A Rapid Method for the Selection of Amidohydrolases from Metagenomic Libraries by Applying Synthetic Nucleosides and a Uridine Auxotrophic Host

: In this study, the development of a rapid, high-throughput method for the selection of amide-hydrolysing enzymes from the metagenome is described. This method is based on uridine auxotrophic Escherichia coli strain DH10B ∆ pyrFEC and the use of N 4 -benzoyl-2’-deoxycytidine as a sole source of uridine in the minimal microbial M9 medium. The approach described here permits the selection of unique biocatalysts, e.g., a novel amidohydrolase from the activating signal cointegrator homology (ASCH) family and a polyethylene terephthalate hydrolase (PETase)-related enzyme.


Introduction
It is estimated that the Earth is inhabited by 10 11 -10 12 microbial species, of which only~10 4 have been cultured in the laboratory [1]. Nonetheless, the unculturable microbiome has provided science with new genes, biocatalysts, natural compounds, and bioproducts [2][3][4][5]. New enzymes that could allow provide novel approaches for designing enzymatic processes are in constant demand, and so are the screening methods for them. Keeping in mind that the functions of 30%-40% of the genes in genomes remain unknown, the detection of new biocatalysts in the sequenced genomes can be difficult if no significant homology with known enzymes is observed [6,7]. Metagenomics, the improvement of high-throughput screening methods, and the development of host expression systems for metagenome-derived genes, systems biology, and gene synthesis may together open the gateway to the useful information that hides in unexplored genetic resources, e.g., new enzymes with unique properties and novel scaffolds applicable for evolution in vitro [8][9][10][11][12].
The amidohydrolases/amidases are large classes (EC 3.5.1 and 3.5.2.) of enzymes which hydrolyse a wide variety of linear and cyclic amides. These enzymes are involved in several metabolic pathways, e.g., amide digestion and absorption by breaking carbon and nitrogen bonds, prokaryotic amide degradation, and synthesis of many important eukaryotic growth factors. Amidohydrolases have potential applications in chemical synthesis, pharmaceuticals, and effluent treatment due to their remarkable chemo-, regio-, and enantio-selectivity. At present, amidases are used for the biosynthesis

Functional Screening of Metagenomic Libraries
To develop this selection method, N 4 -benzoyl-2'-deoxycytidine (4) was used as a sole source of uridine in the minimal M9 medium for the E. coli DH10B ∆pyrFEC::Km [27] strain, harbouring plasmid vectors with random environmental DNA fragments (the metagenomic libraries). The amidohydrolases, supposedly encoded in the metagenomic libraries, convert N 4 -benzoyl-2'-deoxycytidine to 2'-deoxycytidine, and the native E. coli cytidine deaminase [28] converts 2'-deoxycytidine to 2'-deoxyuridine, upon which the growth phenotype of uridine auxotrophic cells is restored. The principle of the selection method is shown in Figure 1.
Metagenomic libraries used for functional selection of amidohydrolases were constructed from different soil DNA samples as described earlier [12]. The DNA fragment length varied from 5 to 17 kb in the different libraries. In total, 15 libraries were tested. The positive hits were identified in 5 of these libraries (Table S1). The clones encoding the desired amidohydrolases were selected using M9 medium containing 100 µg/mL ampicillin, 40 µg/mL kanamycin, and 20 µg/mL N 4 -benzoyl-2'-deoxycytidine (4). In overall, approximately 10 6 clones were analysed on a single agar plate. Altogether, 12 positive clones were selected. To omit the false positive hits, all clones were re-streaked on M9 medium with and without the cytidine derivative. Plasmid DNA was isolated only from the clones forming colonies in the presence of N 4 -benzoyl-2'-deoxycytidine, and restriction analysis was carried out to exclude the recurring ones. Therefore, 7 plasmids were chosen for further analysis. N 4 -benzoylcytidine (12), instead of N 4 -benzoyl-2'-deoxycytidine (4), was also tested as a substrate-all 7 E. coli DH10B ∆pyrFEC::Km transformants harbouring the selected plasmids used the N 4 -benzoylcytidine (12) as a uridine source. The necessity of the auxiliary enzyme for successful selection was demonstrated by deletion of the E. coli cdd gene, which encodes cytidine deaminase. As expected, none of the selected plasmids complemented the uridine deficiency in E. coli DH10B ∆pyrFECcdd::Km-the cells did not grow on the M9 agar medium supplemented with N 4 -benzoyl-2'-deoxycytidine (4) ( Figure S1). To develop this selection method, N 4 -benzoyl-2'-deoxycytidine (4) was used as a sole source of uridine in the minimal M9 medium for the E. coli DH10B ΔpyrFEC::Km [27] strain, harbouring plasmid vectors with random environmental DNA fragments (the metagenomic libraries). The amidohydrolases, supposedly encoded in the metagenomic libraries, convert N 4 -benzoyl-2'deoxycytidine to 2'-deoxycytidine, and the native E. coli cytidine deaminase [28] converts 2'deoxycytidine to 2'-deoxyuridine, upon which the growth phenotype of uridine auxotrophic cells is restored. The principle of the selection method is shown in Figure 1. The principle of amidohydrolase selection. Metagenomic DNA isolated from the environmental samples is fragmented, inserted into an appropriate plasmid vector, and used to transform the competent Escherichia coli DH10B ∆pyrFEC::Km cells. Mineral M9 medium containing N 4 -benzoyl-2'-deoxycytidine as a sole source of uridine is used to select the clones harbouring the amidohydrolase activity: amidohydrolases convert N 4 -benzoyl-2'-deoxycytidine to 2'-deoxycytidine, which is then converted to 2'-deoxyuridine by the native E. coli cytidine deaminase. The positive hits complement the growth of uridine auxotrophic E. coli DH10B ∆pyrFEC::Km strain, allowing the formation of colonies.

Bioinformatic Analysis of the Selected Clones
The sequence analysis showed that 6 of the 7 selected clones contained ORFs with medium (40%) to high (99%) sequence identity to hydrolases found in the NCBI GenBank database [29] ( Table 1). The clone P4FUMM07 contained two ORFs (80%-84% identity) encoding potential amidohydrolases, which were therefore named P4FUMM07_AcAm and P4FUMM07_AmH. The phylogenetic analysis of the selected hydrolases showed that these enzymes belong to various groups of proteins ( Figure 2 and Figure S2). Four of these enzymes (BRM_Am, MO10_Am, K3_Am, and P4FUMM07_AcAm) belong to the amidase signature (AS) enzyme superfamily SSF75304. According to the Uniprot database [30], the amidase signature (AS) enzymes are a large group of hydrolytic enzymes that catalyse the hydrolysis of amide bonds and have diverged widely with regard to substrate specificity and function [31]. Examples of AS enzymes include peptide amidase [32], fatty acid amide hydrolase [33], malonamidase E2 [34], and subunit A of Glu-tRNA(Gln) amidotransferase [35].
MO13 and CIAN4 are representatives of ABhydrolase superfamily SSF53474 and belong to the group C of the tannase/feruloyl_esterase family [38,39]. Feruloyl esterases (E.C. 3.1.1.73), also known as ferulic acid esterases (FAE), cinnamoyl esterases, and cinnamic acid hydrolases, are a subclass of the carboxylic acid esterases (E.C. 3.1.1.1) that are able to hydrolyse the ester bond between hydroxycinnamic acids and sugars present in the plant cell walls [39][40][41]. Tannase is an inducible microbial enzyme that hydrolyses ester and depside bonds in hydrolysable tannins to produce gallic acid and glucose [38]. This family includes fungal tannase and feruloyl esterase [42,43], several bacterial homologues of unknown function, and polyethylene terephthalate (PET) hydrolysing enzymes, e.g., the mono(2-hydroxyethyl) terephthalate hydrolase from the bacterium Ideonella sakaiensis [44,45]. In contrast to its homologous sequences, the MO13_Est631 shows a prolonged C-terminus (approximately 100 amino acids) uncharacteristic for most of the proteins in the tannase/feruloyl_esterase family, except for several hypothetical hydrolases from Variovorax, Serratia, and several other species. A truncated variant of this amidase-MO13_Est537, containing a 95-amino-acid deletion at the C-terminus-complemented the growth of uridine auxotrophic E. coli as efficiently as the wild-type variant (data not shown). Therefore, the trimmed MO13_Est537 protein was used for further characterization.
Of all 8 of the selected amidohydrolases, the most intriguing is the D8_RL enzyme: it is a 128-amino-acid uncharacterized protein without any predictable hydrolytic function. The amino acid sequence analysis using SMART [46] revealed that D8_RL contains the human activating signal cointegrator homology (ASCH) [47] and growth-arrest-specific protein 2 (GAS2) domains [48]. ASCH domain-containing proteins are widespread and diverse but, at present, the vast majority of those proteins have no function assigned to them [47], except for an ASCH domain-containing ribonuclease from Zymomonas mobilis [49] and a recently characterized unique ASCH domain-containing amidohydrolase YqfB from Escherichia coli, which is active towards N 4 -acylcytidines [50]. The comparison of YqfB and D8_RL amino acid sequences showed merely 17 identical residues between these two proteins, and only 2 of the 4 YqfB catalytic amino acids were among them, although the aspartic acid residue in position 72 of the D8_RL and the catalytic glutamic acid residue in position 74 of the YqfB enzyme can also be considered conservative ones (Figure 3 and Figure S2).
Catalysts 2020, 9, x FOR PEER REVIEW 5 of 12 produce gallic acid and glucose [38]. This family includes fungal tannase and feruloyl esterase [42,43], several bacterial homologues of unknown function, and polyethylene terephthalate (PET) hydrolysing enzymes, e.g., the mono(2-hydroxyethyl) terephthalate hydrolase from the bacterium Ideonella sakaiensis [44,45]. In contrast to its homologous sequences, the MO13_Est631 shows a prolonged C-terminus (approximately 100 amino acids) uncharacteristic for most of the proteins in the tannase/feruloyl_esterase family, except for several hypothetical hydrolases from Variovorax, Serratia, and several other species. A truncated variant of this amidase-MO13_Est537, containing a 95-amino-acid deletion at the C-terminus-complemented the growth of uridine auxotrophic E. coli as efficiently as the wild-type variant (data not shown). Therefore, the trimmed MO13_Est537 protein was used for further characterization. Of all 8 of the selected amidohydrolases, the most intriguing is the D8_RL enzyme: it is a 128amino-acid uncharacterized protein without any predictable hydrolytic function. The amino acid sequence analysis using SMART [46] revealed that D8_RL contains the human activating signal cointegrator homology (ASCH) [47] and growth-arrest-specific protein 2 (GAS2) domains [48]. ASCH domain-containing proteins are widespread and diverse but, at present, the vast majority of those proteins have no function assigned to them [47], except for an ASCH domain-containing ribonuclease from Zymomonas mobilis [49] and a recently characterized unique ASCH domain-containing amidohydrolase YqfB from Escherichia coli, which is active towards N 4 -acylcytidines [50]. The comparison of YqfB and D8_RL amino acid sequences showed merely 17 identical residues between these two proteins, and only 2 of the 4 YqfB catalytic amino acids were among them, although the aspartic acid residue in position 72 of the D8_RL and the catalytic glutamic acid residue in position 74 of the YqfB enzyme can also be considered conservative ones (Figure 3 and S2).

Substrate Specificity of the Selected Amidohydrolases
To confirm that the enzymes encoded in the selected fragments of the metagenomic DNA had the amidohydrolytic activity, their genes were PCR-amplified and the resulting fragments were ligated into pLATE31 or pET21a expression vectors. The BL21 (DE3) strain of E. coli was transformed with these plasmids and used for the synthesis of the recombinant amidohydrolase proteins. In total, 7 recombinant proteins were successfully purified by Ni-NTA chromatography ( Figure S3). Due to its hydrophobic nature, the P4FUMM07_AmH protein was not purified to homogeneity, and a partially purified protein fraction was used for the activity assays.

Substrate Specificity of the Selected Amidohydrolases
To confirm that the enzymes encoded in the selected fragments of the metagenomic DNA had the amidohydrolytic activity, their genes were PCR-amplified and the resulting fragments were ligated into pLATE31 or pET21a expression vectors. The BL21 (DE3) strain of E. coli was transformed with these plasmids and used for the synthesis of the recombinant amidohydrolase proteins. In total, 7 recombinant proteins were successfully purified by Ni-NTA chromatography ( Figure S3). Due to its hydrophobic nature, the P4FUMM07_AmH protein was not purified to homogeneity, and a partially purified protein fraction was used for the activity assays.
Seven of the selected hydrolases recognized esters as substrates; only the P4FUMM07_AmH was a "true" amidohydrolase without any detected esterase activity. The enzymes were active towards the short-chain esters (20)(21)(22); most of them used compounds 24 and 25, but only 3 amidases-BRM_Am, MO13Est537, and D8_RL-hydrolysed the pNP-decanoate (23). Thus, the selected hydrolases were the most active towards short-chain esters, in contrast to the larger amides.
Among the identified enzymes, the MO13_Est537 showed distinguished catalytic properties: in addition to cytidine-based substrates, the enzyme hydrolysed dimethyl terephthalate (27) and bis(2-hydroxyethyl) terephthalate (26). However, it was not active towards monomethyl terephthalate (28). HHpred [53] analysis showed that the best structural homologues of MO13_Est537 were feruloyl esterases from Aspergillus oryzae (PDB [54] entries 3WMT_B and 6G21_A) [43] and mono(2-hydroxyethyl) terephthalate hydrolase (MHETase) from Ideonella sakaiensis (PDB entry 6QG9_I), the second key enzyme in degradation of PET [44,45]. However, the previously described MHETases are not active towards dimethyl and bis(2-hydroxyethyl) terephthalates: these compounds are degraded by PET hydrolases (PETases), which are unable to hydrolyse the mono-modified terephthalates [55][56][57]. Hence, it can be concluded that the hydrolase MO13_Est537 from this study belongs to MHETase group according to its amino acid sequence, but its substrate specificity is more similar to that of PETases. There is no data available on the ability of feruloyl esterases to hydrolyse dimethyl/monomethyl terephthalate and bis(2-hydroxyethyl) terephthalate.

Catalytic Properties of the Amidohydrolase D8_RL
A more detailed analysis of the D8_RL hydrolase was carried out. Its kinetic parameters and substrate specificity was compared to those of a recently characterized amidohydrolase YqfB [50]. The kinetic parameters of both D8_RL and YqfB are shown in Table 2. According to these data, it can be concluded that both enzymes exhibit similar catalytic efficiency in the case of N 4 -benzoylated cytidines (4, 12), but, in contrast to D8_RL, the YqfB hydrolase is more active towards N 4 -acetylated cytidine derivatives (1, 11). Table 2. Catalytic parameters of D8_RL and YqfB. Both recombinant enzymes contain C-terminal 6-His tags. Activity was measured in 50 mM potassium phosphate buffer, pH 8.0, at 22 • C. Rate constants were calculated from experiments with at least 3 different initial concentrations of substrate, using a statistical significance value of 0.05.

Enzyme
Substrate N 4 -acetylcytidine not being a natural substrate of D8_RL enzyme might be the reason for the differences between the catalytic constants. The tremendous difference of the isoelectric points of these two small proteins-4.69 for YqfB (103 aa) and 9.00 for D8_RL (128 aa)-attracts attention, but the relationship between the structural features and the catalytic properties remains unclear. Keeping in mind that both proteins belong to the same ASCH family, but have very different features, further studies should be carried out to elucidate the peculiarities of these small hydrolases.

Overexpression and Purification of Hydrolases
Metagenomic amidases encoding genes were amplified with Phusion DNA polymerase using primers listed in Supplementary Table S2. The cloning and overexpression procedures were performed as described previously [12].

Enzymatic Activity of Hydrolases
The activity of hydrolases was analysed by incubating the enzyme with 1 mM pNP-esters and amides: acetate, butyrate, valerate, decanoate, trimethylacetate, benzyl p-nitrophenyl carbonate, p-nitroacetanilide, and p-nitrobenzanilide (from 10 mM stock in DMSO) in 50 mM potassium phosphate buffer, pH 7.5, at 37 • C for 10-30 min in 100 µL of reaction volume. Depending on the enzyme specificity for the substrate, the reaction mixture contained 5-300 ng of protein. The absorption of the reaction mixture at 405 nm was measured against enzyme-free blank to compensate for the substrate auto-hydrolysis [62]. The hydrolytic activity using acylcytidines/cytosines was assayed in reaction mixture, which consisted of 50 mM potassium phosphate buffer, pH 7.5, with 0.1-5 mM substrate and 5-300 ng of protein as described earlier [50]. A kinetic study of D8_RL hydrolase was performed as described previously [50]. The reactions were started by adding an appropriate amount of protein to 50 mM potassium phosphate buffer, pH 8.0, supplemented with 0.1-0.5 mM substrate. Then, a decrease in absorbance at 310 nm was recorded at 22 • C. rModeler software (Ubique Calculus Ltd., Vilnius, Lithuania) was used for the calculation of the kinetic parameters.

Conclusions
In conclusion, a combination of a synthetic cytidine derivative and the uridine auxotrophic mutant of E. coli DH10B ∆pyrFEC allows for the functional selection of amidohydrolases from metagenomic libraries. The proposed selection method is a high-throughput method that is applicable for rapid processing of large metagenomic libraries and offers the possibility to discover representatives of enzymes belonging to both known families of amidases and novel protein groups. Hence, the unique metagenomic amidohydrolases D8_RL, BRM_Am, and MO13_Est631, which hydrolyse a wide spectrum of substrates, have been discovered. It is expected that the proposed method might be further optimized and tailored for the identification of the enzymes with the desired properties by using substrates of different structures, e.g., varying in acyl bulkiness or charge.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/10/4/445/s1. Figure S1: Growth of the selected clones on the M9 medium. Figure S2: The alignment was performed by ClustalW software. Figure S3: Analysis of the purified amidohydrolases by SDS-PAGE. Figures S4-S8: HPLC-MS analysis of the activity of D8_RL amidohydrolase. Figures S9 and S10: HPLC-MS analysis of the activity of MO13_Est631 amidohydrolase. Table S1: Metagenomic libraries used in this work. Table S2: Primers used for the amplification of the genes of the selected amidohydrolases.