Carbohydrate utilization by enterohaemorrhagic Escherichia coli O157:H7 in bovine intestinal content

Carbohydrate utilization by enterohaemorrhagicEscherichia coli O157:H7 in bovine intestinal content

Yolande Bertin,1* Frédérique Chaucheyras-Durand,2

Catherine Robbe-Masselot,3 Alexandra Durand,1

Anne de la Foye,4 Josée Harel,5 Paul S. Cohen,6

Tyrell Conway,7 Evelyne Forano1 andChristine Martin1

1INRA, UR454 Microbiologie, 63122Saint-Genès-Champanelle, France.2Lallemand Animal Nutrition, 19 rue des Briquetiers,31702 Blagnac, France.3Unité de Glycobiologie Structurale et Fonctionnelle, IFR147, Université des Sciences et Technologies de Lille,59655 Villeneuve d’Ascq, France.4INRA, Plate-Forme d’Exploration du Métabolisme,F-63122 Saint-Genès-Champanelle, France.5Groupe de Recherche sur les Maladies Infectieuses duPorc, Université de Montréal, Faculté de médecinevétérinaire, C.P. 5000, Saint-Hyacynthe, Québec,Canada J2S 7C6.6Department of Cell and Molecular Biology, University ofRhode Island, Kingston, RI 02881, USA.7Department of Botany and Microbiology, University ofOklahoma, Norman, OK 73019-0245, USA.

Summary

The bovine gastrointestinal (GI) tract is the mainreservoir for enterohaemorrhagic Escherichia coli(EHEC) responsible for food-borne infections. Char-acterization of nutrients preferentially used by EHECin the bovine intestine would help to develop ecologi-cal strategies to reduce EHEC carriage. However, thecarbon sources that support the growth of EHEC inthe bovine intestine are poorly documented. In thisstudy, a very low concentration of glucose, the mostabundant monomer included in the cattle dietarypolysaccharides, was detected in bovine small intes-tine contents (BSIC) collected from healthy cows atthe slaughterhouse. Six carbohydrates reported to beincluded in the mucus layer covering the enterocytes[galactose, N-acetyl-glucosamine (GlcNAc), N-acetyl-

galactosamine (GalNAc), fucose, mannose andN-acetyl neuraminic acid (Neu5Ac)] have been quan-tified for the first time in BSIC and accounted fora total concentration of 4.2 mM carbohydrates. Thegenes required for enzymatic degradation of the sixmucus-derived carbohydrates are highly expressedduring the exponential growth of the EHEC strainO157:H7 EDL933 in BSIC and are more stronglyinduced in EHEC than in bovine commensal E. coli. Inaddition, EDL933 consumed the free monosaccha-rides present in the BSIC more rapidly than the resi-dent microbiota and commensal E. coli, indicating acompetitive ability of EHEC to catabolize mucus-derived carbohydrates in the bovine gut. Mutations ofEDL933 genes required for the catabolism of eachof these sugars have been constructed, and growthcompetitions of the mutants with the wild-type strainclearly demonstrated that mannose, GlcNAc, Neu5Acand galactose catabolism confers a high competitivegrowth advantage to EHEC in BSIC and probably rep-resents an ecological niche for EHEC strains in thebovine small intestine. The utilization of these mucus-derived monosaccharides by EDL933 is apparentlyrequired for rapid growth of EHEC in BSIC, and formaintaining a competitive growth rate as comparedwith that of commensal E. coli. The results suggest astrategy for O157:H7 E. coli survival in the bovineintestine, whereby EHEC rapidly consumes mucus-derived carbohydrates that are poorly consumed bybacteria belonging to the resident intestinal micro-biota, including commensal E. coli.

Introduction

Enterohaemorrhagic Escherichia coli (EHEC) strains areShiga-toxin-producing E. coli (STEC) that cause humanillnesses ranging from uncomplicated diarrhoea to haem-orrhagic colitis (HC) and haemolytic-uraemic syndrome(HUS) (Law, 2000). The gastrointestinal (GI) tract of cattleand other ruminants is the principal reservoir of EHECstrains and outbreaks have been associated with directcontact with the farm environment, and with the consump-tion of meat, dairy products, water and fruits or vegetablescontaminated with ruminant manure (Cieslak et al., 1993;O’Brien et al., 2001; Yatsuyanagi et al., 2002; Caprioliet al., 2005; Muniesa et al., 2006). It is important to

Received 16 July, 2012; revised 1 October, 2012; accepted2 October, 2012. *For correspondence. E-mail [email protected], Tel. (+33) 04 73 62 45 36; Fax (+33) 04 73 6245 81.

bs_bs_banner

Environmental Microbiology (2012) doi:10.1111/1462-2920.12019

© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd

mailto:[email protected]

mailto:[email protected]

determine the mechanisms underlying EHEC persistencein the bovine intestine in order to develop nutritional orecological strategies to reduce EHEC survival in the GItract and thus limit contamination of food products. There-fore, understanding of EHEC physiology in the ruminantgut is critical for limiting EHEC shedding.

In the rumen, growth of EHEC O157:H7 is limited bythe resident microbiota and strictly anaerobic conditions(de Vaux et al., 2002; Chaucheyras-Durand et al., 2006;2010). However, O157:H7 E. coli strains may survivepassage through the acid barrier of the abomasum to thesmall intestine, which constitutes a more favourable envi-ronment than rumen contents for bacterial growth (Diez-Gonzalez et al., 1998; de Vaux et al., 2002; Chaucheyras-Durand et al., 2010). Although the recto-anal junction isconsidered as the primary E. coli O157:H7 colonizationsite for persistent shedding in cattle (Naylor et al., 2005;Lim et al., 2007; Chase-Topping et al., 2008), the smallintestine and the proximal colon are also minor sites ofEHEC carriage (Baines et al., 2008; Nart et al., 2008).Furthermore, a recent report showed that STEC O157strains are distributed along the entire GI tract of naturallyshedding cattle (Keen et al., 2010). However, little isknown about the nutrients preferentially used by EHEC orthe metabolic pathways required for persistence andgrowth in the bovine GI tract. Snider et al. demonstratedthat fucose is a critical carbon source for maintenance ofEHEC in the bovine rectum (Snider et al., 2009) and invivo colonization experiments showed that the genesagaB and dctA coding for the specific transport of GlcNAcand C4-dicarboxylic acids respectively influence coloniza-tion of the bovine gut by EHEC (Dziva et al., 2004). Morerecently, we demonstrated that ethanolamine is an impor-tant nitrogen source for EHEC in the bovine small intes-tine content (BSIC) and favours EHEC persistence at thissite (Bertin et al., 2011).

The gastrointestinal epithelium is covered by a mucusgel layer (MGL) synthesized and secreted by host gobletcells. The MGL is an integral structural component of themammal intestine, acting as a medium for protection andtransport between the luminal content and the epitheliumlining (Deplancke and Gaskins, 2001). The major functionof the MGL is to lubricate and to protect the intestinalepithelium from damage caused by food and digestivesecretions (Deplancke and Gaskins, 2001). The MGL alsoacts as a trap for microorganisms, including pathogens,preventing their access to the epithelia (Johansson et al.,2011). Mucin, the main constituent of mucus, is a filamen-tous glycoprotein constituted of high-molecular-weightsubunits, each subunit being rich in oligosaccharideslinked to serine or threonine residues. In human,N-acetylglucosamine (GlcNAc), N-acetylgalactosamine(GalNAc), fucose and galactose are the four primarymucin oligosaccharides that are often terminated with

sialic acid or sulfate groups (Deplancke and Gaskins,2001). Interestingly, in the intestine, mucin can constitutea direct source of carbohydrates constantly released inthe luminal content, and can offer numerous ecologicaladvantages to intestinal bacteria (Deplancke andGaskins, 2001). E. coli is limited to growth on mono- ordisaccharides and cannot degrade the complex polysac-charides constituting mucin (Hoskins et al., 1985).However, resident anaerobes in the bovine gut are able todegrade mucin and to release carbohydrate monosaccha-rides into the luminal content (Png et al., 2010).

Infection of streptomycin-treated mice is a usefulanimal model for studying colonization of the mammalgut by enteric bacteria (Wadolkowski et al., 1990). Thismodel has been extensively used to study EHECmetabolism during gut colonization (Miranda et al., 2004;Fabich et al., 2008; Leatham et al., 2009). EHECO157:H7 has been shown to be present both in themucus layer that overlies the mouse intestinal epitheliumand closely associated with epithelial cells (Mirandaet al., 2004). Bacterial competition experiments haveshown that the EHEC strain EDL933 uses seven sugarsknown to be present in the mouse caecal mucus(arabinose, fucose, GlcNAc, galactose, hexuronates,mannose and ribose) (Fabich et al., 2008). Importantly,E. coli EDL933 appears to consume four sugars (galac-tose, hexuronates, mannose and ribose) that are notused by the non-pathogenic E. coli MG1655, suggestinga strategy for EHEC to invade the mouse intestine andcolonize in the presence of commensal E. coli (Fabichet al., 2008). In contrast, the utilization of mucus-derivedcarbohydrates by EHEC in the bovine intestine is poorlydocumented. The composition of the mucin covering theepithelium of the bovine small intestine has beendescribed: proteins and carbohydrates represent 53%and 47% of the bovine mucin respectively, and the mainfermentable monosaccharides constituting the mucincarbohydrate fraction are galactose, GlcNAc, GalNAc,fucose, mannose and N-acetyl neuraminic acid(Neu5Ac) (Montagne et al., 2000).

In this report, we hypothesize that mucus-derived car-bohydrates can be released in the bovine small intestinefrom the mucus layer and used as substrates for EHEC.To test this hypothesis, we compared the consumption ofcarbohydrates in the BSIC by endogenous microbiota,EHEC, and commensal E. coli, and we performed in vitrobacterial competitive experiments using a wild-type EHECand isogenic mutants deficient in carbohydrate catabolicpathways. We demonstrated that utilization of mucus-derived carbohydrates confers a competitive growthadvantage to EHEC in the BSIC and that EHEC con-sumes free carbohydrate monosaccharides more rapidlythan the bacteria belonging to the resident intestinalmicrobiota.

2 Y. Bertin et al.

© 2012 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology

Results

Carbohydrate degradation by E. coli O157:H7

To degrade oligosaccharides, bacteria must produceglycosidases with the appropriate specificities for eachof the glycoside linkages within the oligosaccharidechains. In order to analyse the genetic informationencoding carbohydrate degradation carried by theEDL933 genome, we performed in silico analyses usingthe CAZy (Carbohydrates Active enZYme) databasedescribing the families of glycoside hydrolases (GHs)(EC 3.2.1.-) that break down or modify glycosidic bonds.Fifty-two open reading frames predicted to encodehydrolases included in 19 families of GHs were presentin the bacterial genome. Most of the sequences encodedfamily-24, family-13 and family-23 GHs comprisingessentially lysozyme (EC 3.2.1.17) and amylase (EC3.2.1.1) enzymes. However, genes coding for thehydrolysis of dimers were also present in the genomeof EDL933: bglA and ascB encode a 6-phospho-b-glucosidase (GH1), ebgA and lacZ a b-D-galactosidase(GH2), iudA a b-D-glucuronidase (GH2), bglX a b-D-glucoside-glucohydrolase (GH3), celF a 6-phospho-b-glucosidase, and melA an a-galactosidase (GH4). Insummary, E. coli EDL933 possesses the genetic informa-tion required to hydrolyse disaccharides into monosac-charides (b-D-galactosidase, phospho-b-glucosidase andb-D-glucuronidase) and to hydrolyse the terminal resi-dues from oligosaccharides (a-galactosidase and b-D-glucoside-glucohydrolase). In agreement with previousstudies concerning E. coli (Hoskins et al., 1985), DNAsequence analysis strongly suggests that E. coli EDL933is limited to grow on mono- or disaccharides.

The EDL933 genome contains genes encoding thetransport and the degradation of the six main monosac-charides included in bovine mucin: fucose, galactose,GalNAc, GlcNAc, mannose and Neu5Ac (Table S1). Todetermine whether EDL933 was able to utilize thesesugars, the bacterial strain was incubated in M9 mediumsupplemented with each carbohydrate (10 mM) as thesole carbon source. Growth of EDL933 on glucose,reported to be the most suitable fermentation substrate forE. coli, was monitored as a control. The bacterial growthcurves showed that EDL933 could use each of the sixmonosaccharides as the sole carbon source (Fig. 1).However, differences in growth patterns were observed:(i) the use of GlcNAc or Neu5Ac resulted in growth pat-terns similar to that of EHEC grown on glucose (highgrowth yield and high rate), (ii) EHEC could use mannose,galactose and GalNAc but exhibited lower growth effi-ciency (extended lag phase, lower growth rate and loweryield) and (iii) incubation of EHEC with fucose resulted ina much lower cell yield, which is to be expected sincefucose catabolism requires excretion of propanediol as a

by-product. These results indicate that these sugars werenot equivalent in providing carbon and energy to the bac-terial cell.

Bovine small intestinal content and mucus-derivedcarbohydrate quantification

BSIC samples with a live endogenous microbiota (BSIC-LEM) were collected at the slaughterhouse and storedat -80°C to keep most of the endogenous microbiotaviable (see the Experimental procedure section). Bacterialcounts performed using freshly collected BSIC-LEM samples revealed the presence of 9 ¥ 104 to1.4 ¥ 105 ml-1 of strict anaerobes and 3 ¥ 105 to9.5 ¥ 105 ml-1 of facultative anaerobes. Because themammalian small intestine is neither fully aerobic noranaerobic (Jones et al., 2007; Fabich et al., 2008), strictanaerobic conditions were not physiologically relevantfor in vitro incubation of BSIC. Consequently, BSIC-LEMsamples were incubated at 39°C (internal bovine tem-perature) without shaking (to minimize oxygen availability)to mimic the physiological conditions of the bovine gut.After 7 h of incubation, the endogenous bacterial popula-tion reached 5.6 ¥ 107 to 7.8 ¥ 107 ml-1 of strict anaerobesand 8.5 ¥ 106 to 1.2 ¥ 107 ml-1 of facultative anaerobes.As previously described (Bertin et al., 2011), the endog-enous bacterial population initially present in BSIC wasable to grow under the culture conditions defined in thisreport.

The presence of free mucus-derived carbohydrates inthe BSIC was investigated by gas chromatography (GC).BSIC samples contained 1.43 mM (� 0.028) galactose,0.89 mM (� 0.016) GlcNAc, 0.72 mM (� 0.014) GalNAc,0.64 mM (� 0.013) fucose, 0.50 mM (� 0.014) mannoseand 0.09 mM (� 0.0018) Neu5Ac resulting in a total

Fig. 1. Growth curves of EDL933 incubated in minimal media. M9minimal medium was supplemented with 10 mM of each carbonsource. Cultures were incubated at 37°C with aeration. Each timepoint is the mean of three independent experiments.

Carbon nutrition of EHEC O157:H7 in the bovine intestine 3


concentration of 4.2 mM (� 0.08) carbohydrates (BSICsamples also contained 0.35 mM glucose). In agreementwith the glycoprotein composition of mucin previouslypurified from bovine small intestine by Montagne andcolleagues (2000), galactose and GlcNAc were the mostabundant (34.2 mol per cent and 19.3 mol per cent,respectively, of the total mucosal carbohydrates),whereas Neu5Ac was low (only 2.4 mol per cent of thetotal mucosal carbohydrates). These results stronglysuggest that the six carbohydrates were released into thebovine luminal content from the mucosal surface coveringthe enterocytes.

Expression of genes required forcarbohydrate catabolism

To investigate the ability of EDL933 to degrade themucus-derived carbohydrates present in the BSIC, rela-tive expression of the genes galK, agaF, nagE, fucA,manA and nanA encoding the specific catabolismof galactose, GalNAc, GlcNAc, fucose, mannose andNeu5Ac respectively (Table S1), was investigated byreal-time PCR. RNA samples were collected during theexponential growth phase, when the bacteria entered intothe stationary phase, and during the stationary phase(Fig. S1). The ratio of mRNA levels was calculated forEDL933 incubated in BSIC relative to cells grown in M9minimal medium supplemented with glucose (M9-Glc)and normalized using the tufA gene.

Maximal expression was observed for all the genesexcept manA during the exponential growth of EDL933.Particularly, a high induction of nanA (ª 370-foldincrease), fucA (ª 130-fold increase) and agaF (ª 125-foldincrease) was observed (Fig. 2). In contrast, a maximallevel of manA transcripts was found when EDL933 incu-bated in BSIC entered into the stationary phase, whereasthe remaining genes appeared to be weakly activatedor showed an expression ratio close to 1 (Fig. 2). Duringthe stationary growth phase, all the genes were poorlyexpressed or were downregulated.

Utilization of mucus-derived carbohydrates byendogenous microbiota and EHEC

The utilization of mucus-derived carbohydrates by theendogenous microbiota and by EHEC was next analysed.BSIC-LEM samples [ª 7.5 ¥ 105 anaerobes (strict and fac-ultative) ml-1] and BSIC-LEM samples inoculated withEDL933 (ª 5 ¥ 103 bacteria ml-1) were incubated at 39°Cwithout shaking, and at each time point the mucus-derived carbohydrates were quantified. As shown inFig. 3A, the intestinal endogenous microbiota was able toconsume mucus-derived carbohydrates from the first hourof incubation. However, despite the differences in initial

bacterial populations (endogenous microbiota vs EHEC),the concentration of total mucus-derived carbohydratesdecreased more rapidly when BSIC-LEM samples wereinoculated with EHEC (P < 0.01, BSIC samples inocu-lated with EDL933 vs BSIC samples from 1 to 3 h ofincubation) (Fig. 3A). During the first 4 h of incubation, thedisappearance of 25–45% of each carbohydrate wasassociated with the incubation of EDL933, suggesting ahigh efficiency of EHEC in fermenting mucus-derived car-bohydrates. Determination of the concentration of eachcarbohydrate after 3 h of incubation of BSIC inoculatedwith or without EHEC confirmed that EDL933 contributedsignificantly to degrading mucus-derived carbohydrates(Fig. 3B). In addition, the monosaccharides were not con-sumed in a sequential order since the concentration ofeach sugar decreased simultaneously during the incuba-tion of BSIC-LEM samples (results not shown) or BSIC-LEM samples inoculated with EHEC (Fig. 4). It is also ofinterest to note that carbohydrates did not completelydisappear from BSIC-LEM samples after 6 h of incubation[as 20–25% of the total mucus-derived carbohydratesinitially present in BSIC-LEM were still present (Figs 3Aand 4)]. This suggests that the concentration of carbohy-drate quantified in BSIC samples constitutes the balancebetween the carbohydrates degraded by intestinal bacte-ria and the release of carbohydrate monomers frommucus-derived polysaccharides.

As previously described (Chaucheyras-Durand et al.,2010), the EHEC strain appeared to be well adaptedfor growth on BSIC-LEM since the initial bacterialpopulation (ª 5 ¥ 103 bacteria ml-1) reached ª 106 and

Fig. 2. Relative expression levels of genes required for thecatabolism of mucus-derived carbohydrates during incubation ofEDL933 in BSIC compared with M9-Glc. The ratio of mRNA levelof each gene was measured in EDL933 incubated in filtered BSICin comparison to cells grown in M9-Glc. RNA samples werecollected during the exponential growth phase (grey), when thebacteria entered into the stationary phase (white) and during thestationary phase (black). Values are the mean � 1 SEM of threeindependent experiments.

4 Y. Bertin et al.


ª 5 ¥ 108 bacteria ml-1 after 3 and 7 h of incubationrespectively. Taken together, these results suggestedthat the rapid growth of EDL933 in BSIC-LEM could bedue, at least partially, to its effectiveness in degradingmucus-derived monosaccharides.

Bacterial growth competition in BSIC-LEM

EDL933 was able to grow in undiluted BSIC-LEMsamples without any supplementation [ª 5 log increase incolony-forming units (cfu) ml-1 after 7 h of incubation].Considering the effectiveness of EDL933 in degradingmucus-derived monosaccharides, we hypothesized thatutilization of these carbohydrates gives it a growth advan-tage in BSIC. To test this hypothesis, we used the isogenicmutants of EDL933 described by Fabich and colleagues(2008) with mutations that directly impact the first stepsof galactose, GlcNAc, fucose, mannose and Neu5Accatabolism and we constructed the EDL933DagaF mutantwith a specific defect in GalNAc degradation. Growthcompetition assays are usually performed to compare thegrowth pattern of a wild-type strain and its isogenic mutantco-incubated in biological fluids or liquid growth medium(Farrell and Finkel, 2003; Palchevskiy and Finkel, 2006;Pradhan et al., 2010; Bertin et al., 2011). A mutant strainthat does not compete efficiently for nutrients fails toreach the same population density as the parent strain,whereas similar growth curves indicate that both strainsare able to use limiting nutrients equally well or do notcompete for the same limiting nutrient.

EDL933 and each mutant strain were co-incubatedin BSIC-LEM samples at the same concentration

(5 ¥ 103 cfu ml-1), and at each time point the wild type andthe mutant strains were enumerated. The competitiveindex (CI) was calculated as described in the experimen-tal procedure section. As shown in Fig. 5, an importantgrowth defect of E. coli strains impaired in the mannose,GlcNAc, Neu5Ac or galactose catabolic pathway wasobserved during co-incubation in BSIC-LEM samples.The CI values of EDL933DmanA, EDL933DnagE,EDL933DnanAT and EDL933DgalK versus EDL933 were0.0018, 0.0028, 0.0035 and 0.016 respectively, at 7 h ofco-incubation. These results clearly demonstrate that thecatabolism of mannose, GlcNAc, Neu5Ac and galactose

Fig. 3. Disappearance rate of mucus-derived carbohydrates.A. Concentration of total mucus-derived carbohydrates was monitored during incubation of BSIC-LEM samples (�) and BSIC-LEM samplesinoculated with the EHEC strain EDL933 (•) or with the commensal E. coli strain BG1 (�). The concentration of each carbohydrate wasquantified individually.B. Concentration of each carbohydrate was quantified in bacterial supernatants of BSIC-LEM samples (white) and BSIC-LEM samplesinoculated with BG1 (black) or EDL933 (grey) after 3 h of incubation. Bars represent the SEM of three independent experiments. ***P < 0.01and **P < 0.05 vs BSIC-LEM samples as determined by the Student t-test for independent samples.

Fig. 4. Concentration of each carbohydrate during incubation ofEDL933 in BSIC-LEM samples. Bars represent the SEM of threeindependent experiments.



Fig. 5. Growth competition assays between EDL933 and its isogenic mutants. The BSIC-LEM samples were inoculated with a 1:1 mixture ofthe two strains. The mutant strains tested were defective for the pathway required for the catabolism of GalNAc (EDL933DagaF), fucose(EDL933DfucAO), galactose (EDL933DgalK), GlcNAc (EDL933DnagE), mannose (EDL933DmanA) or Neu5Ac (EDL933DnanAT). Barsrepresent the SEM of three independent experiments. The double asterisk denotes statistical significance, P < 0.05 and the triple asteriskdenotes statistical significance, P < 0.01 as determined by the Student t-test for paired samples.

6 Y. Bertin et al.


confers a competitive growth advantage to EHEC inBSIC-LEM. Mutation of the fucose catabolic pathway hasa weaker but statistically significant impact on bacterialgrowth during co-incubation in BSIC-LEM (Fig. 5). The CIvalue of EDL933DfucAO versus EDL933 was 0.0923 at7 h of incubation. Interestingly, the wild-type strain signifi-cantly outcompeted DmanA and DgalK mutants from 2and 3 h respectively, of co-incubation, whereas a signifi-cant growth defect was only observed after 4 h ofco-incubation of EDL933 with mutant strains defective forGlcNAc, Neu5Ac or fucose catabolism (Fig. 5). In con-trast, similar growth phenotype and CI value close to1 were observed when EDL933 was co-incubated withthe DagaF mutant, demonstrating that GalNAc was notrequired for the growth of EDL933 in BSIC-LEM samples.

The growth curves of DmanA, DnagE, DnanAT, DgalKand DfucAO mutants individually incubated in BSIC-LEMsamples were closely similar to that of EDL933 (data notshown), indicating that (i) the growth defects observedduring co-incubation did not result from an inability of themutants to grow in BSIC but from an inability to competewith the wild-type strain and (ii) in the absence of a straincompeting for the same sugars, the EHEC mutant strainsgrow at the same rate as the wild-type EDL933.

In summary, growth competition assays suggest thatthe capacity of EDL933 to degrade mannose, GlcNAc,Neu5Ac and galactose is required for maximal growth ofEHEC in the bovine small intestine.

Utilization of mucus-derived carbohydrates bycommensal bovine E. coli

BLASTn analysis from bacterial genome libraries showedthat galK, nagE, agaF, nanA, fucA and manA are present

in the genome of numerous pathogenic and commensalE. coli strains present in the ‘microbes genomic’ BLAST

database. However, agaF is absent in the genome of theK12 E. coli MG1655 resulting in its inability to utilizeGalNAc as a nutrient (Fabich et al., 2008; Mukherjeeet al., 2008). The presence of agaF was investigated byPCR amplification among commensal E. coli strains fromour laboratory collection. In contrast to E. coli MG1655, aspecific agaF amplification product was obtained from allthe bovine strains tested (n = 30) (data not shown), indi-cating the presence of the genetic information required forGalNAc utilization in the genome of bovine commensalE. coli.

In this report, we used the E. coli strain BG1 isolatedfrom the small intestine of a cow at slaughter (Bertin et al.,2001) to analyse the growth pattern of a commensalE. coli. Incubation of BG1 in M9 medium supplementedwith carbohydrate demonstrated that the commensalE. coli was able to use each of the mucus-derived carbo-hydrates as the sole carbon source. The relative expres-sion of genes required for the catabolism of mucus-derived carbohydrates was then quantified during growthof BG1 in filtered BSIC in comparison with growth inM9-Glc. The highest induction rates were observed for theagaF and fucA genes (ª 25-fold increase) during theexponential growth phase, whereas nanA, galK, manAand nagE were poorly induced or were repressed duringthe exponential or stationary phases (Fig. S2). Moreimportantly, genes required for the catabolism of carbo-hydrates appeared to be less induced in BG1 than inEDL933 (Fig. 6). Indeed, the level of nanA, fucA, agaF,galK and nagE transcripts in BSIC was six- to 97-foldlower in BG1 than in EDL933 during the exponentialgrowth phase, and the level of manA transcripts wasª sevenfold lower during the stationary growth phase(Fig. 6). The commensal E. coli strain BG1 was thentested for its ability to catabolize mucus-derived carbohy-drates. As described above, BSIC samples were inocu-lated with BG1 (5 ¥ 103 bacteria ml-1) and disappearanceof the carbohydrates was monitored during incubationat 39°C. The patterns of carbohydrate disappearanceshowed no significant difference between BSIC-LEMsamples and BSIC-LEM samples inoculated with BG1(Fig. 3A and B). In addition, the rate of carbohydrate dis-appearance in BSIC-LEM samples inoculated withEDL933 is greater than that of BSIC-LEM samples inocu-lated with BG1 (Fig. 3A and B) suggesting that EDL933uses sugars at a more rapid rate than BG1.

Bacterial growth competition experiments includingBG1 were also performed. Indistinguishable growthcurves and CI value close to 1 were obtained whenBG1 and EDL933 were co-incubated in BSIC (Fig. S3). Incontrast, co-incubation of BG1 and EDL933Dgal,EDL933DfucAO, EDL933DnagE, EDL933DmanA or

Fig. 6. Fold-change comparison of carbohydrate catabolism geneexpression between the EHEC strain EDL933 and the commensalE. coli strain BG1. The ratio of mRNA level of each gene wasmeasured in the E. coli strains incubated in filtered BSIC. RNAsamples were collected during the exponential growth phase (grey),when the bacteria entered into the stationary phase (white) andduring the stationary phase (black). Values are the mean � 1 SEMof three independent experiments.



EDL933DnanAT showed that the commensal E. coli sig-nificantly outcompeted each of the mutant strains(Fig. S3). These results indicated that EHEC needs toconsume galactose, GlcNAc, mannose, Neu5Ac orfucose to maintain its growth rate at the same level as thatof the commensal E. coli strain in the BSIC. Therefore,since in the absence of a strain competing for the samesugars, the EDL933 sugar mutants grow at the same rateas the wild-type EDL933 in BSIC-LEM (Fig. S3), BG1 andEDL933 are competing for the same sugars.

Discussion

Although E. coli strains possess the genetic information tosynthesize sugars de novo, it is clearly in their interestto scavenge sugars from the surrounding environment.Carbohydrates are the largest component of the cow’sdiet (60–70%) and consist essentially in polysaccharidesembedded in plant cell walls. Dietary plant cell wallpolysaccharides are mainly cellulose, consisting in linearchains of glucose units (up to 40% of total plant biomass),and hemicelluloses that are composed of more complexheteropolymers including glucose, arabinose, xylose,mannose, galactose or fucose (Dashtban et al., 2010;Schadel et al., 2010) (but not GalNAc, GlcNAc orNeu5Ac). These polysaccharides are extensively de-graded in the bovine rumen and the resulting monosac-charides are rapidly and almost completely fermented bythe vast microbial population of the rumen (for a review,see Russell et al., 2009; Wilson, 2011). A low proportion ofundegraded plant cell wall polysaccharides can onlytransit through the small intestine and reaches the largeintestine where their efficient breakdown occurs thanks tothe presence of a diversified microbial fibrolytic activity(Michalet-Doreau et al., 2002). Although a part of dietarystarch can be degraded in the small intestine, glucosemonomers are generally rapidly absorbed or directly uti-lized by the intestinal cell wall (Owens et al., 1986) result-ing in the presence of a very few or even no solublecarbohydrates of dietary origin in the ileum. Additionally,the six main monosaccharides reported to be included inthe mucin glycoprotein fraction were found in the BSIC insimilar proportions to those of the mucus layer (Montagneet al., 2000). Taken together, these observations stronglysuggest that the six monosaccharides are not provided bydietary polysaccharide degradation but are released fromthe mucus layer covering the enterocytes.

Bacterial growth patterns clearly showed that Neu5Acand GlcNAc are the most efficient carbon sources for invitro multiplication of EDL933 in minimal medium. Previ-ous studies have shown that EHEC grown in minimalmedium supplemented with a complex mixture of sugarsconsumes the mucus-derived carbohydrates in a preciseorder with cascading carbon source utilization and over-

lapping metabolism (Chang et al., 2004; Fabich et al.,2008). For example, the catabolite-repressing sugarsGlcNAc and galactose are first consumed by EDL933,whereas fucose and GalNAc disappear from the mediawhen all the other sugars are completely exhausted(Fabich et al., 2008). In this report, we demonstratedthat EDL933 simultaneously consumed the six mucus-derived sugars present in the BSIC, suggesting that theintestinal environment must affect the selection of car-bohydrates by EHEC, perhaps owing to the presence ofamino acids and other growth factors present in BSICthat were absent in the minimal medium used in theprevious study.

Based on competitive growth experiments, we demon-strated that mannose, GlcNAc, Neu5Ac and galactosewere the preferred mucus-derived carbohydrates used byEDL933 for its growth in BSIC-LEM. To a lesser extent,fucose also contributed to the growth of EHEC. In agree-ment with our results, in vivo colonization experiments incattle have shown that specific transport of GlcNAc andutilization of fucose (but not GalNAc) are required forcolonization of the bovine gut or the persistence of EHECin the bovine rectum (Dziva et al., 2004; Snider et al.,2009). Competitive advantage due to mucus-derived car-bohydrates in the gut of mammals is also observed forother pathogenic bacteria. For example, Neu5Ac andfucose catabolism confers a competitive advantage topathogenic Vibrio cholerae and Campylobacter jejeuni inthe mouse intestine and in chick caecum respectively,whereas a Dgal mutant is dramatically impaired in EHECcolonization of the infant rabbit intestine (Ho and Waldor,2007; Sheng et al., 2008; Almagro-Moreno and Boyd,2009; Muraoka and Zhang, 2011).

Interestingly, all the carbohydrates disappeared fromBSIC-LEM samples during the first hour of EDL933 incu-bation, whereas each of the DnagE, DnanAT, DgalK andDfucAO mutants showed a significant growth defect after3–4 h of co-incubation. This suggested that during the firsthours of co-incubation the wild-type strain probably con-sumed a part of GlcNAc, Neu5Ac, galactose or fucoseinitially present in BSIC-LEM, whereas the mutant strainsdefective for the catabolism of the corresponding carbo-hydrate could use other carbon sources for growth. Wehypothesized that insufficient carbon sources were avail-able in BSIC-LEM for optimal growth of the mutant strainsafter 3 or 4 h of co-incubation, whereas the wild-typestrain could continue to degrade the mucus-derived car-bohydrates constantly released by the resident anaer-obes from mucus-derived polysaccharides.

In the bovine gut, EHEC is present at a low concentra-tion as compared with the high density of the residentmicrobiota, and must compete for carbon sources topersist and grow. In our experiments, BSIC-LEM con-tained approximately 8 ¥ 105 anaerobic bacteria ml-1 and

8 Y. Bertin et al.


was inoculated at a low population by EHEC (ª 5 ¥ 103

bacteria ml-1). In the bovine intestine, anaerobes of theendogenous microbiota have been adapted for the break-down and fermentation of complex polysaccharides thatescape hydrolysis in the rumen. In contrast to E. coli,which is limited to growth on mono- or disaccharides, theendogenous microbiota can also use oligosaccharides,and even in some cases preferentially uses oligosaccha-rides rather than the corresponding monomers (Amarettiet al., 2006). Moreover, EDL933 grew more rapidly inBSIC-LEM than the resident strict and facultative anae-robic bacteria. Consequently, EDL933 can probablyconsume free monosaccharides more rapidly, suggestinga competitive ability of EHEC to catabolize mucus-derivedcarbohydrates. This is consistent with the nutrient-nichetheory postulating that a microorganism can coexist in themammal intestine by utilizing limiting nutrients better thanthe other bacterial species (Freter et al., 1983). In additionto the resident microbiota, the commensal E. coli BG1also consumes the mucus-derived sugars present inBSIC-LEM more slowly than EHEC. However, based oncompetitive growth experiments, we demonstrated thatEHEC and commensal E. coli grow together at the samerate in BSIC-LEM and compete for mucus-derived sugars.Therefore, it appears that the commensal E. coli strain ismore efficient than EHEC in using sugars for growth inBSIC-LEM, i.e. same rate of growth metabolizing fewersugars. Interestingly, these results suggested that com-mensal strains may be better adapted to limiting nutrientsin the intestine. This may help to explain why O157 infec-tions run their course in a week or two, while commensalscolonize their host for months.

The metabolism of EHEC during gut colonization ofmice has been extensively explored (Miranda et al., 2004;Fabich et al., 2008; Leatham et al., 2009). The strainEDL933 is present both in the mucus layer and closelyassociated with epithelial cells, but the bacterial strain isunable to grow in vitro on intestinal luminal contents,suggesting that EHEC colonizes the mouse intestine bygrowing in the mucus layer (Miranda et al., 2004; Fabichet al., 2008). Moreover, colonization of streptomycin-treated mice by EDL933 is supported by the catabolismof several carbohydrates including galactose, fucose,mannose and GlcNAc, but not Neu5Ac or GalNAc (Fabichet al., 2008). In particular, inactivation of the galactoseand fucose pathways has the largest impact on mousecolonization fitness (Fabich et al., 2008). In contrast, weshowed that EDL933 grows rapidly in BSIC-LEM and wefound that mannose and GlcNAc catabolism conferred thegreatest competitive growth advantage to EHEC in BSIC.Furthermore, we highlight the considerable growth advan-tage conferred by Neu5Ac to EHEC despite its lowabundance. In fact, diet, digestive system and intestinalmicrobiota of ruminant and monogastric animals resulted

in different status of nutrient limitation and, consequently,different strategies may be adopted by EHEC to persist inthe gut according to the infected host.

Freter’s nutrient-niche theory postulates that coloniza-tion of the intestine by a particular bacterium is defined byits ability to occupy nutrient-defined ecological niches thatdiffer from the other species present (Freter et al., 1983).According to this hypothesis, the population density of aparticular bacterium is determined by the available con-centration of its preferred nutrient, for which its affinitylikely is the highest. Our data indicate a competitive abilityof EHEC compared with the resident microbiota or com-mensal E. coli to catabolize mucus-derived carbohydrates(in particular mannose, GlcNAc, Neu5Ac and galactose).These carbon sources probably constitute an ecologicalniche for EHEC strains in the bovine small intestine.In-depth knowledge of the physiology of EHEC in thedigestive tract of the ruminant will help to select nutritionalor ecological strategies (for example probiotics) in order toreduce EHEC carriage prior to slaughter and to limit thedissemination of EHEC into the human food chain.

Experimental procedures

Bacterial strains and growth conditions

Escherichia coli strains used in this study are listed inTable S2. The spontaneous nalidixic acid- and rifampicin-resistant mutants EDL933 NalR and EDL933 RifR respectivelyshowed growth curves similar to that of EDL933 when cul-tured in Luria–Bertani (LB) broth at 37°C. Similarly, the nali-dixic acid-resistant mutant BG1 NalR showed growth curvessimilar to that of BG1. The capacity of EDL933 to use aspecific mucus-derived carbohydrate as a sole carbon sourcewas assessed by using M9 minimal medium (DIFCO) sup-plemented with glucose, galactose, GlcNAc, GalNAc, fucose,mannose or Neu5Ac (10 mM) (Sigma-Aldrich). Broth cultureswere started from a single colony in LB medium and grownovernight at 37°C with aeration. Cells were then pelleted bycentrifugation, resuspended in sterile PBS and diluted 75-foldin M9 minimal medium supplemented with the correspondingsugar. The cultures were then incubated at 37°C with shakingand the growth was monitored spectrophotometrically at anoptical density of 600 nm (OD600).

Preparation of bovine intestinal contents andmicrobial enumeration

BSIC with a live endogenous microbiota (LEM) were col-lected and stored as previously described (Bertin et al.,2011). Briefly, three beef cattle were slaughtered in accord-ance with the guidelines of the local Ethics Committee in theexperimental facility of Unit of Research on Herbivores,INRA, Saint-Genès Champanelle, France. The jejunum andthe ileum were removed as a single piece and the totalluminal contents were collected in O2-free N2 saturated sterileflasks. The BSIC-LEM samples were then pooled, rapidlyfiltered through four layers of cheesecloth and immediatelyfrozen at -80°C until use.



To enumerate total facultative anaerobes, 10-fold serialdilutions of each sample were performed in the mineral solu-tion described by Bryant and Burkey (1953). The bacterialdilutions were plated on Petri dishes containing G20 complexmedium (Chassard et al., 2008) and incubated at 39°C for48 h. To enumerate the total viable counts of strictly anaero-bic bacteria, 10-fold serial dilutions of BSIC samples wereperformed in mineral solution (Bryant and Burkey, 1953)under CO2 flow. Each dilution was inoculated into O2-free CO2

saturated roll tubes (Hungate et al., 1966) using the completeCC medium (Leedle and Hespell, 1980). Three roll tubeswere inoculated per dilution and the colony counts weredetermined after 2–3 days at 39°C.

To obtain sterile BSIC, the intestinal contents were centri-fuged twice at 2000 g for 20 min and the resulting superna-tants were filtered through a 0.22 mm nylon filter. Theefficiency of the filtration step was verified by adding BSICsamples on Petri dishes of G20 complex medium and in CCanaerobic broth. The absence of either colonies on agarplates or bacterial growth in CC medium after 48–72 h at39°C confirmed that BSIC was sterile. Resulting pooled aliq-uots were then stored at -80°C until use.

Mutant construction

The mutant strain defective for the gene agaF (required for aGalNAc-positive phenotype) (Mukherjee et al., 2008) wasconstructed by the replacement of agaF by the gene confer-ring resistance to kanamycin using a one-step PCR-basedmethod (Datsenko and Wanner, 2000). The primers used toconstruct the EDL933DagaF mutant (agaF_Km_F: GCAAATTCTCGGTGAGCAATCGCAGTTTATCGCCATCGATTTTCCGGAAACAGCCACGTTGTGTCTCAAAATC and agaF_Km_R: TGCTCGATGAGTTTTTCTAATAACTCGTCCACCAGACTGGTCAGCCCACGATGTCCACACTCCAGCGCCTG) weredesigned according to the EDL933 genome sequence. Geneknockouts were confirmed by PCR analysis and DNAsequencing. The phenotype of EDL933DagaF was controlledby incubating the mutant strain in M9 minimal medium sup-plemented with GalNAc (10 mM) or glucose (10 mM) andmonitoring bacterial growth spectrophotometrically (OD600).

Complementation of the mutant strains

To generate the complemented mutant strains, PCR ampli-fications were performed using the high-fidelity Pfx50™ DNApolymerase (Invitrogen). The primers used consist of thefirst or last nucleotides of the corresponding ORF andof the restriction sequence of the enzymes HindIII, EcoRIor NcoI (Table S3). DNA sequences corresponding to theagaF, fucAO, galK, manA, nagE, and nanAT genes weredesigned according to the EDL933 genome sequence(Z4488, Z4117 and Z416, Z0927, Z2616, Z0826, Z4583and Z4582 respectively). The PCR products were purifiedusing the Qiaquick Purification PCR kit (Qiagen) anddigested with the relevant enzymes. The PCR fragmentswere then ligated into the expression vector pBAD24 (con-ferring resistance to ampicillin, under the control of thearaC-PBAD promoter) using T4 DNA ligase (Invitrogen) aspreviously described (Guzman et al., 1995). Each of theresulting recombinant plasmids was then electroporated into

the corresponding mutant and selected on an LB agar platecontaining ampicillin (50 mg ml-1) and the relevant antibiotic[kanamycin or chloramphenicol (50 mg ml-1 each)]. Genecomplementation was checked by PCR analysis and DNAsequencing. In addition, the pBAD24 vector was electropo-rated into the wild-type strain EDL933 to create a vector-only control strain. To check the restoration of thephenotype, the resulting strains were inoculated in M9minimal medium supplemented with ampicillin (50 mg ml-1),L-arabinose (0.5 mM) and the corresponding carbohydrates(10 mM). The bacterial suspensions were then incubated at37°C with shaking and growth was monitored spectropho-tometrically (OD600).

RNA extraction and relative mRNA quantification byreal-time PCR (q-PCR)

The strain EDL933 was incubated at 39°C without aeration infiltered BSIC samples and in M9 minimal media supple-mented with glucose (10 mM) as the sole carbon source(M9-Glc). Bacterial suspensions were collected during theexponential growth phase (2.5 h and 5 h in filtered BSIC andM9-Glc respectively), when the bacteria entered into the sta-tionary phase (4.25 h and 7.25 h in filtered BSIC and M9-Glcrespectively), or during the stationary phase (6.5 h and 9.5 hin filtered BSIC and M9-Glc respectively) (Fig. S1). The bac-terial suspensions were then centrifuged at 10 000 g for15 min and the supernatants were stored at -20°C for furtherinvestigations. Total RNAs were isolated from bacterialpellets by using TRIzol as previously described (Vareilleet al., 2007; de Sablet et al., 2008). RNA quantification wasperformed using a NanoDrop spectrophotometer and RNAintegrity was electrophoretically verified by ethidium bromidestaining. One microgram of each RNA sample was thenreverse transcribed using the SuperScript II Reverse Tran-scriptase kit (Invitrogen) with 3 mg of random primer and 100units of SuperScript II Rnase H. Real-time PCR runs werecarried out using the Mastercycler ep realplex apparatus(Eppendorf) with 20 ng of cDNA, 0.5 mM of each primer,3 mM of MgCl2, 10 ml of SYBR® Premix Ex Taq™ mix (TakaraBio) in a final volume of 20 ml. Amplification conditions wereas follows: 95°C for 15 s, 55°C for 15 s and 72°C for 20 s.The tufA mRNA was used for normalization of mRNA quan-tification. The relative mRNA quantification was performedusing primers designed to specifically amplify fragments of90–200 bp (Table S4). Control samples lacking the reversetranscriptase were included to assess DNA contaminationand triplicate samples were amplified in each case. Resultswere calculated using the comparative cycle thresholdmethod.

Bacterial competition experiments

The spontaneous mutant strains EDL933 NalR and BG1NalR were used in competition experiments betweenwild-type strains and isogenic mutants unable to degrademucus-derived carbohydrates. Competition experimentsbetween pathogenic and commensal E. coli were performedusing the EDL933 RifR and BG1 NalR spontaneous mutantsrespectively. Precultures of the E. coli strains, inoculatedfrom a single colony, were incubated in LB broth with

10 Y. Bertin et al.


appropriate antibiotic for 8 h at 37°C with aeration. The pre-cultures were then 50-fold diluted in LB broth and grownovernight at 39°C without shaking. The next day, a sampleof BSIC-LEM was inoculated with approximately 5 ¥ 103

bacteria ml-1 of each of the two strains tested and incubatedat 39°C without shaking. At each time point, the co-culturewas 10-fold serially diluted in phosphate buffer (PBS)pH 7.2 and plated on Sorbitol MacConkey (SMAC) agarplates containing kanamycin (50 mg ml-1), nalidixic acid(50 mg ml-1), rifampicin (100 mg ml-1) or chloramphenicol(50 mg ml-1). The plates were then incubated overnight at37°C and the number of cfu was determined. Each experi-ment was replicated at least three times. The presentedvalues are the log10 mean number of cfu ml-1 � the stand-ard error. Statistical analysis was done using Student’st-test for paired samples (two-tailed). The CI was calculatedas follows: (mutant cfu recovered/wild-type cfu recovered)/(mutant cfu inoculated/wild-type cfu inoculated). CI < 1 indi-cated that the wild-type strain outcompeted the mutant;CI > 1 indicated that the mutant outcompeted the wild-typestrain; CI ª 1 indicated that none of the two strains has acompetitive advantage with regard to the other one.

Monosaccharide analysis

The molar composition of free mucus-derived carbohydrateswas determined by GC on a Shimadzu gas chromatographequipped with a 25 m ¥ 0.32 mm CPSil5 CB Low bleed/MScapillary column, 0.25 mm film phase (Chrompack France,Les Ulis, France), as previously described (Kamerling et al.,1975; Montreuil et al., 1986). Briefly, 5 mg of mesoinositol(internal standard) was added to 250 ml of each BSIC sample.Freeze-dried samples were then submitted to methanolysisfor 24 h at 80°C in 500 ml of 0.5 M HCl-methanol. The acidicsolution was neutralized by adding silver carbonate and re-N-acetylated overnight at room temperature by adding 20 ml ofacetic anhydride. The methanolic phase was washed twicewith 200 ml of heptane and dried under a stream of nitrogen.Monosaccharides were further trimethylsilylated by adding50 ml of BSTFA (bis-silyltrifluoroacetamide) and 50 mlof pyridine. After 2 h, 0.5 ml of the solution was applied toGC. Standard carbohydrates (1 mg ml-1) (Sigma-Aldrich),co-injected with internal standard (mesoinositol), are used inan independent GC experimentation to determine their reten-tion time as well as their relative response factor. Co-injectionof internal standard at a known concentration with BSICsample allows quantifying the relative molar ratio for eachmonosaccharide. Statistical analysis was done using the Stu-dent’s t-test for independent samples.

In silico and statistical analyses

In silico analyses were performed using the xBASE (http://xbase.ac.uk) and BLAST (http://www.ncbi.nlm.nih.gov)servers, and the CAZy (Carbohydrates Active enZYme)(http://www.cazy.org) database. The nucleotide sequenceswere compared with bacterial complete genomes presentin the ‘Microbes genomic’ BLAST database at NCBI andwith DNA sequences from gut metagenomes (http://www.ncbi.nlm.nih.gov/sutils/blast_table.cgi?taxid=Environmental&taxidinf=environ_info&selectall).

Acknowledgements

This work was supported by an EU project (ProSafeBeef)within the 6th Framework Programme (ref. Food-CT-2006–32241).

References

Almagro-Moreno, S., and Boyd, E.F. (2009) Sialic acidcatabolism confers a competitive advantage to pathogenicVibrio cholerae in the mouse intestine. Infect Immun 77:3807–3816.

Amaretti, A., Tamburini, E., Bernardi, T., Pompei, A., Zanoni,S., Vaccari, G., et al. (2006) Substrate preference of Bifi-dobacterium adolescentis MB 239: compared growth onsingle and mixed carbohydrates. Appl Microbiol Biotechnol73: 654–662.

Baines, D., Lee, B., and McAllister, T. (2008) Heterogeneity inenterohemorrhagic Escherichia coli O157:H7 fecal shed-ding in cattle is related to Escherichia coli O157:H7 colo-nization of the small and large intestine. Can J Microbiol54: 984–995.

Bertin, Y., Boukhors, K., Pradel, N., Livrelli, V., and Martin, C.(2001) Stx2 subtyping of Shiga toxin-producingEscherichia coli isolated from cattle in France: detection ofa new Stx2 subtype and correlation with additional viru-lence factors. J Clin Microbiol 39: 3060–3065.

Bertin, Y., Girardeau, J.P., Chaucheyras-Durand, F., Lyan, B.,Pujos-Guillot, E., Harel, J., and Martin, C. (2011) Entero-haemorrhagic Escherichia coli gains a competitive advan-tage by using ethanolamine as a nitrogen source inthe bovine intestinal content. Environ Microbiol 13: 365–377.

Bryant, M.P., and Burkey, L.A. (1953) Cultural methods andsome characteristics of some of the more numerousgroups of bacteria in the bovine rumen. J Dairy Sci 36:205–217.

Caprioli, A., Morabito, S., Brugere, H., and Oswald, E. (2005)Enterohaemorrhagic Escherichia coli: emerging issues onvirulence and modes of transmission. Vet Res 36: 289–311.

Chang, D.E., Smalley, D.J., Tucker, D.L., Leatham, M.P.,Norris, W.E., Stevenson, S.J., et al. (2004) Carbon nutritionof Escherichia coli in the mouse intestine. Proc Natl AcadSci USA 101: 7427–7432.

Chase-Topping, M., Gally, D., Low, C., Matthews, L., andWoolhouse, M. (2008) Super-shedding and the linkbetween human infection and livestock carriage ofEscherichia coli O157. Nat Rev Microbiol 6: 904–912.

Chassard, C., Scott, K.P., Marquet, P., Martin, J.C.,Del’homme, C., Dapoigny, M., et al. (2008) Assessment ofmetabolic diversity within the intestinal microbiota fromhealthy humans using combined molecular and culturalapproaches. FEMS Microbiol Ecol 66: 496–504.

Chaucheyras-Durand, F., Madic, J., Doudin, F., and Martin,C. (2006) Biotic and abiotic factors influencing in vitrogrowth of Escherichia coli O157:H7 in ruminant digestivecontents. Appl Environ Microbiol 72: 4136–4142.

Chaucheyras-Durand, F., Faqir, F., Ameilbonne, A., Rozand,C., and Martin, C. (2010) Fates of acid-resistant andnon-acid-resistant Shiga toxin-producing Escherichia coli



http://xbase.ac.uk

http://xbase.ac.uk

http://www.ncbi.nlm.nih.gov

http://www.cazy.org

strains in ruminant digestive contents in the absence andpresence of probiotics. Appl Environ Microbiol 76: 640–647.

Cieslak, P.R., Barrett, T.J., Griffin, P.M., Gensheimer, K.F.,Beckett, G., Buffington, J., and Smith, M.G. (1993)Escherichia coli O157:H7 infection from a manuredgarden. Lancet 342: 367.

Dashtban, M., Maki, M., Leung, K.T., Mao, C., and Qin, W.(2010) Cellulase activities in biomass conversion: meas-urement methods and comparison. Crit Rev Biotechnol 30:302–309.

Datsenko, K.A., and Wanner, B.L. (2000) One-step inacti-vation of chromosomal genes in Escherichia coli K-12using PCR products. Proc Natl Acad Sci USA 97: 6640–6645.

Deplancke, B., and Gaskins, H.R. (2001) Microbial modula-tion of innate defense: goblet cells and the intestinal mucuslayer. Am J Clin Nutr 73: 1131s–1141s.

Diez-Gonzalez, F., Callaway, T.R., Kizoulis, M.G., andRussell, J.B. (1998) Grain feeding and the dissemination ofacid-resistant Escherichia coli from cattle. Science 281:1666–1668.

Dziva, F., van Diemen, P.M., Stevens, M.P., Smith, A.J., andWallis, T.S. (2004) Identification of Escherichia coliO157:H7 genes influencing colonization of the bovine gas-trointestinal tract using signature-tagged mutagenesis.Microbiology 150: 3631–3645.

Fabich, A.J., Jones, S.A., Chowdhury, F.Z., Cernosek, A.,Anderson, A., Smalley, D., et al. (2008) Comparison ofcarbon nutrition for pathogenic and commensalEscherichia coli strains in the mouse intestine. InfectImmun 76: 1143–1152.

Farrell, M.J., and Finkel, S.E. (2003) The growth advantagein stationary-phase phenotype conferred by rpoS muta-tions is dependent on the pH and nutrient environment. JBacteriol 185: 7044–7052.

Freter, R., Brickner, H., Botney, M., Cleven, D., and Aranki,A. (1983) Mechanisms that control bacterial populations incontinuous-flow culture models of mouse large intestinalflora. Infect Immun 39: 676–685.

Guzman, L.M., Belin, D., Carson, M.J., and Beckwith, J.(1995) Tight regulation, modulation, and high-level expres-sion by vectors containing the arabinose P-BAD promoter.J Bacteriol 177: 4121–4130.

Ho, T.D., and Waldor, M.K. (2007) EnterohemorrhagicEscherichia coli O157:H7 gal mutants are sensitive to bac-teriophage P1 and defective in intestinal colonization.Infect Immun 75: 1661–1666.

Hoskins, L.C., Agustines, M., McKee, W.B., Boulding, E.T.,Kriaris, M., and Niedermeyer, G. (1985) Mucin degradationin human colon ecosystems. Isolation and properties offecal strains that degrade ABH blood group antigens andoligosaccharides from mucin glycoproteins. J Clin Invest75: 944–953.

Hungate, R.E., Smith, W., and Clarke, R.T. (1966) Suitabilityof butyl rubber stoppers for closing anaerobic roll culturetubes. J Bacteriol 91: 908–909.

Johansson, M.E.V., Ambort, D., Pelaseyed, T., Schutte, A.,Gustafsson, J.K., Ermund, A., et al. (2011) Compositionand functional role of the mucus layers in the intestine. CellMol Life Sci 68: 3635–3641.

Jones, S.A., Chowdhury, F.Z., Fabich, A.J., Anderson, A.,Schreiner, D.M., House, A.L., et al. (2007) Respiration ofEscherichia coli in the mouse intestine. Infect Immun 75:4891–4899.

Kamerling, J.P., Gerwig, G.J., Vliegenthart, J.F., and Clamp,J.R. (1975) Characterization by gas-liquidchromatography-mass spectrometry and proton-magnetic-resonance spectroscopy of pertrimethylsilyl methyl glyco-sides obtained in the methanolysis of glycoproteins andglycopeptides. Biochem J 151: 491–495.

Keen, J.E., Laegreid, W.W., Chitko-McKown, C.G., Durso,L.M., and Bono, J.L. (2010) Distribution of Shiga-toxigenicEscherichia coli O157 in the gastrointestinal tract of Natu-rally O157-shedding cattle at necropsy. Appl EnvironMicrobiol 76: 5278–5281.

Law, D. (2000) Virulence factors of Escherichia coli O157 andother Shiga toxin-producing E. coli. J Appl Microbiol 88:729–745.

Leatham, M.P., Banerjee, S., Autieri, S.M., Mercado-Lubo,R., Conway, T., and Cohen, P.S. (2009) Precolonizedhuman commensal Escherichia coli strains serve as abarrier to E. coli O157:H7 growth in the streptomycin-treated mouse intestine. Infect Immun 77: 2876–2886.

Leedle, J.A., and Hespell, R.B. (1980) Differential carbohy-drate media and anaerobic replica plating techniques indelineating carbohydrate-utilizing subgroups in rumen bac-terial populations. Appl Environ Microbiol 39: 709–719.

Lim, J.Y., Li, J., Sheng, H., Besser, T.E., Potter, K., andHovde, C.J. (2007) Escherichia coli O157:H7 colonizationat the rectoanal junction of long-duration culture-positivecattle. Appl Environ Microbiol 73: 1380–1382.

Michalet-Doreau, B., Fernandez, I., and Fonty, G. (2002) Acomparison of enzymatic and molecular approaches tocharacterize the cellulolytic microbial ecosystems of therumen and the cecum. J Anim Sci 80: 790–796.

Miranda, R.L., Conway, T., Leatham, M.P., Chang, D.E.,Norris, W.E., Allen, J.H., et al. (2004) Glycolytic and gluco-neogenic growth of Escherichia coli O157:H7 (EDL933)and E. coli K-12 (MG1655) in the mouse intestine. InfectImmun 72: 1666–1676.

Montagne, L., Toullec, R., and Lalles, J.P. (2000) Calfintestinal mucin: isolation, partial characterization, andmeasurement in ileal digesta with an enzyme-linked immu-nosorbent assay. J Dairy Sci 83: 507–517.

Montreuil, J., Bouquelet, S., Debray, H., Fournet, H., Spik, G.,and Strecker, G. (1986) Glycoproteins. In CarbohydratesAnalysis: A Practical Approach. Chaplin, M.F., andKennedy, J.F. (eds). Oxford, UK: IRL Press, pp. 143–204.

Mukherjee, A., Mammel, M.K., LeClerc, J.E., and Cebula,T.A. (2008) Altered utilization of N-acetyl-D-galactosamineby Escherichia coli O157:H7 from the 2006 spinach out-break. J Bacteriol 190: 1710–1717.

Muniesa, M., Jofre, J., Garcia-Aljaro, C., and Blanch, A.R.(2006) Occurrence of Escherichia coli O157:H7 and otherenterohemorrhagic Escherichia coli in the environment.Environ Sci Technol 40: 7141–7149.

Muraoka, W.T., and Zhang, Q. (2011) Phenotypic and geno-typic evidence for L-fucose utilization by Campylobacterjejuni. J Bacteriol 193: 1065–1075.

Nart, P., Naylor, S.W., Huntley, J.F., McKendrick, I.J., Gally,D.L., and Low, J.C. (2008) Responses of cattle to gastroin-

12 Y. Bertin et al.


testinal colonization by Escherichia coli O157:H7. InfectImmun 76: 5366–5372.

Naylor, S.W., Roe, A.J., Nart, P., Spears, K., Smith, D.G.,Low, J.C., and Gally, D.L. (2005) Escherichia coli O157:H7forms attaching and effacing lesions at the terminal rectumof cattle and colonization requires the LEE4 operon. Micro-biology 151: 2773–2781.

O’Brien, S.J., Adak, G.K., and Gilham, C. (2001) Contact withfarming environment as a major risk factor for Shiga toxin(Vero cytotoxin)-producing Escherichia coli O157 infectionin humans. Emerg Infect Dis 7: 1049–1051.

Owens, F.N., Zinn, R.A., and Kim, Y.K. (1986) Limits to starchdigestion in the ruminant small intestine. J Anim Sci 63:1634–1648.

Palchevskiy, V., and Finkel, S.E. (2006) Escherichia coli com-petence gene homologs are essential for competitivefitness and the use of DNA as a nutrient. J Bacteriol 188:3902–3910.

Png, C.W., Linden, S.K., Gilshenan, K.S., Zoetendal, E.G.,McSweeney, C.S., Sly, L.I., et al. (2010) Mucolytic bacteriawith increased prevalence in IBD mucosa augment in vitroutilization of mucin by other bacteria. Am J Gastroenterol105: 2420–2428.

Pradhan, S., Baidya, A.K., Ghosh, A., Paul, K., and Chowd-hury, R. (2010) The El Tor biotype of Vibrio cholerae exhibitsa growth advantage in the stationary phase in mixed cul-tures with the classical biotype. J Bacteriol 192: 955–963.

Russell, J.B., Muck, R.E., and Weimer, P.J. (2009) Quantita-tive analysis of cellulose degradation and growth of cellu-lolytic bacteria in the rumen. FEMS Microbiol Ecol 67:183–197.

de Sablet, T., Bertin, Y., Vareille, M., Girardeau, J.P., Gar-rivier, A., Gobert, A.P., and Martin, C. (2008) Differentialexpression of stx2 variants in Shiga toxin-producingEscherichia coli belonging to seropathotypes A and C.Microbiology 154: 176–186.

Schadel, C., Richter, A., Blochl, A., and Hoch, G. (2010)Hemicellulose concentration and composition in plantcell walls under extreme carbon source-sink imbalances.Physiol Plant 139: 241–255.

Sheng, H.Q., Lim, J.Y., Watkins, M.K., Minnich, S.A., andHovde, C.J. (2008) Characterization of an Escherichia coliO157:H7 O-antigen deletion mutant and effect of the dele-tion on bacterial persistence in the mouse intestine andcolonization at the bovine terminal rectal mucosa. ApplEnviron Microbiol 74: 5015–5022.

Snider, T.A., Fabich, A.J., Conway, T., and Clinkenbeard,K.D. (2009) E. coli O157:H7 catabolism of intestinal mucin-derived carbohydrates and colonization. Vet Microbiol 136:150–154.

Vareille, M., de Sablet, T., Hindre, T., Martin, C., and Gobert,A.P. (2007) Nitric oxide inhibits Shiga-toxin synthesis byenterohemorrhagic Escherichia coli. Proc Natl Acad SciUSA 104: 10199–10204.

de Vaux, A., Morrison, M., and Hutkins, R.W. (2002) Dis-placement of Escherichia coli O157:H7 from rumen

medium containing prebiotic sugars. Appl Environ Micro-biol 68: 519–524.

Wadolkowski, E.A., Burris, J.A., and O’Brien, A.D. (1990)Mouse model for colonization and disease causedby enterohemorrhagic Escherichia coli O157:H7. InfectImmun 58: 2438–2445.

Wilson, D.B. (2011) Microbial diversity of cellulose hydrolysis.Curr Opin Microbiol 14: 259–263.

Yatsuyanagi, J., Saito, S., and Ito, I. (2002) A case ofhemolytic-uremic syndrome associated with shiga toxin2-producing Escherichia coli O121 infection caused bydrinking water contaminated with bovine feces. Jpn J InfectDis 55: 174–176.

Supporting information

Additional Supporting Information may be found in the onlineversion of this article:

Fig S1. Growth curves of EDL933 incubated in BSICsamples (open circle) and M9-Glc (filled circle). EDL933 wasgrown at 39°C without shaking. The arrows indicate the timeat which the RNA samples were collected. Values are themean � 1 SEM of three independent experiments.Fig. S2. Relative expression of genes required for thecatabolism of mucus-derived carbohydrates during incuba-tion of the commensal E. coli strain BG1 in BSIC comparedwith M9-Glc. The ratio of mRNA level of each gene wasmeasured in BG1 incubated in filtered BSIC in comparisonwith cells grown in M9-Glc. RNA samples were collectedduring the exponential growth phase (grey), when the bacte-ria entered into the stationary phase (white) and during thestationary phase (black). Values are the mean � 1 SEM ofthree independent experiments.Fig. S3. Growth competition assays between the commen-sal E. coli BG1 and the EHEC strain EDL933 or isogenicmutants of EDL933. The BSIC-LEM samples were inoculatedwith a 1:1 mixture of the two strains. The EDL933 mutantswere defective for the pathway required for the catabolism offucose (EDL933DfucAO), galactose (EDL933DgalK), GlcNAc(EDL933DnagE), mannose (EDL933DmanA) or Neu5Ac(EDL933DnanAT). Bars represent the SEM of three inde-pendent experiments. The double asterisk denotes statisticalsignificance, P < 0.05 and the triple asterisk denotes statisti-cal significance, P < 0.01 as determined by Student’s t-testfor paired samples.Table S1. Genes involved in catabolism of mucus-derivedcarbohydrates by the EHEC strain EDL933.Table S2. Bacterial strains and plasmids used in this study.Table S3. Primer pairs used for gene cloning. The DNAsequence of restriction enzyme site is underlined. Start andstop codons of the cloned genes are in bold.Table S4. Sequence of primers used in relative mRNAquantification.



Carbohydrate utilization by enterohaemorrhagic Escherichia coli O157:H7 in bovine intestinal content

Documents