Genome-Wide Screening for Identification of Novel Toxin-Antitoxin Systems in Staphylococcus aureus

Recent advances in genome sequencing and bioinformatics research have demonstrated that most pathogenic bacteria harbor a large number of chromosomally encoded toxin-antitoxin (TA) modules. However, little is known about the TA systems in S. aureus. Here, we newly identified four S. aureus TA systems using a combination of manual base-by-base screening and functional analysis in E. coli. Moreover, all toxins of the identified TA systems caused growth inhibition in the native host S. aureus. Although the newly identified TA systems did not exhibit sequence similarity with known bacterial TA systems, their orthologues were conserved only among other Staphylococcus species, indicating their uniqueness to staphylococci. Our approach opens the possibility for studying unannotated TA systems in various bacterial species.

ABSTRACT

Toxin-antitoxin (TA) systems consist of toxin-inhibiting diverse cellular functions (e.g., DNA replication, transcription, and translation) and a noncoding RNA or protein antitoxin. TA systems are associated with various cellular events, such as stress responses, programmed cell death, and bacterial pathogenicity. Recent advances in genome sequencing and bioinformatics research have demonstrated that most bacteria harbor various kinds of TA modules on their chromosomes; however, there is little understanding of chromosomally encoded TA systems in the Gram-positive pathogen Staphylococcus aureus. Here, we report on newly discovered S. aureus TA systems, each of which is composed of two proteins. Manual search and gene operon prediction analysis identified eight 2-gene operons as potential candidates for TA systems. Subsequently, using an Escherichia coli host killing and rescue assay, we demonstrated that four of the eight candidates worked as TA systems, designated tsaAT, tsbAT, tscAT, and tsdAT. Moreover, the TsaT, TsbT, TscT, and TsdT toxins inhibited S. aureus growth, and the toxicity of TsbT was neutralized by coexpressing the tsbA gene in the native host, S. aureus. Further, the bioinformatics analysis of the gene clusters revealed that TsaAT, TsbAT, TscAT, and TsdAT did not exhibit sequence similarity to known bacterial TA systems, and their homologues were present only within Staphylococcus species and not among any other bacteria. Our results further advance not only the understanding of S. aureus TA systems but also the study of unannotated TA systems in various bacterial species.

IMPORTANCE Recent advances in genome sequencing and bioinformatics research have demonstrated that most pathogenic bacteria harbor a large number of chromosomally encoded toxin-antitoxin (TA) modules. However, little is known about the TA systems in S. aureus. Here, we newly identified four S. aureus TA systems using a combination of manual base-by-base screening and functional analysis in E. coli. Moreover, all toxins of the identified TA systems caused growth inhibition in the native host S. aureus. Although the newly identified TA systems did not exhibit sequence similarity with known bacterial TA systems, their orthologues were conserved only among other Staphylococcus species, indicating their uniqueness to staphylococci. Our approach opens the possibility for studying unannotated TA systems in various bacterial species.

INTRODUCTION

Toxin-antitoxin (TA) systems consist of a toxin that inhibits essential cellular functions (e.g., DNA replication, transcription, translation, cytoskeleton formation, cell wall biosynthesis, and membrane integrity) and its cognate antitoxin, which neutralizes the toxicity (1, 2). The first TA system, CcdBC, was identified in the Escherichia coli F plasmid as a plasmid-located killer gene involved in plasmid maintenance (3). Since then, TA systems have been widely discovered not only on plasmids but also on bacterial chromosomes, with chromosomally encoded TA systems thought to be involved in processes such as stress response, stabilization of genomic parasites, programmed cell death, pathogenicity, and the formation of persister cells in the presence of antibiotics (4–6). TA systems are currently classified into six groups (types I to VI) based on the nature of the antitoxins. The antitoxins are composed of noncoding RNA in type I and type III TA systems, whereas both the toxin and antitoxin are composed of proteins in the type II, IV, V, and VI TA systems. In the type II and IV TA systems, antitoxins directly bind to the toxin (type II) and indirectly compete for binding to a target of the toxin (type IV). The type V antitoxins have RNase activity for toxin mRNA degradation, and the type VI antitoxins are proteolytic adaptors for the degradation of toxin proteins (1, 7). Recent advances in genome sequencing and bioinformatics research have demonstrated that most bacteria harbor a large number of TA modules on their chromosomes, and there is a tendency of pathogenic bacteria to possess many chromosomally encoded TA systems. E. coli strain K-12 contains at least 36 TA systems, and Mycobacterium tuberculosis H37Rv harbors 79 chromosomal TA systems, in contrast to its nonpathogenic counterpart, Mycobacterium smegmatis, which contains only two TA systems (8). Moreover, in M. tuberculosis, uropathogenic E. coli, and Salmonella enterica serovar Typhimurium, type II TA systems contribute to intracellular growth and survival in host animals (2, 9–11).

Staphylococcus aureus is a Gram-positive pathogen that causes a wide range of illnesses from minor skin infections to life-threatening diseases, e.g., abscess, toxic shock syndrome, food poisoning, pneumonia, and sepsis. Several type I and type II TA systems in S. aureus have been identified by sequence similarity analysis with known bacterial TA systems, whereas type III, IV, V, and VI TA systems have not been identified. Two type II TA systems, MazEF_sa and YefMYoeB_sa, are highly conserved on the chromosome of various S. aureus strains. Moreover, certain strains harbor additional TA systems on the mobile genetic elements, such as pathogenicity islands and plasmids (2, 12, 13). Two type I TA systems, SprA1_ASPepA1 and SprG/SprF, are encoded on the mobile genetic element νSaβ and pathogenicity island ΦSa3 of strain N315, respectively (13, 14). Furthermore, a type II TA system, PemIKsa, is present on S. aureus plasmids and is encoded on the chromosomes of other staphylococcal species (15). However, little is known about the TA systems in S. aureus, particularly the chromosomally encoded TA systems that are thought to be involved in S. aureus pathogenicity, such as its persistence, biofilm formation, and multidrug tolerance. Thus, we hypothesized that S. aureus might contain a large number of unannotated TA systems on its chromosome, and, therefore, we attempted to identify novel TA systems in which both the toxin and antitoxin are composed of proteins on the S. aureus chromosome. Genetic variation in S. aureus is very extensive, and most virulence factors and antimicrobial resistance factors are associated with genetic variations, including pathogenicity islands and mobile genetic elements. To eventually elucidate the mechanism of S. aureus pathogenicity, drug resistance, persister formation, and stress response, here, we used the whole-genome sequence of the strain N315, which is one of the representative genomes for S. aureus. N315 is a hospital-acquired methicillin-resistant S. aureus strain and also has a large variety of virulence factors. Some of the newly identified TA systems in this study may play important roles in the physiology of S. aureus and may be involved in its pathogenicity.

The success of our approach provides a better understanding of S. aureus TA systems and the possibility of the identification of unannotated TA systems in other bacteria.

RESULTS

Manual search for putative TA systems on the S. aureus genome.

Rapid Automated Scan for Toxins and Antitoxins in Bacteria (RASTA-Bacteria) software (16) newly detected two pairs of putative type II TA systems on an S. aureus pathogenicity island (SaPIn1). However, the genes were already annotated as having other functions, namely, SA1833/transcriptional activator (Str), SA1832/excisionase (Xis), and SA1804/phage repressor (Cro) (see Table S1 in the supplemental material). In addition, the TA finder found a putative type II TA system consisting of SA2060 (antitoxin) and SA2061 (toxin) (see Table S2 in the supplemental material). However, SA2060, assigned as an antitoxin, exhibited a strong toxicity to E. coli (see Fig. S1 in the supplemental material). Therefore, we manually searched the entire genome sequence of S. aureus strain N315 and qualified putative TA system genes according to certain criteria (see Materials and Methods), not based on sequence similarity to known toxins and antitoxins. Consequently, eight two-gene operons were identified as potential candidate genes of TA systems (1 to 8) on the S. aureus chromosome and were found to be conserved in other S. aureus strains (Table 1; see also Table S3 in the supplemental material). The gene pair of SA0287 and SA0288 was excluded from these candidates because they have similar sequences and, therefore, form a tandem duplication.

TABLE 1.

Staphylococcus aureus candidate TA systems

No.	Gene ID	Size (aa)	Toxicity to E. coli	Neutralization	NCBI Gene ID	Gene name	Characterization
1	SA0576	67	Toxicity	No	1123381
	SA0575	74	No		1123382
2	SAS019	86	No	Yes	1123561	tsaA	Antitoxin
	SAS018	62	Toxicity		1123560	tsaT	Toxin
3	SA1671	199	No	Yes	1124523	tsbA	Antitoxin
	SAS053	50	Toxicity		1124522	tsbT	Toxin
4	SA1799	107	Toxicity	No	1124690
	SAS063	53	Toxicity		1124689
5	SA1831	69	No	Yes	1124723	tscA	Antitoxin
	SA1830	108	Toxicity		1124722	tscT	Toxin
6	SA1890	97	No	Yes	1124791	tsdA	Antitoxin
	SA1889	69	Toxicity		1124790	tsdT	Toxin
7	SA1986	178	Toxicity	No	1124902
	SA1985	79	No		1124901
8	SA2153	147	Toxicity	No	1125081
	SA2154	151	Toxicity		1125082

Identification of novel TA systems in S. aureus.

To qualify the eight two-gene operons as TA systems, we next experimentally validated their function, using a host killing and rescue assay in E. coli strain BL21(DE3). Each candidate gene was cloned in an arabinose-inducible pBAD24 vector and evaluated for toxicity to E. coli. Only one of the two genes in 1, 2, 3, 5, 6, and 7 showed toxicity to E. coli, whereas both genes in 4 and 8 showed toxicity (Fig. 1). Subsequently, a host rescue assay was performed to demonstrate that coinduction with the cognate antitoxin gene could neutralize the toxicity. For 2, 3, 5, and 6, induction of the toxin gene by isopropyl-β-d-thiogalactopyranoside (IPTG) led to cell growth arrest even though His-tagged 6 decreases toxicity, and coinduction of its cognate gene by arabinose neutralized the toxicity, causing the cells to resume their growth on M9 Gly containing both IPTG and arabinose (Fig. 2). However, for 1 and 7, the cognate gene products could not neutralize the toxicity. Therefore, these gene sequences might be one of the other TA systems composed of RNA antitoxins; however, the complementary sequences acting as an antitoxin for base pairing to the toxin mRNA were not detected surrounding the toxin genes. Genes 4 and 8 consisted of two toxin operons, and thus cell growth was not rescued by the coinduction of both toxin genes (Fig. 2). Taken together, these results demonstrated that 2, 3, 5, and 6 worked as TA systems, and they were henceforth termed “toxin-antitoxin system in staphylococci a, b, c, and d” with antitoxin (A) and toxin (T) genes (tsaAT, tsbAT, tscAT, and tsdAT) (Table 1).

FIG 1.

Host killing assay. Expression of various candidate genes under the arabinose-inducible promoter pBAD24 vector used to evaluate toxicity against Escherichia coli BL21(DE3) cells. E. coli cells were streaked onto M9 Gly medium with (+Ara) or without (−Ara) arabinose supplementation. The plates were incubated at 37°C. The absence of bacterial growth in the arabinose-containing plate suggests that the gene encodes a toxin.

FIG 2.

Host rescue assay. Escherichia coli BL21(DE3) cells, carrying both the arabinose-inducible pBAD24 vector with the toxin gene and the IPTG-inducible pET28a vector with its cognate gene sequence, were streaked onto M9 Gly plates containing 0.07 mM IPTG (left) or both 0.2% arabinose and 0.07 mM IPTG (right). The plates were incubated at 37°C. The rescue of E. coli cell growth on the M9 Gly plates containing both arabinose and IPTG suggests that the gene encodes an antitoxin.

Effect of toxins on cell growth and cell viability.

As a first step to understanding the mechanism of toxicity, we investigated the effect of the toxins on cell growth and viability in liquid medium. TsaT robustly induced growth arrest and cell death (Fig. 3A, 4A, and 5A), and the effect of its toxicity was neutralized by coexpression of the TsaA antitoxin (Fig. 3A). TsbT induced growth arrest, and the growth was moderately neutralized by coexpression of the TsbA antitoxin (Fig. 3B). TsbT strongly reduced colony-forming activity. However, the cells expressing TsbT were viable (Fig. 4A and 5A). These results indicate that TsbT induced a bacteriostatic condition in which the cells were viable but unable to proliferate. TscT abruptly lost colony-forming activity and induced cell death via cell lysis, and the coexpression of TscA antitoxin slightly delayed cell lysis (Fig. 3C and 4A). Moreover, microscopic analysis showed that TscT induced cell elongation and protoplast-like cell formation, eventually leading to cell lysis (Fig. 5A and B). TsdT moderately inhibited cell growth and colony-forming ability but did not lead to cell death (Fig. 3D, 4A, and 5A). The toxicity of TsdT was almost completely neutralized by coexpression of TsdA antitoxin (Fig. 3D). Taken together with these data, all of the four toxins (TsaT, TsbT, TscT, and TsdT) inhibited E. coli growth in liquid medium and displayed varying degrees of growth arrest. Moreover, TsaT and TscT led to cell death. These results suggest that the individual toxins act through different mechanisms.

FIG 3.

Effect of individual TA systems on growth in liquid medium. (A) Growth curve of Escherichia coli BL21(DE3) derivatives carrying pBAD24 (control), pBAD-tsaA, pBAD-tsaT, or pBAD-tsaAT. (B) Growth curve of E. coli BL21(DE3) derivatives carrying pBAD24 (control), tsbA, tsbT, or tsbAT. (C) Growth curve of E. coli BL21(DE3) derivatives carrying pBAD24 (control), tscA, tscT, or tscAT. (D) Growth curve of E. coli BL21(DE3) derivatives carrying pBAD24 (control), tsdA, tsdT, or tsdAT. E. coli cells were grown at 37°C in M9 Gly medium supplemented with ampicillin. When the A₆₆₀ of the culture reached 0.3, arabinose was added to a final concentration of 0.2%. The arrows indicate the time point of arabinose induction. The growth curves represent the average value ± standard deviation of triplicate measurements.

FIG 4.

Effect of individual toxin induction on cell viability. Escherichia coli BL21(DE3) derivatives carrying (A) pBAD-tsaT, pBAD-tsbT, pBAD-tscT, pBAD-tsdT or (B) pBAD24 (control) were grown at 37°C in M9 Gly medium supplemented with ampicillin. When the A₆₆₀ of the culture reached 0.3, arabinose was added to a final concentration of 0.2%. To determine the CFU, the cultures were drawn at 0, 1, 2, and 3 h after arabinose induction and spread on M9 Glc plates supplemented with ampicillin. The data represent mean values ± standard deviations.

FIG 5.

Microscopic analysis. (A) E. coli BL21(DE3) derivatives carrying pBAD-tsaT, pBAD-tsbT, pBAD-tscT, pBAD-tsdT, or pBAD24 (control) were stained with the Live/Dead viability kit at 1 h after arabinose induction. SYTO9-labeled cells (green) represent viable cells, while PI-labeled cells (red) represent dead cells. (B) TscT-induced morphological changes. Phase contrast images of E. coli BL21(DE3) derivatives carrying pBAD24 (control) or pBAD-tscT were obtained at 0 h (without arabinose), 1 h, and 2 h after arabinose induction. The arrows indicate protoplast-like cells.

Effect of toxins on S. aureus.

We identified four novel S. aureus TA systems using a combination of manual base-by-base screening and functional assignments using E. coli. Therefore, we evaluated whether these identified toxins actually function in the native host, S. aureus. The induction of each toxin (TsaT, TsbT, TscT, or TsdT) led to the inhibition of S. aureus growth on Trypticase soy broth (TSB) agar plates containing anhydrotetracycline to induce the individual toxin proteins (Fig. 6A). These results indicated that TsaT, TsbT, TscT, and TsdT functioned as toxins in the native host, S. aureus. Moreover, TsbA neutralized the toxicity of TsbT when the gene operon tsbA-tsbT was expressed in S. aureus (Fig. 6B). These results demonstrated that TsbA and TsbT actually functioned as a TA system in S. aureus, in concordance with the E. coli host killing and rescue assay. Similarly, TsaA and TsdA could neutralized the toxicity of TsaT and TsdT, respectively, whereas TscA could not neutralized TscT toxicity by expressing the gene operon tscA-tscT using the agar plate assay (data not shown). This result is in reasonable agreement with the data obtained in the E. coli experiments (Fig. 3C), in which the coexpression of TscA antitoxin slightly delayed cell lysis.

FIG 6.

Host killing rescue assay in S. aureus. S. aureus RN4220 derivatives carrying (A) pRAB11 (control), pRAB-tsaT, pRAB-tsbT, pRAB-tscT, or pRAB-tsdT or (B) pRAB11 (control), pRAB-tsbT, or pRAB-tsbAT were grown at 37°C for 8 h in TSB medium supplemented with chloramphenicol. The cell cultures were diluted and spotted on TSB Cp agar with or without 0.1 μM anhydrotetracycline to induce TsaT, TsbT, TscT, TsdT, or TsbA and TsbT. The plates were incubated at 37°C.

Conservation of new TA system genes across Staphylococcus species.

BLAST search analysis revealed that the newly identified TA systems were widely conserved across Staphylococcus species (see Fig. S2 to S5 in the supplemental material) and did not exhibit sequence similarity to already known bacterial TA systems. Moreover, the gene cluster and genome map tools in the KEGG database showed that these novel TA genes together with their surrounding genes were highly conserved in the chromosomes of only Staphylococcus species (Fig. 7 and Table 2). The tsaAT gene was conserved in 11 Staphylococcus species, including S. aureus, S. argenteus, S. capitis, S. haemolyticus, S. epidermidis, S. pasteuri, and S. warneri, although the tsaA antitoxin genes were absent in S. lugdunensis, S. saprophyticus, S. xylosus, and S. pettenkoferi (Fig. 7A). The tsbAT and tsdAT genes were highly conserved in the chromosomes of S. aureus isolates and almost all Staphylococcus species registered in the GenBank database, with the exception of the tsdAT gene in S. sciuri (Fig. 7B and D; Table 2). The tscAT gene was located on a phage-like infectious particle, SaPIn1, in S. aureus strain N315 and SaPI1 in S. aureus strain COL and was found in S. aureus, S. agnetis, S. haemolyticus, S. lugdunensis, S. pseudintermedius, and S. warneri, even though the tscA antitoxin gene was not found in S. agnetis (Fig. 7C and Table 2). These results indicate that the tsbAT and tsdAT genes were highly conserved as core genes in staphylococci, whereas the tsaAT and tscAT genes were transferred within and between Staphylococcus species.

FIG 7.

Schematic representation of genetic structures surrounding TA genes identified in Staphylococcus species. (A) The tsaAT gene. Arrows indicate the direction of open reading frames. The blue arrows indicate the tsaA antitoxin gene, and the red arrows represent the tsaT toxin gene. The genes and their gene products are as follows: cspC, cold shock protein C; lysE, l-lysine exporter protein; HP, conserved hypothetical protein. (B) The tsbAT gene. Arrows indicate the direction of open reading frames. The blue arrows indicate the tsbA antitoxin gene, and the red arrows represent the tsbT toxin gene. The genes and their gene products are as follows: airR, two-component response regulator protein; airS, two-component sensor protein; rluD, ribosomal large subunit pseudouridine synthase D; fumC, fumarate hydratase; cspR, RNA methyltransferase; queG, tRNA epoxyqueuosine reductase; tnp, transposase; glnQ, glutamate ATP-binding ABC transporter; nagA, N-acetylglucosamine-6-phosphate deacetylase; HP, conserved hypothetical protein. (C) The tscAT gene. Arrows indicate the direction of open reading frames. The blue arrows indicate the tscA antitoxin gene, and the red arrows represent the tscT toxin gene. The genes and their gene products are as follows: rep, replication initiator protein; pri, primase; xis, excisionase; str, transcriptional activator; stl, repressor protein; int, site-specific integrase; groEL, chaperonin GroEL; groES, cochaperonin GroES; rpsR, small subunit ribosomal protein S18; ssb, single-strand DNA-binding protein gene; rpsF, small subunit ribosomal protein S6; SAtmRNA, tRNA; smpB, SsrA binding protein; rnr, RNase R; metQ1, methionine ABC transporter substrate-binding protein; metP1, methionine ABC transporter permease; seq, enterotoxin Q; sek, enterotoxin K; TR, transcriptional regulator. (D) The tsdAT gene. Arrows indicate the direction of open reading frames. The blue arrows indicate the tsdA antitoxin gene, and the red arrows represent the tsdT toxin gene. The genes and their gene products are as follows: murF, UDP-N-acetylmuramoyl-tripeptide–d-alanyl-d-alanine ligase; ddlA, d-alanine-d-alanine ligase; rodA, rod shape-determining protein; cls2, cardiolipin synthase 2; yidC, membrane protein insertase; thiE, thiamine-phosphate pyrophosphorylase; tnp, transposase; copA, Cu⁺-exporting ATPase; HP, conserved hypothetical protein.

TABLE 2.

Orthologous genes in staphylococci

Staphylococcus species	Genea				GenBank accession no.
	tsaAT	tsbAT	tscAT	tsdAT
S. aureus	+	+	+	+	GCA_000009645.1
S. agnetis		+	−	+	GCA_001442815.1
S. argenteus	+	+		+	GCA_000236925.1
S. capitis	+	+		+	GCA_001028645.1
S. carnosus		±		+	GCA_000009405.1
S. cohnii		+		+	GCA_002984565.1
S. condimenti		±		+	GCA_001922405.1
S. epidermidis	+	+		+	GCA_000011925.1
S. equorum		+		+	GCA_001432245.1
S. felis		+		+	GCA_003012915.1
S. haemolyticus	+	+	+	+	GCA_000009865.1
S. hyicus		+		+	GCA_000816085.1
S. kloosii		+		+	GCA_003019255.1
S. lugdunensis	−	+	+	+	GCA_000025085.1
S. lutrae		+		+	GCA_002101335.1
S. pasteuri	+	+		+	GCA_000494875.1
S. pettenkoferi	−	+		+	GCA_002208805.2
S. pseudintermedius		+	+	+	GCA_000185885.1
S. saprophyticus	−	+		+	GCA_000010125.1
S. sciuri		+			GCA_002209165.2
S. schleiferi		+		+	GCA_001188855.1
S. simulans		±		+	GCA_001559115.2
S. warneri	+	+	+	+	GCA_000332735.1
S. xylosus	−	+		+	GCA_000953575.1

^a+, antitoxin and toxin genes; −, toxin gene; ±, antitoxin gene.

DISCUSSION

Recent advances in sequencing technology and bioinformatics provide easy access to bacterial genome information and greatly enhance the understanding of complex biological phenomena on a genome-wide scale. However, a number of genes are still considered to encode hypothetical proteins in genome information, and the functional assignment of hypothetical proteins remains a challenging task. In this study, we newly identified four TA systems from a large number of small hypothetical genes in the S. aureus chromosome. Additionally, the newly identified TA systems (TsaAT, TsbAT, and TsdAT) were highly conserved in not only S. aureus strains but also other Staphylococcus species. Sequence similarity searching against a number of databases is a powerful approach to identify unannotated genes in the genome sequence. In most cases, bacterial TA systems are identified by the Basic Local Alignment Search Tool (BLAST) and bioinformatics approaches that combine sequence homology and characteristics of TA systems specifically developed for the identification of type II TA module genes in prokaryotes (16–18). These previous studies did not allow for the identification of new putative TA systems in S. aureus. More recently, using a combination of RASTA-Bacteria, a toxin-antitoxin database, and PSI-BLAST, a total of 67 putative TA systems (type I to type V) were predicted in S. aureus strain MW2 (19). However, the four novel TA systems identified in the present study are not included in the 67 putative TA systems. Sequence similarity searching lacks the ability to detect novel TA systems that do not show sequence similarity with known TA systems in E. coli and Mycobacterium. Moreover, TA modules are often overlooked in genome annotation because the peptides are too small. Thus, we searched for novel TA systems manually by looking at operons, which encode two small hypothetical proteins (most of which are overlapping), thus obtaining eight potential candidates of TA systems. We then used a host killing and rescue assay in E. coli to verify these potential TA systems and identified four TA systems that functioned in the native host, S. aureus, although further research is still needed to confirm the neutralization effect of TscA in S. aureus. Until now, extensive study of bacterial TA systems has demonstrated that E. coli is a useful host cell for identifying TA systems in a wide range of bacterial species, not only for Gram-negative bacilli but also for Gram-positive bacteria such as Bacillus thuringiensis, Streptococcus suis, and Mycobacterium tuberculosis (20–22). Moreover, unlike E. coli, overexpression of MazF_sa in S. aureus carrying the endogenous mazEF_sa gene induced cell stasis rather than cell death, despite the fact that MazF_sa has the biological function of a toxin degrading mRNA in S. aureus (23). Our results showed that either one or both operons of all of the candidate genes showed toxicity to E. coli. Given that many exogenous genes (e.g., those encoding membrane proteins, restriction enzymes, proteases, and transcriptional regulators) are known to exhibit toxicity to E. coli (24), we further evaluated whether the cognate gene could neutralize this toxicity. Interestingly, the gene order of the newly identified TA systems was similar to that of many TA systems, in which the upstream gene is the antitoxin and the downstream gene is the toxin. This could be considered one of the characteristics of TA systems, although there are TA systems where the gene locus order is reversed, such as SpoIISA/SpoIISB. In this study, we could not determine whether these new TA systems belong to any specific types of TA systems, which are classified based on the nature of the antitoxins. Therefore, further research is needed to identify the neutralization mechanism of the cognate antitoxin.

Now, the most interesting question is what are the cellular targets for the individual toxins and their physiological roles in S. aureus? As a first step to understanding the mechanism of toxicity, we investigated the effect of the individual toxins on cell growth and viability using E. coli. Interestingly, E. coli cells expressing the individual toxins exhibited varying degrees of toxicity. These observations suggest that the individual toxins inhibit different cellular targets. Moreover, an intriguing observation was that the cells expressing TscT toxin formed a protoplast-like cell and lysed. The morphological changes were different from previously reported YeeV and CptA toxins targeting the cytoskeleton proteins FtsZ and MreB (25, 26). A signal peptide for secretion was not found in the amino acid sequence of TscT; therefore, TscT may intracellularly inhibit cell wall synthesis or cell division through an unknown mechanism. Additionally, Pfam analysis (27) suggested that TsdA and TsdT contain a copper-sensing domain (DUF156) and a heavy-metal-associated (HMA) domain, respectively. Interestingly, however, TsdT lacks the two active site cysteine residues that are important in the binding and transfer of metal ions. These findings may provide insight into the molecular mechanism underlying the toxicity of TsdT.

As for the physiological role of S. aureus TA systems, mazEF_sa gene deletion led to slightly increased sensitivity to β-lactam antibiotics and increased biofilm formation in S. aureus (28, 29), whereas deleting three type II TAs in S. aureus, namely, mazEF, yefM_sa1 yoeB_sa1, and yefM_sa2 yoeB_sa2, did not affect the level of persisters, suggesting that unannotated TAs play a role in persister formation (30). Some of the newly identified TA systems in this study may play important roles in the physiology of S. aureus and may be involved in its pathogenicity. Meanwhile, tsbAT and tsdAT were highly conserved in staphylococci, and the surrounding genes were composed of essential genes. Therefore, these TA systems may play an important role in numerous fundamental physiological processes in staphylococci. We are currently performing studies to characterize the mechanisms of the toxins and the physiological roles of these TA systems in S. aureus.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture medium.

E. coli strains and plasmids used in this study are listed in Table 3. E. coli DH5α (TaKaRa Bio, Shiga, Japan) was used for plasmid construction. E. coli BL21(DE3) (Merck, Darmstadt, Germany) was used as a host for the host killing and rescue assay. E. coli cells were cultured overnight at 37°C in M9 medium containing 0.4% Casamino Acids (BD Biosciences, Franklin Lakes, NJ) and 0.4% glucose (M9 Glc) or 0.3% glycerol (M9 Gly). Ampicillin (100 μg/ml) and kanamycin (50 μg/ml) were added to the medium as necessary.

TABLE 3.

Bacterial strains and plasmids

Strain or plasmid	Genotype or descriptiona	Source or reference
E. coli strains
DH5α	F− φ80dlacZΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17(rK− mK+) phoA supE44 λ− thi-1 gyrA96 relA1	TaKaRa
BL21(DE3)	F− ompT gal dcm lon hsdSB(rB− mB−) λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5])	Merck
S. aureus strain
RN4220	Laboratory strain, NCTC 8325-4, r− m+	35
Plasmids
pBAD24	Arabinose-inducible expression vector, Ampr in E. coli	32
pET28a	IPTG-inducible expression vector, Kmr in E. coli	Merck
pRAB11	Anhydrotetracycline-inducible expression vector, Cpr in S. aureus	33
pFK90	pBAD24 with SA1831 gene	This work
pFK91	pBAD24 with SA1830 gene	This work
pFK92	pBAD24 with SA0576 gene	This work
pFK93	pBAD24 with SA0575 gene	This work
pFK94	pBAD24 with SA1671 gene	This work
pFK95	pBAD24 with SAS053 gene	This work
pFK96	pBAD24 with SA1799 gene	This work
pFK97	pBAD24 with SAS063 gene	This work
pFK98	pBAD24 with SA1890 gene	This work
pFK99	pBAD24 with SA1889 gene	This work
pFK100	pBAD24 with SA1986 gene	This work
pFK101	pBAD24 with SA1985 gene	This work
pFK102	pBAD24 with SA2153 gene	This work
pFK103	pBAD24 with SA2154 gene	This work
pFK104	pBAD24 with SAS019 gene	This work
pFK105	pBAD24 with SAS018 gene	This work
pFK106	pBAD24 with SA1831 and SA1830 genes	This work
pFK108	pBAD24 with SA1671 and SAS053 genes	This work
pFK110	pBAD24 with SA1890 and SA1889 genes	This work
pFK113	pBAD24 with SAS019 and SAS018 genes	This work
pFK114	pET28a with SA1830 gene	This work
pFK116	pET28a with SA0576 gene	This work
pFK118	pET28a with SAS053 gene	This work
pFK120	pET28a with SAS063 gene	This work
pFK122	pET28a with SA1889 gene	This work
pFK124	pET28a with SA1986 gene	This work
pFK126	pET28a with SA2153 gene	This work
pFK128	pET28a with SAS018 gene	This work
pFK201	pRAB11 with SAS018 gene	This work
pFK202	pRAB11 with SAS053 gene	This work
pFK203	pRAB11 with SA1830 gene	This work
pFK204	pRAB11 with SA1889 gene	This work
pFK206	pRAB11 with SA1671 and SAS053 genes	This work

^aLocus numbers are based on S. aureus N315 (https://www.ncbi.nlm.nih.gov/nuccore/BA000018.3).

Bioinformatics search for TA systems on the S. aureus chromosome.

To identify novel TA systems in S. aureus, the S. aureus genome sequence was analyzed using the Rapid Automated Scan for Toxins and Antitoxins in Bacteria (RASTA-Bacteria) software (16) and TA finder (17).

Manual search for potential candidate genes of TA systems on the S. aureus chromosome.

Potential TA system gene studies were performed base-by-base from nucleotide 1 to nucleotide 2,814,816 of the entire genome sequence of S. aureus strain N315 (31). The sequences qualified as putative TA system genes based on the following criteria: (i) the two genes, each consisting of less than 200 amino acid (aa) residues, either overlap or are less than 30 bases apart; (ii) the genes consist of an operon encoding two proteins; and (iii) both genes encode a hypothetical protein. According to the first criterion, 20 two-gene pairs were selected, and the second and third criteria further narrowed down this pool to the eight potential candidates (see Table S3 in the supplemental material). Predicted two-gene operons were identified using MicrobesOnline Operon Predictions for Staphylococcus aureus strain N315 (http://www.microbesonline.org/operons/gnc158879.html) (32).

Cloning of putative toxin and antitoxin genes.

Each gene from potential candidate genes of TA systems was amplified by PCR using S. aureus N315 genomic DNA as the template and KOD Plus Neo DNA polymerase (Toyobo, Osaka, Japan) with the primer sets described in Table 4. The amplified DNA fragments were digested using appropriate restriction enzymes and cloned into the same site of an arabinose-inducible pBAD24 vector (33), an IPTG-inducible pET28a vector (Merck, Darmstadt, Germany), and a pRAB11 vector (34).

TABLE 4.

Oligonucleotide primers

Primer	Sequence (5′ to 3′)a
SA1831FN	TAGCATATGAATAATGAACAAAAAGAAG
SA1831RE	CTTGAATTCCCCGATTCATTTTTATTC
SA1830FNh	GAGAGCTAGCATGAATCGGGAAATTAAAG
SA1830RH	AAGAAAGCTTAATTTCTATATTACTCTGC
SA1830FK	AATAGGATCCAATTAGAGAAAGGAGAATAA
SA1830RX	AATTTCTAGATTACTCTGCTACTTTTATTG
SA0576FN	TAACATATGAGTAAAATAAACCACATC
SA0576RE	CATGAATTCGTCCTCCTGATTAAATTG
SA0575FN	ATACATATGACTAAAACACTTGAATTAAG
SA0575RE	ACTGAATTCATTGCGCTTTCTTTATATC
SA1671FN	TAGAACATATGGCAATGAACTTTAAAGTC
SA1671RE	CACAGAATTCTAATTAAGCAAAGCGTGAC
SAS053FN	GTATTGCATATGTCTAAAGACAAAGATCC
SAS053RE	TACAGAATTCTAATACTTTACTGACTGTC
SA1671FBg	TATGAGATCTGCAAACTAAGAGGAGTGTAG
SAS053FBg	TTGCAGATCTAAAAATCTGTGAGGTATTGA
SA1799FN	GATACATATGCCACATATTTTAAACGTAAC
SA1799RE	TTTAGAATTCTTACTCATTTGTTGTGCCC
SAS063FN	GCACAACATATGAGTAAAACTTATAAAAGC
SAS063RE	TCTTGAATTCTTGTTACTGTTACTTGTTGG
SA1890FS	AATGAGCTCACTGAACAAGATAATGCAC
SA1890RE	ATCGAATTCTGATGTATCATCTTAAAC
SA1889FN	GAGTTTCATATGATACATCAAAATACG
SA1889RE	TTCAGAATTCAAATCTATAGCGCACTC
SA1889FBg	AAATAGATCTAAGACTAAAGGAGTTTAAGA
SA1986FN	GTTACATATGAAACGACTTAAAAACTTTATC
SA1986RE	ATGAGAATTCTCCCTTACAACACTCGTGG
SA1985FN	GTATCATATGGCTAACAATCATAACCAAAAC
SA1985RE	AATTGAATTCTAAATTGATGAATTAACTCC
SA2153FN	TTAACATATGATGAAACTCAATTTATTTATC
SA2153RE	TTTCGAATTCTTCATATTGATAAGCGCTC
SA2154FN	CGCTTATCCATATGAAAAATTTGAAAAACTC
SA2154RE	TGGTGAATTCAAATCATTTTAATTGTTTCAG
SAS019FN	CAGTCACATATGAATAACATTTTGTTAAATGC
SAS019RE	AAACAGAATTCCAAAATGTAAAGACATGTG
SAS018FN	GTTTCATATGTCTTTACATTTTGCAATTC
SAS018RE	TATAAAATGAATTCCATTTATAGCACACC
SAS018FBg	TTAAAGATCTCGGTATATAAGGAGGGTTTC

^aSequence letters in boldface represent restriction sites.

Host killing assay for evaluating the gene coding toxin in E. coli.

A host killing assay was performed as previously described (19, 35). To evaluate the toxicity of candidate genes, the pBAD24 libraries were transformed into E. coli BL21(DE3). The transformed cells were precultured in M9 Gly medium supplemented with ampicillin at 37°C overnight, and then E. coli was streaked on M9 Gly agar plates either with or without 0.2% arabinose. The plates were incubated at 37°C overnight. The presence or absence of E. coli colonies on the arabinose-containing plate was used as the readout to evaluate the toxicity of the gene product.

Host rescue assay for evaluating the gene coding antitoxin in E. coli.

A host rescue assay was performed as previously described (35). To evaluate the antitoxin activity by coexpression of the toxin gene and its cognate antitoxin gene, pET28a carrying the paired gene was transformed into E. coli BL21(DE3) carrying the toxin-cloned pBAD24 derivative. The transformed cells were cultured in M9 Gly medium supplemented with ampicillin and kanamycin at 37°C overnight, and then the E. coli cells were streaked on four kinds of M9 Gly agar plates containing ampicillin and kanamycin as well as arabinose, IPTG, arabinose and IPTG, or no inducers. The plates were incubated at 37°C overnight. The antitoxin activity was evaluated based on whether the paired gene could neutralize the cell growth inhibition by the toxin.

Differential staining of live and dead cells.

Differentiation between live and dead E. coli cells was performed using the Live/Dead BacLight bacterial viability kit (ThermoFisher Scientific Inc., Vantaa, Finland) according to the manufacturer’s protocol. E. coli BL21(DE3) strains carrying pBAD24 derivatives were grown with shaking at 37°C in 10 ml M9 Gly medium supplemented with ampicillin (100 μg/ml). When the optical density at 660 nm (OD₆₆₀) of the culture reached 0.3, arabinose was added, and then the cultures were harvested after 1 h of cultivation. The E. coli cells were washed with 0.85% NaCl and stained with SYTO9 dye and propidium iodide (PI). SYTO9 stains all bacterial cells green, while PI is a red fluorescent dye and selectively stains dead cells in a population. Fluorescent images and phase-contrast images were obtained using a Nikon Eclipse E1000 microscope (Nikon, Tokyo, Japan).

Host killing and rescue assay in S. aureus.

The pRAB11 derivatives were transformed into S. aureus RN4220 by electroporation as previously described (29, 36) and selected on Trypticase soy broth (TSB) (Becton, Dickinson Microbiology Systems, Cockeysville, MD) agar plates containing 5 μg/ml chloramphenicol (Cp). S. aureus strains were cultured at 37°C for 8 h in TSB-Cp medium. S. aureus cells were then serially diluted (10-fold) in TSB medium, and 10 μl of each cell dilution was spotted on a TSB-Cp agar plate with or without 0.1 μM anhydrotetracycline (Sigma-Aldrich, St. Louis, MO, USA) for gene expression in S. aureus. The plates were incubated at 37°C overnight. All experiments were performed at least in triplicate.

Identification of homologous genes.

The homologous gene sets with their surrounding genes among different organisms were identified using the gene cluster and genome map in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (37). Multiple sequence alignments were performed using ClustalW2.1 (38).