Plasmid-Encoded Outer Membrane Protein YadA Mediates Specific Binding of Enteropathogenic Yersiniae to Various Types of Collagen

HENDRIK SCHULZE-KOOPS,l HARALD BURKHARDT, I JÜRGEN HEESEMANN,2 KlAUS VON DER MARK,! AND FRANK EMMRICHh

Max-Planck-Society, Clinkal Research Unit tor Rheumarology/lmmunology ut the Institute tor C/inicaflmmuno/ogy 01 the University Erlangen-Nümberg, D-8520 Erlangen, I and Institute

01 Hygiene and Microbiology, University Hospital, D-87oo Wünburg,2 Gennany

Received 9 December 1991/Accepted 6 March 1992

Tbe plasmid-enooded outer membrane protein YadA of eOleropalhogeoic yersiniae is associated with pathogeDicily. Recently, collagen binding ofYadA-positive yenioiae was reponed witboul detailed charaeter­izalioD (L. Emödy, J. HeuemaoD, H. Wolf·Watz, M. Skurnik, G. KappeNd, P. O'Toole, and T. Wadström, J. Baderiol. 171:6674-6679, 1989). To eluddate tbe nature of collagu bioding 10 YadA, we used 11

recombinanl Yel'$inia stnin expressing the cloned VadA gene. Direc:t binding or VadA-posiliYe yersiniae 10 coIlageßs was demoßstrated iß amnity blot experiments Oß nitrocellulose filters. A spe<:trom or collagen types in a wide conceßtralion range were tested ror their ability to block binding or 115I_labeled collagen type 11 to VadA-positiYe yersiniae. The results indicate a spe<:ifie binding slte(s) ror VadA in eollageß types I, 11, 111, IV, V. and XI. la contrast, collagen type VI did not bind to VadA. To eharacterize the binding site(s) more prec::isely, isolated collagen ehains and cyanogen bromide rragments were inyesligated. These studies reyealed that binding orVadA to collagen type I is conftned to the ed(l) ehain, whereas the binding sile within collagen type XI is localized in the w(XI) ehaln. a2(I). ed(XI). and a2(XI) did not bind to VadA. Most interestlngly. in the 0.1(11) ehain the specific binding site ror VadA resides In the cyBnogen bromide rragment CBIO. The laller might indieate a binding site that does not depend on conronnation. Based on these ftndings, rurlher rragmentation and the synthesis or peptides may allow definition or the peptide sequenee(s) relevant ror YadA binding.

Yersinia enterocolitica and Yel'$inia pseudotuberculosis are enteropathogenic for humans (4, 7). The disease spec­trum comprises both intestinal and extraintestinal manifes­talions involving the connective tissue, such as reactive arlhritis, erythema nodosum, and different types ofvaseuli­tis (4, 7, 38). Recently, it was demonstrated that a plasmid of about 70 kb is necessary for full virulence expression in yersiniae (for reviews, see referenees 6, 15, 29, and 41). Different plasmid-encoded proteins, incJuding outer mem­brane proteins with palhogenic funelions, have been de­scribed (2, 3, 6, 13, 14). One of the mosl prominent plasmid­encoded ouler membrane proteins is the Yersinia adhesin VadA. It forms multimeric fibrillae on the surfaee of entero­pathogenic yersiniae (25). There is plenty of evidence that YadA is involved in resistance to complement lysis, phago­cytosis, and adherenee to mammalian cells (14, 16). Binding of collagen types I, 11, and IV 10 YadA-positive yersiniae has heen reported (8). This property might be relevant for the pathogenesis of diseases induced by enleropalhogenic yer­siniac_ However, Ihe specificity of the collagen-YadA inter­action still remains obseure, sinee the reported competition experiments were performed al a 400- to I,OOO-fold molar excess of the unlabeled inhibitor. This prompted us to investigate the nature of collagen binding by YadA in more detail. In a first series of experiments we tested direct binding of YadA-positive yersiniae to a broad speetrum of collagens immobilized on nitrocellulose filters_ We dcmon­strale here that YadA expression mediates binding of yers­iniae 10 all collagens investigaled exccpt collagen type VI. In blocking experiments, different collagens in a wide concen-

• Corresponding aUlhor.

Iralion range were tested for their ability 10 inhibit binding of Ihe main cartilage collagen, collagen type 11 , to YadA­positive yersiniae; blocking occurred at pieomolar concen­trations. The experiments indieate a speeific high-affinity binding site(s) for YadA on many collagen types. Conse­quently, the analysis was exlended to assign the putalive binding site to either isolated collagen chains or cyanogen bromide fragments.

MATERIALS AND METIlODS

Bactenal strains. Three Y. enterocolitica strains of sero­type 0:5 were used in Ihis study. Strain NFO is of environ­menlal origin, plasmid free, and YadA negative. Strain NF-pCB9: :Tn5 harbors the virulenee plasmid of Y. entero­colitica serotype 0:9 with an inaetivated YadA gene. Strain NF-pRK290B9-4 harbors a hybrid plasmid eonsisting of the mobilizable vector pRK290B (16) and a BamHI fragment of the virulcnce plasmid of Y. enterocolilica 0:9 encoding YadA and is otherwise homologous to NFO; il has been demonslraled that the vector pRK290B does not encode collagen binding faetors (8). All strains were described previously (8, 14, 16, 17). Baeteria were grown in Luria· Bertani bouillon at 37"C for 24 h, heal inactivaled al 6O"C for 45 min, and stored in phosphale-buffered salinc (PBS}-0.01% NaN) (pH 7.3) al 4°C.

Collageo preparatioo. Human type I and V collagens from plaeental tissue werc prepared as deseribcd by Benlz et al. (1) and kindly provided by K. Kühn, Martinsried, Germany. Type 11, IX, and Xl collagens were prepared from the ehieken xyphoid process by extraetion with guanidinium hydrochloride for removal of proteoglycans, digestion with pepsin, and fractional sall precipitation as deseribed by von

ocr Mark el al. l4UJ. I ne IDur-slcp pr<K:courc 0 1 I.;ouagen type IV preparation from human placcnta was performed as dc.scribcd by Glanville and Rauter (12) and included pepsin solubilization of tissuc., fractional salt precipitation of solu· bi lized collagens, native reduction and carboxymethylation, a second pepsin treatment, and precipitation of high·molee· ular-weight materials out of the denatured sampie with 3 M NaO. Type 111 collagen was prepared from fetal bovinc skin. Extraction of collagen from the skin, precipitation by am­monium sulfate and NaO, dialysis against 2 M urea contain­ing NaCI-Tris buffer, and separation of collagen type 111 with DEAE-cellulosc columns were pcrformed as described by Timpl et al. (37). Pepsin-solubilized collagen type VI from human placenta was purified as describcd by von der Mark et al. (39).

Proteoglycans. Proteoglycans were prepared by extract ion with I M NaCI of human fetal cartilage followed by DEAE chromatography (23). Thcy were kindly provided by T. Kirsch, MPG-Arbeilsgruppen, Erlangen, Germany.

CN Rr deavage and CR rngment purification. Sampies of 50 mg of collagen type 11 were cleaved with CNBr, and the cyanogen bromide fragments (CB fragments) were purified as recently describcd by Burkhardt et al. (5). Brießy, colla· gen type II was dissolved in 70% HCOOH (10 ml), warmed to 50"C to ensure denaturation, and ßushed with nitrogen. Then CNBr was added in a l00-fold molar excess ovcr methionine rcsidues (1.5 mgfmg of collagen) and allowed 10 react for 4 h at 37"C. For furthe r purification the CB fragments were subjected to fast protein liquid chromatog­raphy in 0.1 M ammonium acetate (pH 5) on a Supcrosc 12 column. Homogencity of the peptides was controlled by S<XIium dodecyl sulfate (SDS)-gcl electrophoresis in 18% acrylamide min igels with the buffeT system of Laemmli (26).

Isolation or 0:1(1) and 012(1). al( l) and 02(1) collagen chains were separated by chromatography on carboxymethyl cellu· lose at 4rC in 0.04 M sodium acetate (pH 4.8) containing 4 M urea as describcd previously (40). The pUTity of the chains WeiS controlled by SDS·polyacrylamide gcl electrophoresis (PAGE) in 7% acryl amide minigels with the buffer systcm of Laemmli (26).

Radiolabeling or collagen type 11. Collagen type 11 was radiolabeled by the iodogen mcthod (10), and a specific activity of 2 x 10" to 3 x IIY' cpml..,.g was dctcrmincd.

Binding or radiolabeled collagen type 11 to yersiniae and blocking experiments. Increasing amounts of 12~ 1 _labc led collagen type JI (2.7 x Ht cpmJ..,.g) were incubated with 100 ..,.1 of particulatc bacterial suspension (2 x 10" bactcria) for 75 min at room temperature under gentle agitation. Then I ml of PBS-O.01 % Tween 20 was addcd. and the suspension was cenlrifuged (30 min, urc, 7,500 x g). Thc supcmatant was aspiratcd, and Ihe radioactivity associated with the pellet was determined in a gamma counter (COBRA 5005; Canbcrra and Packard Instruments). YadA-negative yersin­iae were uscd as a control for unspecific binding of collagen type 11 to yersiniae. In the blocking assays, 100 ng of radiolabclcd collagen type 11 was used; without blocking prolcins. this gave signals of about 2.2 x 10· cpm with YadA-positive yersiniae and abou! 3.5 x IW cpm with YadA-negative ycrsiniae afler Ihe washing procedure. For blocking expcrimenls, the desired amount of protein in 10 ~I of PBS was incubated with 100..,.1 of Ihe particu late baclerial suspension for 75 min at room temperature unless otherwise slated. Sam pies of 100 ng of Il!Il_collagen type 11 in 10 ~l of pas were added and allowcd 10 reac! for 75 min at room temperature unless othcrwise stated. Washing and determi· nation of radioactivity werc performed as dcscribed abovc.

1-\11 ~mI'IC:' WC,C <lLL<I',YLCU 'LI d' IC":" .WV Ll IU ....... "' ...... ,,,

experiments in duplicate assays. The plastic tubes uscd for the assays were precoated wilh bovine serum albumin (BSA) to minimize nonspecific binding of collagen and bacteria to the walls of the tubes.

Affinity blots. SOS-PAGE was performed on collagen and collagen fragments with 18 and 10% acryJamide minigcls. respectively. The proteins were electrophoretically trans· ferred to nitrocellulose filters. After nonspecific protein binding sites were blocked with 1% BSA in PBS for 2 h at room tempcrature. the filter was incubated with a particulate Yersinia suspension (3 x 10" bacteria per ml in PBS) for 2 h at Toom tempcraturc. A rabbit anti-Y. enterocolitica 0:5 antibody was used to detect filter-bound yersiniae, and pcroxidase-conjugated goat anti-rabbit immunoglobulin G was used as the detection system.

ELiSA. FOT the enzyme·linked immunosorbcnt assay (ELlSA), samples (10 ~I) of a collagen solution (various concentrations in PBS) were incubated with 100 ..,.1 of the bacterial suspension (2 x 10" bacteria) in V microtiter plates for 90 min 3t room temperature. The plates were cenlrifuged (20 min, 15°C, 550 x g), and the pellets were washed with 100 ..,.1 of PBS-O.l % Tween by resuspension. Centrifugation and washing were repeated. The bacteria were incubated with an anti-oollagcn type I1-spccific antibody (in 100 ..,.1 o f PB$-I% BSA, 4°C, overnight) and then washed as described above. The pellets weTe incubated with peroxidase-conju­gated rabbit anti·mouse immunoglobutin G (Bio-Rad Labe· ratories) diluted 1;5,000 in PBS-l% BSA for 2 h at room tempcrature. Bacteria were washed again, and the substrate (3,3 '-5,5'-tetramethylbenzidine (50 ml I, 0.01% in 100 mM citrate·phosphate (pH 6) plus 5 ~J of 30% H20 2) was added. The A . .... o was determined for the supernatant after centrifu· gation. The monoclonal anti-chicken collagen type 11 anti· bodies used in this study (BI, Cl, 03) were kindly provided by R. Holmdahl, Uppsala. Sweden. They were described previously (5, 19). The epitopes of these antibodies on chicken collagen type 11 are located on specific CB frag­ments: CB8 (BI), CBlO (0 3), and CBn (Cl). Thc curves for binding of all three antibodies to their corresponding frag­ments were similar, and the antibody concentrations were adjusted 10 similar signal intensities.

RESULTS

Binding orVadA-positive yersiniae to immobllized collagen • To test for direct binding of yersiniae to a spectrum of collagens, we performcd affinity blot experiments with im­mobiJized collagen. After SOS·PAGE and electroblotting of the proteins to a nitrocellulose filter, yersiniae were added; the bacteria bound to the fi lter were detected by using an anti-Y. enterocolitica 0:5 amibody.

YadA-positive yersiniae bound to collagen types 11 , V, and XI. whereas no binding was detectable to type VI collagen (Fig. 1). Within the heterodimeric collagen type V, YadA-positive yersiniae bound to both c hains. whereas, within the heterotrimeric collagen type XI, YadA·mediated binding was confined to the a3(XI) chain. Collagen types 111 and IV also bound YadA-posilive yersiniae (data not shown). YadA·negative yersiniac did not bind to any of the collagens (Fig. 1).

The most abundant collagen is type I collagen, a hetero· trimeric collagen consistingoftwo al(l) chains and one 02(1) chain. We were intercstcd in dctcrmining whcther binding of YadA to collagen type I is restricted to both chains, which show 65% homology at the protein level. The chains were

• 0 , . • • - : --\'1

., FIG. 1. Binding of YadA-positive and YadA-negalive yeTsi"iae

10 immobilized collagen. Collagen types 11 (a, b), XI (c. d). V (c, f). and VI (g, h) wert electroblotted 10 nitrocellulose filters after SOS·PAGE. YadA-posilive (a, c, c, g) or YadA-negalivc (SI rain NFpCB9::TnS) (b, d. f, 11) yers iniae wert added. and filler-bound bacleria were assessed by IIsing a rabbi! an ti-Y. enterQCQlitica 0:5 anlibody. An anti-rabbi! peroxidase conjugale was used as the dctection system. The positions of the prOleins nOI Mund by yersiniae [o:l(XI) aod o.:2(X I) in lane c, collagen type VI in lanes g and h] aft indicaled according 10 the it localizalion by Ponceau red sta ining. Coomassie-5ta ined SDS· PAGE of collagen type VI is also shown (i).

isolated by ion-exchange chromatography and further puri­fi ed by reverse-phase high-pressure liquid chromatography. The affinity blot experiment with the isolated chains of collagen Iype I showed Ihal binding 10 YadA was limiled 10 Ihe a1(1) chain (Fig. 2).

Bioding of IUI_labeled rollagen type 11 10 YadA_ The main carlilage collagen, collagen type 11, was radiolabeled 10 fu rt her characlerize binding of YadA . Increasing amounls o f 12!l1_labeled collagen type 11 (2.7 )( Hf' cpmJ~g) were added 10 2 x 109 YadA-positive and YadA-negative yersiniae. Binding o f collagen type 11 10 YadA-negative yersiniae was used to detect nonspecific binding. The result indicales saturat io n of specific binding of collagen Iype 1110 YadA al higher concentratio ns (Fig. 3). Based on these findings, 100 ng of ' 15I-labeled collagen type 11 was chosen as reference for furlher binding and blocking experiments. In Ihe assay . afler incubation and washing, 100 ng of radiolabeled collagen type 11 (2 )( l OS 10 3 )( 10' cpm) gave signals o f about 22,000 cpm wilh YadA' posilive yersiniae and about 3,500 cpm with YadA·negalive yersiniae.

Blocking of collagen type 1I binding 10 YadA by various collagens. Various collagens and collagen preparations were

• b c d

-

FIG. 2. Binding of YadA·positive yersiniae to immobilized col­lagen type I chains. Isolated chains a l(l) and a2(J) were subjected to SOS·PAGE (a and b. respcctive ly) and then electroblOlled to nitrocellulose filters (d and c. respcclively). The arrowhead marks tne posi tion of the a2(J) chain, according 10 its localiza tion by Po nceau red slai ning. The experiments were pcrformed as describcd in the legend to Fig. 1.

Y-a'dA~~~'iii~~' y~·~~i~·i;e .- H~~I~d~-;;a·t~;ed· ~~~t'~~t i~; IY~ ii collagens were equally effeclive in preventing binding of subsequenlly added native radioaclive collagen Iype 11 (Fig. 4). For Ihis experimenl collagen type 11 was heat denatured (42°C. 60 min I and the blocking assay was performed at 37"C to prevent spontaneous renat uralion. Collagen types 11 , 111, IV, V. and XI showed very similar inhibit ion curves (Fig. 4 and 5) . The dala suggesl high·affinity binding of all five collagens to YadA. About 60% inhibition was observed al ready wilh a fivcfold mo lar excess (500 ng) of the inhibitor. The contro l proteins BSA and cart ilagc proteoglycans did nOI show any inhibilion of collagen Iype 11 binding 10 YadA (Fig.5A).

Howeve r. the abilily to block collagen Iype II -YadA binding is not simply a feature of all triple-helical collagens. Collagen type VI at up to a 20·fold molar excess over radiolabcled collagen type 11 did nOI block collagen type 11-YadA binding. Only at a higher concentratio n (50·fold molar exeess) was in hibition observed. The collagen type IX inhibition curve showed greater inhibition compared with that of collagen type VI and did nol show the typical high effi cienl blocking of collagen types 11 , 111 , IV, V, and XI in the low concentration range.

Blocking of collagen type II-VadA binding by collagen type J and lsolated collagen type I cbains_ Type I collagen did not block collagen type II·YadA binding as efficicntly as the o ther fib ri l·forming collagens. e.g., types 11 , 111 , V, and XI. did (Fig. 5 and 6). Since Ihe affi ni ly blot expe riments with the isolaled collagen type I chains revealed thai the binding site for YadA was confined to the a l (l) c hain only (Fig. 2) and binding of collagen type 11 to YadA was independent of the Iriple-helieal confo rmalion (Fig. 4), we performed blocking experimenls with the isolaled chains o f collagen type I. The blocking eapacilY o f collagen type I at low mo lar concentra­tions resided complele ly in the a1(1) chain (58% inhibition at 500 ng) (Fig. 6). The blocking potenlial of a 1(1) was compa· rable to thaI of collagen Iype TI in nalive and denalured forms (Fig. 4). The a2(1) c hain failed to exhibit signifieanl blocking capacity (9% inhibition at 500 ng).

Blocking of collagen type II-YadA binding by cyanogen bromide fragments of collagen type 11. To further eharaeler­ize the binding si te for VadA on collagen type 11 . eN Br c1eavage al met hionine sites of collagen type 11 was per­formed, and Ihen Ihe fragments were purified by fa st protein liquid c hromatography. The modified solid-phase ELISA wilh Ihe purified fragmenls and fragmenl-speeific monoclo­nal anlibodies demonstrated YadA·mediated binding for CBlO only (data not shown). Therefore, CB10 (346 amino acids) and the approximalcly equal sized CB11 (279 amino acids) were used in the blocking experiments shown in Fig. 7. The amounts o f the Ihree inhibitor proteins shown in Fig. 7 are indicated in molarities for beUer comparison lal(II ), - 95 kOa: a l(II )CBlO, -31 kDa; o:l(II)CBll, - 25 kOal. Only CBlO was efficient in blocking collagen type II ·YadA binding. Up to a concentra tion of 3 pM , Ihe blocking capacities of CBIO and Ihe whole collagen type 11 molecu le were similar. AI this eoneenlralion the maximum inhibition (55%) was reached for the inhibilor C BlO. Furt her rise o f in hibito r concentrat io n up to a 150-fold molar excess (carre­sponding to 5 ~g) did nol inerease the blocking potential for the collagen type Il·YadA in teraction. CBll did not block collagen type 11 binding to YadA. Interaclions al high inhibitor concentrations such as those observed wilh the eomplete polypeptide ehains could nol be detected with the CS fragments used for the binding slUdics.

E 80 fr ~ o

60

40

20

I -t J j

-~

D

~

,~

, , • , • • '~ •• ..!MOI (lo1-t

O~----.----,-----.-----r----~----

o 5 10 15 20 25

'2'I-collagen 1ype 11 addod [106 cpm) FIG. 3. Specilil,': binding of radiolabcled collagen rype 11 10 YadA-positive yersiniae. Increasing amQunts of Il!I l_labeled collagen type 11 (2.7

)( H1' cpmJ~) were incubated with 2 x IOv bacteria . Unspecific binding 10 YadA-negative yersiniae was determined for each concentration aod subtracted. The insert demonstrates total counts obtaincd with YadA-positive (. ) aod YadA-negalive (0) yersiniae as weil as thc resulting specific binding (A). Each value represenls thc average of duplicalc assays. The variation of each individual va lue from thc mean never excceded 10%.

l ,,.

f " = !

1 " • i ,L.,.....L

10 100 ao 500 1000 2000 !OOO 2!5000

WIibIIof [1'1\11

FIG. 4. Inhibition of thc binding of radiolabcled collagen type 11 10 YadA by native (light stipp]ing) aod heal-denatured (4TC, 60 min~ (dark stippling) collagen Iype 11. YadA·posilive yersiniae (2 x 10 bactena) were im.:ubated with different concentrations of the colla· gen preparations. and then radiolabeled native collagen Iype 11 (2 x 10' to 3 x 10' cpm) was added. Thc aSS3ys were performed at 37"C to prevent spontaneous renaluration. Radioaclivity associated with the pellet was measured in a gamma counter. Radioactivity obtained with radiolabeled collagen type 11 and YadA·positive yersiniae without preincubation with other prote;ns was delined as 100% 12.'1J_labeled collagen Iype 11 binding. Each value reprcsents the mean and standard deviation of at least two independent duplicate assays.

D1SCVSSION

Attaehment of baeteria or baeterial products (0 host cells and host tissue is erucial for infectious diseases caused by enteropathogens like Y. enterocolitica. Previous studies have reported on binding of different bacteria to extracellular matrix proteins like laminin (9, 27, 32, 34. 35), fibronectin (24; for a review, see reference 20), and collagen (8, 9,18,33) and have moreover shown the importance of these interac­tions for tissue adhesion (21, 22). The interaction of proteins expressed by Y. enterocolitiea with components of the extracellular matrix, like collagen, may contribute 10 dis­eases affecting the connective tissue. Therefore, we investi­gated the binding to collagen of the virulence-associated outer membrane protein YadA, which is common to all enteropathogenic yersiniae. We identified a broad speetrum of collagen types that bind YadA-positive yersiniae in affin· ity blot experiments (collagen types 1,11, III, IV, V, and XI). Interestingly, YadA-positive yersiniae did not bind to all collagens tested (e.g., did not bind type VI collagen), and in some collagens the binding property for YadA was restrieted to a single chain o nly [Cll(I), a3(XI)J. Collagen binding has been reported for rat hapatocytes (31) and Stophylococeus oureus (33), both of which bound to all collagens tested (collagen types I, 11, 111, IV, and V for rat hepatocytes and collagen types I, 11, III, IV, V, and VI for S. aureus). Different CNBr-generated peptides of the ClI{I) chain and synthetic peptides with collagenlike slruclures [e.g., (Pro­Gly-Pro)"J were also bound indiscriminately. In contrasl to these findings, our results argue in favor of a YadA collagen

A ~ ,~

~I i =

~ 0 80 • f 60 ~

~ Ö Q 40 0

'!i " 20

0

'0' ,()2 ,0' ,0'

inhibitor (ng]

B ~ '00 i!. ~ 80 0 •

I 60

Ö Q 40 0

~

" 20

0 10' , ()2 ' 0' ,0'

inhibitor Ing] FIG. S. Inhibition of the binding of radiolabclcd collagen type 11 10 YadA by different collagens and other peptides. Collagen types IV (. l.

V (. ), and XI (. ) (A) and Iypes 111 (e ). VI (. ). lnd IX (. ) (D)wcre invclitigatcd. The experiments wert pcrformed asdescribed in thc legend 10 Fig. 4, except thai thc temperature was room temperature. earlilage protcoglycans (0) and bovine serum albumin CO) wert used as controls (A). Each value represcnts the mean and standard deviation of at least !Wo independent duplicatc assays.

binding site that is not determined by the simple repetitive sequence of collagen (Gly-X.Y) hut ralher by a specific amino acid molif. Studies with YadA-positive and YadA­negative yersiniae demonstrated that binding of collagen type 11 to YadA is saturable, which is further evidence for a specific interaction.

Preincubation of YadA-positive yersiniae with unlabeled collagen type 11 prevenled binding of subsequenlly added 1ul-collagen type 11 in a concentration.(!ependent manneT. Moreover. native and heal·denatured type 11 collagens in· hibited the collagen type 11-YadA interaClion 10 similar degrees. Furthermore, our experiments revealed thai, in contrast to the a2(I) chain, the al(l) chain and a 346-amino-

acid CB fragment (CBlO) of collagen type 11 harbar a specific binding site for YadA. Therefore, YadA seems 10 interact with a locally restricted peptide sequence, whereas binding 10 a conformationally dependenl epitope formed by seat­lered amino acids and dependcnt on the Iriple helical struc­lure of collagen seems ralher unlikely.

It remains to be shown whether the blocking capacities of collagen type VI and of the a2(I) chain at vast molar excesscs (Fig. SB and 6) were due 10 (i) minute contamina­lions (5%) of the preparations with other collagens not visible in Coomassie-stained SOS-PAGE, (ii) a specific bind· ing sile with lower affinity, or (iii) simply a slickiness because of weak noncovalent fo rces along the whole mole·

"- lUV -

= ~ i> 80 < • ~ : 60

15 ~ 40

i " 2<>

0 10' 10' 10'

inhibitor lngl FIG. 6. Inhibition of the binding of radiolabeled collagen type 11

10 YadA by collagen type I (. ) and isolated oollagen type I chains 011(1) (0) and a2(I) (0). The experiments were performed as described in the legend 10 Fig. 4, excepl that the tempenture was room temperaillre. Each value represents the mean and standard deviation of at least two independent duplicate assays.

eule. Considering one of the lasl two options, the falluTe of the CB fragments invesligatcd in Ihis sludy to further block collagen type II-YadA binding at high inhibitor concentra­tions might be duc 10 (i) an additional binding sile for YadA on collagen type 11, localized on a CD fragment not tested in the blocking assays, or (ii) deSlruction of low-affinity binding 10 YadA by the CNBr eleavage.

The capacities of the collagens tested in this study to bind YadA are different. However, some collagens had similar reaction patterns. Collagen types 11, 111, IV, V, and XI but not types VI and IX bound 10 YadA with high affinity and were able to inhibil collagen Iype I1-YadA binding in block­ing experiments at low inhibitor concenlrations. All fibril ­forming collagens (28) excepttype I collagen (i.e., collagen

!': '" • g " • • ~

!!" " , • ., ~ •

" ,

", ,,. ", ," lnhibllor (pmog

FIG. 7. Inhibition of the binding of radiolabeled collagen type 11 to YadA by collagen type 11 (.) and the collagen type 11 fragments CBIO (0) and CBlt (0). The amounts of the inhibitor proteins are indicated in molarities [011(11). -95 kDa; al{T1)CBIO, - 31 kDa; al{T1)CBll. - 25 kDa). The experiments were performed as de­scribed in the legend to Fig. 4, except that the temperature was room temperature. Each value represenlS Ihe mean and standard devia­tion of at least two independent duplicate assays.

''yl-'''''' " , ..... , "., ...... uu ..... u "'!!;" ' '' '''LI~ '' ...... " ...... '" .v. YadA. Within collagen type I, the a1(1) chain also bound to YadA to a degree comparab1e 10 other fibril-forming colla­gellS. Moreover, collagen type 11 is a homotrimer of three al(ll) chains, whereas the type XI molecule consists of three different c hains (al, a2, aJ), of which only the w(XI) chain was shown to bind YadA in the affinity blot experiments. There is some evidence that the a3(XI) chain and the al(H) chain are encoded by the same gene, with minor differences duc 10 posuranslalional modifications (11. 36). Therefore, it could weil be that collagen types Hand XI interact wilh YadA via the same binding sile. which is conscrved in both evolulionarily c10sely related molecules. It might be specu­lated that the inleraClion s ite remained conserved during the evolution of the collagen family from an ancestral 54-bp uni! (30) in collagens exhibiting YadA binding. Whether the binding site for YadA on collagen type IV is homologous 10 the YadA binding site on collagen type 11 is currenlly under investigation. Collagen types VI and IX, which bound con­siderably less weH to YadA, belong 10 distinct elasses of the collagen supergene family (28).

Further experiments are required 10 localize precisely the intcraction si tes on collagens and on YadA to answer Ihe question ofwhether collagen binding by YadA conlribules to the pathogenicity of yersiniae. Binding of YadA immuno­complexes 10 collagen type I. a major component of the eXlracellular matrix of the skin, could be implicated in the indUClion of erylhema nodosum, a skin affiiction that is frequently causcd by Y enterocolitica. In Ihis study we have demonstrated that the cartilage collagen types 11 and XI specifically bind YadA. Whether binding to cartilage-re­stricted collagens may be linked 10 the arthritogenic poten­tial of enteropathogenic yersiniae remains to bc investigated.

We have provided structural data on collagen-YadA inter­actions thai may slimulate further studies. including in vivo sludies with collagen peptides for modulation of Yersinia­induced experimental animal discascs.

ACKNOWLEDGMENTS

We thank Eva Bauer for expert teehnical assistance. This work was partly supported by the Deutsche Forschungsge­

meinschaft, SFB 263, project C3. The Max-Planck.Arbeitsgruppen in Erlangen are funded by the German Ministry for Research and Teehnology (grant 0IVM8702lO).

REFERENCES L Bentz, H., H. P. Bä(hinger, R. W_ Glanville, and K. Kühn. 1978.

Physical evidence for the assembly of a and b chains of human placental collagen in a single tripIe helix. Eur. J. Biechem. 92:563-567.

2. Bölin, I., 1- NOTlandeT, and H. Wolr_Wat~. 1982. TemperalUre­inducible outer-membrane protein of Yeßinia pseudotuben:ulo­sis and Yeßinia enterocolitiea is associated with the virulence plasmid. Infee!. Immun . 37:506--512.

3. Bölln, I., D. A. Por1noy, and H. Wolr_Wat~. 1985. Expression of the temperature-indudble outer membrane proteins of yersin· iae. Infeet. Immun. 48:234-240.

4. Bolfone, E. J_ 1977. YCßinia cnterocolitiea: a panoramic view of a charismatie mieroorganism. Cri!. Rev. Microbio!. .5:211-241.

5. Rurkhardl, H., R. Holmdahl, R. [)eulzmann, H_ Wltdemann, H. von der Mark. S. Goodmann, and K_ von der Mark.. 1991. Identification of a major anligenie epitope on CNBr-fragment 11 of type 11 collagen reeognized by murine autoreactive B cells. Eur. J. Immunol. 21:49-54.

6. Comelis, G. R., T. Riot, C. Lambertde Rouvroil, T_ Michiel5, 8. Mulder, C. Slullen. M_ P_ Sol')', M. van Bouchaute, and J. C. VanooteghMl. 1989. The YCßin;a yop regulon. Mol. Microbiol. 3:1455- 1459.

' ........ .. ", .. ..... ......... .... ,..."" •.• ~u,. ,~ , ... " ... "''''''v~v.".~" . .•. Engl. J. Med. 321:16--24.

8. Emödy, L, J . H_mlnD, H. Wolr'Wltz, M. Skurnik, G. Kapperud, P. O'Toole, Iod T. Wl d5trom. 1989. Binding to collagen by Yer:rillw eflterocolitica and Yer:ri1lia pseudotube,.. culosis: eviden!;C for YopA·mediated and chromosomally en· ooded mechanisms. J. Bacteriol. 171:6674-6679.

9. Fitzgerald, T. J., L. A. Repnh, D. R. 811neo, Ind J. N. Miller. 1984. Attachment of Treponema pal/idum 10 fibronectin, lami· nin, collagen IV and collagen I, and blockage of attachmenl by immune rabbil IgG. Br. J. Vener. Dis. 60:357-363.

10. Fnker, P. J., Ind J. C. Speck, J r. 1978. Protein and cell membrane iodinalion with a sparingJy soluble ehloroamide 1,3,4,6·letrachloro-3a,6a -diphenylglycoluril. Biochem. Bio­phys. Res. Commun. 80:849-857.

11. Furuto, D. K., and E. J. Miller. 1983. Different levels of gJycolisalion contribule 10 Ihe helerogeneity of al( lT) collagen chains derived horn a transplantable rat chondrosarcoma. Arch. Biochem. Biophys. 226:6()4....611.

12. Gllnville, R. W., Iod A. Rauter. 1981. Pepsin fragments of human pla!;Cntal basement-membrane collagens showing inter­rupted triple-helical amino acid sequences. Hoppe-Seyler's Z. Physiol. ehem. 362:943-951.

13. Guan, K., alld J. E. Dlxon. 1990. Protein tyrosine phosphatase activity of an essential virulence determinant in Yersinw. Sei­ence 249:553-556.

14. " _ mann, J., 8. Algermlssen, Ind R. Laurs. 1984. Genetically manipuJated virulence of Yer:ri1lw c1Iterocolitiea. Tnfeet. Tm­mun.46:105-110.

15 . H_ ml nn, J.,and K. Gaede.1989. Mechanisms involved in the pathogenesis of Yer:rillw infections. Rheumato1. Int. ':213-217.

16. H_mlnn, J., alld L. Griler. 1987. Genetie evidence that ouler membrane prOlein YOP1 of Yersillia enteroco/itica mediales adherence and phagocytosis resistance to human epithelial !;Clls. FEMS Microbio1. Let!. 40:37-41.

17. H_mlon, J., Ind R. Laur,. 1983. Construction of a mobiJiz­able Yer:rillia enterocolitica virulence plasmid. J. Bacteriel. 155:761-767.

18. Holderbaum, D., R. A. Specb, and L. A. Erhart. 1985. Specific binding of collagen 10 Staphylococeus aureus. Collagen Relal. Res. 5:261-271.

19. HoImdahl, R.. K. Rubin, L. Kllreskot!:, E. Larssen, Ind H. WipeIl.1986. Characlerization ofthe antibody response in mice with type 11 collagen-induced arthritis, using monoclonal anti· type 11 collagen antibodies. Arthritis Rheum. 2'=400-412.

20_ HMk, M., L. M. Switalskl, T. Wldstrom, and M. Lindbel""l. 1989. Inleractions of pathogenie microorganisms with fibronec­tin, p. 295-308. In D. F. Mosher (cd.), Fibronectin . Academic Press, Inc., San Diego, Cali!.

21. Isbel""l, R. R. 1991. Discrimination between inlracellular uptake and surface adhesion of bacterial pathogenes. Seience .152:934-938.

22_ Isbel""l, R. R •• and J. M. Leo0l. 1988. Cultured mammalian eells atlaeh to the invasin protein of Yer:rillia p$eudotuberculosis. Proc. Natl. Acad. Sei. USA 85:6682-6686.

23_ IGncb, T., and K. ,"OD der Mlrk. 1990. Isolation of bovine type X collagen and immunolocalization in growth.plale cartilage . Biochem. J. 265:453-459.

" .... ~._, •• • "~ •• ,~,~ .. ~~ .... ~ ... ~~ ,y Y'-I'''J'-~V~~_ -_. __ .

Nature (london) 276:718-720. 25. Latbica, R. V., D. L. Zink, Ind W. R. Fernes. 1984 . Association

of fibril structure forma tion with te il surfaee propertics of Yersinia e1lterocolitica . InfeCI. Immun. 46:272-275.

26. Laemmli, U. K. 1970. Cieavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (london) 227:680-685.

27. Lopes, J. D., M. Dos Reis, Ind R. R. 8rentani. 1985. Presence of laminin receplors in Staphylococeus aureus. Seience 229:275-277.

28. Miller, E. J., and S. Gay. 1987. The collagens: an overview and update. Methods Enzymo!. 144:3-41.

29. Portnoy, D. A., and R. J. Martinez. 1985. Role of a plasmid in the pathogenicity of Yersi1lia species, p. 29-51. 111 W. Gocbel (ed.), Genetic approaches 10 microbial pathogenicity. Springer­Verlag, New York.

30. Ramirez, F. 1989. Organization and evolulion of the fibrillar collagen genes, p. 21-30./11 B. R. Olsen, and M. E. Nimni (ed. ), Collagen, vol. 4. Molecular biology. CRC Press, Inc., Boca Raton, Fla.

31. Rubin, K •• M. HMk, 8. Öbrink, and R. Timpi. 1981. Substrate adhesion of rat hepatocyle: mechanism of allaehment to colla­gen substrales. Cell 24:463-470.

32. Speziale, P., M. HMk, T. Wadslriim, . nd R. TImpi. 1982. Binding of the basement membrane prolein laminin to Esche­riehia eoli. FEBS Lell. 146:55-58.

33. Speziale, P., G. Rautd, L. Vlsal, L. M. Switalskl, R. Timpi, and M. Höök.. 1986. Binding of collagen (0 Staphylococcus aureus Cowan 1. J. Bacteriol. 167:77-81.

34. Switalskl, L. M., H. Murchlson, R. Timpi, R. CurtisslU, Ind M. Höök.. 1987. Binding of laminin 10 oral and endocarditis strains of viridans streptococci. J. Baeleriol. 16':1095- 1101.

35. SwitaJskl, L. M., P. Spezille, M. HMk, T. Wadstrilm, and R. TImpI. 1984. Binding of Streptococeus pyogenes 10 laminin. J. Biol. Chem. 25'=3734-3738.

36. Swoboda, B., R. HoImdahl, H. Slöp, . nd K. ,"on der Mark. 1989. Cellular heterogeneity in cullured human ehondrocytes idenli­lied by anli" 'dies spc:cilie for a2{XI) collagen ehains. J . Cel1 Biol. 109:130 ...... 1369.

37. TImpl, R., R. W. Glanville, H. Nowatk, H. Wiedemann, P. P. F1etuk, and K. Kühn. 1975. Isolation, chemical and eleelron microscopical characlerization of neutral-salt-soluble type 111 collagen and procollagen fr('''' fetal bovine skin . Hoppe-Seyler', Z. Physiol. Chem. 356.. .792.

38. ToivaDen, A •• K. Granron, R. Labesmaa·Rantala. R. LeillO, T. Stahlbel""l, and R. Vuento. 1985. Pathogenesis of Yersi1lia­Iriggered reactlve arthri tis: immunologieal, microbiological and clinical aspects. Jmmunol. Rev. 86:47-70.

39. ,"on der Mark, H., M. Aumailley, G. Wk:k, R. FleiKbmajer, I nd R_ TImpi. 1984. Tmmunochemistry, genuine siu and tissue localisation of collagen VI. Eur. J. Biochem. 142:49J-..S02.

4{1. ,"on der Mark, K., M. van Menxd, and H. Wiedeml nn. 1982. Isolation and characterisation of new collagen from chiek ear­tilage. Eur. J . Biochem. 124:57-620.

41. WoIr,Wltt, H. 1987. Plasmid mediated virulence of Yer:riniae. Prog. Vet o Microbiol. ImmuneI. 3:159- 178.