Functional Characterization of the mazEF Toxin-Antitoxin System in the Pathogenic Bacterium Agrobacterium tumefaciens

Agrobacterium tumefaciens is a pathogen of various plants which transfers its own DNA (T-DNA) to the host plants. It is used for producing genetically modified plants with this ability. To control T-DNA transfer to the right place, toxin-antitoxin (TA) systems of A. tumefaciens were used to control the target site of transfer without any unintentional targeting. Here, we describe a toxin-antitoxin system, Atu0939 (mazE-at) and Atu0940 (mazF-at), in the chromosome of Agrobacterium tumefaciens. The toxin in the TA system has 33.3% identity and 45.5% similarity with MazF in Escherichia coli. The expression of MazF-at caused cell growth inhibition, while cells with MazF-at co-expressed with MazE-at grew normally. In vivo and in vitro assays revealed that MazF-at inhibited protein synthesis by decreasing the cellular mRNA stability. Moreover, the catalytic residue of MazF-at was determined to be the 24th glutamic acid using site-directed mutagenesis. From the results, we concluded that MazF-at is a type II toxin-antitoxin system and a ribosome-independent endoribonuclease. Here, we characterized a TA system in A. tumefaciens whose understanding might help to find its physiological function and to develop further applications.


Introduction
Many toxin-antitoxin (TA) systems are present on plasmids or chromosomes and consist of two small genetic elements, namely toxin and antitoxin [1]. The TA system is known to regulate cell growth or the arrest of bacteria for maintaining life under stressful environmental conditions, but its various proposed functions are still under debate, and its role is still unclear [1]. The TA systems were first identified in the plasmid F as ccdAB and in the plasmid R1 as hok-sok and kis-kid [2][3][4]. Among the type II TA systems on plasmid, the CcdAB TA system consisting of antitoxin CcdA and toxin CcdB was identified in the Escherichia coli F plasmid. CcdB inhibits DNA replication through the interaction with

Cloning of mazE, mazF, mazEF, and mazF Mutants
The mazE-at, mazF-at, and mazEF-at genes in the mazEF-at operon were separately amplified through PCR using A. tumefaciens gDNA as a template, and cloned into pBAD24, pET28a, and pGEX 6p-1 with specific primer sets ( Table 2). The mazF-at mutants (MazFat-E24N, -E24Q, -E24K, -E24D, and -E24A) were constructed by site-directed mutagenesis with the use of the mutagenic oligonucleotide primers ( Table 2) using pBAD24-mazF-at as template. The PCR products were digested with DpnI to remove the template DNA and transformed into the E. coli DH5α strain. All point mutations were confirmed using DNA sequencing.

Cellular Target Assay of DNA, RNA, and Protein Synthesis In Vivo
The pBAD24-mazF-at was transformed into E. coli BW25113. The transformed cells were inoculated in M9 medium with 0.5% glycerol and cultured at 37 • C. The culture was grown until the OD 600nm of the culture reached 0.4, and then MazF-at was overexpressed by induction with arabinose to a final concentration of 0.2%. Each 0.4 mL cell culture was sampled at 0, 5, 10, 30, and 60 min after the induction. Each sample was mixed with 10 µCi of [ 3 H]-thymidine, 10 µCi of [ 3 H]-uridine, or 30 µCi of [ 35 S]-methionine, and 30 µg of nonradioactive thymidine, 30 µg of nonradioactive uridine, or 80 µg of nonradioactive methionine, respectively, and incubated at 37 • C for 30 sec. The rates of DNA, RNA, and protein synthesis were analyzed as previously described [19]. Table 2. Primers used in this study.

Protein Expression and Purification
To purify the N-terminal His 6 -tagged MazE-at, MazF-at, and MazEF-at complex, E. coli BL21(DE3) harboring pET28a-mazE-at, -mazF-at, or -mazEF-at was grown in LB medium at 37 • C. The culture was grown until the OD 600nm of the culture reached 0.5, and then MazE-at, MazF-at, or MazEF-at was overexpressed by the induction with isopropyl-βd-1-thiogalactoside (IPTG) (Merck, St. Louis, MO, USA) to a final concentration of 0.5 mM. After the induction, the cells were grown for 6 h at 18 • C. The cultured cells were harvested by centrifugation (4392× g for 30 min at 4 • C) and disrupted using a French pressure cell press. After the centrifugation (12,000× g for 30 min at 4 • C), soluble fractions were applied to Ni-NTA agarose (Qiagen, Hilden, Germany) following the manufacturer's protocol. The purified proteins were applied to 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene difluoride membrane (PVDF), and a Western Blot analysis was performed using the polyclonal anti-histidine tag antibody (#2365) (cell signaling Technology, Danvers, MA, USA).

mRNA Stability Analysis Using Northern Blot
The pBAD24-mazF-at was transfected into E. coli BW25113 cells. The transformed cells were cultured in M9 medium with 0.5% glycerol at 37 • C. The culture was grown until the OD 600nm of the culture reached 0.4, then MazF-at was overexpressed by induction with arabinose to a final concentration of 0.2%. Non-induced samples were prepared without the induction by arabinose in the same growth conditions as the induced samples. Aliquots of the cell cultures were collected at 0, 5, 10, 30, and 60 min after the induction. Total RNA was extracted by TRIzol (Thermo Fisher Scientific, Waltham, MA, USA) using the manufacturer's protocol. The 30 µg of total RNA was applied to each lane onto denatured agarose gel. Northern blot was performed as previously described, with the modification that the probes were labeled with biotin-14-dCTP [20].

Secondary Structure Analysis by Circular Dichroism
Far-UV CD measurements were obtained using an automated Chirascan CD spectrometer (Applied Photophysics Ltd., Leatherhead, UK). The far-UV spectra were determined over a wavelength range of 190-260 nm using 0.5-mm path length cell at 25 • C. All experiments were performed in 20 mM Tris HCl, pH 7.5, and then adjusted to a 0.25 mg/mL protein concentration of purified MazE-at and MazF-at for CD analysis. The background CD spectrum of the buffer was subtracted from each spectrum. The percentage of secondary structure content was calculated using the CDNN software [21].

Protein Interaction Analysis by GST Pull-Down Assay with Beads
The pGEX6p-1 (Glutathione S-transferase (GST) only), pGEX-6-1-mazE-at (GST-MazEat), and pET28-mazF-at (6H-MazF-at) were transformed into E. coli BL21(DE3) and cultured in LB medium with 100 µg/mL ampicillin for pGEX6p-1 and 50 µg/mL kanamycin for pET28. When the OD 600nm of the cultures reached 0.5, 0.5 mM IPTG was added to induce the expression of GST, GST-MazE-at, and N-terminal His 6 -tagged MazF-at for 6 h at 18 • C. The cultured cells were harvested by centrifugation (4392× g for 30 min at 4 • C) and disrupted using a French pressure cell press. The cell debris were removed by centrifugation (12,000× g for 30 min at 4 • C), and soluble fractions were collected. GST only and GST-MazE-at were immobilized onto glutathione-Sepharose 4B beads by a gentle shaking motion on a rotating platform for 3 h at 4 • C, collected by centrifugation (1250× g for 1 min), and washed three times with PBS solution. The soluble fraction of MazF-at was incubated with GST beads by a gentle shaking motion on a rotating platform for 3 h at 4 • C and was collected by centrifugation (1250× g for 1 min). Beads were washed three times with PBS solution, separated by SDS-PAGE, and detected by Western Blotting with an anti-histidine antibody (#2365) (cell signaling Technology, Danvers, MA, USA).

MazF-at Is Toxic in E. coli and Is Neutralized by MazE-at
Using the toxin-antitoxin database (TADB 2.0), we identified 13 candidates (Table S1) for TA systems in A. tumefaciens C58. We found that six clones (Atu0940, Atu2017, Atu0674, Atu0934, Atu1004, and Atu8169) out of 13 showed toxicity ( Figure S1, Table S1) [22]. With this screening of TA systems, we also found that two genes, mazE-at (Atu0939) and mazF-at (Atu0940), later classified as one of the mazEF homologs, had been predicted as a potential TA system in the chromosome of A. tumefaciens. The two genes, mazE-at and mazF-at, are located in the same operon, and the stop codon of mazE-at overlaps with the start codon of mazF-at by one nucleotide (Figure 1a, asterisk). The two proteins, MazE-at and MazF-at, consist of 89 and 119 amino acid residues, respectively, and the pI values are 5.35 and 8.31, respectively ( Figure 1a). MazF-at (Atu0940) was identified as pemIK TA system in the previous study [15]. They found six candidates of TA modules in A. tumefaciens C58, and five of those six candidates showed growth inhibition in A. tumefaciens gv3101 strains [15]. In this study, we further evaluated whether the two genes, mazE-at and mazF-at, function as a TA system. mazE-at, mazF-at, and mazEF-at were amplified using genomic DNA extracted from A. tumefaciens and cloned into the pBAD24 vector. The plasmids pBAD24-mazE-at, mazF-at, and mazEF-at were transformed into E. coli BW25113 cells, and cell toxicity was tested on plates with or without 0.2 % L-arabinose. When MazE-at or MazEF-at were induced, transformant colonies were formed on the plate. However, when MazF-at was induced in the presence of 0.2 % arabinose, no colonies were formed. We also examined the effect of MazF-at inducing cell growth in a liquid culture ( Figure 1b). As shown in Figure  1c, growth inhibition was immediately observed after the induction of MazF-at but not MazE-at or MazEF-at in the presence of 0.2% arabinose. The results indicated that MazF-at inhibits cell growth as a toxin of the TA system, and MazE-at neutralizes the MazF-at toxicity as an antitoxin of the TA system in the same operon. cells, and cell toxicity was tested on plates with or without 0.2 % L-arabinose. When MazEat or MazEF-at were induced, transformant colonies were formed on the plate. However, when MazF-at was induced in the presence of 0.2 % arabinose, no colonies were formed. We also examined the effect of MazF-at inducing cell growth in a liquid culture ( Figure  1b). As shown in Figure 1c, growth inhibition was immediately observed after the induction of MazF-at but not MazE-at or MazEF-at in the presence of 0.2% arabinose. The results indicated that MazF-at inhibits cell growth as a toxin of the TA system, and MazE-at neutralizes the MazF-at toxicity as an antitoxin of the TA system in the same operon.  (1) were incubated on the M9 plate with or without 0.2% L-arabinose; (c) Effect of MazF-at on cell growth induction. The culture of E. coli BW25113 harboring pBAD24-mazE-at (black circle), -mazF-at (black square), or -mazEF-at (black rhombus) was grown until the OD 600 nm (optical density) of the culture reached ~0.4, followed by adding 0.2% L-arabinose at 0 min, and bacterial growth was measured for 6 h at OD600 nm. Error bars represent the standard error of the mean (SEM).

The MazF-at belongs to the MazF Family and E24 of MazF-at is an Active Residue for its Toxicity
The MazF-at is one of the MazF homologs in A. tumefaciens. MazF is reported in E. coli as an mRNA endoribonuclease that has the ability to cleave mRNA. To explore this  (1) were incubated on the M9 plate with or without 0.2% L-arabinose; (c) Effect of MazF-at on cell growth induction. The culture of E. coli BW25113 harboring pBAD24-mazE-at (black circle), -mazF-at (black square), or -mazEF-at (black rhombus) was grown until the OD 600nm (optical density) of the culture reached~0.4, followed by adding 0.2% L-arabinose at 0 min, and bacterial growth was measured for 6 h at OD 600 nm. Error bars represent the standard error of the mean (SEM).

The MazF-at Belongs to the MazF Family and E24 of MazF-at Is an Active Residue for Its Toxicity
The MazF-at is one of the MazF homologs in A. tumefaciens. MazF is reported in E. coli as an mRNA endoribonuclease that has the ability to cleave mRNA. To explore this possibility, we analyzed the global sequence alignment using the amino acid sequences of MazF-at and MazF-ec. The results showed that these two proteins have 33.3% identity and 45.5% similarity (Figure 2a). From a previous study on the tertiary protein structure of MazF-ec, a purified MazF-ec E24A mutant was used because it is impossible to purify wild-type MazF-ec from E. coli due to its toxicity. The E24A mutant is shown to have less toxicity than the non-mutated form, and the glutamic acid (E24) of MazF-ec has been determined as an active residue for the mRNA endoribonuclease activity [23]. In a recent study on the crystal structure of the E24 mutant, it is shown that the E24A mutant of MazF-ec affected substrate recognition [24,25]. Therefore, we created MazF-at mutants such as E24N, E24Q, E24K, E24D, and E24A in MazF-at and tested the cell growth on plates and in liquid medium with or without 0.2% L-arabinose. Glutamic acid was replaced by a different type of amino acids, such as polar uncharged side chain (N and Q), positive charged side chain (K), negative charged chain (D), or hydrophobic side chain (A). The results showed that the MazF-at mutants (E24D) have only preserved their toxicity, like wild-type MazF-at, and that the other mutants, MazF-at E24N, E24Q, E24K, and E24A, did not inhibit cell growth under 0.2 % arabinose (Figure 2b). In a liquid culture, the growth curve also indicated that MazF-at (E24D) had a cell growth arrest similar to that of wild-type MazF-at (Figure 2c). This might be because aspartic acid is a negatively charged amino acid, as glutamic acid. These results suggest that the glutamic acid (E24) amino acid of MazF-at is one of the active residues responsible for the toxicity of MazF-at and the MazF homolog, which is an endoribonuclease.
Microorganisms 2021, 9, x FOR PEER REVIEW 7 of 13 possibility, we analyzed the global sequence alignment using the amino acid sequences of MazF-at and MazF-ec. The results showed that these two proteins have 33.3% identity and 45.5% similarity (Figure 2a). From a previous study on the tertiary protein structure of MazF-ec, a purified MazF-ec E24A mutant was used because it is impossible to purify wild-type MazF-ec from E. coli due to its toxicity. The E24A mutant is shown to have less toxicity than the non-mutated form, and the glutamic acid (E24) of MazF-ec has been determined as an active residue for the mRNA endoribonuclease activity [23]. In a recent study on the crystal structure of the E24 mutant, it is shown that the E24A mutant of MazFec affected substrate recognition [24,25]. Therefore, we created MazF-at mutants such as E24N, E24Q, E24K, E24D, and E24A in MazF-at and tested the cell growth on plates and in liquid medium with or without 0.2% L-arabinose. Glutamic acid was replaced by a different type of amino acids, such as polar uncharged side chain (N and Q), positive charged side chain (K), negative charged chain (D), or hydrophobic side chain (A). The results showed that the MazF-at mutants (E24D) have only preserved their toxicity, like wildtype MazF-at, and that the other mutants, MazF-at E24N, E24Q, E24K, and E24A, did not inhibit cell growth under 0.2 % arabinose (Figure 2b). In a liquid culture, the growth curve also indicated that MazF-at (E24D) had a cell growth arrest similar to that of wild-type MazF-at (Figure 2c). This might be because aspartic acid is a negatively charged amino acid, as glutamic acid. These results suggest that the glutamic acid (E24) amino acid of MazF-at is one of the active residues responsible for the toxicity of MazF-at and the MazF homolog, which is an endoribonuclease.

MazF-at Inhibits Protein Synthesis
Toxins in TA systems have various targets, such as the essential cellular process of DNA, RNA, or protein synthesis to regulate cell growth according to environmental conditions. Therefore, to evaluate the effect of MazF-at on cell growth, we examined the effect Figure 2. The identification of active residues in MazF-at for ribonuclease activity. (a) Sequence alignments between MazF-at and MazF-ec. Residues were shaded according to BLOSUM62 score. Dark-and grey-shaded residues represent identical and similar amino acid residues, respectively; (b) Toxicity of MazF-at mutants, MazF-at-E24N, -E24Q, -E24K, -E24D, and -E24A, on plates. The cells of E. coli BW25113 harboring each mazE-at mutant in pBAD24 were incubated on the M9 plate with or without 0.2% L-arabinose; (c) Effect of MazF-at mutants, MazF-at-E24N, -E24Q, -E24K, -E24D, and -E24A induction on cell growth. The culture of E. coli BW25113 harboring each mazE-at mutant in pBAD24 was grown until the OD 600nm of the culture reached~0.4, followed by adding 0.2% L-arabinose at 0 min, and bacterial growth was measured for 6 h at OD 600 nm. Error bars represent standard error of the mean (SEM).

MazF-at Inhibits Protein Synthesis
Toxins in TA systems have various targets, such as the essential cellular process of DNA, RNA, or protein synthesis to regulate cell growth according to environmental conditions. Therefore, to evaluate the effect of MazF-at on cell growth, we examined the effect of MazF-at induction with or without 0.2 % arabinose on DNA, RNA, and protein synthesis using  3c). This indicates that MazF-at inhibits protein synthesis, but not RNA and DNA synthesis.

MazF-at is an Endoribonuclease
Next, we purified MazE-at, MazF-at, and MazEF-at complex proteins using pET28-mazE-at, -mazF-at, and -mazEF-at with a His6-tag at the N-terminal end to confirm the activity of MazF-at in vitro. The molecular weight and purity of purified MazE-at, MazF-at, and MazEF-at were confirmed by SDS-PAGE and Western blot (Figure 4a). In addition, we evaluated whether MazF-at inhibited protein synthesis by decreasing cellular mRNA stability as MazF or YhaV does in E. coli using Northern blot. When MazF-at was overexpressed in the presence of 0.2% arabinose, the degradation of cellular mRNAs (ompA, ompF) was observed 5 min after induction (Figure 4b). These results suggest that MazF-at in the MazEF-at TA system has potential as an endoribonuclease in A. tumefaciens.
Next, to evaluate the in vitro endoribonuclease activity of purified MazF-at, the MazF-at or MazEF-at complex was incubated with MS2 phage RNA, and its cleavage patterns were compared with those of MazF-ec(E24A) (Figure 5c). MazF-at cleaved MS2 phage RNA from 1 min and completely digested the full-size MS2 phage RNA within 10 min (Figure 5a). Notably, MazEF-at did not cleave MS2 phage RNA even after 30 min of incubation ( Figure 5b). The results indicated that MazF-at has in vitro endoribonuclease activity and its cognate antitoxin, MazE-at, can inhibit the activity of MazF-at ( Figure 5).

MazF-at Is an Endoribonuclease
Next, we purified MazE-at, MazF-at, and MazEF-at complex proteins using pET28-mazE-at, -mazF-at, and -mazEF-at with a His 6 -tag at the N-terminal end to confirm the activity of MazF-at in vitro. The molecular weight and purity of purified MazE-at, MazF-at, and MazEF-at were confirmed by SDS-PAGE and Western blot (Figure 4a). In addition, we evaluated whether MazF-at inhibited protein synthesis by decreasing cellular mRNA stability as MazF or YhaV does in E. coli using Northern blot. When MazF-at was overexpressed in the presence of 0.2% arabinose, the degradation of cellular mRNAs (ompA, ompF) was observed 5 min after induction (Figure 4b). These results suggest that MazF-at in the MazEF-at TA system has potential as an endoribonuclease in A. tumefaciens.   Next, to evaluate the in vitro endoribonuclease activity of purified MazF-at, the MazFat or MazEF-at complex was incubated with MS2 phage RNA, and its cleavage patterns were compared with those of MazF-ec(E24A) (Figure 5c). MazF-at cleaved MS2 phage RNA from 1 min and completely digested the full-size MS2 phage RNA within 10 min (Figure 5a). Notably, MazEF-at did not cleave MS2 phage RNA even after 30 min of incubation (Figure 5b). The results indicated that MazF-at has in vitro endoribonuclease activity and its cognate antitoxin, MazE-at, can inhibit the activity of MazF-at ( Figure 5). (a) Protein purification and Western blot assay using anti-his6 antibody of MazE-at and MazF-at. 1.5 ng of purified MazE-at and MazF-at proteins with N-terminal His6-tag were separated by 15% SDS-PAGE gel, and the gel was stained with Coomassie blue (left panel). Western blot analysis was performed on PVDF, and the blots were detected by anti-his6 antibody (right panel); (b) Effect of MazF-at on cellular mRNA stability using Northern blot assay. The total RNAs were extracted from E. coli BW25113 cells harboring pBAD24-mazF-at with (induced) or without 0.2% L-arabinose (noninduced) at 0, 5, 10, 30, and 60 min after the induction, and Northern blotting was performed using the ompF or ompA mRNA specific probes labeled with biotin-14-dCT.

Characterization of Purified MazE-at and MazF-at Proteins
To investigate the secondary structure of MazE-at and MazF-at, we used far-UV circular dichroism (CD) with 0.25 mg/mL of purified proteins. The CD spectrum of MazE-at showed two negative peaks at approximately 208 and 222 nm that consist of α-helices (30.9%), β-sheets (31.5%), turns (19.7%), and random coils (18.0%) (Figure 6a). Moreover, MazF-at also has two negative peaks at approximately 208 and 222 nm, which consist of α-helices (28.9%), β-sheets (32.1%), turns (19.6%), and random coils (21.9%) (Figure 6b). The components of the secondary structures were calculated through CDNN [21]. These results suggest that the two proteins have similar secondary structure compositions. Normally, the antitoxin of type II TA systems has the ability to block the activity of its cognate toxins by forming a TA complex through protein-protein interactions. To confirm the in vitro interaction of the antitoxin MazE-at with MazF-at, we performed a pull-down assay using GST-MazE-at fusion protein. When MazF-at was incubated with GST or GST-MazEat immobilized beads, GST-MazE-at only showed the band of MazF-at with His6-tag at

Characterization of Purified MazE-at and MazF-at Proteins
To investigate the secondary structure of MazE-at and MazF-at, we used far-UV circular dichroism (CD) with 0.25 mg/mL of purified proteins. The CD spectrum of MazEat showed two negative peaks at approximately 208 and 222 nm that consist of α-helices (30.9%), β-sheets (31.5%), turns (19.7%), and random coils (18.0%) (Figure 6a). Moreover, MazF-at also has two negative peaks at approximately 208 and 222 nm, which consist of αhelices (28.9%), β-sheets (32.1%), turns (19.6%), and random coils (21.9%) (Figure 6b). The components of the secondary structures were calculated through CDNN [21]. These results suggest that the two proteins have similar secondary structure compositions. Normally, the antitoxin of type II TA systems has the ability to block the activity of its cognate toxins by forming a TA complex through protein-protein interactions. To confirm the in vitro interaction of the antitoxin MazE-at with MazF-at, we performed a pull-down assay using GST-MazE-at fusion protein. When MazF-at was incubated with GST or GST-MazE-at immobilized beads, GST-MazE-at only showed the band of MazF-at with His 6 -tag at the N-terminal end from the Western blot assay using anti-his 6 antibody. This result showed that MazE-at directly interacts with MazF-at in vitro (Figure 6c).

Discussion
Here, we identified a MazEF TA system candidate in the A. tumefaciens genome and showed that MazF-at is a toxin and that the gene overlaps by one base with MazE-at, which functions as an antitoxin for MazF-at. The PemK family in the conserved domain

Discussion
Here, we identified a MazEF TA system candidate in the A. tumefaciens genome and showed that MazF-at is a toxin and that the gene overlaps by one base with MazE-at, which functions as an antitoxin for MazF-at. The PemK family in the conserved domain database includes different toxins, such as MazF, Kid, PemK, ChpA, ChpB, and ChpAK in TA systems [26]. The Kis-Kid TA system is also known as the PemIK TA system. The Kis-Kid in plasmid R1 is known to encode the addiction modules to prevent the loss of plasmids through a mechanism known as post-segregational killing [7,25,26]. Recently, a MazEF homolog (MazF-dr) in Deinococcus radiodurans responds by an increase in ROS accumulation upon DNA damage stress [27]. Moreover, MazF in Staphylococcus aureus may contribute to reversible bacteria dormancy through activation of translation rescue, ribosome hibernation, the increase of cell wall thickness, and the decrease of cell division [28].
The amino acid alignment showed a high homology between MazF-ec from E. coli and MazF-at with 33.3% identity and 45.5% similarity. In addition, MazF-at has a high structural similarity with MazF-ec using a J-pred server ( Figure S2 and Table S2) [29]. The active residue of MazF-ec was predicted to be the 24th glutamic acid because the toxicity of MazF-ec was reduced tenfold when the 24th glutamic acid was replaced by alanine (E24A) [23]. Based on this result, we changed the 24th glutamic acid of MazF-at (MazF-at-E24A) to alanine and confirmed that the E24 of MazF-at is an active residue for its toxicity using site-directed mutagenesis in a manner similar to MazF-ec. However, the MazF-homolog from Bacillus subtilis has its catalytic key residues at R25 and T48 from the crystal structure and site-directed mutagenesis assay. Therefore, it is necessary to study whether there are more active residues of MazF-at besides E24. In addition, it is not known whether they are involved in binding to target RNA, endoribonuclease activity, or both. In order to prove this, further studies are required to determine the tertiary structure of MazF-at.
Previous studies have shown that MazF homologs in other bacteria have mRNA interferase activity to cleave mRNA. Here, we evaluated whether MazF-at also has endoribonuclease activity, as MazF does in E. coli. First, we examined whether MazF-at affects cell growth by inhibiting important cellular processes and found that MazF-at is an inhibitor for the synthesis of all cellular proteins, but not DNA and RNA synthesis. In addition, to evaluate whether MazF-at inhibits protein synthesis by decreasing cellular mRNA stability, MazF-at was overexpressed in vivo, and we found that the cellular mR-NAs (ompA, ompF) were degraded following specific time intervals. To confirm that this RNA degradation was due to MazF-at, we purified MazE-at, MazF-at, MazEF-at complex, and MazF-E24A, and the decrease in RNA stability after the incubation of purified MazF-at with MS2 phage RNA was evaluated in vitro. Ideally, all of these experiments should be done in its original host, A. tumefaciens. However, it has been proven to have toxicity with overexpressed MazF-at in A. tumefaciens, and it is difficult to obtain a sufficient amount of proteins from A. tumefaciens for in vitro studies, because its doubling time is 2.5-4 hrs. For this reason, this study was conducted using E. coli, although MazF-at may have different physiological function in A. tumefaciens. From these results, we showed that MazF-at has a ribosome-independent endoribonuclease activity because MazF-at showed endoribonuclease activity without ribosomes in vitro. Usually, ribosome-independent mRNA interferases recognize and cleave specific RNA sequences and specific lengths. The MqsR homolog from E. coli specifically cleaves intracellular mRNAs at a GCU sequence [30]. Recently, it was reported that an MazF homolog from E. coli recognized about seven nucleotide regions with extended recognition specificity for ACA and its flaking sequences [31]. The protein purification of MazF-ec or MqsR has not been possible without the modification of the protein, but an MazF homolog (MazF-cd) from Clostridium difficile, which cleaves 5bp UACAU sequences of mRNA, was successfully purified [32]. This might be because of the different level of toxicity of those toxins, but what makes this difference in toxicity is still unknown. Furthermore, the global analysis of the cleavage site of MazF showed that E. coli MazF cleaved rRNA precursors and ribosomal protein transcripts, which may inhibit ribosome biogenesis [31,33]. However, AtYoeB showed ribosome-independent non-specific nuclease activity cleaving both RNA and DNA [19]. Therefore, further study is needed to identify whether MazF-at has specific recognition sites and the length of the cleavage site.
The antitoxin of the type II TA system inhibits toxin activity by forming a TA complex through protein-protein binding. To confirm whether MazE-at also binds to MazFat through direct protein-protein interaction, we examined the pull-down assay, which showed the direct binding between MazE-at and MazF-at. Moreover, we evaluated the secondary structure of MazE-at and MazF-at. The two proteins have typical helix structures. Usually, although many antitoxins in type II TA systems have an unstructured region of free antitoxin, the disordered domain of antitoxin is highly susceptible to stress-induced proteases [34]. Interestingly, MazE-at has a stable secondary structure, unlike other antitoxins. Almost all antitoxins block the toxicity of their cognate toxins by direct binding to their active sites. MazE-at antitoxin binding to MazF-at may induce a conformational change to inhibit the activity of MazF-at.
In this study, we showed that the MazEF-at toxin-antitoxin system is located in the A. tumefaciens chromosome and one of the MazF homologs. MazF-at in the MazEF TA system has endoribonuclease nuclease activity, and its cognate antitoxin MazE-at may neutralize the toxicity of MazF-at. This might help us to find the function of TA systems in A. tumefaciens and to develop its application.