Rhs proteins from diverse bacteria mediateintercellular competitionSanna Koskiniemia,1, James G. Lamoureuxa,1, Kiel C. Nikolakakisb,2, Claire t’Kint de Roodenbekea,2, Michael D. Kaplana,David A. Lowa,c, and Christopher S. Hayesa,c,3

Departments of aMolecular, Cellular and Developmental Biology and bChemistry and Biochemistry and cBiomolecular Science and Engineering Program,University of California, Santa Barbara, CA 93106

Edited by Susan Gottesman, National Institutes of Health, Bethesda, MD, and approved March 21, 2013 (received for review January 11, 2013)

Rearrangement hotspot (Rhs) and related YD-peptide repeat pro-teins are widely distributed in bacteria and eukaryotes, but theirfunctions are poorly understood. Here, we show that Gram-nega-tive Rhs proteins and the distantly related wall-associated protein A(WapA) from Gram-positive bacteria mediate intercellular competi-tion. Rhs and WapA carry polymorphic C-terminal toxin domains(Rhs-CT/WapA-CT), which are deployed to inhibit the growth ofneighboring cells. These systems also encode sequence-diverse im-munity proteins (RhsI/WapI) that specifically neutralize cognate tox-ins to protect rhs+/wapA+ cells from autoinhibition. RhsA and RhsBfrom Dickeya dadantii 3937 carry nuclease domains that degradetarget cell DNA. D. dadantii 3937 rhs genes do not encode secretionsignal sequences but are linked to hemolysin-coregulated proteinand valine-glycine repeat protein G genes from type VI secretionsystems. Valine-glycine repeat protein G is required for inhibitor cellfunction, suggesting that Rhs may be exported from D. dadantii3937 through a type VI secretion mechanism. In contrast, WapAproteins from Bacillus subtilis strains appear to be exported throughthe general secretory pathway and deliver a variety of tRNasetoxins into neighboring target cells. These findings demonstratethat YD-repeat proteins from phylogenetically diverse bacteriashare a common function in contact-dependent growth inhibition.

bacterial competition | CDI | cell-cell interaction | DNase activity |toxin/immunity genes

Rearrangement hotspot (rhs) elements were first identified assites that promote recombination in Escherichia coli (1). The

region between rhsA and rhsB is frequently duplicated in E. colicells, presumably because these genes share large regions (∼3.7kbp) of sequence identity and are positioned relatively close toone another in the genome. The large invariant rhs sequence wasoriginally termed the “core” (2) and encodes the conservedN-terminal 1,240 residues of E. coli Rhs proteins including theYD-peptide repeats that define this protein family (Pfam ID:PF03527 and PF05593). More recent analysis has revealed thatenterobacterial Rhs proteins are composed of four distinctregions and that the core can vary among family members (3).Rhs C-terminal regions (Rhs-CT) are highly variable and enco-ded by “core extensions” with GC content and codon bias that aredifferent from sequences encoding the N-terminal regions (2–4).The Rhs-CT region is sharply demarcated by a PxxxxDPxGLpeptide motif within the adjacent hyperconserved domain (3).This structure suggests that rhs genes are composites formed bythe insertion of horizontally transferred core-extension sequences(2, 3). Rhs-CT polymorphism is extensive, and different strains ofa given species typically contain a distinct complement of rhsalleles (2, 3, 5). Rhs are distributed throughout β-, γ-, and δ-pro-teobacteria, and genes encoding more distantly related YD-pep-tide repeat proteins are found in Gram-positive bacteria, somefungi, and higher metazoans through the vertebrates (6–8). Thisbroad distribution suggests that Rhs/YD-repeat proteins playa fundamental role in biology. Indeed, the rhs genes of E. coli andShigella strains show evidence of strong positive selection (9);however, the functions of this gene family are not well understood.

We recently found that some Rhs-CT regions share sequenceidentity with the toxic effector domains of contact-dependentgrowth inhibition (CDI) systems (10). CDI systems mediate inter-bacterial competition and encode large CdiA proteins with highlyvariable C-terminal toxin domains (CdiA-CT) (11, 12). CdiA isexported to the inhibitor cell surface where it engages receptors onrelated bacteria and delivers its toxin into target cells (13, 14).CDI+ bacteria also carry cdiI immunity genes immediately down-stream of cdiA, and together these genes form a family of poly-morphic toxin/immunity pairs. CdiI immunity proteins bind andneutralize CdiA-CT toxins to protect CDI+ cells from auto-inhibition (11). Because CdiI proteins are highly variable, theyprotect cells from cognate CdiA-CT but not from the toxinsdeployed by other CDI systems. Similar to CDI, the rhsB locus ofDickeya dadantii 3937 encodes a toxin/immunity protein pair.RhsB-CT inhibits growth when expressed in E. coli, but this toxicityis blocked by coexpression of an immunity protein (RhsIB) encodedimmediately downstream of rhsB (10). These parallels with CDI ledus to examine the role of Rhs proteins in intercellular competition.We find that D. dadantii 3937 uses two of its three Rhs proteins toinhibit the growth of neighboring cells in a contact-dependentmanner. Additionally, we show that distantly related wall-asso-ciated protein A (WapA) from Bacillus subtilis strains also carriesvariable C-terminal toxin domains and is encoded with a linkedimmunity gene. Like D. dadantii 3937, B. subtilis 168 deploysWapA to inhibit adjacent nonimmune cells. Together, theseresults demonstrate that YD-repeat proteins from diverse bac-teria share a common function in intercellular competition.

ResultsRhs Proteins Mediate Intercellular Competition in D. dadantii 3937.To explore the role of Rhs in intercellular competition, we ex-amined the three rhs loci in Dickeya dadantii 3937 (Fig. S1).We first generated target strains deleted for individual rhs/rhsIgene pairs, reasoning that cells lacking rhsI immunity genesshould be susceptible to inhibition by the wild-type rhs+ strain.Indeed, ΔrhsA-ΔrhsIA andΔrhsB-ΔrhsIBmutants were outcompetedby wild-type D. dadantii 3937 during coculture on solid medium,whereas growth of the ΔrhsC-ΔrhsIC mutant was unaffected (Fig.1A). These results suggest that D. dadantii 3937 uses RhsA andRhsB for intercellular competition. The Rhs toxin/immunity modelpredicts that RhsI proteins should protect cells from growth

Author contributions: S.K., J.G.L., C.t.d.R., D.A.L., and C.S.H. designed research; S.K., J.G.L.,K.C.N., C.t.d.R., and M.D.K. performed research; S.K., J.G.L., K.C.N., C.t.d.R., and M.D.K.contributed new reagents/analytic tools; S.K., J.G.L., K.C.N., M.D.K., D.A.L., and C.S.H.analyzed data; and C.S.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1S.K. and J.G.L. contributed equally to this work.2K.C.N. and C.t.d.R. contributed equally to this work.3To whom correspondence should be addressed. E-mail: chayes@lifesci.ucsb.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1300627110/-/DCSupplemental.

7032–7037 | PNAS | April 23, 2013 | vol. 110 | no. 17 www.pnas.org/cgi/doi/10.1073/pnas.1300627110

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

inhibition. Therefore, we reintroduced the rhsIA and rhsIB genesinto ΔrhsA-ΔrhsIA and ΔrhsB-ΔrhsIB target cells, respectively,using a Tn7-based integration vector and tested the complementedstrains in competitions with rhs+ cells. As predicted, each rhsI geneprotected target cells from growth inhibition (Fig. 1A). Together,these results indicate that D. dadantii cells deploy Rhs to inhibitother cells and protect themselves with immunity proteins.RhsA and RhsB are nearly identical over 1,303 N-terminal

residues, but their C-terminal toxin domains are divergent (Fig.S2A). Similarly, the associated RhsIA and RhsIB proteins areunrelated (Fig. S2B), suggesting that immunity is specific foreach Rhs-CT. To test this hypothesis, we cocultured inhibitorsthat express only RhsA (rhsA+/ΔrhsB) or RhsB (ΔrhsA/rhsB+)with ΔrhsA-ΔrhsIA/ΔrhsB-ΔrhsIB target cells complemented with

either rhsIA or rhsIB. The rhsIA gene conferred immunity torhsA+/ΔrhsB inhibitors but not to ΔrhsA/rhsB+ cells (Fig. 1B).Similarly, rhsIB protected only targets from ΔrhsA/rhsB+ inhib-itors (Fig. 1B), demonstrating the specificity of Rhs immunity.Because Rhs shares many features with CDI, we asked whetherRhs-mediated inhibition also requires close contact betweencells. We repeated the competitions but separated inhibitorsfrom target cells using membranes of different porosities. RhsA-and RhsB-mediated inhibition was not observed when cells weresegregated with a membrane containing 0.45-μm pores, whichallow transfer of soluble factors but prevent cell mixing (Fig. 1C).However, target cells were inhibited when separated frominhibitors by 8.0-μm pores (Fig. 1C), which allow bacterial cellpassage. These results suggest that soluble Rhs is not inhibitoryand that cells must be in close proximity for inhibition.

D. dadantii 3937 Rhs Toxins Degrade Target Cell DNA. RhsA andRhsB inhibit cell growth, but their biochemical activities areuncharacterized. The RhsA-CT domain is a member of the en-donuclease NS_2 family (PF13930), and RhsB-CT contains anHNH endonuclease motif (PF01844), suggesting that both toxinsare nucleases. To test these predictions, we expressed each toxininside E. coli and stained the cells with DAPI to visualize DNAby microscopy. Growth was inhibited after ∼1.5 h of RhsA-CTinduction, and cells became elongated with condensed, centrallylocated nucleoids (Fig. S3A). The rhsB-CT coding sequencecould not be isolated from the linked immunity gene, presumablybecause of toxicity. Therefore, we used controlled proteolysis ofRhsIB to activate the RhsB-CT toxin inside E. coli cells (10, 15).RhsB-CT activation immediately arrested growth, and the in-hibited cells became filamentous and lost DNA staining (Fig.S3B). Additionally, plasmid DNA was degraded rapidly in cellsoverproducing either RhsA-CT or RhsB-CT (Fig. S4), indicatingthat changes in DAPI staining probably reflect DNA damage.We also tested RhsC-CT and found that it inhibits E. coli growthbut does not alter nucleoid structure (Fig. S3C). For each toxin,growth inhibition and the associated changes in cell morphologywere blocked specifically by the cognate RhsI immunity protein(Fig. S3). We next asked whether DNA is degraded in D.dadantii 3937 target cells during coculture with rhs+ inhibitors.We labeled ΔrhsA-ΔrhsIA/ΔrhsB-ΔrhsIB target cells with GFP todifferentiate them from unlabeled inhibitors by fluorescencemicroscopy. DAPI staining revealed that many target cells hadlost DNA during coculture with either rhsA+/ΔrhsB or ΔrhsA/rhsB+ inhibitors (Fig. 2A). However, targets complemented withcognate rhsI retained their DNA (Fig. 2A), as did target cells thatwere cocultured with mock inhibitors that lack rhsAIA and rhsBIBloci (Fig. 2B). Quantification of these data showed that 40–45% ofthe targets lost DNA after 12 h of coculture with inhibitors (Fig.2C). In contrast, none of the targets expressing cognate rhsI lostDNA staining (Fig. 2C). Together, these findings demonstrate thatRhsA and RhsB toxins act by degrading target cell DNA.

Valine-Glycine Repeat Proteins Are Required for RhsB-MediatedInhibition. RhsA and RhsB deliver DNase domains into targetcells, but both proteins lack recognizable secretion signal sequen-ces, raising the question of how these toxins are exported. The rhsgenes in many bacteria are linked to type VI secretion systems(T6SS), suggesting a possible mechanism of Rhs delivery. T6SSencode cell-puncturing structures that deliver effector proteins intoboth eukaryotic and prokaryotic cells (16). This apparatus resem-bles a tailed bacteriophage assembly and is composed of a fila-mentous tube of hemolysin-coregulated protein (Hcp) protomerscapped by valine-glycine repeat proteins (VgrG) (16). The D.dadantii 3937 rhsA and rhsB loci are linked to hcp and vgrG genes,whereas rhsC is adjacent to a full complement of T6SS genes (Fig.S1). To test whether T6S plays a role in RhsB-mediated inhibition,we deleted vgrG genes from ΔrhsA/rhsB+ inhibitor cells and

A

competitiveindex

inhibitors:

targets:

wild-type ( +)rhs

A B

rhsAIA rhsBIB rhsCIC101

100

10-1

10-2

10-3

:rhsI

C

0.45 m 0.45 m8.0 m 8.0 m

inhibitors:

targets:

rhsA+/ rhsB rhsA/rhsB+

:rhsIA A B B

competitiveindex

101

100

10-1

10-2

B

competitiveindex

inhibitors:

targets:

rhsA+/ rhsB rhsA/rhsB+

A AB B

rhsAI / rhsBIA B

101

100

10-1

10-2

10-3

:rhsI

rhsAI / rhsBIA B

Fig. 1. Rhs mediates contact-dependent inhibition. (A) Rhs mediates competi-tion between D. dadantii 3937 cells. Wild-type inhibitors (rhs+) were coculturedwith target cells lacking the indicated rhs loci. (B) Specificity of Rhs immunity.Inhibitors expressing either RhsA (rhsA+/ΔrhsB) or RhsB (ΔrhsA/rhsB+) werecocultured with targets lacking both rhsAIA and rhsBIB loci. (C) Rhs-mediatedgrowth inhibition requires cell–cell contact. Inhibitor and target cells were sepa-rated by membranes containing 0.45-μm or 8.0-μm pores. In each panel, targetcells were complemented with rhsIA or rhsIB where indicated. For all competi-tions, cells were cocultured at a 10:1 inhibitor-to-target ratio, and target cellgrowth is expressed as the competitive index as described in Materials andMethods.

Koskiniemi et al. PNAS | April 23, 2013 | vol. 110 | no. 17 | 7033

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

examined the effects in growth competitions. Inhibitor cells lackingvgrGB still outcompeted targets, but deletion of both vgrGA andvgrGB genes abrogated this competitive advantage (Fig. 3A). In-hibitor activity was restored when vgrGB was reintroduced into theΔvgrGA ΔvgrGB strain (Fig. 3A). We also tested the role of thelinked hcp genes and again found that inhibitors lacking both hcpAand hcpB lost their competitive advantage over target cells (Fig.3B). However, the ΔhcpA ΔhcpB phenotype could not be com-plemented with reintroduced hcpA or hcpB (Fig. 3B). The ΔhcpAΔhcpB inhibitor strain grows more slowly than ΔrhsA-ΔrhsIA/ΔrhsB-ΔrhsIB target cells. Therefore, the apparent loss of inhibitor cellfunction could reflect a difference in growth rates. Nevertheless,a copy of either vgrGA or vgrGB is required for the inhibitor cellphenotype, suggesting that Rhs proteins are exported usinga T6S mechanism.

WapA Proteins from Bacillus subtilis Strains Contain C-terminal ToxinDomains. Rhs proteins are found only in Gram-negative bacteria,but many Gram-positive bacteria express large WapA proteinsthat contain YD-peptide repeats (8). As with Rhs and CdiA, theWapA-CT region is variable among different species and strains.All Bacillus subtilis strains contain a single wapA locus thatencodes one of four distinct WapA-CT sequences (Fig. S5A).Moreover, the genes immediately following wapA also are vari-able, with four sequences that cosegregate with the wapA-CTalleles (Fig. S5B). These observations suggest that wapA and thedownstream ORF encode cognate toxin/immunity pairs. To de-termine if WapA-CT domains are toxic, we cloned wapA-CTsequences from various B. subtilis strains under the arabinose-inducible araBAD promoter (PBAD) and tested whether theresulting plasmids inhibit E. coli cell growth. Although eachwapA-CT plasmid could be maintained stably when expressionwas repressed with D-glucose, none of the plasmids could beintroduced into E. coli cells in the presence of L-arabinose (Fig.4). However, cells expressing predicted wapI immunity genesfrom a given B. subtilis strain were readily transformed withwapA-CT constructs from the same strain (Fig. 4). These findingssuggest that WapA-CTs have toxic activities that are neutralizedby cognate WapI immunity proteins. We next sought to identifythe biochemical activities of WapA-CTs by examining nucleic

acids from inhibited E. coli cells. Gel analysis revealed extensivetRNA cleavage in cells expressing wapA-CT168 from B. subtilis168 but not in cells coexpressing the toxin and cognate wapI(yxxG) immunity gene. Northern blot analysis confirmed thatseveral different tRNA isoacceptors are cleaved by this activity(Fig. 5A), and the cleavage sites were mapped to the 3′ end of E.coli tRNA2

Arg using S1 nuclease protection analysis (Fig. S6).This RNase activity could account for the observed growth in-hibition because tRNAs lacking the universally conserved 3′-CCA sequence cannot support protein synthesis. AdditionalNorthern blot screening revealed that the other WapA-CT toxinsare specific tRNases. Induction of wapA-CTnatto from B. subtilissubsp. ‘natto’ led to tRNAGlu cleavage, whereas tRNA3

Ser wascleaved in cells expressing wapA-CTT-UB-10 from B. subtilissubsp. spizizenii T-UB-10 (Fig. 5B). In each instance, tRNaseactivity was specifically blocked by coexpression of the cognatewapI gene (Fig. 5B). Thus, each WapA-CT possesses a distincttRNase activity capable of inhibiting cell growth.

WapA Mediates Intercellular Competition in Bacillus subtilis 168. Fi-nally, we tested whether WapA mediates intercellular competi-tion. We deleted the wapA-CT/yxxG region from B. subtilis 168 togenerate a target strain for coculture with wild-type wapA+ cells.Initially, we conducted competitions in transwell chambers, whichseparate the two populations with membranes containing either0.45-μm or 8.0-μm pores. The ΔwapA-CT-ΔyxxG targets wereinhibited if they could make contact with wapA+ cells but not whensegregated from the inhibitors by 0.45-μm pores (Fig. 6A). More-over, ΔwapA-CT-ΔyxxG targets carrying a plasmid-borne copy ofyxxG were protected from inhibition (Fig. 6A), demonstrating thatthis gene is sufficient to confer immunity. WapA-CT variabilitysuggests that this system is modular and can deliver different toxins.To test this hypothesis, we generated chimeric inhibitor strains byreplacing the wapA-CT/yxxG coding region in B. subtilis 168 with thecorresponding sequences from B. subtilis subsp. spizizenii str. W23and B. subtilis subsp. ‘natto’. Both chimeric strains inhibited ΔwapA-CT-ΔyxxG target cells, and targets were protected only by cognatewapI immunity genes (Fig. 6B). Together, these results demonstratethat B. subtilis WapA functions like Rhs as a modular contact-dependent inhibition system.

rhsA+/ rhsB

rhsA+

target :rhsI

inhibitors

targets

%anucleatetargets rhsB

+

DAPI

DAPI

GFP

0

0

20

40

60

12 12 12 12 hr0 0 0

GFP

GFP

DAPI

rhsA/rhsB+

rhsA/ rhsB

+rhsIA

rhsA/

rhsB

+rhsIB

rhsA/

rhsB

rhsA/

rhsB

0 hr 0 hr12 hr 12 hr

A

A AB B

B

C

0 hr 12 hr

:inhibitor

Fig. 2. RhsA and RhsB degrade target cell DNA. (A) Fluorescence microscopy of D. dadantii 3937 cocultures. Inhibitors expressing RhsA (rhsA+/ΔrhsB) or RhsB(ΔrhsA/rhsB+) were cocultured at a 1:1 ratio with GFP-labeled target cells lacking both rhsAIA and rhsBIB loci. Targets were complemented with either rhsIA orrhsIB where indicated. Samples from 0 and 12 h were stained with DAPI to visualize genomic DNA. Anucleate target cells are indicated by white arrows. (B)Mock competition between ΔrhsA/ΔrhsB inhibitors and GFP-labeled ΔrhsA/ΔrhsB target cells. (C) Quantification of anucleate target cells. GFP-labeled targetswere examined visually and scored as anucleate if they completely lack DAPI staining. Reported values represent the mean ± SEM.

7034 | www.pnas.org/cgi/doi/10.1073/pnas.1300627110 Koskiniemi et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

DiscussionRhs elements were discovered more than 30 y ago, but only noware their physiological functions being elucidated. Early studiesfound that the E. coli rhsA core extension inhibits recovery fromstationary phase (17), and recent work suggests that this toxininhibits protein synthesis (18). Additionally, a plasmid-encodedRhs protein has been implicated in bacteriocin production inPseudomonas savastanoi (19). Bioinformatic analyses also sup-port the conclusion that bacterial Rhs proteins commonly carrytoxin domains (10, 20). We propose that intercellular growthinhibition is the primary function of these proteins because thedistantly related WapA proteins of Bacillus and Listeria speciesshare a similar architecture with Rhs and also deliver toxicC-terminal domains. The divergence of Rhs and WapA probablyreflects the different strategies required to deliver proteintoxins across Gram-negative and Gram-positive cell envelopes.D. dadantii 3937 appears to use a T6S apparatus to introduceRhs toxins into target cells. It is unclear whether an entire Rhs

protein could be delivered in this process because these systemstypically secrete much smaller effector proteins (16). Remarkably,T6SS can inject toxins into a variety of Gram-negative bacteriaand eukaryotes (21–23), suggesting that D. dadantii 3937 also maydeliver Rhs to other species. On the other hand, WapA carriesa canonical secretion signal sequence and is stably associated withthe peptidoglycan wall of B. subtilis (8). By analogy with bacter-iocins and CDI systems (13, 24), WapA probably binds to cell-surface receptors on target cells and subsequently delivers its C-terminal toxin domain. Such surface receptor–ligand interactionsare specific, and therefore we predict that WapA-mediated in-hibition is restricted to closely related bacteria. Despite the dif-ferences in delivery modalities, Rhs and WapA act only onnearby target cells. Thus, both systems are designed to influencethe immediate environment and therefore could serve to estab-lish a niche and defend it against other bacteria. The fact thatdiverse bacteria possess several distinct systems (CDI, Rhs, andT6SS) to mediate contact-dependent inhibition suggests that thisactivity is fundamental to bacterial biology.YD-repeat proteins play an important role in bacterial inter-

actions with eukaryotic host cells. The insecticidal toxin-complexC proteins (TccC) of Photorhabdus luminescens strains containYD-peptide repeats and have a core/core-extension architecturethat is reminiscent of Rhs (25). TccC core extensions carry toxindomains that are used to destroy the midgut of insect hosts.The Serratia entomophila pathogenicity C protein (SepC) shareshomology with TccC and very likely has the same insecticidalfunction (26). Similarly, RhsT (UniProt ID: A6N5U6) fromPseudomonas aeruginosa PSE9 recently was shown to be trans-located into phagocytic cells, where it induces inflammasome-mediated cell death (27). Kung et al. (27) also showed that micesurvive infection with ΔrhsT mutants in an acute pneumoniamodel, demonstrating that RhsT is a bona fide virulence factor.However, we note that a potential immunity gene lies immedi-ately downstream of rhsT (GenBank ID: EF611305.1). Thetranslation initiation signals for this unannotated gene overlapwith the rhsT 3′-coding sequence, which is very similar to the

competitiveindex

competitiveindex

:vgrG

:hcp

rhsA/rhsB+

rhsA/rhsB+

rhsA/rhsB

rhsA/rhsB

vgrGB

hcpB

hcpA

vgrG

vgrG

A

B

/

hcpA

hcpB

/

B

A

B

B

A

B

101

101

100

100

10-1

10-1

10-2

10-2

10-3

10-3

10-4

inhibitors:

inhibitors:

Fig. 3. Role of T6SS genes in RhsB activity. (A) VgrG is required for RhsB-mediated inhibition. RhsB+ (ΔrhsA/rhsB+) inhibitors carrying vgrG deletionswere tested in competition with ΔrhsAIA ΔrhsBIB target cells. Inhibitors werecomplemented with vgrGB where indicated. (B) RhsB+ (ΔrhsA/rhsB+) inhib-itors carrying hcp deletions were tested in competition with ΔrhsAIA ΔrhsBIBtarget cells. Inhibitors were complemented with hcpBwhere indicated. Mockcompetitions using ΔrhsA/ΔrhsB inhibitors are included in both panels.Inhibitors and targets were cocultured at a 10:1 ratio in all experiments.Target cell growth is expressed as the competitive index as described inMaterials and Methods.

(no toxin)

vector

no

wapI

wapA-CT

natto

wapA-CT

T-UB-10

wapI

natto

wapI

T-UB-10

yxxG

168

wapA-CT

168

Fig. 4. The wapA loci from B. subtilis strains encode different toxin/im-munity pairs. B. subtilis wapA-CT/wapI genes encode a network of cognatetoxin/immunity pairs. Arabinose-inducible wapA-CT constructs were in-troduced into E. coli cells expressing various wapI genes and transformantsselected on media supplemented with L-arabinose.

Koskiniemi et al. PNAS | April 23, 2013 | vol. 110 | no. 17 | 7035

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

arrangement of other rhs toxin/immunity gene pairs. Theseobservations suggest that P. aeruginosa PSE9 is also a potentialRhsT target and needs to protect itself from toxin delivery.

Perhaps RhsT is delivered into both eukaryotic and prokaryoticcells through a T6S mechanism. BecauseD. dadantii 3937 is a soft-rot pathogen, it also could deploy Rhs as virulence factors againstplant host cells. RhsC may be a good candidate for delivery intoplant cells because its gene is linked to a full complement of T6SSgenes. Although the biochemical activity of RhsC-CT is unknown,this domain clearly inhibits E. coli and conceivably could exerta similar effect if introduced into plant hosts. We suspect thatRhs evolved to exchange protein toxins between bacteria andthat some bacteria have co-opted these systems for a role inpathogenesis.In addition to their growth-inhibiting function, Rhs also ap-

pear to coordinate multicellular behavior. Disruption of an rhsgene in Myxococcus xanthus ablates social motility (28), sug-gesting a broader role in communication between kin. Perhapscontact-dependent exchange of Rhs-CTs between isogenic/immune cells provides information about population density,analogous to quorum sensing mediated by soluble factors. Thetoxin/immunity activities may ensure that only isogenic cells areallowed to participate and could exclude “cheaters” who wish tobenefit from cooperative behavior but not bear the burden of rhsexpression. This general hypothesis is supported by a recent re-port that CDI is required for robust biofilm formation in Bur-kholderia thailandensis E264 (29). Thus, immune sibling cellsappear to exploit CDI-mediated contact to build a multicellularcommunity. Intercellular communication also is a function ofthe YD-repeat–containing teneurin proteins of higher metazoans.Teneurins are type II integral membrane proteins that help toestablish neuronal cell connections during development (30, 31).Remarkably, some teneurins carry C-terminal sequences thatresemble neuroendocrine signaling peptides (32). These sequen-ces are adjacent to predicted furin cleavage sites, suggesting thatthe C-terminal peptides are released and transmit signals toneighboring cells. The parallels with the bacterial inhibitionsystems described here are striking and suggest that all YD-repeat proteins share a primordial function in cell–cell contactand communication.

Materials and MethodsBacterial Strains and Plasmids. All bacterial strains and plasmids are listed inTables S1 and S2, respectively. D. dadantii 3937 mutants were constructed byallelic exchange as described (33). Briefly, regions upstream and downstreamof the target gene were amplified by PCR. The two products then werecombined into one fragment by overlapping end-PCR and were ligated intoplasmid pRE112 (33). D. dadantii 3937 gene deletions were complementedusing the mini-Tn7 delivery plasmid pUC18R6KT-mini-Tn7T to introduce genesin a single copy at the glmS locus (34). B. subtilis 168 mutants were generatedby double-crossover integration using a series of plasmid constructs describedin SI Materials and Methods. B. subtilis wapI genes were amplified by PCR andligated to plasmid pHB201 (35) for analysis of immunity function in B. subtilis

A PBAD-wapA-CT

168

tRNA3

Ser

tRNA1B

Ala

tRNA1

Ile

tRNA2

Arg

L-ara

none 168 T-UB-10 wapI:natto

+ + ++

B PBAD-wapA-CT

nattoPBAD-wapA-CT

T-UB-10

tRNA3

Ser

tRNAGlu

tRNA2

Arg

L-ara

none none168 168T-UB-10 T-UB-10natto natto

+ + ++ ++ ++

wapI:

Fig. 5. WapA-CT toxins have diverse tRNase activities. (A) WapA-CT168 is a general tRNase. The PBAD-wapA-CT168 construct was induced with L-arabinose (+)in E. coli cells that coexpress wapI from the indicated strains. RNA was analyzed by Northern blot using probes to the indicated tRNAs. (B) WapA-CTnatto andWapA-CTT-UB-10 toxins are specific tRNases. PBAD-wapA-CT constructs were induced with L-arabinose (+) in E. coli cells that coexpress wapI from the indicatedstrains. tRNAs were analyzed by Northern blot.

target wapI

competitiveindex

inhibitor

wapA-CT

none

B 101

100

10-1

10-2

10-3

10-4

168

W23

none

168

natto

W23

none

168

natto

none

168

W23

natto

W23 natto

A

competitiveindex

0.45m

0.45m

8.0

m

8.0

m

101

100

10-1

10-2

10-3

pHB201 pHB201-

yxxG

wapA-CT/ yxxG

inhibitors:

targets:

wild-type (+)wapA

Fig. 6. WapA functions in intercellular competition. (A) WapA mediates con-tact-dependent inhibition. Wild-type B. subtilis 168 (wapA+) was coculturedwith ΔwapA-CT-ΔyxxG targets in a transwell apparatus. Growth chambers wereseparated by membranes containing 0.45-μm or 8.0-μm pores. Target cells car-ried plasmid-borne yxxG or vector alone (pHB201) where indicated. (B) WapA isa modular toxin delivery system. Wild-type B. subtilis 168 and chimeric inhibitorsexpressing wapA-CT/wapI sequences from the indicated B. subtilis strains werecocultured with ΔwapA-CT-ΔyxxG targets at a 1:1 ratio. Targets expressed wapIfrom the indicated B. subtilis strains. The bar at the far left represents mockcompetition with B. subtilis 168 ΔwapA-CT mutant cells. Target cell growth isexpressed as the competitive index as described in Materials and Methods.

7036 | www.pnas.org/cgi/doi/10.1073/pnas.1300627110 Koskiniemi et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0

168. Sequences encoding Rhs-CT and WapA-CT toxins and corresponding im-munity proteins were amplified by PCR and cloned into derivatives of plasmidspCH450 (36) or pTrc99A for analysis in E. coli cells. Oligonucleotide sequencesare listed in Table S3, and the details of all plasmid constructions are providedin SI Materials and Methods.

Growth Conditions and Competition Assays. D. dadantii 3937 and B. subtilisstrains were routinely grown in LB broth/agar at 30 °C and 37 °C, re-spectively. When appropriate, medium was supplemented with antibiotics asdescribed in SI Materials and Methods. D. dadantii 3937 competitions wereperformed on LB agar buffered at pH ∼7.3 with 50 mM potassium phos-phate and supplemented with 2% (wt/vol) D-glucose. Inhibitors and targetswere mixed at a 10:1 ratio and were plated for coculture at 30 °C. Contactdependence was tested by separating the inhibitor and target cells withmembranes as described in SI Materials and Methods. B. subtilis 168 com-petitions were performed at a 1:1 inhibitor-to-target ratio in shaking LB brothcultures. Contact dependence was determined by coculture in a transwell ap-paratus as described in SI Materials and Methods. Viable inhibitor and targetcells were enumerated as colony-forming units at the beginning and endof coculture by plating on selective media. All competition data are ex-pressed as the competitive index (CI) of target cells:CI= ½CFUtargets

t=n =CFUinhibitorst=n �=

½CFUtargetst=0 =CFUinhibitors

t=0 �.

Microscopy. Inhibitor and GFP-labeled target cells were cocultured at a 1:1cell ratio for 12 h. Cells were fixed in 4% (vol/vol) formaldehyde for 12 h,quenched with 125 mM glycine (pH 7.5) for 15 min, and washed with 1×PBS. Bacteria were plated onto glass slides coated with poly-D-lysine andwere stained with DAPI supplemented with 0.1% Triton X-100. Imageswere collected and processed with the GIMP imaging suite as described inSI Materials and Methods. Anucleate target cells were quantified by visualinspection of at least 300 target cells from two independent cocultures(∼150 target cells from each biological replicate). Cells were scored asanucleate if they lacked visible DAPI staining.

RNA Analyses. WapA-CT toxins were expressed in E. coli cells from the arabi-nose-inducible PBAD promoter. Cells were frozen at −80 °C and extracted withguanidinium isothiocyanate-phenol to isolate total RNA. tRNAs were analyzedby Northern blot hybridization and S1 nuclease protection as described(14, 36). Oligonucleotide probes (Table S3) were radiolabeled with [32P] fornorthern and S1 protection analyses as described in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank Steve Lindow for pPROBE plasmids andAmy Charkowski for strains and discussions. This work was supported byGrant 0642052 from the National Science Foundation (to D.A.L.) and byNational Institutes of Health Grants U54 AI065359 (to D.A.L.) and R01GM078634 (to C.S.H.). S.K. is supported by fellowships from the Carl Tryggersand Wenner-Gren Foundations.

1. Lin RJ, Capage M, Hill CW (1984) A repetitive DNA sequence, rhs, responsible forduplications within the Escherichia coli K-12 chromosome. J Mol Biol 177(1):1–18.

2. Hill CW, Sandt CH, Vlazny DA (1994) Rhs elements of Escherichia coli: A family of geneticcomposites each encoding a large mosaic protein. Mol Microbiol 12(6):865–871.

3. Jackson AP, Thomas GH, Parkhill J, Thomson NR (2009) Evolutionary diversification ofan ancient gene family (rhs) through C-terminal displacement. BMC Genomics 10:584.

4. Zhao S, Hill CW (1995) Reshuffling of Rhs components to create a new element.J Bacteriol 177(5):1393–1398.

5. Liu K, Knabel SJ, Dudley EG (2009) rhs genes are potential markers for multilocussequence typing of Escherichia coli O157:H7 strains. Appl Environ Microbiol 75(18):5853–5862.

6. Tucker RP, Beckmann J, Leachman NT, Schöler J, Chiquet-Ehrismann R (2012) Phylo-genetic analysis of the teneurins: Conserved features and premetazoan ancestry. MolBiol Evol 29(3):1019–1029.

7. Minet AD, Rubin BP, Tucker RP, Baumgartner S, Chiquet-Ehrismann R (1999) Teneurin-1, a vertebrate homologue of the Drosophila pair-rule gene ten-m, is a neuronalprotein with a novel type of heparin-binding domain. J Cell Sci 112(Pt 12):2019–2032.

8. Foster SJ (1993) Molecular analysis of three major wall-associated proteins of Bacillussubtilis 168: Evidence for processing of the product of a gene encoding a 258 kDaprecursor two-domain ligand-binding protein. Mol Microbiol 8(2):299–310.

9. Petersen L, Bollback JP, Dimmic M, Hubisz M, Nielsen R (2007) Genes under positiveselection in Escherichia coli. Genome Res 17(9):1336–1343.

10. Poole SJ, et al. (2011) Identification of functional toxin/immunity genes linked tocontact-dependent growth inhibition (CDI) and rearrangement hotspot (Rhs) systems.PLoS Genet 7(8):e1002217.

11. Aoki SK, et al. (2010) A widespread family of polymorphic contact-dependent toxindelivery systems in bacteria. Nature 468(7322):439–442.

12. Aoki SK, et al. (2005) Contact-dependent inhibition of growth in Escherichia coli.Science 309(5738):1245–1248.

13. Aoki SK, et al. (2008) Contact-dependent growth inhibition requires the essentialouter membrane protein BamA (YaeT) as the receptor and the inner membranetransport protein AcrB. Mol Microbiol 70(2):323–340.

14. Diner EJ, Beck CM, Webb JS, Low DA, Hayes CS (2012) Identification of a target cellpermissive factor required for contact-dependent growth inhibition (CDI). Genes Dev26(5):515–525.

15. McGinness KE, Baker TA, Sauer RT (2006) Engineering controllable protein degra-dation. Mol Cell 22(5):701–707.

16. Silverman JM, Brunet YR, Cascales E, Mougous JD (2012) Structure and regulation ofthe type VI secretion system. Annu Rev Microbiol 66:453–472.

17. Vlazny DA, Hill CW (1995) A stationary-phase-dependent viability block governed bytwo different polypeptides from the RhsA genetic element of Escherichia coli K-12.J Bacteriol 177(8):2209–2213.

18. Aggarwal K, Lee KH (2011) Overexpression of cloned RhsA sequences perturbs thecellular translational machinery in Escherichia coli. J Bacteriol 193(18):4869–4880.

19. Sisto A, et al. (2010) An Rhs-like genetic element is involved in bacteriocin productionby Pseudomonas savastanoi pv. savastanoi. Antonie van Leeuwenhoek 98(4):505–517.

20. Iyer LM, Zhang D, Rogozin IB, Aravind L (2011) Evolution of the deaminase fold andmultiple origins of eukaryotic editing and mutagenic nucleic acid deaminases frombacterial toxin systems. Nucleic Acids Res 39(22):9473–9497.

21. Hood RD, et al. (2010) A type VI secretion system of Pseudomonas aeruginosa targetsa toxin to bacteria. Cell Host Microbe 7(1):25–37.

22. MacIntyre DL, Miyata ST, Kitaoka M, Pukatzki S (2010) The Vibrio cholerae type VIsecretion system displays antimicrobial properties. Proc Natl Acad Sci USA 107(45):19520–19524.

23. Zheng J, Ho B, Mekalanos JJ (2011) Genetic analysis of anti-amoebae and anti-bacterial activities of the type VI secretion system in Vibrio cholerae. PLoS ONE 6(8):e23876.

24. Cascales E, et al. (2007) Colicin biology. Microbiol Mol Biol Rev 71(1):158–229.25. Waterfield NR, Bowen DJ, Fetherston JD, Perry RD, ffrench-Constant RH (2001) The tc

genes of Photorhabdus: A growing family. Trends Microbiol 9(4):185–191.26. Hurst MR, Glare TR, Jackson TA, Ronson CW (2000) Plasmid-located pathogenicity

determinants of Serratia entomophila, the causal agent of amber disease of grassgrub, show similarity to the insecticidal toxins of Photorhabdus luminescens. J Bac-teriol 182(18):5127–5138.

27. Kung VL, et al. (2012) An rhs gene of Pseudomonas aeruginosa encodes a virulenceprotein that activates the inflammasome. Proc Natl Acad Sci USA 109(4):1275–1280.

28. Youderian P, Hartzell PL (2007) Triple mutants uncover three new genes required forsocial motility in Myxococcus xanthus. Genetics 177(1):557–566.

29. Anderson MS, Garcia EC, Cotter PA (2012) The Burkholderia bcpAIOB genes defineunique classes of two-partner secretion and contact dependent growth inhibitionsystems. PLoS Genet 8(8):e1002877.

30. Hong W, Mosca TJ, Luo L (2012) Teneurins instruct synaptic partner matching in anolfactory map. Nature 484(7393):201–207.

31. Mosca TJ, Hong W, Dani VS, Favaloro V, Luo L (2012) Trans-synaptic Teneurin sig-nalling in neuromuscular synapse organization and target choice. Nature 484(7393):237–241.

32. Kenzelmann D, Chiquet-Ehrismann R, Tucker RP (2007) Teneurins, a transmembraneprotein family involved in cell communication during neuronal development. Cell MolLife Sci 64(12):1452–1456.

33. Edwards RA, Keller LH, Schifferli DM (1998) Improved allelic exchange vectors andtheir use to analyze 987P fimbria gene expression. Gene 207(2):149–157.

34. Choi KH, et al. (2005) A Tn7-based broad-range bacterial cloning and expressionsystem. Nat Methods 2(6):443–448.

35. Bron S, et al. (1998) Protein secretion and possible roles for multiple signal peptidasesfor precursor processing in bacilli. J Biotechnol 64(1):3–13.

36. Hayes CS, Sauer RT (2003) Cleavage of the A site mRNA codon during ribosomepausing provides a mechanism for translational quality control. Mol Cell 12(4):903–911.

Koskiniemi et al. PNAS | April 23, 2013 | vol. 110 | no. 17 | 7037

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Dec

embe

r 9,

202

0