Escherichia coli ASKA Clone Library Harboring tRNA-Specific Adenosine Deaminase (tadA) Reveals Resistance towards Xanthorrhizol

Xanthorrhizol is a potent antimicrobial compound isolated from the rhizome of Curcuma xanthorrhiza. However, the mechanism of xanthorrhizol action is unknown. To screen for probable target(s), we introduced the ASKA pooled-plasmid library into Escherichia coli W3110 imp4213 and enriched the library for resistant clones with increasing concentrations of xanthorrhizol. After three rounds of enrichment, we found nine genes that increased xanthorrhizol resistance. The resistant clones were able to grow in LB medium containing 256 µg/mL xanthorrhizol, representing a 16-fold increase in the minimum inhibitory concentration. Subsequent DNA sequence analysis revealed that overexpression of tadA, galU, fucU, ydeA, ydaC, soxS, nrdH, yiiD, and mltF genes conferred increased resistance towards xanthorrhizol. Among these nine genes, tadA is the only essential gene. tadA encodes a tRNA-specific adenosine deaminase. Overexpression of E. coli W3110 imp4213 (pCA24N-tadA) conferred resistance to xanthorrhizol up to 128 µg/mL. Moreover, overexpression of two tadA mutant enzymes (A143V and F149G) led to a twofold increase in the MIC. These results suggest that the targets of xanthorrhizol may include tadA, which has never before been explored as an antibiotic target.


ASKA Library Enrichment and Screening for Xanthorrhizol Resistance
In this study, we used a complete set of ORF clones from the E. coli ASKA library to screen for gene(s) that confer resistance to xanthorrhizol. This is a plasmid-based genomic screening system that was modified from the protocol used for the triclosan target screen [11]. Cells that can grow at high concentrations of xanthorrhizol may bear plasmids containing genes that confer xanthorrhizol resistance. Such clones would likely out-compete others in the culture. The library grown in the absence of xanthorrhizol was used as a growth control. After three rounds of enrichment, clones from E. coli W3110 imp4213 harboring the ASKA pooled-plasmid library were isolated from the cultures grown with the highest concentrations of xanthorrhizol ( Figure 1). The library culture showed delayed growth in LB medium supplemented with xanthorrhizol, compared to E. coli W3110 imp4213 harboring an empty plasmid pCA24N (Figure 1). After the final round of enrichment, the clones could grow at a xanthorrhizol concentration of 256 µg/mL within 20 h of inoculation. Figure 1. Growth of the E. coli W3110 imp4213 (pCA24N) and E. coli W3110 imp4213 ASKA pooled-plasmid library culture in Luria broth media supplemented with chloramphenicol (30 µg/mL) and with various concentrations of xanthorrhizol. Growth was monitored for 20 h in an automatic incubator (TVS1 Bio-Photorecorder, Advantec, Shiga, Japan). (A) E. coli W3110 imp4213 (pCA24N) was grown in the present of xanthorrhizol 0, 8, 16 and 32 µg/mL. The growth was inhibited at 16 µg/mL; (B) E. coli W3110 imp4213 ASKA pooled-plasmid library was grown in LB media with various concentrations of xanthorrhizol after three rounds of enrichment. The xanthorrhizol concentrations used were 0, 32, 64, 128, and 256 µg/mL. The library clones that grew after the delay exhibited resistance towards xanthorrhizol.
The tadA gene encodes a tRNA-specific adenosine deaminase that specifically converts A34 to I34 in bacterial tRNA Arg2 [16]. The tadA gene is known to be essential in E. coli MG1665 [16][17][18] and Shewanella oneidensis MR-1 [19]. However, this gene is noted as non-essential in Bacillus subtilis 168 and Pseudomonas aeruginosa UCBPP-PA14 in the Online Gene Essentiality Database [20]. The tadA protein is conserved (>83%) among Gram-negative and Gram-positive bacteria, and yeast [16]. Prokaryotic tadA shares amino acid sequence similarity with the Tad2P enzymes in humans, Drosophila, and Saccharomyces [21]. Tad2P deaminates adenosine, but its deaminase domain is similar to that of the cytidine deaminase (CDA) superfamily [22]. However, tadA is specific for prokaryotic tRNA substrates and it specifically deaminates tRNA Arg2 . No eukaryotic tRNA substrates were found to be modified by tadA, except for yeast tRNA Arg [16]. Because tadA is essential in E. coli, we used the tadA gene for further analysis. Genes [15,[24][25][26]; c The Database of Essential Genes classifies nrdH as an essential gene, but a knockout mutant is present in the KEIO collection [17].
The other eight genes that were isolated from the enrichment are listed as non-essential genes. However, among these genes, the essentiality of nrdH is ambiguous because, although Gerdes et al. classified nrdH as an essential gene [18], a knockout-mutant strain of nrdH has been constructed and is present in the KEIO non-essential gene-knockout collection [17]. Although galU is not essential for bacterial growth, this gene plays an important role in bacterial virulence, for example in Francisella tularensis [27]. ydaC (synonym: rcbA) encodes a protein that is involved in maintaining the integrity of the bacterial chromosome by lowering the steady-state level of double-strand breaks [28]. Overexpression of ydaC leads to increased resistance to erythromycin [10]. mltF encodes a membrane-bound lytic transglycosylase that is responsible for the release of 1,2-anhydromuropeptides from peptidoglycan [29]. MltF (YfhD) is located in the periplasmic space and is involved in organic solvent tolerance. However, the deletion or loss of mltF does not cause any loss of such tolerance [30]. SoxS is a DNA-binding transcriptional dual regulator. This regulator activates the transcription of the soxRS regulon, which is involved in the oxidative stress-defense system of E. coli [31]. Overexpression of soxS has also been reported to increase resistance to triclosan in E. coli [32]. This increased resistance is due to the expression control of genes involved in drug efflux, thereby leading to drug resistance [33]. Furthermore, modifications of lipopolysaccharide that are mediated by SoxS can protect E. coli against various compounds and increase resistance to multiple drugs [34]. The yiiD gene encodes a predicted acetyltransferase. Little information is available regarding this gene, but in a protein interaction study, Butland et al. revealed that yiiD interacts with an acyl carrier protein that is a key player in fatty-acid chain elongation [35].

Verification of tadA Overexpression-Induced Resistance to Xanthorrhizol in E. coli W3110 imp4213 and Bacillus subtilis 168, Site-Directed Mutagenesis, and Binding Study
Our results revealed that an essential gene, tadA, was a candidate that might be responsible for increasing the resistance of the E. coli W3110 imp4213 ASKA pooled library to xanthorrhizol. The level of tadA gene overexpression was confirmed by cDNA analysis ( Figure S1). tadA cDNA was increased approximately twofold compared to the control ( Figure S1). We isolated plasmid pCA24N-tadA from the original ASKA library, re-introduced it into E. coli W3110 imp4213, and measured the MIC of xanthorrhizol against the resulting strain. E. coli W3110 imp4213 (pCA24N-tadA) exhibited xanthorrhizol resistance, with an MIC of 128 µg/mL ( Table 2). This MIC was not as high as that of the enriched library culture, a result that is likely due to the library being a mixed culture in which each clone contributes to the overall resistance towards xanthorrhizol. Table 2. Xanthorrhizol minimum inhibitory concentration s of wild-type and mutant tadA-overexpressing strains.
As described in the Introduction Section, xanthorrhizol is more effective against Gram positive bacteria, such as B. subtilis. To confirm whether tadA is involved in increasing resistance towards xanthorrhizol, we treated B. subtilis 168 overexpressing tadA with xanthorrhizol. We observed that xanthorrhizol MICs increased fourfold (Table 2). This result suggests that tadA may be one of the xanthorrhizol targets in the bacterial cell.
Site-directed mutagenesis was used to confirm that tadA was the xanthorrhizol target. Using docking analysis, we found that substitutions at Ala143 and Phe149 would be likely to have only a small effect on overall protein folding ( Figure S2). Ala143 and Phe149 are located at the C-terminus of the tadA protein. The subsequent determination of xanthorrhizol MICs confirmed that the tadA mutants A143V and F149G confer twofold increased resistance towards xanthorrhizol ( Table 2).
The study of xanthorrhizol and tadA interaction was carried out using tryptophan quenching assay. Tryptophan quenching assay has been used to study binding activity of an enzyme inhibitor [36,37]. tadA protein contains three intrinsic tryptophan residues that make it suitable for quenching assay. The binding activity is indicated by the decreasing of fluorescence signal. In this study, 4 µM tadA protein was titrated with xanthorrhizol. Fluorescence titration demonstrated that xanthorrhizol was able to bind to the protein as revealed by fluorescence change with a Kd = 19.04 μM, 109.7 μM, and 32.78 μM for wild type tadA, A143V, and F149G, respectively ( Figure 2). These results showed increasing Kd of xanthorrhizol in tadA mutants. However, the Kd for both tadA mutants was different even the MIC was the same. This could be because xanthorrhizol may have multiple intracellular targets that might contribute to its inhibition activity.
It should be noted that there have been no reports describing tRNA-specific adenosine deaminase as an antibiotic target. However, some information regarding antibiotics that target microbial RNA processing/editing enzymes is available. Among them, aminoacyl-tRNA synthetase (aaRS) is an essential enzyme and has for years been an attractive target for antibiotic development [38,39]. This enzyme catalyzes the synthesis of aminoacyl-tRNA, which plays an important role in protein biosynthesis. Targeting aaRS has also become one of the strategies to combat multidrug resistance Gram-negative bacteria [40]. Mupirocin, a natural product isolated from Pseudomonas fluorescens, is the only approved antibiotic targeting aaRS and is widely used as a topical antibiotic against S. aureus [41].  (F149G). The assays showed a response, as revealed by fluorescence signal quenching, that xanthorrhizol bound to tadA protein. Tryptophan quenching assay demonstrated that binding capacity of xanthorrhizol to tadA mutants were lower than wild type, as shown by calculated Kd 19.04 μM, 109.7 μM, and 32.78 μM for wild type, A143V, and F149G, respectively.

Modeling of the Binding Mechanism of Xanthorrhizol to Wild-Type and Mutant EcTadA
To identify the drug-resistance mechanism induced by the two mutations at sites 143 and 149, the interaction between EcTadA and xanthorrhizol was explored in detail from energetic and structural perspective. The binding free energies of xanthorrhizol with WT EcTadA and with the A143V and F149G mutants are −9.94 kcal/mol, −6.35 kcal/mol, and −6.69 kcal/mol, respectively (Table S1). Relative to wild-type EcTadA, the contribution of ΔGnonpolar is obviously decreased in the mutants (Table S1). The decrease in nonpolar interaction contribution can be caused by the loss of the van der Waals interaction contribution. The van der Waals contributions for both mutants decreased more than 5 kcal/mol relative to the wild type (Table S1).
Among the contributions of different energy components, the hydrophobic interaction, with a 50.82-kcal/mol contribution (Table S1), provides the main driving force for the binding of xanthorrhizol. A comparison of the hydrophobic surfaces (Figure 3) of the wild-type vs. either mutated EcTadA reveals that both the hydrophobicity and the size of the hydrophobic pocket are influenced by the mutations. In the wild type, the phenolic group of xanthorrhizol was embedded well into the corresponding hydrophobic pocket containing the catalytic active site. In the A143V mutant, with the substitution from a small, hydrophobic alanine to a large, hydrophobic valine, the relative cavity size of the hydrophobic pocket is decreased. Furthermore, the phenolic group of xanthorrhizol is moved away from the favorable binding pocket ( Figure 3B). In the F149G mutant, the substitution of Phe149 with a side chain-lacking glycine also results in an unfavorable hydrophobic interaction; as with the A143V mutation, the xanthorrhizol is forced to leave the favorable position ( Figure 3C). A superimposition of the representative structures of the mutants relative to the wild-type complex showed the evidence that, in the mutants, the phenolic group of xanthorrhizol is forced to leave the original and favorable substrate-binding (Figure 4). Such a change in position can influence the hydrophobic/hydrophilic interactions between the phenolic group and the residues (Asp53, His57, Glu59, Glu85, Cys87, and Cys90) of the binding pocket. Furthermore, the hydroxyl group of xanthorrhizol has moved away from the cysteine at position 87 in the mutant complexes. Thus, the hydrogen bonding interaction between the phenolic hydroxyl group and the sulfur hydrogen of Cys87 is obviously affected.  Moreover, in Figure 4, in the two studied mutant versions of EcTadA, xanthorrhizol is forced away from residues Asp108/Lys110/Thr111 (the favorable position) when compared with the wild-type system. In the mutant versions of EcTadA, the increased nonpolar interactions of residues A143V and F149G with their surrounding residues are likely responsible for a decreased binding-pocket size, and the xanthorrhizol is thus moved away from the binding pocket. Furthermore, the dimethylethenyl group of xanthorrhizol is flipped away from the binding pocket in the two mutant systems compared to wild-type EcTadA. As a result, the binding site of xanthorrhizol was moved by the mutations, and several important interactions were disturbed, including (1) the key hydrophobic interactions between the dimethylethenyl group and the surrounding hydrophobic residues; (2) the loss of the hydrogen bonds formed by the phenolic group with the side chains of Glu59/Glu85/Cys87; and (3) the gain of π-π stacking interactions between the phenolic benzene ring and His57, further decreasing binding affinity.
As described in the wet-experimental section, the mutants A143V and F149G each exhibit a twofold-decreased activity towards xanthorrhizol. Based on the above analysis, we speculate that the additional group in the side chain of each mutated residue will not form unfavorable contacts with the xanthorrhizol but will change its binding mode, which is in agreement with the root-mean-square deviation (RMSD) analysis of xanthorrhizol (see Supplemental Information Figure S3). As a result of this binding mode change, a π-π stacking interaction between the phenolic benzene moiety and His57 appears, and there is no obvious formation of the weaker hydrogen bonds between the hydroxyl group of Glu59, Glu85, and the sulfur hydrogens of Cys87.

Chemicals, Bacterial Strains, and Plasmids
The ASKA clone (−) library and plasmid pCA24N were obtained from the National Bioresource Project, National Institute of Genetics, Japan. Plasmid pHT255 was obtained from MoBiTec GmbH (Goettingen, Germany). Bacterial strains used in this study are listed in Table 3. Xanthorrhizol was prepared in the Biomaterial Research Laboratory of Prof. Hwang at Yonsei University, Korea. Table 3. Bacterial strains and primer pairs used for the site-directed mutagenesis of the tadA gene. The mutation site in the primer is underlined. Restriction enzyme site was written in bold letter.

Name
Description Source
The plates were incubated at 37 °C overnight. The following day, E. coli W3110 imp4213 cells harboring the ASKA pooled-plasmid library were scraped up and pooled into 10 mL of LB.

Xanthorrhizol-Resistance Library Enrichment and Screening
The library enrichment was done according to the procedure described by Yu et al. [11] with some modifications. The expression of the genes in the plasmid library was induced by adding IPTG at a final concentration of 100 µM. The first round of enrichment included xanthorrhizol concentrations of 0, 8, 16, 32, 64, and 128 μg/mL. The bacterial cultures were incubated in a TVS1 Bio-Photorecorder, a temperature-controlled shaking incubator (Advantec), at 37 °C for 20 h. The library culture that grew at the highest xanthorrhizol concentration was used as the inoculum for the second and third rounds of enrichment with increasing xanthorrhizol concentrations. These enrichments were repeated for a total of 3 independent experiments. After the third enrichment, the xanthorrhizol-enriched cell cultures were spread onto LB agar media supplemented with tetracycline (10 μg/mL), chloramphenicol (30 μg/mL), and an appropriate xanthorrhizol concentration. Individual colonies were chosen for plasmid isolation followed by DNA sequencing to identify the gene inserts. An overall scheme for the ASKA pooled-library screen is presented in Figure 5.

Figure 5.
A scheme of the ASKA pooled-plasmid library enrichment screen for targets that confer xanthorrhizol resistance on E. coli W3110 imp4213. The ASKA plasmid library was isolated and re-introduced into competent E. coli W3110 imp4213 cells by electroporation. The E. coli W3110 imp4213 ASKA plasmid library clones were pooled and cultivated in LB medium containing chloramphenicol (30 µg/mL) in the absence or presence of xanthorrhizol (8-512 µg/mL). Cells cultured from the highest xanthorrhizol concentration were spread onto LB agar medium containing chloramphenicol (30 µg/mL), tetracycline (10 µg/mL) and appropriate concentrations of xanthorrhizol. The resulting colonies were randomly chosen for plasmid isolation followed by DNA sequencing analysis.

Verification of tadA Overexpression-Induced Resistance to Xanthorrhizol and Site-Directed Mutagenesis
To verify resistance towards xanthorrhizol, we isolated the pCA24N-tadA plasmid from the original ASKA clone and then re-introduced it into E. coli W3110 and E. coli W3110 imp4213. E. coli W3110 imp4213 (pCA24N-tadA) was grown overnight in 3 mL of LB supplemented with tetracycline (10 μg/mL) and chloramphenicol (30 μg/mL). Approximately 1% of the culture was inoculated into fresh medium with the same antibiotics and grown in a shaking incubator at 37 °C to an OD600nm of 0.5. Plasmid-insert expression was induced in the culture by the addition of IPTG at a final concentration of 100 μM. The cells were allowed to grow for 6 h. This culture was then used as the inoculum for an MIC assay. The microdilution broth method [42] was employed to determine the MIC of xanthorrhizol against E. coli W3110 imp4213 (pCA24N-tadA). The inoculum was adjusted to an OD625nm of 0.08, and the clone was tested against a range of xanthorrhizol concentrations (0, 4, 8 16, 32, 64, 128, 256, and 512 μg/mL). The MIC determination was performed in duplicate and repeated for a total of 3 independent experiments.
We predicted that substitutions of the amino acids at position A143 to V143 and at position F149 to G149 would increase resistance towards xanthorrhizol. Thus, we designed primer pairs to introduce these substitutions (Table 3). Plasmid pCA24N-tadA from the ASKA clone was used as the template. Site-directed mutagenesis was performed using the Exchange™ Site-Directed Mutagenesis Kit (Enzynomics, Daejeon, Korea). The mutant plasmid was introduced into chemically competent DH5α (Enzynomics) and spread onto LB supplemented with chloramphenicol (30 µg/mL). The culture was then incubated overnight at 37 °C. Colonies from each transformation were chosen for plasmid isolation followed by DNA sequencing analysis to confirm the mutation. DNA sequencing analysis was performed by SolGent Co. Ltd. Plasmid isolation was performed using the HiGene™ Plasmid Isolation Kit (SolGent Co. Ltd.). All mutants were tested for their resistance to xanthorrhizol by measuring MICs. To measure the MICs of the tadA mutants, we used the same protocol as described above. This experiment was performed in duplicate and repeated for a total of 3 independent experiments.

tadA Overexpression in B. subtilis 168
B. subtilis tadA gene was amplified from genomic DNA as a template employing PCR amplification. Primer used in the amplification was provided in Table 3. The amplified gene was digested and inserted into pHT255 vector plasmid at XbaI and AatII restriction site. Construction pHT255 carrying tadA gene was done in E. coli DH5α. Subsequently, pHT255 and pHT255-tadA was introduced into naturally competent B. subtilis 168 using Groningen method [43].
B. subtilis 168 harboring empty plasmid (pHT255) and tadA gene (pHT255-tadA) was grown on LB agar plate supplemented by Cm (5 µg/mL) overnight at 37 °C. On the following day, about 1-loop colony of the cultures were inoculated into a LB + Cm (5 µg/mL) and grown until OD600nm of 0.7-0.8. Subsequently, 1 mM IPTG was added to induce tadA gene expression. The cultures continued to grow for another 6 h. These induced cultures were used as inoculum for MIC determination. MIC determination was done as mentioned above. This experiment was performed in duplicate and repeated for a total of 3 independent experiments.

Xanthorrhizol-tadA Binding Study
E. coli strains producing wild-type tadA or the tadA (A143V) and tadA (F149G) mutants were grown overnight at 37 °C in LB supplemented with chloramphenicol (30 µg/mL). About 1% of each culture was inoculated into fresh medium and incubated for approximately 2-3 h until the OD600nm reached 0.5. The cultures were then induced by the addition of 100 µM IPTG and allowed to grow for an additional 8 h. To purify his-tagged tadA protein, cells were harvested and lysed by sonication. Proteins were bound to agarose resin (His•Bind Agarose Resin (Ni-IDA), Elpis Biotech, Daejeon, Korea) and purified through column chromatography as described in the kit protocol (His•Bind Agarose Resin (Ni-IDA) Elpis Biotech). Protein was desalted by using PD-10 desalting columns following the protocols described by the manufacturer (GE Healthcare Bio-Sciences AB, Uppsala, Sweden).
A tryptophan quenching assay was used to study the xanthorrhizol-tadA interaction. tadA (4 µM) in 50 mM Tris-Cl (pH 8) and 10% glycerol was titrated with 2 µL xanthorrhizol (5 mM) and was then immediately excited at 295 nm; the emission was recorded from 300 to 500 nm using a FluoroMate FS-2 fluorescence spectrophotometer (Scinco Co. Ltd., Seoul, Korea). The excitation and emission monochromator slit widths were 2. The assay was repeated at least three times.

Molecular Docking
The EcTadA crystal structure (PDB ID: 1Z3A) was used as the input for xanthorrhizol docking using two different docking programs, GLIDE [44] and GOLD [45]. The substrate and cofactor (tRNA and Zn 2+ , respectively) were imported from the crystal structure by structural superposition [46]. Proton addition was performed with the corresponding algorithms implemented in the different docking programs. In GLIDE (version 5.8), the considered search volume was confined within a cube of 20 Å 3 . A total of 5000 poses per xanthorrhizol were retained for the initial phase of docking, and the best 400 poses per xanthorrhizol were retained for energy minimization. In GOLD (version 5.1), the search was confined within a 10-Å-radius sphere around the center defined above with the slowest and most accurate genetic algorithm search options, generating 10 poses per xanthorrhizol. All 4 built-in scoring functions in GOLD (ChemPLP, GoldScore, ChemScore, and ASP) were used in parallel docking simulations. Finally, the xanthorrhizol structure was built using the Maestro software [47] included in the Schrödinger Suite 2011 package and was energy-minimized using Macromodel [48] with the OPLS_2005 force field.

Molecular Dynamics Simulations
Molecular dynamics simulations were performed using the program Q [49] with the OPLS-AA force field [50]. Subsequently, the binding free energy was calculated as described in the supplementary materials. The parameters needed for xanthorrhizol that were not present in the original version of the force field were retrieved from the automatic parametrization performed with Macromodel [48]. Spherical boundary conditions were used with a 20-Å-radius simulation sphere centered on the same point as defined for the docking calculations. This sphere was solvated with TIP3P [51] water molecules and subjected to polarization and radial constraints according to the surface-constrained all-atom solvent (SCAAS) model [49,52] at the sphere surface to mimic the properties of bulk water. Non-bonded interactions were calculated explicitly up to a 10-Å cutoff, except for the xanthorrhizol atoms, for which no cutoff was used. Beyond the cutoff, long-range electrostatics were treated with the local reaction field multipole-expansion method [53]. All solvent bonds and angles were constrained with the SHAKE algorithm [54], and a 1-fs molecular dynamics (MD) step size was used. The MD simulations were performed at a temperature of 298 K. Non-bonded pair lists were updated every 25 steps, and the same interval was used to sample the xanthorrhizol-surrounding residue interaction energies.

Conclusions
By enriching an E. coli W3110 imp4213 ASKA pooled-plasmid library, we found that TadA may be involved in the mechanism of xanthorrhizol action. We have shown that xanthorrhizol targets TadA and, more importantly, that tRNA-specific adenosine deaminase may be a new target for antibiotic development in the future.