- Email: [email protected]
Two-step desymmetrization of dipyrazolidyl 3-phenylglutarate via lipase-catalyzed hydrolysis in organic solvents
Chemical Engineering Science 139 (2016) 41–48
Contents lists available at ScienceDirect
Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces
Two-step desymmetrization of dipyrazolidyl 3-phenylglutarate via lipase-catalyzed hydrolysis in organic solvents Po-Hao Chan, Shau-Wei Tsai n Institute of Biochemical and Biomedical Engineering, Chang Gung University, Kwei-Shan District, Tao-Yuan 33302, Taiwan, ROC
H I G H L I G H T S
G R A P H I C A L
A B S T R A C T
A sufficient condition for obtaining enantiomerically enriched product was proposed. Dipyrazolidyl 3-phenylglutarate but not the diester was employed as the substrate. Results agreed with the experimental data from the kinetic and thermodynamic analysis.
art ic l e i nf o
a b s t r a c t
Article history: Received 1 July 2015 Received in revised form 14 September 2015 Accepted 21 September 2015
A theoretical analysis on comparing the enzyme performance in a single-step desymmetrization, singlestep kinetic resolution, and two-step desymmetrization (i.e. a single-step desymmetrization followed by a sequent kinetic resolution) is reported. On the basis of ee* ≧ 0.95, a sufficient condition of E3E2 1 ≧ 10 and E1 ≧ 2 is proposed for obtaining an acceptable yield of X2R* 40.412 for the desired enantiomer in the two-step desymmetrization process, in comparison with E1 ≧ 39 for the single-step desymmetrization and E3E2 1 ≧ 20 for the single-step kinetic resolution. With the CALB-catalyzed hydrolytic desymmetrization of dipyrazolidyl 3-phenylglutarate (1) in MTBE as the model system, enantiomerically pure (R)monopyrazolidyl 3-phenylglutarate ((R)-2) can then bfdee prepared. Moreover from the kinetic analysis, the best reaction condition of using 20% water-saturated MTBE as the medium at 45 °C is selected for improving the enzyme activity and stereoselectivity. The thermodynamic analysis also indicates that the enzyme stereo-discrimination in the desymmetrization and sequent kinetic resolution is mainly entropic-driven. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Desymmetrization Kinetic resolution CALB Dipyrazolidyl 3-phenylglutarate (R)-Monopyrazolidyl 3-phenylglutarate
1. Introduction Lipases (E.C. 3.1.1.3) have been employed as the biocatalysts for synthesizing a variety of lipids, food and flavors, pharmaceuticals, fine chemicals, cosmetics, biodiesels, and polymers (Hasan et al., 2006; Kobayashi, 2009; Tan et al., 2010; Soumanou et al., 2013). A very useful feature of these enzymes is the enantiodiscrimination with which the preparation of single enantiomer of alcohol, acid, and amine is fulfilled
n
Corresponding author. Tel.: þ 886 3 2118800x3415; fax: þ886 3 2118668. E-mail address: [email protected] (S.-W. Tsai).
http://dx.doi.org/10.1016/j.ces.2015.09.024 0009-2509/& 2015 Elsevier Ltd. All rights reserved.
(Bornscheuer and Kazlauskas, 2006; Ghanem, 2007; Kamal et al., 2008; Faber, 2011; Paravidino et al., 2012). Many important intermediates and building blocks for various synthetic applications belong to diol, diamine, dicarboxylic acid, or anhydride that contains one or more prochiral or chiral centers. When using a symmetrical prochiral or meso compound as the substrate, the first-step desymmetrization followed by a subsequent kinetic resolution can generally increase the optical purity, yet with the price of decreasing the yield of the required enantiomer (Bornscheuer and Kazlauskas, 2006; Faber, 2011). Quantitative analysis in enzyme-catalyzed two-step desymmetrization of prochiral or meso compounds is more complicated than that in a single-step desymmetrization or kinetic resolution, as at least two
42
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
more kinetic parameters should be considered. This formulation remains still unclear and the relation of the theoretical to the experimental analysis has not been clearly discussed. Enantiomerically pure 3-substituted glutarates are key structural elements of many drug intermediates and building blocks in organic synthesis (Fryszkowska et al., 2005; Gopinath et al., 2012; Jung et al., 2013). Enantiomerically enriched 3-arylglutarates have been prepared from alcoholic ring-opening of cyclic anhydrides via organocatalysts or enzymes (Chaubey et al., 2008; Park et al., 2010; Roy et al., 2014; Fryszkowska et al., 2006; García-Urdiales et al., 2011; Palomo and Cabrera, 2012; Liu et al., 2014) and enzymecatalyzed hydrolysis, alcoholysis, aminolysis, or ammonolysis of dialkyl 3-arylglutarates (Yu et al., 2000; Homann et al., 2001; Lopez-Garcia et al., 2003; Cabrera et al., 2008; B. Wang et al., 2010; P.Y. Wang et al., 2010; Cabrera and Palomo, 2011; Liu et al., 2012; Nojiri et al., 2013). In general it is difficult to develop an efficient desymmetrization process leading to high enantiomeric purity and yields for the desired enantiomer at the mild reaction condition after inspecting the experimental data reported in the references of this paragraph. It is not only about the reaction conditions but also about the biocatalysts. As a part of our ongoing efforts toward using azolides as the substrate for preparing optically active compounds (Wang et al., 2009, B. Wang et al., 2010; P.Y. Wang et al., 2010; Cheng et al., 2012; Tsai, in press), we aimed to employ dipyrazolidyl 3phenylglutarate (1) as the model substrate for preparing an enantiomer of high enantiomeric purity and yield via CALBcatalyzed hydrolysis in MTBE (Scheme 1). A thorough kinetic analysis is firstly performed for proposing a sufficient condition leading to the efficient desymmetrization process giving ee* ≧ 0.95 and X2R* 40.412. From the kinetic analysis, the kinetic constants shown in Scheme 1 are estimated from experimental data for selecting the best reaction condition. The thermodynamic analysis is moreover addressed and elucidated.
CO2H O
N N
k1 N
OO
k3 CO2H CO2H
N N
N
PR ((R)-2)
S (1)
k2
k4 CO2H O
Q (3)
N N
PS ((S)-2) Scheme 1. Two-step desymmetrization of dipyrazolidyl 3-phenylglutarate in MTBE containing different water contents via lipase-catalyzed hydrolysis.
reaction. Similarly when performing a single-step kinetic resolution by using (R,S)-2 as the substrate, one obtains Eqs. (7) and (8), with which the enantiomeric excess for (R)-2 defined as ee¼(X2RX2S)(X2R þX2S) 1 may vary from zero to one. X2 R ¼
E1 ½1–expð tÞ 1 þ E1
ð5Þ
X2 S ¼
1–expð tÞ 1 þ E1
ð6Þ
X 2 R ¼ 0:5 expð E2 tÞ
ð7Þ
X 2 S ¼ 0:5 expð E3 tÞ
ð8Þ
2. Model development By using an excess of water for the hydrolysis and assuming the substrate concentrations to be much lower than the MichaelisMenten constants, an irreversible first-order kinetics for 1, (R)-2, and (S)-2 can be derived and solved for the time-course molar fractions as follows (Faber, 2011): X 1
¼ expð tÞ
ð1Þ
X2 R ¼
E1 ½expð E2 tÞ–expð tÞ ð1 þE1 Þð1 E2 Þ
ð2Þ
X2 S ¼
expð E3 tÞ–expð tÞ ð1 þ E1 Þð1 E3 Þ
ð3Þ
X 3 ¼ 1 X 1 X 2 R X 2 S
ð4Þ
The dimensionless parameters are defined as: t* ¼(k1 þk2)t as dimensionless time, E1 ¼k1k2 1 as stereoselectivity for the single-step desymmetrization, E2 ¼k3(k1 þk2) 1, E3 ¼k4(k1 þk2) 1, ee* ¼ (X2R*-X2S*)(X2R* þX2S*) 1 as enantiomeric excess for (R)-2, and hence E3E2 1 ¼k4k3 1 as enantiomeric ratio for the second-step kinetic resolution. The kinetic parameter combination (k1 þk2) may be firstly estimated from the regression of time-course data of X1* to Eq. (1), and then k2 and k4 from the time-course X2S* and Eq. (3). With the known k1 value, k3 is then regressed from the time-course X2R* and Eq. (2). Apparently for E2 ¼E3 ¼0, Eqs. (2) and (3) reduce to Eqs. (5) and (6) for a single-step desymmetrization, and lead to a constant enantiomeric excess of ee* ¼ (E1-1)(E1 þ1) 1 throughout the
3. Materials and methods 3.1. Materials Novozym 435 (Candida antarctica lipase B (CALB) immobilized on acrylic resins, containing 1–2% (w/w) water and has 7000 PLU/g by using lauric acid and 1-propanol as substrates at 60 °C) was purchased from Novozymes (Bagsvaerd, Denmark). 1 PLU is the amount of enzyme activity which generates 1 μmol of propyl laurate per minute under the defined conditions. Other chemicals of analytical grade were commercially available: pyrazole and 1 Hbenzotriazole from Acros (Geel, Belgium); thionyl chloride from Seedchem (Camberwell, Australia); chloroform-D from Cambridge Isotope Laboratories (Andover, MA); calcium hydride from SigmaAldrich (St. Louis, MO); acetic acid glacial (AA), benzene, isopropanol (IPA), methyl tert-butyl ether (MTBE), n-hexane (HEX) and triethylamine from Tedia (Fairfield, OH); 3-phenylglutaric acid from Alfa (Ward Hill, MA). 100% Water-saturated MTBE was prepared from a biphasic aqueous-MTBE solution kept in a water bath of specified temperature and with stirring for more than 24 h. Anhydrous MTBE was made by adding calcium hydride to the solvent for 24 h. Both solvents were then employed for preparing MTBE containing different water contents. For example, 20% water-saturated MTBE containing 94.5 mM of water at 45 °C was prepared by mixing 1:4 (v/v) of 100% water-saturated MTBE containing 472.7 mM of water and anhydrous MTBE (Alkandary et al., 2001).
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
3.2. Substrate preparation and analysis To 5 mL benzene containing 1.0 mmol 3-phenylglutaric acid, 2.4 mmol pyrazole and 5 mmol triethylamine was added dropwise a mixture containing 1 mL benzene and 2.5 mmol thionyl chloride at 0 °C, and then kept at room temperature with stirring for 4 h. The resultant mixture was stored at 4 °C for 30 min, and then quenched in succession with 10 mL of 10 mM HCl solution (0.1 M), NaOH solution (0.1 M), and NaCl solution (0.1 M) for 15 min for three times. The organic phase was separated, dried over anhydrous MgSO4 for 12 h, filtered, and concentrated under reduced pressure for giving an oily dipyrazolidyl 3-phenylglutarate (1). The 1 H NMR spectrum on Bruker Avance DRX 500 spectrometer in CDCl3 solution with tetramethylsilane as an internal standard was recorded and illustrated in Fig. S1. The hydrolysis of 1, (R)-2, and (S)-2 was monitored by HPLC using a Chiralcel AS-H column from Daicel (Tokyo, Japan). UV detection at 220 nm, 1 H-benzotriazole of 20 mM as the internal standard (IS), and a mobile phase consisting of n-hexane/isopropanol/acetic acid glacial ¼ 86.8/13/0.2 (v/v) at a flow rate of 1.2 mL/min were employed for quantification at ambient temperature. 1, 3, (R)-2, (S)-2, and the internal standard were eluted with the retention time of 9.06, 11.28, 12.67, 14.9, and 19.07 min, respectively, as demonstrated in Fig. S2. 3.3. Kinetic analysis and product separation Unless specified, to 5 mL of 100% water-saturated MTBE containing 10 mM of 1 was added 20 mg/mL of Novozym 435 at 45 °C. The resultant solution was stirred with a magnetic stirrer at 400 rpm. Samples were removed at different time intervals and injected onto the HPLC for analysis. The time-course molar fractions X1* and X3* were first calculated via the calibration curves for 1 and 3 (Fig. S3). From the area ratio of (R)-2 and (R)-2 in HPLC spectrograms and (X2R* þX2S*) calculated from the total material balance for all compounds, the time-course X2R* and X2S* and hence enantiomeric excess ee* were calculated. After depleting the substrate and removing the enzyme via filtration, the resultant solution was quenched with 10 mL of KH2PO4 buffers (pH 5 and 25 mM) for 2 h at 4 °C for three times. The organic phase was then separated, dried over anhydrous MgSO4 for 12 h, filtered, and concentrated under reduced pressure for obtaining (R)-2 of 54% yield and ee* ¼0.96 as a yellowish powder. The analysis was furthermore carried out in MTBE containing different water contents. Effects of substrate concentrations varied from 10.6 and 40.8 mM in 20% water-saturated MTBE were repeated at 45 °C. Unfortunately owing to the solubility limitation of substrate 1 in MTBE at 45 °C, we did not employ substrate concentrations higher than 40.8 mM for the reaction. Similarly in order to study the temperature influence on enzyme performance, the reactions in 60% water-saturated MTBE were also performed at 25 and 35 °C.
4. Results and discussion 4.1. Theoretical modeling The sufficient condition for carrying out an effective desymmetrization process followed by a sequent kinetic resolution for obtaining (R)-2 is that the kinetic parameters should follow the order: k1 ≧ k2, k1 4k3, and k4 4k3 and hence E1 ≧ 1, E3E2 1 41, and E2(E1 þ1)E1 1 o1 (or k2 ≧ k1, k2 ≧ k4, and k3 4 k4 and hence E1 ≦ 1, E3E2 1 o1, and E2(E1 þ1)E1 1 41 if (S)-2 is desired). If the reaction condition can only lead to a single-step desymmetrization, results of E1(E1 þ 1) 1≧X2R* ≧ 0 and ee* ¼(E1-1)(E1 þ1) 1 throughout the
43
reaction are obtainable. Therefore when an optical purity of ee*≧ 0.95 is required, the selectivity for the desymmetrization E1≧39 is needed. Otherwise, one should adjust the reaction condition such that the second-step kinetic resolution can follow and leads to ee* ≧ (E1-1)(E1 þ1) 1 and E1(E1 þ1) 1≧X2R* ≧ 0. Apparently, the penalty of giving ee* values higher than (E1-1)(E1 þ 1) 1 is to reduce the yield of X2R*. The disadvantages of only carrying out the second-step kinetic resolution for (R,S)-2 include X2R ≦ 0.5 and conversions to be higher than 0.5 in order to obtain high ee values. It is well recognized in academy and industry that the acceptable enantiomeric ratio k4k3 1 should be greater than 20, and hence giving X2R 40.412 calculated from k4k3 1ln[(X2R þX2S)(1 þee)]¼ ln [(X2R þX2S)(1-ee)] for ee¼0.95. When the ratio is less than 20, it is imperative to add a single-step desymmetrization before the kinetic resolution for increasing X2R*. Fig. 1A and B illustrates ee* varied with (X2R* þX2S*), ee varied with (X2R þX2S), and time-course (X2R* þX2S*) and (X2R þX2S) for E1 ¼2, E3E2 1 ¼ 2, and k4k1 1 ¼E3(E1 þ1)E1 1 equal to 0.1, 1, and 2, respectively. At the low E1 value, it is still possible to improve the ee* value from 0.333 for a single-step desymmetrization to 0.95, yet giving very low X2R* ¼0.0343 at t* ¼90 for E2 ¼0.0333 and E3 ¼0.0666 (i.e. E3(E1 þ1)E1 1 ¼0.1). Increasing of E2 ¼0.333 and E3 ¼0.666 (i.e. E3(E1 þ1)E1 1 ¼1) as well as E2 ¼ 0.666 and E3 ¼1.333 (i.e. E3(E1 þ1)E1 1 ¼2) results in even lower X2R* values for ee* ¼ 0.95, yet with the benefit of decreasing the required dimensionless time. A change of E3E2 1 to 10 may improve X2R* value to 0.481 at t* ¼50 and ee* ¼0.948 for E2 ¼ 0.00666 and E3 ¼0.0666 (Figs. 1C and 1D). Similarly by further increasing E2 and E3 to 0.0666 and 0.666, respectively, a slight reduction of X2R* ¼0.477 at t* ¼6.0 and ee* ¼0.979 is found (or X2R* ¼ 0.413 at t*¼4.5 and ee* ¼0.959 for E2 ¼ 0.133 and E3 ¼ 1.33). By moreover increasing E3E2 1 to 50 (Figs. 1E and 1F), an enhancement of X2R* ¼ 0.629 at ee* ¼0.945 and t* ¼ 45 for E2 ¼0.00133 and E3 ¼ 0.0666 is calculated. Similarly for a higher E2 ¼0.0133 and E3 ¼0.666, only a slight reduction of X2R* ¼0.622 at t* ¼6.0 and ee* ¼ 0.950 is found (or X2R* ¼ 0.603 at t* ¼4.0 and ee* ¼ 0.956 for E2 ¼0.0266 and E3 ¼ 1.33). As the enantiomeric ratio E3E2 1 is now higher than 20, it is interesting to compare the results when only carrying out a single-step kinetic resolution. In general one obtains X2R* 4X2R at high E1, ee*, and ee values, e.g. X2R ¼0.471 and ee¼0.961 at t* ¼60 for E2 ¼0.00133 and E3 ¼0.0666, at t* ¼ 6.0 for E2 ¼ 0.0133 and E3 ¼0.666, as well as at t* ¼3.0 for E2 ¼0.0266 and E3 ¼1.333. At E3E2 1 ¼2 (Figs. S4A and S4B), one only obtains X2R* ¼0.244 and ee* ¼0.979 at t* ¼ 30 for E2 ¼0.0455 and E3 ¼0.0909, X2R* ¼ 0.161 and ee* ¼0.952 at t* ¼5.0 for E2 ¼ 0.454 and E3 ¼0.909, as well as X2R* ¼ 0.0803 and ee* ¼ 0.953 at t* ¼4.0 for E2 ¼0.909 and E3 ¼ 1.818. More improvements of X2R* at the higher selectivity are perceived, e.g. at E3E2 1 ¼10 in Figs. S4C and S4D, one obtains X2R* ¼0.779 and ee* ¼ 0.951 at t* ¼ 18 for E2 ¼0.00909 and E3 ¼0.0909, X2R* ¼0.711 and ee* ¼0.957 at t* ¼3.0 for E2 ¼0.0909 and E3 ¼0.909, as well as X2R* ¼0.622 and ee* ¼0.962 at t* ¼2.0 for E2 ¼0.182 and E3 ¼1.818. Apparently at an even higher value of E3E2 1 ¼50 (Figs. S4E and S4F), more favorable results are demonstrated, e.g. X2R* ¼0.885 and ee* ¼ 0.949 at t* ¼ 16 for E2 ¼ 0.00182 and E3 ¼0.0909, X2R* ¼ 0.809 and ee* ¼0.949 at t* ¼2.5 for E2 ¼0.0182 and E3 ¼0.909, as well as X2R* ¼0.683 and ee* ¼0.950 at t* ¼1.5 for E2 ¼0.364 and E3 ¼1.818. Therefore for a desired high ee* value, E1 in general has a more impact on decreasing the required dimensionless time, yet with E3E2 1 having more influence on enhancing the yield of X2R*. These arguments are perceived if one compares the results at ee* ¼0.95 for E1 ¼ 2 and E3E2 1 ¼10 with those for E1 ¼10 and E3E2 1 ¼2. By moreover comparing the results with those for E1 o2 and E3E2 1 ¼10, a sufficient condition of E3E2 1≧10 and E1≧2 is hence
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
*) or (X 2R + X 2S)
44
1.0
0.1
2
0.4
1
2S
0.6
*+X
ee* or ee
0.8
(X
2R
*+X
2S
* ) or ( X 2R + X
(X
0.0 0.2 0.4 0.6 0.8 1.0 2S
1.0
1
2
0.1
2S
0.4
2R
0.2
(X
2R
*+X
2S
* ) or ( X 2R + X
2S
(X
0.0 0.2 0.4 0.6 0.8 1.0
1
0.1
2S
0.4
*+X
ee* or ee
2
0.6
2R
0.2 0.0 0.2 0.4 0.6 0.8 1.0
(X
2R
*+X
2S
* ) or ( X 2R + X
2S
(X
0.0
1
0.4 0.2 0.0
2 0
5
10
1.0
)
15
20
15
20
15
20
0.1
0.8 0.6
1
0.4
2
0.2 0.0
0
5
10
)
1.0 0.8
0.6
t*
*) or (X 2R + X 2S)
0.0
0.1
t*
*+X
ee* or ee
0.8 0.6
0.8
) *) or (X 2R + X 2S)
0.0
2R
0.2
1.0
1.0
0.1
0.8
1
0.6
2
0.4 0.2 0.0
0
5
10
t*
Fig. 1. Theoretical ee* (——) and ee (— -) varied with (X2R* þ X2S*) and (X2R þ X2S), respectively, for (A) E1 ¼ 2, E3/E2 ¼2, (C) E1 ¼2, E3/E2 ¼ 10, and (E) E1 ¼ 2, E3/E2 ¼50. Timecourse (X2R* þX2S*) (——) and (X2R þ X2S) (— -) for (B) E1 ¼2, E3/E2 ¼ 2, (D) E1 ¼ 2, E3/E2 ¼10, and (F) E1 ¼ 2, E3/E2 ¼50.
proposed for leading to an efficient two-step desymmetrization process. 4.2. Kinetic analysis The time-course X1*, X2R*, X2S*, X3*, and ee* in Fig. 2A, as well as X2R*, ee*, and (X2R* X2R) varied with (X2R* þX2S*) and ee varied with (X2R þX2S) in Fig. 2B are illustrated, when Scheme 1 is carried out in 100% water-saturated MTBE containing 472.7 mM of water at 45 °C. The kinetic constants estimated and represented in Table 1 are employed for calculating the best-fitted results in agreements with the experimental data in the figures. In comparison with
X2R ¼ 0.266 and ee40.99 at t¼ 70.3 h for only performing a singlestep kinetic resolution, a lower reaction time of 60 h is required for giving X2R* ¼0.255 and ee* 40.99. Moreover, the time-course X2R, X2S, ee, and (X2R* X2R) are calculated and represented in Fig. 2B, giving a maximum enhancement of (X2R* X2R) ¼0.192 for X2R* ¼0.437 and ee* ¼0.95 at t¼ 30.4 h. An investigation of the kinetic constants indicates that the higher value of E3E2 1 ¼9.4, in comparison with E1 ¼3.3 and k3k2 1 ¼1.25, is mainly contributed for giving k4k1 1 ¼ 3.54. Yet no explanation is found for elucidating why the pyrazolidyl moiety in 1 and the acidic group in (R)-2 and (S)-2 can induce different steric hindrances for substrate affinity to
1.0 0.8 0.6 0.4 0.2 0.0
0
20
40
60
80
X 2R*, ee*, (X 2R* - X 2R), ee
X 1*, X 2R*, X 2S*, X 3*, ee*
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
45
1.0 0.5 0.0 -0.5
0.0 0.2 0.4 0.6 0.8 1.0
(X 2R* + X 2S*) or ( X 2R + X 2S)
t (h)
Fig. 2. (A) Time-course X1* (●), X2R* (⎕), X2S* (Δ), X3* (▲), and ee* (∇). (B) Variations of X2R* (⎕), ee* (∇), and (X2R*-X2R) (— -) with (X2R* þ X2S*), and ee (— —) varied with (X2R þX2S). Reaction condition: 100% saturated-water MTBE, 20 mg/mL Novozym 435, (S)0 ¼9.2 mM, and 45 °C. (——) Best-fitted results from Eqs. (1)–(4).
Table 1 Effects of water content in MTBE on kinetic parameters for desymmetrization of 1 at 45 °C. Water content (% saturation)
(k1 þ k2) (1/h) k1 (1/h)
10 20 60 100
1.11 9.46E 1 1.23E 1 8.12E 2
9.60E 1 8.17E-1 9.69E 2 6.24E 2
k2 (1/h)
k3 (1/h)
k4 (1/h)
E1 ( ¼ k1k2 1) E3E2 1 ( ¼ k4k3 1)
E2(E1 þ 1) ( ¼ k3k2-1)
E3(E1 þ 1)E1 1 ( ¼ k4k1 1)
1.51E 1 1.29E 1 2.61E 2 1.88E 2
7.87E 2 5.15E 2 2.90E 2 2.35E 2
1.14 8.75E 1 2.97E 1 2.21E 1
6.3 6.3 3.7 3.3
0.52 0.40 1.11 1.25
1.18 1.07 3.06 3.54
14.5 17.0 10.2 9.4
Reaction condition: 20 mg/mL Novozym 435 and (S)0 around 10 mM. Symbol of E 1 as 10 1.
the enzyme acyl pocket and proton transfers from the triad imidazolium to the leaving pyrazolidyl group. The desymmetrization in 60% water-saturated MTBE is also demonstrated in Fig. S5, and employed for estimating the kinetic constants (Table 1). Agreements of the best-fitted results with experimental data are perceived from the figure. Apparently, the reduction of the water content has improved the enzyme activity but not the selectivity, i.e. 1.23- to 1.55-fold enhancements for all kinetic constants leading to nearly invariable E1 ¼3.7 and E3E2 1 ¼10.2. The lipase performances are further improved if the two-step desymmetrization is carried out in 20% water-saturated MTBE (Fig. 3A and B). Similarly the estimated kinetic constants (Table 1) are employed for calculating the best-fitted results and show agreements with the experimental data. Enhancements of the kinetic constant do yield E1 ¼6.3 and E3E2 1 ¼17, giving favorable results of X2R* ¼0.667 and ee* 40.99 at t ¼6 h as well as (X2R* X2R)¼0.315 for X2R* ¼0.722, and ee* ¼0.963 at t¼4 h. Although the employed substrate and reaction condition are different, a better enzyme performance for our data is concluded after comparing with the results of X2R* ¼ 0.55 and ee* 40.99 at 240 h when dimethyl 3-phenylglutarate of 25 mM was hydrolyzed in pH 5 sodium phosphate containing 300 mg/mL Novozym 435 at 25 °C (Cabrera et al., 2009), as well as X2S* ¼0.42 and ee* ¼ 0.96 at 144 h when dimethyl 3-phenylglutarate of 250 mM was ammonolyzed in MTBE containing 30 mg/mL Novozym 435 at 30 °C (Lopez-Garcia et al., 2003). The time-course data in 10% water-saturated MTBE are furthermore illustrated in Fig. S6. The kinetic constants then estimated and tabulated in Table 1 are employed for preparing the best-fitted curves that are in agreement with the experimental data as demonstrated in the figure. A slight enhancement for each kinetic constant, but not the stereoselectivity, is perceived and leads to X2R* ¼0.624 and ee* 4 0.99 at t ¼5 h. Therefore, 20% water-
saturated MTBE is selected as the best reaction medium by considering water also acting as the substrate for hydrolysis. When anhydrous and 100% water-saturated MTBE were, respectively, employed as the reaction media for hydrolyzing (R,S)2-phenylpropionic pyrazolide (or 1,2,4-triazolide), an order-ofmagnitude higher Michaelis-Menten constants for the later were reported (Wang et al., 2009; B. Wang et al., 2010; P.Y. Wang et al., 2010). This implies that water adsorbed on the active site region may impede substrate affinity to the binding pockets, and leads to a higher Michaelis–Menten constant for each enantiomer in 100% water-saturated MTBE. Yet, the water content also strongly affects the proton transfer and hence the nucleophilic attack of triad serine to the carbonyl moiety. This can result in higher lipase activity in anhydrous MTBE, just like the monotonic increase of kinetic constants (Table 1) when decreasing the water content. However it is still difficult to elucidate k3k2 1 and k4k1 1 varied with the water content when a pyrazolidyl moiety in 1 is replaced with the acidic group in (R)-2 and (S)-2. In order to check if the irreversible first-order kinetics is still valid for high substrate concentrations, more experiments in 20% watersaturated MTBE at 45 °C were carried out. By using the estimated kinetic constants of k1 and k2 from the time-course data (not shown here), the initial rates for converting 1 to (R)-2 and (S)-2 are calculated and illustrated in Fig. 4. One may then estimate the kinetic parameter combination of (Km,1 1 þ Km,2 1)¼1.458 10 2 mM 1 from Fig. 4, k1 and k2 for 20% water-saturated MTBE in Table 1, and Eqs. (S3) and (S4) of Supplementary data. Agreements between the experimental data and best-fitted results calculated from V2R ¼ 68.13 (S)0[68.56þ(S)0] and V2S ¼ 14.80(S)0[68.56þ(S)0] are also illustrated in the figure. Moreover if the same Michaelis-Menten constants are assumed, one obtains Km,1 ¼Km,2 ¼137.1 mM, implying that it is reasonable to assume the irreversible first-order kinetics when the employed substrate concentration is below 20 mM. Otherwise, more complicated rate equations such as those derived from Michaelis-
1.0 0.8 0.6 0.4 0.2 0.0
0
1
2 3 t (h)
4
5
X 2R*, ee*, (X 2R* - X 2R), ee
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
X 1*, X 2R*, X 2S*, X 3*, ee*
46
1.0 0.5 0.0 -0.5
0.0 0.2 0.4 0.6 0.8 1.0
(X 2R* + X 2S*) or ( X 2R + X 2S)
Fig. 3. (A) Time-course X1* (●), X2R* (⎕), X2S* (Δ), X3* (▲), and ee* (∇). (B) Variations of X2R* (⎕), ee* (∇), and (X2R*-X2R) (—) with (X2R* þX2S*), and ee (— —) varied with (X2R þX2S). Reaction condition: 20% saturated-water MTBE, 20 mg/mL Novozym 435, (S)0 ¼ 10.6 mM, and 45 oC. (——) Best-fitted results from Eqs. (1)–(4).
(K mol) are calculated. By comparing the absolute values of ΔΔH and TΔΔS at a specific temperature, the stereo-discrimination for the productive 1 and (S)-2 converting to (R)-2 and 3, respectively, is entropic-driven with compensation from the enthalpy loss.
-1
V2R, V2S (mM h )
40
30
20
5. Conclusions
10
0
0
10
20
30
40
50
(S)0 ( m M ) Fig. 4. Initial rates of (R)-2 (∇) and (S)-2 (○) varied with (S)0. Reaction condition: 20% water-saturated MTBE, 20 mg/mL Novozym 435, and 45 oC. (——) Best-fitted results from Eqs. (S3) and (S4).
Menten kinetics for all reaction steps in Scheme 1 are needed for modeling the whole reaction. 4.3. Thermodynamic analysis In order to check if the enzyme stereoselectivity can be improved at low temperature, more experiments in 60% watersaturated MTBE were carried out at 25 and 35 °C. The kinetic constants estimated from the time-course data (not shown here) are tabulated in Table 2. About 4.5-fold increments for each kinetic constant, and hence yielding nearly invariable E1 and E3E2 1, are found when increasing the temperature from 25 to 45 °C. Therefore, we have selected 45 °C as the best temperature for giving the highest enzyme activity. The good linear relationships between logarithms of kinetic constant and the inverse of absolute temperature, i.e. ln[k1]¼ 7187 T 1 þ20.30, ln[k2]¼ 6896 T 1 þ 18.05, ln[k3]¼ 7042 T 1 þ 18.62, and ln[k4]¼ 7670 T 1 þ22.90, are furthermore illustrated in Fig. 5. Therefore from ln[E1]¼ 291 T 1 þ2.25 for converting 1 to (R)-2 and (S)-2, the differences of enthalpy and entropy between the transition states for both substrates are estimated as ΔΔH¼ 2.42 kJ/mol and –ΔΔS¼ 18.7 J/(K mol) (Wang et al., 2009). Similarly from ln[E3E2 1]¼ 628 T 1 þ 4.28 for converting (R)-2 and (S)-2 to 3, ΔΔH¼ 5.22 kJ/mol and ΔΔS¼-35.58 J/
The theoretical analysis for comparing the enzyme performances in a single-step desymmetrization, a single-step kinetic resolution, and the two-step desymmetrization for preparing chiral compounds is investigated, in which the dimensionless groups of E1, E3E2 1, and E3(E1 þ1)E1 1 are employed as the parameters. An effective two-step desymmetrization process should fulfill the sufficient condition of E1 ≧ 1, E3E2 1 41, and E2(E1 þ1)E1 1 o1. Moreover in order to obtain an acceptable yield with ee* ≧ 0.95 for (R)-2, the quantitative condition of E3E2 1 ≧ 10 and E1≧2 is furthermore proposed. Optically pure (R)-monopyrazolidyl 3-phenylglutariate can be prepared from the hydrolytic desymmetrization of dipyrazolidyl 3-phenylglutarate via Candida antarctica lipase B, in which the best reaction condition of 20% water-saturated MTBE at 45 °C, leading to E1 ¼ 6.3 and E3E2 1 ¼17.0, is selected from the kinetic analysis. The thermodynamic analysis moreover indicates that the enzyme stereodiscrimination is mainly entropic-driven in the desymmetrization and sequent kinetic resolution, giving nearly invariable E1 and E3E2 1 values when changing the reaction temperature.
Nomenclature selectivity defined as k1k2 1 dimensionless parameters defined as k3(k1 þk2) 1and k4(k1 þk2) 1, respectively ee enantiomeric excess defined as (X2R X2S)(X2R þX2S) 1 Km,1, Km,2 Michaelis–Menten constants leading to (R)-2 and (S)-2 from 1, respectively (mM) k1, k2,k3, k4 kinetic constants shown in Schemes 1 (h 1) (S)0 initial concentration for dipyrazolide substrate (mM) T absolute temperature (K) t time (h) t* dimensionless time defined as (k1 þ k2)t V2R, V2S initial rates for (R)-2 and (S)-2 from 1, respectively (mM/ h) Vmax,1, Vmax,2 maximum velocity leading to (R)-2 and (S)-2 from 1, respectively (mM/h) E1 E2, E3
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
47
Table 2 Temperature influence on kinetic parameters for desymmetrization of 1 in 60% water-saturated MTBE. Temp (°C)
(k1 þk2) (1/h)
k1 (1/h)
k2 (1/h)
k3 (1/h)
k4 (1/h)
E1 ( ¼ k1k2 1)
E3E2 1 (¼ k4k3 1)
E2(E1 þ 1) ( ¼ k3/k2)
E3(E1 þ1)E1 1 (¼ k4/k1)
25 35 45
2.74E 2 6.70E 2 1.23E 1
2.13E 2 5.32E 2 9.69E 2
6.12E 3 1.39E 2 2.61E 2
6.59E 3 1.55E 2 2.90E-2
5.90E 2 1.39E 1 2.97E 1
3.5 3.8 3.7
9.0 9.0 10.2
1.07 1.15 1.11
2.77 2.61 3.06
-1
ln[k1], ln[k2], ln[k3], ln[k4], ln[E1], ln[E3E2 ]
Reaction condition: 20 mg/mL Novozym 435, (S)0 around 10 mM, and 45°C. Symbol of E 1 as 10 1.
3
0
-3
-6
3.1
3.2
3.3 -1
3.4
-1
1000T (K ) Fig. 5. Linear variations of ln(k1) (○), ln(k2) (Δ), ln(k3) (⎕), ln(k4) (∇), ln(E1) (●), and ln(E3E2 1) (▲) with inverse of absolute temperature. Reaction condition: 60% watersaturated MTBE, 20 mg/mL Novozym 435, and (S)0 ¼ 11.1 mM.
X1, X2R,X2S, X3 molar fractions based on (S)0 for 1, (R)-2 and (S)-2, and 3, respectively ΔΔH enthalpy difference for two transition states (kJ/mol) ΔΔS entropy differences for two transition states (J/(mol K)) Superscript * for a single-step or two-step desymmetrization Subscript max
as maximum value
Acknowledgments Financial supports of NSC 104-2221-E-182-058-MY2 from Ministry of Science and Technology are appreciated.
Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.ces.2015.09.024.
References Alkandary, J.A., Aljimaz, A.S., Fandary, M.S., Fahim, M.A., 2001. Liquid-liquid equilibria of waterþ MTBEþ reformate. Fluid Phase Equilib. 187-188, 131–138.
Bornscheuer, U.T., Kazlauskas, R.J., 2006. Hydrolases in organic synthesis: regioand stereoselective biotransformations, 2nd ed. Wiley-VCH, Weinheim. Cabrera, Z., Fernandez-Lorente, G., Palomo, J.M., Guisan, J.M., Fernandez-Lafuente, R., 2008. Asymmetric hydrolysis of dimethyl 3-phenylglutarate catalyzed by s Lecitase Ultra . Effect of the immobilization protocol on its catalytic properties. Enzyme Microb. Technol. 43, 531–536. Cabrera, Z., Fernandez-Lorente, G., Fernandez-Lafuente, R., Palomo, J.M., Guisan, J. M., 2009. Novozym 435 displays very different selectivity compared to lipase from Candida antarctica B adsorbed on other hydrophobic supports. J. Mol. Catal. B: Enzym. 57, 171–176. Cabrera, Z., Palomo, J., 2011. Enantioselective desymmetrization of prochiral diesters catalyzed by immobilized Rhizopus oryzae lipase. Tetrahedron Asymmetr 22, 2080–2084. Chaubey, N., Date, S.M., Ghosh, S.K., 2008. An efficient asymmetric desymmetrization of prochiral glutaric anhydrides with SuperQuat chiral oxazolidin-2ones. Tetrahedron Asymmetry 19, 2721–2730. Cheng, Y.L., Wu, A.C., Wang, P.Y., Tsai, S.W., 2012. Kinetic and thermodynamic analysis for lipase-catalyzed hydrolytic resolution of (R,S)-alcohols though their azolyl carbamates. Bioprocess Biosyst. Eng. 35, 953–962. Faber, K., 2011. Biotransformations in organic chemistry: a textbook, 6th ed. Springer-Verlag, New York. Fryszkowska, A., Komar, M., Koszelewski, D., Ostaszewski, M., 2005. Enzymatic desymmetrization of 3-arylglutaric acid anhydrides. Tetrahedron Asymmetry 16, 2475–2485. Fryszkowska, A., Komar, M., Koszelewski, D., Ostaszewski, R., 2006. Studies on enzymatic synthesis of chiral non-racemic 3-arylglutaric acid monoesters. Tetrahedron Asymmetry 17, 961–966. García-Urdiales, E., Alfonso, I., Gotor, V., 2011. Update 1 of: enantioselective enzymatic desymmetrizations in organic synthesis. Chem. Rev. 111, PR110–PR180. Ghanem, A., 2007. Trends in lipase-catalyzed asymmetric access to enantiomerically pure/enriched compounds. Tetrahedron 63, 1721–1754. Gopinath, P., Watanabe, T., Shibasaki, M., 2012. Studies on catalytic enantioselective total synthesis of caprazamycin B: construction of the western zone. J. Org. Chem. 77, 92605–99267. Hasan, F., Shah, A.A., Hameed, A., 2006. Industrial applications of microbial lipases. Enzyme Microb. Technol. 39, 235–251. Homann, M.J., Vail, R., Morgan, B., Sabesan, V., Levy, C., Dodds, D.R., Zaks, A., 2001. Enzymatic hydrolysis of a prochiral 3-substituted glutarate ester, an intermediate in the synthesis of an NK1/NK2 dual antagonist. Adv. Synth. Catal. 343, 744–749. Jung, J.-H., Yoon, D.-H., Kang, P., Lee, W.K., Eum, H., Ha, H.-J., 2013. CAL-B catalyzed desymmetrization of 3-alkylglutarate: "olefin effect" and asymmetric synthesis of pregabalin. Org. Biomol. Chem. 11, 3635–3641. Kamal, A., Azhar, M.A., Krishnaji, T., Malik, M.S., Azeeza, S., 2008. Approaches based on enzyme mediated kinetic to dynamic kinetic resolutions: a versatile route for chiral intermediates. Coord. Chem. Rev. 252, 569–592. Kobayashi, S., 2009. Recent developments in lipase-catalyzed synthesis of polyesters. Macromol. Rapid Commun. 30, 237–266. Liu, W., Hu, Y., Jiang, L., Zou, B., Huang, H., 2012. Synthesis of methyl (R)-3-(4fluorophenyl)glutarate via enzymatic desymmetrization of a prochiral diester. Proc. Biochem. 47, 1037–1041. Liu, W., Hu, Y., Zhang, Y., Ma, Y., Huang, H., 2014. Enzymatic desymmetrization of 3(4-fluorophenyl)glutaric anhydride through enantioselective alcoholysis in organic solvents. Biotechnol. Bioprocess. Eng. 19, 449–455. Lopez-Garcia, M., Alfonso, I., Gotor, V., 2003. Desymmetrization of dimethyl 3substituted glutarates through enzymatic ammonolysis and aminolysis reactions. Tetrahedron Asymmetry 14, 603–609. Nojiri, M., Uekita, K., Ohnuki, M., Taoka, N., Yasohara, Y., 2013. Microbial asymmetric hydrolysis of 3-substituted glutaric acid diamides. J. Appl. Microb. 115, 1127–1133. Palomo, J.M., Cabrera, Z., 2012. Enzymatic desymmetrization of prochiral molecules. Curr. Org. Syn. 9, 791–805. Paravidino, M., Bohm, P., Groger, H., Hanefeld, U., 2012. Hydrolysis and formation of carboxylic acids esters, 3rd ed. In: Drauz, K., Gröger, H., May, O. (Eds.), Enzyme Catalysis in Organic Synthesis, Vol. 1. Wiley-VCH, Weinheim. Park, S.E., Nam, E.H., Jang, H.B., Oh, J.S., Some, S., Lee, Y.S., Song, C.E., 2010. Enantioselective alcoholysis of meso-glutaric anhydrides catalyzed by cinchonabased sulfonamide catalysts. Adv. Synth. Catal. 352, 2211–2217. Roy, S., Chen, K.F., Gurubrahamam, R., Chen, K., 2014. Organocatalytic kinetic resolution of racemic secondary nitroallylic alcohols combined with simultaneous desymmetrization of prochiral cyclic anhydrides. J. Org. Chem. 79, 8955–8959.
48
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
Soumanou, M.M., Perignon, M., Villeneuve, P., 2013. Lipase-catalyzed interesterification reactions for human milk fat substitutes production: a review. Eur. J. Lipid Sci. Technol. 115, 270–285. Tan, T., Lu, J., Nie, K., Deng, L., Wang, F., 2010. Biodiesel production with immobilized lipase: a review. Biotechnol. Adv 28, 628–634. Tsai, S.W., 2015. Enantiopreference of Candida antarctica lipase B toward carboxylic acids: Substrate models and enantioselectivity thereof. J. Mol. Catal B: Enzym, in press, http://dx.doi.org/10.1016/j.molcatb.2014.07.010. Wang, B., Liu, J., Tang, X., Cheng, C., Gu, J., Dai, L., Yu, H., 2010a. Enzymatic synthesis of (S)-glutaric acid monoesters aided by molecular docking. Tetrahedron Lett. 51, 309–312.
Wang, P.Y., Chen, Y.J., Wu, A.C., Lin, Y.S., Kao, M.F., Chen, J.R., Ciou, J.F., Tsai, S.W., 2009. (R,S)-Azolides as novel substrates for lipase-catalyzed hydrolytic resolution in organic solvents. Adv. Synth. Catal. 351, 2333–2341. Wang, P.Y., Wu, C.H., Ciou, J.F., Wu, A.C., Tsai, S.W., 2010b. Kinetic resolution of (R,S)pyrazolides containing substituents in the leaving pyrazole for increased lipase enantioselectivity. J. Mol. Catal. B: Enzym. 66, 113–116. Yu, M.S., Lantos, I., Peng, Z.-Q., Yu, J., Cacchio, T., 2000. Asymmetric synthesis of (-)-paroxetine using PLE hydrolysis. Tetrahedron Lett. 41, 5647–5651.
Contents lists available at ScienceDirect
Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces
Two-step desymmetrization of dipyrazolidyl 3-phenylglutarate via lipase-catalyzed hydrolysis in organic solvents Po-Hao Chan, Shau-Wei Tsai n Institute of Biochemical and Biomedical Engineering, Chang Gung University, Kwei-Shan District, Tao-Yuan 33302, Taiwan, ROC
H I G H L I G H T S
G R A P H I C A L
A B S T R A C T
A sufficient condition for obtaining enantiomerically enriched product was proposed. Dipyrazolidyl 3-phenylglutarate but not the diester was employed as the substrate. Results agreed with the experimental data from the kinetic and thermodynamic analysis.
art ic l e i nf o
a b s t r a c t
Article history: Received 1 July 2015 Received in revised form 14 September 2015 Accepted 21 September 2015
A theoretical analysis on comparing the enzyme performance in a single-step desymmetrization, singlestep kinetic resolution, and two-step desymmetrization (i.e. a single-step desymmetrization followed by a sequent kinetic resolution) is reported. On the basis of ee* ≧ 0.95, a sufficient condition of E3E2 1 ≧ 10 and E1 ≧ 2 is proposed for obtaining an acceptable yield of X2R* 40.412 for the desired enantiomer in the two-step desymmetrization process, in comparison with E1 ≧ 39 for the single-step desymmetrization and E3E2 1 ≧ 20 for the single-step kinetic resolution. With the CALB-catalyzed hydrolytic desymmetrization of dipyrazolidyl 3-phenylglutarate (1) in MTBE as the model system, enantiomerically pure (R)monopyrazolidyl 3-phenylglutarate ((R)-2) can then bfdee prepared. Moreover from the kinetic analysis, the best reaction condition of using 20% water-saturated MTBE as the medium at 45 °C is selected for improving the enzyme activity and stereoselectivity. The thermodynamic analysis also indicates that the enzyme stereo-discrimination in the desymmetrization and sequent kinetic resolution is mainly entropic-driven. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Desymmetrization Kinetic resolution CALB Dipyrazolidyl 3-phenylglutarate (R)-Monopyrazolidyl 3-phenylglutarate
1. Introduction Lipases (E.C. 3.1.1.3) have been employed as the biocatalysts for synthesizing a variety of lipids, food and flavors, pharmaceuticals, fine chemicals, cosmetics, biodiesels, and polymers (Hasan et al., 2006; Kobayashi, 2009; Tan et al., 2010; Soumanou et al., 2013). A very useful feature of these enzymes is the enantiodiscrimination with which the preparation of single enantiomer of alcohol, acid, and amine is fulfilled
n
Corresponding author. Tel.: þ 886 3 2118800x3415; fax: þ886 3 2118668. E-mail address: [email protected] (S.-W. Tsai).
http://dx.doi.org/10.1016/j.ces.2015.09.024 0009-2509/& 2015 Elsevier Ltd. All rights reserved.
(Bornscheuer and Kazlauskas, 2006; Ghanem, 2007; Kamal et al., 2008; Faber, 2011; Paravidino et al., 2012). Many important intermediates and building blocks for various synthetic applications belong to diol, diamine, dicarboxylic acid, or anhydride that contains one or more prochiral or chiral centers. When using a symmetrical prochiral or meso compound as the substrate, the first-step desymmetrization followed by a subsequent kinetic resolution can generally increase the optical purity, yet with the price of decreasing the yield of the required enantiomer (Bornscheuer and Kazlauskas, 2006; Faber, 2011). Quantitative analysis in enzyme-catalyzed two-step desymmetrization of prochiral or meso compounds is more complicated than that in a single-step desymmetrization or kinetic resolution, as at least two
42
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
more kinetic parameters should be considered. This formulation remains still unclear and the relation of the theoretical to the experimental analysis has not been clearly discussed. Enantiomerically pure 3-substituted glutarates are key structural elements of many drug intermediates and building blocks in organic synthesis (Fryszkowska et al., 2005; Gopinath et al., 2012; Jung et al., 2013). Enantiomerically enriched 3-arylglutarates have been prepared from alcoholic ring-opening of cyclic anhydrides via organocatalysts or enzymes (Chaubey et al., 2008; Park et al., 2010; Roy et al., 2014; Fryszkowska et al., 2006; García-Urdiales et al., 2011; Palomo and Cabrera, 2012; Liu et al., 2014) and enzymecatalyzed hydrolysis, alcoholysis, aminolysis, or ammonolysis of dialkyl 3-arylglutarates (Yu et al., 2000; Homann et al., 2001; Lopez-Garcia et al., 2003; Cabrera et al., 2008; B. Wang et al., 2010; P.Y. Wang et al., 2010; Cabrera and Palomo, 2011; Liu et al., 2012; Nojiri et al., 2013). In general it is difficult to develop an efficient desymmetrization process leading to high enantiomeric purity and yields for the desired enantiomer at the mild reaction condition after inspecting the experimental data reported in the references of this paragraph. It is not only about the reaction conditions but also about the biocatalysts. As a part of our ongoing efforts toward using azolides as the substrate for preparing optically active compounds (Wang et al., 2009, B. Wang et al., 2010; P.Y. Wang et al., 2010; Cheng et al., 2012; Tsai, in press), we aimed to employ dipyrazolidyl 3phenylglutarate (1) as the model substrate for preparing an enantiomer of high enantiomeric purity and yield via CALBcatalyzed hydrolysis in MTBE (Scheme 1). A thorough kinetic analysis is firstly performed for proposing a sufficient condition leading to the efficient desymmetrization process giving ee* ≧ 0.95 and X2R* 40.412. From the kinetic analysis, the kinetic constants shown in Scheme 1 are estimated from experimental data for selecting the best reaction condition. The thermodynamic analysis is moreover addressed and elucidated.
CO2H O
N N
k1 N
OO
k3 CO2H CO2H
N N
N
PR ((R)-2)
S (1)
k2
k4 CO2H O
Q (3)
N N
PS ((S)-2) Scheme 1. Two-step desymmetrization of dipyrazolidyl 3-phenylglutarate in MTBE containing different water contents via lipase-catalyzed hydrolysis.
reaction. Similarly when performing a single-step kinetic resolution by using (R,S)-2 as the substrate, one obtains Eqs. (7) and (8), with which the enantiomeric excess for (R)-2 defined as ee¼(X2RX2S)(X2R þX2S) 1 may vary from zero to one. X2 R ¼
E1 ½1–expð tÞ 1 þ E1
ð5Þ
X2 S ¼
1–expð tÞ 1 þ E1
ð6Þ
X 2 R ¼ 0:5 expð E2 tÞ
ð7Þ
X 2 S ¼ 0:5 expð E3 tÞ
ð8Þ
2. Model development By using an excess of water for the hydrolysis and assuming the substrate concentrations to be much lower than the MichaelisMenten constants, an irreversible first-order kinetics for 1, (R)-2, and (S)-2 can be derived and solved for the time-course molar fractions as follows (Faber, 2011): X 1
¼ expð tÞ
ð1Þ
X2 R ¼
E1 ½expð E2 tÞ–expð tÞ ð1 þE1 Þð1 E2 Þ
ð2Þ
X2 S ¼
expð E3 tÞ–expð tÞ ð1 þ E1 Þð1 E3 Þ
ð3Þ
X 3 ¼ 1 X 1 X 2 R X 2 S
ð4Þ
The dimensionless parameters are defined as: t* ¼(k1 þk2)t as dimensionless time, E1 ¼k1k2 1 as stereoselectivity for the single-step desymmetrization, E2 ¼k3(k1 þk2) 1, E3 ¼k4(k1 þk2) 1, ee* ¼ (X2R*-X2S*)(X2R* þX2S*) 1 as enantiomeric excess for (R)-2, and hence E3E2 1 ¼k4k3 1 as enantiomeric ratio for the second-step kinetic resolution. The kinetic parameter combination (k1 þk2) may be firstly estimated from the regression of time-course data of X1* to Eq. (1), and then k2 and k4 from the time-course X2S* and Eq. (3). With the known k1 value, k3 is then regressed from the time-course X2R* and Eq. (2). Apparently for E2 ¼E3 ¼0, Eqs. (2) and (3) reduce to Eqs. (5) and (6) for a single-step desymmetrization, and lead to a constant enantiomeric excess of ee* ¼ (E1-1)(E1 þ1) 1 throughout the
3. Materials and methods 3.1. Materials Novozym 435 (Candida antarctica lipase B (CALB) immobilized on acrylic resins, containing 1–2% (w/w) water and has 7000 PLU/g by using lauric acid and 1-propanol as substrates at 60 °C) was purchased from Novozymes (Bagsvaerd, Denmark). 1 PLU is the amount of enzyme activity which generates 1 μmol of propyl laurate per minute under the defined conditions. Other chemicals of analytical grade were commercially available: pyrazole and 1 Hbenzotriazole from Acros (Geel, Belgium); thionyl chloride from Seedchem (Camberwell, Australia); chloroform-D from Cambridge Isotope Laboratories (Andover, MA); calcium hydride from SigmaAldrich (St. Louis, MO); acetic acid glacial (AA), benzene, isopropanol (IPA), methyl tert-butyl ether (MTBE), n-hexane (HEX) and triethylamine from Tedia (Fairfield, OH); 3-phenylglutaric acid from Alfa (Ward Hill, MA). 100% Water-saturated MTBE was prepared from a biphasic aqueous-MTBE solution kept in a water bath of specified temperature and with stirring for more than 24 h. Anhydrous MTBE was made by adding calcium hydride to the solvent for 24 h. Both solvents were then employed for preparing MTBE containing different water contents. For example, 20% water-saturated MTBE containing 94.5 mM of water at 45 °C was prepared by mixing 1:4 (v/v) of 100% water-saturated MTBE containing 472.7 mM of water and anhydrous MTBE (Alkandary et al., 2001).
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
3.2. Substrate preparation and analysis To 5 mL benzene containing 1.0 mmol 3-phenylglutaric acid, 2.4 mmol pyrazole and 5 mmol triethylamine was added dropwise a mixture containing 1 mL benzene and 2.5 mmol thionyl chloride at 0 °C, and then kept at room temperature with stirring for 4 h. The resultant mixture was stored at 4 °C for 30 min, and then quenched in succession with 10 mL of 10 mM HCl solution (0.1 M), NaOH solution (0.1 M), and NaCl solution (0.1 M) for 15 min for three times. The organic phase was separated, dried over anhydrous MgSO4 for 12 h, filtered, and concentrated under reduced pressure for giving an oily dipyrazolidyl 3-phenylglutarate (1). The 1 H NMR spectrum on Bruker Avance DRX 500 spectrometer in CDCl3 solution with tetramethylsilane as an internal standard was recorded and illustrated in Fig. S1. The hydrolysis of 1, (R)-2, and (S)-2 was monitored by HPLC using a Chiralcel AS-H column from Daicel (Tokyo, Japan). UV detection at 220 nm, 1 H-benzotriazole of 20 mM as the internal standard (IS), and a mobile phase consisting of n-hexane/isopropanol/acetic acid glacial ¼ 86.8/13/0.2 (v/v) at a flow rate of 1.2 mL/min were employed for quantification at ambient temperature. 1, 3, (R)-2, (S)-2, and the internal standard were eluted with the retention time of 9.06, 11.28, 12.67, 14.9, and 19.07 min, respectively, as demonstrated in Fig. S2. 3.3. Kinetic analysis and product separation Unless specified, to 5 mL of 100% water-saturated MTBE containing 10 mM of 1 was added 20 mg/mL of Novozym 435 at 45 °C. The resultant solution was stirred with a magnetic stirrer at 400 rpm. Samples were removed at different time intervals and injected onto the HPLC for analysis. The time-course molar fractions X1* and X3* were first calculated via the calibration curves for 1 and 3 (Fig. S3). From the area ratio of (R)-2 and (R)-2 in HPLC spectrograms and (X2R* þX2S*) calculated from the total material balance for all compounds, the time-course X2R* and X2S* and hence enantiomeric excess ee* were calculated. After depleting the substrate and removing the enzyme via filtration, the resultant solution was quenched with 10 mL of KH2PO4 buffers (pH 5 and 25 mM) for 2 h at 4 °C for three times. The organic phase was then separated, dried over anhydrous MgSO4 for 12 h, filtered, and concentrated under reduced pressure for obtaining (R)-2 of 54% yield and ee* ¼0.96 as a yellowish powder. The analysis was furthermore carried out in MTBE containing different water contents. Effects of substrate concentrations varied from 10.6 and 40.8 mM in 20% water-saturated MTBE were repeated at 45 °C. Unfortunately owing to the solubility limitation of substrate 1 in MTBE at 45 °C, we did not employ substrate concentrations higher than 40.8 mM for the reaction. Similarly in order to study the temperature influence on enzyme performance, the reactions in 60% water-saturated MTBE were also performed at 25 and 35 °C.
4. Results and discussion 4.1. Theoretical modeling The sufficient condition for carrying out an effective desymmetrization process followed by a sequent kinetic resolution for obtaining (R)-2 is that the kinetic parameters should follow the order: k1 ≧ k2, k1 4k3, and k4 4k3 and hence E1 ≧ 1, E3E2 1 41, and E2(E1 þ1)E1 1 o1 (or k2 ≧ k1, k2 ≧ k4, and k3 4 k4 and hence E1 ≦ 1, E3E2 1 o1, and E2(E1 þ1)E1 1 41 if (S)-2 is desired). If the reaction condition can only lead to a single-step desymmetrization, results of E1(E1 þ 1) 1≧X2R* ≧ 0 and ee* ¼(E1-1)(E1 þ1) 1 throughout the
43
reaction are obtainable. Therefore when an optical purity of ee*≧ 0.95 is required, the selectivity for the desymmetrization E1≧39 is needed. Otherwise, one should adjust the reaction condition such that the second-step kinetic resolution can follow and leads to ee* ≧ (E1-1)(E1 þ1) 1 and E1(E1 þ1) 1≧X2R* ≧ 0. Apparently, the penalty of giving ee* values higher than (E1-1)(E1 þ 1) 1 is to reduce the yield of X2R*. The disadvantages of only carrying out the second-step kinetic resolution for (R,S)-2 include X2R ≦ 0.5 and conversions to be higher than 0.5 in order to obtain high ee values. It is well recognized in academy and industry that the acceptable enantiomeric ratio k4k3 1 should be greater than 20, and hence giving X2R 40.412 calculated from k4k3 1ln[(X2R þX2S)(1 þee)]¼ ln [(X2R þX2S)(1-ee)] for ee¼0.95. When the ratio is less than 20, it is imperative to add a single-step desymmetrization before the kinetic resolution for increasing X2R*. Fig. 1A and B illustrates ee* varied with (X2R* þX2S*), ee varied with (X2R þX2S), and time-course (X2R* þX2S*) and (X2R þX2S) for E1 ¼2, E3E2 1 ¼ 2, and k4k1 1 ¼E3(E1 þ1)E1 1 equal to 0.1, 1, and 2, respectively. At the low E1 value, it is still possible to improve the ee* value from 0.333 for a single-step desymmetrization to 0.95, yet giving very low X2R* ¼0.0343 at t* ¼90 for E2 ¼0.0333 and E3 ¼0.0666 (i.e. E3(E1 þ1)E1 1 ¼0.1). Increasing of E2 ¼0.333 and E3 ¼0.666 (i.e. E3(E1 þ1)E1 1 ¼1) as well as E2 ¼ 0.666 and E3 ¼1.333 (i.e. E3(E1 þ1)E1 1 ¼2) results in even lower X2R* values for ee* ¼ 0.95, yet with the benefit of decreasing the required dimensionless time. A change of E3E2 1 to 10 may improve X2R* value to 0.481 at t* ¼50 and ee* ¼0.948 for E2 ¼ 0.00666 and E3 ¼0.0666 (Figs. 1C and 1D). Similarly by further increasing E2 and E3 to 0.0666 and 0.666, respectively, a slight reduction of X2R* ¼0.477 at t* ¼6.0 and ee* ¼0.979 is found (or X2R* ¼ 0.413 at t*¼4.5 and ee* ¼0.959 for E2 ¼ 0.133 and E3 ¼ 1.33). By moreover increasing E3E2 1 to 50 (Figs. 1E and 1F), an enhancement of X2R* ¼ 0.629 at ee* ¼0.945 and t* ¼ 45 for E2 ¼0.00133 and E3 ¼ 0.0666 is calculated. Similarly for a higher E2 ¼0.0133 and E3 ¼0.666, only a slight reduction of X2R* ¼0.622 at t* ¼6.0 and ee* ¼ 0.950 is found (or X2R* ¼ 0.603 at t* ¼4.0 and ee* ¼ 0.956 for E2 ¼0.0266 and E3 ¼ 1.33). As the enantiomeric ratio E3E2 1 is now higher than 20, it is interesting to compare the results when only carrying out a single-step kinetic resolution. In general one obtains X2R* 4X2R at high E1, ee*, and ee values, e.g. X2R ¼0.471 and ee¼0.961 at t* ¼60 for E2 ¼0.00133 and E3 ¼0.0666, at t* ¼ 6.0 for E2 ¼ 0.0133 and E3 ¼0.666, as well as at t* ¼3.0 for E2 ¼0.0266 and E3 ¼1.333. At E3E2 1 ¼2 (Figs. S4A and S4B), one only obtains X2R* ¼0.244 and ee* ¼0.979 at t* ¼ 30 for E2 ¼0.0455 and E3 ¼0.0909, X2R* ¼ 0.161 and ee* ¼0.952 at t* ¼5.0 for E2 ¼ 0.454 and E3 ¼0.909, as well as X2R* ¼ 0.0803 and ee* ¼ 0.953 at t* ¼4.0 for E2 ¼0.909 and E3 ¼ 1.818. More improvements of X2R* at the higher selectivity are perceived, e.g. at E3E2 1 ¼10 in Figs. S4C and S4D, one obtains X2R* ¼0.779 and ee* ¼ 0.951 at t* ¼ 18 for E2 ¼0.00909 and E3 ¼0.0909, X2R* ¼0.711 and ee* ¼0.957 at t* ¼3.0 for E2 ¼0.0909 and E3 ¼0.909, as well as X2R* ¼0.622 and ee* ¼0.962 at t* ¼2.0 for E2 ¼0.182 and E3 ¼1.818. Apparently at an even higher value of E3E2 1 ¼50 (Figs. S4E and S4F), more favorable results are demonstrated, e.g. X2R* ¼0.885 and ee* ¼ 0.949 at t* ¼ 16 for E2 ¼ 0.00182 and E3 ¼0.0909, X2R* ¼ 0.809 and ee* ¼0.949 at t* ¼2.5 for E2 ¼0.0182 and E3 ¼0.909, as well as X2R* ¼0.683 and ee* ¼0.950 at t* ¼1.5 for E2 ¼0.364 and E3 ¼1.818. Therefore for a desired high ee* value, E1 in general has a more impact on decreasing the required dimensionless time, yet with E3E2 1 having more influence on enhancing the yield of X2R*. These arguments are perceived if one compares the results at ee* ¼0.95 for E1 ¼ 2 and E3E2 1 ¼10 with those for E1 ¼10 and E3E2 1 ¼2. By moreover comparing the results with those for E1 o2 and E3E2 1 ¼10, a sufficient condition of E3E2 1≧10 and E1≧2 is hence
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
*) or (X 2R + X 2S)
44
1.0
0.1
2
0.4
1
2S
0.6
*+X
ee* or ee
0.8
(X
2R
*+X
2S
* ) or ( X 2R + X
(X
0.0 0.2 0.4 0.6 0.8 1.0 2S
1.0
1
2
0.1
2S
0.4
2R
0.2
(X
2R
*+X
2S
* ) or ( X 2R + X
2S
(X
0.0 0.2 0.4 0.6 0.8 1.0
1
0.1
2S
0.4
*+X
ee* or ee
2
0.6
2R
0.2 0.0 0.2 0.4 0.6 0.8 1.0
(X
2R
*+X
2S
* ) or ( X 2R + X
2S
(X
0.0
1
0.4 0.2 0.0
2 0
5
10
1.0
)
15
20
15
20
15
20
0.1
0.8 0.6
1
0.4
2
0.2 0.0
0
5
10
)
1.0 0.8
0.6
t*
*) or (X 2R + X 2S)
0.0
0.1
t*
*+X
ee* or ee
0.8 0.6
0.8
) *) or (X 2R + X 2S)
0.0
2R
0.2
1.0
1.0
0.1
0.8
1
0.6
2
0.4 0.2 0.0
0
5
10
t*
Fig. 1. Theoretical ee* (——) and ee (— -) varied with (X2R* þ X2S*) and (X2R þ X2S), respectively, for (A) E1 ¼ 2, E3/E2 ¼2, (C) E1 ¼2, E3/E2 ¼ 10, and (E) E1 ¼ 2, E3/E2 ¼50. Timecourse (X2R* þX2S*) (——) and (X2R þ X2S) (— -) for (B) E1 ¼2, E3/E2 ¼ 2, (D) E1 ¼ 2, E3/E2 ¼10, and (F) E1 ¼ 2, E3/E2 ¼50.
proposed for leading to an efficient two-step desymmetrization process. 4.2. Kinetic analysis The time-course X1*, X2R*, X2S*, X3*, and ee* in Fig. 2A, as well as X2R*, ee*, and (X2R* X2R) varied with (X2R* þX2S*) and ee varied with (X2R þX2S) in Fig. 2B are illustrated, when Scheme 1 is carried out in 100% water-saturated MTBE containing 472.7 mM of water at 45 °C. The kinetic constants estimated and represented in Table 1 are employed for calculating the best-fitted results in agreements with the experimental data in the figures. In comparison with
X2R ¼ 0.266 and ee40.99 at t¼ 70.3 h for only performing a singlestep kinetic resolution, a lower reaction time of 60 h is required for giving X2R* ¼0.255 and ee* 40.99. Moreover, the time-course X2R, X2S, ee, and (X2R* X2R) are calculated and represented in Fig. 2B, giving a maximum enhancement of (X2R* X2R) ¼0.192 for X2R* ¼0.437 and ee* ¼0.95 at t¼ 30.4 h. An investigation of the kinetic constants indicates that the higher value of E3E2 1 ¼9.4, in comparison with E1 ¼3.3 and k3k2 1 ¼1.25, is mainly contributed for giving k4k1 1 ¼ 3.54. Yet no explanation is found for elucidating why the pyrazolidyl moiety in 1 and the acidic group in (R)-2 and (S)-2 can induce different steric hindrances for substrate affinity to
1.0 0.8 0.6 0.4 0.2 0.0
0
20
40
60
80
X 2R*, ee*, (X 2R* - X 2R), ee
X 1*, X 2R*, X 2S*, X 3*, ee*
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
45
1.0 0.5 0.0 -0.5
0.0 0.2 0.4 0.6 0.8 1.0
(X 2R* + X 2S*) or ( X 2R + X 2S)
t (h)
Fig. 2. (A) Time-course X1* (●), X2R* (⎕), X2S* (Δ), X3* (▲), and ee* (∇). (B) Variations of X2R* (⎕), ee* (∇), and (X2R*-X2R) (— -) with (X2R* þ X2S*), and ee (— —) varied with (X2R þX2S). Reaction condition: 100% saturated-water MTBE, 20 mg/mL Novozym 435, (S)0 ¼9.2 mM, and 45 °C. (——) Best-fitted results from Eqs. (1)–(4).
Table 1 Effects of water content in MTBE on kinetic parameters for desymmetrization of 1 at 45 °C. Water content (% saturation)
(k1 þ k2) (1/h) k1 (1/h)
10 20 60 100
1.11 9.46E 1 1.23E 1 8.12E 2
9.60E 1 8.17E-1 9.69E 2 6.24E 2
k2 (1/h)
k3 (1/h)
k4 (1/h)
E1 ( ¼ k1k2 1) E3E2 1 ( ¼ k4k3 1)
E2(E1 þ 1) ( ¼ k3k2-1)
E3(E1 þ 1)E1 1 ( ¼ k4k1 1)
1.51E 1 1.29E 1 2.61E 2 1.88E 2
7.87E 2 5.15E 2 2.90E 2 2.35E 2
1.14 8.75E 1 2.97E 1 2.21E 1
6.3 6.3 3.7 3.3
0.52 0.40 1.11 1.25
1.18 1.07 3.06 3.54
14.5 17.0 10.2 9.4
Reaction condition: 20 mg/mL Novozym 435 and (S)0 around 10 mM. Symbol of E 1 as 10 1.
the enzyme acyl pocket and proton transfers from the triad imidazolium to the leaving pyrazolidyl group. The desymmetrization in 60% water-saturated MTBE is also demonstrated in Fig. S5, and employed for estimating the kinetic constants (Table 1). Agreements of the best-fitted results with experimental data are perceived from the figure. Apparently, the reduction of the water content has improved the enzyme activity but not the selectivity, i.e. 1.23- to 1.55-fold enhancements for all kinetic constants leading to nearly invariable E1 ¼3.7 and E3E2 1 ¼10.2. The lipase performances are further improved if the two-step desymmetrization is carried out in 20% water-saturated MTBE (Fig. 3A and B). Similarly the estimated kinetic constants (Table 1) are employed for calculating the best-fitted results and show agreements with the experimental data. Enhancements of the kinetic constant do yield E1 ¼6.3 and E3E2 1 ¼17, giving favorable results of X2R* ¼0.667 and ee* 40.99 at t ¼6 h as well as (X2R* X2R)¼0.315 for X2R* ¼0.722, and ee* ¼0.963 at t¼4 h. Although the employed substrate and reaction condition are different, a better enzyme performance for our data is concluded after comparing with the results of X2R* ¼ 0.55 and ee* 40.99 at 240 h when dimethyl 3-phenylglutarate of 25 mM was hydrolyzed in pH 5 sodium phosphate containing 300 mg/mL Novozym 435 at 25 °C (Cabrera et al., 2009), as well as X2S* ¼0.42 and ee* ¼ 0.96 at 144 h when dimethyl 3-phenylglutarate of 250 mM was ammonolyzed in MTBE containing 30 mg/mL Novozym 435 at 30 °C (Lopez-Garcia et al., 2003). The time-course data in 10% water-saturated MTBE are furthermore illustrated in Fig. S6. The kinetic constants then estimated and tabulated in Table 1 are employed for preparing the best-fitted curves that are in agreement with the experimental data as demonstrated in the figure. A slight enhancement for each kinetic constant, but not the stereoselectivity, is perceived and leads to X2R* ¼0.624 and ee* 4 0.99 at t ¼5 h. Therefore, 20% water-
saturated MTBE is selected as the best reaction medium by considering water also acting as the substrate for hydrolysis. When anhydrous and 100% water-saturated MTBE were, respectively, employed as the reaction media for hydrolyzing (R,S)2-phenylpropionic pyrazolide (or 1,2,4-triazolide), an order-ofmagnitude higher Michaelis-Menten constants for the later were reported (Wang et al., 2009; B. Wang et al., 2010; P.Y. Wang et al., 2010). This implies that water adsorbed on the active site region may impede substrate affinity to the binding pockets, and leads to a higher Michaelis–Menten constant for each enantiomer in 100% water-saturated MTBE. Yet, the water content also strongly affects the proton transfer and hence the nucleophilic attack of triad serine to the carbonyl moiety. This can result in higher lipase activity in anhydrous MTBE, just like the monotonic increase of kinetic constants (Table 1) when decreasing the water content. However it is still difficult to elucidate k3k2 1 and k4k1 1 varied with the water content when a pyrazolidyl moiety in 1 is replaced with the acidic group in (R)-2 and (S)-2. In order to check if the irreversible first-order kinetics is still valid for high substrate concentrations, more experiments in 20% watersaturated MTBE at 45 °C were carried out. By using the estimated kinetic constants of k1 and k2 from the time-course data (not shown here), the initial rates for converting 1 to (R)-2 and (S)-2 are calculated and illustrated in Fig. 4. One may then estimate the kinetic parameter combination of (Km,1 1 þ Km,2 1)¼1.458 10 2 mM 1 from Fig. 4, k1 and k2 for 20% water-saturated MTBE in Table 1, and Eqs. (S3) and (S4) of Supplementary data. Agreements between the experimental data and best-fitted results calculated from V2R ¼ 68.13 (S)0[68.56þ(S)0] and V2S ¼ 14.80(S)0[68.56þ(S)0] are also illustrated in the figure. Moreover if the same Michaelis-Menten constants are assumed, one obtains Km,1 ¼Km,2 ¼137.1 mM, implying that it is reasonable to assume the irreversible first-order kinetics when the employed substrate concentration is below 20 mM. Otherwise, more complicated rate equations such as those derived from Michaelis-
1.0 0.8 0.6 0.4 0.2 0.0
0
1
2 3 t (h)
4
5
X 2R*, ee*, (X 2R* - X 2R), ee
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
X 1*, X 2R*, X 2S*, X 3*, ee*
46
1.0 0.5 0.0 -0.5
0.0 0.2 0.4 0.6 0.8 1.0
(X 2R* + X 2S*) or ( X 2R + X 2S)
Fig. 3. (A) Time-course X1* (●), X2R* (⎕), X2S* (Δ), X3* (▲), and ee* (∇). (B) Variations of X2R* (⎕), ee* (∇), and (X2R*-X2R) (—) with (X2R* þX2S*), and ee (— —) varied with (X2R þX2S). Reaction condition: 20% saturated-water MTBE, 20 mg/mL Novozym 435, (S)0 ¼ 10.6 mM, and 45 oC. (——) Best-fitted results from Eqs. (1)–(4).
(K mol) are calculated. By comparing the absolute values of ΔΔH and TΔΔS at a specific temperature, the stereo-discrimination for the productive 1 and (S)-2 converting to (R)-2 and 3, respectively, is entropic-driven with compensation from the enthalpy loss.
-1
V2R, V2S (mM h )
40
30
20
5. Conclusions
10
0
0
10
20
30
40
50
(S)0 ( m M ) Fig. 4. Initial rates of (R)-2 (∇) and (S)-2 (○) varied with (S)0. Reaction condition: 20% water-saturated MTBE, 20 mg/mL Novozym 435, and 45 oC. (——) Best-fitted results from Eqs. (S3) and (S4).
Menten kinetics for all reaction steps in Scheme 1 are needed for modeling the whole reaction. 4.3. Thermodynamic analysis In order to check if the enzyme stereoselectivity can be improved at low temperature, more experiments in 60% watersaturated MTBE were carried out at 25 and 35 °C. The kinetic constants estimated from the time-course data (not shown here) are tabulated in Table 2. About 4.5-fold increments for each kinetic constant, and hence yielding nearly invariable E1 and E3E2 1, are found when increasing the temperature from 25 to 45 °C. Therefore, we have selected 45 °C as the best temperature for giving the highest enzyme activity. The good linear relationships between logarithms of kinetic constant and the inverse of absolute temperature, i.e. ln[k1]¼ 7187 T 1 þ20.30, ln[k2]¼ 6896 T 1 þ 18.05, ln[k3]¼ 7042 T 1 þ 18.62, and ln[k4]¼ 7670 T 1 þ22.90, are furthermore illustrated in Fig. 5. Therefore from ln[E1]¼ 291 T 1 þ2.25 for converting 1 to (R)-2 and (S)-2, the differences of enthalpy and entropy between the transition states for both substrates are estimated as ΔΔH¼ 2.42 kJ/mol and –ΔΔS¼ 18.7 J/(K mol) (Wang et al., 2009). Similarly from ln[E3E2 1]¼ 628 T 1 þ 4.28 for converting (R)-2 and (S)-2 to 3, ΔΔH¼ 5.22 kJ/mol and ΔΔS¼-35.58 J/
The theoretical analysis for comparing the enzyme performances in a single-step desymmetrization, a single-step kinetic resolution, and the two-step desymmetrization for preparing chiral compounds is investigated, in which the dimensionless groups of E1, E3E2 1, and E3(E1 þ1)E1 1 are employed as the parameters. An effective two-step desymmetrization process should fulfill the sufficient condition of E1 ≧ 1, E3E2 1 41, and E2(E1 þ1)E1 1 o1. Moreover in order to obtain an acceptable yield with ee* ≧ 0.95 for (R)-2, the quantitative condition of E3E2 1 ≧ 10 and E1≧2 is furthermore proposed. Optically pure (R)-monopyrazolidyl 3-phenylglutariate can be prepared from the hydrolytic desymmetrization of dipyrazolidyl 3-phenylglutarate via Candida antarctica lipase B, in which the best reaction condition of 20% water-saturated MTBE at 45 °C, leading to E1 ¼ 6.3 and E3E2 1 ¼17.0, is selected from the kinetic analysis. The thermodynamic analysis moreover indicates that the enzyme stereodiscrimination is mainly entropic-driven in the desymmetrization and sequent kinetic resolution, giving nearly invariable E1 and E3E2 1 values when changing the reaction temperature.
Nomenclature selectivity defined as k1k2 1 dimensionless parameters defined as k3(k1 þk2) 1and k4(k1 þk2) 1, respectively ee enantiomeric excess defined as (X2R X2S)(X2R þX2S) 1 Km,1, Km,2 Michaelis–Menten constants leading to (R)-2 and (S)-2 from 1, respectively (mM) k1, k2,k3, k4 kinetic constants shown in Schemes 1 (h 1) (S)0 initial concentration for dipyrazolide substrate (mM) T absolute temperature (K) t time (h) t* dimensionless time defined as (k1 þ k2)t V2R, V2S initial rates for (R)-2 and (S)-2 from 1, respectively (mM/ h) Vmax,1, Vmax,2 maximum velocity leading to (R)-2 and (S)-2 from 1, respectively (mM/h) E1 E2, E3
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
47
Table 2 Temperature influence on kinetic parameters for desymmetrization of 1 in 60% water-saturated MTBE. Temp (°C)
(k1 þk2) (1/h)
k1 (1/h)
k2 (1/h)
k3 (1/h)
k4 (1/h)
E1 ( ¼ k1k2 1)
E3E2 1 (¼ k4k3 1)
E2(E1 þ 1) ( ¼ k3/k2)
E3(E1 þ1)E1 1 (¼ k4/k1)
25 35 45
2.74E 2 6.70E 2 1.23E 1
2.13E 2 5.32E 2 9.69E 2
6.12E 3 1.39E 2 2.61E 2
6.59E 3 1.55E 2 2.90E-2
5.90E 2 1.39E 1 2.97E 1
3.5 3.8 3.7
9.0 9.0 10.2
1.07 1.15 1.11
2.77 2.61 3.06
-1
ln[k1], ln[k2], ln[k3], ln[k4], ln[E1], ln[E3E2 ]
Reaction condition: 20 mg/mL Novozym 435, (S)0 around 10 mM, and 45°C. Symbol of E 1 as 10 1.
3
0
-3
-6
3.1
3.2
3.3 -1
3.4
-1
1000T (K ) Fig. 5. Linear variations of ln(k1) (○), ln(k2) (Δ), ln(k3) (⎕), ln(k4) (∇), ln(E1) (●), and ln(E3E2 1) (▲) with inverse of absolute temperature. Reaction condition: 60% watersaturated MTBE, 20 mg/mL Novozym 435, and (S)0 ¼ 11.1 mM.
X1, X2R,X2S, X3 molar fractions based on (S)0 for 1, (R)-2 and (S)-2, and 3, respectively ΔΔH enthalpy difference for two transition states (kJ/mol) ΔΔS entropy differences for two transition states (J/(mol K)) Superscript * for a single-step or two-step desymmetrization Subscript max
as maximum value
Acknowledgments Financial supports of NSC 104-2221-E-182-058-MY2 from Ministry of Science and Technology are appreciated.
Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.ces.2015.09.024.
References Alkandary, J.A., Aljimaz, A.S., Fandary, M.S., Fahim, M.A., 2001. Liquid-liquid equilibria of waterþ MTBEþ reformate. Fluid Phase Equilib. 187-188, 131–138.
Bornscheuer, U.T., Kazlauskas, R.J., 2006. Hydrolases in organic synthesis: regioand stereoselective biotransformations, 2nd ed. Wiley-VCH, Weinheim. Cabrera, Z., Fernandez-Lorente, G., Palomo, J.M., Guisan, J.M., Fernandez-Lafuente, R., 2008. Asymmetric hydrolysis of dimethyl 3-phenylglutarate catalyzed by s Lecitase Ultra . Effect of the immobilization protocol on its catalytic properties. Enzyme Microb. Technol. 43, 531–536. Cabrera, Z., Fernandez-Lorente, G., Fernandez-Lafuente, R., Palomo, J.M., Guisan, J. M., 2009. Novozym 435 displays very different selectivity compared to lipase from Candida antarctica B adsorbed on other hydrophobic supports. J. Mol. Catal. B: Enzym. 57, 171–176. Cabrera, Z., Palomo, J., 2011. Enantioselective desymmetrization of prochiral diesters catalyzed by immobilized Rhizopus oryzae lipase. Tetrahedron Asymmetr 22, 2080–2084. Chaubey, N., Date, S.M., Ghosh, S.K., 2008. An efficient asymmetric desymmetrization of prochiral glutaric anhydrides with SuperQuat chiral oxazolidin-2ones. Tetrahedron Asymmetry 19, 2721–2730. Cheng, Y.L., Wu, A.C., Wang, P.Y., Tsai, S.W., 2012. Kinetic and thermodynamic analysis for lipase-catalyzed hydrolytic resolution of (R,S)-alcohols though their azolyl carbamates. Bioprocess Biosyst. Eng. 35, 953–962. Faber, K., 2011. Biotransformations in organic chemistry: a textbook, 6th ed. Springer-Verlag, New York. Fryszkowska, A., Komar, M., Koszelewski, D., Ostaszewski, M., 2005. Enzymatic desymmetrization of 3-arylglutaric acid anhydrides. Tetrahedron Asymmetry 16, 2475–2485. Fryszkowska, A., Komar, M., Koszelewski, D., Ostaszewski, R., 2006. Studies on enzymatic synthesis of chiral non-racemic 3-arylglutaric acid monoesters. Tetrahedron Asymmetry 17, 961–966. García-Urdiales, E., Alfonso, I., Gotor, V., 2011. Update 1 of: enantioselective enzymatic desymmetrizations in organic synthesis. Chem. Rev. 111, PR110–PR180. Ghanem, A., 2007. Trends in lipase-catalyzed asymmetric access to enantiomerically pure/enriched compounds. Tetrahedron 63, 1721–1754. Gopinath, P., Watanabe, T., Shibasaki, M., 2012. Studies on catalytic enantioselective total synthesis of caprazamycin B: construction of the western zone. J. Org. Chem. 77, 92605–99267. Hasan, F., Shah, A.A., Hameed, A., 2006. Industrial applications of microbial lipases. Enzyme Microb. Technol. 39, 235–251. Homann, M.J., Vail, R., Morgan, B., Sabesan, V., Levy, C., Dodds, D.R., Zaks, A., 2001. Enzymatic hydrolysis of a prochiral 3-substituted glutarate ester, an intermediate in the synthesis of an NK1/NK2 dual antagonist. Adv. Synth. Catal. 343, 744–749. Jung, J.-H., Yoon, D.-H., Kang, P., Lee, W.K., Eum, H., Ha, H.-J., 2013. CAL-B catalyzed desymmetrization of 3-alkylglutarate: "olefin effect" and asymmetric synthesis of pregabalin. Org. Biomol. Chem. 11, 3635–3641. Kamal, A., Azhar, M.A., Krishnaji, T., Malik, M.S., Azeeza, S., 2008. Approaches based on enzyme mediated kinetic to dynamic kinetic resolutions: a versatile route for chiral intermediates. Coord. Chem. Rev. 252, 569–592. Kobayashi, S., 2009. Recent developments in lipase-catalyzed synthesis of polyesters. Macromol. Rapid Commun. 30, 237–266. Liu, W., Hu, Y., Jiang, L., Zou, B., Huang, H., 2012. Synthesis of methyl (R)-3-(4fluorophenyl)glutarate via enzymatic desymmetrization of a prochiral diester. Proc. Biochem. 47, 1037–1041. Liu, W., Hu, Y., Zhang, Y., Ma, Y., Huang, H., 2014. Enzymatic desymmetrization of 3(4-fluorophenyl)glutaric anhydride through enantioselective alcoholysis in organic solvents. Biotechnol. Bioprocess. Eng. 19, 449–455. Lopez-Garcia, M., Alfonso, I., Gotor, V., 2003. Desymmetrization of dimethyl 3substituted glutarates through enzymatic ammonolysis and aminolysis reactions. Tetrahedron Asymmetry 14, 603–609. Nojiri, M., Uekita, K., Ohnuki, M., Taoka, N., Yasohara, Y., 2013. Microbial asymmetric hydrolysis of 3-substituted glutaric acid diamides. J. Appl. Microb. 115, 1127–1133. Palomo, J.M., Cabrera, Z., 2012. Enzymatic desymmetrization of prochiral molecules. Curr. Org. Syn. 9, 791–805. Paravidino, M., Bohm, P., Groger, H., Hanefeld, U., 2012. Hydrolysis and formation of carboxylic acids esters, 3rd ed. In: Drauz, K., Gröger, H., May, O. (Eds.), Enzyme Catalysis in Organic Synthesis, Vol. 1. Wiley-VCH, Weinheim. Park, S.E., Nam, E.H., Jang, H.B., Oh, J.S., Some, S., Lee, Y.S., Song, C.E., 2010. Enantioselective alcoholysis of meso-glutaric anhydrides catalyzed by cinchonabased sulfonamide catalysts. Adv. Synth. Catal. 352, 2211–2217. Roy, S., Chen, K.F., Gurubrahamam, R., Chen, K., 2014. Organocatalytic kinetic resolution of racemic secondary nitroallylic alcohols combined with simultaneous desymmetrization of prochiral cyclic anhydrides. J. Org. Chem. 79, 8955–8959.
48
P.-H. Chan, S.-W. Tsai / Chemical Engineering Science 139 (2016) 41–48
Soumanou, M.M., Perignon, M., Villeneuve, P., 2013. Lipase-catalyzed interesterification reactions for human milk fat substitutes production: a review. Eur. J. Lipid Sci. Technol. 115, 270–285. Tan, T., Lu, J., Nie, K., Deng, L., Wang, F., 2010. Biodiesel production with immobilized lipase: a review. Biotechnol. Adv 28, 628–634. Tsai, S.W., 2015. Enantiopreference of Candida antarctica lipase B toward carboxylic acids: Substrate models and enantioselectivity thereof. J. Mol. Catal B: Enzym, in press, http://dx.doi.org/10.1016/j.molcatb.2014.07.010. Wang, B., Liu, J., Tang, X., Cheng, C., Gu, J., Dai, L., Yu, H., 2010a. Enzymatic synthesis of (S)-glutaric acid monoesters aided by molecular docking. Tetrahedron Lett. 51, 309–312.
Wang, P.Y., Chen, Y.J., Wu, A.C., Lin, Y.S., Kao, M.F., Chen, J.R., Ciou, J.F., Tsai, S.W., 2009. (R,S)-Azolides as novel substrates for lipase-catalyzed hydrolytic resolution in organic solvents. Adv. Synth. Catal. 351, 2333–2341. Wang, P.Y., Wu, C.H., Ciou, J.F., Wu, A.C., Tsai, S.W., 2010b. Kinetic resolution of (R,S)pyrazolides containing substituents in the leaving pyrazole for increased lipase enantioselectivity. J. Mol. Catal. B: Enzym. 66, 113–116. Yu, M.S., Lantos, I., Peng, Z.-Q., Yu, J., Cacchio, T., 2000. Asymmetric synthesis of (-)-paroxetine using PLE hydrolysis. Tetrahedron Lett. 41, 5647–5651.