Corrections

PHYSICSCorrection for “Distortions and stabilization of simple-cubiccalcium at high pressure and low temperature,” byWendy L.Mao,Lin Wang, Yang Ding, Wenge Yang, Wenjun Liu, Duck YoungKim, Wei Luo, Rajeev Ahuja, Yue Meng, Stas Sinogeikin, JinfuShu, and Ho-kwang Mao, which appeared in issue 22, June 1,2010, of Proc Natl Acad Sci USA (107:9965–9968; first publishedMay 17, 2010; 10.1073/pnas.1005279107).The authors note that due to a printer’s error, the affiliation

information for Wendy L. Mao was incorrect. The correct in-stitution name is SLAC.The authors would also like to note that Fig. 5 was incorrect as

shown. The corrected figure and its legend appear below.

www.pnas.org/cgi/doi/10.1073/pnas.1007813107

MICROBIOLOGYCorrection for “Chemical sensing in mammalian host-bacterialcommensal associations,” by David T. Hughes, Darya A. Terek-hova, Linda Liou, Carolyn J. Hovde, Jason W. Sahl, Arati V.Patankar, Juan E. Gonzalez, Thomas S. Edrington, David A.Rasko, and Vanessa Sperandio, which appeared in issue 21, May25, 2010, ofProcNatl Acad Sci USA (107:9831–9836; first publishedMay 10, 2010; 10.1073/pnas.1002551107).The authors note that the following statement should be

added to the Acknowledgments: “V.S. was supported by Na-tional Institutes of Health Grant AI077613.”

www.pnas.org/cgi/doi/10.1073/pnas.1008458107

NEUROSCIENCECorrection for “GluN2B subunit-containing NMDA receptor an-tagonists prevent Aβ-mediated synaptic plasticity disruption invivo,” by Neng-WeiHu, Igor Klyubin, Roger Anwy, andMichael J.Rowan, which appeared in issue 48,December 1, 2009, ofProcNatlAcad Sci USA (106:20504–20509; first published November 16,2009; 10.1073/pnas.0908083106).The authors note that the author name Roger Anwy should

have appeared as Roger Anwyl. The corrected author line appearsbelow. The online version has been corrected.Neng-Wei Hua,b, Igor Klyubina,b, Roger Anwylb,c, and Michael J.

Rowana,b,1

www.pnas.org/cgi/doi/10.1073/pnas.1007564107

10

20

30

40

50

Fre

qu

ency

(m

eV)

0

Y T R S

Phonon DOS (states/meV atom)

0.1 0.2

Fig. 5. Phonon dispersion relation (Left) and density of states (DOS) (Right)for pm Ca-VI.

12734 | PNAS | July 13, 2010 | vol. 107 | no. 28 www.pnas.org

Dow

nloa

ded

by g

uest

on

Mar

ch 5

, 202

1 D

ownl

oade

d by

gue

st o

n M

arch

5, 2

021

Dow

nloa

ded

by g

uest

on

Mar

ch 5

, 202

1 D

ownl

oade

d by

gue

st o

n M

arch

5, 2

021

Dow

nloa

ded

by g

uest

on

Mar

ch 5

, 202

1 D

ownl

oade

d by

gue

st o

n M

arch

5, 2

021

Dow

nloa

ded

by g

uest

on

Mar

ch 5

, 202

1 D

ownl

oade

d by

gue

st o

n M

arch

5, 2

021

Chemical sensing in mammalian host–bacterialcommensal associationsDavid T. Hughesa,b, Darya A. Terekhovaa,b, Linda Liouc, Carolyn J. Hovdec, Jason W. Sahld, Arati V. Patankare,Juan E. Gonzaleze, Thomas S. Edringtonf, David A. Raskod, and Vanessa Sperandioa,b,1

Departments of aMicrobiology and bBiochemistry, University of Texas Southwestern Medical Center, Dallas, TX 75390; cDepartment of Microbiology,University of Idaho, Moscow, ID 83844; dInstitute for Genome Sciences, Department of Microbiology and Immunology, University of Maryland School ofMedicine, Baltimore, MD 21201; eDepartment of Molecular and Cell Biology, University of Texas, Dallas, TX 75083; and fAgricultural Research Services, USDepartment of Agriculture, College Station, TX 77845

Edited* by Stanley Falkow, Stanford University, Stanford, CA, and approved April 16, 2010 (received for review March 1, 2010)

The mammalian gastrointestinal (GI) tract is colonized by a complexconsortium of bacterial species. Bacteria engage in chemical signalingto coordinate population-wide behavior. However, it is unclear ifchemical sensingplaysa role in establishingmammalianhost–bacterialcommensal relationships. Enterohemorrhagic Escherichia coli (EHEC) isa deadly human pathogen but is a member of the GI flora in cattle, itsmain reservoir. EHEC harbors SdiA, a regulator that senses acyl-homo-serine lactones (AHLs) produced byother bacteria. Here,we show thatSdiA is necessary for EHEC colonization of cattle and that AHLs areprominent within the bovine rumen but absent in other areas of theGI tract. We also assessed the rumen metagenome of heifers, and weshow that it is dominated by Clostridia and/or Bacilli but also harborsBacteroidetes.Ofnote, somemembersof theBacteroidetesphylahavebeen previously reported to produce AHLs. SdiA-AHL chemical signal-ing aids EHEC in gauging these GI environments, and promotes adap-tation to a commensal lifestyle. We show that chemical sensing in themammalianGI tractdetermines thenichespecificity for colonizationbya commensal bacterium of its natural animal reservoir. Chemical sens-ingmay be a generalmechanismused by commensal bacteria to senseand adapt to their mammalian hosts. Additionally, because EHEC islargely prevalent in cattle herds, interference with SdiA-mediated cat-tle colonization is an exciting alternative to diminish contamination ofmeat products and cross-contamination of produce crops because ofcattle shedding of this human pathogen.

bovine | enterohemorrhagic Escherichia coli | metagenomics | rumen

Bacteria thrive in complex multispecies communities within thegastrointestinal (GI) tracts of mammals (1). Mammals and

bacteria have amicable and detrimental interactions, and enter-ohemorrhagicEscherichia coli (EHEC) can behave as a commensalor a pathogen depending on its host. EHEC is a commensal in theGI tract of cattle, its main reservoir, but is a human pathogen (2).EHECcauses bloody diarrhea, and it colonizes the large intestine ofhumans to formattachingandeffacing (AE) lesions that are thoughtto be largely responsible for promoting disease (2). The genes forAE lesion formation are encoded within the locus of enterocyteeffacement (LEE) (2). The LEE andAE lesion formations are alsonecessary for EHEC colonization of the recto-anal junction (RAJ)of cattle, facilitating its shedding to the environment (3–5).WhereasAE lesion formation leads to disease in humans, it is innocuous inadult cattle. The site of AE lesion formation in the GI tract may beresponsible for the different outcomes.AE lesions occur in the largeintestine in humans, leading to diarrheal disease,whereasAE lesionformation in the RAJ does not compromise the electrolyte balancein the bovine GI tract (2, 6).To adapt to different hosts and environments, EHEC exploits

chemical signaling (7). Quorum sensing (QS) is a signaling mecha-nism that allows bacteria to respond to chemicals by altering geneexpression. EHEC uses several QS systems for intercellular signaling(7), including the autoinducer-3 (AI-3)/epinephrine/norepinephrine(8) system and the LuxR homolog SdiA that senses acyl-homoserinelactones (AHLs) (9–12).AHLshave a conservedhomoserine lactone

connected to a variable acyl chain. Different acyl chains ensure thatdifferent AHLs will be recognized by different LuxR-type proteins.LuxR-type proteins are transcription factors that regulate transcrip-tion of their target geneswhenbinding toAHL.AHLbinding tomostof these proteins stabilizes them; otherwise, in the absence of signal,they are targeted to degradation (13–16). Congruent with AHLsbeingusedas folding switchesby theseproteins, theNMRstructureofSdiA shows that SdiA requires binding of this signal for folding andfunction (16). EHECencodes SdiAbut does not contain a luxI (AHLsynthase) gene, and it does not produceAHLs (11, 12, 17). TheLuxRsignaling system has been primarily associated with intraspecies sig-naling, but there are examples of LuxR-type proteins (such as SdiA)that are primarily involved in interspecies signaling (9–12). Al-though the AI-3/epinephrine/norepinephrine QS system is impor-tant to activate EHEC virulence in animal models of pathogenesis(18), the role of the SdiA system inEHEC–host associations has notbeen established.Here, we report that SdiA-AHL signaling regulates expression

of EHEC genes known to facilitate the commensal EHEC colo-nization of cattle. We show that AHLs are present within the bo-vine rumen but absent in other areas of theGI tract and that rumenAHLs through SdiA modulate expression of these EHEC genes.

Results and DiscussionSdiA-AHL Represses Transcription of the LEE Genes. Transcriptomestudies were conducted to investigate SdiA-AHL signaling inEHEC. SdiA-AHL signaling altered the expression of 49 genes.Within these 49 genes, the LEE and glutamate decarboxylase(gad) acid-resistance system are present. Both the LEE and gadsystem have been shown to be essential for EHEC colonizationof cattle (4, 19, 20). There was no difference in transcription ofthe LEE genes between wild type (WT) and ΔsdiA in the ab-sence of AHLs (EHEC produces no endogenous AHLs). How-ever, when oxo-C6-homoserine lactone was added, transcriptionof the LEE genes was decreased in the WT strain but not inΔsdiA (Fig. 1B and Fig. S1). These results suggested that AHLsrepress transcription of the LEE genes and that this repression ismediated through SdiA. These results were also consistent withthe role of AHLs in stabilizing the SdiA protein, given that, likeTraR, SdiA will misfold and be targeted for degradation in the

Author contributions: D.T.H., D.A.T., C.J.H., D.A.R., andV.S. designed research; D.T.H., D.A.T.,L.L., C.J.H., J.W.S., A.V.P., and D.A.R. performed research; D.A.T., C.J.H., J.E.G., T.S.E., D.A.R.,and V.S. contributed new reagents/analytic tools; D.T.H., D.A.T., L.L., C.J.H., J.W.S., J.E.G.,D.A.R., and V.S. analyzed data; and D.T.H. and V.S. wrote the paper.

The authors declare no conflict of interest.

Data deposition: This sequence reported in this paper has been deposited in the NationalCenter of Biotechnology Information Gene Expression Omnibus database (accession no.GAE13562).

*This Direct Submission article had a prearranged editor.1Towhom correspondence should be addressed. E-mail: vanessa.sperandio@utsouthwestern.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1002551107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1002551107 PNAS | May 25, 2010 | vol. 107 | no. 21 | 9831–9836

MICRO

BIOLO

GY

absence of AHLs (16). Transcription of the LEE genes was re-pressed in the complemented strain both in the presence andabsence of AHLs (Fig. 1B), suggesting that enough SdiA isproduced to overcome the lack of the AHL through even mildplasmid overexpression (pACYC177 is a low copy-number vec-tor). Of note, these results are consistent with a report that ex-pression of SdiA from a high copy-number plasmid in EHECcaused reduced expression of the LEE genes (21). However, nosdiA mutant was constructed and tested in this previous study(21). SdiA repressed transcription of the LEE genes by directlyrepressing transcription of LEE1 (Fig. 1C). LEE1 encodes theLEE-encoded regulator (Ler), which is required for the expres-sion of all LEE genes (22).The LEE encodes for a type III secretion system (TTSS),

a needle-like structure used to inject bacterial effectors into hostcells, as well as the translocon of this system comprised of theEspA, EspB, and EspD proteins, which are themselves secretedthrough this TTSS (Fig. 1A) (2). There was no difference in theexpression of EspA between the WT and ΔsdiA in the absence ofAHLs (Fig. 1D). However, expression of EspA was reduced inthe WT in the presence of AHL and decreased in the com-plemented strain in both the presence and absence of signal (Fig.1D). These results were consistent with the SdiA-AHL re-pression of LEE transcription (Fig. 1B). Type III secretion ofEspA on WT and ΔsdiA in the absence of AHLs was similar;however, type III secretion in the presence of AHL was strikinglyreduced in WT and complemented strains but not in the sdiAmutant (Fig. 1E). SdiA-AHL controls the expression of all LEEgenes, including the genes encoding the TTSS. Hence, one wouldexpect a more striking phenotype on type III secretion of EspA,where one observes defects in secretion coupled to lesser ex-pression of espA.

SdiA-AHL Activates Transcription of the gad Genes. Congruent withthe transcriptome studies, AHLs activated expression of the gadacid-resistance genes (Fig. 2 A and B). EHEC has other acid-resistance systems, but only the gad system is necessary forEHEC survival within the acidic stomachs of the cow (19). Ofnote, the arginine acid-resistance system (adi) was not regulatedby SdiA or the addition of AHLs (Fig. S2). Expression of the gadgenes was significantly enhanced by AHLs (Fig. 2 A and B).However, in contrast with the regulation of the LEE genes, theexpression of the gad genes was dramatically decreased in ΔsdiA,even in the absence of AHLs (Fig. 2 A and B). The observationthat the gad system can be activated by SdiA even in the absenceof AHLs is consistent with a previous report from Dyszel et al.(9). A potential explanation is that within the bacterial cell, SdiAmay have a half-life in the absence of AHLs that is sufficient toexert an activation role in the transcription of the gad genes.SdiA alone did not directly activate transcription of the gadsystem (Fig. 1C), suggesting that SdiA-AHL either requiresa second factor to directly activate transcription of gad or acti-vates expression of an unidentified transcription factor that di-rectly activates transcription of the gad system. Consequently, thesdiA mutant was less resistant to acidic environments than WTEHEC (Fig. 2C), and this diminished acid resistance was causedby the down-regulation of the gad system (Fig. 2 A and B).

AHLs Are Present in the Bovine Rumen. Because SdiA is active inthe presence of AHLs (16) and previous reports of AHLs statethat they are present within the cattle rumen (23, 24), thepresence of AHLs in rumen fluid was assessed. AHLs wereextracted from the rumen fluid of cannulated cattle and testedwith an Agrobacterium tumefaciens (25) reporter strain for theirpresence. This strain has the traI gene (activated by TraR-AHL)fused to lacZ that, in the presence of AHLs and its substrate,converts to blue on a reverse-phase TLC overlay assay. This

A

B

0

0.2

0.4

0.6

0.8

1

1.2

1.4

WT WT

+AHL

sdiA sdiA

+AHL

Cpl Cpl

+AHL

Rel

ativ

e m

RN

A e

xpre

ssio

n *

*

* p<0.0001

p is not

significant

espA (LEE4)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

WT WT

+AHL

sdiA sdiA

+AHL

Cpl Cpl

+AHL

Rel

ativ

e m

RN

A E

xpre

ssio

n

*

*

* p<0.0001

p is not

significant

ler (LEE1)

C

EspA

WT sdiA WT sdiA compl

No AHL Oxo-C6-HSL

EspA

WT sdiA WT sdiA compl

No AHL Oxo-C6-HSLE D

LEE5 promoter LEE4 promoter

SdiA (uM) 0 0.071 0.14 0.28 0.84 0 0.071 0.14 0.28 0.84

LEE5 promoter LEE4 promoter

SdiA (uM) 0 0.071 0.14 0.28 0.84 0 0.071 0.14 0.28 0.84

SdiA oxo-C6-HSL

NP

SdiA (pMol) 0 2.5 5 10 20 30 0 10 20 30 0 2.5 5 7.5 10 20 30

LEE1 -173 to +86

NP

SdiA oxo-C6-HSL

kan (Neg. cont)

NP

SdiA oxo-C6-HSL

ftsQ (Positive control)

0 2.5 5 7.5 10 20 30

NPSdiA oxo-C6-HSL

gadW

SdiA (pM)

1.8

1.6

1.4

1.2

1

0.8

0.6

0.4

0.2

0Re

lative

mR

NA

exp

ressio

n

WT WT sdiA sdiA Cpl Cpl

+AHL +AHL +AHL

Re

lative m

RN

A e

xpre

ssio

n

1.4

1.2

1

0.8

0.6

0.4

0.2

0

WT WT sdiA sdiA Cpl Cpl

+AHL +AHL +AHL

0 0.071 0.14 0.26 0.84 0 0.071 0.14 0.26 0.84

WT sdiA WT sdiA Cpl

No AHL Oxo-C6-HSL

LEE1 LEE2 LEE3 LEE5 LEE4

(ler) (escV) (escC) (tir) (espADB)

TWTW AidsAids lpClpC

LHA+LHA+LHA+

Fig. 1. SdiA-mediated AHL regulation of the EHEC LEE genes. (A) Schematic depiction of the LEE pathogenicity island. (B) Quantitative RT-PCR (qRT-PCR) ofthe LEE genes (ler/LEE1 and espA/LEE4) in WT EHEC (86-24), isogenic sdiAmutant, and complemented (Cpl) strain (ΔsdiAwith pDH6; sdiA cloned in pACYC177)grown in DMEM to an OD600 of 1.0 in the absence and presence of AHL (10 μM oxo-C6-homoserine lactone). (C) EMSAs of LEE1, LEE4, LEE5, gadW, ftsQ(positive control), and kanamycin-resistant gene, kan promoter, (negative control) with SdiA-AHL. (D) Western blots of whole-cell lysates of WT, sdiA mutant,and sdiA mutant complemented with pDH6 probed with antisera against EspA and RpoA (loading control). (E) Western blots of the secreted proteins of WT,sdiA mutant, and sdiA mutant complemented with pDH6 strains in the absence and presence of AHL (10 μM oxo-C6-HSL) probed with an antiserum againstEspA. In conditions where no AHLs were added, the same amount of the ethyl-acetate solvent was added to ensure that the solvent had no effect in geneexpression.

9832 | www.pnas.org/cgi/doi/10.1073/pnas.1002551107 Hughes et al.

detection strain was chosen, because it detects a wide range ofAHLs (25). AHLs were detected in all rumen extracts (Fig. 3A),further showing that AHLs are indeed present within the rumenof cattle. Alkalinization (alkaline pH hydrolyzes the homoserinelactone ring of AHLs and inactivates these signals) caused loss ofactivity of the rumen-extracted AHLs, confirming that thesesignals were AHLs (Fig. 3B). The activity of these signals wasrestored in the rumen sample on acidification of this reaction(Fig. 3B), which allows for reformation of the lactone ring. Al-though AHLs were detected in all rumen extracts (Fig. 3A),these signals were not detected in other portions of the ruminantGI tract, suggesting that these signals are restricted to the rumen(Fig. 3C and Fig. S3). These results are consistent with thechemistry of AHLs. The homoserine ring of these signals is hy-drolyzed at alkaline pH (26), which is the pH in the intestine;thus, it inactivates these signals. These signals, however, are

stable in acidic pH, such as the pH in the rumen (the averagerumen pH is 5.98 in animals on a grain diet) (Tables S1–S3).

Rumen AHLs Modulate Transcription of EHEC Genes. Given thatAHLs repressed expression of the LEE, activated expression ofthe gad genes (Figs. 1 and 2) and that the rumen harbors AHLs(Fig. 3A), we then tested if AHLs extracted from the rumencould mimic the effect of purified AHLs on transcription. TheAHLs extracted from the rumen repressed transcription of theLEE (Fig. 3 D and E) and activated expression of gadX (Fig. 3F).However, not all rumen AHL effects on LEE transcription oc-curred through SdiA (Fig. 3 D and E). There are two potentialexplanations for this observation: (i) the presence of anotherreceptor for AHLs in EHEC or (ii) another(s) molecule(s) isresponsible for this further repression. Indeed, de Sablet et al.(27) showed that, in gnotobiotic rats colonized with humanmicrobiota, these bacteria produce chemical signals that repressexpression of the genes encoding Shiga toxin in EHEC in a SdiA-independent manner. These studies highlight that complex mi-crobial communities may produce different combinations ofsignals. Conversely, gadX activation by rumen-extracted AHLswas completely abrogated in the sdiA mutant (Fig. 3F).Because one cannot directly assess the levels of AHLs within the

rumen fluid, one potential caveat of the experiments using rumenextracts could be that these signals in the extracts are concentratedto levels that are not physiologically relevant. To address this issue,the aiiA gene fromBacillus cereus, which encode for a lactonase thatspecifically hydrolyses the lactone ring fromAHLs (28), was clonedinto WT EHEC and the sdiA mutant. Expression of gadX in WTgrown in filtered, nonconcentrated, rumen fluid was decreased inthe presence of the aiiA-encoded lactonase (which inactivates anyAHLs present in the rumen fluid) (Fig. 3G), whereas expression ofgadX in the sdiA mutant did not change with or without this lacto-nase (Fig. 3G). Thesedata support thatAHLsat physiological levelsin the rumen fluid activate transcription of gadX and that this acti-vation is SdiA-dependent.

Rumen Metagenomic Studies. The AHLs in the rumen are prob-ably synthesized by the microbial flora that reside in this GIcompartment. However, there is limited information on thecomposition of the rumen microbial flora, and previous meta-genomic studies were conducted only in forage- (grass-legumehay) fed steers (29). To further assess the composition of therumen flora, we performed metagenomic studies on DNAextracted from the ruminant contents of eight ruminally cannu-lated heifers on a grain diet. Forage is the usual diet of cattle onfarms, whereas grain is the preferred diet on feedlots before theyare sent to the abattoir.The general sequence data obtained from this study are sum-

marized in Table S4. More than 40,000 sequences with acceptablecoverage and quality were included in further analysis with ameanof >100 per rumen sample and a range from 4,534 to 5,887 reads.Using a 97% similarity threshold value, we identified a wide rangeof operational taxonomic units (OTUs) for each sample. Therumen sample group displayed significant variability among thenumber of OTUs that could be identified (640 in sample 8 and2,576 in sample 1), suggesting that there is variability between therumen contents of each of the animals. This finding is similar tothe level of variability observed in the human gastrointestinal tract(1). The calculations on the species richness using ACE and Chao(30, 31) and diversity estimators, Shannon and Weaver (32), in-dicated that there is significant variation between each of eightrumen samples (Tables S4–S6).We also examined the most abundant phyla in the rumen

microbiota. In seven of eight samples, the Clostridia class wasdominant in the rumen microbiota, similar to the findings in therumen metagenomic studies of steers on a forage diet (29).However, in sample 4, the Bacilli were dominant with Clostridia,

Δ

Δ Δ

*

*

A

B

C

1.00E+02

1.00E+03

1.00E+04

1.00E+05

1.00E+06

0 60 120

time (minutes)

CF

U

86-24 AHL

sdiA AHL

compl

P<0.0001

3

2.5

2

1.5

1

0.5

0

Re

lative m

RN

A E

xp

ressio

n

Re

lative m

RN

A E

xp

ressio

n

- + - + - + - + - + - + - + - +WT sdiA sdiA sdiA

pACYC psdiAWT sdiA sdiA sdiA

pACYC psdiA

AHL

10μM AHL 10nM AHL

2.5

2

1.5

1

0.5

0

- + - + - + - + 10 M AHL

WT sdiA sdiA sdiApACYC psdiA

μ

Δ Δ Δ Δ Δ

Δ

Fig. 2. SdiA regulation of the gad acid-resistance system. (A) qRT-PCR of thegadW gene encoding the master regulator of the gad system in WT EHEC(86-24), sdiA mutant, sdiA mutant with pACYC177, and sdiA mutant com-plemented with pDH6 (sdiA in pACYC177) strains grown in DMEM to anOD600 of 1.0 in the absence and presence of AHL (10 μM oxo-C8-HSL and 10nM oxo-C8-HSL). (B) qRT-PCR of the gadX gene in WT EHEC (86-24), sdiAmutant, sdiA mutant with pACYC177, and sdiA mutant complemented withpDH6 (sdiA in pACYC177) strains grown in DMEM to an OD600 of 1.0 in theabsence and presence of AHL (10 μM oxo-C8-HSL). (C) Acid resistance (sur-vival in acidic pH) through the gad system in the absence and presence of 10μM oxo-C8-HSL. In conditions where no AHLs were added, the same amountof the ethyl-acetate solvent was added to ensure that the solvent had noeffect in gene expression.

Hughes et al. PNAS | May 25, 2010 | vol. 107 | no. 21 | 9833

MICRO

BIOLO

GY

the second most prevalent class. Other phyla encountered inthese rumen samples include Erysipelotrichi and Bacteroidetes. Ofnote, members of the Bacteroidetes phyla have been previouslyreported to produce AHLs (33) (Fig. 3H). The analysis on thisphylogenetic level shows that the rumen constituents are stableamong these eight samples.The variability in the number and quantity of OTU that are

observed in the rumen samples is similar to the level of variabilityobserved in the human GI tract (1). Not surprisingly, the bac-terial community that is present is different. The microbiota ofthe human colonic GI tract has been characterized by the ex-amination of the fecal material and is dominated by Bacteroidetesand Firmicutes, whereas the bovine rumen samples are domi-nated by Clostria or Bacilli, both within the Firmicutes (Fig. 3H).This difference in the microbiota is most likely linked to thedifference in environmental conditions (pH, anaerobiosis, etc.)

and tissue type. However, both bovine rumen and human colonicsamples show sample-to-sample variation. Additionally, althoughthis study represented the deep sequencing of bovine rumensamples, we still have not explored the rare microbiome. Furtherdeep sequencing of these environments will provide insight intothe population structure in these animals and may providea mechanism of preventing or interrupting the carriage of theseimportant human pathogens.The current analysis does not allow the determination of the

species of the members of the microbiota. Although we do notknow which members of the rumen microbiota are producingAHLs, there has been one species of Bacteroidetes associatedwith fish that was found to produce AHLs (33). Future studies ofthese complex communities and their interactions will furtherilluminate the role of chemical signaling within the rumen.

C6-HSL

C8-HSL

Oxo-C6-HSL

Oxo-C8-HSL

RU1 RU2 RU3

R R RALK ALK+AC

AHL AHL AHLALK ALK+AC

A B C

D

G gadX mRNA levels in the presence of AHL-lactonase

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

wt wt + aiiA sdiA sdiA + aiiA

mR

NA

co

pie

s/rp

oA

mR

NA

co

pie

s

*

n/s

*

* - p<0.05Ñ

sdiA

Ñ

sdiA + aiiAwt + aiiA

H

Rectum Rectum

SI SI

Cecum Cecum

Rectum

Colon

Positive control

Rectum Rectum

SI SI

Cecum Cecum

Rectum

Colon

Positive control

10

0

-10

-20

-30

-40

-50

-60

-70

Rela

tive m

RN

A e

xp

ressio

n

Rumen extract - + - +

WT

Ñ ÑsdiA

50

0

-50

-100

-150

-200

-250

-300

-350

-400

Rumen extract - + - +WT

Ñ

sdiAE

Rela

tive a

bundance

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

Rumen Sample

1 2 3 4 5 6 7 8

Erysipelotrichi

Clostridia

Bacilli

Unclassified Firmicutes

Bacteroidetes

Unclassified

F

Fig. 3. RumenAHLs affect expressionof EHECgenes. (A) TLC fromAHLs extracted from50mL (evaporated to 5 μL thatwereadded to theTLC) of rumenfluid (RU)collected from three different cows (RU1, RU2, and RU3). C6-HSL (402 pmol), C8-HSL (15.8 pmol), oxo-C6-HSL (400 pmol), and oxo-C8-HSL (15 pmol) were used ascontrols. (B) TLC fromAHLs extracted from 50mL (evaporated to 5 μL thatwere added to the TLC) of rumenfluid (R) that has been subjected to alkaline treatment(ALK; hydrolyses the homoserine lactone of AHLs) and acidification (AC; ALK+AC restores the homoserine lactone); oxo-C8-HSL (AHL) undergoing the sametreatments was used as a control. (C) Preparative TLC fromAHLs extracted from 50mL (evaporated to 5 μL thatwere added to the TLC) of other portions of the GIof ruminants. The positive control turns blue. (D) qRT-PCR of ler (LEE1) fromWT and sdiAmutant in the presence of rumenAHLs (5 μL extract). In conditionswhereno rumen extracts were added, the same amount of the dichloromethane solventwas added to ensure that the solvent had no effect in gene expression. (E) qRT-PCR of eae (LEE5) gene fromWT and sdiAmutant in the presence of rumen AHLs (5 μL extract). (F) qRT-PCR of gadX fromWT and sdiAmutant in the presence ofrumenAHLs (5 μL extract). (G) qRT-PCR of gadX fromWTand sdiAmutant containing either empty vector or the aiiA gene fromB. cereus (encodes a lactonase thatinactivates AHLs) cloned into pBADMYcHisA grown in filtered, nonconcentrated rumen fluid in the presence of arabinose. The presence of AHLs in this rumenfluid was previously confirmed (Fig. S3). (H) Graph depicting the bacterial composition of the rumen of eight heifers on a grain diet.

9834 | www.pnas.org/cgi/doi/10.1073/pnas.1002551107 Hughes et al.

sdiA Mutant Is Defective for Colonization of Cattle. Inasmuch asrumen AHLs in a SdiA-dependent fashion modulate expressionof EHEC genes necessary for cattle colonization, we assessed thecontribution of SdiA to EHEC survival within the rumen andsubsequent colonization of the RAJ. For this purpose, a compe-tition trial was done in ruminally cannulated heifers on a graindiet (we used the same eight animals in which the rumen pH wasassessed and the metagenomic studies were performed). Com-petition studies were performed to avoid issues concerning in-dividual variation between different heifers in scoring the abilityof these strains to establish themselves in the GI tract. As aninitial control, an in vitro competition experiment between theWT and the sdiA mutant was performed to ensure that therewere no growth defects in the sdiAmutant (Fig. 4A). The ratio ofWT to mutant bacteria was determined in the inoculum, rumen,and RAJ mucosal swab (RAMS) samples. The competitive indexfor each day was determined by dividing the ratio of mutant toWT bacteria in the rumen and RAJ. Fig. 4 B and C shows thatWT EHEC outcompeted the sdiA mutant in both the rumen andRAJ of cattle, confirming that SdiA is important for cattle col-onization (further details on these infection studies can be foundin SI Results and Discussion).This study shows that AHLs present in the rumen down-

regulate expression of the LEE genes whose expression in this GIcompartment would constitute a superfluous expenditure of en-ergy (Fig. 3D andE). However, expression of the LEE is necessaryfor colonization of the RAJ (20), where AHLs are absent (Fig.3C), allowing for LEE expression and efficient AE lesion forma-tion. Conversely, rumenAHLs activate expression of the gad acid-resistant system (Fig. 3 F and G) necessary for survival within thebovine rumen and the subsequent acidic stomachs (Fig. 4D).

ConclusionsIn this study, we show that the rumen of cattle harbor AHLs andthat these chemical signals can be sensed in part through SdiA to

modulate gene expression in EHEC, leading to successful colo-nization of these animals. However, EHEC also uses otherchemical signaling systems to modulate expression of its viru-lence and colonization genes (8, 34). In contrast to the SdiA-AHL system, the AI-3/epinephrine/norepinephrine system acti-vates virulence traits in EHEC, and in animal infection modelsthought to mimic pathogenic interactions, such as the rabbitinfant model (18), mutants that cannot sense these chemicals areattenuated for disease. QS contribution to the pathogenic and/orcommensal lifestyle of EHEC may constitute an example of yin-and-yang relationship in chemical signaling in bacteria.EHEC is carried by an estimated 70–80%of the cattle herds in the

United States, and thus, interference with AHL signaling in cattlecolonizationmay engender a strategic alternative to diminishEHECshedding and consequently, human disease. Chemical sensing maybe an important mechanism used by commensal bacteria to senseand adapt to specific niches in a complex host environment.

Materials and MethodsDetailed materials and methods describing microbiological examination,molecular biology techniques, AHL extraction from rumen samples, micro-array and metagenomic studies, and animal infection studies can be found inSI Materials and Methods. Detailed strains and plasmid constructions can befound in SI Materials and Methods.

ACKNOWLEDGMENTS.Wethank J.Kaper for anti-EspAantisera.Wealso thankS.McKnight, L. Hooper,M. V. Norgard, and E.Olsen for reading themanuscript.The contents of this work are solely the responsibility of the authors and do notrepresent the official views of the National Institutes of Health National In-stitute of Allergy and Infectious Diseases. This work was supported by NationalInstitutes of Health Grant AI053067, the Burroughs Wellcome Fund, and theNational Cattlemen’s Beef Association. D.T.H. was supported through NationalInstitutes of Health Training Grant 5-T32-AI007520-07. We would like to thankLonie Austin for cattle care and handling. This work was supported, in part, bythe IdahoAgriculture Experiment Station and PHSgrant P20-RR16454 (to C.J.H.)from the NIH NCRR.

AHLspH5.98

Rumen

1 h 4 h day 1 day 2 day 30.00

0.25

0.50

0.75

1.00

Time post-inoculation

Co

mp

etit

ive

ind

ex:

sdiA

/wt

RAJ

day 1 day 2 day 3 day 5 day 80.00

0.25

0.50

0.75

1.00

1.25

Day post-inoculation

Co

mp

etit

ive

ind

ex:

sdiA

/wt

A B

C D

* *

* * * * *

*p<0.05

*P<0.05

Fig. 4. Competition of WT and sdiA mutant in the bovine rumen and RAJ. Eight 1.5-year-old fully ruminant and ruminally cannulated Charolais heifers wereinoculated with equal cfus of WT and sdiA mutant. (A) As a control, equal cfus of WT and sdiA were grown in coculture in vitro, and their competitive indexeswere determined throughout growth (early, mid, late, and stationary phases of growth). A competitive index of 1 means no difference, a competitive index <1means that WT was in higher numbers than the sdiAmutant, and a competitive index >1 means that the mutant was in higher numbers than WT. The ratio ofWT tomutant bacteria was determined in the (B) rumen and (C) RAJ. (D) Schematic model of SdiA-AHL–dependent EHEC gene expression in the bovine GI tract.

Hughes et al. PNAS | May 25, 2010 | vol. 107 | no. 21 | 9835

MICRO

BIOLO

GY

1. Gill SR, et al. (2006) Metagenomic analysis of the human distal gut microbiome.Science 312:1355–1359.

2. Kaper JB, Nataro JP, Mobley HL (2004) Pathogenic Escherichia coli. Nat Rev Microbiol2:123–140.

3. Sheng H, Lim JY, Watkins MK, Minnich SA, Hovde CJ (2008) Characterization of anEscherichia coli O157:H7 O-antigen deletion mutant and effect of the deletion onbacterial persistence in the mouse intestine and colonization at the bovine terminalrectal mucosa. Appl Environ Microbiol 74:5015–5022.

4. Sheng H, Lim JY, Knecht HJ, Li J, Hovde CJ (2006) Role of Escherichia coli O157:H7virulence factors in colonization at the bovine terminal rectal mucosa. Infect Immun74:4685–4693.

5. Naylor SW, et al. (2003) Lymphoid follicle-dense mucosa at the terminal rectum is theprincipal site of colonization of enterohemorrhagic Escherichia coli O157:H7 in thebovine host. Infect Immun 71:1505–1512.

6. Naylor SW, et al. (2007) Impact of the direct application of therapeutic agents to theterminal recta of experimentally colonized calves on Escherichia coli O157:H7shedding. Appl Environ Microbiol 73:1493–1500.

7. Hughes DT, Sperandio V (2008) Inter-kingdom signalling: Communication betweenbacteria and their hosts. Nat Rev Microbiol 6:111–120.

8. Sperandio V, Torres AG, Jarvis B, Nataro JP, Kaper JB (2003) Bacteria-host communication:The language of hormones. Proc Natl Acad Sci USA 100:8951–8956.

9. Dyszel JL, et al. (2010) Salmonella enterica serovar Typhimurium can detect acylhomoserine lactone production by Yersinia enterocolitica in mice. J Bacteriol 192:29–37.

10. Smith JN, et al. (2008) SdiA, an N-acylhomoserine lactone receptor, becomes activeduring the transit of Salmonella enterica through the gastrointestinal tract of turtles.PLoS ONE 3:e2826.

11. Ahmer BM (2004) Cell-to-cell signalling in Escherichia coli and Salmonella enterica.Mol Microbiol 52:933–945.

12. Michael B, Smith JN, Swift S, Heffron F, Ahmer BM (2001) SdiA of Salmonella entericais a LuxR homolog that detects mixed microbial communities. J Bacteriol 183:5733–5742.

13. Zhu J, Winans SC (1999) Autoinducer binding by the quorum-sensing regulator TraRincreases affinity for target promoters in vitro and decreases TraR turnover rates inwhole cells. Proc Natl Acad Sci USA 96:4832–4837.

14. Zhu J, Winans SC (2001) The quorum-sensing transcriptional regulator TraR requires itscognate signaling ligand for protein folding, protease resistance, and dimerization. ProcNatl Acad Sci USA 98:1507–1512.

15. Zhang RG, et al. (2002) Structure of a bacterial quorum-sensing transcription factorcomplexed with pheromone and DNA. Nature 417:971–974.

16. Yao Y, et al. (2006) Structure of the Escherichia coli quorum sensing protein SdiA:Activation of the folding switch by acyl homoserine lactones. J Mol Biol 355:262–273.

17. Huisman GW, Kolter R (1994) Sensing starvation: A homoserine lactone—dependentsignaling pathway in Escherichia coli. Science 265:537–539.

18. Rasko DA, et al. (2008) Targeting QseC signaling and virulence for antibioticdevelopment. Science 321:1078–1080.

19. Price SB, Wright JC, DeGraves FJ, Castanie-Cornet MP, Foster JW (2004) Acid resistance

systems required for survival of Escherichia coli O157:H7 in the bovine gastrointestinal

tract and in apple cider are different. Appl Environ Microbiol 70:4792–4799.20. Dziva F, van Diemen PM, Stevens MP, Smith AJ, Wallis TS (2004) Identification of

Escherichia coli O157:H7 genes influencing colonization of the bovine gastrointestinal

tract using signature-tagged mutagenesis. Microbiology 150:3631–3645.21. Kanamaru K, Kanamaru K, Tatsuno I, Tobe T, Sasakawa C (2000) SdiA, an Escherichia

coli homologue of quorum-sensing regulators, controls the expression of virulence

factors in enterohaemorrhagic Escherichia coli O157:H7. Mol Microbiol 38:805–816.22. Mellies JL, Elliott SJ, Sperandio V, Donnenberg MS, Kaper JB (1999) The Per regulon of

enteropathogenic Escherichia coli: Identification of a regulatory cascade and a novel

transcriptional activator, the locus of enterocyte effacement (LEE)-encoded regulator

(Ler). Mol Microbiol 33:296–306.23. Erickson DL, et al. (2002) Evidence of quorum sensing in the rumen ecosystem:

Detection of N-acyl homoserine lactone autoinducers in ruminal contents. Can J

Microbiol 48:374–378.24. Edrington TS, et al. (2008) Acyl-homoserine-lactone autoinducer in the gastrointesinal

tract of feedlot cattle and correlation to season, E. Coli O157:H7 prevalence, and diet.

Curr Microbiol 58:227–232.25. Fuqua C, Winans SC (1996) Conserved cis-acting promoter elements are required for

density-dependent transcription of Agrobacterium tumefaciens conjugal transfer

genes. J Bacteriol 178:435–440.26. Dunny GM, Winans SC (1999) Cell-Cell Signaling in Bacteria (American Society for

Microbiology, Washington, DC).27. de Sablet T, et al. (2009) Human microbiota-secreted factors inhibit shiga toxin

synthesis by enterohemorrhagic Escherichia coli O157:H7. Infect Immun 77:783–790.28. Dong YH, Xu JL, Li XZ, Zhang LH (2000) AiiA, an enzyme that inactivates the

acylhomoserine lactone quorum-sensing signal and attenuates the virulence of

Erwinia carotovora. Proc Natl Acad Sci USA 97:3526–3531.29. Brulc JM, et al. (2009) Gene-centric metagenomics of the fiber-adherent bovine

rumen microbiome reveals forage specific glycoside hydrolases. Proc Natl Acad Sci

USA 106:1948–1953.30. Chao A (1984) Non-parametric estimation of the number of classes in a population.

Scand J Stat 11:265–270.31. Chao A (1987) Estimating the population size for capture-recapture data with

unequal catchability. Biometrics 43:783–791.32. Shannon CE, Weaver W (1949) The Mathematical Theory of Communication (University

of Illinois Press, Urbana, IL).33. RomeroM, Avendano-Herrera R, Magarinos B, Camara M, Otero A (2010) Acylhomoserine

lactone production and degradation by the fish pathogen Tenacibaculum maritimum,

a member of the Cytophaga-Flavobacterium-Bacteroides (CFB) group. FEMSMicrobiol Lett

304:131–139.34. Rasko DA, et al. (2008) Targeting QseC signaling and virulence for antibiotic

development. Science 321:1078–1080.

9836 | www.pnas.org/cgi/doi/10.1073/pnas.1002551107 Hughes et al.