation has never been employed as the sole source of chirality for this purpose.


Findings
Chiral chromans are prevalent components of a large number of natural products, natural product-like molecules and pharmaceutical drugs, possessing diverse biological activities [1]. In view of their wide spectrum of biological profiles, chiral chromans have become attractive synthetic targets in academia and pharmaceutical industry [1]. Nebivolol (1, Figure 1) is a chroman-based antihypertensive drug that was first reported in the racemic form [2,3]. Chiral HPLC was subsequently employed to access various stereoisomers of 1 in enantiomerically pure form [4,5]. Out of the ten possible stereoisomers of 1, (S,R,R,R)-nebivolol or (+)-nebivolol (1a, Figure 1) was found to be a potent β 1 -adrenergic receptor blocker [2,3]. On the other hand, the corresponding enantiomeric form, (R,S,S,S)-nebivolol or (−)-nebivolol (1b, Figure 1) was found to be devoid of β 1 -antagonist activity, but it showed a significant synergistic effect on the antihypertensive proficiency of the (S,R,R,R) isomer [6][7][8][9].
(obtained by intramolecular epoxide ring-opening of 8 which is the enantiomer of 6) has been followed to obtain 2 (Scheme 1, method 3) whereupon the overall yield of the reaction sequence diminished [11].
On the other hand, the Sharpless asymmetric dihydroxylation (SAD) has been a workhorse as a synthetic tool for accessing enantiopure vicinal diols [21]. The extensive work in this field has resulted in the discovery of a number of cinchona alkaloidderived ligands which allow the dihydroxylation of alkenes of almost all substitution patterns with high enantioselectivity. Noteworthy is that the SAD is not limited to only E-allylic alcohols in its choice of substrates as is the SAE process. Moreover, the SAD is much more superior in terms of operational simplicity as SAE, it can be run at 0 °C in water as a co-solvent and under an atmosphere open to air.
The application of SAD-derived vicinal diols in the synthesis of acyclic molecules and saturated heterocycles has been astonishing. However, their utilities in the synthesis of chiral benzoannulated heterocycles are relatively limited [22][23][24][25][26]. In this paper, we describe our efforts toward the synthesis of 2, 3 and 5 using different cyclization strategies. To the best of our knowledge, nebivolol or its intermediates have never been synthesized using Sharpless asymmetric dihydroxylation as the sole source of chirality [27].
For the synthesis of chroman derivative 2, first a base-mediated intramolecular S N Ar reaction was envisioned for the aryl C-O bond formation under transition-metal-free conditions [28][29][30]. The additional benefit of this strategy would be the non-requirement of any protecting group to construct the chroman ring. To test this seemingly straightforward approach, we initially attempted to synthesize (±)-2 utilizing the cyclization of (±)triol 14 (Scheme 2) as the key step.
Thus, commercially available 2,5-difluorobenzaldehyde (10) was treated with the Wittig reagent Ph 3 P=CHCO 2 Et and the resulting unsaturated ester was then hydrogenated with Pd-C and H 2 at room temperature to obtain compound 11 in 91% yield over two steps (Scheme 2). DIBAL-H (1.1 equiv, -78 °C) reduction of the ester group of 11 followed by Wittig olefination of the resulting crude aldehyde with Ph 3 P=CHCO 2 Et provided (E)-α,β-unsaturated ester 12 (80% over two steps). A further DIBAL-H (2.5 equiv, 0 °C) reduction of 12 delivered (E)-allylic alcohol 13 (90%) which was then dihydroxylated under Upjohn conditions to obtain triol (±)-14 (92%). With access to (±)-14 we were in a position to investigate the key cyclization involving an intramolecular S N Ar to deliver 2. Unfortunately, all attempts of cyclizing (±)-14 to obtain chroman derivative (±)-2 under various S N Ar reaction conditions were not successful. Treatment of (±)-14 with KOt-Bu/ THF (65 °C), NaH/DMF (80 °C), NaH/DMSO (100 °C) and KOt-Bu/toluene (110 °C) did not lead to any conversion. However, more forcing conditions such as NaH/NMP (130 °C) resulted in a to partial decomposition of the starting material. These results indicated that an intramolecular S N Ar reaction of triol (±)-14 to form (±)-2 is not feasible. Thus, the presence of an activating substituent (e.g., a nitro group) at the C-5 position of the benzene ring might be helpful in synthesizing molecules similar to 2 [31,32].
The failure to achieve an intramolecular cyclization of the diol via an S N Ar reaction caused us to investigate other cyclization approaches towards these chroman derivatives. In 2005, Borhan and co-workers described the construction of tetrahydrofuran and tetrahydropyran structures from 1,2,n-triols via an elegant cyclization involving Lewis acid-mediated cyclization of in situ generated cyclic orthoesters [33]. We speculated that a similar reaction on an appropriately positioned diol with a tethered o-hydroxyphenyl group might produce 2-substituted chroman derivatives via a 6-exo-tet cyclization (Scheme 3).
To substantiate this hitherto unexplored approach in the context of synthesizing (+)-nebivolol intermediates, we first needed to synthesize the syn-dihydroxy esters 19 and 20 (Scheme 4). Toward that objective, 2-allyl-4-fluorophenol (15)   Debenzylation of diol 20 with Pd-C and H 2 at room temperature produced compound 21 having a tethered o-hydroxyphenyl group (Scheme 5). Next, compound 21 was exposed to Borhan's reaction conditions with the hope to obtain 22. To our dismay, however, the reaction furnished the hydrolyzed product 23 instead of cyclized product 22. It is important to mention that strict anhydrous conditions were maintained for this transformation in order to prevent the nucleophilic attack of H 2 O leading to the formation of 23. However, the formation of 23 instead of 22 clearly indicates that the in situ generated cyclic orthoester, after getting activated by Lewis acid, did not experience a nucleophilic attack of the phenolic hydroxy group; instead it reacted with water during the work-up process. This different outcome of this reaction compared to Borhan's results might be attributed to the lower nucleophilicity of phenols compared to alcohols.
Previously, Panda and co-worker converted succesfully a syn-2,3-dihydroxy ester into a 2-substituted chroman derivative [23]. The above-described unfortunate failures eventually forced us to turn our attention to utilize this methodology for the synthesis of 2-substituted chroman derivatives. Thus, diols 19 and 20 were subjected to a monotosylation reaction [34] to obtain the β-hydroxy-α-tosyloxy esters 24 and 25, respectively (Scheme 6). Panda and co-worker applied a three-step reaction sequence involving epoxidation/debenzylation/epoxide ring-opening to convert the β-hydroxy-α-tosyloxy ester into the corresponding 2-substituted chroman derivative. However, it has been reviewed that not only benzylic epoxides but also non-benzylic epoxides are sensitive to the standard hydrogenation/debenzylation conditions [35]. Whereas benzylic epoxides are highly sensitive to hydrogenation conditions, non-benzylic epoxides, depending on the reaction conditions, may produce traces to significant amounts of side-products via hydrogenolysis. Thus, we decided to modify Panda's synthetic route to significantly increase the overall yield. We hypothesized that this problem might be circumvented by performing the debenzylation reaction prior to the epoxide-ring formation. Further we speculated that compound 26, in the presence of a base, might undergo a simultaneous epoxidation-intramolecular epoxide-ring opening to produce 27 (Scheme 7) as the corresponding benzoxepin ring formation via intramolecular displacement of -OTs group by ArO − is unresponsive [23]. To test this hypothesis, first compound 24 was subjected to the debenzylation reaction with 10% Pd-C in EtOH under an H 2 atmosphere (Scheme 8) at room temperature. After completion of the debenzylation process, K 2 CO 3 was added to the reaction mixture in the same reaction vessel. The reaction mixture, after being run for additional 6 h, provided compound 27 in 70% yield which is significantly higher than the literature yield (53%) for the similar transformation [23]. The similar strategy was applied in converting 25 into chroman derivative 28. Next, LiAlH 4 reduction of 27 and 28 provided 2 and 29, respectively, in 93% yield. It is to be mention that the NMR spectra and spe-cific rotations of 2 and 29 matched with those reported in the literature [11,14,15]. For the synthesis of compound 3, stereoisomer 29 was subjected to classical two-step Mitsunobu inversion protocol which was successful but poor yielding (Scheme 9).
Not surprised by the poor yield of this transformation, we focused on the conversion of 28 into 5 which has also been used as a late-stage nebivolol intermediate (5 is a more advanced intermediate compared to 3). Thus, the tosylation of compound 28 followed by LiBH 4 reduction of the resulting tosylate and subsequent epoxidation of the obtained tosyloxy alcohol with anhydrous K 2 CO 3 in absolute ethanol (Scheme 10) provided compound 5. The NMR spectra and specific rotations of 5 also matched those reported in the literature [17][18][19]36].

Supporting Information
Supporting Information File 1 Experimental procedures, characterization data and copies of 1 H and 13 C NMR spectra for final compounds are available.