HicA of Escherichia coli Defines a Novel Family of Translation-Independent mRNA Interferases in Bacteria and Archaea

Toxin-antitoxin (TA) loci are common in free-living bacteriaand archaea. TA loci encode a stable toxin that is neutralizedby a metabolically unstable antitoxin. The antitoxin can beeither a protein or an antisense RNA. So far, six differentTA gene families, in which the antitoxins are proteins, havebeen identified. Recently, Makarova et al. (K. S. Makarova,N. V. Grishin, and E. V. Koonin, Bioinformatics 22:2581-2584,2006) suggested that the hicAB loci constitute a novel TA genefamily. Using the hicAB locus of Escherichia coli K-12 as amodel system, we present evidence that supports this inference:expression of the small HicA protein (58 amino acids [aa]) inducedcleavage in three model mRNAs and tmRNA. Concomitantly, theglobal rate of translation was severely reduced. Using tmRNAas a substrate, we show that HicA-induced cleavage does notrequire the target RNA to be translated. Expression of HicB(145 aa) prevented HicA-mediated inhibition of cell growth.These results suggest that HicB neutralizes HicA and thereforefunctions as an antitoxin. As with other antitoxins (RelB andMazF), HicB could resuscitate cells inhibited by HicA, indicatingthat ectopic production of HicA induces a bacteriostatic ratherthan a bactericidal condition. Nutrient starvation induced stronghicAB transcription that depended on Lon protease. Mining of218 prokaryotic genomes revealed that hicAB loci are abundantin bacteria and archaea.

INTRODUCTION

Top

INTRODUCTION

Toxin-antitoxin (TA) loci that encode protein antitoxins sharecertain characteristics. They are two-component systems thatencode two small proteins, a toxin and an antitoxin; the antitoxinneutralizes the toxin by direct protein-protein contact; theantitoxin has a DNA binding domain and autoregulates TA operontranscription; many such toxins cleave mRNA; the toxin familiesexhibit complex phylogenetic patterns due to extensive horizontalgene transfer; and members of the same family are often detectedon phages and plasmids, as well as on chromosomes (3, 14, 33).A previous comprehensive analysis of ${approx}$ 120 prokaryotic genomesled to the identification of ${approx}$ 600 loci belonging to the six knownTA gene families (relBE, mazEF, vapBC, phd-doc, ccdAB, and parDE).The genes are distributed in a striking pattern: slowly growingfree-living organisms particularly have many TA loci, whileobligatory intracellular organisms have few or none (27, 33).For example, Mycobacterium tuberculosis has more than 60, andSulfolobus spp. has ${approx}$ 30 TA loci, while Mycobacterium leprae andrickettsiae have none.

The RelE and MazF families of toxins inhibit translation bymRNA cleavage (34, 45) and were coined mRNA interferases (45).Doc of P1 also inhibits translation (26). Two other TA families,typified by ccd of plasmid F and parDE of plasmid RK2, encodeinhibitors of DNA gyrase (1, 4, 20, 30). The biological functionsof chromosome-encoded TA loci have been the subjects of considerabledebate. In almost all cases investigated, chromosome-encodedTA loci are induced by nutritional stress, raising the possibilitythat these TA loci function to help the cells cope with environmentalstress (5, 14). In further support of the stress response hypothesis,transcription of TA loci was induced by heat shock in Sulfolobussolfataricus and by exposure to chloroform in Nitrosomonas europaea(18, 40). There is also evidence that a MazF toxin homologueelicits programmed cell death during Myxococcus species development(32). It has been suggested that mazEF of Escherichia coli alsomediates programmed cell death (2, 24, 25), although we andothers have not been able to reproduce these results (10, 42).

A bicistronic locus closely linked to the pilus gene clusterof Haemophilus influenzae (hif) was called hic (for hif contiguous)(29). In a recent bioinformatic study, it was suggested thatthe hicAB loci constitute a novel TA gene family with many membersin bacteria and archaea (27). Predicted HicA proteins (COG1724)are small and have a double-stranded RNA-binding fold, whilepredicted HicB proteins (COG1598/4226) have a DNA binding domainfused to a degraded RNase H fold (27). The bioinformatic analysesdid not predict with certainty which of the proteins would bethe toxin or the antitoxin, although HicB has a DNA bindingdomain similar to that of known antitoxins.

Here, we present evidence showing that the hicAB genes of E.coli K-12 have properties very similar to those of previouslycharacterized TA loci. Thus, induction of hicA inhibited translationand induced multiple cleavages in three model mRNAs and tmRNA.Induction of hicB neutralized the detrimental effects of HicA,indicating that HicB can function as an antitoxin. Similar tobona fide TA loci, transcription of hicAB was induced by aminoacid and carbon starvation by a Lon-dependent mechanism. Finally,we present a comprehensive update of TA gene phylogeny. Ouranalysis revealed 1,340 TA loci in 218 prokaryotic genomes,119 of which belong to the new hicAB TA gene family.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Growth conditions and media. Cells were grown either in standard Luria-Bertani (LB) brothor in M9 minimal medium at 37°C, supplemented with 0.2%glucose, 1 µg/µl thiamine, and amino acids in definedconcentrations. When appropriate, the medium was supplementedwith ampicillin (30 or 100 µg/ml) or chloramphenicol (50µg/ml). When amino acid starvation was induced in M9 minimalmedium by the addition of 0.4 mg/ml serine hydroxymate (SHX;Sigma-Aldrich), serine was excluded from the medium. Glucosestarvation was induced in M9 minimal medium containing 0.05%glucose by the addition of methyl ${alpha}$ -D-glucopyranoside (Sigma-Aldrich).Expression of the P_A1/O4/O3 promoter was induced by the additionof 2 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), andexpression of the P_BAD promoter was induced by the additionof 0.2% arabinose.

Bacterial strains and plasmids. Bacterial strains and plasmids used and constructed are listedin Table 1, and DNA primers are listed in Table 2.

View this table:
[in this window]
[in a new window]

TABLE 1. Strains and plasmids used and constructed

View this table:
[in this window]
[in a new window]

TABLE 2. DNA primers

(i) Strain MGJ36. Strain MG1655 ${Delta}$ hicAB was constructed by deletion of hicAB fromthe chromosome of MG1655, by the procedure described in reference11. A PCR product was synthesized with the primers delta hicAB-fand delta hicAB-r, with pKD3 as template. The PCR product waselectroporated into strain BW25113/pKD46, and the cells werespread on Luria agar plates containing 25 µg/ml chloramphenicoland incubated at 37°C. The ${Delta}$ hicAB::cat allele from BW25113 ${Delta}$ hicABwas then transduced into MG1655. The cat gene was removed withthe plasmid pCP20, resulting in MG1655 ${Delta}$ hicAB.

(ii) Plasmid pMJ221. The hicA gene was amplified from MG1655 chromosomal DNA, usingthe primers hicA-f and hicA-r. The PCR product was digestedwith BamHI and EcoRI and inserted into pNDM220. This plasmidexpresses HicA upon addition of IPTG.

(iii) Plasmid pMJ331. The hicB gene was amplified from MG1655 chromosomal DNA, usinghicB-f and hicB-r as primers. The PCR product was digested withHindIII and XbaI and inserted into pBAD33. Cells carrying thisplasmid express hicB upon addition of arabinose. PCR fragmentswere confirmed by sequencing.

Rates of protein, RNA, and DNA synthesis. Cells were grown at 37°C in M9 minimal medium with 0.2%glucose and amino acids in defined concentrations to an opticaldensity at 450 nm (OD₄₅₀) of ${approx}$ 0.5. The cultures were diluted10 times, and 2 mM IPTG was added at an OD₄₅₀ of 0.3. Samplesof 0.5 ml were added to 5 µCi of [³⁵S]methionine (proteinsynthesis), 2 µCi of [methyl-³H]thymidine (DNA synthesis),or 0.1 µCi of [2-¹⁴C]uracil (RNA synthesis) plus a 100-foldexcess of unlabeled carrier. After 1 min of incorporation, sampleswere chased for 10 min with 0.5 ml/mg of cold methionine, 0.5mg/ml thymidine, or 0.5 mg/ml uracil, respectively. The sampleswere harvested and resuspended in 200 µl cold 20% trichloroaceticacid and centrifuged at 20,000 x g for 30 min at 4°C. Thesamples were washed twice with 200 µl cold 96% ethanol.Precipitates were transferred to vials, and the amounts of incorporatedradioactivity were counted in a liquid scintillation counter.

Primer extension analysis. Cells were grown in LB medium at 37°C. At an OD₄₅₀ of 0.5,the cultures were diluted 10 times and grown to an OD₄₅₀ of0.5, and transcription of the toxin was induced by the additionof 2 mM IPTG. To inhibit translation, chloramphenicol (50 µg/ml)was added. Primer extension analysis was used to map the hicABpromoter and the cleavage patterns of the dksA, ompA, rpoD,and, ssrA mRNAs, using ³²P-labeled primers, following extensionby reverse transcriptase. The primer (3 pmol) was labeled with2 µl of [ ${gamma}$ -³²P]ATP at a concentration of 6,000 Ci/mmolby addition of 0.4 µl polynucleotide kinase (New EnglandBiolabs Inc.) in polynucleotide kinase buffer and incubatedfor 1 h at 37°C. Labeled primer was hybridized to 10 to20 µg total RNA and extended with reverse transcriptase(Finnzymes). The labeled cDNA was fractionated by using a 6%polyacrylamide gel electrophoresis, which was dried and placedon a phosphorimager screen.

Reverse transcription qPCR. RNA was extracted from all cell samples using an RNeasy mini-kit(Qiagen) according to the manufacturer's instructions. cDNAsynthesis was performed using a high-capacity cDNA reverse transcriptionkit (Applied Biosystems). Quantitative PCR (qPCR) reactionswere run in duplicate simultaneously. Briefly, 10 to 20 ng cDNAwas mixed with 0.3 µm primers and 10 µl of 2x PCRMaster Mix for Sybr green kit from Eurogentec (Seraing, Belgium),and the qPCR was run on a LightCycler 480 real-time PCR system(Roche). Relative expression was analyzed using qBase software(19).

RESULTS

Top

RESULTS

Overproduction of HicA is bacteriostatic. The hicAB locus of E. coli K-12 is shown schematically in Fig.1A. The hicA reading frame (formerly yncN) consists of 58 codonsand has a GTG start codon and a predicted Shine-Dalgarno sequence(AGGGAGG). The hicB gene (formerly ydcQ) has 145 codons andstarts 24 bp downstream of hicA. We used primer extension analysisto map the transcriptional start site upstream of hicA. Interestingly,the hicA mRNA 5' end was not detectable with total RNA preparedfrom exponentially growing cells (Fig. 1B, –2' sample).However, the addition of chloramphenicol induced the appearanceof a distinct cDNA band, consistent with a transcriptional startsite located 66 bp upstream of the hicA start codon (Fig. 1A and B).Thus, blockage of translation induced hicAB transcription. Noadditional promoter was detected in the region between hicAand hicB (data not shown).

View larger version (32K):
[in this window]
[in a new window]

FIG. 1. Genetic organization of the hicAB locus. (A) Genetic map of hicAB. The enlargement shows the hicAB promoter region. (B) The 5' ends of the hicAB mRNA were mapped by primer extension using a ³²P-labeled hicA-20 primer (see Materials and Methods). Cells of E. coli MG1655 were grown exponentially in LB medium at 37°C. At an OD₄₅₀ of ${approx}$ 0.5, chloramphenicol (Cml) (50 µg/ml) was added, and samples were taken at the time points indicate promoter sequences. Numbers at left SD, Shine-Dalgarno sequence.

The hicA gene was PCR amplified and inserted downstream of theIPTG-inducible P_A1/03/04 promoter of the low-copy-number vectorpNDM220, resulting in pMJ221. Similarly, hicB was inserted downstreamof the arabinose-inducible P_BAD promoter of pBAD33, resultingin pMJ331. Strains MG1655/pMJ221 (P_A1/03/04::hicA) and MG1655/pMJ331(P_BAD::hicB) were grown exponentially in rich medium. Inductionof hicA resulted in a rapid cessation of cell growth (Fig. 2A)and simultaneously inhibited cell proliferation (Fig. 2B). Theseresults show that ectopic production of HicA is bacteriostatic.By contrast, induction of hicB had no effect on cell growth.

View larger version (15K):
[in this window]
[in a new window]

FIG. 2. hicAB of E. coli is a TA locus. (A) Growth of MG1655 wild-type (wt) cells carrying pMJ221 (P_A1/O3/O4::hicA) or pMJ331 (P_BAD::hicB) were grown exponentially in LB medium at 37°C. At an OD₄₅₀ of ${approx}$ 0.5, transcription of hicA and hicB was induced by the addition of 2 mM IPTG and 0.2% arabinose, respectively. (B) CFU counts after induction of hicA and hicB in wild-type and in ${Delta}$ hicAB strains (hicA induction only) are shown. Cell samples were diluted and plated on rich solid medium containing ampicillin (30 µg/ml) for the hicA-carrying plasmid pMJ221 and chloramphenicol (50 µg/ml) plates for the hicB-carrying plasmid pMJ331. (C) Growth of MG1655/pMJ221 (P_A1/O3/O4::hicA)/pBAD33 (vector control lacking HicB [–HicB]) and MG1655/pMJ221 (P_A1/O3/O4::hicA)/pMJ331 (P_BAD::hicB) (+HicB). Cultures were grown in LB medium at 37°C to exponential phase and then diluted to an OD₄₅₀ of ${approx}$ 0.1, and growth was continued for 1 h. At this point, hicA was induced by addition of IPTG (2 mM). Ninety minutes later, hicB was induced by addition of arabinose (0.2%).

HicB neutralizes HicA-mediated inhibition of growth. Next, we deleted the chromosome-encoded copy of the hicAB locusand repeated the hicA induction experiment described above.In this case, HicA also inhibited cell growth (data not shown)and, moreover, reduced the number of colony-forming cells (Fig.2B). This result indicated that HicA was more active in the ${Delta}$ hicAB strain than in the wild-type strain. Therefore, we testedwhether HicB could neutralize HicA. Cells containing plasmidscarrying the inducible hicA (IPTG) and hicB (arabinose) geneswere grown exponentially. As described before, induction ofhicA inhibited cell growth (Fig. 2C). Strikingly, however, laterinduction of hicB resulted in an immediate resumption of cellgrowth, showing that HicB can indeed counteract HicA. Moreover,the rapid resumption of cell growth indicated that HicA wasbacteriostatic rather than bactericidal, at least for the 90min of hicA induction used here.

HicA inhibits the global rate of translation. Since the molecular target of HicA is unknown, we measured therates of protein, DNA, and RNA syntheses after induction ofhicA. As shown in Fig. 3, the rate of translation was stronglyinhibited (20-fold after 1 h with hicA induction), while replicationand transcription continued unaffected during this period.

View larger version (21K):
[in this window]
[in a new window]

FIG. 3. Rates of protein, DNA, and RNA synthesis after induction of hicA. Cells of E. coli MG1655 carrying pMJ221 (P_A1/O3/O4::hicA) were grown exponentially in M9 minimal medium, and hicA was induced by addition of IPTG (2 mM) at time zero. Samples were taken at the time points indicated, and synthesis rates were measured by pulse-labeling for 1 min and a 10-min chase as described in Materials and Methods. The preinduction rates were set to 100%.

Induction of hicA induces mRNA cleavage. Ectopic production of mRNA interferases, such as RelE and MazF,results in a dramatic inhibition of translation (7, 10). Therefore,we investigated whether ectopic production of HicA would inducemRNA cleavage in three different test mRNAs. The primer extensionanalysis illustrated in Fig. 4 shows that HicA induced multiplecleavage sites in the ompA, dksA, and rpoD mRNAs. By comparisonof the band patterns obtained with RNA prepared from cells treatedwith chloramphenicol, HicA-mediated cleavages were distinguishedfrom premature terminations of reverse transcriptase (Fig. 4,Cml). The HicA-mediated cleavage patterns in the three testmRNAs did not yield any obvious consensus recognition motifs.

View larger version (80K):
[in this window]
[in a new window]

FIG. 4. HicA induces cleavage of the ompA, dksA, and rpoD mRNAs and tmRNA. Primer extension analysis of the ompA (A), dksA (B), and rpoD (C) mRNAs and tmRNA (D) using the ³²P-labeled primers ompA-ctr-ccw, dksA PE1, rpoD PE1, and 10SA-2, respectively, are shown. (A to C) Total RNA was prepared from cells growing exponentially in LB medium at 37°C. At time zero, 50 µg/ml chloramphenicol (first panels, labeled Cm1) or 2 mM IPTG (second panels, labeled hicA) was added. The strains used were MG1655 (first panels) and MG1655/pJM221 (P_A1/O4/O3::hicA) (second panels). (D) The first panel shows results with strain MG1655; second panel with strain MG1655/pJM221 (P_A1/O4/O3::hicA); third panel with strain MG1655 ${Delta}$ ssrA/pMJ221 (P_A1/O3/O4::hicA)/pSC320 (encoding wild-type tmRNA and smpB); and fourth panel with strain MG1655 ${Delta}$ ssrA/pMJ221 (P_A1/O3/O4::hicA)/pSC321 (encoding tmRNA carrying a resume codon mutation and smpB). Square dots indicate HicA-induced cleavages in the RNA sequences. The numbers above the lanes are time points of cell sampling relative to hicA induction or the addition of chloramphenicol (left panels). GCA* and UAA* are wild-type and mutated tmRNA resume codons, respectively.

HicA-induced RNA cleavage does not depend on translation. Next, we investigated whether HicA also induced cleavages intmRNA. Figure 4D shows a primer extension analysis of the codingregion of tmRNA before and after induction of hicA. As shown,HicA induced cleavage of wild-type tmRNA at two sites (A ${downarrow}$ AAC).Addition of chloramphenicol did not lead to any appreciablecleavage of tmRNA, thus showing that the cleavages were inducedby HicA. Primer extension analysis of the nontranslated regionsof tmRNA did not reveal additional cleavage sites (data notshown).

Both of the HicA-induced cleavage sites were located withinthe translated part of tmRNA. Therefore, we tested HicA cleavageof a nontranslated variant of tmRNA that had its GCA resumecodon changed to a UAA ochre codon (44). The mutant tmRNA exhibiteda cleavage pattern indistinguishable from that of the wild-typetmRNA (Fig. 4D, third and fourth panels). Thus, the HicA-mediatedcleavages in tmRNA were independent of tmRNA translation.

Amino acid and carbon starvation induce hicAB transcription. We investigated whether hicAB would be induced during nutrientstarvation. Amino acid starvation was induced by the additionof SHX to exponentially growing cells, and the relative levelof hicAB mRNA was measured by qPCR. As shown in Fig. 5A, SHXinduced a rapid and dramatic increase in the level of hicABtranscripts. Sixty minutes after the onset of amino acid starvation,the level of hicAB mRNA was ${approx}$ 15-fold increased. Assuming thatthe metabolic stability of hicAB mRNA was not changed by aminoacid starvation, as was the case with relBE mRNA (9), we concludethat amino acid starvation activates transcription of hicAB.

View larger version (12K):
[in this window]
[in a new window]

FIG. 5. Nutrient starvation induces hicAB transcription by a Lon-dependent mechanism. (A) Cells were grown exponentially in LB medium, and samples were withdrawn before and after the addition of SHX (1 mg/ml). (B) Cells were grown exponentially in M9 minimal medium containing 0.05% glucose, and samples were withdrawn before and after the addition of methyl- ${alpha}$ -D-glucopyranoside (2%). (C) Samples were withdrawn before and after the addition of 50 µg/ml chloramphenicol as indicated. Strains SG22093 ( ${Delta}$ clpP) (D) and SG22095 ( ${Delta}$ lon) (E) were grown exponentially in LB medium at 37°C, and starvation was induced by the addition of SHX (1 mg/ml).

Next, we induced glucose starvation by the addition of ${alpha}$ -methylglucoside to cells growing in glucose minimal medium (Fig. 5B).In this case, hicAB transcription was also stimulated, althoughto a lesser degree ( ${approx}$ sixfold). Finally, we measured the effectof chloramphenicol on hicAB transcription (Fig. 5C). As shown,hicAB transcription was stimulated strongly by chloramphenicol( ${approx}$ 12-fold), consistent with the result shown in Fig. 1B.

Activation of hicAB depends on Lon. The above-described results are consistent with the notion thatHicB is an unstable antitoxin that autoregulates hicAB transcription.If so, induction of hicAB transcription might depend on a cellularprotease. To test this possibility, we investigated the transcriptionalresponse to amino acid starvation in lon or clpP protease mutantstrains. Deletion of clpP did not reduce induction of hicABduring amino acid starvation (Fig. 5D). However, deletion oflon severely reduced the activation of hicAB transcription duringamino acid starvation (Fig. 5E). This observation suggests thatHicB is an autorepressor of hicAB transcription and that Londegrades HicB during starvation.

Distribution of hicAB and other TA loci in prokaryotic genomes. Previously, we generated a comprehensive phylogenetic analysisof the six known TA locus families in 126 completely sequencedprokaryotic genomes (RelE, MazF, VapC, Doc, CcdB, and ParE)(33). Here, we extended the previous survey to encompass 218(22 archaeal and 196 bacterial) genomes and included the hicABgene family in the analysis. In total, we identified 1,340 TAloci divided among 1,069 loci of bacteria ( ${approx}$ 5.4 loci per genome)and 272 loci of archaea ( ${approx}$ 12.4 loci per genome), of which 119were hicAB loci (Table 3). Figure 6 shows the two bacterialand two archaeal genomes with the highest number of hicAB locitogether with members of the six other TA families. Treponemadenticola, a gram-negative spirochete, has at least 32 TA loci,9 of which belong to the hicAB family. Photorhabdus luminescens,an enteric insect pathogen, has at least 59 TA loci, 8 of whichbelong to the hicAB family. Methanosarcina activorans and M.mazei have 21 and 18 TA loci, respectively. In both cases, 8of these are hicAB loci. Thus, genomes carrying multiple TAloci are common.

View this table:
[in this window]
[in a new window]

TABLE 3. Phyletic distribution of seven TA families in 218 prokaryotic genomes^a

View larger version (35K):
[in this window]
[in a new window]

FIG. 6. Bacterial and archaeal genomes with many hicAB loci. The bacterial (two upper diagrams) and archaeal (two lower diagrams) genomes with the highest number of hicAB loci are shown. Other TA loci are included for comparison.

Table S1 in the supplemental material gives an overview of theTA loci that we identified in 218 prokaryotic genomes. Detailedinformation for all these genes is given in Table S2 in thesupplemental material. Our extensive survey supports conclusionsdrawn from a previous study that encompassed 118 genomes (33).All archaea, even Nanoarchaeum spp., with its tiny ${approx}$ 490-kb genome,have at least two TA loci (see Table S1 in the supplementalmaterial), whereas almost all obligate intracellular organismsare devoid of TA loci (49 bacterial species) (see Table S3 inthe supplemental material). Thus, the average number of TA lociin free-living bacteria is estimated here to be ${approx}$ 7.2 per genome.Even with this correction, it appears that archaea have, onaverage, significantly more TA loci than bacteria. Some organisms,both archaeal and bacterial particularly, have many TA loci.Organisms with more than 10 TA loci are listed in Table S4 inthe supplemental material. Many of these organisms are characterizedby slow growth (e.g., Mycobacterium tuberculosis and speciesof Geobacter, Synechocystis, Nostoc, Nitrosomonas, and Caulobacterand many archaea).

DISCUSSION

Top

DISCUSSION

Here, we show that hicAB of E. coli K-12 has many characteristicsin common with a number of well-characterized TA loci such asrelBE and mazEF. (i) The hicA gene encodes an inhibitor of translationwhose ectopic production leads to mRNA cleavage. (ii) hicB encodesa protein that neutralizes HicA. (iii) Nutritional stress induceshicAB transcription by a Lon-dependent mechanism. (iv) Chloramphenicolalso induced transcription of the hicAB locus. Observations(iii) and (iv) are consistent with the proposal that HicB isan unstable transcriptional repressor of hicAB transcriptionand that Lon degrades HicB. This proposal is consistent withthe presence of an HTH DNA binding motif in the C terminus ofHicB (6). (v) The phylogenetic pattern of the hicAB locus issimilar to that of the relBE and vapBC loci (Table 3 and seeTable S1 in the supplemental material).

Ectopic production of HicA-induced cleavages in three model mRNAs and in tmRNA. We do not know if these cleavages are direct or indirect, thatis, if HicA overproduction activates an endogenous RNase orif HicA itself is an RNase (Fig. 4). Although many TA loci encodemRNA-cleaving enzymes, there are examples of such translationalinhibitors that mediate indirect mRNA cleavage (13), and thefinal test to determine whether HicA is an RNase awaits biochemicalexperiments.

The hicA (formerly yncN) gene of E. coli K-12 has not been describedbefore (23), whereas hicB (formerly ydcQ) has been implicatedin genetic interaction with ${sigma}$ ^E, the sigma factor of E. coli thatresponds to stress in the cell membrane and periplasmic space(12). Thus, in wild-type cells, ${sigma}$ ^E is essential. However, suppressormutations that render ${sigma}$ ^E nonessential are easily obtained, andone class of such suppressors had IS1 elements inserted intohicB (6). Deletion of hicB from wild-type cells resulted ina significant downregulation of extracytoplasmic stress responses(both the ${sigma}$ ^E and the Cpx-dependent responses) and a severe reductionin the formation of outer membrane vesicles (6). The reason(s)for these complex phenotypes is not known. However, we suggestthat deletion of hicB (ydcQ) leads to hyperactivation of HicAthat, in turn, leads to degradation of mRNAs encoding proteinsinvolved in extracytoplasmic stress. We are now testing thisproposal.

Including that of hicAB, seven TA families are now known (Table3). Some free-living organisms have a plethora of these genesystems. For example, M. tuberculosis has more than 60 TA lociand N. europaea has at least 49 (see Table S1 and S3 in thesupplemental material). The phylogenetic pattern is consistentwith the proposal that TA loci are particularly beneficial toslowly growing organisms. The reason for this apparent correlationis not known. E. coli has at least seven TA loci that encodemRNA interferases: four relBE loci (relBE [15], yefM yoeB [16],dinJ yafQ [31], and prlF yhaV [35]), two mazEF loci (mazEF andchpB [28]), and hicAB (this work). In all cases, transcriptionof these loci is induced by amino acid starvation. Thus, evensome rapidly growing organisms have multiple TA loci. The reasonfor this apparent redundancy is not known.

The literature contains a number of hypotheses that explainthe physiological function(s) of TA loci. One group suggestedthat mazEF mediates programmed cell death during amino acidstarvation (2), although we and others (10, 42) have not beenable to confirm this. By contrast, we have obtained solid evidencethat relBE and mazEF of E. coli are induced during nutrientstarvation and reduce the global level of translation withoutcell killing (8-10). There are several reasons why a reducedglobal level of translation may be beneficial rather than detrimentalto starving cells. Most importantly, it may reduce the levelof translational errors by reducing the level of uncharged tRNAs(37, 38). Second, a reduced cellular level of translation mayhelp the cells adapt more rapidly to nutrient starvation. Thephylogenetic distribution of TA loci (i.e., their absence fromobligate intracellular organisms) is consistent with the proposalthat they function as stress response elements that are triggeredby nutrient starvation and other forms of environmental stresses.This hypothesis is supported by the observations that TA lociwere induced by heat shock in S. solfataricus and by exposureto chloroform in Nitrosomonas europaea (18, 40). However, weare aware that this hypothesis does not explain the apparentmassive redundancy of TA loci.

It has also been proposed that TA loci may function to helpthe cells enter a dormant state called persister cells (22,36, 43). Persister cell formation can be viewed as an extremecase of translational inhibition, and the phenomenon is compatiblewith the proposal that TA loci reduce the cellular level oftranslation during nutritional stress.

Due to the differential metabolic stabilities of the toxinsand antitoxins, TA loci may also lead to genetic stabilizationof the chromosomal regions to which they are closely linked(39). To test these hypotheses more stringently, we are nowanalyzing model organisms devoid of all known TA loci.

ACKNOWLEDGMENTS

This work was supported by the Danish Centre for mRNP Biogenesisand Metabolism and the Wellcome Trust.

FOOTNOTES

* Corresponding author. Mailing address: Institute for Cell and Molecular Biosciences, Medical School, Newcastle University, Newcastle NE2 4HH, United Kingdom. Phone: 44 0191 222 5318. Fax: 44 0191 222 2467. E-mail: kenn.gerdes{at}ncl.ac.uk

Published ahead of print on 5 December 2008.

Supplemental material for this article may be found at http://jb.asm.org/.

REFERENCES

Top