The bacterial strains used in this study are listed in Table 1. For all studies, bacterial strains were grown with aeration at 37°C. E. coli BL21(DE3) was grown in LB or M9 minimal media supplemented with casamino acids and 0.2% glucose or 0.2% glycerol. M. smegmatis mc² 155 was grown in Middlebrook 7H9 (a liquid medium, Difco) containing 0.2% glycerol and 0.02% Tween 80 or Middlebrook 7H10 (a solid agar medium, Difco) containing 0.2% glycerol. Chloramphenicol (Cm) or ampicillin (Amp) was added to a concentration of 50 μg/ml to cultures of E. coli, and kanamycin was added to a concentration of 50 μg/ml to cultures of E. coli and 25 μg/ml to cultures of M. smegmatis.

To construct plasmids, toxin and antitoxin genes were amplified from Mtb genomic DNA by polymerase chain reaction (PCR). The amplified PCR fragments were subcloned into the SmaI site of pUC19, and the sequences of the amplicons were verified using DNA sequencing. Using DNAs digested by the appropriate restriction enzymes, toxin genes were subsequently cloned into the NdeI-HindIII sites (for Rv2019, Rv3180c, Rv3697c, or Rv0299) or NdeI-XbaI sites (for Rv3749c) of the pBAD33 vector containing pA15 origin of replication (∼20 copies per cell) the arabinose-inducible (araBAD) promoter (Guzman et al., 1995) to generate pBAD33-Rv2019, pBAD33-Rv3180c, pBAD33-Rv3697c, pBAD33-Rv0299, and pBAD33-Rv3749c. Antitoxin genes were cloned into the NdeI-EcoRI sites of the pET21c vector containing an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible T7 promoter to generate pET21c-Rv2018, pET21c-Rv3181c, pET21c-Rv3697A, pET21c-Rv0298, and pET21c-Rv3750c.

To construct a plasmid expressing a chimeric glutathione S-transferase (GST) Rv2019 protein (pGST-Rv2019), pET21c-Rv2019 was digested with NdeI and XhoI, and the Rv2019 fragment was ligated into the NdeI-XhoI sites of the pGST vector containing an IPTG-inducible T7-lac promoter. To express the toxins in M. smegmatis, pUC19-toxin subclones were digested with NdeI and XbaI, and the resulting fragment was ligated into the NdeI-XbaI sites of the pMV306AC vector containing an acetamide-inducible promoter (Stover et al., 1991). To construct Rv2019 mutant clones, the primers listed in Supplementary Table S1 were used to amplify gene fragments, and the resulting DNA fragments harboring mutation were inserted into pBAD33, yielding pBAD33-Rv2019_D4K, pBAD33-Rv2019_E41K, and pBAD33-Rv2019_D98K. All plasmids used in this study are listed in Table 1.

Overnight cultures of BL21(DE3) harboring compatible plasmid pairs (pBAD33 and pET21c, pBAD33-Rv2019 and pET21c, pBAD33 and pET21c-Rv2018, or pBAD33-Rv2019 and pET21c-Rv2018) were diluted 10²-fold in fresh M9 minimal media containing Cm and Amp, and the diluted cultures were grown to the exponential phase at 37°C. The cells were harvested by centrifugation and resuspended in M9 minimal media to an optical density at 600 nm (OD₆₀₀) = 0.2. The resuspension was diluted 10^-1, 10^-2, 10^-3, and 10^-4-folds, and each diluted sample was spotted onto M9 agar plates containing Cm and Amp with 0.2% L-arabinose, 0.05 mM IPTG, or both or without inducers. The plates were incubated overnight at 37°C. For a neutralization assay in liquid culture, BL21(DE3) harboring pBAD33-Rv2019 and pET21c-Rv2018 were grown at 37°C to the exponential phase in M9 minimal media containing Cm and Amp. The cultures were diluted fivefold in the same minimal media, and inducers were added as follows: 0.2% L-arabinose, 0.05 mM IPTG, or both. The diluted cultures were incubated at 37°C for 2 h and continuously diluted fivefold in fresh M9 minimal media containing inducers every 2 h, and OD₆₀₀ was measured.

Overnight cultures of BL21(DE3) transformed with pGST or pGST-Rv2019 were diluted 50-fold in 1 L of fresh LB media containing kanamycin and incubated at 37°C until OD₆₀₀ reached 0.5–0.6. IPTG was added to 0.5 mM, and the cultures were further incubated with shaking (200 rpm) at 37°C for 2 h. The cells were centrifuged for 20 min at 4,000 × g, and the cell pellet was washed in 30 ml of 20 mM Tris–HCl (pH 6.8). The washed cells were centrifuged again and resuspended in 20 ml (for GST) or 66 ml (for GST-Rv2019) of lysis buffer [50 mM Tris–HCl (pH 8.0), 100 mM NaCl, 5 mM β-mercaptoethanol] with (for GST-Rv2019) or without (for GST) sarkosyl (N-lauroylsarcosine). The resuspended cells were lysed using a sonicator. Lysates were centrifuged for 25 min at 10,000 × g, and the supernatant was centrifuged for 1 h at 180,000 × g in a Beckman 70Ti rotor. The final supernatant was loaded onto a glutathione agarose column that was equilibrated with 10 column volumes of lysis buffer. The column was washed with 30 column volumes of lysis buffer, and bound proteins were eluted with 10 1-ml aliquots of elution buffer containing 10 mM reduced glutathione. The eluted proteins were dialyzed twice overnight against 2 L of storage buffer [50 mM Tris–HCl (pH 8.0), 100 mM NaCl].

The hot phenol method was used to extract total cellular RNA (Sarmientos et al., 1983). BL21(DE3) was grown to OD₆₀₀ = 1.0, and 1.5 ml of the culture was centrifuged for 2 min at 17,000 × g. Pelleted cells were resuspended in 500 μl of solution A (20 mM sodium acetate, 10 mM EDTA, 0.5% SDS), and an equal volume of hot phenol was added, followed by incubation for 10 min at 60°C. After centrifugation for 10 min at 17,000 × g, 400 μl of the aqueous phase was mixed with 40 μl of 3 M sodium acetate (pH 4.3). In total, 1 ml of ice-cold ethanol was added to the mixture, which was incubated for 30 min at -20°C. RNA was collected by centrifugation for 10 min at 17,000 × g. The pellet was resuspended in 200 μl of solution B (20 mM sodium acetate, 10 mM EDTA) and mixed with 20 μl of 3 M sodium acetate (pH 4.3). Next, 1 ml of ice-cold ethanol was added, and the sample was incubated for 30 min at -20°C. RNA was collected by centrifugation for 10 min at 17,000 × g. The pellet was washed with 200 μl of 75% ethanol and dissolved in 50 μl of distilled water. The quality of RNA was assessed using agarose gel electrophoresis, and the RNA concentration was determined using Thermo Scientific NanoDrop 2000.

For the in vivo assay, BL21(DE3) harboring pBAD33 or pBAD33-Rv2019 was grown to OD₆₀₀ = 0.6–0.7 in M9 minimal media containing Cm. The cultures were diluted fivefold in fresh M9 media supplemented with 0.2% L-arabinose, and the diluted cultures were incubated at 37°C for 2 h. Cultures were continuously diluted fivefold in fresh M9 minimal media containing inducer every 2 h. After induction for 0, 2, 4, or 6 h, equal numbers of cells were harvested according to OD₆₀₀, and the pellets were resuspended in 100 μl of 20 mM Tris–HCl (pH 6.8). An equal volume of phenol (pH 8.0) was added, which was vortexed to lyse the cells, and the aqueous phase was electrophoresed through 1.2% agarose gel. For the in vitro assay, total RNAs were extracted from BL21(DE3), as described earlier. The reaction buffer containing 50 mM Tris–HCl (pH 8.0) and the purified protein (15 pmole of GST or GST-Rv2019) was incubated with 2 μg of RNA at 37°C. At the indicated times, reactions were stopped by adding 3 μl of 6x RNA loading buffer [10 mM Tris–HCl (pH 7.5), 0.03% bromophenol blue, 0.03% xylene cyanol FF, 60% glycerol, and 60 mM EDTA]. The samples were electrophoresed through 1.2% agarose gel, and RNAs were visualized using ethidium bromide.

BL21(DE3) harboring pBAD33 or pBAD33-Rv2019 was grown to the exponential phase, and the culture was diluted 10-fold in fresh M9 minimal medium containing Cm. L-arabinose was added to the diluted culture to a final concentration of 0.2%. Next, the cultures were incubated with shaking (200 rpm) at 37°C for 4 h; the cultures were then diluted 10-fold in fresh M9 minimal media containing Cm and 0.2% L-arabinose and incubated at 37°C. After 8 h, the same amounts of cells were harvested, and total RNAs were extracted using the hot phenol method, as described above. RNase-free DNase (Qiagen) was treated to remove contaminating genomic DNA, according to the manufacturer’s instructions. After DNase treatment, the quality of total RNA was assessed using agarose gel electrophoresis, and the RNA concentration was determined using Thermo Scientific NanoDrop 2000. Complementary DNA (cDNA) was synthesized using the Superscript III First Strand Synthesis for RT-PCR Kit (Invitrogen) with 1 μg of total RNA. The StepOne Plus Real-time PCR System (Applied Biosystems) was used to amplify 1 μl of cDNA added to SYBR Green master mix (Qiagen). The eight pairs of primers used for qRT-PCR are listed in Supplementary Table S1. Thermocycling conditions were as follows: initial denaturation for 15 min at 95°C, followed by 40 cycles at 95°C for 5 s, 55°C or 59°C (only for reaction using the glyS2 primers) for 30 s, and 72°C for 50 s.

The shuttle vector pMV306AC lacking a mycobacterial origin of replication was used for integration, which occurs between the attP site of the vector and the attB site of M. smegmatis genomic DNA (Stover et al., 1991). Rv2019, Rv2018-2019, Rv3180c, Rv3180c-3181c, Rv3697c, Rv3749c, or Rv0299 was each cloned into the NdeI-XbaI sites of the acetamide-inducible pMV306AC vector to generate pMV306AC-Rv2019, pMV306AC-Rv2018-2019, pMV306AC-Rv3180c, pMV306AC-Rv3180c-3181c, pMV306AC-Rv3697c, pMV306AC-Rv3749c, or pMV306AC-Rv0299, respectively. Electroporation was used to introduce the plasmids into M. smegmatis mc² 155 (Snapper et al., 1990). Single colonies of transformed cells were selected by plating the cultures onto agar plates containing kanamycin. Colony PCR was performed to confirm the integration of the plasmid into the bacterial genome (data not shown).

The Mtb genome encodes 79 TA systems as follows: 10 MazF homologs, 50 VapC homologs, 2 RelE homologs, 3 HigB homologs, 2 ParE homologs, 1 YoeB homolog, and 11 putative TA systems (Sala et al., 2014). Notably, because of the numerous TA systems in Mtb, some TA modules were characterized to identify their TA functions; however, the functional mechanisms of most Mtb TA systems are unknown. Therefore, in the present study, in order to elucidate functions of those TA systems, we selected 4 VapBC TA systems (Rv2019-Rv2018, Rv3697c-Rv3697A, Rv3180c-Rv3181c, and Rv3749c-Rv3750c) and 1 MazEF TA system (Rv0299-Rv0298) of Mtb whose toxicities are still unknown in Mtb as well as in E. coli.

VapC toxin homologs share sequence similarities with the PIN domain, which is a PilT N-terminal domain (Arcus et al., 2011). The three-dimensional structure of the active site of PIN domain includes four highly conserved acidic residues that coordinate the divalent metal ion. Notably, a cluster of metal ion-associated acidic residues forms a cavity and plays an important role in ribonuclease activity (Das et al., 2014). First, we compared the amino acids sequences of four VapC homologs with those of Mtb VapC1 and VapC30, each of which contains well-conserved acidic residues. As shown in Figure 1A, the alignment of the amino acids sequence of the Rv2019 toxin with those of the Mtb VapC homologs (VapC1 and VapC30) showed relatively low identities (6.87–10.52%) and similarities (10.68–20.61%). Although the overall sequence conservation was poor, the three acidic residues Asp4, Glu41, and Asp98 in Rv2019 were well conserved. However, the fourth acidic residue was not conserved. When the other three VapC homologs (Rv3180c, Rv3697c, and Rv3749c) were aligned with Mtb VapC1 and VapC30, three acidic residues were less well conserved (Supplementary Figure S1). All Mtb MazF toxins appeared distantly related to known MazF proteins of other bacteria (Ramage et al., 2009), and in the case of Rv0299 toxin as a MazF homolog, the similarity and identity scores of Rv0299 and MazF homologs in E. coli, Bacillus subtilis, and Staphylococcus aureus were 9.9–12.61% and 5–8%, respectively (Figure 1B).

To assess the toxic effects of five putative Mtb TA systems on E. coli, we transformed BL21(DE3) with the pBAD33-toxin plasmid and induced the expression of toxin genes using 0.2% L-arabinose. The expression of the five mycobacterial TA systems in cells grown on solid M9 media showed only that the expression of Rv2019 inhibited colony formation of BL21(DE3) (Figure 2A). Further, the expression of Rv3180c, Rv3697c, Rv3749c, and Rv0299 did not influence colony formation of BL21(DE3). The open reading frames of Rv2018, Rv3181c, Rv3697A, Rv3750c, and Rv0298 were ligated to the pET21c vector containing an IPTG-inducible (T7) promoter. In the presence of 0.05 mM IPTG, the expression of the antitoxins did not affect cell growth, suggesting that the antitoxin proteins were not toxic. Further, only the Rv2019 toxin severely inhibited the growth of E. coli when each of the five toxins was expressed in cells cultured in liquid M9 media (Figure 2B). These results suggest that Mtb Rv2019 functions as a toxin, at least in E. coli.

Next, to investigate whether the three acidic residues are indispensable for the toxicity of the Rv2019 toxin, the mutations D4K, E41K, and D98K were individually introduced into the primary sequence of the Rv2019 toxin. In Figure 2C, the expression of the Rv2019_D4K mutant inhibited colony formation of BL21(DE3) to an extent similar to that of wild-type Rv2019. In contrast, the Rv2019_E41K and Rv2019_D98K mutants did not inhibit colony formation of BL21(DE3). These results suggest that Glu41 and Asp98, but not Asp4, play a pivotal role in the toxicity of Rv2019 and coordination of metal ion and that the toxicity is solely due to the expression of Rv2019.

In exponentially growing healthy cells, all toxins are constitutively expressed and interact with their cognate antitoxins, either protein or RNA molecules, except for the Type IV TA system. This association prevents the function of a toxin by neutralizing its toxicity. Therefore, to examine whether the toxic effect of Rv2019 was neutralized by its cognate Rv2018 antitoxin, E. coli BL21(DE3) was cotransformed with compatible plasmid pairs (pBAD33 and pET21c, pBAD33-Rv2019 and pET21c, pBAD33 and pET21c-Rv2018, or pBAD33-Rv2019 and pET21c-Rv2018). In BL21(DE3) cotransformants, the expression of the gene encoding the Rv2019 toxin or the Rv2018 antitoxin was independently induced by 0.2% L-arabinose or 0.05 mM IPTG and coinduced when 0.2% L-arabinose and 0.05 mM IPTG were added together. We found that the expression of the Rv2019 toxin inhibited colony formation of BL21(DE3) in the presence of arabinose (Figure 3A), which is consistent with the previous result (Figure 2A). However, coexpression of the Rv2019 toxin and the cognate Rv2018 antitoxin recovered normal colony formation of BL21(DE3) in the presence of both inducers. Similar results were obtained using liquid cultures (Figure 3B). Furthermore, a reciprocal exchange in expression vectors yielded the same results (Supplementary Figure S2A). Thus, these findings demonstrate that the expression of the Rv2019 toxin is toxic to E. coli and its cognate antitoxin Rv2018 counteracts the toxic effect of the Rv2019 toxin.

In many bacteria, various VapC homologs exhibit ribonuclease activity in vitro or in vivo (Ramage et al., 2009; Maezato et al., 2011; Winther and Gerdes, 2011; McKenzie et al., 2012a; Winther et al., 2013; Lopes et al., 2014). Therefore, we decided to investigate the ribonuclease activity of the Rv2019 toxin in vivo. First, we determined the amount of and pattern changes of total nucleic acids extracted from cells after inducing the expression of the Rv2019 toxin. The expression of the Rv2019 toxin was induced by adding L-arabinose to exponential phase cells harboring pBAD33 or pBAD33-Rv2019, and OD₆₀₀ was measured at 2-h intervals. The same number of cells was harvested at each time point, and total nucleic acids were extracted using phenol. In the presence of L-arabinose, control cells exhibited normal exponential growth, and the level of total nucleic acids remained constant after induction (Figure 4A). However, after the expression of the Rv2019 toxin, we found that 23S rRNA level gradually decreased, and that more interestingly, the amount of small RNAs increased for 6 h. The intensities of the 23S and 16S rRNA and small RNA bands were analyzed using ImageJ software. Figure 4B shows that after the induction of the Rv2019 toxin, the levels of 23S and 16S rRNAs substantially declined after 6 h, and the levels of small RNAs increased (Figure 4C), indicating the accumulation of fragments of 23S and 16S rRNAs. The amount of genomic DNA at the top of the gel remained constant, indicating the RNA-specific activity of Rv2019. These results suggest that the Rv2019 toxin degrades rRNA and that its activity inhibits the growth of BL21(DE3).

To detect the ribonuclease activity of the Rv2019 toxin in vitro, we cloned pGST-Rv2019, which produces a GST-Rv2019 chimera, and tested its effects on the growth of BL21(DE3) (Supplementary Figure S3B). The coexpression of GST-Rv2019 and Rv2018 recovered the normal growth of BL21(DE3) (Supplementary Figure S3C). Next, we extracted total RNAs from exponentially growing BL21(DE3) and purified GST and GST-fused Rv2019 (GST-Rv2019). Purified GST or GST-Rv2019 (15 pmoles each) was incubated with 2 μg of total E. coli RNAs for 90 min at 37°C in 50 mM Tris–HCl (pH 8.0). As shown in Figure 5A, total E. coli RNAs were cleaved by purified GST-Rv2019, and the activity increased with time. Moreover, RNA was not detectably digested in the absence of GST-Rv2019 or by GST. Notably, the intensity of 23S rRNA abruptly decreased after 10 min; however, 16S rRNA began to disappear after 30 min (Figure 5B, left) with a concomitant increase in the intensities of smaller RNAs (Figure 5B, right). Therefore, the purified Rv2019 toxin degrades rRNA in vitro as well as in vivo. Interestingly, the in vitro results showed that the Rv2019 toxin preferentially cleaved 23S rRNA. Unfortunately, although the Rv2018 antitoxin antagonized the activity of the GST-Rv2019 toxin in vivo (Supplementary Figure S3C), the addition of a purified N-terminal His₆-tagged Rv2018 (antitoxin) to the reaction mixture did not inhibit the ribonuclease activity of the toxin (Supplementary Figure S4A). It is therefore very likely that improper association conditions inhibit the binding of the antitoxin to the toxin, as we confirmed the counteracting activity of His-Rv2018 to GST-Rv2019 in vivo (Supplementary Figure S4B).

The activities of VapC toxins of various bacteria depend on Mg²⁺, Mn²⁺, or both (Ahidjo et al., 2011; McKenzie et al., 2012a,b; Lopes et al., 2014). Therefore, we tested the effects of Mg²⁺ and Mn²⁺ on the ribonuclease activity of the Rv2019 toxin. In this assay, the reaction was incubated for 10 min instead of 90 min to detect the enhanced activity in the presence of 0.0015, 0.015, 0.15, 1.5, or 15 mM metal ions. Increasing the concentrations of metal ions did not enhance the catalytic activity, and the 23S and 16S RNAs were not digested in the presence of EDTA. When we performed the reactions in the presence of Cu²⁺, Ca²⁺, or Co²⁺, the results were the same (Supplementary Figure S4C). This is likely due to the incorporation of metal ions into the proteins during cultivation and purification, suggesting that metal ions are required for the ribonuclease activity of Rv2019 (Figure 5C).

Next, we determined whether Rv2019 digested mRNAs. For this purpose, we selected the E. coli BL21(DE3) genes (ileS and dnaA, ∼3 kb; glyS, ∼2 kb; and holA and rpoA, ∼1 kb) and used qRT-PCR to amplify specific regions of each. The expression of the Rv2019 toxin was induced by adding L-arabinose to exponential phase cells harboring pBAD33 or pBAD33-Rv2019, and total RNAs were extracted after induction for 8 h. To amplify specific regions of five mRNAs, we designed primer sets that generate an approximately ∼400-bp amplicon (Figure 6A). The cDNA synthesized from total RNA was subjected to qRT-PCR, and the differences in relative expression for each gene are shown in Figure 6B. The relative mRNA amounts of ileS1, ileS2, dnaA1, dnaA2, glyS1, glyS2, and holA were significantly lower (∼9-fold) when the Rv2019 toxin was expressed. However, the transcript amount of rpoA was marginally lower in the presence of the Rv2019 toxin, suggesting that Rv2019 may not cleave the rpoA transcript. Therefore, our results show that ileS, dnaA, glyS, and holA mRNAs serve as substrates for the Rv2019 toxin and that the seven specific regions of these mRNAs contain the sequence cleaved by Rv2019. Based on these results, mRNA was likely digested by the ribonuclease activity of the Rv2019 toxin.

So far, we determined the toxicities of five putative Mtb toxins to E. coli and found that only the Rv2019 toxin caused toxicity, presumably through its ribonuclease activity. Therefore, we assessed the toxic effects of five candidate Mtb TA systems on a non-pathogenic species M. smegmatis, which is closely related to Mtb. For this purpose, we constructed expression vectors in pMV306AC, which harbors an acetamide-inducible promoter and integrates into the attB site of M. smegmatis genomic DNA. The resulting plasmids pMV306AC-Rv2019, pMV306AC-Rv2018-2019, pMV306AC-Rv3697c, pMV306AC-Rv3180c, pMV306AC-Rv3180c-3181c, pMV306AC-Rv0299, and pMV306AC-Rv3749c were individually transformed and integrated into the genome of M. smegmatis, as described in the section “Materials and Methods.” The initial transformation results showed that the transformation efficiency of pMV306AC-Rv2019 was significantly lower than those of the other vectors (Figure 7A, plate 2). However, the transformation of pMV306AC-Rv2018-2019 resulted in the similar colony number with that obtained from control vector (Figure 7A, plate 6). This suggests that transient expression of Rv2019 is toxic to M. smegmatis and this toxicity was reverted by the expression of antitoxin, Rv2018. Interestingly, when we streaked colonies formed on the initial plates, only the Rv3180c toxin inhibited colony formation of M. smegmatis in the presence of 0.2% acetamide (Figure 7B, sector 3). This toxicity was neutralized by the antitoxin Rv3181c (Figure 7B, sector 7). Further, we found that only the Rv3180c toxin inhibited the growth of M. smegmatis after the addition of acetamide to cells cultured in liquid Middlebrook 7H9 containing 0.2% glycerol and 0.02% Tween 80 (data not shown). Peculiarly, however, the cells transformed with pMV306AC-Rv2019 did not exhibit an inhibitory phenotype upon induction (Figure 7B, sector 2). These findings raise the possibility that mutation, rearrangement, or both of the promoter or the locus harboring Rv2019 prevents the expression of Rv2019, and sequencing analysis revealed that transformants of pMV306AC-Rv2019 harbor several spontaneous mutations in the open reading frame of Rv2019 (data not shown). Therefore, our results demonstrate that the expression of Rv2019 and Rv3180c inhibited the growth of M. smegmatis, and their cognate antitoxins counteracted the toxicities of toxins.