Glucokinase Activator: Practical Asymmetric Hydrogenation and

Glucokinase Activator: Practical Asymmetric Hydrogenation and Scalable Synthesis of an API Fragment ... Publication Date (Web): October 9, 2013. Copyr...
1 downloads 3 Views 394KB Size
Communication pubs.acs.org/OPRD

Glucokinase Activator: Practical Asymmetric Hydrogenation and Scalable Synthesis of an API Fragment Stephan Bachmann,*,†,§ Alec Fettes,†,§ Christian Lautz,‡ and Michelangelo Scalone†,§ †

pRED, Process Research and Synthesis, and ‡Pharma Technical Development, F. Hoffmann-La Roche Ltd., Grenzacherstrasse 124, CH-4070 Basel, Switzerland S Supporting Information *

Scheme 1. API structure and main disconnection

ABSTRACT: The enantioselective synthesis of (R)-2-(3chloro-4-methanesulfonyl-phenyl)-3-cyclopentyl propionic acid (R)-2 is described. The key intermediate (E)-7, a trisubstituted α-aryl β-alkyl acrylic acid, was conveniently accessed as its dicyclohexylamine salt by Perkin reaction in good yield and purity. Subsequent asymmetric hydrogenation with ruthenium catalysts was achieved with complete conversion and catalyst loadings up to S/C 75000 and enantiomeric excess up to 99% after crystallization.

A

ctivation of glucokinase (GK) is one of the proposed mechanisms for the treatment of type 2 diabetes.1 Several GK activators, including our development candidate R4940 (1), have shown promising results in clinical trials, thus requiring the supply of larger amounts of active pharmaceutical ingredient (API). The obvious disconnection at the amide bond gives acid (R)2 and pyrazine (R)-3, two building blocks of comparable synthetic complexity (Scheme 1). The present publication focuses on the efficient synthesis of (R)-2 suitable for largescale supply.2 Given the anticipated relatively high daily dose, cost of goods was a critical concern. Asymmetric hydrogenation of trisubstituted acrylic acids is often considered an established approach for the synthesis of enantiopure carboxylic acids.3 Whereas α-alkyl β-aryl derivatives or α,β-diaryl derivatives have been used as substrates,4−6 examples with α-aryl β-alkyl acrylates are scarce. To the best of our knowledge, asymmetric hydrogenation of such substrates has only been described using rhodium7 or iridium8 catalysts. The enantioselectivities were moderate (up to 85% ee for Rh, up to 95% ee for Ir) but the substrate-to-catalyst molar ratio (S/C, in the range 100−4000) was too low for technical application. Ru-based hydrogenation has not been reported with this type of substrate.9 For acid (R)-2, an enantiomeric excess of ≥95% was needed. Indeed, we showed that this quality leads to API of >99% ee by enrichment during the final crystallization. (R)-2 of lower enantiomeric purity would require an additional crystallization prior to the final coupling. The requisite substrate for the asymmetric hydrogenation is an acrylic ester, an acrylic acid or a salt thereof with the sulfurcontaining functional group either at the sulfide or sulfone oxidation state. The geometry of the double bond was expected to be very important for high enantioselectivity of the © 2013 American Chemical Society

hydrogenation, and thus, an expedient route to the isomerically pure substrate was pursued. The acrylic acid could conceivably be derived from a suitably substituted aryl acetic acid derivative by Perkin reaction.10 Alternatively, the transformation could proceed stepwise through an isolated aldol with subsequent acid- or basemediated elimination. Since preliminary experiments (Perkin reaction and hydrogenation of the corresponding olefin) confirmed the superiority of sulfone (E)-7 compared to the corresponding sulfide, our efforts focused exclusively on the preparation of sulfone (E)-7. The route commenced with the oxidation of commercially available 3-chloro-4-(methylthio)benzene acetic acid (4) (Scheme 2). Comparison of a variety of common oxidation agents11 including Oxone,12 H2O2/Na2WO4,13 magnesium monoperoxyphthalate (MMPP),14 and HCO3H, revealed that performic acid is the preferred reagent for oxidation in terms of Received: August 12, 2013 Published: October 9, 2013 1451

dx.doi.org/10.1021/op4002164 | Org. Process Res. Dev. 2013, 17, 1451−1457

Organic Process Research & Development

Communication

Scheme 2. Synthesis of the hydrogenation precursora

Under the preferred conditions, the protocol prescribes the dropwise addition over 6 hours of a 30% aq H2O2 solution (2.1 equiv) to a 20% solution of sulfide 4 in formic acid at 50 °C. After quenching with saturated aqueous sodium bisulfite, upon stripping most of the solvent, sulfone 5 crystallized by addition of water. Sulfone 5 was isolated as white crystals by filtration in 93% yield and excellent purity. The two-step exothermal oxidation process (ΔrH = −555 kJ/mol sulfide 4) can clearly be seen in the specific heat output of the reaction, i.e. the dosingcontrolled oxidation to the sulfoxide during addition of the first equivalent of H2O2 showed a slightly lower heat output than the following oxidation to the sulfone 5 during continued addition. The total adiabatic temperature rise for this step was high with ΔadiaTmax = 135 °C; the maximum thermal accumulation, corresponding to the in situ generated performic acid, was 9% for a dosing time of 6 hours, leading to a potential temperature increase of ΔadiaTaccu = 12 °C (Figure 1). At the end of the addition, stirring was maintained overnight to ensure complete conversion. The excess performic acid was quenched by careful addition of saturated aqueous sodium bisulfite solution (0.2 equiv) with negligible total adiabatic temperature rise of ΔadiaTmax = 1 °C. Examples of Perkin reactions with enolizable aldehydes are scarce and mostly proceed in modest yields.16 Typically, Perkin reactions require high temperatures in the presence of Et3N. Using these standard conditions, self-condensation product 8 was isolated as the main product, along with its hydrolyzed counterpart 9 (Scheme 3). The formation of these side products can be minimized by lowering the reaction temperature and using alkali metal acetates instead of amine bases.

a

Reagents and conditions: a) aq H2O2 (2.1 equiv, slow addition), HCO2H, 50 °C, 93% (cryst.). b) i. Ac2O (2.5 equiv), AcONa (1 equiv), THF, 40 °C, 25 h; ii. Cy2NH, Me2CO, heptane, 72%.

reaction rate, workup simplicity, waste stream and cost. A potential drawback of this method is the hazard associated with the use of performic acid on large scale:15 a very detailed study of the stability of performic acid solutions has shown its vigorous decomposition even at relatively low temperatures with the generation of a large amount of heat. On larger scale, accumulation of performic acid must be avoided.

Figure 1. Specific heat flow and enthalpy of reaction (left) as well as dosing, thermal conversion and accumulation (right) for the oxidation of sulfide 4. 1452

dx.doi.org/10.1021/op4002164 | Org. Process Res. Dev. 2013, 17, 1451−1457

Organic Process Research & Development

Communication

excerpt of the results is shown in Table 1. Several chiral diphosphine classes such as atropisomeric ligands of the MeOBIPHEP type, chiral 1,2-diphosphines (e.g., L7 or L17) and the majority of the ferrocene-based ligands gave only low to moderate enantioinduction. Nevertheless, three ligands in combination with Rh and five ligands in combination with Ir were identified to give good levels of enantioselectivity.17 With the Mandyphos-derived system ((R,S)-L13) 88−91% ee could be achieved (Table 1, runs 3 and 7). The Skewphos type ligands (S,S)-L8 and (S,S)-L9 were viable ligands for our substrate, giving 87% ee in combination with Rh (runs 1−2) and up to 97% ee in combination with Ir (runs 5−6). Additionally, two of the recently reported SIPHOX ligands (L14, L15) gave higher enantioselectivity in combination with Ir as well (runs 8−9).18 THF turned out to be the solvent of choice for the Rh catalysts, whereas for the Ir catalysts only CH2Cl2 was tested. As expected on the basis of literature precedent, the activity of Rh complexes remained rather low for this type of substrate, the highest S/C ratio of 1000−1500 being achieved with the Rh/ (R,S)-L13 catalyst. The Rh/(S,S)-L8 and (S,S)-L9 catalysts were even less active (maximum S/C ratio of 500−1000), with enantioselectivities decreasing at higher S/C ratios.19 As a result, the cost of the rhodium catalyst combined with the required high catalyst loadings would not allow us to achieve the ambitious cost target. The iridium-based catalysts in general showed comparable or better activities than the analogous Rh systems. The highest S/C ratio of 2500 was achieved with the Skewphos-based catalyst [Ir(COD)(S,S)-L9]BF4, with enantiomeric excesses in the range of 91−96% ee (Table 2). The activity of the Ir catalysts based on ligands (S,S)-L8 or (S,S)-L9 was strongly dependent on the hydrogen pressure, whereas the enantioselectivity was only marginally affected.20 Thus, the best performance in terms of conversion and optical purity was obtained with both cationic or neutral Ir catalysts at a hydrogen pressure of 10−20 bar. Even under the best conditions the S/C ratio remains moderate (2000−2500 at best), whereas ee values in the range of 91−94% still fell short of the targeted ≥95%. Finally, a wide range of Ru catalysts (140 in total) was screened; selected results are reported in Table 3. Only two of them gave results that deserved further optimization (runs 5 and 13), whereas all complexes containing other diphosphines gave only low to moderate ee. MeOH turned out to be the ideal solvent for [Ru(OAc)2((S)-L2)] (12), while CH2Cl2 was very good for [Ru((S,R)-L 10 )(η 5 -2,4-dimethylpentadienyl)(CH3CN)]BF4 (13).21 A short investigation of the influence of temperature and pressure on the activity and selectivity of the most selective Ru catalysts revealed quite different behaviors for the two complexes. As shown in Table 4, catalyst 12 tolerated hydrogen pressure of 180 bar (run 2) and temperatures of 80 °C (run 3). In contrast, with 13 as the catalyst the ee dropped when the hydrogenation was run above 60 °C (runs 5 and 7). A temperature of 50 °C combined with a hydrogen pressure of 50 bar proved ideal as to activity and selectivity with both 12 and 13. Under these conditions the lowest catalyst loading tested was at S/C 75000 using 12 (with 91.4% ee, Table 5, run 3), and S/C 22500 with catalyst 13 (run 5). The S/C ratio was further pushed to 25000 using catalyst 13 at 70 bar of hydrogen, and only a marginal decrease in ee was observed (run 6). No lower catalyst loadings were tested. The combined results of the hydrogenation clearly showed that the

Scheme 3. Main products isolated from Perkin reactions under standard conditions

Perkin reaction with acid 5 was best carried out using acetic anhydride (2.5 equiv), aldehyde 6 (1.5 equiv), sodium acetate (1.0 equiv) in THF (50% concentration) at 40 °C. After hydrolysis of the intermediate anhydride with water and catalytic amounts of DMAP, the crude acrylic acid was obtained. Purification was achieved by crystallization as the dicyclohexylamine salt. Finally, (E)-7·HNCy2 was isolated in 72% yield as a 98.4:1.6 E/Z mixture. In the crude reaction mixture of the Perkin reaction, two impurities were identified (Scheme 4); β-lactone 10 and βScheme 4. Main byproducts isolated from Perkin reactions under optimized conditions

hydroxy acid 11. β-Lactones are frequent side products in Perkin reactions. The amount of 10 formed is insensitive to all reaction parameters screened. Lactone 10 was easily removed during crystallization of (E)-7·HNCy2, whereas β-hydroxy acid 11 was very difficult to remove by crystallization. Hence, its formation had to be avoided. By strict control of the pH during workup (pH < 6), the formation of 11 could be minimized and its content reduced to acceptable levels (99.9 >99.9 >99.9 >99.9 99 >99.9 >99.9 >99.9 >99.9 >99.9

Table 3. Ru-catalyzed hydrogenation of (E)-7·HNCy2a

ee [%]c

run

Ru catalyst

conv. [%]b

87.2 87.8 88.2 81.8 89.6 97.0 90.8 93.6 91.0 82.0

1 2 3c 4 5 6 7 8 9 10 11 12 13d,e 14d,e 15d,e

[Ru(OAc)2((S,S)-L7)] [Ru(OAc)2((S,S)-L8)] [Ru(TFA)2((S,S)-L17)] [Ru(OAc)2((S)-L1)] [Ru(OAc)2((S)-L2)] (12) [Ru(OAc)2((S)-L3)] [Ru(OAc)2((S)-L5)] [Ru(TFA)2(S)-L6)] [Ru(OAc)2((S)-L18)] [Ru(OAc)2((R)-L19)] [Ru(OAc)2((S)-L4)] [Ru(OAc)2((R)-L20)] [Ru((S,R)-L10)(D)(S)]BF4 (13) [Ru((S,R)-L11)(D)(S)]BF4 [Ru((R,S)-L12)(D)(S)]BF4

82 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9 >99.9

(R) (R) (R) (R) (R) (R) (R) (R) (R) (R)

a Conditions: 50 mg scale, substrate-to-catalyst molar ratio (S/C) 25, [M(COD)2]BF4, chiral diphosphine ligand, solvent (1 mL), 50 °C, 50 bar H2, 17 h. bDetermined by HPLC. cDetermined by chiral LC. d Precursor: [Rh(COD)2]SbF6. fIsolated complex used. gCounterion: BArF ([B[3,5-(CF3)2C6H3]4]−).

ee [%]b 32.6 45.0 53.8 80.0 90.0 37.6 80.6 62.4 86.0 58.0 80.8 76.0 91.4 81.8 39.2

(S) (R) (S) (R) (R) (S) (R) (R) (R) (S) (R) (S) (R) (R) (S)

Conditions: 50 mg scale, S/C 25, Ru catalyst, MeOH (1 mL) 50 °C, 50 bar, 18 h. bDetermined by chiral LC. cTFA = trifluoroacetate. d CH2Cl2 used as solvent. eD = η5-2,4-dimethylpentadienyl; S = CH3CN. a

Table 2. Influence of temperature and pressure on the activity of the [Ir(COD)((S,S)-L9)]BF4 catalysta run

S/C

p [bar]

conv. [%]b

ee [%, (R)]c

1 2 3 4 5d 6d,e 7d 8f 9 10f 11g

2000 2000 2000 2000 1000 2000 2000 2000 2500 2500 2500

150 50 20 10 20 20 50 20 20 20 20

63.8 89 99.7 99.5 >99.9 81.5 77.5 >99.9 80 85 99

96.0 95.8 94.4 93.8 94.8 95.2 95.8 94.8 94.8 94.8 91.0

Table 4. Influence of temperature and pressure on the activity of 12 and 13a run

catalyst

S/C

solvent

1d 2 3e 4f 5d 6 7

12 12 12 13 13 13 13

500 1000 50 1000 1000 1000 1000

MeOH MeOH MeOH THF THF CH2Cl2 CH2Cl2

a

Conditions: 185-mL autoclave, 15 mmol of (E)-7·HNCy2, [Ir(COD)((S,S)-L9)]BF4 as catalyst, CH2Cl2, 50 °C, 18 h. bDetermined by HPLC. cDetermined by chiral LC. dCatalyst prepared in situ from [Ir(COD)2]BF4 and (S,S)-L9. e20 mmol scale. fCatalyst prepared in situ from [Ir(COD)Cl]2 and (S,S)-L9. gReaction run at 60 °C.

p [bar] T [°C] 20 180 50 20 20 180 50

50 50 80 50 80 50 60

conv. [%]b

ee [%, (R)]c

>99.9 >99.9 >99.9 98 >99.9 >99.9 >99.5

89.8 82.2 88.2 93.8 74.2 90.4 91.8

a

Conditions: 0.5 g of (E)-7·HNCy2, 35-mL autoclave, solvent, H2, 17 h. bDetermined by HPLC. cDetermined by chiral LC. d1 g scale, 50mL autoclave. e50 mg scale, 35-mL autoclave. f0.3 g scale, 35-mL autoclave.

1454

dx.doi.org/10.1021/op4002164 | Org. Process Res. Dev. 2013, 17, 1451−1457

Organic Process Research & Development

Communication

the removal of metal residues unnecessary. Indeed, the amount of ruthenium used during the hydrogenation was already below the threshold of 10 ppm allowed for oral application in the API.22 Levels of residual ruthenium after crystallization at −10 °C were determined to be 1.4 ppm (using catalyst 12 at S/C 50000). In conclusion, a highly efficient, asymmetric synthesis of acid (R)-2 has been developed. The three-step process starts with the oxidation of the commercially available sulfide 4 to sulfone 5, followed by high-yielding Perkin reaction of 5 with cyclopentane carboxaldehyde 6 to the acrylic acid dicyclohexylamine salt (E)-7·HNCy2. Its subsequent asymmetric hydrogenation affords the targeted acid (R)-2 with 97−99% ee after crystallization in 62% overall yield (Scheme 7). For the asymmetric hydrogenation, three commercially available catalysts have been identified with the ruthenium catalyst 12 being the preferred one for scale-up. In terms of cost, this new process compares favorably to the original process24 based on classical resolution by diastereomeric salt formation.

Table 5. Activity test of the catalysts 12 and 13 in the asymmetric hydrogenation of (E)-7·HNCy2a runb

catalyst

S/C

solvent

p [bar]

ee [%, (R)]c

1 2 3 4 5 6

12 12 12 13 13 13

10000 30000 75000 10000 22500 25000

MeOH MeOH MeOH CH2Cl2 CH2Cl2 CH2Cl2

50 50 50 50 50 70

90.8 91.6 91.4 93.0 92.2 91.6

Conditions: 10 g scale, 185-mL autoclave, solvent, H2, 50 °C, 17 h. Full conversion was achieved for all reactions as determined by HPLC. cDetermined by chiral LC.

a b

MeOBIPHEP-based ruthenium catalyst 12 is preferred for further scale-up. The detrimental impact of contamination with diastereomeric (Z)-7 in the hydrogenation substrate was demonstrated in a hydrogenation experiment with a (Z):(E) 80:20 mixture: (R)-2 was obtained with only 8.4% ee. Under otherwise identical conditions (catalyst 12, S/C 50, MeOH, 50 °C, 50 bar H2), diastereomerically pure (E)-isomer gave (R)-2 with 90% ee. Several issues were addressed before scale-up. (i) The thick suspension obtained at the end of the hydrogenation complicated the emptying of the autoclave. The addition of an acid (aq HCl or aq H2SO4) led to the formation of a solution which was much easier to handle. Aqueous sulfuric acid was preferred because of the corrosion potential of HCl. The obtained salt (Cy2NH·H2SO4) was efficiently extracted into the aqueous phase. (ii) Since the requirement of 95% ee for crude acid (R)-2 (crude ee: 91−92%) was not met, enrichment of optical purity by crystallization was investigated. Crude (R)-2 of 92% ee (obtained by ruthenium-catalyzed hydrogenation and liberated from dicyclohexyl amine by extraction with aqueous sulfuric acid) was crystallized conveniently from 2-propanol at −10 °C in 85−92% yield with 97% ee, thus allowing us to reach the desired 95% ee. (iii) For the ruthenium catalysts, the extraordinary activity renders



EXPERIMENTAL SECTION General. Optical rotation, MS, IR and NMR spectra were measured by the central analytical service of F. Hoffmann-La Roche Ltd. The 1H NMR and 13C NMR spectra were measured on Bruker 600 MHz NMR spectrometers at 600 and 150 MHz, respectively. The relative chemical shifts are reported in ppm relative to TMS. HPLC and chiral LC analyses were measured on Agilent spectrometers as described below. The reaction calorimetry measurements were performed on a Mettler-Toledo RC1e calorimeter with RTCal technology. Autoclaves were purchased from Premex Reactor AG, Lengnau, Switzerland and were usually made from Hastelloy C22. Elemental analyses and trace metal determinations (ICP-MS) have been carried out by Solvias AG, Römerpark 2, Kaiseraugst, Switzerland. Materials. All starting materials and reagents were purchased and used without further purification. All solvents were of p.a. quality, and the solvents for hydrogenation reactions were distilled under argon prior to use. All complexes,

Scheme 7. Summary of hydrogenation and purification stepsa

a Reagents and conditions: a) aq H2O2 (2.1 equiv, slow addition), HCO2H, 50 °C, 93% (cryst.); b) i. Ac2O (2.5 equiv), AcONa (1.0 equiv), 6 (2.5 equiv), THF, 40 °C, 25 h; ii. Cy2NH, Me2CO, heptane, 72%; c) i. [Ru(OAc)2(S)-L2] 12, S/C 50000, MeOH, 50 °C, 50 bar H2, 17 h; ii. aq H2SO4; d) 2-PrOH, rt to −10 °C.

1455

dx.doi.org/10.1021/op4002164 | Org. Process Res. Dev. 2013, 17, 1451−1457

Organic Process Research & Development

Communication

mixture was degassed with argon. Then, dicyclohexylamine (7.20 mL) was added at ambient temperature. Before the end of the addition, crystallization of the title compound was observed, giving a thick off-white suspension. Stirring was maintained for 17 h at ambient temperature and for 4.5 h at 0 °C. The crystals were filtered off, washed with a mixture of acetone (10 mL) and n-heptane (10 mL), filtered, and dried under reduced pressure at 50 °C for 19 h to yield the title compound (15.0 g, 72%) as white crystals (98.4% (E)-7· HNCy2). Mp: 195−196 °C. 1H NMR (600 MHz, CDCl3) δ 8.07 (d, J = 8.1 Hz, 1 H), 7.43 (d, J = 1.6 Hz, 1 H), 7.28 (dd, J = 8.1, 1.6 Hz, 1 H), 6.78 (d, J = 10.6 Hz, 1 H), 3.28 (s, 3 H), 2.82 (tt, J = 11.3, 3.75 Hz, 2 H), 2.28−2.37 (m, 1 H), 1.87− 1.97 (m, 4 H), 1.60−1.82 (m, 10 H), 1.47−1.58 (m, 2 H), 1.35−1.45 (m, 2 H), 1.25−1.34 (m, 4 H), 1.15−1.25 (m, 4 H), 1.04−1.15 (m, 2 H) . 13C NMR (150 MHz, CDCl3) δ 171.3, 147.5, 145.7, 136.2, 135.3, 133.3, 131.3, 129.8, 128.9, 52.6, 42.9, 40.0, 33.8, 29.5, 25.7, 25.3, 24.8. MS (EI, Turbo Spray): [Mamine + H]+, 182.0; [Macid + NH4]+, 346.1. Anal. Calcd for C27H40ClNO4S: C 63.57, H 7.90, N 2.75; found C 63.43, H 7.91, N 2.74. (R)-2-(3-Chloro-4-methanesulfonyl-phenyl)-3-cyclopentyl Propionic Acid ((R)-2). In a glovebox (