The Use of Specific Antibodies to Demonstrate theGlycocalyx and Spatial Relationships of a K99-, F41-Enterotoxigenic Strain of Escherichia coli Colonizing

the Ileum of Colostrum-deprived Calves

R. Chan, C.J. Lian, J.W. Costerton and S.D. Acres*

ABSTRACT

Electron microscopy was usedto study the interaction betweenthe glycocalyx of enterotoxi-genic Escherichia coli strain210 (09:K30+;K99-;F41-:H-) andthe glycocalyx of epithelial cellsin the ileum of experimentallyinfected newborn colostrum-deprived calves. Fixation oftissues in anti-K30 antibody andruthenium red was used to sta-bilize the bacterial glycocalyx sothat the spatial relationshipbetween the bacteria and theintestinal epithelial cells couldbe characterized.When strain 210 was grown in

vitro and reacted with anti-K30antibody prior to staining withruthenium red, the extensiveglycocalyx could be clearly vis-ualized surrounding the bacte-rial cells. By negative staining,an unidentified pilus was alsoseen. Sections of ileum frominfected calves, which were notfixed in antibody nor stainedwith ruthenium red, revealedattached bacteria which weresurrounded by an electron-translucent zone and no visiblebacterial glycocalyx. Whenruthenium red staining wasused, the bacterial glycocalyxpartially collapsed during thedehydration steps of fixation,but could be seen as either afibrous capsule or an electron-

dense accretion on the bacterialcell surface. When ileal tissuewas reacted for one hour in anti-K30 antibody before stainingwith ruthenium red, the bacte-rial glycocalyx was seen as a dis-crete electron-dense structureup to 1.0 ,um thick which was inintimate contact with the glyco-calyx of the epithelial cells. Theimportance of the bacterial exo-

polysaccharide to microcolonyformation on the villi could beclearly visualized.

R:SUMP

Cette experience consistait autiliser la microscopie electro-nique pour etudier l'interactionentre le glycocalice de la soucheenterotoxinogene 210 d'Esche-richia coli (09: K30+; K99-;F41-: H-) et celui des cellulesepitheliales de l'ileon de veauxnouveau-nes et prives de colos-trum, soumis a une infectionexperimentale. On utilisa la fixa-tion des tissus dans des anticorpsanti K30 et la coloration aurouge de ruthenium pour stabili-ser le glycocalice des coliba-cilles, afin de pouvoir caracte-riser la relation spatiale entreces bacteries et les cellulesepitheliales de I'Mlon.Quand on cultiva la souche 210

in vitro et qu'on la soumit a l'ac-tion des anticorps anti K30,

avant de la colorer avec le rougede ruthenium, on pouvait facile-ment visualiser le glycocaliceextensif qui entourait les cellulesbacteriennes. La colorationnegative permit aussi de voir unfilament protoplasmique nonencore identifie. Les sections del'ileon des veaux experimentauxqu'on ne soumit pas a l'action desanticorps et qu'on ne colora pasavec le rouge de ruthenium,recelaient des bacteries atta-chees a la muqueuse, entoureesd'une zone translucide, a lamicroscopie electronique, maisdepourvues d'un glycocalice visi-ble. Quand on utilisa la colora-tion au rouge de ruthenium, leglycocalice bacterien s'affaissapartiellement, au cours desetapes de deshydratation de lafixation; on pouvait cependant levoir, soit comme une capsulefibreuse, soit comme une accr&tion dense, a la microscopie elec-tronique, sur la surface des cellu-les bacteriennes. Par ailleurs,lorsqu'on soumit des sections del'ileon des veaux experimentauxa l'action des anticorps anti K30,durant une heure, avant deproceder a la coloration au

rouge de ruthenium, le glycoca-lice bacterien se presenta sous laforme d'une structure discrkteet dense, au microscope electro-nique; cette structure mesuraitjusqu'a 1,um d'epaisseur et elleetait en contact etroit avec le

Can J Comp Med 1983; 47: 150-156.

*Department of Biology, The University of Calgary, Calgary, Alberta T2N 1N4 (Chan and Costerton) and Veterinary Infectious DiseaseOrganization, 124 Veterinary Road, University of Saskatchewan, Saskatoon, Saskatchewan S7N OWO (Lian and Acres).Reprint requests to Dr. S.D. Acres.Submitted October 6, 1982.

150

glycocalice des cellules 6pith6-liales. On put aussi visualiserclairement l'importance de 1'exo-polysaccharide bact6rien A1'6gard de la formation demicrocolonies sur les villositksintestinales.

INTRODUCTION

In a previous paper we describedthe attachment and spatial rela-tionship between enterotoxigenicEscherichia coli (ETEC) strainB44 (09:K30+;K99+;F41+:H-) andthe ileal mucosa of colostrum-fedcalves (4). Attachment, multiplica-tion and the subsequent formationof microcolonies covering theintestinal epithelial cells are criti-cal steps in the pathogenesis ofdiarrhea caused by ETEC. Pilihave been shown to be an impor-tant mechanism of bacterial at-tachment to the intestinal mucosain several animal species (12). Mostepidemiological studies of bovinestrains of E. coli have found a highcorrelation between the presenceof K99 pili and enterotoxigenicity(9, 13, 17), suggesting that K99antigen is the main attachmentmechanism in calves. However,there may be other mechanisms ofadherence in addition to K99 anti-gen. Girardeau and co-workersdemonstrated the attachment ofK99- strains of E. coli to calf intes-tinal villi in vitro, and suggestedthe occurrence of two other pilusattachment factors on calf ETECwhich they called F(Y) and F (31a)(8, Girardeau, J.P., H.C. Dubour-guier and M. Contrepois. 1979.Attachment des E. coli enteropa-thogenes a la muqueuse intesti-nale. In gastro-enterites neon-atales du veau, Societe FranCaisede Buiatrie. pp. 53-66). Morerecently, Morris et al described asecond pilus in addition to K99,which occurs on ETEC strainspossessing 0 antigens 9 or 101, andwhich they have labelled F41 (6,18, 19). Smith and Huggins werealso able to reproduce diarrhea incolostrum-deprived calves byorally inoculating ETEC strains

which lacked the K99 antigen (21).Using one of these strains obtainedfrom H.W. Smith (VIDO strain210, serotype 09:K30+;K99-;F41-:H-), Bellamy and Acres also rep-roduced diarrhea in colostrum-deprived calves and demonstratedthat the organisms attached tointestinal mucosa. However, incontrast to K99+ ETEC whichform layers covering intestinalvilli in the ileum and jejunum,colonization of the villus surface bystrain 210 was much more focaland was confined to the ileum (2).The present study was conductedto determine if the spatial relation-ship between the glycocalyx of att-ached bacterial cells and the gly-cocalyx of the ileal mucosa incolostrum-deprived calves chal-lenged with ETEC strain 210 is asintimate as that seen in calvesinfected with ETEC strain B44.

MATERIALS AND METHODSSTRAINS OF E. COLIEscherichia coli strains of the

serotypes 09:K30+;K99-;F41-;H-(VIDO strain 210) and 09:K30-;K99-(VIDO strain 211) wereobtained from Dr. H.W. Smith,Houghton Poultry Research Sta-tion, England. Both strains werederived from an 09:K30+:K99+ calfenteropathogenic strain, believedto be reference strain B44, bymethods previously described (21).Strains 210 and 211 were pre-viously designated as the 0+K+99-and 0+K-99- variants respectivelyby Smith and Huggins (21). Bothstrains produced heat-stable ente-rotoxin as detected by the infantmouse assay (17). Neither strainagglutinated with standard K99antiserum when grown on Mincaagar containing 1% Isovitalex1(Minca-IS) (10). Strain 210, but notstrain 211, agglutinated in K30antiserum. Strain 210 grown onMinca agar was negative whenexamined for the presence of F41antigen using an indirect immuno-fluorescent technique (personalcommunication, J.A. Morris), andhad previously been shown not topossess K88 or 987P pili (2).

PREPARATION OF ANTISERAThe antisera used were those

described previously (4). The K99antiserum, which was preparedusing purified K99 antigen (11),had an agglutinating titer of1/2048 when tested against K99reference strain B41 (0101:K99)and the K30 antiserum had anagglutinating titer of 1/64 whentested against strain 210 grown onblood agar.PREPARATION OFCULTURED CELLS FORELECTRON MICROSCOPY

Aliquots (4 mL) of cells fromexperimental cultures of strains210 and 211 grown on Minca-IS for18 h at 37°C and washed once inphosphate buffered saline (PBS),were mixed with 0.1 mL of K30antiserum (diluted 1:2 with PBS)and held for one hour at roomtemperature. Following this spe-cific antibody stabilization, thecells were washed in PBS andfixed and processed in the pre-sence of ruthenium red by themethods of Mackie et al (16). Cellsto be examined by negative stain-ing were mixed with an equalvolume of 2% (v/v) ZrO2 (pH 7.0,containing 0.2% sucrose) and adrop of this mixture was placed onpolyvinyl Formvar-coated coppergrid, blotted and air dried.INOCULATION OF CALVES

Two newborn Holstein bullcalves were removed from theirdams immediately after birthbefore nursing had occurred andwere placed in individual isolationrooms. At three hours of age theywere challenged orally by slowlyinoculating a trypticase soy brothculture ofE. coli strain 210 into theback of the mouth using a 20 mLsyringe as previously described(2).COLLECTION OF TISSUE SAMPLES

Calves were examined at regu-lar intervals following challengeinoculation for the presence ofdiarrhea. At 42 (Calf 80-40) and 25(calf 80-42) hours of age the calveswere euthanized by an intravenous

IBBL, Cockeysville, Maryland.

151

sule, tissue sections were immersedin a 1:5 dilution of K30 antiserumfor one hour at room temperatureprior to staining with rutheniumred (15) and fixation.

RESULTSWhen strains of ETEC were

A

a

AA

.. .. .. ."

NQ

.r, .. k

, 4.r;i >r$$ ;. .i ; * , t ' B ^v a

igi|- r- t si ¢ 4 2 v |¢;x1'

Fig. 1. Electron micrographs of anti-K30 antibody-stabilized, ruthenium red-stained ETEC strains grown on Minca-IS medium. Strain 211 (09:K30-:K99-) showsno glycocalyx and no structures outside the outer membrane of the cell wall (Fig. la),whereas strain 210 (09:K30+:K99-;F4 1-:H-) shows distinct consolidated glycocalycesvarying from a relatively thin radially-arranged fibrous matrix (Fig. lb), to veryextensive fibrous masses surrounding the cells in a radial pattern and occupyingmore than 1 Am of intercellular space (Fig. lc). The bars in these and all subsequentelectron micrographs indicate 0.1 Am.

grown on Minca-IS medium andexamined by TEM (as anti-K30antibody-stabilized, rutheniumred-stained preparations) the K30-strain (211) was seen to lack anydiscernable structure outside theouter membrane of the cell wall(Fig. la), while cells of K30+ strain210 were seen to be surroundedeither by thin radially-structuredglycocalyces (Fig. lb), or by veryextensive glycocalyces composedof fibrous exopolysaccharidematerial up to 1.0,m thick (Fig.lc). Negatively-stained prepara-tions of strain 210 showed the pre-sence of variable numbers of verythin flexible pilus-like structures.A comparatively heavily "piliated"cell is shown in Fig. 2.Electron microscopy of the

ileum (intestinal locations 4 and 5)of colostrum-deprived neonatalcalves orally infected with strain210 showed numerous bacteria inimmediate proximity to themicrovillus border (Figs. 3-7).When both ruthenium red stainingand antibody stabilization wereomitted the bacterial cells wereseen to be separated from thetissue surface, and from adjacentbacteria and debris by an electron-translucent zone of variabledimensions (Fig. 3). When thepreparations were stained withruthenium red, but not stabilizedby specific antibody, the exopoly-saccharide of the infecting bacte-rial cells, which had collapsed to avariable extent during dehydra-tion, was seen to form either afibrous capsule (Fig. 4a) or a veryhighly condensed electron-denseaccretion on the bacterial cell sur-face (Fig. 4b).When preparations of infected

tissue were treated with specificanti-K30 antibody before fixationand staining with ruthenium red,the bacterial glycocalyces wereseen as thick consolidated electron-dense structures surrounding thebacterial cells and in intimate con-tact with the brush border (Figs. 5,6 and 7). These very extensive gly-cocalyces were seen to completelysurround the bacteria and to playan important role in the formationof planar microcolonies at thetissue surface. In areas into which

152

inoculation of sodium pentobar-bitol (Euthanyl, MTC Pharma-ceuticals, Hamilton, Ontario) andtissue sections were taken fromfive equally spaced sites in thesmall intestine and processed asdescribed previously (4). For sta-bilization of the bacterial cell cap-

A.;.i-

.*-4": ....." lm.::::.:.: W..Q

the stabilizing antibodies hadpenetrated less well (Fig. 5, inset,and the inner glycocalyces in Fig.6) the bacterial glycocalyx is lesscompletely stabilized and consoli-dated. The specificity of this stabil-ization of the bacterial glycocalyxis supported by the total lack ofstabilization of the tissue cell gly-cocalyces at the microvillarborders seen in Fig. 6 and in Fig. 7(arrows). Well stabilized glycoca-lyces are seen at high magnifica-tion in Fig. 7, and both the fibrousnature of the exopolysaccharideand its role in microcolony forma-tion are clearly illustrated.

DISCUSSION

The ability of K99- strains ofencapsulated ETEC to colonize thesmall intestine and cause diarrheain colostrum-deprived calves waspreviously demonstrated by Smithand Huggins (21); however, theydid not attempt to visualize at-

tachment of the bacteria to villusepithelial cells. Bellamy and Acresconfirmed that strain 210, obtainedfrom H.W. Smith, caused diarrheain colostrum-deprived calves andalso demonstrated focal attach-ment of this strain to the ilealepithelium (2). The present reportconfirms that strain 210 attachesto intestinal epithelial cells, anddemonstrates that the degree ofcontact between the glycocalyx ofadherent bacteria and the glyco-calyx of epithelial cells in the ileumis just as intimate as that seen withstrain B44 (4).The molecular architecture of

the cell surface of this strain is ofinterest because it bears a pilus ofunknown function and is sur-rounded by K30 capsular antigen,

an extensive fibrous polyanionicglycocalyx composed of D-glu-curonic acid, D-mannose and D-galactose (3). This extensivelyhydrated structure (22) condensesduring the dehydration steps usedin routine preparation of tissue forelectron microscopy (16), andruthenium red staining reveals theremnants of this collapsed matrixas an even electron dense "crust" atthe cell surface (Fig. 4b). When thecollapse of the glycocalyx is lesscomplete (Fig. 4a), or when it isprevented by cross-linking withspecific antibodies (16), this com-plete fibrous surface structure isseen to form a very coherent andhighly organized matrix close (ca0.3 ,m) to the cell wall, and a muchmore diffuse matrix that may

* *.

Y.

*.."1,

V...

Fig. 2. Negative stain of cells of strain 210 showing very fineflexible pilus-like structures. This strain had previously beenshown, by fluorescent antibody staining of intestinal sections,to lack K99, K88, 987P antigens.

Fig. 3. Electron micrograph of a preparation of the ileum(location 5) of a colostrum-deprived neonatal calf infectedwith ETEC strain 210 (09:K30+:K99-;F41-:H-). Note theabsence of any electron-dense projections from the surfaces ofthese cells and the extensive electron-translucent zonesseparating these bacteria from the tissue surface.

153

fsS' ^.

7-0-f -* j."

414 -.N" :;

extend several ,m into the sur-rounding milieu (Fig. lc). The gly-cocalyx is not seen when theK30 antigen is absent (Fig. la), orwhen ruthenium red fixation is notused (Fig. 3). Similar results wereobserved when the attachment ofK99+ strains to the brush border ofcolostrum-fed calves was exam-ined using the same techniques (4).The specificity of the stabiliza-

tion of the glycocalyces by anti-K30 antibodies is attested to by thelack of reaction with K30- cells andby the fact that the chemically sim-ilar (20) glycocalyx of the epithe-lial cells of the ileum is not affectedin any way by these sera. Consoli-dation of the glycocalyx is more

complete where this structure ismost accessible to stabilizingantibodies (Fig. 6), but this greateraccess to antibodies changes onlythe degree of consolidation and notthe spatial extent of the glyco-calyx.When the real extent and distri-

bution of the glycocalyces of thesebacteria is seen in antibody-stabilized preparations, the role ofthese fibrous exopolysaccharidestructures in the formation ofmicrocolonies is obvious. Stabili-zation of the bacterial glycocalyxreveals that this exopolysaccha-ride material connects neighbour-ing bacterial cells so that planarmicrocolonies form on the surface

of the infected tissue. Similarmicrocolonies are often seen innatural ecosystems such as streams(7), soils (1), and the bovine rumen(5), and in pathogenic systems suchas the infected lung in cystic fibro-sis (14).The mechanism of attachment of

strain 210 to the ileal mucosa ofcalves is unknown. By negativestaining, cells of this strain grownon Minca-IS had pili that did notreact with antiserum to the knownattachment factors (K99, F41,987P, K88) found on ETEC iso-lated from calves, sheep or swine.It is possible that adhesion ismediated by this undefined pilus.In addition, this strain possesses a., . s

. ....... s a*.+.. .. ,. .N, .

.;_;i .

,.,= t.4t_E (_t s

tr .

; \ --tih;; .., \l5SY *6 s4.;.S'0 i_,.s +e > t -.. i. 23. ^. .. :

,>. .. .... ° e . .} Si - .sd I

.: ^ @

:: ^ i b4. _.{

J

Fig. 4. Electron micrograph of a ruthenium red-stainedpreparation of the ileum (location 5) of a colostrum-deprivedneonatal calf infected with ETEC strain 210. Collapse of theexopolysaccharide glycocalyces of the bacteria produceseither a matrix of electron-dense fibres (Fig. 4a) or a very thinelectron-dense accretion at the bacterial cell surface (Fig. 4b).

Fig. 5. Electron micrograph of a K30 antibody-stabilized,ruthenium red-stained preparation from the same sourceused in Fig. 4. Note the stabilization which produced a highlyconsolidated and very extensive electron-dense glycocalyxsurrounding the infecting bacteria and the obvious roleplayed by this structure in the formation of planar microcolo-nies at the surface of the infected tissue. In the intervillusspace of the ileal epithelium, where access of the specificantibody was limited, a less consolidated stabilization of thebacterial glycocalyx is seen (inset).

154

A. :,;..

.4 ....

- ,:

,.l-.AN-.-,.;

'o0 , ., .,44,IN? lp. -

t'

capsular antigen (K30) and theexact role of this glycocalyx in at-tachment is not known. From thisstudy, as well as the previous oneon strain B44, it is evident thatlarge amounts of this dense fibrousmaterial is formed in vivo, encap-sulates the attached bacterial cells,and is in intimate contact with thebrush border. A clearer definitionof the specific role of the pilus andthe glycocalyx in colonization ofthe ileum requires further study.

ACKNOWLEDGMENTS

This study was supported by theAgricultural Research Council ofAlberta, Farming for the Future.The technical assistance of SandraFeschuk, Joyce Nelligan andSheila Costerton is gratefullyacknowledged.

REFERENCES

1. BAE, H.C., E.H. COTA-ROBLESand L.E. CASIDA, Jr. Microflora ofsoil as viewed by transmission electronmicroscopy. Appi. Microbiol. 23:623-648. 1972.

2. BELLAMY, J.E.C. and S.D. ACRES.A comparison of histopathologicalchanges in calves associated with K99-and K99+ strains of enterotoxigenicEscherichia coli. Can. 3. comp. Med.47:143-149.1983.

3. CHAKRABORTY, A.K., H. FRIE-BOLIN and S. STIRM. Primarystructure of the Escherichia coli sero-type K30 capsular polysaccharide. 3.Bact. 141:971-972. 1980.

4. CHAN, R., S.D. ACRES and J.W.COSTERTON. The use of specificantibody to demonstrate glycocalyx,K99 pili, and the spatial relationships ofK99+ enterotoxigenic E. coli in theileum of colostrum-fed calves. Infection& Immunity 37:1170-1180. 1982.

5. CHENG, J.J., R.P. McCOWAN andJSW. COSTERTON. Adherent epithe-lial bacteria in ruminants and their

roles in digestive tract function. Am. 3.dlin. Nutr. 32:139-148. 1979.

6. DE GRAAF, F.K. and I. ROORDA.Production, purification and charac-terization of the fimbral adhesive anti-gen F41 isolated from calf enteropa-thogenic Escherichia coli strain B41 M.Infection & Immunity 36:751-758.1982.

7. GEESEY, G.G., W.T. RICHARD-SON, H.G. YEOMANS, R.T. IRVINand J.W. COSTERTON. Microscopicexamination of natural sessile bacterialpopulations from an alpine stream.Can. J. Microbiol. 23:1733-1736. 1977.

8. GIRARDEAU, J.P. A new in vitrotechnique for attachment to intestinalvilli using enteropathogenic Escheri-chia coli. Ann. Microbiol. (Inst. Pas-teur) 131B:31-37. 1980.

9. GUINEE, P.AM. and W.H. JAN-SEN. Detection of enterotoxigenicityand attachment factors in Escherichiacoli strains of human, porcine andbovine origin; a comparative study.Zentbl. Bakt. Hyg., I. Abt. Orig. A243:245-257. 1979.

10. GUINEE, P.A.M., J. VELDKEMPand W.H. JANSEN. Improved Minca

- 4

I

rr. N~~~~P

48

a..)~~~~~~~~~~~~~~~~~~~~~~.

VA -~~~~~~~~~~~~~~~~~~~.

Fig. 6. As Figure 5. Note the heavily consolidated stabiliza-

tion of the more accessible outer bacterial glycocalyces and

the less consolidated stabilization of the less accessible inner

glycocalyces.

Fig. 7. As Figure 5. Note the consolidated stabilization of the

bacterial glycocalyx, whose component fibers are clearly

seen at this magnification and the lack of stabilization of the

tissue cell glycocalyx (arrows) seen at the microvillar border.

155

I--,"-.'1I--.

-4

!K..-

"us"

i4-

medium for the detection of K99 anti-gen in calf enterotoxigenic strains ofEscherichia coli. Infection & Immunity15:676-678. 1977.

11. ISAACSON, R.E. K99 surface antigenof Escherichia coli: purification andpartial characterization. Infection &Immunity 15:272-279. 1977.

12. ISAACSON, R.E. Pili of enterotoxi-genic Escherichia coli. In S.D. Acres,A.J. Forman, H. Fast, Eds. Proceed-ings Third International Symposiumon Neonatal Diarrhea, VeterinaryInfectious Disease Organization, Sas-katoon, Saskatchewan, Canada. pp.213-236. 1981.

13. ISAACSON, R.E., H.W. MOON andR.A. SCHNEIDER. Distribution andvirulence of Escherichia coli in thesmall intestines of calves with andwithout diarrhea. Am. J. vet. Res.39:1750-1755. 1978.

14. LAM, J., R. CHAN, K. LAM and J.W.COSTERTON. Production of mucoidmicrocolonies by Pseudomonas aerugi-nosa within infected lungs in cystic

fibrosis. Infection & Immunity 28:546-556. 1980.

15. LUFT, J.H. Ruthenium red and ruth-enium violet. I. Chemistry, purifica-tion, methods of use for electronmicroscopy, and mechanism of action.Anat. Rec. 171:347-368. 1971.

16. MACKIE, E.B., K.N. BROWN, J.LAM and J.W. COSTERTON. Mor-phological stabilization of capsules ofgroup B streptococci, types Ia, Ib, IIand III with specific antibody. J. Bact.138:609-617. 1979.

17. MOON, H.W., S.C. WHIPP and S.M.SKARTVEDT. Etiologic diagnosis ofdiarrheal diseases of calves: frequencyand methods for detecting enterotoxinand K99 antigen production by Esche-richia coli. Am. J. vet. Res. 37:1025-1029. 1976.

18. MORRIS, J.A., C. THORNS, A.C.SCOTT, W.J. SOJKA and G.A.WELLS. Adhesion in vitro and in vivoassociated with an adhesive antigen(F41) produced by a K99- mutant of thereference strain Escherichia coli B41.

Infection & Immunity 36:1146-1153.1982.

19. MORRIS, J.A., C.J. THORNS andW.J. SOJKA. Evidence for two adhe-sive antigens on the K99 referencestrain Escherichia coli B41. J. Gen.Microbiol. 118:107-113. 1980.

20. ROSEMAN, S. Complex carbohy-drates and intercellular adhesion. InBiology and Chemistry of EukaryoticCells Surfaces. Vol. 7. p. 317. E.Y.C.Lee and E.E. Smith, Eds. London:Academic Press. 1974.

21. SMITH, H.W. and M.B. HUGGINS.The influence of plasmid-determinedand other characteristics of enteropa-thogenic Escherichia coli on their abil-ity to proliferate in the alimentary tractof piglets, calves and lambs. J. Med.Microbiol. 11:471-492. 1978.

22. SUTHERLAND, I.W. Bacterial exo-polysaccharides - their nature andproduction. In Surface Carbohydratesof the Prokaryotic Cell. Sutherland,I.W., Ed., London: Academic Press.1977.

156