Rational Regulation of Reaction Specificity of Nitrilase for Efficient Biosynthesis of 2-Chloronicotinic Acid through a Single Site Mutation

ABSTRACT

Nitrilase-catalyzed hydrolysis of 2-chloronicotinonitrile (2-CN) is a promising approach for the efficient synthesis of 2-chloronicotinic acid (2-CA). The development of nitrilase with ideal catalytic properties is crucial for the biosynthetic route with industrial potential. Herein, a nitrilase from Rhodococcus zopfii (RzNIT), which showed much higher hydration activity than hydrolysis activity, was designed for efficient hydrolysis of 2-CN. Two residues (N165 and W167) significantly affecting the reaction specificity were precisely identified. By tuning these two residues, a single mutation of W167G with abolished hydration activity and 20-fold improved hydrolysis activity was obtained. Molecular dynamics simulation and molecular docking revealed that the mutation generated a larger binding pocket, causing the substrate 2-CN to bind more deeply in the pocket and form a delocalized π bond between the residues W190 and Y196, which reduced the negative influence of steric hindrance and electron effect caused by chlorine substituent. With mutant W167G as biocatalyst, 100 mM 2-CN was exclusively converted into 2-CA within 16 h. The study provides useful guidance in nitrilase engineering for simultaneous improvement of reaction specificity and catalytic activity, which are highly desirable in value-added carboxylic acids production from nitriles hydrolysis.

IMPORTANCE 2-CA is an important building block for agrochemicals and pharmaceuticals with a rapid increase in demand in recent years. It is currently manufactured from 3-cyanopyridine by chemical methods. However, during the final step of 2-CN hydrolysis under high temperature and strong alkaline conditions, the byproduct 2-CM was generated except for the target product, leading to low yield and tedious separation steps. Nitrilase-mediated hydrolysis is regarded as a promising alternative for 2-CA production, which proceeded under mild conditions. Nevertheless, nitrilase capable of efficient hydrolysis of 2-CN has not been reported because the enzymes showed either extremely low activity or surprisingly high hydration activity toward 2-CN. Herein, the reaction specificity of RzNIT was precisely tuned through a single site mutation. The mutant exhibited remarkably enhanced hydrolysis activity without the formation of byproducts, providing a robust biocatalyst for 2-CA biosynthesis with industrial potential.

INTRODUCTION

A derivative of pyridine with broad applications in the synthesis of pharmaceuticals and agrochemicals, such as the most effective herbicides, diflufenican, and nicosulfuron, is 2-chloronicotinic acid (2-CA) (1, 2). Currently, the chemical manufacturing process for 2-CA began with N-oxidation of 3-cyanopyridine followed by chlorination and hydrolysis (3). However, this route suffered from harsh reaction conditions and serious environmental problems. Moreover, except for the target product 2-CA, some 2-chloronicotinamide (2-CM) was formed during the final step hydrolysis of 2-chloronicotinonitrile (2-CN) under strongly alkaline conditions, leading to low yields and tedious separation steps (4). Consequently, there is an increasing demand for large-scale production of 2-CA with higher efficiency and eco-friendliness.

Nitrilases (EC 3.5.5.1) are a class of industrially important enzymes that catalyze the direct hydrolysis of nitriles to produce corresponding carboxylic acids and ammonia (5–7). Over the past years, eco-friendly nitrilase-mediated bioprocess has received considerable interest for the preparation of value-added carboxylic acids (8), which are important building blocks for a variety of polymers (9, 10), pharmaceuticals (11–15), pesticides (16), and feed additives (17). Accordingly, nitrilases are regarded as promising alternatives to chemical catalysts for efficient hydrolysis of 2-CN, which run at ambient temperature without the requirement of strongly acidic or basic reaction conditions and produce the desired product with high selectivity. Nevertheless, to the best of our knowledge, known nitrilase showed either extremely low hydrolysis activity or surprisingly high hydration activity toward 2-CN that converted 2-CN to 2-CM (2, 18).

Although nitrilases are defined by their ability to hydrolyze nitriles into carboxylic acids, several nitrilases exhibiting both hydrolysis and hydration activity were reported in recent years. For example, the nitrilase NIT4 from Arabidopsis thaliana catalyzed the conversion of β-cyanoalanine to asparagine and aspartic acid, and its hydration activity was higher than hydrolysis activity (19, 20). The nitrilase from Pseudomonas sp. strain UW4 produced significantly more indole-3-acetamide (>75%) than indole-3-acetic acid from indole-3-acetonitrile (21). The nitrilase from Pseudomonas fluorescens EBC191 generated approximately the same amounts of (S)-mandelic amide and (S)-mandelic acid from (S)-mandelonitrile (22–25). The occurrence of hydration activity of nitrilase severely restricted its application in the synthesis of carboxylic acids with high yield and purity (26). Due to the deficiency of precise crystal structure information of nitrilase (27–29), it is still quite challenging to tailor nitrilase to overcome trade-offs between reaction specificity and catalytic activity.

In our present study, nitrilase from Rhodococcus zopfii (RzNIT) was found to show relatively high catalytic activity toward 2-CN. Frustratingly, its hydration activity dominated the whole reaction. With RzNIT as biocatalyst, 2-CM generated from 2-CN accounted for as high as 88% of the product. The occurrence of hydration activity of RzNIT significantly reduced product yield and caused tedious separation steps. Hence, it is highly desirable to engineer RzNIT with high activity and strict reaction specificity. A single mutant of W167G was obtained and its hydration activity toward 2-CN was completely abolished, while its hydrolysis activity was remarkably enhanced by 20-fold compared to the wild type, indicating that RzNIT was precisely switched into a robust nitrilase with only hydrolysis activity toward 2-CN.

RESULTS

Screening of nitrilases with activity toward 2-CN.

Various nitrilases from different sources were screened for activity toward 2-CN. However, except for the nitrilase from Rhodococcus zopfii (RzNIT), most of the tested nitrilases showed no activity or extremely low activity (Table 1). In addition, some enzymes exhibited hydration and hydrolysis activity simultaneously. For example, the nitrilase from Acidovorax facilis (AfNIT), an industrially important nitrilase (9, 10), exhibited low hydrolysis activity (0.79 U/g DCW) and hydration activity (0.93 U/g DCW). Although RzNIT showed the highest activity toward 2-CN, its hydration activity for 2-CN (18.60 U/g DCW) was much higher than hydrolysis activity (2.65 U/g DCW). With RzNIT as biocatalyst, 88% of 2-CM and 12% of 2-CA were detected as products in the reaction mixture (Fig. 1), indicating poor reaction specificity of the enzyme toward 2-CN. However, RzNIT displayed much higher hydrolysis activity (4132 U/g DCW) toward the nonchlorinated substrate nicotinonitrile, and no formation of nicotinamide was observed.

TABLE 1.

Activity and amide formation capability of nitrilases from different sources toward 2-CN^d

Nitrilase	Sequence ID	Source	Hydration activity (U/g DCWa)	Hydrolysis activity (U/g DCW)	Amide formation (%)
AtNIT1	NP_851011	Arabidopsis thaliana	Ndb	Nd	-c
BrNIT2	BAG72074	Brassica rapa	Nd	Nd	-
AaNIT	KFK44999	Arabis alpina	Nd	Nd	-
ZmNIT1	AAO11743	Zea mays	Nd	Nd	-
OsNIT	BAA77679	Oryza sativa	Nd	Nd	-
PgNIT	WP_006050412	Paraburkholderia graminis	Nd	Nd	-
BnNIT	AAK57436	Brassica napus	Nd	Nd	-
TpNIT	XP_002290043	Thalassiosira pseudonana	Nd	Nd	-
KpNIT	P10045	Klebsiella pneumoniae	Nd	Nd	-
GpNIT	ABH04285	Geobacillus pallidus	Nd	Nd	-
ApNIT	BAJ17399	Arthrobacter pascens	Nd	Nd	-
RrNIT	BAA02127	Rhodococcus rhodochrous K22	Nd	0.37 ± 0.05	0
RzNIT	WP_138999863	Rhodococcus zopfii	18.60 ± 0.81	2.65 ± 0.20	88
BjNIT	BAC48662	Bradyrhizobium japonicus	Nd	Nd	-
AlfNIT	WP_042484030	Alcaligenes faecalis	Nd	Nd	-
PpNIT	WP_093971852	Pseudomonas putida	Nd	Nd	-
PaNIT	CAB50304	Pyrococcus abyssi	Nd	Nd	-
AfNIT	ABD98457.1	Acidovorax facilis	0.93 ± 0.03	0.79 ± 0.06	54
TmNIT	WP_010865067	Thermotoga maritima	Nd	Nd	-
PfNIT	AAW79573	Pseudomonas fluorescens	Nd	Nd	-
CsNIT	XP_010514771	Camelina sativa	Nd	Nd	-
PgNIT	QCT24552.1	Paraburkholderia graminis	0.02 ± 0.01	0.03 ± 0.01	40
CrNIT	XP_006291365	Capsella rubella	Nd	Nd	-
EsNIT	XP_006391934	Eutrema salsugineum	Nd	Nd	-
NdNIT	KPM40743.1	Neonectria ditissima	Nd	Nd	-

^aDCW, dry cell weight.

^bNd indicates that the nitrilase tested did not display hydration activity or hydrolysis activity toward 2-CN.

^cA dash (-) indicates that amide formation was not calculated.

^dThe reactions were performed in the standard reaction mixtures (1 mL) containing resting cells (4 mg DCW) and 2-CN (50 mM) at 30°C.

FIG 1.

Bioconversion of 2-CN and nicotinonitrile with RzNIT as biocatalyst.

Rational design of RzNIT for regulation of reaction specificity.

To improve the reaction specificity of RzNIT toward 2-CN, homologous modeling and docking studies were performed. The 3D structure model of nitrilases was constructed using homology modeling and optimized after a 30 ns molecular dynamic (MD) simulation (Fig. S1). Subsequently, these equilibrated conformers were extracted for detailed structural analysis. Their geometries were validated by a Ramachandran plot. As shown in Fig. S2, 92.2% of residues were in the most favored regions, 5.9% were in additional allowed regions, 2.0% were in the generously allowed regions, and no residues were in disallowed regions, suggesting that the constructed model of RzNIT was within the acceptable range. The conserved catalytic triad of RzNIT was located at positions 49 (Glu), 132 (Lys), and 166 (Cys). The substrates 2-CN and nicotinonitrile were docked into the active center pocket of the wild-type (WT), respectively (Fig. 2A). Fifteen residues, including Y55, T136, V138, E139, N165, W167, E168, W190, P191, Y196, Q197, and P198 as well as E49, K132, and C166 that formed the catalytic triad, were within 5 Å of the substrates and, together, formed the substrate-binding pocket. According to the elucidated reaction mechanism of nitrilase (30), C166 was responsible for nucleophilic attack of the carbon atom of the cyano group by the sulfur atom or itself (cysteine residue). Thus, the distance (D_C-S) between the carbon atom of the cyano group of the substrate and the sulfur atom of catalytic C166 played a crucial role in determining the activity of nitrilase. D_C-S of 2-CN and nicotinonitrile were 3.31 Å and 3.12 Å, respectively (Fig. 2B). The increased distance prevented the efficient nucleophilic attack, leading to its lower activity toward 2-CN, which was consistent with the experimental observation.

FIG 2.

The active site of RzNIT with predicted binding mode for 2-CN and nicotinonitrile. (A) Surface figures of superimposed docking models. All the atoms of the catalytic triad of E49-K132-C166 are shown in blue, magenta, and red, respectively. All the atoms of N165 and W167 are shown in yellow and green, respectively. 2-CN and nicotinonitrile are shown in cyan and orange, respectively. (B) Magnified view of the models. The chlorine, sulfur, oxygen, hydrogen and carbon, and nitrogen atoms are shown in purple, yellow, red, gray, green, and blue, respectively. The cyano group is shown in magenta. D_C-S of 2-CN and nicotinonitrile are shown in the dashed yellow line and dashed blue line, respectively. Red dashed line indicates the axis cyano group of 2-CN. Red circular arrow indicates the rotation of 2-CN.

Comparison of the models revealed that the pyridine rings of 2-CN and nicotinonitrile were in significantly different orientations, while their cyano groups were fitted similarly into the active center (Fig. 2B). The chlorine atom of 2-CN formed a spatial barrier and initiated the rotation of the pyridine ring (red circular arrow in Fig. 2B) of 2-CN on the axis of the cyano group (red dash line in Fig. 2B). The cyano groups of the two substrates were in similar orientation, which might attribute to the steric hindrance in the substrate-binding cavity. Two bulky amino acids N165 and W167 were in direct proximity to the cysteine residue in the binding pocket, and W167 was located at the entrance of the binding pocket. They were involved in a steric clash with the cyano group and chlorine atom, respectively, hindering 2-CN binding with the catalytic residues. By fine-tuning these two amino acids with hydrophobic and hindered side chains, substrate 2-CN was expected to be bound more deeply in the pocket. Therefore, N165 and W167 were chosen as candidate residues for engineering reaction specificity and catalytic activity.

Identification of RzNIT variants with improved reaction specificity and activity.

Based on site-directed saturation mutagenesis, a small library of mutations at positions 165 and 167 was constructed. The reaction specificities and activities of whole cells harboring all mutants were investigated. As shown in Fig. 3, mutagenesis at residue 165 and 167 significantly affected the reaction specificity toward 2-CN. The hydration activity of N165A toward 2-CN was decreased from 18.60 to 0.35 U/g DCW, and that of N165G, W167A, and W167G were too low to be determined. As a result, the formation of 2-CA was increased from 12% up to 94%, 99%, 99%, and 100% with N165A, N165G, W167A, and W167G as biocatalyst, respectively. In addition, compared with WT, the hydrolysis activity of mutants N165A, W167A, and W167G were increased from 2.65 to 5.75, 8.77, and 53.50 U/g DCW, respectively. Among all the mutants, W167G exhibited the highest activity and strictest reaction specificity. These results strongly demonstrated that the reaction specificity was fine-tuned through a single mutation of W167.

FIG 3.

Activity of the mutants from saturation mutagenesis at positions 165 (A) and 167 (B).

The nitrilases with His-tags were purified by nickel affinity chromatography and analyzed by SDS-PAGE (Fig. S3). The specific activities of the WT and W167G mutant toward nicotinonitrile and its chlorinated nicotinonitriles were further investigated. Byproduct formation was detected only with 2-CN as the substrate. The mutant W167G exhibited the specific hydrolysis activity of 118.68 × 10⁻³ U/mg, which was 20-fold higher than that of the WT (5.48 × 10⁻³ U/mg) (Table 2), and its hydration activity was eliminated. The specific activity of W167G for nicotinonitrile and 6-CN were dramatically decreased, and neither WT nor W167G showed activity for 4-CN and 5-CN (Table 3).

TABLE 2.

Specific activities of the WT and its variants for 2-CN

Mutants	Specific hydration activity (U/mg) (×10−3)	Specific hydrolysis activity (U/mg) (×10−3)	Total activity (U/mg) (×10−3)	Relative activity (%)
WT	31.62 ± 0.93	5.48 ± 0.10	37.10 ± 1.03	100
N165C	12.07 ± 0.28	16.18 ± 0.65	28.25 ± 0.93	76
N165A	Nda	13.04 ± 0.23	13.04 ± 0.23	35
N165G	Nd	1.85 ± 0.05	1.85 ± 0.05	5
W167A	Nd	12.75 ± 0.22	12.75 ± 0.22	34
W167G	Nd	118.68 ± 4.15	118.68 ± 4.15	320

^aNd indicates that the protein tested did not display hydration activity toward 2-CN.

TABLE 3.

Specific activities (U/mg) of the WT and W167G toward nicotinonitrile and its chlorinated derivatives

Mutants	Nicotinonitrile	2-CN	4-CN	5-CN	6-CN
WT	20.36 ± 1.58	37.10 ± 1.03 (×10−3)	Nda	Nd	0.90 ± 0.05
W167G	1.17 ± 0.03	118.68 ± 4.15 (×10−3)	Nd	Nd	0.02 ± 0.01

^aNd indicates that the protein tested did not display activity toward the substrate.

Kinetic analysis of RzNIT and its variants.

The kinetic parameters of the WT and its mutants N165C, N165A, N165G, W167A, and W167G, which exhibited different 2-CM formation capacities (88%, 54%, 6%, 1%, 1%, and 0%, respectively) (Fig. 4), were investigated. Kinetic studies on the hydrolysis and hydration activities were performed with 2-CN as the substrate. As listed in Table 4, the K_m for hydrolysis activity of N165C, N165A, and N165G was found to be slightly decreased from 3.15 mM to 2.47, 2.58, and 2.55 mM, while that for W167A and W167G was increased to 8.89 and 6.78 mM, respectively. However, W167G displayed a 20-fold increase in k_cat compared with WT, leading to a 9-fold improvement in catalytic efficiency (k_cat/K_m). For hydration activity, N165C decreased the hydration catalytic efficiency by 40% compared with WT (Table 4). The 2-CM formation capabilities of N165A, N165G, W167A, and W167G were too low to be determined. The results also revealed that the K_m of the WT and N165C toward 2-CN for both hydrolysis and hydration reactions is very similar if not identical, suggesting that the catalytic center for both reactions might be the same (19).

FIG 4.

HPLC analysis of the reaction mixture catalyzed by the wild-type RzNIT and its variants.

TABLE 4.

Kinetic parameters of the WT and its mutants toward 2-CN

Mutants	Vmax (μmol/mg/min) (×10−3)	Km (mM)	kcat (s−1) (×10−3)	kcat/Km (s−1×mM−1) (×10−3)
Kinetic parameters of hydrolysis activity
WT	6.51 ± 0.31	3.15 ± 0.06	4.48 ± 0.21	1.42 ± 0.04
N165C	21.14 ± 0.87	2.47 ± 0.04	14.55 ± 0.60	5.89 ± 0.15
N165A	14.72 ± 0.27	2.58 ± 0.19	10.12 ± 0.19	3.94 ± 0.22
N165G	1.99 ± 0.06	2.55 ± 0.03	1.37 ± 0.04	0.54 ± 0.01
W167A	17.46 ± 1.12	8.89 ± 0.39	11.99 ± 0.76	1.35 ± 0.03
W167G	130.43 ± 2.63	6.78 ± 0.26	89.52 ± 1.81	13.21 ± 0.24
Kinetic parameters of hydration activity
WT	37.67 ± 2.60	3.78 ± 0.09	25.93 ± 1.79	6.85 ± 0.31
N165C	16.56 ± 0.17	2.87 ± 0.05	11.39 ± 0.12	3.97 ± 0.03
N165G	-a	-	-	-
W167A	-	-	-	-
W167G	-	-	-	-

^aA dash indicates that the kinetic parameters were not calculated.

Docking analysis and MD simulation.

To probe the mechanism of the improved reaction specificity and activity, the 3D structure models of the mutants (N165C, N165A, N165G, W167A, and W167G) were constructed (Fig. S1). Subsequently, the substrate 2-CN was docked into the active center pocket of the equilibrated conformers of the mutants. New molecular interactions were found in all 2-CN-mutant models but did not exist in the 2-CN-WT model (Fig. 5A). In the 2-CN-N165C model (Fig. 5B), the nitrogen atom of the pyridine ring formed a hydrogen bond with hydroxyl hydrogen of Y55 (2.97 Å). In the 2-CN-N165A, 2-CN-N165G, and 2-CN-W167A models (Fig. 5C to E), the nitrogen atom of the pyridine ring formed a hydrogen bond with hydroxyl hydrogen of T136 (2.03 Å, 1.92 Å, and 2.00 Å), respectively. In the 2-CN-W167G model (Fig. 5F), the pyridine ring formed π–π stacking with indole ring of W190 (6.54 Å and 7.79 Å) and benzene ring of Y196 (5.25 Å), respectively.

FIG 5.

The substrate-binding pocket of the WT RzNIT and its variants with docked substrate 2-CN. (A) 2-CN-WT model. (B) 2-CN-N165C model. (C) 2-CN-N165A model. (D) 2-CN-N165G model. (E) 2-CN-W167A model. (F) 2-CN-W167G model. The catalytic triad of E49-K132-C166 and the residue at sites 55, 136, 165, 167, 190, and 196 are shown in stick figures. The chlorine, sulfur, oxygen, hydrogen and carbon, and nitrogen atoms are shown in purple, yellow, red, gray, green, and blue, respectively. The cyano group is shown in magenta. The D_C-S are shown in dashed yellow lines. The hydrogen bonds are shown in dashed green lines. The π-π bonds are shown in dashed red lines.

Dynamic structural analysis was performed to investigate how the substitutions contributed to the enhanced reaction specificity. The mutation site was located at loop A (165 to 167) and replacements of bulky N165 and W167 with smaller amino acids reduced its fluctuation. As a result, the root-mean-square fluctuation (RMSF) of loop A was significantly decreased in all mutants (Fig. S4B to F), indicating its increased rigidity. Y55 and T136 (formed a hydrogen bond with 2-CN) were located at loop B (47 to 57) and loop C (132 to 136), respectively (Fig. S4A). The RMSF of these two loops were decreased in N165C and N165A, N165G, and W167A, respectively (Fig. S4B to E), which was partially due to the reduced fluctuation because of additional hydrogen bonds. Conformational flexibility was proposed to affect the catalytic promiscuity (31, 32). In this study, the decreased conformational flexibility of A, B, and C loops resulted in enhanced reaction specificity of RzNIT.

Generality of the key residue for reaction specificity tuning among nitrilases.

To clarify the general effect of the redesigned residue on reaction specificity and activity of nitrilases, three nitrilases from Fusarium euwallaceae (FeNIT), Macrophomina phaseolina (MpNIT), and Paraburkholderia graminis (PgNIT), which shared 38%, 39%, and 33% identities to RzNIT, were chosen for further test. The putative sensitive residues of these nitrilases were identified by multisequence alignment as W163 (FeNIT), W164 (MpNIT), and F164 (PgNIT), respectively (Fig. S5). As shown in Table S2, the hydrolysis activity of mutant FeNIT-W163G, MpNIT-W164G, and PgNIT-F164G was 7.7-fold, 3.5-fold, and 629-fold higher than that of the wild-types, respectively. Furthermore, the hydration activities of all the mutants were completely abolished. These results strongly indicated that the predicted hot spot also played an essential role in regulating the reaction specificity and activity of nitrilases possessing over 33% identities to RzNIT.

Biosynthesis of 2-CA by the whole-cell biocatalyst.

To verify the improved reaction specificity and activity of W167G mutant, biosynthesis of 2-CA was investigated using the whole cells harboring the wild-type RzNIT, mutant W167G, and AfNIT as biocatalysts. As shown in Fig. 6, the W167G mutant could completely hydrolyze 100 mM 2-CN within 16 h without any formation of 2-CM, while the WT converted 2-CN after 26 h and only 12.48 mM 2-CA was produced. In addition, the industrially important nitrilase AfNIT only produced 2.45 mM 2-CA after 26 h. These results highlighted that RzNIT was successfully turned into a robust biocatalyst for efficient production of 2-CA with only hydrolysis activity. To our knowledge, mutant W167G is so far the best choice of biocatalyst with the highest reaction specificity and activity for 2-CA among all reported nitrilases.

FIG 6.

Conversion of 2-CN mediated by the wild-type RzNIT, W167G mutant, and AfNIT. The reactions were carried out at 30°C and pH 7.0 in a 50 mL reaction mixture containing 0.2 g DCW recombinant cells and 100 mM 2-CN.

DISCUSSION

In contrast with the common wisdom that nitrilase exclusively hydrolyze nitrile into the corresponding carboxylic acid, its hydration activity has been neglected for a long time (33, 34). In recent years, reasonable catalytic mechanisms of nitrilases have been proposed. It is commonly accepted that nitrilases convert nitriles into carboxylic acids and amides in two different pathways (Fig. 7) (35–37). In pathway A, elimination of ammonia from the enzyme-substrate tetrahedral intermediate to produce acids requires protonation of the nitrogen atom originated from the cyano group. In pathway B, the thiol elimination from the intermediate to produce amides requires protonation of the thiol in catalytic Cys. In addition, the competition between the hydrolysis and hydration reactions based on the cleavage of the C-N or C-S bond of the tetrahedral intermediate is profoundly influenced by the stereochemical and electronic properties of the R group of the substrate. When the R group destabilized protonation of the nitrogen atom of the cyano group, pathway A of nitrilase would be blocked, and simultaneously the pathway B would be initiated. Furthermore, it was found that the extent of amide increases with the electronegativity of α-substituent: CH₃ < H < Cl (34). For example, a nitrilase from A. thaliana was highly sensitive to electronic effects (38). It catalyzed 2-crotonitrile to 99% crotonic acid, whereas 95% amide was formed from the electron-deficient substrate 3-nitroacrylonitrile (38). Similarly, when 2-fluoroarylacetonitrile was hydrolyzed by this nitrilase, the main product was (R)-2-fluoroarylacetamides, but not the expected 2-fluorarylacetic acids (39).

FIG 7.

The proposed catalytic mechanism of RzNIT with hydration and hydrolysis activity.

The wild-type RzNIT was investigated for its catalytic properties toward nicotinonitrile and its chlorinated derivatives. Owing to the electron-donating chlorine atom, the cyano group of 2-CN has a larger electron density compared to that of nicotinonitrile. The increased electron density enhances the electronegativity of the enzyme-substrate intermediate and destabilizes the positive charge on the nitrogen atom of the cyano group. Thus, pathway B occupies the dominant position, resulting in the production of appreciable amounts of byproduct 2-CM.

Docking analysis showed that new molecular interactions were found in all 2-CN-mutant models. We noted that both hydrogen bond and π-π bond could decrease the electronegativity of the intermediate. As shown in Fig. 8, the newly formed hydrogen bond enabled the hydroxyl hydrogen of Y55 and T136 to attract the lone pair electrons of the nitrogen atom in the pyridine ring of 2-CN, which protonates the nitrogen atom much easier, bending the selectivity of N165C, N165A, N165G, and W167A toward 2-CA. By delocalization π bond, the electrons are delocalized over indole ring of W190 and benzene ring of Y196, making protonation of the nitrogen atom in pathway A much easier, resulting in complete abolishment of hydration activity. On the other hand, double mutants W167G/W190G and W167G/Y196G were constructed to verify the role of the π-π bond. Consequently, neither of them showed any activity toward 2-CN, suggesting that the newly formed π-π bond was very critical for both reaction specificity and catalytic activity regulation. These results suggested that the charge distribution in the enzyme-substrate intermediate depends not only on the stereochemical and electronic properties of the substrate but also on the molecular interactions between the substrate and the amino acid residues, leading to a precise switch between hydration and hydrolysis reactions.

FIG 8.

The effects of electronegativity of enzyme-substrates intermediate and protonation of the nitrogen atom of cyano group on reaction specificity of RzNIT and its mutants. The electronegativity of 2-CN-WT was set as “–,” and the protonation of the nitrogen atom of the cyano group was set as “+”. The symbols “–––,” “–” and “–” represent the electronegativity of enzyme-substrates intermediate was decreased gradually. The symbols “+,” “++,” “+++” and “++++” represent protonation of the nitrogen atom of the cyano group were much easier.

Despite improved reaction specificity, the specific activities of most mutants were dramatically decreased because of increased nucleophilic attack distance. As illustrated in Fig. 5A to E, the D_C-S in the 2-CN-N165C, 2-CN-N165A, 2-CN-N165G, and 2-CN-W167A models were increased from 3.31 Å to about 3.80 Å compared with the WT. The structural changes in the pocket made the formation of hydrogen bonds easier, but nucleophilic attack much more difficult. While in the 2-CN-W167G model, the smallest side chain of the mutant generated a larger binding pocket, and the cyano group of 2-CN was orientated toward the sulfur atom of the catalytic C166 residues. Furthermore, under the strong π–π interaction, the 2-CN appeared to be stabilized in a more productive position where the shortened D_C-S (3.23 Å) made it easier to initiate a nucleophilic attack (Fig. 5F). W190 and Y196 (formed π–π bond with 2-CN) were located at loop D (190 to 197) (Fig. S4A). Compared with WT, the RMSF of loop D in the W167G was increased, indicating that this loop was relatively flexible (Fig. S4F). Substitution of Trp by Gly at position 167 reduced the hydrophobic interactions, which enhanced the flexibility of loop D and further resulted in improvement in catalytic efficiency. Thus, W167G exhibited remarkably enhanced hydrolysis activity.

Although many efforts have been devoted to regulating the reaction specificity of nitrilases, the trade-offs between reaction specificity and catalytic activity remain a thrilling challenge to be resolved. For instance, mutant F193N of the nitrilase from Synechocystis sp. PCC6803 improved its amide product up to 73%, which was 35-fold than that of the wild type, but only maintained 50% of its activity (37). The mutant W188K of nitrilase from P. fluorescens EBC191 increased amide formation up to 90%, while it just possessed 20% of the wild-type enzyme activity (25). The activity of double mutant I128L/N161Q of nitrilase from G. intermedia was improved by 100%, however, a small amount of amide was still detected (27). In this study, simultaneous regulation of reaction specificity and catalytic activity of RzNIT was achieved by a single site mutation. The hydration activity of the mutant W167G toward 2-CN was abolished, and its hydrolysis activity was improved by 20-fold.

We succeeded in the rational design of nitrilase for efficient synthesis of 2-CA with the highest catalytic activity and strictest reaction specificity. Analysis of the substrate-binding conformations revealed that the newly formed hydrogen bond, delocalization π bond, and shortened nucleophilic attack distance together contributed to overcoming the trade-offs between reaction specificity and catalytic activity. The strategies provide valuable information for the rational design of nitrilases, which would greatly expand its applications for high value-added carboxylic acids production.

MATERIALS AND METHODS

Reagents and materials.

Nicotinonitrile, 2-CN, 4-chloronicotinonitrile (4-CN), 5-chloronicotinonitrile (5-CN), 6-chloronicotinonitrile (6-CN), and their corresponding amides and acids were purchased from J&K Chemicals (Shanghai, China). Kanamycin and isopropyl-β-d-thiogalactopyranoside (IPTG) were obtained from Sigma (China). Phanta Max Super-Fidelity DNA polymerase, 2×Phanta Max buffer, and deoxynucleoside triphosphates (dNTPs) for PCR were purchased from Vazyme (Nanjing, China). All other reagents were of analytical grade and commercially available.

Bacterial strains and plasmids.

Twenty-five amino acid sequences of nitrilase from different sources were codon-optimized against Escherichia coli as the host. The synthesized genes with 6× His tags were inserted into plasmid pET-28b (+) between restriction sites of Nco I and Xho I, generating the recombinant plasmids pET-28b (+)-NIT, which was further transformed into competent E. coli BL21(DE3) cells. The recombinant plasmids were then confirmed by sequencing.

Site-directed saturation mutagenesis.

Site-directed saturation mutagenesis was performed by introducing degenerate oligonucleotides (NNK) at the selected hot spots (amino acid residues 165 and 167) using the primers listed in Table S1. The PCRs were set up in a total volume of 50 μL containing 5 ng pET-28b (+)-RzNIT template plasmid, 0.4 μM (each) these primers, 25 μL 2 × phanta Max buffer, 0.2 mM dNTPs, 2.5 U PrimeSTAR HS DNA polymerase (TaKaRa) and distilled water. PCR program used was 3 min at 95°C followed by 30 cycles of 15 s at 95°C, 5 s at 60°C and 3.5 min at 72°C, and a final extension step with 10 min at 72°C. The amplification products were digested for 3 h at 37°C using DpnI to digest the parent plasmid. The mutant plasmids were transformed into E. coli BL21 (DE) cells and then confirmed by sequencing.

Expression and purification of nitrilases.

The recombinant E. coli BL21(DE3) cells harboring the recombinant plasmids were cultured in LB medium (10 g/liter peptone, 5 g/liter yeast extract, 10 g/liter NaCl) supplemented with kanamycin (50 μg/mL) at 37°C and 200 rpm. When the optical density at 600 nm (OD₆₀₀) reached about 0.6, IPTG at a final concentration of 0.1 mM was added to induce protein expression at 28°C for 10 h. The cells were harvested by centrifugation at 10,000 × g for 10 min 4°C. The collected cells were resuspended in potassium phosphate buffer (50 mM, pH 7.0) and disrupted by sonication. The cell debris was removed by centrifugation at 12,000 × g for 10 min at 4°C. The supernatant was applied to a 5 mL Ni-NTA Superflow column (Bio-Rad) that was preequilibrated with binding buffer (20 mM potassium phosphate, 300 mM sodium chloride, pH 8.0). The unbound proteins were eluted with a washing buffer (20 mM potassium phosphate, 300 mM sodium chloride, and 50 mM imidazole, pH 8.0), and the target protein was eluted with elution buffer (pH 8.0) containing 20 mM potassium phosphate, 300 mM sodium chloride and 250 mM imidazole. The eluted fractions containing the target protein were dialyzed in potassium phosphate buffer (50 mM, pH 7.0). The purified proteins were analyzed by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Protein concentration was determined using a protein assay kit from KeyGEN BioTECH (Jiangsu, China) with bovine serum albumin as a standard.

Enzyme assay and analytical methods.

The whole-cell activity of nitrilase was assayed in a reaction mixture (1 mL) containing potassium phosphate buffer (50 mM, pH 7.0), 2-CN (50 mM), and an appropriate number of recombinant cells. The activity of purified nitrilase was evaluated in a reaction mixture (1 mL) containing potassium phosphate buffer (50 mM, pH 7.0), 2-CN (20 mM), and an appropriate amount of purified nitrilase. The reaction was performed at 30°C for 10 min and quenched by adding 10 μL HCl (6.0 M). One unit of hydrolysis and hydration activity of nitrilase toward 2-CN was defined as the amount of enzyme required to produce 1 μmol of 2-CA and 2-CM per minute, respectively.

The formation of 2-CM and 2-CA in the reaction mixture was quantified by high-performance liquid chromatography (HPLC) equipped with a C18 column (5 μm × 250 mm × 4.6 mm, Welch Materials, Inc., Shanghai, China). The mobile phase was composed of acetonitrile and 0.1% phosphoric acid (25:75, vol/vol). The system was operated at a flow rate of 1 ml/min at 40°C. The detection wavelength was set as 210 nm. The other chlorinated amides and acids were detected by the same condition.

To determine the kinetic parameters, the initial velocity was measured by varying the concentrations of 2-CN from 1 to 20 mM. The values of K_m and V_max were obtained by performing a nonlinear regression analysis of the Michaelis-Menten equation using OriginPlot 8.0. The turnover number (k_cat) was calculated by using the equation k_cat=V_max/[E], where [E] was the molarity of the enzymes.

Computational methods.

The three-dimensional (3D) structures of the wild-type (WT) RzNIT and its variants were constructed via protein modeling software Modeller 9.19 (Accelrys, Inc.) Based on homology modeling concerning the crystal structure of Syechocystis sp. PCC6803 nitrilase (PDB ID accession number 3WUY), which shared the highest sequence identity to RzNIT. The C terminus of RzNIT (residues 294 to 367) was not modeled due to low homology. The models were checked by Procheck (http://nihserver.mbi.ucla.edu/).

The molecular dynamic (MD) simulation for WT and its variants was performed using GROMACS version 4.0 and OPLS-AA force field. Each protein was placed under periodic boundary conditions in the center of a cubic box consisting of SPC216 water molecules. After the redundant charges were neutralized, a reasonable geometrical structure was obtained by minimizing the system’s energy with the steepest descents method. To maintain the simulated systems at a constant temperature and pressure, 100 ps NVT and 100 ps NPT ensembles were used to achieve stable environment parameters. After completion of equilibration, 30 ns long MD simulations were performed with a step time of 2 fs at a constant pressure of 1 bar pressure and a constant temperature of 300 K, respectively. The dynamics trajectories were analyzed by using Visual Molecular Dynamics (VMD) software and Grace 5.1.25.

The equilibrated conformers of the nitrilases were extracted from their dynamic trajectories by VMD software and then docked to substrate 2-CN. The molecular docking studies were carried out by AutoDock 4.0 and the results were visualized by PyMol ver 1.7 (Schrodinger, Portland, OR).

Whole-cell biocatalytic hydrolysis of 2-CN.

The conversion of 2-CN was catalyzed by whole cells of recombinant E. coli harboring WT, mutant W167G, and AfNIT. The biotransformation was carried out at 30°C in a 50 mL reaction mixture containing potassium phosphate buffer (200 mM, pH 7.0), recombinant cells (4 g/liter dry cell weight, DCW), and 2-CN (100 mM). Samples were withdrawn periodically for HPLC analysis as described above.