Highly Regioselective Protecting-Group-Free Synthesis of the Antimalarial Drug MMV693183

MMV693183 is a promising antimalarial drug candidate that works for uncomplicated malaria treatment and resistance management. Herein, we report an efficient and highly regioselective synthesis of MMV693183. This novel synthetic method highlights a three-step route with an overall yield of 46% from readily available starting materials. The key to the success lies in (1) utilizing the subtle difference of the two amino groups in the starting material (S)-propane-1,2-diamine dihydrochloride without amino protection and (2) identifying the L-(+)-tartaric acid as the counter acid for the organic salt formation, yielding the desired regioisomer up to 100:0. The efficient and scalable three-step protocol operates under mild conditions with a high chemo/regioselectivity, providing effective access to MMV693183.


■ INTRODUCTION
−5 Pantothenic acid (vitamin B5) (Figure 1) is an important precursor to the enzyme cofactor coenzyme A (CoA), on which the predominant pathogen for Malaria, Plasmodium falciparum, is dependent during the intraerythrocytic stage of its life cycle. 6n the last few decades, many analogues of pantothenic acid have been synthesized that hinder pantothenic acid utilization and thus block the parasite life cycle. 7−17 Medicines for Malaria Venture (MMV) has developed the inverted pantothenamide MMV693183 (Figure 1) as a single dose treatment for uncomplicated malaria and resistance management.Developing a cost-effective process for the synthesis of MMV693183 will make the therapy more affordable and likely increase its impact.Unfortunately, only one synthetic route for MMV693183 has been published so far, and this route would be quite limiting to employ as a production route with a low price point in mind (Scheme 1). 14his route started with a Mitsunobu reaction of Cbz-protected aminoalcohol 1 with phthalimide to provide 2 in 65% yield, which was immediately subjected to phthalimide deprotection to provide the monoprotected diamine 3 in 96%.The resulting Cbz-protected diamine 3 was then reacted with (R)pantolactone (4) to afford the diol 5 in 74% yield, which was then protected by 2,2-dimethoxypropane to provide acetonide 6 in 63% yield.Cbz deprotection by hydrogenolysis provided the free amine 7 in quantitative yield, which allowed for selective acylation of the amine with 2,4,5-trifluorobenzoic acid (8) to provide amide 9 in 78% yield.Finally, the acetonide protecting group was removed to furnish MMV693183 in 61% yield (14% overall yield).
While this 7-step sequence was successfully employed to make MMV693183 on a decagram scale, it also offers several opportunities for improvement.For example, the overall yield was only ∼14% and 3 of the 7 steps were used to manipulate protecting groups.The introduction of the less sterically hindered primary amine was accomplished via the Mitsunobu reaction and subsequent hydrazine deprotection, which are challenging transformations to scale-up due to their inherent wastefulness, cost, and the safety risks associated with handling diazodicarboxylates and hydrazine at scale.Thus, a more efficient and scalable route is needed for the synthesis of MMV693183 that would accommodate cost-effective commercial implementation and maximize access to this drug should it become commercially available.

■ RESULTS AND DISCUSSION
Herein, we report a three-step scalable synthesis of MMV693183 using readily available and low-cost starting materials, which avoids the use of any protecting group. 18Our approach is based on the hypothesis that the steric differences of both primary amines in (S)-1,2-diaminopropane dihydrochloride (10) would alone be sufficient to direct acylation to the desired less hindered amine (in blue) in a regioselective fashion (Table 1).To test this hypothesis, diamine 10 (readily available by resolution of the corresponding racemic diamine) 19−21 was reacted directly with (R)-pantolactone (4) in the presence of 3 equiv of base (Table 1, entry 1).In this initial reaction, we observed good reactivity of the starting diamine; however, the mixture of amide products formed in the reaction was difficult to quantify and characterize.
In order to deconvolute the reaction mixture, we next discretely synthesized compounds 11′ and 12′ to utilize as analytical standards (Scheme 2).These were prepared by the reaction of (R)-pantolactone (4) with both Boc-protected amines 14 and 16 followed by Boc deprotection to provide pure 11′ and 12′ as their HCl salts.
Having the standards in hand allowed for the development of an HPLC method for the identification of the ratio of products in each direct amidation reaction (Table 1).A reversed-phase HPLC method was developed that was capable of separating the diamide products 13 from monoamide products 11′ and 12′, but it could not competently separate the regioisomers 11′ and 12′.A separate hydrophilic interaction liquid chromatography (HILIC) method, however, was able to separate regioisomers 11′ and 12′, so a combination of HPLC and HILIC methods was used to characterize the reaction mixtures formed.
The initial reaction (Table 1, entry 1) with 3.0 equiv of Na 2 CO 3 provided 52% assay yield (by HPLC area %) of a 9:1 regioisomeric mixture of monoamide products (11 and 12)  favoring the desired product.It was also observed that 8% of the diacylated side product 13 formed in this reaction.For further optimization, a systematic solvent screening was conducted utilizing Na 2 CO 3 as the base (Table 1, entries 2− 8).It was determined that the reaction occurred only in polar protic solvents, such as MeOH, EtOH, and i-PrOH (Table 1, entries 1−3), but no reaction was observed in polar aprotic solvents probably due to insolubility of the inorganic base  1, entries 4−7).In i-PrOH, the reaction was significantly slower, however, the addition of water to the solvent system (entries 9−15) gave superior results and up to 86% assay yield (by HPLC area %) of the monoamide products (Table 1, entry 8).Notably, the regioisomeric ratio was consistently 9:1 in favor of the desired isomer, regardless of the conditions screened.Due to the incomplete solubility of all species, 10 V of solvent was necessary to ensure proper

Organic Process Research & Development
mixing of the reaction mass as the reaction mixture in 5 V conditions (Table 1, entry 9) was difficult to stir.More dilute conditions (20 V of solvent) gave the same results as 10 V of the solvent (Table 1, Entry 10), so 10 V was deemed ideal.With this optimized solvent system in hand, various bases were screened (Table 1, entries 11−13).Among these different bases, Na 2 CO 3 and Et 3 N offered the best results; however, Et 3 N proved difficult in the workup, presumably due to the formation of the Et 3 N hydrochloride salt.Notably, the mixed solvents of i-PrOH/H 2 O (7:3) worked well; however, the product isolated by column chromatography from this mixed solvent system was of relatively low purity.The product was contaminated with inorganic salts, likely due to the high water content in the reaction mixture leading to isolation of the water-soluble product mixture along with significant amounts of dissolved inorganic salts.Decreasing the amount of water in the reaction from 7:3 to 9:1 (i-PrOH/H 2 O) allowed for effective isolation of the product with a higher isolated yield and purity (11 + 12 86% yield and >95%, qNMR purity) (Table 1, entry 14).Ultimately, the optimal conditions for this reaction (Table 1, entry 14) provided the mixture (∼9:1) of monoamide products 11 and 12 in 86% yield after column chromatography.The mixture of monoamides (11 and 12)  was then taken to the next step for further separation.
With conditions to provide predominately the desired amide product 11, the separation of the mixture to exclude the undesired secondary amide 12 (Table 2) was studied.The free amines are oily after chromatography, so, for depletion of the undesired isomer 12, an acidic partner was sought that would generate a crystalline material.Thus, the 9:1 mixture of 11 and 12 was treated with various acids to screen for a suitable crystalline adduct.Mineral acids tend to cleave the amide bonds before crystallization occurs (Table 2, entries 1−3).
With weaker H 3 PO 4 or organic acids, the amide bond proved stable; however, most resulting salts that formed were not crystalline making them unsuitable for further purification All of the reactions were carried out with ∼9:1 regioisomeric mixture of 11:12 (1.0 g, 1.0 equiv), acid (1.0 equiv), solvent (10 V), rt, 3h.b HILIC ratio at 210 nm.ND = Not determined.NA = Not applicable.c 80% of mass recovery.d 78% of isolated yield.

Scheme 3. Completion of MMV693183 Synthesis
Organic Process Research & Development (Table 2, entries 4−8).Intriguingly, when L-(+)-tartaric acid was utilized, a stable white solid (18) was formed, and more importantly, the precipitate was enriched to a 25:1 ratio of desired/undesired amides in ∼80% mass recovery and 73 area % (210 nm) purity (Table 2, entry 9).A variety of solvents screened for salt formation with L-(+)-tartaric acid, and it was found that cosolvents of i-PrOH/MeOH (9:1) gave a stable tartrate salt of the desired regioisomer (Table 2, entry 14), while other solvent systems were less effective.The salt formation process rejected the undesired regioisomer, effectively affording exclusively the desired regioisomer 18 along with some excess L-(+)-tartaric acid in the isolate, which created purification problems in the following step.With a substoichiometric loading of the L-(+)-tartaric acid (0.9 equiv), the desired amine salt 18 was isolated with 78% yield and 97 area % (210 nm) purity as the exclusive regioisomer.
To complete the synthesis of MMV693183, L-(+)-tartrate salt 18 was reacted with 2,4,5-trifluorobenzoyl chloride (19) in the presence of 1.5 equiv of Na 2 CO 3 , providing the final API in 72% yield in >99 area % (210 nm) purity as determined by HPLC (Scheme 3).Both potassium and sodium carbonate can be used as a base to promote this final reaction, and both provide similar results; however, Na 2 CO 3 is lower cost and was ultimately preferred.
To further showcase the synthetic utility of our three-step protocol for preparation of MMV693183, two multigram batches were carried out (Scheme 4).Starting with 10 g of (S)-1,2-diaminopropane dihydrochloride (10), the monoamide product was isolated as a regioisomeric mixture (∼9:1) with 83% yield after column purification to remove the diamide impurity.Future work will explore the possibility to omit column chromatography.Treatment of the mixture of 11 and 12 with 0.9 equiv of L-(+)-tartaric acid furnished 18 in 78% yield with >99 area % (210 nm) HPLC purity.The resulting salt 18 was then acylated with 2,4,5-trifluorobenzoyl chloride (19) to afford MMV693183 in 72% yield with >99 area % (210 nm) HPLC purity.The overall yield of the three-step process from 10 to MMV693183 was 46%.

■ CONCLUSIONS
Contrasting this newly developed synthetic route of MMV693183 with the prior reported route, a number of critical advantages have been achieved in this body of work, including (1) dramatically reducing the step count (3 vs 7) with a considerably higher overall yield (46 vs 14%); (2) eliminating the need for protecting groups by taking advantage of the native steric differences of the primary amines in 10; (3) separating regioisomeric amides via an acid/base crystallization with L-(+)-tartaric acid; and (4) utilizing 2,4,5-trifluorobenzoyl chloride for acylation to avoid using expensive coupling reagents, all of this resulting in a cost-effective and scalable strategy to this promising API.These findings will hopefully serve to improve the commercial-scale manufacturing of MMV693183 in its effort to combat malaria.
■ EXPERIMENTAL SECTION General Information.Reagents and solvents were purchased from Sigma-Aldrich Chemical Co., Fisher Scientific, Alfa Aesar, Acros Organics, Oakwood, or TCI.Liquid reagents were purified by distillation when necessary.Unless otherwise noted, solid reagents were used without further purification.The key starting materials, (S)-(−)-1,2-diaminopropane dihydrochloride and D-(−)-pantolactone, were purchased from Sigma-Aldrich with 98% purity and >99% ee, and they were used as is without further purification.Column chromatography was carried out using a Biotage Isolera automated flash chromatography system.Melting point was measured using the Stuart melting point apparatus SMP10.For all compounds, 1 H, 13 C, and 19 F NMR spectra were recorded on a Bruker Avance III 600 MHz spectrometer.Chemical shifts were measured relative to the residual solvent resonance for 1 H and 13 C NMR (CDCl 3 = 7.26 ppm for 1 H and 77.2 ppm for 13 C, DMSO-d 6 = 2.50 ppm for 1 H and 39.5 ppm for 13 C, and CD 3 OD = 3.31 ppm for 1 H and 49.0 ppm for 13 C).Coupling constants J are reported in Hertz (Hz).The following abbreviations were used to designate signal multiplicity: s, singlet; d, doublet; t, triplet; dd, doublet of doublet; ddd, doublet of doublet of doublet; dt, double of triplet; m, multiplet; br, broad.Reactions were monitored by TLC, Scheme 4. Gram-Scale Demonstration of Protecting-Group-Free Synthesis of MMV693183

Organic Process Research & Development
HPLC, or GC-MS by using various methods.Exact mass measurements were obtained on a Thermo Scientific LTQ Orbitrap Velos.Glassware was oven-dried at 120 °C, assembled while hot, and cooled to ambient temperature under an inert atmosphere.Unless noted otherwise, reactions involving air-sensitive reagents or requiring anhydrous conditions were performed under a nitrogen atmosphere.HRMS was recorded using PerkinElmer Axion 2 ToF MS, ionization mode: positive with scan range: 100−1000 m/z, flight tube voltage: 8 kV, spray voltage: 3.5 kV, and solvent: methanol.

Figure 1 .
Figure 1.Chemical Structures of MMV693183 and related pantothenic acid derivatives.Scheme 1. Reported Synthetic Route to MMV693183