Concise Large-Scale Synthesis of Tomatidine, A Potent Antibiotic Natural Product

Tomatidine has recently generated a lot of interest amongst the pharmacology, medicine, and biology fields of study, especially for its newfound activity as an antibiotic agent capable of targeting multiple strains of bacteria. In the light of its low natural abundance and high cost, an efficient and scalable multi-gram synthesis of tomatidine has been developed. This synthesis uses a Suzuki–Miyaura-type coupling reaction as a key step to graft an enantiopure F-ring side chain to the steroidal scaffold of the natural product, which was accessible from low-cost and commercially available diosgenin. A Lewis acid-mediated spiroketal opening followed by an azide substitution and reduction sequence is employed to generate the spiroaminoketal motif of the natural product. Overall, this synthesis produced 5.2 g in a single pass in 15 total steps and 15.2% yield using a methodology that is atom economical, scalable, and requires no flash chromatography purifications.


Introduction
The genus Solanum consists of over 2000 species of plants widely distributed over the world, from which over 600 natural products have been isolated [1]. Ten percent of these products are alkaloid steroids, which are classified into the Veratrum and Solanum alkaloid families [2]. Tomatidine (TO, 1), which belongs to the solanidine subfamily of Solanum, has recently gathered great interest from the scientific community for its biological and medicinal applications stemming from its anabolic [3], antioxidant [4], neuroprotective [5], antiviral [6,7], anticancer [8], and most importantly, antibiotic [9][10][11][12][13] properties. Our research group originally discovered the antibiotic potential of 1 against Staphylococcus aureus and its virulent small-colony variants (SCV), which are responsible for resistance and persistence of the pathogen [9,10]. Structure-activity relationship (SAR) studies based on modifications of the natural product showed that it was possible to bring the minimum inhibitory concentration (MIC) of S. aureus from >128 µg/mL to 2 µg/mL using the most active derivative FC04-100 [11,12]. Furthermore, the MIC could be brought down from 64 µg/mL to 8 µg/mL for the SCVs (8-fold potency increase) when TO was used synergistically with the aminoglycoside antibiotic gentamycin [12].
Accessibility is a challenge commonly encountered in the study of biologically active natural products [14]. Recently, our group has shown that the potent alkaloid tomatidine could be synthesized on gram-scale [15], with only one other methodology producing a close derivative (tomatidenol, ∆5-6 -1) [16] in small quantities (50 mg) through non-strategic redox [17]. With SAR studies still on-going in our laboratory and considering the common profound scientific interest, high costs (> USD 1500/g) and scarce natural availability of this steroid, we tackled the ambitious challenge of developing a new synthesis capable of decagram scale quantities of 1.
Despite providing over 2 g of tomatidine, we identified multiple shortcomings in our previous synthetic methodology that would render a scale-up largely inefficient. Our previous methodology (Scheme 1a, red) [15] relied on a chiral auxiliary to set the stereocenter of Molecules 2021, 26, 6008 2 of 15 iodo-3, followed by its introduction to 5 as a temperature-sensitive organolithium reagent to obtain adduct 4, which then underwent acid-mediated spiroketalization to obtain 2. To develop a scalable and atom-economical synthesis, inefficient steps of the first-generation synthesis, such as the use of the heavy TBDPS protecting group, requirement of a chiral auxiliary, and the arduous temperature control of the organolithium 3 generation on higher scales, had to be addressed. To further optimize the overall practicality of this methodology, we also set the ambitious objective of avoiding the use of flash chromatography purifications for the whole sequence, aiming to develop clean transformations and to obtain pure intermediates from recrystallizations, distillations, and silica pads only.
of avoiding the use of flash chromatography purifications for the whole sequence, aiming to develop clean transformations and to obtain pure intermediates from recrystallizations, distillations, and silica pads only.
We recognized early in our new retrosynthetic analysis (Scheme 1a, green) that the C22-C23 bond was an efficient disconnection strategy for introducing a side chain synthon containing the required (S)-C25 of the natural product. We opted to use the Suzuki-Miyaura reaction to combine the steroid scaffold 5 to the required asymmetric synthon 8 (as organoborane 6). This reaction is known to be a generally efficient and cost-effective method to give access to pharmaceutical intermediates and natural products even on industrial scales [18]. The union of an iodo enol ether (as iodo dihydropyran) to an alkylborane generated in situ from a parent olefin through Pd catalysis has been previously demonstrated by Tan et al. [19] to afford the desired adducts in high yields (Scheme 1,b). We planned to use and optimize this methodology by circumventing the use of the stannyl intermediate in the preparation of the required iodo enol ether. Succeeding at this critical step would, following cyclisation, lead us to spirostanol 2, which would lead us to tomatidine 1 through our previously developed sequence of five reactions [15]. We recognized early in our new retrosynthetic analysis (Scheme 1a, green) that the C 22 -C 23 bond was an efficient disconnection strategy for introducing a side chain synthon containing the required (S)-C 25 of the natural product. We opted to use the Suzuki-Miyaura reaction to combine the steroid scaffold 5 to the required asymmetric synthon 8 (as organoborane 6). This reaction is known to be a generally efficient and cost-effective method to give access to pharmaceutical intermediates and natural products even on industrial scales [18]. The union of an iodo enol ether (as iodo dihydropyran) to an alkylborane generated in situ from a parent olefin through Pd catalysis has been previously demonstrated by Tan et al. [19] to afford the desired adducts in high yields (Scheme 1,b). We planned to use and optimize this methodology by circumventing the use of the stannyl intermediate in the preparation of the required iodo enol ether. Succeeding at this critical step would, following cyclisation, lead us to spirostanol 2, which would lead us to tomatidine 1 through our previously developed sequence of five reactions [15].
Preparing the required enol ether following this strategy would start from functionalization of lactone 5, followed by coupling to alkylborane 6 to obtain adduct 7, which, after an acid-catalyzed spiroketalization, would deliver 2. This approach requires less of the valuable asymmetric synthon 8 (1.5 eq., vs. 2.0 eq. required in the first-generation synthesis) and avoids the temperature dependency of its union to the steroid backbone. Scaffold 5 could be obtained in two steps on large scale from diosgenin [20], an inexpensive (USD 0.30/g) sapogenin, obtained mainly from the acidic hydrolysis of wild yam (Dioscorea) on industrial scale [21].

Results and Discussion
Our efforts started with the preparation of the first anchor for the Pd-mediated Suzuki-Miyaura reaction, the steroid core. We initially probed the reactivity of lactone 5 as its C 3 -protected congeners 9a-c (Scheme 2) in the hope of obtaining enol triflates 10a-c, which would give access to the required coupling partner following our retrosynthetic strategy. Conversion of ketones and lactones to their corresponding enol triflates followed by crosscoupling is a strategic way to access diverse complex natural products in a convergent manner [22]. Despite extensive efforts, lactones 9a-c resisted triflation under common conditions (LiHMDS/KHMDS/LDA then Ph 2 NTf/Tf 2 O/Comin's reagent) [22,23], most likely because of the strong steric hindrance imparted by the C 18 and C 21 methyl groups restricting the deprotonation at C 20 . This hypothesis was rationalized when attempts at quenching the supposed enolates with deuterated protic sources (D 2 O, AcOD) yielded no 2 H incorporation at C 20 . Preparing the required enol ether following this strategy would start from functionalization of lactone 5, followed by coupling to alkylborane 6 to obtain adduct 7, which, after an acid-catalyzed spiroketalization, would deliver 2. This approach requires less of the valuable asymmetric synthon 8 (1.5 eq., vs. 2.0 eq. required in the firstgeneration synthesis) and avoids the temperature dependency of its union to the steroid backbone. Scaffold 5 could be obtained in two steps on large scale from diosgenin [20], an inexpensive (USD 0.30/g) sapogenin, obtained mainly from the acidic hydrolysis of wild yam (Dioscorea) on industrial scale [21].

Results and Discussion
Our efforts started with the preparation of the first anchor for the Pd-mediated Suzuki-Miyaura reaction, the steroid core. We initially probed the reactivity of lactone 5 as its C3-protected congeners 9a-c (Scheme 2) in the hope of obtaining enol triflates 10ac, which would give access to the required coupling partner following our retrosynthetic strategy. Conversion of ketones and lactones to their corresponding enol triflates followed by cross-coupling is a strategic way to access diverse complex natural products in a convergent manner [22]. Despite extensive efforts, lactones 9a-c resisted triflation under common conditions (LiHMDS/KHMDS/LDA then Ph2NTf/Tf2O/Comin's reagent) [22,23], most likely because of the strong steric hindrance imparted by the C18 and C21 methyl groups restricting the deprotonation at C20. This hypothesis was rationalized when attempts at quenching the supposed enolates with deuterated protic sources (D2O, AcOD) yielded no 2 H incorporation at C20. Following this roadblock, we investigated the work of Boeckman et al., which probed the reactivity of cyclic enol ethers (as 3,4-dihydropyrans) by deprotonation using strong bases, followed by treatment with a wide range of electrophiles [24,25]. To the same extent, we aimed to prepare the equivalent enol ether (as 2,3-dihydrofuran) derivative of 5 to create the required reactive handle for the Suzuki-Miyaura reaction following these conditions. Heavy silyl protecting groups (TBS, TBDPS) were not considered to protect the C3 alcohol since studies have previously shown unpredictable migratory behavior in the required strongly basic conditions [26,27] and were not ideal in terms of mass economy. In contrast, the MOM-protecting group was chosen owing to its ability to withstand strong basic media, low molecular weight, and acid lability, enabling its removal during the acid-promoted spiroketalization whilst saving a deprotection step and enhancing the overall efficiency.
We began our synthesis (Scheme 3) by treating 5 with MOMCl in the presence of DIPEA/TBAI [28] to afford MOM-protected lactone 9a, followed by reduction with i Bu2AlH [20] to produce the corresponding hemiacetal 11 in 95% yield in two steps on 2 g scale as an inconsequential mix of diastereoisomers (2:1). Despite the screening of different catalytic acidic conditions previously used in dehydrations of hemiacetals [29], 12 could not be obtained cleanly without removing the MOM-protecting group. Fortunately, the activation of the hemiacetal using MsCl and subsequent elimination with Following this roadblock, we investigated the work of Boeckman et al., which probed the reactivity of cyclic enol ethers (as 3,4-dihydropyrans) by deprotonation using strong bases, followed by treatment with a wide range of electrophiles [24,25]. To the same extent, we aimed to prepare the equivalent enol ether (as 2,3-dihydrofuran) derivative of 5 to create the required reactive handle for the Suzuki-Miyaura reaction following these conditions. Heavy silyl protecting groups (TBS, TBDPS) were not considered to protect the C 3 alcohol since studies have previously shown unpredictable migratory behavior in the required strongly basic conditions [26,27] and were not ideal in terms of mass economy. In contrast, the MOM-protecting group was chosen owing to its ability to withstand strong basic media, low molecular weight, and acid lability, enabling its removal during the acid-promoted spiroketalization whilst saving a deprotection step and enhancing the overall efficiency.
We began our synthesis (Scheme 3) by treating 5 with MOMCl in the presence of DIPEA/TBAI [28] to afford MOM-protected lactone 9a, followed by reduction with i Bu 2 AlH [20] to produce the corresponding hemiacetal 11 in 95% yield in two steps on 2 g scale as an inconsequential mix of diastereoisomers (2:1). Despite the screening of different catalytic acidic conditions previously used in dehydrations of hemiacetals [29], 12 could not be obtained cleanly without removing the MOM-protecting group. Fortunately, the activation of the hemiacetal using MsCl and subsequent elimination with an excess of Et 3 N at high temperatures [30] afforded the desired enol ether 12 in a modest 65% yield on a 2 g scale after passage through a short silica pad. 3, Method B), a niche electrophilic source of iodine. [34] To our delight, this methodology cleanly afforded 13 in 91% yield on a 2 g scale. A slight optimization of this protocol allowed us to use 3.0 eq. of both reagents ( t BuLi and 1,2-DIE) instead of the original 4.0 eq. prescribed, while maintaining >97% conversion on gram scale. Further lowering of the equivalents (<3.0 eq.) led to poor conversions (<75%). With the anchor of the Pd-mediated cross coupling installed on the steroidal scaffold, we then devised a scalable route to the desired olefin 8. Commercially available (R)-Roche ester 14 was protected as THP followed by reduction using LiAlH4 to yield (S)monoprotected diol 15 in 81% yield on a 25 g scale after distillation (Scheme 4) [35]. Subsequent oxidation using the Cornforth reagent (PDC) in conjunction with silica gel [36] cleanly yielded the corresponding aldehyde, which was then immediately subjected to a Wittig olefination [37]. The desired olefin 8 was cleanly obtained following a short silica pad in 62% yield after two steps on a 30 g scale. Having obtained both coupling partners, we then explored the scalability of the reported one-pot transformation to obtain adduct 7 from 6 and 13. Fortunately, we did Next, we subjected 12 to similar conditions as those described by Boeckman [24,25]. Following lithiation, attempts at reacting various electrophilic halogen sources such as NBS, DBDMH, NCS, NIS, and I 2 failed to produce the expected halogenated derivative. However, treatment of the organolithium intermediate with Bu 3 SnCl cleanly yielded the stannyl enol ether, which afforded the iodo enol ether 13 after reaction with molecular iodine (Scheme 3, Method A) [31]. Other interesting protocols have been reported for the synthesis of the equivalent obtained stannyl enol ether starting from a lactone derivative; however, they were too inefficient to be conducted on a large-scale [32,33]. Although the reaction with Bu 3 SnCl provided a direct route to 13, the use of a tin intermediate is chemically inefficient, dangerous to handle on such scales, and traces of the toxic heavy metal would likely be carried to the final product used in in vivo and SAR studies. We circumvented this liability by reacting lithio-12 with 1,2-diiodoethane (1,2-DIE) (Scheme 3, Method B), a niche electrophilic source of iodine. [34] To our delight, this methodology cleanly afforded 13 in 91% yield on a 2 g scale. A slight optimization of this protocol allowed us to use 3.0 eq. of both reagents ( t BuLi and 1,2-DIE) instead of the original 4.0 eq. prescribed, while maintaining >97% conversion on gram scale. Further lowering of the equivalents (<3.0 eq.) led to poor conversions (<75%).
With the anchor of the Pd-mediated cross coupling installed on the steroidal scaffold, we then devised a scalable route to the desired olefin 8. Commercially available (R)-Roche ester 14 was protected as THP followed by reduction using LiAlH 4 to yield (S)monoprotected diol 15 in 81% yield on a 25 g scale after distillation (Scheme 4) [35]. Subsequent oxidation using the Cornforth reagent (PDC) in conjunction with silica gel [36] cleanly yielded the corresponding aldehyde, which was then immediately subjected to a Wittig olefination [37]. The desired olefin 8 was cleanly obtained following a short silica pad in 62% yield after two steps on a 30 g scale. synthesis of the equivalent obtained stannyl enol ether starting from a lactone derivative; however, they were too inefficient to be conducted on a large-scale [32,33]. Although the reaction with Bu3SnCl provided a direct route to 13, the use of a tin intermediate is chemically inefficient, dangerous to handle on such scales, and traces of the toxic heavy metal would likely be carried to the final product used in in vivo and SAR studies. We circumvented this liability by reacting lithio-12 with 1,2-diiodoethane (1,2-DIE) (Scheme 3, Method B), a niche electrophilic source of iodine. [34] To our delight, this methodology cleanly afforded 13 in 91% yield on a 2 g scale. A slight optimization of this protocol allowed us to use 3.0 eq. of both reagents ( t BuLi and 1,2-DIE) instead of the original 4.0 eq. prescribed, while maintaining >97% conversion on gram scale. Further lowering of the equivalents (<3.0 eq.) led to poor conversions (<75%).

Scheme 3. Preparation of iodo enol ether 13.
With the anchor of the Pd-mediated cross coupling installed on the steroidal scaffold, we then devised a scalable route to the desired olefin 8. Commercially available (R)-Roche ester 14 was protected as THP followed by reduction using LiAlH4 to yield (S)monoprotected diol 15 in 81% yield on a 25 g scale after distillation (Scheme 4) [35]. Subsequent oxidation using the Cornforth reagent (PDC) in conjunction with silica gel [36] cleanly yielded the corresponding aldehyde, which was then immediately subjected to a Wittig olefination [37]. The desired olefin 8 was cleanly obtained following a short silica pad in 62% yield after two steps on a 30 g scale. Having obtained both coupling partners, we then explored the scalability of the reported one-pot transformation to obtain adduct 7 from 6 and 13. Fortunately, we did Having obtained both coupling partners, we then explored the scalability of the reported one-pot transformation to obtain adduct 7 from 6 and 13. Fortunately, we did not encounter any problems during the scale-up testing (up to 8 g) using the reported conditions [19]. Attempts at optimization by lowering the equivalents of alkylborane 6 Molecules 2021, 26, 6008 5 of 15 (from 1.5 through to 1.1 eq.) and catalyst loading (20 mol% through to 1 mol%) were met with high amounts of reduction back to parent enol ether 12 (up to 20%), so the originally reported conditions were employed. The transformation sequence from 7 to the desired spirostanol 2 was then completed under strong mineral acidic conditions (methanolic HCl), as previous work highlighted the undesired isomerization at C 25 when weaker mineral acids (e.g., AcOH) were used [38]. Spirostanol 2 was obtained in 8.1:1 d.r. at C 25 in favor of the desired 25-(S) diastereoisomer in 74% yield on a 1 g scale after recrystallization. The observed imperfect diastereoselectivity could be attributed to the loss of optical purity of the intermediate aldehyde of 15 during the Wittig olefination through uncontrolled racemization under the strong basic conditions.
After individually optimizing all the steps on a gram scale, we were able to streamline multiple steps to facilitate processing without compromising on either purity or yield (Scheme 5). We found that the protection, reduction, and dehydration sequence of 5 to 12 could be accomplished using crude intermediates in 65% yield on a 13 g scale and ultimately in 61% yield on a 46 g scale over three steps (avg. 85%/step), requiring only one silica pad after the third step. Minor decomposition products could be observed during the reduction step when the mixture warmed above −65 • C on the 13 g scale batch. In consequence, we decided to divide the 46 g scale batch in two equal parts to maintain better internal temperature control of the cryogenic mixture. The handling of the air-sensitive i Bu 2 AlH did not prove to be a safety hazard at this scale. Since the next reaction proved to be a safety hazard, we limited the scale of the transformation according to the t BuLi reagent bottle size and concentration. Transfer by canulation of a single reagent container (100 mL) of t BuLi to the reaction mixture limited our exposure to the pyrophoric reagent, which is important to avert potential accidents on this scale. [39] The iodination, cross-coupling, and spiroketalization steps from 12 to 2 were also streamlined to afford the desired spiroketal in 44% yield over three steps (avg. 76%/step) on a 19 g scale after recrystallization directly from the reaction mixture. Overall, this sequence yielded 5.2 grams of tomatidine in a single batch, starting from 46 g of lactone 5 in 15.2% yield after 11 steps LLS (longest linear sequence). Another smaller-scale iteration of this sequence enabled us to produce a total of >7.5 grams of synthetic tomatidine in the laboratory. An analysis of a single crystal by X-ray diffraction (XRD) unequivocally confirmed the identity of 1 (Scheme 5) as its methanol co-crystal. With a large quantity of the desired spiroketal 2 in hand, we applied our previously used successful endgame sequence of transformations with no complications [15]. Compound 2 was acetylated using Ac 2 O in pyridine and then subjected to Lewis acid-mediated spiroketal opening using BF 3 ·Et 2 O and LiBr, followed by substitution using NaN 3 in DMF to yield the desired azide 16 in 93% yield over three steps (avg. 98%/step) and high purity (>99%) over three steps, requiring no purification. Subsequent reduction of the azide, followed by instant cyclization using TMSI (TMSCl/NaI) delivered the spiroaminoketal moiety, which, after ester hydrolysis (NaOH), produced the natural product 1 in a 5.9:1 ratio with the undesired 5,6-dihydrosolasodine 17. The obtention of 17 as a minor product is a consequent of the unperfect diastereoselectivity in the obtention of 2 and the uncontrolled inversion of the C 25 (S→R) during the TMSI-mediated reduction of 16 [40]. Purification by crystallization (vapor diffusion, see Supplementary Materials) generated crystals of tomatidine in a 9.0:1 ratio of 1:17 in 61% yield in two steps from azide 16 on a 10 g scale.

General Remarks
Overall, this sequence yielded 5.2 grams of tomatidine in a single batch, starting from 46 g of lactone 5 in 15.2% yield after 11 steps LLS (longest linear sequence). Another smaller-scale iteration of this sequence enabled us to produce a total of >7.5 grams of synthetic tomatidine in the laboratory. An analysis of a single crystal by X-ray diffraction (XRD) unequivocally confirmed the identity of 1 (Scheme 5) as its methanol co-crystal.

General Remarks
Unless otherwise stated, all reactions were performed using commercial reagents bought from Sigma-Aldrich (Oakville, Canada), TCI America (Portland, OR, USA), Combi-Blocks (San Diego, CA, USA), Chem-Impex (Bensenville, IL, USA) and Strem Chemicals (Newburyport, MA, USA) without further purification and Teflon-coated stir bars. Reactions that required heating above room temperature (23 • C) were heated in an oil bath. Silica pads were performed using SiliaFlash™ P60 40 to 63 µm thick silica gel from Silicycle. Thin layer chromatography (TLC) was performed on glass-backed Silicycle SiliaPlate F-254 precoated plates with a thickness of 250 µm and a porosity of 60 Å. Compounds were visualized with UV light (254 nm), followed by staining with cerium ammonium molybdate (steroids) or potassium permanganate (nonsteroids) and heating. Nuclear magnetic resonance (NMR) spectra were obtained on a Bruker Ascend 400 spectrometer (Bruker, Ettlingen, Germany) and are reported in parts per million (ppm) for chemical shifts and hertz (Hz) for coupling constants (J). Residual chloroform (CHCl 3 ) signals were used as an internal reference for the 1 H (δ = 7.26 ppm) and 13 C (δ = 77.16 ppm) spectra. The following abbreviations are used to describe the NMR multiplicities: br = broad, s = singlet, d = doublet, t = triplet, q = quartet, quint = quintet, sext = sextuplet, sept = septuplet, and m = multiplet. For the 1 H NMR data of steroid intermediates, only assignable signals are listed. Infrared (FTIR) spectra were obtained on an ABB Bomem MB104 spectrophotometer (ABB Inc., Québec City, Canada) using a diamond-attenuated total reflectance (ATR) accessory. High-resolution mass spectra (HRMS) were obtained on a Nexera (Shimadzu) LC-QTOF mass spectrometer (Shimadzu Scientific Instruments, Columbia, MD, USA) coupled to a maXis (Bruker) mass spectrometer (ESI) using sodium formate as an internal standard. Melting points were measured on an Electrothermal MEL-TEMP 1101D melting point apparatus (Mettler-Toledo, Portland, OR, USA). XRD was performed on a KAPPA APEX-DUO (Bruker, Karlsruhe, Germany) diffractometer using a Cu radiation source.
A 1 L, two-neck round-bottom flask was charged with Et 2 O (175 mL) and lithium aluminum hydride (8.00 g, 211 mmol, 1.0 eq.). A thermometer was fitted, and the suspension was vigorously stirred and cooled to 0 • C using an ice bath. To this suspension was added a solution of the THP ester prepared above in Et 2 O (200 mL + 50 mL wash) via canulation at a slow rate that kept the internal temperature below 15 • C. After the canulation was complete, the mixture was warmed to room temperature and stirred overnight. The mixture was then cooled back to 0 • C and quenched by careful addition of sodium sulfate decahydrate (50.9 g, 158 mmol, 0.75 eq.). The suspension was warmed to room temperature and stirred vigorously for 1 h. Subsequently, this suspension was filtered using a Büchner funnel and concentrated in vacuo. The obtained residue was distilled under reduced pressure to yield monoprotected diol 15 (29.6 g, 81% over two steps) as a colorless oil. 1 13
Under Ar atmosphere, a 250 mL round-bottom flask was charged with freshly purified [43] 1,2-diiodoethane (42.5 g, 151 mmol, 3.0 eq.) and dissolved in anhydrous THF (100 mL + 10 mL wash). This solution was then added to the lithiated enol ether flask by canulation over 15 min. After the addition, the cooling bath was replaced with a room temperature water bath, and the reaction was allowed to proceed for 1 h as the mixture slowly developed a deep purple color. Sat. Na 2 S 2 O 3(aq) (250 mL) was then added to the flask and stirred vigorously. After 15 min, this mixture was transferred to an extraction funnel containing water (300 mL). The organic layer was collected, and the aqueous layer was extracted with EtOAc (3 × 200 mL). The organic layers were combined, washed with brine (75 mL), dried over Na 2 SO 4 , and concentrated in vacuo. The product was passed through a silica pad (6 cm × 12 cm) using 15% EtOAc/hexanes (2 L) as an eluent to yield iodo enol ether 13 (24.7 g) as a light-yellow solid following concentration, which was used immediately in the next step. NMR analysis of the crude mixture showed 99% conversion of the starting material [44]. *NOTE: 13 was shielded from light at all times. 13 was found to be unstable even when stored under Ar (g) in a dark cold freezer after one week and needs to be consumed directly after it is synthesized. After being stored for 24 h in CDCl 3 in an NMR tube at room temperature, 13 underwent decomposition, and the solution in the tube turned to a dark purple color. 1 (7) A 1 L round-bottom flask was charged with 8 (12.8 g, 75.2 mmol, 1.5 eq.) and a stir bar. The flask was purged of air and moisture by alternately applying vacuum and Ar (g) (three cycles), and anhydrous THF (300 mL) was then added. A 9-borabicyclo[3.3.1]nonane solution (0.5 M in THF, 300 mL, 150 mmol, 3.0 eq.) was added dropwise over 30 min. The resulting mixture was stirred for 3 h at room temperature. Then, NaOH (aq.) (1 M, 150 mL, 150 mmol, 3.0 eq.) was added carefully, and the mixture was stirred for a further 30 min at room temperature to prepare intermediate alkylborane 6.
Meanwhile, a 3 L flask was charged with crude 13 and a stir bar, followed by a 3:1 THF/H 2 O mixture (500 mL). [1,1 -Bis(diphenyl-phosphino)ferrocene]dichloropalladium (II) complex with dichloromethane (8.19 g, 10.0 mmol, 0.2 eq.) was added in one portion and then stirred under Ar (g) for 10 min. The hydroboration reaction mixture was then canulated to the main reaction over 15 min in a final THF/H 2 O ratio of 3.5:1. After the transfer was complete, the mixture, which turned from bright red to dark brown upon addition of the alkylborane solution, was stirred at room temperature. After 5 h, a TLC analysis of the reaction showed full consumption of 13. Hexanes (800 mL) were then added, and the resulting suspension was filtered through a 1:1 Celite™/silica gel pad, which was further washed with 20% EtOAc/hexanes (300 mL). The filtrates were combined and added to an extraction funnel containing water (600 mL). The organic layer was collected, and the aqueous layer was extracted with EtOAc (1 × 200 mL). The organic layers were combined, washed with brine (150 mL), dried over Na 2 SO 4 , and concentrated in vacuo to obtain a brown oil. The crude material was passed through a silica pad (6 cm × 12 cm) using 20% EtOAc/hexanes (1.5 L) as an eluent. The filtrate was concentrated in vacuo to yield adduct 7 (26.1 g) as a light-brown pasty solid, which was used immediately in the next step. A small sample (35 mg) was purified by preparative TLC to collect analytical data. 1 (2) A 1 L round-bottom flask was charged with crude 7 and a stir bar, followed by addition of MeOH (390 mL). The solution was adjusted to pH 1 by careful addition of acetyl chloride (5.34 mL, 75.2 mmol, 1.5 eq.), and the resulting mixture was stirred at room temperature. After 30 h, a thick white precipitate was formed, and an NMR analysis of the crude mixture showed full conversion of the starting material and its intermediates. The crude mixture was then transferred to an Erlenmeyer, recrystallized directly from the reaction mixture, and filtered through a Büchner funnel to yield 2 (9.15 g, 44%, three steps) as a white solid with an 8.1:1 d.r. at C 25 *.
A 1 L, two-neck round-bottom flask was charged with crude OAc-2, lithium bromide (19.1 g, 220 mmol, 10.0 eq.), and a stir bar. This flask was purged of air and moisture by alternately applying vacuum and Ar (g) (three cycles)*. DCM (190 mL) and anhydrous acetonitrile (95 mL) were then added, and the mixture was stirred vigorously before adding BF 3 ·Et 2 O (27.2 mL, 220 mmol, 10.0 eq.) dropwise over 10 min at room temperature. After 3 h, sat. NaHCO 3(aq.) (50 mL) was slowly added, followed by water (200 mL). This mixture was transferred to an extraction funnel, the organic layer was collected, and the aqueous layer was extracted with DCM (3 × 100 mL). The organic layers were combined, washed with brine (40 mL), dried over Na 2 SO 4 , and concentrated in vacuo. The crude mixture was then transferred to a 250 mL round-bottom flask, dissolved in DMF (95 mL), and heated to 70 • C. After stirring for 2 h at this temperature, sodium azide (4.29 g, 66.1 mmol, 3.0 eq.) was added, and the mixture was stirred for further 3 h at 70 • C. Subsequently, the mixture was cooled to room temperature, transferred to an extraction funnel containing water (300 mL), and extracted with DCM (4 × 100 mL). The organic layers were combined, washed with brine (100 mL), dried over Na 2 SO 4 , and then concentrated in vacuo to yield 16 (9.91 g, 93% over three steps, >99% purity by NMR) as a beige waxy solid. This material was used directly in the next step without purification.
*NOTE: Glass vacuum adaptors were used to keep the mixture under Ar (g) during the reaction-BF 3 ·Et 2 O corroded and clogged regular needles. OAc