doi.org/10.26434/chemrxiv.14502744.v1

Development of an Efficient Route to 2-Ethynylglycerol for the Synthesisof IslatravirStephan M. Rummelt, Ji Qi, Yonggang Chen, James F. Dropinski, Gregory Hughes, Jeffrey T. Kuethe,Donghong Li, Kevin M. Maloney, Eric Margelefsky, Rose Mathew, Daniel J. Muzzio, Christopher C. Nawrat,Justin A. Newman, Honggui Ouyang, Niki R. Patel, Zhen Qiao, Gao Shang, Eric Sirota, Zhiguo Jake Song,Lushi Tan, Richard J. Varsolona, Baoqiang Wan, Brian M. Wyvratt, Feng Xu, Yingju Xu, Jingjun Yin,Shaoguang Zhang, Ralph Zhao

Submitted date: 28/04/2021 • Posted date: 29/04/2021Licence: CC BY-NC-ND 4.0Citation information: Rummelt, Stephan M.; Qi, Ji; Chen, Yonggang; Dropinski, James F.; Hughes, Gregory;Kuethe, Jeffrey T.; et al. (2021): Development of an Efficient Route to 2-Ethynylglycerol for the Synthesis ofIslatravir. ChemRxiv. Preprint. https://doi.org/10.26434/chemrxiv.14502744.v1

The unnatural, alkyne-containing nucleoside analog islatravir (MK-8591) is synthetically accessed through abiocatalytic cascade starting from 2-ethynylglycerol as a building block. Herein, we describe the developmentof an efficient synthesis of this building block including the initial route, route scouting and final processdevelopment. Key challenges that have been overcome are the development of an efficient and safeacetylenic nucleophile addition to an appropriate ketone, and the identification of a2-ethynylpropane-1,2,3-triol derivative with favorable physical properties. An acid-catalyzed cracking ofcommercially available 1,3-dihydroxyacetone dimer and subsequent 1,2-addition of an acetylenic nucleophilehas been discovered and optimized into the manufacturing process

File list (2)

download fileview on ChemRxivTMS Triol_26-Apr-2021.pdf (325.79 KiB)

download fileview on ChemRxivTMS Triol_SI_26-Apr-2021.pdf (0.92 MiB)

Development of an Efficient Route to 2-Ethynylglycerol for the Synthe-sis of Islatravir Stephan M. Rummelt‡, Ji Qi*,‡, Yonggang Chen‡, James F. Dropinski‡, Gregory Hughes‡, Jeffrey T. Kuethe‡, Donghong Li#, Kevin M. Maloney‡, Eric Margelefsky‡, Rose Mathew‡, Daniel J. Muzzio‡, Christopher C. Nawrat‡, Justin A. Newman‡, Honggui Ouyang§, Niki R. Patel‡, Zhen Qiao#, Gao Shang‡, Eric Sirota‡, Zhiguo Jake Song‡, Lushi Tan‡, Richard J. Varsolona‡, Baoqiang Wan§, Brian M. Wyvratt‡, Feng Xu‡, Yingju Xu‡, Jingjun Yin‡, Shaoguang Zhang‡, Ralph Zhao‡

‡ Department of Process Research and Development, Merck & Co., Inc., Rahway, New Jersey 07065, United States § Process R&D, WuXi AppTec Co., Ltd., 288 Fute Zhong Road, Shanghai 200131, China # Department of Synthetic Chemistry, Pharmaron Beijing Co., Ltd. 6 Taihe Road BDA, Beijing, 100176, China

ABSTRACT: The unnatural, alkyne-containing nucleoside analog islatravir (MK-8591) is synthetically accessed through a biocata-lytic cascade starting from 2-ethynylglycerol as a building block. Herein, we describe the development of an efficient synthesis of this building block including the initial route, route scouting and final process development. Key challenges that have been overcome are the development of an efficient and safe acetylenic nucleophile addition to an appropriate ketone, and the identification of a 2-ethynylpropane-1,2,3-triol derivative with favorable physical properties. An acid-catalyzed cracking of commercially available 1,3-dihydroxyacetone dimer and subsequent 1,2-addition of an acetylenic nucleophile has been discovered and optimized into the manu-facturing process.

INTRODUCTION The alkyne moiety is one of the two unnatural structural fea-

tures in the nucleoside analog islatravir (1, MK-8591) and en-ters the biocatalytic cascade of the manufacturing route in the form of 2-ethynylglycerol 2 (Scheme 1).1 Retrosynthetic analy-sis of this key building block mostly relied on 1,2-addition of a metalated alkyne into a 1,3-disubstituted ketone. While appear-ing straightforward at first sight, the most direct approach of ethynyl addition into 1,3-dihydroxyacetone (DHA) is compli-cated by the two protic primary hydroxy groups and the dimeric nature of the starting material. Furthermore, the ethynyl source needed to be chosen carefully, as acetylene is a highly explosive gas and its use or generation are highly undesirable in a phar-maceutical manufacturing process.2 The challenging physical properties of triol 2 combined with the purity requirements de-manded by the downstream biocatalytic cascade towards inor-ganic and organic impurities, constituted an additional key driver for route selection and development.1 Here, we describe our efforts to develop a short and efficient synthesis of triol 2 leading from a inital supply route, through route scouting, and to the manufacturing process.

Scheme 1. Biocatalytic Cascade for the Synthesis of Is-latravir

RESULTS AND DICUSSION Properties and Crystalline Forms of Triol 2. At the outset

of the project it was crucial to obtain information about the physical properties of triol 2. The biocatalytic cascade for the commercial synthesis of islatravir (1) starts with triol 2 as the key building block (Scheme 1). Triol 2 therefore marks the tran-sition from traditional synthetic chemistry to a biocatalytic re-action setup and thus occupies a strategically crucial position in the overall route. Hence, triol 2 or a suitable precursor should ideally be isolated in a form that exhibits physical properties suitable for storing and shipping, allowing flexibility in the sup-ply chain. A crystalline intermediate would also provide a con-trol point, which can ensure purity control before entering the biocatalytic cascade. Solid triol 2, obtained through crystalliza-tion at –20 °C, has a low melting point of 43 °C and deliquesces in air at ambient temperature (Figure 1). These unfavorable physical properties preclude the use of free triol 2 as an isolated intermediate. As triol 2 enters the biocatalytic cascade as a so-lution in water, intermediates that can be transformed to an aqueous solution of triol 2 in an operationally simple way would constitute a viable alternative. Due to the high solubility of triol 2 in water, efficient extraction from an organic phase into an

aqueous phase is feasible. At the same time, extraction of triol 2 from an aqueous into an organic phase was found inefficient, prohibiting any process that relies on a standard aqueous workup to remove water soluble byproducts in the final step.

Figure 1. Solid, crystalline forms and derivatives of triol 2

and their physical properties. mp = melting point; nd = not de-termined; na = not available; RH = relative humidity.

At the onset of our studies we aimed to identify alternative crystalline precursors of triol 2 (Figure 1). One strategy in-volved deprotonation of triol 2 and isolation of the correspond-ing alkoxide species. Addition of sodium methoxide to a solu-tion of triol 2 in isopropyl alcohol results in the crystallization of a triol sodium salt 3, which can be isolated with 91% yield. While no melting point of the so obtained crystalline material could be determined, differential scanning calorimetry (DSC) analysis revealed a sharp exotherm of 369 J/g with an onset tem-perature of 91 °C. At a relative humidity (RH) of 43% the ma-terial underwent a phase change as observed by powder x-ray diffraction (pXRD) and it ultimately turned deliquescent at 55% relative humidity. While the physical properties of the triol so-dium salt 3 at ambient conditions were improved compared to free triol 2, the strong exothermic behavior with a low onset temperature combined with the hygroscopicity are not ideal for storage, shipping and handling. Crystalline materials with sim-ilar unfavorable physical properties were also obtained when lithium or potassium salts were generated by a similar crystalli-zation process. (see SI for further details).

Alternatively, triol 2 could be isolated as DABCO co-crystal 4 (DABCO = 1,4-diazabicyclo[2.2.2]octane) after addition of DABCO (1.2 equiv) to a solution of triol 2 in THF in 93% yield. DABCO complex 4 has a melting point of 113 °C and an ex-otherm of 1208 J/g with an onset temperature of 130 °C as de-termined by DSC analysis. The pXRD spectrum showed no change up to a relative humidity of 58%. In a slurry in 1-propa-nol at room temperature DABCO co-crystal 4 converts to hemi-DABCO co-crystal 5. The hemi-DABCO co-crystal 5 melts at 95 °C as determined by DSC and therefore has a melting point 52 °C higher than that of triol 2 itself and 18 °C lower than that of the mono-DABCO co-crystal 4. DSC again revealed an ex-otherm of 1482 J/g with an onset temperature of 119 °C. Both DABCO co-crystals 4 and 5 were characterized by single crys-tal x-ray diffraction (see SI for details). An aqueous solution of free triol 2 with sufficient purity for the downstream biocata-lytic cascade could be accessed through treatment of an aqueous solution of the triol sodium salt 3 or DABCO co-crystals 4 and

5 with an acidic resin.1 Therefore these three solid forms of triol 2 could potentially be used as intermediates.

A different approach involved the isolation of a non-terminal alkyne that can be converted to parent triol 2 under mild reac-tion conditions. Crystalline TMS-triol 6 showed a melting point around 80 ºC, and exothermic decomposition starting around 226 °C with 1489 J/g energy by DSC analysis. Although 6 is a highly energetic compound, the exotherm onset temperature is quite high, and drop weight testing indicated that it is not impact sensitive at an energy level of 29 J. Because the SADT (self-accelerating decomposition temperature) for 50 kg was esti-mated to be greater than 75 °C, temperature control during ship-ping and storage is not required. Additionally, the purified solid is not hygroscopic as determined by DVS (dynamic vapor sorp-tion) analysis. TMS-triol 6 was also characterized using single crystal x-ray diffraction (see SI for details). Preliminary TMS deprotection condition screenings revealed that TMS-triol 6 could be readily converted to triol 2 under aqueous conditions at elevated temperature or mild basic conditions.3 Further de-velopment of this reaction is described in the following paper of this issue.4 An additional advantage of this approach pre-sented itself, as TMS-triol 6 could easily be extracted into or-ganic solvents, providing a way to remove inorganic and highly aqueous soluble impurities by organic/aqueous extraction.

It should be noted that solid TMS-triol 6 has a minimum ig-nition energy of <1 mJ with and without inductance. This means that the compound is extremely ignition sensitive and any elec-trostatic spark could ignite a dust cloud. Therefore, specific safety handling precautions have to be followed in order to han-dle the product safely.

In comparison with sodium salt 3 and DABCO co-crystal 4, the hemi-DABCO co-crystal 5 was slightly favored as it has a higher onset temperature for exothermic decomposition and is a more atom-economic alternative respectively. However, if synthetically accessible, TMS-triol 6 would be the preferred choice because of its improved physical properties including exotherm onset temperature and feasibility for organic/aqueous extraction.

Initial Route to Triol 2. As in previous syntheses of is-latravir (1),5 the initial route to triol 2 which was used to prepare the material necessary for development of the biocatalytic cas-cade installed the alkyne through 1,2 addition into a protected form of a 1,3-dihydroxyketone. Using 1,3-dihydroxyacetone di-acetate 7, prepared from acetyl protection of commerical dihy-droxyacetone, as starting material allowed introduction of the ethynyl group using commercially available ethynylmagnesium chloride and isolation of alkyne 8 after aqueous workup as an oil in 74% yield (Scheme 2). Deprotection of the hydroxy groups could be achieved by reacting diacetate 8 in refluxing methanol in the presence of the acidic resin Amberlite IR-120 to yield triol 2. The material obtained after treatment of the crude reaction stream with activated carbon, filtration and re-moval of all volatiles had a purity of 56 wt%. Triol 2 was further purified and isolated through crystallization as DABCO co-crystal 4, by treating a THF solution of crude triol 2 with DABCO.

Scheme 2. First Route for the Synthesis of Co-crystal 4

Efforts to streamline the synthesis focused on the use of tet-rahydropyran (THP) as protecting group, as deprotection can be performed under milder reaction conditions (Scheme 3). Start-ing from inexpensive and readily-available dihydroxyacetone dimer 9, THP protection could be achieved in 93% assay yield using 3,4-dihydropyran and catalytic pyridinium p-toluenesul-fonate (PPTS) in toluene at 35 °C. The crude reaction mixture was added to a slight excess of ethynylmagnesium chloride in THF at a temperature slightly below 0 °C to afford alkyne 10. After an aqueous workup and concentration to remove THF, the crude solution was reacted with ethanol in the presence of the acidic resin Amberlyst 15 at 70 °C to remove the THP groups. The crude reaction stream was purified through a carbon treat-ment, partial solvent switch to THF and subsequent isolation of the DABCO co-crystal 4 as outlined above. This sequence af-forded a yield of 47% from dihydroxyacetone dimer 9 without intermediate isolation and requiring one aqueous workup.

Scheme 3. Through Process to DABCO Co-crystal 4

Route Scouting for the Manufacturing Route of Triol 2. While the route described above was sufficient to ensure mate-rial supply for the development of the biocatalytic cascade, sev-eral issues remained that should ideally be addressed in the manufacturing route: (1) upon quench of the acetylenic Gri-gnard reagent used in the 1,2-addition, acetylene is generated. As acetylene is a highly explosive gas with a very low minimum ignition energy (MIE), a strict electrical code is required for its handling and venting in a manufacturing facility. Hence, acety-lene use and generation presents a major safety concern, and ideally would be avoided.2 (2) Due to the introduction and re-moval of the protecting groups for the hydroxyl group, the syn-thetic sequence consists of three chemical steps while only one carbon-carbon bond is formed. A shorter route, ideally forgoing hydroxy protection, was therefore highly desirable. (3) The iso-lation of triol 2 through formation of a highly crystalline triol-DABCO co-crystal 4 or 5 is a practical way to address the purity requirements and the lack of a crystalline intermediate through-out the route to triol 2. Nevertheless, salt formation and salt break operations are necessary from the crude triol 2 stream.

Therefore, a new route should ideally allow a more straightfor-ward way to an aqueous solution of triol 2, that meets the purity requirements of the subsequent biocatalytic cascade.

To address these challenges two main strategies were devised (Scheme 4). The first was based on the use of an alternative ketone starting material, which was monomeric and contained non-protic functional groups which could subsequently be con-verted to hydroxy groups, thereby allowing a direct 1,2-addition of an acetylide. To this end, oxetanone 11 and 1,3-dichloroace-tone 12 were investigated. Starting from oxetanone 11, triol 2 was successfully generated through a two-step procedure (Scheme 4a). Following a literature procedure,6 1,2-addition of ethynylmagnesium bromide into oxetane 11 afforded alkyne 13 as a crystalline solid in 98% yield. The oxetane was cleaved through subsequent treatment of alkyne 13 with the acidic resin Amberlite IR-120 in an aqueous solution, resulting in formation of the desired triol 2 in 82% yield. Despite the low step-count and high yield this route was deprioritized due to the high cost and low availability of the starting material 11 on commercial scale. Another route used 1,3-dichloroacetone 12 as starting material which was converted to alkyne 14 through 1,2-addition of a Grignard reagent obtained from TMS-acetylene (TMS = trimethylsilyl) and MeMgBr. Subsequent treatment of the prod-uct 14 with aqueous NaOH yielded triol 2 in moderate yield,7 which could be improved to >85% through phase transfer catal-ysis and the use of a milder carbonate base (Scheme 4b). Mech-anistic studies of the reaction from dichloride 14 to triol 2 showed that the desilylation of the alkyne happened first, pro-hibiting isolation of TMS-triol 6. The alkyl chloride groups were cleaved under the same reaction conditions through gen-eration of a series of epoxides and subsequent hydrolysis. From this reaction triol 2 could be isolated either as an aqueous solu-tion or as hemi-DABCO co-crystal 4. This route offers a high-yielding two-step access to the desired triol 2 while avoiding the use and generation of acetylene by employment of TMS-acetylene as alkyne source. However, the high toxicity and lach-rymatory nature of 1,3-dichloroacetone 12 prompted us to pur-sue another route.

Scheme 4. Alternative Routes to Triol 2

The second strategy targeted direct 1,2-addition of an alkyne into readily available 1,3-dihydroxyacetone dimer 9 (Scheme 4c). Our investigations of this approach started with the screen-ing a combination of bases and metal acetylides. For example, deprotonation of the hydroxyl groups with different bases in-cluding MeOK, EtOK, tBuOK, and LiHMDS in THF followed by addition of ethynylmagnesium bromide only yielded trace amounts of desired product 2. In addition, lithium, zinc, alumi-num and manganese acetylides were prepared and studied in the reaction with dimer 9. However, under all the conditions

screened, the desired triol product 2 was generated in less than 5% yield with the observation of significant decomposition of the starting material.

One of the main challenges for this approach was identified as the equilibrium between the dimer 9 and monomer 15 of 1,3-dihydroxyacetone, as 1,2-addition was thought to only success-fully occur on monomeric species 15. 1,3-Dihydroxyacetone can exist as a dimer or monomer in the solid state, although the latter slowly converts to the thermodynamically favored dimer.8 In our experience, commercial 1,3-dihydroxyacetone either consists of pure dimer or a monomer/dimer mixture even though it is sold as monomer. Dissolution of the dimer in protic solvents such as water and methanol shifts the solution equilib-rium towards the monomer and allows for its isolation, although it reverts to the dimer upon storage.9 However, these solvents prohibit the use of most organometallic 1,2-addition protocols. In order to deconvolute issues with the monomer/dimer equilib-rium from potential issues with a 1,2-addition on a substrate with two unprotected hydroxy groups, our early studies used isolated 1,3-dihydroxyacetone monomer 15 as starting material.

During high throughput experimentation (HTE) studies it was found that combinations of a dialkylzinc reagent with an amino alcohol or diamine ligand enabled addition of phenyla-cetylene into freshly prepared 1,3-dihydroxyacetone monomer 15. However, catalytic turnover could not be achieved and the reactions were stoichiometric in Zn(II) and ligand, rendering the approach unattractive. Further evaluation of the reaction was focused on the use of 1.5 equivalents of ethynylmagnesium bromide as the nucleophile and two equivalents of base to deprotonate the hydroxyl groups. The use of strong organic ba-ses (e.g. LDA, LiHMDS, NaHMDS, KHMDS, n-BuLi) led to significant decomposition. Detectable amounts of desired prod-uct triol 2 were observed in reactions with alkoxide bases (KOMe and NaOMe) or a sacrificial Grignard reagent (i-PrMgCl). However, significant decomposition of starting mate-rial was always observed under these conditions and the com-bined yield of starting material and desired product did not ex-ceed 25%. In contrast, the addition of 3 equiv. of ethynylmagne-sium bromide to dihydroxyacetone monomer 15 generated 63% yield of desired triol product 2, likely because the magnesium acetylide had the appropriate balance of nucleophilicity and ba-sicity to avoid decomposition experienced with other strong ba-ses. Similar reactivity could also be observed with TMS-ethynylmagnesium bromide, offering a potential to avoid the generation of acetylene and isolation of the more preferred TMS-triol 6.

Based on these results the route involving direct 1,2-addition into dihydroxyacetone 9 (Scheme 4c) was pursued for further development.

Development of the Dihydroxyacetone Route. As men-tioned above, proof of concept that direct 1,2-addition of an acetylenic Grignard into dihydroxyacetone monomer 15 is pos-sible, was achieved through the use of 3 equiv. of ethynyl-magnesium bromide as nucleophile and base to deprotonate the hydroxy groups, generating 63% yield of desired triol product 2 (Table 2, entry 1). Under the same conditions, reaction of di-hydroxyacetone monomer 15 with TMS-ethynylmagnesium bromide proceeded smoothly and generated TMS-triol 6 in 80% yield (Table 2, Entry 2). When dihydroxyacetone dimer 9 was used as starting material addition product TMS-triol 6 was ob-tained in 45% yield (Table 2 entry 4), while a reaction with ethynylmagnesium bromide did not yield appreciable amounts

of triol product 2. (Table 2, entry 3) It should be noted that re-actions with dimer 9 as starting material required longer reac-tion times when compared with reactions using monomer 15 as starting material.

Table 2. Acetylide Grignard addition to Dihydroxyacetone Dimer 9 and Monomer 15 a)

a) Reactions were carried out with addition of dihydroxy ace-tone dimer/monomer into Grignard reagents (1 M in THF, 3.0 equiv.) at –20 ⁰C in 6 batches over 1 h. The reactions were then warmed to 25 ⁰C and analyzed by quantitative-NMR.

Due to the direct accessibility from readily available dihy-droxyacetone dimer 9, the feasibility to avoid generation of acetylene, the favorable physical properties, and the apparent ease of converting it to triol 2, TMS-triol 6 was chosen as an isolated intermediate en route to islatravir (1) and further devel-opment work focused on optimizing its synthesis. The key areas of improvement were identified to be efficiency, cycle time, and cost.

As TMS-ethynylmagnesium halide was the main contributor to the cost of the reaction, in-situ generation protocols were evaluated. One approach involved the use of 2 equiv. of less-expensive Grignard reagents for the deprotonation of the hy-droxy groups, thereby reducing the amount of more-expensive TMS-ethynylmagnesium halide needed. It was found that reac-tions with 2 equiv of t-BuMgCl and only 1.5 equiv. of TMS-ethynylmagnesium bromide generated the desired product 6 in 44% yield, although at a much slower rate than with 3 equiv TMS-ethynylmagnesium bromide (120 h vs. 18 h). No desired product 6 was observed when i-PrMgCl and n-HexMgCl were used as bases. The TMS-ethynylmagnesium halide reagent could also be prepared directly from TMS-acetylene and t-BuMgCl giving TMS-triol 6 in moderate yields. Higher reac-tion temperatures (45 ºC) helped to improve the reaction rate while an age-time of the in-situ TMS-ethynylmagnesium chlo-ride generation greater than 10 h ensured consistent yields be-tween 70-75%. Using this optimized reaction protocol, a 40 kg scale reaction was successfully demonstrated and afforded 75% isolated yield, proving the robustness and scalability of the pro-cess. (Scheme 9)

Scheme 9. Pilot plant demonstration

Alternatively, TMS-ethynylmagnesium chloride could be generated in-situ from MeMgCl and TMS-acetylene. Although in this way three equivalents of TMS-acetylene were necessary, the higher concentration and better commercial availability of MeMgCl outweighed the higher cost and was ultimately chosen for the manufacturing route. It was also observed that the result-ing TMS-acetylenic Grignard reagent was stable for at least 7 days and performed without any diminished yield in the subse-quent chemistry.

The slow equilibrium between dihydroxyacetone monomer 15 and dimer 9 was still thought to be the main reason for the long reaction times, as 1,2-addition likely only occurs on ketone 15. Our efforts therefore focused on optimizing the cracking of dimer 9 during or before the reaction and we turned to screening of acid-catalyzed conversion to monomer 15. A series of acids were examined where it was found that utilizing 1-2 mol% TFA (TFA = trifluoroacetic acid) afforded the best results. Optimal conditions involved treating dimer 9 with 2 mol% TFA in 10 L/kg of THF at 60 ℃ for 1-2 h. Initially, the reaction starts out as a slurry and becomes homogeneous once the dimer 9 was converted to monomer 15. The solution of monomer is stable above 40 ℃ for 24 h but will start to crystallize and convert back to dimer below 30 ℃. Solutions of monomer 15 obtained in this way could directly be used in the 1,2-addition, allowing much shorter reaction times.

The conditions chosen for manufacturing involved treatment of 3.2 equivalents of TMS-acetylene relative to the monomer 15 with 3.15 equivalents of MeMgCl at 30 ℃ and aging for 2 h. The resulting solution was then added to a solution of mono-mer 15, prepared as described above from dimer 9 and catalytic amounts of TFA, followed by heating for 2-4 h at 55 ℃. This resulted in full conversion to product 6 as determined by GC analysis of a quenched sample. The isolation of TMS-triol 6 in-volved inversely quenching into an aqueous solution of 3.4 equivalents of 3 M aqueous AcOH and extraction with 10 L/kg IPAc. After washing the organic layer with an 18 wt% solution of brine, the mixture was azeotropically dried and concentrated to a final volume of 7 L/kg in IPAc which resulted in a slurry of the product. The internal temperature was adjusted to 45 ℃ and heptane (7 L/kg) was added over 4-5 h. The resulting slurry was cooled to room temperature, followed by filtration which af-forded a TMS-triol 6 in 80% isolated yield. Throughout the pro-cess acetylene was not detected in headspace of the reaction vessels by gas-phase Fourier-transform infrared (FT-IR) spec-troscopy. This process has safely and successfully been con-ducted on >100 kg scale at the commercial-scale manufacturing site. (Scheme 10).

Finally, TMS-triol 6 enters the biocatalytic cascade after re-moval of the TMS protecting group with catalytic Bis-Tris, providing triol 2 as an aqueous solution. Enzymatic phosphor-ylation then provides triol phosphate 16 in excellent conversion and ee.1

Scheme 10. Final Process

Concluding Remarks

This study reports the details behind the development of the synthesis to the 2-ethynyl glycerol building block 6. The route includes an in situ acid-catalyzed cracking of 1,3-dihydroxyace-tone dimer and subsequent 1,2-addition of TMS-ethynylmagnesium chloride. The resulting product 6 is a crys-talline and stable solid, addressing the concerns about unfavor-able physical properties of triol 2 and derivatives. The route was succefully demonstrated on scale and directly used in the down-stream biocatalytic cascade.

EXPERIMENTAL SECTION General Considerations. All reactions were carried out under a ni-

trogen atmosphere. All solvents and reagents were used as received from commercial sources.

In a vessel was charged 146.6 kg of THF followed by 57.7 kg (587.6 mol) of trimethylsilyl acetylene and the mixture was warmed to an in-ternal temperature of 35 ℃. To this solution was added 187.9 kg (559.8 mol) of 3M MeMgCl at a rate of 0.63 kg/min using a mass meter. The resulting mixture was stirred at 35 ℃ for 4 h and was cooled to 20 ℃. In a separate vessel was added 134.7 kg of THF, 16.8 kg (93.3 mol) of dihydroxyacetone dimer, and 213 g (1.87 mol) of TFA. The slurry was heated to an internal temperature of 60 ℃ for two hours and was cooled to 40 ℃ to give a homogeneous solution of dihydroxyacetone mono-mer. This solution was then added to the above Grignard solution at a maximum rate of 2.1 kg/min using a mass meter. The vessel was fol-low-flushed with 15 kg of THF. The mixture was stirred at 20 ℃ for four hours and inversely quenched into 213.2 kg of a 3 M aqueous so-lution of AcOH. To the quenched mixture was added 146.5 kg of IPAc, the bi-phasic mixture mixed for 30 min, and the layers were separated. The resulting organic solution was concentrated under reduced pres-sure of a final volume of ~ 135 L. To the solution was added 79.2 kg of IPAc and the distillation continued under reduced pressure to a vol-ume of ~ 135 L. This flush was repeated 2 X 70 L IPAc and the distil-lation stopped at a final volume of 135 L. The mixture was then warmed to 45 ℃ and 80.4 kg of heptanes was charged over 1 h. The mixture was then cooled to 20 ℃, the resulting slurry was filtered and the wet cake washed with 115 kg of heptanes. The product was dried under vacuum/N2 sweep for 38 h to provide 29.5 kg (84%) yield of TMS-triol 6 as a colorless solid. 1H NMR (400 MHz, DMSO-d6) δ 5.12 (s, 1H), 4.65 (t, J = 6.1 Hz, 2H), 3.46 – 3.35 (m, 4H), 0.12 (s, 9H).13C NMR (75 MHz, DMSO-d6) δ 0.2, 64.9, 71.5, 87.2, 109.1. HRMS MS (m/z): [M + Na]+ calcd for [C8H16O3Si Na] 211.0766; found 211.0766.

AUTHOR INFORMATION Corresponding Author* * ji_qi@merck.com

ACKNOWLEDGMENT We thank Paul G. Bulger for his feedback on the manuscript and input on the project. Anna Fryszkowska is thanked for help with biocatalytic reactions as well as feedback on the manuscript and input on the project. We thank L.-C. Campeau, Benjamin D.

Sherry, and Cheol K. Chung for their feedback on the manuscript. We thank François L. Levesque, Cecilia Bottecchia, and Timothy J. Wright for evaluating reactions in flow. We thank Mike Toth and Brittany Armstrong for experimental assistance with the safety evaluation. We thank Zhanqiang Guo, Yongwei Wu, Shunjia Zheng, Guiquan Liu, Jun Zhang, Jianfeng Zhan, Wanyong Fu, Minyi Feng, and Fang Wang for experimental help.

REFERENCES (1) Huffman, M. A.; Fryszkowska, A.; Alvizo, O.; Borra-Garske,

M.; Campos, K. R.; Canada, K. A.; Devine, P. N.; Duan, D.; Forstater, J. H.; Grosser, S. T.; Halsey, H. M.; Hughes, G. J.; Jo, J.; Joyce, L. A.; Kolev, J. N.; Liang, J.; Maloney, K. M.; Mann, B. F.; Marshall, N. M.; McLaughlin, M.; Moore, J. C.; Murphy, G. S.; Nawrat, C. C.; Nazor, J.; Novick, S.; Patel, N. R.; Rodriguez-Granillo, A.; Robaire, S. A.; Sherer, E. C.; Truppo, M. D.; Whittaker, A. M.; Verma, D.; Xiao, L.; Xu, Y.; Yang, H. Design of an in Vitro Biocatalytic Cascade for the Manufacture of Islatravir. Science 2019, 366 (6470), 1255–1259. https://doi.org/10.1126/science.aay8484.

(2) Schobert, H. Production of Acetylene and Acetylene-Based Chemicals from Coal. Chem. Rev. 2014, 114 (3), 1743–1760. https://doi.org/10.1021/cr400276u.

(3) Full conversion of TMS deprotection of 19 was observed in water at 70 °C for 3 h or with 10 wt% Ambersep® 900 at ambient temperature for 3 h.

(4) Further details on the phosphorylation will be described in a fu-ture publication.

(5) (a) Nawrat, C. C.; Whittaker, A. M.; Huffman, M. A.; McLaugh-lin, M.; Cohen, R. D.; Andreani, T.; Ding, B.; Li, H.; Weisel, M.; Tschaen, D. M. Nine-Step Stereoselective Synthesis of Islatravir from Deoxyribose. Org. Lett. 2020, 22 (6), 2167–2172. https://doi.org/10.1021/acs.orglett.0c00239. (b) Patel, N. R.; Nawrat, C. C.; McLaughlin, M.; Xu, Y.; Huffman, M. A.; Yang, H.; Li, H.; Whittaker, A. M.; Andreani, T.; Lévesque, F.; Fryszkowska, A.; Brunskill, A.; Tschaen, D. M.; Maloney, K. M. Synthesis of Islatravir Enabled by a Catalytic, Enantioselective Alkynylation of a Ketone.

Org. Lett. 2020, 22 (12), 4659–4664. https://doi.org/10.1021/acs.or-glett.0c01431. (c) Patel, N. R.; Huffman, M. A.; Wang, X.; Ding, B.; McLaughlin, M.; Newman, J. A.; Andreani, T.; Maloney, K. M.; John-son, H. C.; Whittaker, A. M. Five-Step Enantioselective Synthesis of Islatravir via Asymmetric Ketone Alkynylation and an Ozonolysis Cas-cade. Chem. – Eur. J. n/a (n/a). https://doi.org/10.1002/chem.202003091. (d) McLaughlin, M.; Kong, J.; Belyk, K. M.; Chen, B.; Gibson, A. W.; Keen, S. P.; Lieberman, D. R.; Milczek, E. M.; Moore, J. C.; Murray, D.; Peng, F.; Qi, J.; Reamer, R. A.; Song, Z. J.; Tan, L.; Wang, L.; Williams, M. J. Enantioselective Synthesis of 4′-Ethynyl-2-Fluoro-2′-Deoxyadenosine (EFdA) via En-zymatic Desymmetrization. Org. Lett. 2017, 19 (4), 926–929. https://doi.org/10.1021/acs.orglett.7b00091. (e) Kageyama, M.; Miyagi, T.; Yoshida, M.; Nagasawa, T.; Ohrui, H.; Kuwahara, S. Con-cise Synthesis of the Anti-HIV Nucleoside EFdA. Biosci. Biotechnol. Biochem. 2012, 76 (6), 1219–1225. https://doi.org/10.1271/bbb.120134. (f) Kageyama, M.; Nagasawa, T.; Yoshida, M.; Ohrui, H.; Kuwahara, S. Enantioselective Total Syn-thesis of the Potent Anti-HIV Nucleoside EFdA. Org. Lett. 2011, 13 (19), 5264–5266. https://doi.org/10.1021/ol202116k.

(6) Wang, Z.; Chen, Z.; Sun, J. Catalytic Enantioselective Intermo-lecular Desymmetrization of 3-Substituted Oxetanes. Angew. Chem. Int. Ed. 2013, 52 (26), 6685–6688. https://doi.org/10.1002/anie.201300188.

(7) Chênevert, R.; Simard, M.; Bergeron, J.; Dasser, M. Chemoen-zymatic Formal Synthesis of (S)-(−)-Phosphonotrixin. Tetrahedron Asymmetry 2004, 15 (12), 1889–1892. https://doi.org/10.1016/j.tetasy.2004.04.046.

(8) Davis, L. The Structure of Dihydroxyacetone in Solution. Bioor-ganic Chem. 1973, 2 (3), 197–201. https://doi.org/10.1016/0045-2068(73)90023-0.

(9) Insight into the Unexpectedly Rapid Degradation of Dihydroxy-acetone‐Based Hydrogels - Ricapito - 2016 - Macromolecular Chemis-try and Physics - Wiley Online Library https://onlineli-brary.wiley.com/doi/10.1002/macp.201600170 (accessed Oct 8, 2020).

download fileview on ChemRxivTMS Triol_26-Apr-2021.pdf (325.79 KiB)

S1

Supporting Information

Development of an Efficient Route to 2-Ethynyl Glycerol for the Synthesis of Islatravir

Stephan M. Rummelt‡, Ji Qi*,‡, Yonggang Chen‡, James F. Dropinski‡, Gregory Hughes‡, Jeffrey T. Kuethe‡, Donghong Li#, Kevin M. Maloney‡, Rose Mathew‡, Daniel J. Muzzio‡, Christopher C. Nawrat‡, Justin A. Newman‡, Honggui Ouyang§, Niki R. Patel‡, Zhen Qiao#, Gao Shang‡, Eric Sirota‡, Zhiguo Jake Song‡, Lushi Tan‡, Richard J. Varsolona‡, Baoqiang Wan§, Brian M. Wyvratt‡, Feng Xu‡, Yingju Xu‡, Jingjun Yin‡, Shaoguang Zhang‡, Ralph Zhao‡

‡ Department of Process Research and Development, Merck & Co., Inc., Rahway, New Jersey 07065, United States § Process R&D, WuXi AppTec Co., Ltd., 288 Fute Zhong Road, Shanghai 200131, China # Department of Synthetic Chemistry, Pharmaron Beijing Co., Ltd. 6 Taihe Road BDA, Beijing, 100176, China

ji_qi@merck.com

Table of Contents I. Synthesis and Solid-State Characterization of Crystalline Forms of Triol 2 ............................................. 2

II. Additional Synthetic Procedures .............................................................................................................. 6

III. Single Crystal X-Ray Data ...................................................................................................................... 8

V. NMR Spectra.......................................................................................................................................... 15

VI. References............................................................................................................................................. 16

S2

I. Synthesis and Solid-State Characterization of Crystalline Forms of Triol 2

Crystallization of triol 2. Crystalline triol 2 was obtained by storing purified triol 2 at –20 °C.

Figure S 1. Differential scanning calorimetry (DSC) of triol 2.

Crystallization of sodium-triol salt 3. To a stirred solution of triol 2 (40.0 g, 344 mmol, 1.0 equiv) in isopropyl alcohol (200 mL) at RT was added a solution of sodium methoxide in methanol (25 wt%, 158 mL, 689 mmol, 2.0 equiv) dropwise over 1 h. The resulting suspension was aged for 1 h at room temperature and filtered. The resulting solid was washed with isopropyl alcohol (160 mL) and 2-MeTHF (160 mL) and dried in a vacuum oven at 50 °C to yield sodium salt 3 as an off-white, crystalline solid in 91% yield (43.2 g, 313 mmol).

Figure S 2. DSC of sodium-triol salt 3.

S3

Crystallization of lithium-triol salt SI1. To a stirred solution of triol 2 (2.00 g, 17.2 mmol, 1.0 equiv) in isopropyl alcohol (10 mL) at RT was added a solution of lithium methoxide in methanol (2 M, 8.6 mL, 17.22 mmol, 1.0 equiv) over 5 min. The resulting suspension was aged for 1 h at room temperature and filtered. The resulting solid was washed with isopropyl alcohol and 2-MeTHF and dried in a vacuum oven at 40 °C to yield lithium salt SI1 as an off-white, crystalline solid in 90% yield (1.90 g, 15.57 mmol).

Figure S 3. DSC of lithium-triol salt SI1.

Crystallization of potassium-triol salt SI2. To a stirred solution of triol 2 (2.42 g, 20.8 mmol, 1.0 equiv) in isopropyl alcohol (10 mL) at RT was added a solution of potassium methoxide in methanol (25 wt%, 9.2 mL, 31.2 mmol, 1.5 equiv) dropwise over 1 h. The resulting suspension was aged for 1 h at room temperature and filtered. The resulting solid was washed with isopropyl alcohol and MTBE and dried in a vacuum oven at 40 °C to yield potassium salt SI2 as an off-white, crystalline solid in 81% yield (2.61 g, 16.93 mmol).

169.59°C

158.79°C480.6J/g

117.19°C

265.74°C

233.30°C198.2J/g

198.39°C

-1

1

3

Hea

t Flo

w (

W/g

)

25 75 125 175 225 275 325

Temperature (°C)

Sample: MK8591 Camp 0 Q108Size: 3.0000 mgMethod: Cell constant calibrationComment: Li complex 5000894-0116

DSCFile: T:...\Q108 5000894-0116 5-SEPT-18.001Operator: DCBRun Date: 05-Sep-2018 15:08Instrument: DSC Q100 V24.11 Build 124

Exo Up Universal V4.5A TA Instruments

S4

Figure S 4. DSC of potassium-triol salt SI2.

Crystallization of DABCO complex 4.1 To a stirred solution of triol 2 (95 wt%, 20.0 g, 164 mmol, 1.0 equiv) in THF (50 mL) at RT was added a solution of DABCO (22.0 g, 196 mmol, 1.2 equiv) in THF (50 mL) dropwise over 1 h. The resulting suspension was aged for 14 h at room temperature and filtered. The resulting solid was washed with 2-MeTHF (20 mL) and dried in a vacuum oven at 40 °C to yield DABCO complex 4 as an off-white, crystalline solid in 93% yield (34.7 g, 152 mmol).

Figure S 5. DSC of DABCO complex 4.

89.66°C

172.40°C

159.62°C1208J/g

129.75°C

108.73°C

105.05°C140.4J/g

-2

0

2

4

He

at F

low

(W

/g)

0 50 100 150 200 250 300 350

Temperature (°C)

Sample: Triol-Mono-DABCOSize: 5.5900 mgMethod: Standard DSC MethodComment: NB-0416528-0053-FS

DSCFile: TRIOL-MONO-DABCO Q1636 21-JUN-2019.00Operator: ARRun Date: 21-Jun-2019 11:07Instrument: DSC Q200 V24.11 Build 124

Exo Up Universal V4.5A TA Instruments

S5

Crystallization of hemi-DABCO complex 5. A 100 mL reaction vessel equipped with an overhead stirrer was charged with a solution of triol 2 (4.00 g, 34.5 mmol, 1.0 equiv) in 1-propanol (10 mL) and the solution was heated to 30 °C. DABCO (1.16 g, 10.3 mmol, 0.3 equiv) was added causing precipitation of a white solid after ca. 5 min and the resulting mixture was aged for 1.5 h. A second portion of DABCO (1.16 g, 10.3 mmol, 0.3 equiv) was added and the resulting mixture was aged for 1 h. Then heptane (24 mL) was added over 4 h and the resulting suspension was cooled to 15 °C over 1 h and aged at that temperature for 14 h. After filtration, the solid was washed with heptane (20 mL) and dried in a vacuum oven at 40 °C to yield hemi-DABCO complex 5 as an off-white, crystalline solid in 82% yield (4.89 g, 14.1 mmol).

Figure S 6. DSC of hemi-DABCO complex 5.

82.58°C

119.14°C

166.38°C

151.34°C1482J/g

95.18°C

93.67°C177.8J/g

-4

-2

0

2

4

Hea

t Flo

w (

W/g

)

0 50 100 150 200 250 300 350

Temperature (°C)

Sample: Triol-Hemi-DABCOSize: 5.2200 mgMethod: Standard DSC MethodComment: NB-0416528-0135-FS

DSCFile: Q1571 TRIOL-HEMI-DABCO 21-JUN-2019.001Operator: ARRun Date: 21-Jun-2019 11:09Instrument: DSC Q200 V24.11 Build 124

Exo Up Universal V4.5A TA Instruments

S6

DSC Data for TMS-Triol 6.

Figure S 7. DSC of TMS-triol 6.

II. Additional Synthetic Procedures

Hydrolysis of oxetane 13 to triol 2. To a solution of oxetane 132 (780 mg, 8.0 mmol, 1.0 equiv) in water (6 mL) was added Amberlite IR120 acidic resin (1.1 g) and the resulting mixture was stirred for 18 h at 60 °C. The resulting mixture was allowed to cool to RT filtered and the resin was rinsed with water yielding an aqueous solution of triol X in 82 % yield (7.25 g, 10 wt% as determined by quant. 1H NMR against an internal standard, 6.52 mmol). The identity was confirmed through direct comparison with an authentic sample by 1H NMR spectroscopy.1 1H NMR (400 MHz, D2O): δ 3.72 (d, J = 11.6 Hz, 2H), 3.67 (d, J = 11.5 Hz, 2H), 2.95 (s, 1H).

Through process from DHA dimer 9 via THP-protection to DABCO complex 4. A flask was charged with dihydroxy acetone dimer 9 (20.0 g, 111.0 mmol, 1.0 equiv) and toluene (100 mL). To the resulting stirred slurry was added PPTS (0.70 g, 2.8 mmol, 2.5 mol%) followed by 3,4-dihydro-2H-pyran (60.4 mL, 666 mmol, 6.0 equiv) and the resulting mixture was stirred at 33-35 °C for 20 h. To the mixture was added another portion of PPTS (0.70 g, 2.8 mmol, 2.5 mol%) and 3,4-dihydro-2H-pyran (18.0 mL, 222 mmol, 2.0 equiv) and the resulting mixture was stirred at 33-35 °C for 24 h. The reaction stream was taken directly into the next step without further purification. A flask equipped with a stirrer was charged with a solution of ethynyl magnesium chloride in THF (0.5 M, 497 mL, 249 mmol, 1.2 equiv) and cooled to –5 °C. The toluene solution obtained above was slowly transferred into the stirred ethynyl magnesium chloride solution, maintaining an internal temperature below 0 °C. The resulting solution was stirred for 3 h, after which full consumption of starting material was observed by thin-layer chromatography (TLC). Sat. NH4Cl aq. (400 mL) and toluene (50 mL) was added to the reaction mixture and the layers were separated. The aqueous layer was extracted with toluene (100 mL) and the combined organic phases were concentrated to a volume of ~325 mL. The resulting solution

S7

was washed with sat. NH4Cl aq. until the aqueous washing solution was slightly acidic (three times: 200, 150 and 100 mL). To the resulting solution was added acidic Amberlyte IR-120 (prewashed with EtOH, 14.7 g, 25 w%) and ethanol (177 mL) and the stirred reaction mixture was heated to 70 °C for 4 h, after which analysis by TLC indicated full consumption of the starting material. The resulting solution was allowed to cool to RT and charcoal (14.7 g, 25 w%) was added. After stirring at RT for x h the resulting mixture was filtered through Celite and concentrated to a volume of ~62 mL. To the resulting stirred solution was added a solution of DABCO (34.9 g, 311 mmol, 1.5 equiv) in THF (250 mL) dropwise. During the addition a white precipitate formed. After the addition, the resulting mixture was stirred for 30 min at RT and the white precipitate was collected by filtration. The filter cake was washed with 2Me-THF (150 mL) and dried under a nitrogen stream to yield DABCO complex 4 in 47% yield (24.2 g, 106 mmol) as a white crystalline solid. The identity was confirmed through direct comparison with an authentic sample by 1H NMR spectroscopy.1 1H NMR (400 MHz, D2O, 25 °C): δ 3.70 (d, J = 11.5 Hz, 2H), 3.66 (d, J = 11.5 Hz, 2H), 2.78 (s, 12H). Synthesis of triol 2 from 1,3-dichloroacetone. A 3-necked 250 mL flask equipped with a temperature probe, a reflux condenser and a magnetic stir bar was charged with THF (32 mL) and a solution of methyl magnesium chloride in THF (3.0 M, 23.1 mL, 69.3 mmol, 1.1 equiv) and the resulting stirred mixture was heated to 30 °C. TMS-acetylene (10.7 mL, 76 mmol, 1.2 equiv) was added dropwise over 2 h. The resulting mixture was stirred for 2 h and cooled to RT with a water bath. To the resulting solution was added a solution of 1,3-dichloroacetone (8.0 g, 63 mmol, 1.0 equiv) in THF (16 mL) over 5 h. The resulting mixture was stirred for 0.5 h before it was slowly transferred into a cooled 1 M aqueous solution of citric acid (32 mL, 32 mmol, 0.5 equiv) maintaining the internal temperature below 25 °C. To the resulting mixture was added heptane (32 mL) and the phases were separated. The organic layer was concentrated to obtain crude alkyne 13 in 93% yield (14.65 g, 90.2 wt% as determined by quant. 1H NMR spectroscopy against an internal standard, 58.7 mmol) as a dark orange oil. 1H NMR (500 MHz, CDCl3, 25 °C): δ 3.78 (d, J = 11.2 Hz, 2H), 3.75 (d, J = 11.3 Hz, 2H), 0.19 (s, 9H). 13C{1H} NMR (126 MHz, CDCl3, 25 °C): δ 101.9, 92.7, 70.5, 48.8, -0.3. A 100 mL Easymax reactor equipped with a temperature probe and an overhead stirrer was charged with a solution of crude dichloride 13 (20.1 g, 91 wt%, 81 mmol, 1.0 equiv) in heptane (18 mL), an aqueous solution of K2CO3 (3 M, 54 mL, 162 mmol, 2.0 equiv) and (oct)4NBr (444 mg, 0.81 mmol, 1 mol%) and the resulting mixture was stirred (500 rpm) for 20 h at RT. The phases were separated and an aqueous solution of K2CO3 (3 M, 17 mL, 51 mmol, 0.6 equiv) was added to the organic phase. The resulting mixture was heated to 65 °C and stirred (500 rpm) for 30 h at that temperature. The reaction mixture was allowed to cool to RT and filtered. The phases were separated, and the aqueous layer was analyzed by quant. 1H NMR spectroscopy to determine an assay yield of 83% (26.98 g, 28.9 wt%, 67.1 mmol) for the desired triol 2. Nitrogen was bubbled through the resulting aqueous solution and conc. HCl aq. was added slowly until a stable pH = 3.5 was reached. The mixture was stirred for 30 min at RT before 2 N KOH aq. was added until a pH = 7 was reached. To this solution with a total volume of ~25 mL was added isopropyl alcohol (150 mL) and the resulting slurry was stirred at 4 °C for 20 h. After filtration, the isopropyl alcohol was removed on a rotary evaporator to obtain triol 2 as an aqueous solution in 81% yield (31.17 g, 24.0 wt% as determined by quant. 1H NMR spectroscopy against an internal standard, 64.6 mmol). The identity was confirmed through direct comparison with an authentic sample by 1H NMR spectroscopy.1 1H NMR (400 MHz, D2O, 25 °C): δ 3.60 (d, J = 11.6 Hz, 2H), 3.55 (d, J = 11.6 Hz, 2H), 2.84 (s, 1H). 13C{1H} NMR (101 MHz, D2O, 25 °C): δ 83.0, 75.6, 71.7, 65.2. Dihydroxyacetone monomer 15 preparation.3 To 400 mL of 2-propanol was added 12 g of DHA dimer and stirred for 70 minutes, while partially immersed in a 60 °C oil bath. The resulting solution was filtered and solvent concentration by roto-evaporation at 50-55 °C over 95 minutes to remove 100 mL of 2-propanol. The remaining solution was held at -20 °C overnight to crystallize dihydroxyacetone in the monomeric form. The crystallized product was recovered by filtration and dried under vacuum at room

S8

temperature to yield 6.77 g of white crystalline in 56.4% yield. 1H NMR (400 MHz, DMSO-d6) δ 5.020 (t, J = 5.9 Hz, 1H), 4.159 (d, J = 5.9 Hz, 2H). 13C NMR (101 MHz, CDCl3) δ 216.6, 70.8. TMS triol preparation with t-BuMgCl as base. Charge tert-butylmagnesium chloride (8.23 L, 1.7 M,14 mol) into a 20-L reactor equipped with an overhead stir and inert atmosphere of nitrogen at 25-30 °C. Ethynyltrimethylsilane (589 g, 6 mol) was charged at 25-30°C in 30 mins, keep the reaction temperature at 25-30°C and stirred for 16 h. 2,5-bis(hydroxymethyl)-1,4-dioxane-2,5-diol (360 g, 2 mol) was charged into the reaction mixture in 6 portions (45 min for each) and keep reaction temp at 25-30°C. After addition, reaction mixture was warmed up to 45°C and aged for 66 hours. Gas release was found and N2 sweep was applied. The reaction mixture was cooled to 25°C. The reaction system was added slowly into a biphasic mixture of acetic acid (964.8 g, 8 eq) in 20 V 5% NaCl (7200 mL) and 10 V IPAc(3600 mL) at 0±5 °C over 5.5 hours. Keep the internal temperature of quenching solution at 0±5°C. During quench, small amount of gas emission was observed. N2 sweep was applied. Stirred the mixture for 1h at 0±5°C, pH of aq. phase was ~5.50. Phase cut. The organic phase was concentrated under reduced pressure to about 5000 ml and solvent switch to IPAc (~3600 mL X 3) at 50 °C azetropic to remove THF and water. Control 2% THF residue and KF keep at 0.08%. Heat the IPAc solution (~5760 mL) to 50°C for 30 min. Then n-heptane (8640 mL) was charged dropwise to the mixture over 1.5 h at 50°C. Product was precipitated over the addition. Slowly cool down of the mixture to 25°C over 3 hour and stirred for 16 h. 2-((trimethylsilyl)ethynyl)propane-1,2,3-triol (530 g, 2815 mmol, 70.4 % yield) was collected by filtration as white crystalline solid after drying at 40±5 °C. 1H NMR (400 MHz, DMSO-d6) δ 5.135 (s, 1H), 4.661 (t, J = 6.1 Hz, 2H), 3.463 – 3.346 (m, 4H), 0.134 (s, 9H). 13C NMR (75 MHz, DMSO) δ 0.6, 65.3, 71.9, 87.6, 109.6. HRMS MS (m/z): [M + Na]+ calcd for [C8H16O3Si Na] 211.0766; found 211.0766.

III. Single Crystal X-Ray Data

Crystal Data and Structure Refinement for Compound 4 (CCDC 2044297) A single crystal grown by cooling a 42°C saturated 1:3 1-PrOH/heptane solution to room temperature was selected for single crystal X-ray data analysis. The crystal was a plate with dimensions of 0.14 mm x 0.11 mm x 0.05 mm. Data collection was performed on a Bruker Apex II system at 100K. The unit cell was determined to be triclinic in space group P-1. The structure contained one molecule of triol and one molecule of DABCO in the crystallographic asymmetric unit. Crystallographic data is summarized in Table S1. Figure S8 shows a thermal ellipsoid representation of Compound 4 with thermal ellipsoids set at the 50% probability level. Coordinates, refinement details and structure factors have been deposited with the Cambridge Crystallographic Data Centre (CCDC 2044297).

S9

Figure S 8. Thermal ellipsoid representation of Compound 4 with thermal ellipsoids set at the 50% probability level.

S10

Table S1. Crystal Data and Structure Refinement for Compound 4 [CCDC 2044297] Identification code mdo034 Empirical formula C11H20N2O3 Formula weight 228.29 Temperature 100.15 K Radiation CuKα (λ = 1.54178) Crystal system triclinic Space group P-1 Unit Cell Dimensions a = 12.4218(4) Å α = 81.3420(10)° b = 12.9239(4) Å β = 82.2980(10)° c = 16.6033(6) Å γ = 81.3420(10)° Volume 2364.85(14) (Å3) Z 8 Density 1.282 (g/cm3) Absorption coefficient 0.765 mm-1 F(000) 992.0 Crystal size 0.14 × 0.11 × 0.05 mm3 2Θ range for data collection 5.4° to 141.382° Index ranges -15 ≤ h ≤ 14, -15 ≤ k ≤ 15, -19 ≤ l ≤ 19 Reflections collected 62563 Independent reflections 8485 [Rint = 0.0461, Rsigma = 0.0281] Completeness to Θ=67.679° 0.981 Absorption correction multi-scan Max. and min. transmission 0.7474 and 0.6715 Refinement Method Full-matrix least-squares on F2 Data/restraints/parameters 8485/0/589 Goodness-of-fit on F2 1.071 Final R indexes [I>=2σ (I)] R1 = 0.0438, wR2 = 0.1342 Final R indexes [all data] R1 = 0.0491, wR2 = 0.1419 Largest diff. peak/hole 0.42/-0.26 eÅ-3

S11

Crystal Data and Structure Refinement for Compound 5 (CCDC 2044296) A single crystal grown by cooling a 30°C saturated 1-PrOH solution to room temperature was selected for single crystal X-ray data analysis. The crystal was a block with dimensions of 0.12 mm x 0.12 mm x 0.10 mm. Data collection was performed on a Bruker Apex II system at 100K. The unit cell was determined to be orthorhombic in space group Pca21. The structure contained two triol molecules and one DABCO molecule in the crystallographic asymmetric unit. Crystallographic data is summarized in Table S2. Figure S9 shows a thermal ellipsoid representation of Compound 5 with thermal ellipsoids set at the 50% probability level. Coordinates, refinement details and structure factors have been deposited with the Cambridge Crystallographic Data Centre (CCDC 2044296).

Figure S 9. Thermal ellipsoid representation of Compound 5 with thermal ellipsoids set at the 50% probability level.

S12

Table S2. Crystal Data and Structure Refinement for Compound 5 [CCDC 2044296] Identification code mdo030 Empirical formula C16H28N2O6 Formula weight 344.40 Temperature 100(2) K Radiation CuKα (λ = 1.54178) Crystal system orthorhombic Space group Pca21 Unit Cell Dimensions a = 23.7690(5) Å α = 90° b = 7.8559(2) Å β = 90° c = 9.7538(2) Å γ = 90° Volume 1821.30(7) (Å3) Z 4 Density 1.256 (g/cm3) Absorption coefficient 0.797 mm-1 F(000) 744.0 Crystal size 0.12 × 0.12 × 0.1 mm3 2Θ range for data collection 7.438/° to 133.188° Index ranges -28 ≤ h ≤ 27, -9 ≤ k ≤ 8, -11 ≤ l ≤ 11 Reflections collected 11344 Independent reflections 3142 [Rint = 0.0245, Rsigma = 0.0209] Completeness to Θ=66.594° 0.999 Absorption correction multi-scan Max. and min. transmission 0.7528 and 0.7101 Refinement Method Full-matrix least-squares on F2 Data/restraints/parameters 3142/1/223 Goodness-of-fit on F2 1.071 Final R indexes [I>=2σ (I)] R1 = 0.0237, wR2 = 0.0647 Final R indexes [all data] R1 = 0.0239, wR2 = 0.0650 Flack parameter 0.13(4) Largest diff. peak/hole 0.17/-0.15 eÅ-3

S13

Crystal Data and Structure Refinement for Compound 19 (CCDC 2044962) A single crystal grown from 1:1 acteone/water by solvent evaporation was selected for single crystal X-ray data analysis. The crystal was a block with dimensions of 0.10 mm x 0.09 mm x 0.06 mm. Data collection was performed on a Bruker Apex II system at 100K. The unit cell was determined to be triclinic in space group P-1. The structure contained one molecule in the crystallographic asymmetric unit. The hydroxy hydrogens exhibit disorder (disorder removed from Figure S10 for clarity) Crystallographic data is summarized in Table S3. Figure S10 shows a thermal ellipsoid representation of Compound 19 with thermal ellipsoids set at the 50% probability level. Coordinates, refinement details and structure factors have been deposited with the Cambridge Crystallographic Data Centre (CCDC 2044962).

Figure S 10. Thermal ellipsoid representation of Compound 6 with thermal ellipsoids set at the 50% probability level.

S14

Table S3. Crystal Data and Structure Refinement for Compound 19 [CCDC 2044962] Identification code mdo039 Empirical formula C8H16O3Si Formula weight 188.30 Temperature 100.15 K Radiation CuKα (λ = 1.54178) Crystal system triclinic Space group P-1 Unit Cell Dimensions a = 5.312(5) Å α = 97.15(3)° b = 6.200(5) Å β = 97.40(3)° c = 16.454(18) Å γ = 97.15(3)° Volume 519.2(9) (Å3) Z 2 Density 1.204 (g/cm3) Absorption coefficient 1.776 mm-1 F(000) 204.0 Crystal size 0.1 × 0.09 × 0.06 mm3 2Θ range for data collection 5.484° to 134.154° Index ranges -6 ≤ h ≤ 6, -7 ≤ k ≤ 7, -19 ≤ l ≤ 19 Reflections collected 11990 Independent reflections 1852 [Rint = 0.0271, Rsigma = 0.0149] Completeness to Θ=67.077° 0.995 Absorption correction multi-scan Max. and min. transmission 0.7528 and 0.6790 Refinement Method Full-matrix least-squares on F2 Data/restraints/parameters 1852/0/118 Goodness-of-fit on F2 1.107 Final R indexes [I>=2σ (I)] R1 = 0.0490, wR2 = 0.1291 Final R indexes [all data] R1 = 0.0502, wR2 = 0.1299 Largest diff. peak/hole 0.56/-0.22 eÅ-3

S15

V. NMR Spectra

Figure S 11. 1H NMR spectrum of TMS-triol 6.

Figure S 12. 13C NMR spectrum of TMS-triol 6.

S16

VI. References

[1] Huffman, M. A.; Fryszkowska, A.; Alvizo, O.; Borra-Garske, M.; Campos, K. R.; Canada, K. A.; Devine, P. N.; Duan, D.; Forstater, J. H.; Grosser, S. T.; Halsey, H. M.; Hughes, G. J.; Jo, J.; Joyce, L. A.; Kolev, J. N.; Liang, J.; Maloney, K. M.; Mann, B. F.; Marshall, N. M.; McLaughlin, M.; Moore, J. C.; Murphy, G. S.; Nawrat, C. C.; Nazor, J.; Novick, S.; Patel, N. R.; Rodriguez-Granillo, A.; Robaire, S. A.; Sherer, E. C.; Truppo, M. D.; Whittaker, A. M.; Verma, D.; Xiao, L.; Xu, Y.; Yang, H. Design of an in Vitro Biocatalytic Cascade for the Manufacture of Islatravir. Science 2019, 366 (6470), 1255–1259. [2] Wang, Z.; Chen, Z.; Sun, J. Catalytic Enantioselective Intermolecular Desymmetrization of 3-Substituted Oxetanes. Angew. Chem. Int. Ed. 2013, 52 (26), 6685–6688. [3] Davis, L. The Structure of Dihydroxyacetone in Solution. Bioorganic Chem. 1973, 2 (3), 197–201.

download fileview on ChemRxivTMS Triol_SI_26-Apr-2021.pdf (0.92 MiB)