Binding of Escherichia coli to Fibronectin - The Journal of Biological ...

2 downloads 0 Views 4MB Size Report
and the binding sites on the fibronectin molecule are partly located. .... containing 10% fetal calf serum, 1% glutamine, 25 units/liter of penicillin and 25 pglliter of ...
Vol. 259, No. 23, Issue of Decernher 10, pp. 14899-14905,1984 Printed in U.S.A

THE.JOURNAL OF BIOLOGICAL CHEMISTRY 01984 by The American Society of Biological Chemists, Inc.

Binding of Escherichia coli to Fibronectin A MECHANISMOFTISSUEADHERENCE* (Received for publication, April 18, 1984)

Gunnar Froman$#, Lech M. SwitalskiS, Ahmed FarislI,Torkel Wadstroml, and MagnusHook$ll From the SConnectiue Tissue Laboratory, Uniuersity of Alabama, Birmingham, Alabama 35294 and the TDepartment of Bacteriology and Epizootology, Swedish University of Agricultural Sciences, Biomedicum, Box 583, S-751 23 Uppsala, Sweden

Four out of 17 enterotoxigenic strainsof Escherichia coli isolated from infantile diarrhea bound '''I-fibronectin. This binding, which was inhibited by unlabeled fibronectin but not by other proteins, appears to involve two classesof receptors, one of which binds the ligand reversibly. Consistent with thepresence of two classes of receptors the bacteria bound to at least two distinct sites of the fibronectin molecule, one being the amino-terminal domain which also contains the binding sites for Gram-positive bacteria and the other located outside this domain. The E. coli strain expressing fibronectin receptors adhered to fibroblasts and to fibronectin but not to ovalbumin-coated coverslips. Bacteria grown at40 "C did not express fibronectin receptors and did not adhere to either substrate. Saturation of receptors with fibronectin blocked adhesion to both fibronectin-coated coverslipsand to cultured fibroblasts. These data suggest that binding to fibronectin represents a mechanism of tissue adherence of E. coli.

-

Fibronectin is a large glycoprotein ( M , 440,000) found in the extracellular matrix of loose connectivetissueandin plasma. Fibronectin has beenshown to interact with both eucaryotic cells and some bacteria as well as with a number of connectivetissue molecules such as collagens, sulfated glycosaminoglycans, and hyaluronic acid. These interactions involve different distinct sites in the protein. A large portion of the primary structureof the protein has been determined, and the site binding to eucaryoticcells has been localized to a tetrapeptide (1). The major biological function of fibronectin is most likely related to its ability to mediate adhesion of eucaryotic cells to the extracellular matrix. Presumably fibronectin interacts withspecific receptors present on the surface of cells. These receptors have not yet beenidentified(for reviews on fibronectin see Refs. 2-4). Several pathogenic Gram-positive bacteria suchas staphylococci ( 5 , 6) andstreptococci (7, 8) were showntobind fibronectin. Receptorsfor fibronectin havebeen isolated from bacteria, and the fibronectin-bacteria interactionshave been

* This work was supported by Public Health Service Grants AM 27807 from the National Institutesof Health and Grants 16X-04723 and K84-13P-7001-01 from the Swedish Medical Research Council. The costs of publication of this article were defrayed in part by the payment of pagecharges. Thisarticlemusttherefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate thisfact. § Present address, Department of Bacteriology and Epizootology, Swedish University of Agricultural Sciences, Biomedicum, Box 583, S-751 23 Uppsala, Sweden. 11 Recipient of an Established Investigator Award from the American Heart Association. To whom correspondence should be addressed.

characterizedin some detail (9-13). Fibronectinbindsto streptococci and staphylococci with a much higher affinity than to eucaryotic cells. Binding to Gram-positive bacteria occurs preferentially at a site located in the amino-terminal domain of fibronectin (13,14),whereas the eucaryotic binding site is located in the center of the polypeptide chain. The amino acid sequence of the amino-terminal domain hasbeen determined both for bovine (15) and human (16) fibronectin. A recent study has indicated that fibronectinmay also associate with Gram-negative bacteria(17). The biological significance of the binding of fibronectin to bacteria is unclear. Fibronectin was suggested to act as an opsonin mediating the uptake of bacteria by phagocytic cells. However, when examined closer, this hypothesis was questioned (17, 18). Fibronectin has also been postulated to mediate the adhesion of bacteria to eucaryotic cells (19, 20) although conclusive evidence for this hypothesis hasnot previously been presented. In this communication we report that some strains of E. coli are capable of binding fibronectin. The binding of ' T labeled fibronectin to one high binding strainis characterized, and the binding sites on the fibronectin molecule are partly located. Furthermore, thehypothesis that binding of bacteria to fibronectin representsa mechanism of tissue adherence is analyzed. EXPERIMENTALPROCEDURES

Chemicals-Human fibronectin was prepared from freshly frozen plasma according to the method of Engvall and Ruoslahti (21) as modified by Miekka et al. (22) and labeled with iodine by the chloramine-T method (23). N a T (specific activity, 15 mCi/pg) was obtained from Amersham Corp., Percoll was from Pharmacia (Uppsala, Sweden). Trypsin (L-1-tosylamido-2-phenylethyl chloromethylketone-treated, 12,000 units/mg),fetuintype IV, bovine serumand chicken egg albumin, a-acid glycoprotein (orosomucoid), human immunoglobulin G, and phenylmethylsulfonyl fluoride were purchased from Sigma. Bacteria-Strains of Escherichia coli of different serotypes were originallyisolatedfrom infantile diarrhea (24). Strain selected for this study (B34289 c) belonging to serotype06:KZ:M was maintained in deep brain heart infusion agar (Difco) a t 4 "C. Bacteria were grown on CFA-agar (colonization factor antigen) medium (25) at 33 "C for 36 h, suspended in PBS' (0.14 M sodiumchloride, 0.02% sodium azide, 10 mM phosphate, pH 7.4) washed, and resuspended in PBS to a cell density of 8 X lo9 cells/ml. Bacterial suspensions were kept frozen a t -70 "C without significant loss of binding activity for several months. For a matter of comparison Staphylococcus a u r e u cells were used in one series of experiments. 5'. aureus strain Cowan 1 to which binding of fibronectin has been characterized earlier (11)was grown in tryptic soy broth (BBL/Microbiology Systems, Cockeysville, MD) at 37 "C for 18 b, centrifuged, and washed with PBS. Staphylococcal cells were killed a t 88 "C for 20 min, washed, and suspended in PBS to a density of 10'" cells/ml. Binding Assay-Fibronectin binding assay was performed essen-

14899

The abbreviation used is: PBS, phosphate-buffered saline.

14900

E. coli-Fibronectin Binding

tially as described earlier (11).Briefly, 8 X 10' cells were incubated with 5 X lo4 cpm of 1251-fibronectin(specific activity, 2 X lo6 cpm/ pg) in PBS containing 1 mg/ml bovine serum albumin in a total volume of 0.5 ml. After end-over-end incubation (for 1 h, unless otherwise stated), a 40O-pl sample was added to 0.5 ml of PBS layered on top of 3 ml of 10% Percoll in PBS (density, 1.020 g/ml) and centrifuged at 1350 X g for 15 min. After aspiration of the supernatant the radioactivity associated with the pellet was quantitated in a y counter (LKB Wallac, Turku, Finland). Background values were subtracted. All incubations were done in duplicates. Identification of Tryptic Fragments of Fibronectin Binding to Bacteria-Labeled fibronectin (250 pl, 0.1 pg, 5 X 1O'cpm) in 0.1% bovine serum albumin in PBS was incubated with 0.1 pg of trypsin for 1 h at 20 "C. The digestion was stopped by adding 2.5 plof a freshly prepared 0.1 M phenylmethylsulfonyl fluoride solution in 95% ethanol. Fifty p1 of the labeled degradation products were incubated with 50 ~1 of S. aureus Cowan 1 (5 X 10' cells) or 50 pl of E. coli (4 X 10' cells) for 1 h at 20 "C, followed by centrifugation at 1350 X g for 15 min. The pellet was washed once with 0.1% bovine serum albumin in PBS. Proteins bound to bacterial cells were released by boiling for 1 min in the electrophoresis sample buffer containing 4% sodium dodecyl sulfate, 2 M dithiothreitol, 30% sucrose, 0.01% bromphenol blue, 82 mM Tris-HC1, pH 8.8. Following centrifugation the supernatant was subjected to electrophoresis in a 5-15% polyacrylamide gradient gel (26). The gel was stained with Coomassie Brilliant Blue R-250, dried (27), and autoradiograpbed using X-Omat AR film (Kodak, Rochester, NY). Amino-terminal Domain of Fibronectin-Fibronectin was digested with plasmin-Sepharose at 37 "C for 20 h (28) and dialyzed against 50 mM NH4HC03. The sample was applied on a DEAE-Sephacel column equilibrated with 50 mM NH4HC03.The nonbound material was collected, freeze dried, and dissolved in 1 M acetic acid. The sample was applied on a Sephadex G-100 column (1 X 100 cm) in 1 M acetic acid and chromatograpbed. The chromatogram showed one minor peak at the void volume of the column and a major peak with a K., of0.3 containing the 29,000 amino-terminal domain. The purified material showed a single band upon SDS-gel electrophoresis migrating as a24,000 protein when unreduced and as a29,000 protein when it was reduced and alkylated. The amino acid composition of this preparation was verysimilar to theone reported by Garcia-Pardo et al. (16). Labeled Bacteria-Bacteria were grownfor 40 h at 30 "C in a liquid CFA medium containing 1 nM [35S]methionine (Amersbam Corp., specific activity 1400 Ci/mmol). Bacteria were rinsed with PBS, suspended in a modification of Eagle's medium (Flow Laboratories, McLean, VA) without serum, and their density was spectrophotometrically adjusted to 10' cells/ml. The suspension of labeled bacteria was kept at -70 "C. Fibronectin-coated Couerslips-Glass coverslips (18-mm diameter) were incubated in a solution of fibronectin (200 pg/ml) for 1 h at 20 "C followed by incubation in a solution of ovalbumin in PBS (1 mg/ml), rinsed with PBS, and used in the attachmentassay. Control coverslips were coated with ovalbumin alone. The coverslips were not allowed to dry. Fibrobhts-Rat embryonic fibroblasts obtained from Flow Laboratories (McLean, VA) were seeded on sterile glass coverslips (18mm diameter) submerged in the a modification of Eagle's medium containing 10% fetal calf serum, 1% glutamine, 25 units/liter of penicillin and 25 pglliter of streptomycin and allowed to grow for 35 days a t 37 "C. Coverslips were subsequently rinsed and stored in serum-free medium. Attachment Assay-Coverslips coated with proteins or with confluently grown fibroblasts were used throughout the study. Radioactively labeled bacteria cells (300 pl; density, 1 X 10' cells/ml) were incubated on the slips at 20 "C. Following the incubation the coverslips were thoroughly rinsed with PBS andtransferred to scintillation vials for counting. In one series of experiments the bacteria were preincubated end-over-end for 1 h in the presence of proteins (200 pg/ml) to be tested as inhibitors of bacterial attachment, and this suspension was applied on the surface of the coverslips as described. Electron Microscopy-The specimens were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.3, for 2 h at 20 "C, rinsed, and kept at 4 "C in 0.1 M cacodylate buffer, pH 7.3, containing 7% sucrose. Specimens were dehydrated through a series of ethanol baths, critical point dried, coated with gold-palladium in a high-vacuum evaporator (HUS-4, Hitachi Ltd., Tokyo, Japan), andexamined in a

ISI-100 scanning electron microscope operated at 25kV Seisakusho, Tokyo, Japan).

(Akashi-

RESULTS

Initial screening of a collection enterotoxigenic of E. coli strains isolated from human infantile diarrhea (24) showed that 4 out of 17 strains bound '251-fibronectin.The strain exhibiting the highest binding capacity (strain B 34289 C) was selected for further studies. Characterization of Fibronectin Binding-E. coli cells bound lZ5I-fibronectininatime-dependentmanner (Fig. 1). This reaction was rapid, and a plateau of bound fibronectin was reached within 15-20 min. Continuing the incubation up to 4 h did not affect the amount of protein bound. In thefollowing experiments an incubation time of 1 h was chosen to ensure maximum binding. Binding of '251-fibronectintook place over a broad pH range with a maximum at pH6.0 (data not shown). A t pH 7.4 which was chosen in the following experiments the binding was 90% of the maximum. Quantitation of "'I-fibronectin binding as a function of increasing concentration of radiolabeled ligand yielded a hyperbolic curve (Fig. 2), indicating that thebacteria contain alimited number of fibronectin receptors. A Scatchard plot (29) of the binding data gave a biphasic curve (Fig. 2) which suggested the presence of more than one type of binding site. If we assume the presence of two classes of fibronectin receptors, we can calculate Kd for one class of receptors (class A) to 3 X M and for the second class of receptors (class B) to 4 x lo-' M. The number of binding sites/cell was estimated to be 1 x lo4 for class A and 5 X lo3for class B, respectively. Reversibility of binding, which is a prerequisite for Scatchard plot analysis, was tested by adding increasing amounts of unlabeled fibronectin to displace bacteria-bound "'I-fibronectin. The amount of added unlabeled fibronectin exceeded the amount of labeled protein by a factor of 200 to 2000. A maximum of 59% couldbe displaced by 100 pg of fibronectin (Fig. 3). These data indicate that a portion of the bacteria-bound fibronectin can be displaced within 1 h by an excess of unlabeled protein and, therefore, is reversibly bound to the E. coli cell. On the other hand some of the '251-fibronectinwas not displaceable; hence it appears to be bound irreversibly to bacterial receptors. Addition of fetuin, orosomucoid (not shown) and IgG (Fig. 3) did not displace the bacteria-bound '251-fibronectindem-

I I

h

15

10

5

I

2

4

TIME ( hours )

FIG. 1. Kinetics of binding of '261-fibronectinto E. coli strain B 34289 c. Bacteria were incubated with labeled fibronectin (see "Experimental Procedures") for the indicated periods of time. Presented data refer to 400-pl samples containing 6.4 X 10' bacteria.

E. coli-Fibronectin Binding

14901 TABLE I

Specificity of'261-fibronectinbinding by E. coli strain B 34289 c Bacteria were incubated with '251-fibronectinin the presence of indicated concentrations of proteins for 1 h, and theamount of bound fibronectin was determined as described under "Experimental Procedures." Data are expressed as percentages of control (bacteria incubated with '=I-fibronectin but without additional unlabeled proteins).

W W

a

,

LL

- ."...

n

Pmtein

3 0

added/0.5

z m

Percentage of 'mI-fibronectinbound by bacteria incubated in the presence of FibreOvalbumin Fetuin OrosoHuman nectin mucoid I&

23

10 ag 100 ae

I

93 81

103 89 94

100 94

78 ~~~

BOUND ( p Y )

I

I

20 FIBRONECTIN ADDED

(

pg 1

FIG.2. Saturability of binding of '*'I-fibronectin to E. coli strain B 34289 c. Bacteria were incubated in the presence of increasing concentrations of labeled fibronectin (specific activity, 4100 cpm/pg). Data refer to the amount of fibronectin bound by 6.4 X 10' bacteria. Background values were determined for each concentration of added 1251-fibronectin and subtracted from the'incubations containing bacteria. Inset, Scatchard plot analysis of the above data. The binding data for kinetic analysis were calculated by anonweighted linear regression computer program.

200 K

I,

20

30

50

40

PROTEIN ADDED

(

3

IO0

ng 1

FIG.3. Displacement of ""I-fibronectin with unlabeled proteins. Bacteria were preincubated with lZSI-fibronectinin the standard conditions (see "Experimental Procedures") followed by 1-h incubation in the presence of an excess of unlabeled fibronectin (O), unlabeled 29.000-fragment of fibronectin (O),or unlabeled immunoglobulin G (A). The radioactivity associated with the bacteria was determined, and the data are expressed as percentages of control, i.e. incubations in the absence of unlabeled proteins. onstrating thespecificity of the displacement reaction. Specificity of the binding of '251-fibronectinto E. coli cells was also demonstrated in experiments where an excess of unlabeled fibronectin or other proteins was added together with the labeled fibronectin (Table I). An almost complete (94%) inhibition of binding was observed with 100 pgof unlabeled fibronectin while the inhibition by other proteins did not exceed 22%. Localization of the Binding Site in the Fibronectin Mokcuk-In order to identify domains inthe fibronectin molecule participating in binding to the E. coli strain, peptides gener-

-

"

I16K94 K

-

43 K 67 K

30 K

-

20. I K

-

14.4 K

IO

2

50

30

IO

-

FIG.4. Gel electrophoresis of fibronectin fragments binding to s.aureus strain Cowan I or to E. coli strain €3 34289 c. Bacteria were incubated with '251-fibronectinfragments for 1 h. The mixtures were then centrifuged, and the bacterial pellet was boiled in the presence of buffer containing sodium dodecyl sulfate and dithiothreitol. Material released from bacterial cells was analyzed by gel electrophoresis. The gel was dried and subjected to autoradiography. Lane 1, tryptic fragments of '251-fibronectin;lane 2, fragments bound by S. aureus; lane 3, fragments bound by E. coli. Numbers in the left margin indicate the molecular weights and migration distances of the standard proteins. ated by trypsin digestion of lZ5I-fibronectinwere incubated with bacteria. Subsequentlybacteria were centrifuged and bound peptides were solubilized by boiling bacteria with 4% sodium dodecyl sulfate for 1 min. Released peptides were analyzed bygel electrophoresis (Fig. 4). Fora matter of comparison, fragments binding to S. aureus strain Cowan I were also analyzed, and this strainbound only a M , = 29,000 fragment, presumably the amino-terminal domain (13). The E. coli strain bound fragments with a molecular weight of about 150,000 as well as small amounts of the M , = 29,000 domain. Thus, the E. coli strain appears to bind to at least two domains inthe fibronectin molecule, one being the aminoterminal. This observation was confirmed in an experiment where a purified M , = 29,000 amino-terminal fragment was tried as aninhibitor of binding of '251-fibronectinto bacteria. A complete inhibition of binding to S. aureus strain Cowan I was obtained with 5 pg of the M , = 29,000 fragment (data not

E. coli-Fibronectin Binding

14902

shown in Fig. 3 addition of an unlabeled M , = 29,000 fragment did not displace '251-fibronectin already bound to bacterial 1 cells. " Fibronectin Binding as a Mechanism of Bacterial Adher0 2 ence-The possibility that binding of bacteria to fibronectin 0" represents amechanism of tissue adherence was investigated. m Two model systems were used; fibronectin-coated coverslips z 50 Ior fibroblasts (which produce a fibronectin-rich extracellular o W matrix) grown on coverslips. Bacterial suspensions were z 0 placed on the coverslips and incubated at room temperature. a m Loosely bound bacteria were removed by extensive washing, LL and substrate-attached bacteriawere quantitated or examined by scanning electron microscopy. 5 IO 15 20 Initial experimentsshowed that theE. coli readily attached to coverslips coated with fibronectin but not to coverslips PROTEINADDED ( pg ) coated with ovalbumin (Fig. 7 , a and b). By using [35S] FIG. 5. Inhibition of binding of '261-fibronectinby an excess methionine-labeled bacteria it was possible to quantitate adof unlabeled fibronectin or its M. 29,000 amino-terminal domain. Increasing amounts of unlabeled fibronectin (0)or its hering bacterial cells. The time dependence of attachment of amino-terminal domain (0)were added together with lZ51-fibronectin bacteria to fibronectin-coated glass was linear up to at least to bacteria, incubated for 1 h, and the amount of bound fibronectin 60 min (Fig. 8a). During this incubation time essentially no was determined as described under "Experimental Procedures." Data bacteria attached to the coverslip coated with ovalbumin. A are expressed as in Fig. 3. similar series of experiments was also performed using fibroblasts as a substratum for bacterial adherence. Attachment of E. coli to cultured fibroblasts was rapid, and the process was essentially completed within 20 min (Fig. 8b). Scanning electron microscopy showed that a substantial number of bacteria adhered to the fibroblasts (Fig. 9a). Higher magnification (Fig. 9c) of bacteria attachingto fibroblasts showed that bacteria did not attach randomly to the surface of fibroblasts but seemed to be lined up along fibrils which may represent fibronectin fibers of the pericellular matrix. In initial experiments the expression of the fibronectin binding components was found to be dependent on the growth temperature and was maximal for bacteria grown at 33 "C. Binding of fibronectin by bacteria grown at 40 "C did not exceed 2% of the binding by bacteria grown at standardconditions. Attachment of bacteria grown at 40 "C was very low to both fibronectincoated coverslips and tofibroblast cultures (data not shown) N K which correlates to thelack of fibronectin receptors. 100 200 300 400 500 1000 Inhibition of Attachment-If the adherence of E. coli to the 29K FIBRONECTIN FRAGMENT ADDED ( ng 1 fibroblast cultures is mediated only through the fibronectin FIG. 6. Saturability of binding of '261-labeledM , 29,000 receptors then it should be possible to inhibit adherence by (29K) fibronectin fragment to E. coli strain B 34289 c. Inset, blocking the fibronectin receptors. To test this hypothesis, we Scatchard plot analysis of the above data. Experimental conditions measured the adherence of bacteria preincubated with differare as in the legend to Fig. 2; specific activity of the labeled 29,000 ent proteins. Bacteria preincubated with fibronectin attached fragment was 27,000 cprnlrg. neither to fibronectin-coated coverslips nor to fibroblast cultures, as shown by scanning electron microscopy (Fig. 7c and shown). The inhibition of binding to the E. coli strain was 8b) or by determining the number of adhering bacteria (Fig. only 30% as compared to an almost complete inhibition by 8, a and b, and Table 11). Pretreatment of bacteria with intact fibronectin (Fig. 5 ) . The molar concentration of the M , proteins other than fibronectin did not affect the adherence = 29,000 fragment was 16 times higher than thatof fibronecof bacteria to cultured fibroblasts (Table 11) but somewhat tin. These data indicate that theE. coli strain used recognizes surprisingly, preincubation with fetuin resulted in a 30% the amino-terminal domain of fibronectin but that themajor decreased adherence of bacteria to fibronectin-coated glass. binding site in the fibronectin molecule is located outside this The proteins tested did not compete with '251-fibronectinfor domain. binding to bacteria (see Table I). These data suggest that Incubation of bacterial cells with isolated lz5I-1abeledM , = binding of bacteria to fibronectin represents a mechanism of 29,000 fragments resulted in a binding of the labeled peptide tissue adherence and if this interaction is blocked, bacteria to the bacteria (Fig. 6). By increasing the amounts of labeled are no longer capable of adhering to the extracellular matrix ligand the binding was shown to be saturable, and Scatchard of cultured fibroblasts. plot analysis of the binding isotherms revealed the presence of only one class of binding sites. An apparent K d of 8 X DISCUSSION M was calculated for the binding reaction which is of the same Recent studies in our and other laboratories have shown order as theKd of the binding of intact fibronectin to class B receptors described above (Fig. 2). The number of binding that a number of pathogenic Gram-positive bacteria bind sites per one bacterial cell for the M , = 29,000 fragment was fibronectin, whereas it has generally been believed that Gramestimatedto be 7 X lo3 compared to 5 X IO3 for class B negative bacteria do not have fibronectin-specific receptors. receptors obtained by using intact fibronectin as a ligand. As Our data indicate that the ability to bind fibronectin is not h

100

Binding E. coli-Fibronectin

14903

N

E E

\

0

w X

o a

I-

t

a

5 a

W I-

o

a

m

40

20

TIME

(

60

min 1

N

E E

I

\

0 W

x

o a

1000

-b

0

I-

I-

a 5 a

W I-

o a m

20

A

A

I

I

40

60

FIG.8. Kinetics of attachment of E. coli strain B 34289 c to different substrata.Suspensions of [:'sS]methionine-labeled bacteria were placed on the surface of: a, glass coverslips coated with fibronectin (0)or ovalbumin (0);b, rat embryonic fibroblasts (0). Bacteria preincubated for 1 h with fibronectin (200 pg/ml) were also tested for attachment to fibronectin-coated glass (A) or to fibroblasts (A).

FIG. 7. Attachment of E. coli strain B 34289 c to proteincoated glass (scanning electron microscopy). Bacteria were incubated for 1 h on the coverslips coated with ( a ) fibronectin or ( b ) ovalbumin. (c) Bacteria were preincubated with fibronectin (200 pg/ mi) and then incubated onthe fibronectin-coated coverslip. Bars indicate 10 pm.

an uncommon phenomenon among strainsof E. coli isolated from infantile diarrhea. Binding of fibronectin to strainB 34289 c characterized in this study isspecific since excess amounts of unlabeled fibronectin almost completely inhibited the binding of Iz5I-fibronectin whereas other proteins tested had nolow or inhibitory effect. The E. coli cells appear to have two classes of fibronectin receptors which bind fibronectin with apparent dissociation constants of 3 X lo-" M (class A) and 4 X lo-' M (class B), respectively. When, in fact, fibronectin was used as a ligand the number of receptor molecules/cell was calculated

to be 1 X IO4 (class A) and 5 x 10* (class B), respectively. Since only a portion (60%) of the bacteria-bound '251-fibronectin canbe displaced by a subsequent additionof unlabeled fibronectin it islikely that class B receptors bind fibronectin functionally irreversibly whereas the class A receptors interact reversibly with fibronectin. At least two distinct sites in the fibronectin molecule are recognized by the bacterial receptors. One of these binding sites located is within the aminoterminal region of the protein. Since added isolated aminoterminal fragments did notdisplace bacteria-bound '*'I-fibronectin it is likely that the class B receptors are interacting with this domain. Binding of the amino-terminal fragment to bacteria engaged a maximum of 7 X lo3receptor molecules/ cell which bound the ligand with an apparent K d of 8 X lo-' M. These values are reasonably similar to those estimated for as a ligand. The the classB receptors using intact fibronectin Kd values and number of receptorsreported herefor the different ligands should be regarded as estimates. Because of uncertainties in thespecific activities of the ligands, nonspecific binding and possible heterogeneity in the bacterial cell population precise values cannot be determined at thisstage. Theaffinities of the E. coli receptors forfibronectin are similar to those reported for fibronectin receptors S. aureus on (10) and Streptococcus pyogenes (13). These Gram-positive bacteria bind irreversibly to the N-terminal domainof fibronectin similarly to the interactionbetween class B receptors of E. coli and fibronectin. Earlier studies on tissue components involved in E. coli

E. coli-Fibronectin Binding TABLEI1 Specificity of adherence of E. coli strain R 34289 c to rat embryonic fibroblasts and fibronectin-coatedglass Bacteria labeled with ["S]methionine were preincubated with proteins, 200 pglml, for 1 h and then incubated on fibroblasts or fibronectin-coated glass for 40 min. Data are expressed as percentages of attachment relative to control (untreated bacteria). Substratum

Per cent of bacteria attached after pretreatment with Fetuin

FibroOvalnectin bumin

C)zO;;

Human IRC ~~

Fibroblasts Fibronectincoated glass -

"

8 11

95 99

113 111 69

101 94

96

adhered to fibronectin-coated glass. This adherence was specific since (a) it did not occur with ovalbumin-coated glass and ( b ) it wasabolished by pretreatment of bacteria with fibronectin but not with other proteins. The E. coli strain adhered also to cultured fibroblasts although the number of attached bacteria was somewhat lower as compared to fibronectin-coated glass. This difference could possibly be ascribed to the different amountof fibronectin available for bacterial attachment on the two surfaces. Bacteria grown under conditions when fibronectin receptors were not expressed did not adhere to fibroblast cultures or to fibronectin-coated coverslips. Furthermore, preincubationof the E. coli strain B34289 c with solublefibronectin but not with other proteins inhibited bacterial adherence to fibroblasts which indicates that binding to fibronectin is indeed an important mechanism of tissue adherence. The relative importance of this phenomenon in the etiology of E. coli diarrhea is unclear. It is not known if fibronectin is available in the linings of the intestine. Intestinal epithelial cells grown in vitro appear to contain fibronectin exclusively in regions of cell-to-cell contact (33), and it is probable that only small quantities of fibronectin are present in vivo on luminal aspects of these cells. Fibronectin has also been demonstrated on epithelial cells of rectal mucosa by ultrastructural immunohistochemistry (34). Even if the exposure of fibronectin in the lumen of healthy intestine is questionable, the lysis or damage of the epithelium would expose fibronectinpresent in the basal matrix that could serve as a substratum for bacterial adhesion. Acknowledgments-We gratefully acknowledge the expert assistance of Francis R. Denys in the electronmicroscopy. Sharon Brackner in preparing the manuscript, andDr. S. Barnes in computer analysis of binding data.

PIC:.9. Attachment of E. coli s t r a i n B 34289 c to rat embryonic fibroblasts (scanning electron microscopy). a and c, hacteria were incubated for 1 hon the fibroblasts grown on glass coverslips. h, bacteria were preincubated for 1 h with fibronectin (200 pg/ml) and then incubated on the fibroblasts. Bars indicate 10 pm ( a and h ) or 2 pm (c).

adherence havebeen focused oncarbohydrate-containing molecules of eucaryotic cell membranes. In the case of some pyelonephritic strains of E. coli the structure of the sugars, interacting with the bacteria, has been determined (30, 31). Connective tissue proteinswith the exception of laminin (32) have not previously been considered as substrates for E. coli adherence. Our observation that some strains of E. coli express receptors for fibronectin prompted us to investigate the possible importance of this interaction in the adherence of bacteria to eucaryotic cells. The E. coli strain chosen for thisstudy

REFERENCES 1. Pierschbacher, M. D., and Ruoslahti, E. (1984) Nature (Lond.) 309,30-33 2. Ruoslahti, E., Engvall, E., and Hayman, E. G . (1981) Collagen Relat. Res. 1.95-128 3. Hynes, R. O., and Yamada, K. M. (1982) J. Cell Riol. 95, 369377 4. Yamada, K. M. (1983) Annu. Rev. Riochem. 5 2 , 761-799 5. Kuusela, P. (1978) Nature (Lond.)276, 718-720 6. Switalski, L.M., R y d h , C., Rubin, K., Ljungh, A., Hook, M., and Wadstrom, T. (1983)sInfect.Immun. 42,628-633 7. Switalski, L. M., Ljungh, A., Ryden, C., Rubin, K., Hook, M., and Wadstrom, T.(1982) Eur. J. Clin. Microbiol. 1, 381-387 8. Myhre, E. B., and Kuusela, P. (1983) Infect. Immun. 40, 29-34 9. Espersen, F., and Clemmensen, I. (1982) Infect. Immun. 37,526531 10. Proctor, R. A., Mosher, D. F., and Olbrantz, P. J. (1982) J. Biol. Chem. 257, 147S14794 11. R y d h , C., Rubin, K., Speziale, P., Hook, M., Lindberg, M., and Wadstrom, T. (1983) J. Riol. Chem. 258,3396-3401

E. coli-Fibronectin Binding 12. Courtney, H. S., Simpson, W.A, and Beachey, E. H. (1983) J. Bacteriol. 1 5 3 , 763-770 13. Speziale, P., HGk, M., Switalski, L.M., and Wadstrom, T. (1984) J. Bacteriol. 1 5 7 , 420-427 14. Mosher, D. F., and Proctor, R. A., (1980) Science (Wash. D. C.) 209,927-929 15. Petersen, T. E., Thfigersen, H.C., Skorstensgaard, K., VibePedersen, K., Sahl, P., Sottrup-Jensen, L., and Magnusson, S. (1983) Proc. Natl. Acad. Sci. U. S. A. SO, 137-141 16. Garcia-Pardo, A., Pearlstein, E., and Prangione, B. (1983) J.Biol. Chem. 258, 12670-12674 17. Van De Water, L., Destree, A. T., andHynes, R. 0. (1983) Science ( Wash. D.C.) 220, 201-204 18. Verbrugh, H. A., Peterson, P. K., Smith, D. E., Nguyen, B. T., Hoidal, J. R., W i l k k o n , B. J., Verhoef, J., and Furcht, L. T. (1981) Infect. Immun. 33,811-819 19. Abraham, S. N., Beachey, E. H., and Simpson, W. A. (1983) Infect. Immun. 4 1 , 1261-1268 20. Simpson, W. A., and Beachey, E.H. (1983) Infect. Immun. 3 9 , 275-279 21. Engvall, E., and Ruoslahti, E. (1977) Int. J. Cancer 2 0 , 1-5 22. Miekka, S. I., Ingham, K. C., and Menache, D. (1982) Thromb. Res. 27, 1-14 23. Hunter, W. M. (1978) in Handbook of Experimental Immunology

14905

(Weir, D. M., ed) pp. 14.1-14.40, Blackwell, London 24. Faris, A., Wadstrom, T., and Freer, J. H. (1981) Curr. Microbiol. 5,67-72 25. Evans, D. G., Evans, D. J., Jr., and Tjoa, W. (1977) Infect. Immun. 18,330-337 26. Blobel, G., and Dobberstein, B. (1975) J. Cell Biol. 67,835-851 27. Wallevik, K., and Jensenius, J. C. (1982) J. Biochem. Biophys. Methods 6, 17-21 28. Rubin, K., Johansson, S., Pettersson, I., Ocklind, K., Obrink, B., and Hook, M. (1979) Biochem. Biophys. Res. Commun. 91,8694 29. Scatchard, G . (1949) Ann. N. Y. Acad. Sci. 51, 660-672 30. Kallenius, A., Mollby, R., Svenson, S. B., Winberg, J., Lundblad, A., Svensson, S., and Cedergren, B. (1980) FEMS Microbiol. Lett. 7 , 297-302 31. Svanborg-Ed&, C., Freter, R., Hagberg,L., Hull, R., Hull, S., Leffler, S., and Schoolnik, G. (1982) Nature (Lond.) 298,560562 32. Speziale, P.,Hook, M., and Wadstrom, T. (1982) FEBS Lett. 146,55-58 33. Quaroni, A., Isselbacber, K. J., and Ruoslahti, E. (1978) Proc. Natl. Acad. Sci. U. S. A. 75, 5548-5552 34. Scott, D. L., Morris, C. J., Blake, A. E., Low-Beer, T. S., and Walton, K. W. (1981) J. Clin. Pathol. (Lond.) 3 4 , 749-758