Toxin-Antitoxin (TA) Systems are Prevalent and Transcribed in Clinical Isolates of Pseudomonas aeruginosa and Methicillin-Resistant Staphylococcus aureus

Abstract

The percentage of bacterial infections refractory to standard antibiotic treatments is steadily increasing. Among the most problematic hospital and community-acquired pathogens are methicillin-resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa (PA). One novel strategy proposed for treating infections of multidrug resistant bacteria is the activation of latent toxins of toxin-antitoxin (TA) protein complexes residing within bacteria; however, the prevalence and identity of TA systems in clinical isolates of MRSA and PA has not been defined. We isolated DNA from 78 MRSA and 42 PA clinical isolates and used PCR to probe for the presence of various TA loci. Our results showed that the genes for homologs of the mazEF TA system in MRSA, and the relBE and higBA TA systems in P. aeruginosa were present in 100% of the respective strains. Additionally, RT-PCR analysis revealed that these transcripts are produced in the clinical isolates. These results indicate that TA genes are prevalent and transcribed within MRSA and PA and suggest that activation of the toxin proteins could be an effective antibacterial strategy for these pathogens.

Introduction

Initially discovered on plasmids, toxin-antitoxin (TA) systems were termed “plasmid-addiction” modules to describe their role in plasmid maintenance through a post-segregational killing mechanism (Gerdes, et al., 1986, Hayes, 2003). TA systems ensure plasmid maintenance in the bacterial host population through the differential stability of the stable toxin and labile antitoxin, both encoded by the plasmid. When present, the plasmid enables the continued expression of antitoxin, which binds to and inactivates the toxin. However, if the plasmid is lost during cell division, the antitoxin protein is rapidly degraded and not replenished, thus releasing the stable toxin to kill the bacterial cell.

TA genes are also found on bacterial chromosomes, although their precise role in this setting is debated (Keren, et al., 2004, Buts, et al., 2005, Gerdes, et al., 2005, Engelberg-Kulka, et al., 2006, Szekeres, et al., 2007, Nariya & Inouye, 2008). Two of the most well studied TA systems are MazEF and RelBE encoded by the E. coli chromosome. The MazEF system in E. coli may function as an irreversible mediator of cell death under stressful conditions (Amitai, et al., 2004) or as a modulator of translation to induce a reversible state of bacteriostasis (Pedersen, et al., 2002, Christensen, et al., 2003). RelBE modulates the stringent response induced by amino acid starvation (Christensen, et al., 2001), causing global translation inhibition and leading to bacteriostasis (Pedersen, et al., 2002, Pedersen, et al., 2003). Similar to plasmid-encoded systems, chromosomal TA modules derive their intrinsic killing/growth inhibition ability from the shift in the balance towards free toxin (Christensen, et al., 2004).

Exploitation of the inherent toxicity of TA systems has been proposed as a novel antibacterial target, as activation of the latent toxin via direct TA complex disruption or some alternative mechanism would result in bacterial cell death (Engelberg-Kulka, et al., 2004, DeNap & Hergenrother, 2005, Alonso, et al., 2007, Williams & Hergenrother, 2008). However, a prerequisite for the success of this strategy is the identification of clinically important bacteria that would be susceptible to a compound that activates TA systems. Surveys of clinical isolates to determine the prevalence and identity of TA systems could support and guide the development of this strategy by establishing which TA loci are most frequently encountered and would thus serve as the best target candidates.

One such survey discovered that TA systems were frequently encoded on plasmids carried by vancomycin-resistant enterorocci (VRE) (Moritz & Hergenrother, 2007). The observation that TA systems are ubiquitous and functional on plasmids in VRE (Moritz & Hergenrother, 2007, Sletvold, et al., 2007, Halvorsen, et al., 2011) raises the possibility that other pathogenic bacteria may also harbor the genes for TA systems.

A bioinformatics survey of 126 prokaryotic genomes identified TA loci belonging to the 7 known TA gene families in the completed genomes of three S. aureus strains (Pandey & Gerdes, 2005). The genomes of all three S. aureus strains studied contained two loci belonging to the relBE gene family and one locus belonging to the mazEF gene family, which was later demonstrated to be a functional TA module in S. aureus (Fu, et al., 2007). The toxin, MazF_Sa, is a sequence-specific endoribonuclease that inhibits cell growth when expressed in both E. coli and S. aureus (Fu, et al., 2009, Zhu, et al., 2009). The MazEF_Sa system is cotranscribed with the alternative transcription factor σ^B under certain stress conditions (Donegan & Cheung, 2009).

Additionally, the bioinformatics survey identified three TA loci on P. aeruginosa strain PAO1, relBE, parDE, and higBA (Pandey & Gerdes, 2005). Although no additional work has been published on these TA systems, functional homologs have been described in other pathogenic bacteria, including RelBE in Streptococcus pneumoniae (Nieto, et al., 2006), Yersinia pestis (Goulard, et al., 2010) and Mycobacterium tuberculosis (Yang, et al., 2010); ParDE in Vibrio cholerae (Yuan, et al., 2011); and HigBA in V. cholerae (Christensen-Dalsgaard & Gerdes, 2006, Budde, et al., 2007), Proteus vulgaris (Hurley & Woychik, 2009) and Y. pestis (Goulard, et al., 2010).

While the analysis of sequenced genomes has been informative, there is no data on the prevalence and identity of TA loci in a large cadre of methicillin-resistant S. aureus (MRSA) and P. aeruginosa (PA) clinical isolates. In the current study, we find that mazEF, relBE, higBA, and parDE are widespread in collections of MRSA and PA clinical isolates.

Materials and Methods

Bacterial strains

Clinical isolates of MRSA were obtained from 3 medical centers and the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA) for a total of 78 strains. The medical centers were Carle Foundation Hospital (Urbana, IL), Memorial Medical Center (Springfield, IL), and Delnor Community Hospital (Geneva, IL). The clinical isolates of PA designated CI01–CI20 were obtained from the sputum of 20 different CF patients at Carle Foundation Hospital, as described previously (Musk, et al., 2005). The remaining 22 PA clinical isolates were a kind gift from Cubist Pharmaceuticals, Inc. (Lexington, MA) and had been obtained from over eight geographically diverse clinical sites in the US.

MLVA

To assess the clonality of the clinical MRSA and PA isolates, basic molecular typing was performed by PCR-based multiple-locus variable number of tandem repeats analysis (MLVA) previously described (Sabat, et al., 2003, Vu-Thien, et al., 2007). For MRSA, a minor modification was made to the reported protocol, in that a greater amount of Taq polyermase was added to the PCR mix (5 units) and 6 μL of PCR products were analyzed in 1.8% Low-Range Ultra agarose (Biorad) for 3 hrs at 6.5 V/cm. For PA, 10 of the 15 minisatellites described by Vu-Thien and co-workers were analyzed (ms 142, ms 211, ms 212, ms 213, ms 214, ms 215, ms 216, ms 217, ms 222, ms 223) and 1 μL of PCR products was electrophoresed in 2.0% Molecular Biology Grade agarose (Fisher) at 10 V/cm. For MRSA or PA respectively, TIFF or JPEG files of the MLVA gel images were visually evaluated with BioNumerics software (Applied Maths) and a dendrogram of banding patterns was constructed using the Dice or Pearson coefficients, respectively, and the unweighted-pair group method using average linkages (UPGMA).

PCR Analysis

For all MRSA and PA strains, PCR amplification was performed from purified total DNA. Gene-specific internal primers were used to amplify the mazEF_Sa, relBE_Pa, parDE_Pa, and higBA_Pa TA genes and separate intergenic primers were used to amplify the upstream and dowstream flanking regions. The oligonucelotide sequences of the primers are listed in Table 1, and Figure 2 depicts the homologous region of the primers for the PCR-based screen and the flanking region primers. PCR amplification was carried out in a DNA thermal cycler (PTC-200, MJ Research, Inc.) under reaction conditions as described previously (Moritz & Hergenrother, 2007) with a lowering of the annealing temperature to 49°C for most primers. PCR amplified products were analyzed by agarose gel electrophoresis in 1% agarose and stained with ethidium bromide.

Table 1.

Gene-specific primers used for PCR and RT-PCR analysis

TA system	Primer sequence (5′ to 3′)	Length (nt)	Application
mazEFSa	(+)ATCATCGGATAAGTACGTCAGTTT (−)AGAAGGATATTCACAAATGGCTGA	408	PCR, RT-PCR
parDEPa	(+)GCGGCTGACCTGGATTTATC (−)CCAAGCAGTAGCGGATCAATTG	556	PCR, RT-PCR
relBEPa	(+)CAGGGGGTAATTTCGACTCTG (−)ATGAGCACCGTAGTCTCGTTC	505	PCR
higBAPa	(+)CTCATGTTCGATCTGCTTGC (−)CAATGCTTCATGCGGCTAC	469	PCR
relBEPa	(+)CGCAGTACCTGGAAAGGCAGC (−)GCCTTTAACCCGAAACGGG	349	RT-PCR
higBAPa	(+)GGCCAACATAGCATCAGGATC (−)GGACGTATCAAAGTAACGCCC	305	RT-PCR
mazEFSa Up	(+)GTCTTGAACACATCTTCACGCG (−)GCGAAAATACCGACACATGTAGAG	753	PCR, flanking
mazEFSa Down	(+)GCTTCGTTCGCTAGGGAGAG (−)CTACAAGCGGGTGAGTCTGTAA	556	PCR, flanking
parDPa Up	(+)CGGTGATCTTTGCCAACACTAAG (−)CTTCCGCTCAGCATATGACTC	679	PCR, flanking
parEPa Down	(+)TGAGTCTTCTGGGGGTGCTG (−)GGAATTCCACACCATCCGC	676	PCR, flanking
relEPa Up	(+)CCGGAAAAAGCGCGAGAAAGC (−)GGGGGCTGCAATGAGCCTG	548	PCR, flanking
relBPa Down	(+)GTGCTCATTTTCTGATCAACTTCG (−)GTGACGCTCTCCGACAGCTTC	646	PCR, flanking
higAPa Up	(+)GATCCGACCCCTTCCGTCTAAACG (−)GTAGCCGCATGAAGCATTG	823	PCR, flanking
higBPa Down	(+)CAGGTGGAGAGCGCAGGTC (−)CAATTGTCCCAACGCCTCCTTCG	712	PCR, flanking
higAPa Up-PA7	(+)GTTTGCCACGTTTGCATGCAG (−)CGCTCAGTTCTGGATGAATCTCC	803	PCR, flanking
higBPa Down-PA7	(+)GCATCGCCGATTCCAAGTG (−)GCAACGTGTGTTCGTCACC	655	PCR, flanking

(+) sense primer

(−) antisense primer

Figure 2. Locations of primary homology for primers used in PCR screen, RT-PCR and flanking region PCR.

(A) Primer sequences were based off the S. aureus COL genome. The same internal primers were used to amplify a region of mazEF_Sa for both the PCR-based screen and RT-PCR. Flanking mazEF_Sa are the genes rbsU and alr. Primers were designed to amplify the sequences from rbsU to mazF_Sa and from mazE_Sa to alr. (B–D) Primer sequences were based off the P. aeruginosa PAO1 genome. (B) The same internal primers were used to amplify a region of parDE_Pa for both the PCR-based screen and RT-PCR. Flanking parDE_Pa are genes encoding an integrase, a tRNA and a transferase. Primers were designed to amplify the sequence between the integrase gene and parD_Pa and between parE_Pa and the transferase gene. (C) Separate sets of internal primers were used to amplify regions of relBE_Pa for the PCR-based screen and for RT-PCR. Flanking relBE_Pa are genes encoding a transcriptional regulator and a hypothetical protein. Primers were designed to amplify the sequence from the transcriptional regulator gene to relB_Pa and from relE_Pa to the hypothetical protein gene. (D) Separate sets of internal primers were used to amplify regions of higBA_Pa for the PCR-based screen and for RT-PCR. Flanking higBA_Pa are genes encoding a hypothetical protein and a Ton-B dependent receptor. Primers were designed to amplify the sequence from the hypothetical protein gene to higA_Pa and from higB_Pa to the Ton-B dependent receptor gene. (E) In P. aeruginosa PA7, higBA_Pa is flanked by genes encoding a restriction endonuclease and a hypothetical protein. Primers were designed to amplify the sequence from the restriction endonuclease gene to higA_Pa and from higB_Pa to the hypothetical protein gene.

RT-PCR Analysis

RT-PCR was performed by using the SuperScript One-Step RT-PCR with Platinum Taq kit (Invitrogen). The primers used to amplify the mazEF_Sa, and parDE_Pa sequence for RT-PCR are the same as those designed for PCR analysis, whereas RT-PCR for higBA_Pa and relBE_Pa was performed with separate specific intragenic primers designed from the P. aeruginosa PAO1 sequence. The sequences of all primers used in RT-PCR are listed in Table 1 and the homologous regions are depicted in Figure 2. The extracted total RNA (up to 40 ng) was used in RT-PCR as well as PCRs with Platinum Taq DNA polymerase (Invitrogen) to detect DNA contamination. Reverse transcription and PCR amplification were carried out in a DNA thermal cycler (PTC-200, MJ Research, Inc.) under reaction conditions as described previously (Moritz & Hergenrother, 2007), with modifications made to the PCR annealing temperature as follows: 55°C for mazEF_Sa, 58°C for relBE_Pa, and 50.8°C for both higBA_Pa and parDE_Pa. RT-PCR amplification products were analyzed by agarose gel electrophoresis in 1% agarose and stained with ethidium bromide.

Results

MLVA of Clinical Isolates

MRSA isolates (collected from the three medical centers and NARSA) and PA isolates (collected from Carle Foundation Hospital and from Cubist Pharmaceuticals) were analyzed by the DNA-based typing method, MLVA, to assess intra-species relatedness. Although the MLVA for the 17 NARSA strains had been previously characterized, 8 of these isolates were included in the MLVA for comparison (See SI Methods). For PA, two standard laboratory strains (PAO1 and PA14) were included for comparison.

The experimental variation between duplicate experiments was determined for the MLVA profile using 5 MRSA or 5 PA isolates and applied to establish a cutoff value of 91% or 97%, respectively, for typing strains with identical DNA banding patterns. Using the 91% cutoff, 46 MLVA patterns were defined out of the 69 MRSA strains evaluated. Applying a 75% similarity value generated 13 clusters from 56 strains, excluding 13 strains from these clusters (Figure 1A). Isolates belonging to the same cluster differed by up to four bands. MLVA of the clinical S. aureus isolates using either cutoff value revealed that the majority of isolates do not belong to the same clusters as U.S. S. aureus clones (USA100, USA200, USA300, and USA600) and suggest that the isolates collected from the Illinois area are not clonal.

Figure 1. MLVA of the MRSAand PAclinical isolates.

The 91% or 97% clonal cutoff value and 75% similarity cutoff value are indicated by solid and dashed vertical lines, respectively for MRSA (A) and PA (B). The clusters generated are shown in corresponding solid and dashed brackets alongside the dendrogram.

For PA, using the 97% cutoff value, 43 MLVA banding patterns were formed out of the 44 strains. When a cutoff value of 75% was applied, 10 clusters were generated comprising 28 strains, and 26 MLVA banding patterns were discerned (Figure 1B). Strains that group according to these two cutoff values are in a variety of clusters, demonstrating that the isolates studied were not clonal.

Presence of TA genes

Armed with the knowledge that the collection of MRSA and PA clinical isolates were sufficiently diverse, an effort was made to define the prevalence of TA genes in the strains. For the MRSA isolates, gene-specific PCR primers were used to amplify the genes for the mazEF homolog (called mazEF_Sa) observed on the S. aureus COL genome (Pandey & Gerdes, 2005). The PA isolates were probed for homologs of the higBA, parDE and relBE systems identified in PA strain PAO1 (Pandey & Gerdes, 2005). The oligonucleotide sequences of all PCR primers used to amplify TA genes are listed in Table 1 and the homologous regions are represented in Figure 2.

Total DNA preparations from each of the 78 MRSA and 42 PA strains were analyzed by PCR, and results were designated positive if a distinct band was observed at the expected size on an agarose gel. The PCR screen revealed that the mazEF_Sa TA system was present in all MRSA isolates (78/78, 100%). For the 42 PA isolates, relBE_Pa (42/42, 100%) and higBA_Pa (42/42, 100%) were ubiquitous, whereas parDE_Pa (13/42, 30%) was less prevalent. Table S1 contains a complete list of all MRSA and PA isolates and the TA genes detected by PCR. Comparison of the MLVA genotypes of PA strains that carry parDE_Pa showed that these strains are dispersed throughout the dendrogram, indicating that there is no correlation between genome relatedness and carriage of parDE_Pa.

DNA sequencing was performed on ~10% of all PCR products. For the MRSA isolates, sequenced PCR products revealed strong sequence identity (95.6 – 99.5%) to the reference TA system sequence (mazEF_Sa alignments are shown in Figure S1). For the PA isolates, sequenced PCR products also revealed strong sequence identity (97.8 – 100%) to the reference TA system sequences (higBA_Pa, 99.4% average identity; parDE_Pa, 99.6% average identity; and relBE_Pa, 98.7% average identity); alignments are shown in Figures S2–4).

TA gene localization

It was next investigated whether the TA genes were located on a plasmid or the chromosome of the MRSA and PA isolates. The sequences directly upstream and downstream of the mazEF_Sa and relBE_Pa TA genes are highly conserved among the completed S. aureus and P. aeruginosa genomes in the National Center for Biotechnology Information (NCBI) Genome database, whereas the flanking regions of parDE_Pa and higBA_Pa are conserved in P. aeruginosa PAO1, LESB85 and UCBPP-PA14, but are different in strain PA7. Primers were designed (Table 1 and Figure 2) to amplify the sequences flanking the TA genes based on the conserved sequence in S. aureus strains and in P. aeruginosa strains PAO1 and PA7. In this experiment the presence of a PCR product would suggest chromosomal location of the TA systems.

PCR analysis revealed that in 100% (78/78) of the MRSA isolates, the regions upstream and downstream of the mazEF_Sa genes were amplified with the flanking region primers, suggesting a chromosomal location with sufficient homology to the S. aureus reference strains in the NCBI database. In the PA isolates, both flanking regions of the parDE_Pa genes in all isolates (13/13, 100%) were amplified using primers homologous to the PAO1 reference sequence. The flanking regions of nearly all relBE_Pa genes (41/42, 97%) were amplified, except for strain 1284, for which no flanking region could be amplified. Amplification was observed for the downstream sequence of every higBA_Pa loci (42/42, 100%) as well as for the region upstream of higBA_Pa except for in 10 strains (32/42, 76%). For these 10 strains, PCR was performed with various primers designed based on the PAO1 reference sequence, as well as primers designed to probe the upstream sequence of higBA_Pa observed in P. aeruginosa PA7; however, no product was amplified in any of these cases. All results from the flanking region PCR are listed in Table S2.

DNA sequencing was performed on >10% of the PCR products to confirm the identity of the amplified sequence. Sequenced PCR products revealed a strong sequence identity for the mazF_Sa upstream and downstream regions (91.5–98.6%) compared to the reference sequence from the S. aureus COL genome (Figure S5). The flanking region PCR products of parDE_Pa (92.6 – 98.2%), relBE_Pa (96.2 – 99.4%) and higBA_Pa (91.8 – 99.4%) also showed strong sequence identity to the reference P. aeruginosa PAO1 sequence (Figures S6–8).

Transcription of TA genes

To determine if the TA systems were transcribed by the clinical isolates, RT-PCR was performed with total RNA isolated from >10% of strains shown by PCR to contain the genes for each TA system. The oligonucleotide sequences of all primers used for RT-PCR are listed in Table 1 and Figure 2 depicts the regions of homology. The mazEF_Sa transcript was detected from the total RNA of all nine MRSA strains probed by RT-PCR (Figure 3A). Similarly, the transcripts for relBE_Pa (6/6), higBA_Pa (5/5) and parDE_Pa (3/3) transcripts were detected in all PA strains probed by RT-PCR (Figure 3B). For all samples, no amplification products were observed in the absence of reverse transcriptase, confirming that the products seen by RT-PCR were due to the presence of the TA transcript in the clinical isolates and not DNA contamination (Figures 3A and B).

Figure 3. RT-PCR analysis of MRSA and P. aeruginosa clinical isolates.

RT-PCR with primers complementary to the genes encoding each TA system indicates that transcripts are produced in the clinical isolates of (A) MRSA and (B) P. aeruginosa, as shown in the (+) RT row. Controls for DNA contamination, in which the reverse transcriptase is omitted from the reaction mix, yield no product, as shown in the (−) RT row. Clinical isolates analyzed are indicated by the strain number above each column.

Discussion

Bioinformatics analyses of published prokaryotic genomes have demonstrated the pervasive nature of TA loci (Makarova, et al., 2009); however, little effort has been made to survey large collections of clinical bacterial strains for the presence and functionality of TA systems. Herein we use PCR to determine that mazEF_Sa is ubiquitous in a collection of MRSA clinical isolates, and higBA_Pa and relBE_Pa are ubiquitous in a collection of PA clinical isolates, whereas parDE_Pa is less commonly observed. This PCR method is complementary to the whole genome sequencing that has previously been used to examine the presence of TA systems in MRSA and PA, and the results reveal the value of inspecting large numbers of clinical isolates in the manner. For example, of the three sequenced PA clinical isolates that have been analyzed, PA14 does not have the genes for parDE_Pa, whereas PAO1 and PA7 do (Makarova, et al., 2009). However, the results presented herein show that PA clinical isolates that cluster with PA14 (via MLVA) are just as likely to have the genes for parDE_Pa as those PA strains that do not cluster with PA14.

Assessment of the flanking sequence of the TA systems in MRSA and PA revealed that the chromosomal location was conserved across all strains carrying mazEF_Sa and parDE_Pa, in nearly all strains for relBE_Pa and in the majority of strains for higBA_Pa. The inability to amplify the upstream sequence of higBA_Pa in 10 strains suggests that the upstream sequence has diverged or that the higBA loci of these 10 strains is located elsewhere; however, the conservation of the downstream sequence implies that higBA_Pa is chromosomally encoded.

Defining the identity of TA systems in clinical isolates satisfies the first requirement in validating TA systems as a viable antibacterial target. However, it is imperative to establish which TA systems are functional in clinical isolates. Thus RT-PCR analysis was performed to determine if the TA systems were transcribed. Importantly, it was shown by RT-PCR that mazEF_Sa, higBA_Pa, relBE_Pa and parDE_Pa were transcribed in strains that carried the genes. Collectively, the results presented herein indicate that the TA genes detected in the MRSA and PA strains reside on the chromosome and are active TA modules.

It has been suggested that activation of TA systems could be an attractive antimicrobial strategy, as the freed toxin would kill the host bacterial cell (Engelberg-Kulka, et al., 2004, DeNap & Hergenrother, 2005, Gerdes, et al., 2005, Alonso, et al., 2007, Williams & Hergenrother, 2008). While the presence of TA systems in sequenced prokaryotic genomes has been established, prior to this work the prevalence of TA systems in clinical isolates of MRSA and PA was unknown. In addition, this is the first time that the mazEF_Sa, higBA_Pa, parDE_Pa, and relBE_Pa transcripts have been shown to be produced in these bacteria, and one of the few examples of demonstrated transcription of any TA genes from a clinical isolate. Given the results of a previous study showing that TA systems are ubiquitous in VRE (Moritz & Hergenrother, 2007), and the current survey showing that TA systems are also highly prevalent and transcribed in MRSA and PA, it appears that these problematic bacterial pathogens would indeed be susceptible to TA-based antibacterial strategies. Specifically, activation of MazEF_Sa should be considered for MRSA, and activation of RelE_Pa or HigBA_Pa appear to be attractive strategies against Pseudomonus aeruginosa.