Binding of uropathogenic Escherichia coli R45
to glycolipids extracted from vaginal epithelial
cells is dependent on histo-blood group
secretor status.
A Stapleton, … , S Hakomori, W E Stamm
J Clin Invest.
1992;
90(3)
:965-972.
https://doi.org/10.1172/JCI115973
.
Women with a history of recurrent Escherichia coli urinary tract infections (UTIs) are two to
three times more likely to be nonsecretors of histo-blood group antigens than are women
without such a history. Further, uroepithelial cells from women who are nonsecretors show
enhanced adherence of uropathogenic E. coli compared with cells from secretors. To
investigate the hypothesis that nonsecretors express unique receptors for uropathogenic E.
coli related to their genetic background, we extracted glycosphingolipids (GSLs) from
vaginal epithelial cells collected from nonsecretors and secretors and used an assay in
which radiolabeled uropathogenic E. coli were bound to these GSLs separated on TLC
plates. An E. coli strain (R45) expressing both P and F adhesins, which was isolated from
one of these patients' UTIs, was metabolically labeled with 35S for the TLC binding assay.
The radiolabeled E. coli R45 bound to two extended globo-series GSLs, sialosyl
gal-globoside (SGG) and disialosyl gal-gal-globoside (DSGG), found in the GSL extracts from
nonsecretors but not from secretors. The identity of SGG in the nonsecretor GSL extracts
was confirmed in radioimmunoassays using an mAb to SGG and in immunofluorescence
assays with this mAb and native vaginal epithelial cells. We show that SGG and DSGG are
selectively expressed by epithelial cells of nonsecretors, presumably as a result of
sialylation of the gal-globoside precursor glycolipid, which in secretors […]
Research ArticleFind the latest version:
http://jci.me/115973/pdf
Binding
of
Uropathogenic Escherichia coil R45
to
Glycolipids
Extracted from
Vaginal
Epithelial Cells Is Dependent
onHisto-Blood
Group
Secretor Status
AnnStapleton,* Edward Nudelman,t Hennk Clausen, Sen-itiroh Hakomori,*andWalter E. Stamm*
*Department ofMedicine, University of Washington, Seattle, Washington 98195;*TheBiomembraneInstitute, Seattle, Washington 98119;and§Department ofOralDiagnosis, TheRoyalDentalCollege, Copenhagen,Denmark
Abstract
Women with a history of recurrent Escherichia coli
urinary
tractinfections
(UTIs)
are two tothree timesmorelikely
tobe nonsecretors of histo-blood group antigens than are womenwithout such ahistory. Further,
uroepithelial
cellsfromwomenwho are nonsecretorsshow enhanced adherence of
uropatho-genicE.coli
compared
with cells fromsecretors.Toinvestigate
thehypothesis thatnonsecretorsexpressuniquereceptorsfor uropathogenicE. coli relatedto their
genetic background,
weextracted
glycosphingolipids (GSLs)
fromvaginal
epithelial
cells collected from nonsecretors and secretors and used an
assay inwhichradiolabeled
uropathogenic
E.coliwerebound to these GSLs separated on TLCplates.
An E. coli strain (R45)expressing bothPand Fadhesins, whichwasisolated from one ofthese patients' UTIs, wasmetabolically
labeled with35S
fortheTLC bindingassay. Theradiolabeled E.coli R45 boundto twoextendedglobo-seriesGSLs, sialosyl
gal-glo-boside
(SGG)
anddisialosyl
gal-globoside
(DSGG),
foundin the GSLextracts from nonsecretors but not from secretors.Theidentity of SGGinthenonsecretorGSLextracts was
con-firmed in radioimmunoassays
using
an mAb to SGG and in immunofluorescenceassays with thismAb andnativevaginal
epithelial
cells.We showthat SGG and DSGGareselectively
expressed by
epithelial
cellsofnonsecretors,presumably
as aresult ofsialylation ofthe
gal-globoside
precursorglycolipid,
whichin secretorsis
fucosylated
andprocessed
toABH anti-gens. The presence ofSGGand DSGG may accountfor the increased binding ofE.colitouroepithelial
cellsfromnonse-cretors and fortheirincreased
susceptibility
torecurrent UTI. (J. Clin.Invest.1992.90:965-972.) Key
words:recurrent
uri-nary tractinfection* nonsecretor *glycosphingolipid
receptorsIntroduction
Acute
uncomplicated urinary
tractinfections(UTIs)
' occurin millions of youngwomen each year.Although
most of thesePortions of this workwerepresentedatthe annualmeetingofthe
Asso-ciation of AmericanPhysicians/American Societyfor Clinical
Investi-gation/American Federation ofClinical Research, May 1991, and have beenpublishedas anabstract( 1991. Clin. Res. 39:214a).
Addresscorrespondenceto Dr. AnnStapleton, Department ofMed-icine, Harborview Medical Center, 325 NinthAvenue,ZA-89, Seattle,
WA 98104.
Receivedforpublication 22 October 1991 and in revisedform 16 March 1992.
1. Abbreviations used in this paper: DSGG, disialosyl gal-globoside; GSL,glycosphingolipid; HPTLC, high-performancethin-layer
chroma-womenexperience only single or sporadic infections, - 20%
sufferveryfrequent(three or more per year)recurrences(1). The apparentincreased susceptibilityto UTI in these patients cannotbeexplainedbyunderlyingfunctionalor anatomic ab-normalities oftheurinary tract, but instead appears to arise fromtheinteraction of infecting Escherichiacolistrains with thesepatients' epithelial cells. Thus,women prone tofrequent recurrencesdemonstrate prolongedcolonizationof the vaginal mucosawithE.
coli,
the predominant causative species inthese infections (2) and threefold more E.coli
adhereto vaginal, buccal, anduroepithelial cells fromwomenwith recurrent UTI than tocellsfrom control patients (3,4). Womenwitha his-toryofrecurrent UTI arealsomorelikelytobenonsecretorsof histo-bloodgroupantigensthan are womenwithoutahistory ofinfections (relative risk=3-4[5-8]),and E.coli
adherein greater numbers touroepithelialcellsfromnonsecretors (8).E.
coli
adhesins also playanimportant role in the pathogen-esis of UTI (9-20),particularly
thegenetically
relatedPandF adhesins. Padhesinsare presentin 50-65% ofE. coli strainscausing cystitis
(9-13) and 75-90% of isolates frompyelone-phritis
(10, 1 1, 14-19). The minimalreceptorfor thePadhesin is the galactosea1-4galactose moiety, present in the globo-se-riesglycolipids
and theP,
blood groupantigen(20-28). F ad-hesinsare expressedin 30-65% ofUTI isolates(9, 10). The minimalbinding moiety forthe Fadhesinis less well defined andprobably morecomplex; proposed receptors include the Forssman and para-Forssman antigens, globoside, galactosyl globoside,globo-A, and globo-H, and stage-specific embryonicantigen-4
(29-32). The globo-series glycolipidshave recently been showntobemodifiedbythe histo-bloodgroup status, and ABH active globo-seriesglycosphingolipids(GSLs) are found in renal epithelium and in voideduroepithelial cells as well as onerythrocytes
(21, 28, 33-38).Given the
foregoing,
wehypothesized
thatoneexplanation
for
theincreased risk of UTI innonsecretors and for the in-creased attachment ofuropathogenic
bacteriatotheiruroepi-thelial cells
might
bethepresence ofunique
receptors for P-andF-fimbriatedE.coli
in suchpatients.
Toevaluatethishy-pothesis,
westudiedbinding
of radiolabeleduropathogenic
E.coli
R45toGSLs extractedfrom
thevaginal epithelial
cells of young womenof knownsecretorstatusandseparatedonTLC (39, 40). We demonstratedbinding moieties
uniqueto nonse-cretorswithstructuresconsistent with Se gene-mediated con-trol ofantigen expression inglobo-series
GSLsfrom vaginalepithelial
cells.Methods
Preparationofglycolipids.Vaginal epithelial cellswerecollected from healthy female college students whose secretor statuses were
deter-tography; IHW,isopropanol/hexane/water; SGG, sialosyl
gal-globo-side;UTI,urinarytractinfection. J. Clin.Invest.
© TheAmericanSociety for ClinicalInvestigation,Inc.
0021-9738/92/09/0965/08 $2.00
minedusingsalivahemagglutination inhibition assays and whose red bloodcell Lewisaantigen phenotypesweredeterminedusing hemagglu-tinationassayswithianantibodytotheLewisaantigen (7).Cellswere
collected bysalinerinsingandgentle scrapingwith aspatula, washed four times inPBS,pH7.3,and stored in afreezingmedium(85%M199
[Sigma Chemical Co., St. Louis, MO], 10% fetal calf serum, 5% DMSO) at -70°Cuntiluse(7, 41).Beforeglycolipid extraction,the cells were washedfour timesinPBS,quantitatedinahemocytometer,
andequalized for extraction procedures.The cells constituted a homo-geneouspopulation possessing typical morphologyasviewed onlight microscopy.6-8xI07cellspooled frommultiple patientsorI0'cells/ individualpatientwere used in each TLCassay. The total upper- and lower-phase glycolipid fractionswereobtained asfollows: cells were extracted twice with 10 vol isopropanol/hexane/water (IHW) (55:25:20byvol) with sonicationin a warmbathandcentrifugationat 2,500 rpmfor10 min. Thecombined supernatant fractionsweredried
undernitrogenand twice resuspended in
chloroform/
methanol(2:1 byvol)with 0.166 vol water andcentrifugedat500 g for 10 min (42). Total upper and lowerphaseswereevaporatedunder anitrogenstream andresuspendedin IHWforchromatography.
Forseparation oftotal upper neutralglycolipidsandgangliosides,
upperphaseglycolipidswerefirstresuspendedin 0.1% KCI in water,
subjected to C18 Sep-Pak reverse-phase column chromatography,
washed with water, eluted with methanol, dried, andpassedover a DEAE Sephadex A-25 column. Gangliosideswerethen eluted with 0.45 Mammoniumacetate inmethanol,dried,andpassedover aC18 column.
Glycolipids were preparativelyseparated bychromatography on
glass high-performance thin-layer chromatography (HPTLC) plates
(WhatmanInc., Clifton, NJ) inchloroform/methanol/water50:40:10 with 0.05% CaCI2. The bands were visualized with primuline (43) underultraviolet(UV)light,markedwithapencil,scrapedfromthe
silica plates, extractedtwicein IHW, anddried. Glycolipidstandards were isolated previously in this laboratory (38, 44-48). To generate
sialosyl gal-globoside (SGG), disialosyl gal-globoside (DSGG) stan-dard was partially desialylated with 1% acetic acid at 100°C for 5-10 min.
Bacterial labeling. E. coli R45 was isolated from awoman with cystitisandhas the papfamilygenotype,expressingboth the P and F adhesins(9). Organismswerekeptfrozenin 50%glycerol/50%Luria broth (49) at -70°C until the night before use, when they were streaked ontoLuria agarplates (49)topromotefimbrialexpression
and grownovernightat37°C.On theday ofuse,bacteriawerescraped from the plate,resuspendedinM9medium(49)withoutaminoacids,
and shaken at37°Cfor40 min todeplete cellularmethioninestores.
[35S]Methionine
(Trans3"S-label,
1137Ci/mmol,ICNBiochemicals, Irvine, CA)wasthenadded at 200,Ci!
1010 organisms,the cells were shakenfor1h,washedthreetimesin PBS andresuspendedin PBS (pH7.3),and counted in ascintillationcounter togiveafinal activityof
0.01-0.02cpm perorganism.
Bacterialoverlay assays.Glycolipids (10-40
/g/lane)
werechro-matographedonglassHPTLCplates (Whatman Inc.)inchloroform/
methanol/water 50:40:10 with 0.05% CaCI2, with oneplaterunin
parallel for orcinol staining. Platesweredried, dippedfor 1 min in
diethyl ether containing0.5% polyisobutylmethacrylate, dried, and
preincubatedin 1%BSA/PBS for 1 h, and then washed three times in PBS.Radiolabeled bacteria were overlaid (108cpm total per plate) and theplates were gently rocked for 1 h, washedfourtimes inPBS,and
subjectedtoautoradiography.
Radioimmunoassay and antibody overlay assays. Glycolipids
sepa-rated onHPTLC were immunostainedaccordingto theprocedure of
Magnaniet al.(50) asmodified by Kannagi et al. (51 ). Radioimmuno-assays of GSLsimmobilizedinmicrotiter wells wasperformed
accord-ingtothemethodof Karlsson (40), with some modifications.
Specifici-ties and sourcesof mAbs aregivenin TableI.
Immunofluorescence
staining ofvaginal epithelial cells. Vaginalepi-thelial cells were washed three times in PBS (pH 7.3), counted in a
Table I.
Speciicities
ofMonoclonal AntibodiesMonoclonal
Antigen antibody Reference
Lewis' CA3F4 ATCC
Sialosyl gal- ID4 Produced in this laboratory globoside
DisialosylI NUH2 Produced in this laboratory (48) A type I,II,III AH16 Produced in this laboratory (55) A typeIII,IV HH5 Produced in this laboratory (56)*
RabbitIa CRL 1760 ATCC
*mAb HH5 has the same antigenic specificity as mAb MBrl
de-scribedin thereference.
hemocytometer, and - 3 x 104cellswere resuspended in PBS. Cells
were incubated on ice or at room temperature with the primary mAb or noantibody for 1 h, washed three times in PBS, and incubated with the
FITC-conjugated secondary antibody (diluted 1:100) on ice for 30 min.Primary mAbs were undiluted ID4 or one oftwo control mAbs to unrelated antigens: NUH2 (undiluted) or CRL1760(ATCC; undi-luted).Afterthreeadditionalwashes in PBS, stained cells were evalu-ated in ablinded fashion byexaminingeachfield sequentially using fluorescent microscopythenlight microscopy. Cellswithfaintor no
stainingwerescored asunstainedandall others wereconsidered posi-tive.
Results
Binding of
E.coli
R45toGSLstandards.
Thebinding ofmeta-bolically radiolabeled
E.coli strain
R45toglycolipid
standards separated on HPTLC plates was determined using a broad panel of GSLs with varying carbohydrate moietiesasshownin TableII. Aspredicted
from itsgenotypeand phenotype (P and Fadhesinpositive),
thebacteriabound toglobo-series GSLs containingthe minimalpap-binding moiety gal a 1-4galaswellas several
of
the suggestedreceptors for the F adhesin, includinggloboside,
gal-globoside, the Forssman antigen, and globo-A and -H. The organism did notbind to GSLs with structures unrelated to theglobo-series
GSLs, such as B1 (ABH-Bantigenon a type 1 core) and thegangliosides asialo
GM1 andasialo GM2. In
pilot experiments
whereorganisms were grown underconditions thatinhibit
P fimbrialexpres-sion,
binding
of severalE.coli
isolates withPand/or
Fpheno-types toGSLsinbacterialoverlayassayswasminimal or ab-sent.Threewild-typeE.coli isolatesnegativefor either the Por
theF
phenotypes
failed tobind GSLs, confirmingpublished reportsof thebinding of otherphenotypically
negative isolates toGSLs inbacterial overlayassays(30, 32).Binding of
E.coli R45toglycolipids extractedfrom vaginal
epithelial cells.
Total upper- andlower-phase
GSLswereex-tracted from
pooled vaginal epithelial
cells collected from three groupsof healthy collegestudents: 10nonsecretors(8 withahistory ofrecurrentUTI and 2without), 12 secretorswitha
history of recurrent UTI, and 18 secretorswithout recurrent
infections.Thedistribution of ABO histo-bloodgroup
pheno-types amongthewomenin thegroups was
comparable.
E.coli strainR45(isolated
fromoneofthepatient's UTIs)
wasTable II. Binding ofEscherichia coli R45 to Glycosphingolipids
Symbol Structure Binding
CMH Glc61-Icer
CDH Gal,B1-4
Glcfl
-IcerCTH Gala 1-4
Gal#1l-4
Glc#l-lcer +Globoside Ga1NAc 1-3Galal-4 Gal,B1-4 Glc,Bl-lcer +
Gal-globoside
Ga,1#1-3 GalNAcfl1-3
Gala 1-4 Gal, 1-4GlcjtI-Icer
+Forssman GalNAcal-3
Ga#1
1-3GalNAc#31-3
Galal-4 Gal,B1-4 Glc,Bl-lcer + Globo-H (Fuca1-2)Gal,B1-3 GalNAc,B1-3Galal-4Gal#11-4
Glc, I-Icer +Globo-A GalNAca1-3 (Fuca1-2)Gal,B1-3 GalNAc,B1-3 Galal-4
Ga#1
1-4Glc,Bl-lcer +SGG NeuAca2-3 Gal 1-3Ga1NAc,81-3Gala 1-4
Gal#
1-4Glc(31-lcer +DSGG NeuAca2-3 (NeuAca2-6)Gal,B1-3 GalNAcB1-3 Gala 1-4
Gal#1-4
Glc(Il-lcer +ASGM1
Gal#
1-3GalNAc 1l-4 Gal1 1-4GlctIl-lcer
-ASGM2 GalNAcB1-4 Gal 1-4GlcjIl-lcer
-nLc6 Gal,1l-4GlcNAc,1-3Gal3l1-4GIcNAcl1-3Gal,14Glcl1-1lcer
-BI Galal-3(Fucal-2)Gal,Bl-3GlcNAc#1-3Gal,B1-4GlcIl-lcer
-CMH, ceramidemonohexoside;CDH, ceramidedihexoside; CTH,ceramide trihexoside(globotriaosylceramide); globoside, globotetraosylcera-mide;gal-globoside,galactosylgloboside;SGG,sialosyl gal-globoside; DSGG, disialosyl gal-globoside;ASGM 1, asialo GM 1; ASGM2, asialo
GM2;nLc6,lacto-N-norhexaosylceramide;B1, type 1 chain B antigen; Glc, glucose; cer, ceramide; Gal, galactose; GalNAc,
N-acetylgalp-os-amine; Fuc,fucose;NeuAc,neuraminic acid;GlcNAc,N-acetylglucosamine.
unique glycolipid band in the total upper-phase GSLs from nonsecretors, which was not seen in GSLs from either group of secretors(datanotshown). This experimentwasrepeated us-ingpooled vaginal cells from 1O nonsecretors and 20 secretors (10withand 10withoutahistory ofrecurrentUTI) with iden-tical results. No differencesinbindingweredetectedin either experimentwhen secretors withand without a history of recur-rentUTIwerecompared.
Subsequent experiments were conducted to identify the unique E. coli R45-binding GSL band detectedin nonsecre-tors'
vaginal
epithelial cells. Totalupper neutral GSLs and total uppergangliosides
wereextracted from vaginal epithelialcellsA
pooledinequal quantitiesfrom five nonsecretors and nine se-cretorsand reactedwith radiolabeled E.coliR45. In total gan-glioside fractionsfrom nonsecretors but not from secretors, the organism boundtobandscomigratingwith SGG and DSGG standardsand also reacted with the standards themselves (Fig.
1 A). The totalamountofglycolipid placedineachlane was normalized baseduponequivalentnumbers of cells. As can be seenfrom theorcinolTLC,the chemical amounts of total up-per-phase
gangliosides
fromnonsecretorsandsecretors are vi-sually comparable (Fig. 1 B, lanes 8 and9).InFig. 1 A, E. coli R45 also boundto abandseen in the totalupperneutralGSLs from nonsecretors' vaginal epithelial
B
CTH-Globoside
pw~~~~~~i
§ -GG
3-DSGG
! e
_6 0 *:.
1246 9
1 2 3 4 5 6 7 8 9 10
i .s
.a_...
.. .....
,.*:
-,},sES
...zi-;
k
+
_
_ i. _
'. 3; '8.
*_ W.
1 2 3 4 5 6 7 8 9 10
Figure1. E.coli R45binding to glycolipids extracted from vaginal epithelial cells. Glycolipids extracted from vaginal epithelial cells were sepa-ratedonHPTLC plates and overlaid with the metabolically[35S]methionine-labeleduropathogenic E.coli strain R45 as described in Methods.
Ashowsarepresentative autoradiogram of bacterial binding to GSLs extracted from equal quantities of pooled vaginal epithelial cells from five
nonsecretorsandninesecretorsandtoGSL standards.Anorcinol-stained HPTLC of the samples is shown in B. Lane 1 containsceramrde
trihexoside(globotriaosylceramide)andgloboside standards; lane 2 contains total lower phase GSLs from the nonsecretors; lane 3 contains total lowerphase GSLs from the secretors; lane 4 contains the Forssman antigen standard; lanes 5 and 6 contain total upper neutral GSLs from
nonsecretorsandsecretors, respectively; lanes 7 and 10 contain gal-globoside, SGG, and DSGG standards; lanes 8 andQcontaintota- upper phasegangliosidesfromnonsecretorsand secretors, respectively.
z"..,L...
'-70
60
.--N 50 v o 40
2 30
0-O 20
10
---GG
---SGG
o...4.
I .It
I I I
II
I I I
II II
I I I I
200 100 50 25 12.5 6.25 3.1 1.56
Glycolipid (ng)
1
2
3
4
5
67
Figure2.Immunostaining ofglycolipids extractedfrompooled
vagi-nal epithelial cells ofnonsecretorsandsecretors.Total gangliosides
wereextracted from vaginal epithelialcellsofnonsecretorsand
secre-torspooled inequal quantitiesand separatedonHPTLC platesas
describedinMethods. Bands fromnonsecretors' vaginal epithelial
cells that comigrated with SGGandDSGG standardswereelutedas
described, chromatographed withthe totalganglioside fractionsand GSLstandards, and overlaid withmAbID4(SGG). LanesI and 2 contain totalgangliosides from nonsecretorsandsecretors,
respec-tively. Lane 3 contains theeluted bandfromnonsecretorsthat
comi-gratedwithSGGand lane 4 contains the eluted band from
nonsecre-torsthatcomigrated with DSGG.Lanes5and 7containSGG
stan-dards fromhumanpancreasand lane6contains gal-globoside
standard from humanpancreas(notetraceof SGG).
cells(lane 5) butnotseeninthe totalupperneutralGSLs from secretors' cells(lane 6). Although itappearsto beuniqueto nonsecretors, we have not attempted to characterize this moiety, which correspondstoaneutral, extended globo-series GSL(probablyextendedgal-globoside).Theentityappearsto
bepresent intoosmallaquantity for feasible characterization,
basedontheintensity of orcinolstaininginthisregionand the
strength of the signal from bound bacteria.Inaddition,
bacte-ria bindtolower-phase lipids inasimilar fashion insecretors
(lane 2) andnonsecretors(lane 3). This binding appearedto
be nonspecific and similartoaphenomenonseenin immuno-staining experiments (Nudelman and Hakomori, unpublished data).
Immunostaining and radioimmunoassays of glycolipids extractedfrom pooled vaginal epithelialcells. To confirm the identity of SGG in the total upperganglioside fraction from
nonsecretors' vaginal epithelial cells, bands comigratingwith
SGGandDSGGwereeluted from HPTLCplates,
chromato-graphed along with total gangliosides fromnonsecretors and secretors, and reacted with mAbID4directedagainstSGG. As
showninFig. 2, the mAb stainedabandcomigratingwith SGG
in lane 1, containing the totalupperganglioside fraction from
nonsecretorsand in lane3,containing the putative SGG eluted from thenonsecretors.Nostainingwas seeninthe total
ganglio-side fraction from the secretors(lane 2) nor in theputative
DSGG material from thenonsecretors(lane 4).
TheSGGand DSGG bands from the nonsecretors' vaginal
Figure 3.Radioimmunoassay with mAb to sialosyl gal-globoside and glycolipids from nonsecretors. Glycolipid standards and the eluted glycolipid bands comigrating with SGG and DSGG from nonsecre-tors'vaginal epithelialcells wereapplied to microtiter wells and sub-jectedto aradioimmunoassayusingamonoclonal antibody to SGG (ID4), asoutlined in Methods. Dotted line indicates SGG standard from human pancreas; solid line indicates band comigrating with SGG eluted from TLC of nonsecretors' extracted vaginal epithelial cells; dashed line indicates eluted band comigrating with DSGG from nonsecretors and thefollowing GSL standards: gal-globoside andDSGG from human pancreas, GDla from bovine brain, and disialosyl I from humanplacenta.
epithelial cellextracts werethen reacted with ID4 ina radioim-munoassay shown in Fig. 3. The antibody reacted with the SGG band fromnonsecretors and withthe SGG standard but did not bindto thenonsecretor DSGG bandnor to theDSGG, gal-globoside,GDl a, ordisialosylI standards.
To determine if the nonsecretor phenotypes of the patients asdeterminedbyblood and salivatestingwasalsoexpressedin
vaginal epithelialcells, totalupper-phase GSLsfrom nonsecre-tors werechromatographed and reacted withmAbs"anti-Lea" and AH16(againstA type I, II, and III). Inthenonsecretor
GSLs,abandcomigratingwithLewisa standardwasbrightly stained but no stainingwasobserved with mAb AH16 (data notshown). Inaddition, mAbs against ceramidetnhexoside (globotriaosylceramide) and globoside/gal-globoside stained
TableIII. Secretor,Lewisa, andABOPhenotypes andAntibody Binding
Antibody
binding
Patient Secretor ABO Lewise
number phenotype phenotype phenotype HH5 ID4
I NS A + - +
2 NS 0 + - +
3 NS 0 + - +
4 NS A + - +
5 NS B + - +
6 S 0 - -
-7 S 0 - -
-8 S AB - +
-9 S AB - +
-10 S A - +
-NS, nonsecretor; S,secretor.
...
11
B
D
=SGG
1
2
3
4 5 6 7 8 9
101
2
3 4
5 6
7
8 9
10 11Figure4.Immunostainingofglycolipids extractedfromvaginal epithelial cells ofindividualnonsecretorsandsecretors.Totalupperphase gly-colipidsfromvaginal epithelialcells of five individualnonsecretorsand fiveindividualsecretors.(IO'cells/patient)wereextracted,
chromato-graphedonHPTLC, and overlaidwith(A) mAbs ID4 (SGG) and (B)HH5(AtypeIII,IV)asdescribed in Methods.Autoradiographsfrom
antibodyoverlayassays areshown.Secretor,Lewisa,and redblood cell ABOphenotypesofpatients representedin A and Baregivenin Table
III.Lanes1-10contain samplesinsequencefrompatients1-10.Lane11in AcontainsSGGstandardfrom humanpancreasand in Bcontains upperneutral GSLs from type A human red blood cells.
the cell extracts, confirming that globo-series GSLs are
ex-pressedinvaginal epithelial cells (datanotshown).
Immunostaining
ofglycolipids extractedfrom vaginal
epi-thelial cellsfrom
individual patients. To determine whether SGG isfoundinallnonsecretors(and is absentinallsecretors)orpresentonlyinsomenonsecretors,weextracted totalupper
and, lower phase GSLs from vaginal epithelial cells collected fromfiveseparate nonsecretorindividualsaswellasfrom five
separate secretor individuals. Equal quantities of cells (107 cellsperpatient)wereextracted andchromatographedas
de-scribed and reactedwith antibodies ID4 (against SGG)and
HH5 (against A types III and IV). Thesecretor, ABO, and Lewisared bloodcellphenotypes of the patients (determined by hemagglutination assays)aregiveninTable III. As shownin
Fig. 4A, SGGwasdetected intotal upper-phaseGSLs from
eachof thefive nonsecretors(lanes 1-5)but not fromanyof
the secretors(lanes 6-10). In Fig. 4B, mAb HH5 detected
A-reactive substancesinGSLsfrom the threesecretorswith A
orABphenotypes (lanes8-10)butnotfrom the nonsecretors
(lanes 1-5).
Immunofluorescence
assays with nativevaginal epithelial cells.Todemonstratethat SGG ispresentandaccessibleonthesurfaceofnativevaginal epithelial cells fromnonsecretorsbut
notsecretors,immunostaining of cells fromtwoindividual
se-cretors andtwo nonsecretors in threetrialsusing mAb ID4was
compared. Asshown in Fig. 5, the majority of nonsecretors' vaginal epithelial cellswerestained whereas littleor nostaining
ofsecretors' cells was observed. Staining of cells from both patient groups with control mAbs NuH2 and CRL1 760 (against unrelated antigens)wascomparable and minimal.
Discussion
Severallinesofevidencehavesuggestedthat theincreased
sus-ceptibility to recurrent UTI observed in some otherwise
healthywomen maybe explained by genetic factors
influenc-ingthedensity and/or specificity of bacterialreceptors avail-abletomediate colonization of their uroepithelial cells (3-8). Although it is known thatnonsecretors ofhisto-bloodgroup
antigensareoverrepresentedamong womenwith ahistory of
recurrentUTIs (5-7) and that uroepithelial cells from nonse-cretorsshowenhanced E. coliadherencecompared with cells fromsecretors(8), the biochemical basis for these observations hasnotbeen clarified. In this study,wehave shown that
nonse-cretorsexpresstwo unique GSLreceptorsforE. coliR45 on
theirvaginal epithelial cells, SGG and DSGG, whichwerenot
expressed in secretors' vaginal epithelial cells. SGGwas
accessi-
100-c
_
=
D
CO)
.0 E
z
80
60
40
20,
o0
|
Se/-[]z
se/seNo ID4
Figure5. Degreeof immunofluorescentstaining ofvaginal epithelial
cells collected from individualnonsecretorsandsecretorsandstained
with mAbtoSGG. Immunofluorescenceassaysperformedonvaginal epithelialcellsfromtwosecretorsandtwononsecretorsusingmAb
ID4(against SGG). 50cells/patientintwoassaysand20 cells/pa-tient inoneassay wereexamined andscored forintensityof
fluores-centstaining.The totalnumber of cellsstaininginthreeassays
com-binedareshown.Cells with faintornostainingwereconsidered neg-ativewhereasanyotherdegreeofstainingwasconsideredpositive.
ble tobinding by mAbs on the surface of the cells; we were
unable to test for the surface accessibilityofDSGG to mAb binding because of the lack of anappropriatemAb. SGG and DSGG were detected in extracts ofvaginal epithelial cells pooled from a total of 10 nonsecretors and also in cells col-lected and separately extracted fromfiveadditionalindividuals withthese/sephenotype,suggestingthat theexpression ofthis antigenis uniform for nonsecretors.Considering all bacterial-bindingexperiments using both pooled and individuals' vagi-nal epithelial cells, westudieda total of 35 secretors' and 15 nonsecretors' epithelial cells. A large epidemiological study would be necessary toconfirmtheuniversalityofourfindings, butourdatasupportthehypothesis that the Se locuscontrols theexpressionofABO bloodgroupantigen variants of globo-seriesGSLs invaginalepithelial cells,ananatomic site which plays animportant role in thesequenceofcolonizationevents preceding the development ofUTI in women. As described above, severalglobo-seriesGSLs are known receptors for uro-pathogenic E. coliand are present in otherurinarytract sites (20-32,34, 35), but the influences of the Segene on globo-se-riesbiosynthesis invaginalepithelium hasnotbeen previously demonstratedto ourknowledge. Leffler etal. (52) suggested that globo-series GSL
synthesis
in voideduroepithelial
cells from adults depended onABO and secretor status. The pres-enceof SGGandDSGGinuroepithelial
cellsfrom nonsecre-tors may accountfor the increasedbinding
ofE. colitotheir cells and fortheirincreasedsusceptibility
torecurrent UTI.Inadditiontosupporting theroleofagenetic hostfactor in thepathogenesisof recurrent UTIin women, our data clarify themechanism bywhich this occurs. Others havespeculated thatE. colireceptors onuroepithelial cells from nonsecretors are moreaccessible because of thelackof fucosyltransferase-mediated synthesis ofA, B, and Hantigens, whereas in secre-tors,thepresence ofhisto-bloodgroup
antigens
onepithelial cellsmight
shield thereceptorsandpreventbacterial binding (6, 8, 53). However,reportscorrelating
secretor statuswith the expression of histo-bloodgroupantigens
ontype I-IV carbohy-dratecore structuresin theurinary
tract areconflicting
(54). Thecorrelationofsecretor statewithexpression ofABH anti-genshasonlybeenclearly demonstrated forantigens
carriedon type Ichains (52, 54), but uropathogenicE.coli havenotbeen reportedtobindto type I coreGSLs(52). Thiswasalsotruein ourstudy.InGSLsextractedfromvaginalepithelial
cellsfrom nonsecretors and secretors,respectively,
weidentified major bandscorresponding
in TLCmigration
toLewisa and Lewisbantigens.
In extracts from nonsecretors, the Lewisa band reactedwithanmAbdirectedtoLewisa. However,E.coli R45 didnotbindtoeither of the LewisaorLewisb bands. Incon-trast, we have shown that onvaginal
epithelial
cells,thereisa correlation between secretor statusandexpression of well-de-scribedreceptors foruropathogenicE.coli, the type IV chain (globo-series)GSLs. Rather than failing to shield bacterial re-ceptors because of their lack offucosyltransferase-mediated
synthesis ofA,B, and Hantigensontype 1 chains, nonsecre-torssynthesize unique
sialylated,
E.coli-binding
derivatives of theglobo-series GSLs.To ourknowledge, influences oftheSe gene onglobo-series biosynthesisinvaginal epithelium
have notbeenpreviouslydemonstrated.Genetic
variability
in glycosylation encoded by blood group genesisgenerally thought to beinvolvedin"masking" orshielding glycan
receptors formicrobial pathogens.Globo-series GSLs characterized byaterminal orinternal
galactose
a1-4 galactose moietyarepreferredreceptorsforthe E.coli P fimbrial adhesin (20-28). Previous studies have shownthat the binding of P-fimbriated E. coli to globo-series GSLs changes when the disaccharidereceptorsite is further modified by elongation ofthe saccharide chain (28). The genetic variabil-ityinherent to bloodgroup antigens implies that a proportion of the population lacks certain glycanstructures.In such per-sons,competition forterminalglycosylationof the precursor gal-globosideis sharedbyboth afucosyltransferaseand a sialyl-transferase. Apparently, the affinity of thefucosyltransferase for the terminal galactose is greater and hence theGlobo H structureissynthesizedtotheexclusion ofanyterminally sialy-lated structures. In contrast, nonsecretors lack such a fucosyl-transferase inepithelialcells andsynthesizeSGGand DSGG in vaginal epithelium throughsialylationof theprecursor gal-glo-boside,with no competition for terminal fucosylation bythe absent Se gene-encoded fucosyltransferase. As evidencedby ourresults,thissialylationdoes not interfere with the receptor activityof theglobocore.
In summary, we have shown that nonsecretors of histo-blood group antigens synthesize unique GSLs, SGG and DSGG,ontheirvaginalepithelial cells, which arenotfoundin cells from secretors. As these moieties serve asreceptors for uropathogenicE.
coli,
thisfindingmay be a biochemical expla-nation for the increasedadherenceofbacteriatothesewomen's uroepithelialcells and for theirpropensitytodeveloprecurrent UTIs. However, we have also shown that otherglobo-series GSLs known to bereceptorsforuropathogenicE.coliare pres-entin GSL extracts from vaginal epithelial cells and bind bacte-ria inHPTLCoverlay assays.Innonsecretors, SGG and DSGG maybe more abundant oraccessibleonvaginal epithelialcells than otherE.coli receptorsortheaffinityofbacterialbinding for these moieties may be higherthan the affinity for other globoseriesGSLs. We arepresently pursuing studies compar-ingtherelativedensities and bacterialbinding affinities
of the variousE.coli receptorsonvaginal epithelial cells fromnonse-cretorsandsecretors todetermine the relativecontributionsof thesereceptorstotheobserveddifferencesinbacterial
adher-ence to
vaginal
epithelial
cells from the twopatient
popula-tions.Finally,
ourfindings
may havepossible
clinical applica-tions. Forexample,bacterialreceptoranalogues, perhaps incor-porated in a spermicidalpreparation,
could be applied topicallytothevaginal
epitheliumtoinhibitbinding
of E. coli andsubsequent development
ofUTI.Wearepresently
pursu-ingthesepossibilitiesinvariouspatient
populations.Wethank ThomasM.Hooton,M.D.forproviding patientsandhelpful suggestions;CarolWinter,R.N. andLisa Kuwamurafor assistance in
collectingspecimens; and Sharon Baker-Johnson and Dr.Sheila Luke-hartforproviding assistance with materials for and adviceon
immuno-fluorescence assays.
This work was supported in part by Grants DK-40045 and
Al-07044 fromtheNational Institutes of Health.
References
1.Mabeck,C.E.1972. Treatmentof uncomplicatedurinarytract infectionin
nonpregnant women.Postgrad.Med.J. 48:69-75.
with Enterobacteriaceae in recurrent urinary tract infections. J. Urol. 113:214-217.
3.Schaeffer,A.J., J. M. Jones, and J. K. Dunn. 1981.Association of in vitro Escherichia coliadherence tovaginalandbuccalepithelialcellswith
susceptibil-ityofwomen torecurrenturinary-tract infections. N. Engl. J. Med. 304:1062-1066.
4.Svanborg-Eden, C., and U. Jodal. 1979. Attachment of Escherichia coli to urinary sedimentepithelialcells fromurinarytractinfection-prone and healthy
children.Infect.Immun.26:837-840.
5. Kinane, D. F.,C. C. Blackwell,R. P.Brettle, D. M.Weir,F. P.Winstanley,
and R.A.Elton. 1982. ABO blood group,secretor stateandsusceptibilityto
recurrenturinary tractinfectioninwomen. Br.Med.J.285:7-9.
6.Sheinfeld,J.,A.J.Schaeffer,C.Cordon-Cardo, A.Rogatko, and W. R. Fair. 1989.Associationof theLewisblood-group phenotype withrecurrenturinary
tract infections inwomen.N.Engl.J. Med.320:773-777.
7. Hooton, T. M., C. Johnson, P. L. Roberts, and W. E. Stamm. 1989. The
associationof sex, withorwithoutdiaphragmandspermicideuse, with UTI. In Program and Abstracts. 29th Intersci.ConfAntimicrob. Agents Chemother. Proc. 8.(Abstr.)
8.Lomberg, H., B. Cedergren,H.Leffler,B.Nilsson,A.-S.Carlstrom,and C.
Svanborg-Eden. 1986.Influenceof blood groupontheavailabilityof receptors
for attachmentofuropathogenicEscherichiacoli.Infect.Immun. 51:919-926. 9.Stapleton,A., S.Moseley,and W. E.Stamm. 1991. Urovirulence
determi-nants inEscherichia coliisolatescausing first-episodeandrecurrentcystitisin women.J. Infect.Dis.163:773-779.
10. Arthur, M., C. E. Johnson,R. H.Rubin,R. D.Arbeit,C.Campanelli,C. Kim, S. Steinbach, M. Agarwal, R.Wilkinson,and R.Goldstein.1989. Molecu-larepidemiology ofadhesin andhemolysinvirulence factors among uropatho-genic Escherichia coli. Infect.Immun.57:303-313.
11.Sandberg, T., B.Kaijser,G. Lidin-Janson, K. Lincoln, F. Orskov,I. Ors-kov, E. Stokland, and C. Svanborg-Eden. 1988. Virulence ofEscherichia coliin relation to host factors in women withsymptomaticurinary tract infection. J.
Clin.Microbiol.26:1471-1476.
12.O'Hanley, P.,D.Low,I. Romero,D.Lark, K. Vosti, S. Falkow, and G.
Schoolnik.1985.Gal-gal bindingandhemolysin phenotypesand genotypes
asso-ciatedwithuropathogenic Escherichia coli.N.Engl.J.Med. 313:414-420.
13. Lidefelt, K.-J., I. Bollgren, G. Kallenius, and S. B. Svenson. 1987. P fimbriatedEscherichia coliinchildren with acutecystitis.ActaPaediatr.Scand. 76:775-780.
14.Johnson, J. R.,P. L.Roberts, and W. E. Stamm. 1987.Pfimbriaeand other virulencefactorsinEscherichiacoli urosepsis: associationwithpatients' characteristics.J.Infect.Dis.156:225-229.
15.Jacobson, S. H., L.-R. Lins, S. B. Svenson, and G. Kallenius. 1985.P
fimbriatedEscherichiacoliin adults with acutepyelonephritis. J. Infect.Dis. 152:426-427.
16. Vaisenen, V., L. G. Tallgren,P. H.Makela, G. Kallenius, H. Hultberg, J. Elo,A. Siitonen,C.Svanborg-Eden, S.B.Svenson, and T. Korhonen. 1981.
Mannose-resistanthaemagglutinationandPantigen recognitionare characteris-tic ofEscherichia colicausing primary pyelonephritis.Lancet2:1366-1369.
17.Kallenius,G., S. B. Svenson,H.Hultberg, R. Mollby,I.Helin, B. Ceder-gren, and J.Winberg. 1981. Occurrence of P-fimbriatedEscherichiacoliin uri-nary tract infections. Lancet 2:1369-1372.
18.Vaisanen-Rhen, V., J. Elo, E.Vaisanen,A.Siitonen,I. Orskov, F. Orskov, S. B.Svenson, P. H.Makela, andT. K.Korhonen. 1984. P-fimbriated clones amonguropathogenic Escherichia colistrains.Infect.Immun.48:149-155.
19.Westerlund, B.,A.Siitonen,J.Elo, P. H.Williams,T.K.Korhonen, and P. H.Makela. 1988.Properties of Escherichia coliisolates from urinary tract
infectionsin boys. J.Infect.Dis. 158:996-1002.
20. Kallenius, G.,R.Mollby, S. B. Svenson, J. Winberg, A. Lundblad, S. Svennsson, and B.Cedergren.1980. The pkantigenasreceptorforthe haemagglu-tinin ofpyelonephritic Escherichia coli.FEMS(Fed.Eur.Microbiol.Soc.)
Mi-crobiol.Lett.7:297-302.
21.Leffler,H., andC. Svanborg-Eden. 1980. Chemical identification of a
glycosphingolipidreceptorfor Escherichia coliattaching to human urinary tract cells andagglutinatinghumanerythrocytes. FEMS(Fed.Eur.Microbiol.Soc.) Microbiol.Lett.8:127-134.
22.Svanborg-Eden, C., and H. Leffler. 1980. Glycosphingolipids of human
urinarytract aspossible receptors foradheringEscherichia coli.Scand.J.Infect.
Dis.Suppl.24:144-147.
23.Kallenius, G.,R.Mollby, S.B.Svenson,J.Winberg, and H. Hultberg. 1981. Identification ofacarbohydrate receptor recognized by uropathogenic
Escherichia coli.Infection.8(Suppl. 3):S288-S293.
24.Kallenius, G., S.B.Svenson,R.Mollby,B.Cedergren, H. Hultberg, and J. Winberg. 1981. Structure of carbohydrate part of receptor on human uroepithe-lial cells forpyelonephritogenic Escherichia coli.Lancet2:604-606.
25.Leffler,H., and C. Svanborg-Eden. 1981. Glycolipid receptors for
uro-pathogenic Escherichia colionhumanerythrocytes and uroepithelial cells. Infect.
Immun.34:920-929.
26. Kallenius, G., S. B. Svenson, R. Mollby, T. Korhonen, J. Winberg, B. Cedergren, I. Helin, and H. Hultberg. 1982. Carbohydrate receptor structures recognized by uropathogenic Escherichiacoli.Scand. J. Infect. Dis. Suppl. 33:52-60.
27. Svenson, S. B., H. Hultberg, G. Kallenius, T. K. Korhonen, R. Mollby, and J. Winberg. 1983. P-fimbriae of pyelonephritogenic Escherichia coli: identifi-cation andchemicalcharacterization of receptors.Infection. 11:73/61-79/67.
28. Bock, K., M. E. Breimer, A. Brignole, G. C. Hansson, K.-A. Karlsson, G. Larson, H. Leffler, B. E. Samuelsson, N. Stromberg, C. Svanborg-Eden, et al. 1985. Specificity of binding of a strain of uropathogenicEscherichia colito gala
1-4gal-containing glycosphingolipids. J. Bio.Chem.260:8545-8551. 29. Lund, B., B.-I. Marklund, N. Stromberg, F. Lindberg, K.-A. Karlsson, and S.Normark. 1988. UropathogenicEscherichia colican express serologically iden-ticalpili of different receptor bindingspecificities. Mo. Microbiol.2:255-263.
30. Lindstedt, R., N. Baker, P. Falk, R. Hull, S. Hull, J. Karr, H.Leffler,C. Svanborg-Eden, and G. Larson. 1989. Binding specificities of wild-type and clonedEscherichia colistrains that recognize globo-A.Infect.Immun. 57:3389-3394.
31. Karr, J. F., B. J. Nowicki, L. D. Truong, R. A. Hull, J. J. Moulds, and S.I.
Hull. 1990. pap-2-encoded fimbriae adhere to the P blood group-related
glyco-sphingolipid stage-specific embryonicantigen 4 in the human kidney. Infect.
Immun. 58:4055-4062.
32. Stromberg, N., B.-I. Marklund, B. Lund, D. Ilver, A. Hamers, W. Gaastra, K.-A. Karlsson, and S. Normark. 1990.Host-specificity ofuropathogenic
Esche-richiacolidepends on differences in binding specificity to gal a 1-4 gal-containing isoreceptors. EMBO (Eur. Mol. Biol.Organ)J.9:2001-2010.
33. Breimer, M. E., and P.-A. Jovall. 1985. Structuralcharacterization ofa
blood groupAheptaglycosylceramidewithglobo-seriesstructure: themajor
glyco-lipid based blood group A antigen of human kidney. FEBS (Fed. Eur. Biochem. Soc.) Lett. 179:165-172.
34. Breimer, M. E., G. C. Hansson, and H.Leffler.1985. The specific
glyco-sphingolipid composition of human ureteral epithelial cells. J. Biochem. 98:1169-1180.
35. Lindstedt, R., G. Larson, P. Falk, U. Jodal, H. Leffler, and C. Svanborg. 1991. The receptor repertoire defines the host range for attaching Escherichia coli strains that recognize globo-A. Infect. Immun. 59:1086-1092.
36.Clausen, H., and S.-I. Hakomori. 1989. ABH and related histo-blood group antigens: immunochemical differences in carrier isotypes and their distri-bution. Vox Sang. 56:1-20.
37. Kannagi, R., S. B. Levery, and S.-I. Hakomori. 1984. Blood group H antigen with globo-series structure. Isolation and characterization from human blood group 0 erythrocytes. FEBS(Fed.Eur.Biochem.Soc.) Lett. 175:397-401. 38. Clausen, H., K. Watanabe, R. Kannagi, S. B. Levery, E. Nudelman, Y. Arao-Tomono, and S.-I. Hakomori. 1984. Blood group A glycolipid(A')with globo-series structure which is specific for blood group A, erythrocytes: one of the chemical bases for A, and A2 distinction. Biochem. Biophys. Res. Commun. 124:523-529.
39. Hansson, G. C., K.-A. Karlsson, G. Larson, N. Stromberg, and J. Thurin. 1985. Carbohydrate-specific adhesion of bacteria to thin-layer chromatograms: a rationalized approach to the study of host cell glycolipid receptors.Anal.
Bio-chem. 146:158-163.
40.Karlsson, K.-A., and N. Stromberg. 1987. Overlay and solid-phase
analy-sisofglycolipidreceptorsforbacteria and viruses.MethodsEnzymol. 138:220-232.
41.Daifuku, R., and W. E. Stamm. 1986. Bacterial adherence to bladder
uroepithelialcells in catheter-associated urinary tract infection. N. Engl. J. Med. 314:1208-1213.
42.Folch, J., M. Lees, and G. H. Sloane Stanley. 1957. A simple method for theisolation and purification of total lipides from animal tissues. J.Bio.Chem. 226:497-509.
43.Skipski,V. P.1975.Thin-layer chromatography of neutral glycosphingo-lipids.MethodsEnzymol. 35:396-425.
44.Clausen,H.,M.Stroud, J. Parker, G.F.Springer, and S.-I. Hakomori. 1988. Monoclonal antibodies directed to the blood group A associated structure, galactosylA:specificity and relation to theThomsen-Friedenreichantigen.Mo. Immunol.25:199-204.
45.Hakomori, S., S.-M. Wang, and W. W. Young. 1977. Isoantigenic expres-sion of Forssmanglycolipid in human gastric colonic mucosa: its possible identity with "A-like antigen" in human cancer. Proc. Natl.Acad. Sci. USA. 74:3023-3027.
46.Kannagi,R., S.B.Levery,F.Ishigami, S.-I. Hakomori,L. H.Shevinsky,
B. B.Knowles, and D. Solter. 1983. New globoseries glycosphingolipids in
hu-manteratocarcinomareactive with the monoconal antibody directed to develop-mentally regulated antigen, stage-specific embryonic antigen 3. J.Bio.Chem. 258:8934-8942.
47.Nudelman, E., Y. Fukushi, S. B. Levery, T. Higuchi, and S.-I. Hakomori. 1986. Novel fucolipids of human adenocarcinomas: disialosyl Le' antigen
monoclonalantibody(FH7)definingthis structure. J. Biol.Chem. 261:5487-5495.
48. Nudelman, E.D.,U.Mandel,S. B.Levery,T.Kaizu,and S. Hakomori. 1989. Aseriesofdisialogangliosideswithbinary2-3sialosyllactosamine
struc-ture,defined by monoclonal antibody NUH2,areoncodevelopmentally regu-latedantigens. J. Biol. Chem. 264:18719-18725.
49.Maniatis,T., E. F.Fritsch,and J.Sambrook. 1982. MolecularCloning:A Laboratory Manual. ColdSpringHarborLaboratory, ColdSpring Harbor,NY. 545pp.
50.Magnani,J.L., D. F.Smith, and V.Ginsburg.1980. Detection of ganglio-sides thatbindcholeratoxin: direct bindingof'25I-labeled toxintothin-layer
chromatograms.Anal. Biochem. 109:399-402.
51.Kannagi,R., E.Nudelman, S. B. Levery, and S. Hakomori. 1982. A series of human erythrocyteglycosphingolipids reactingtothemonoclonalantibody directed to a developmentally regulated antigen, SSEA-1. J. Biol. Chem. 257:14865-14872.
52.Leffler,H., H.Lomberg,andC.SvanborgEden. 1989.Glycolipid
recep-torsforbacterial adhesiononhumanurinarytractepithelium:relationtoblood group and age. In Host Parasite Interactions inUrinaryTract Infections. E. H. Kass andC.Svanborg-Eden,editors.UniversityofChicago Press, Chicago. 93-99.
53.Schoolnik,G.K. 1989. How Escherichiacoliinfectsthe urinary tract. N.
Engl.J.Med. 320:804-805.
54.Orntorft,T.F.,H.Wolf,H.Clausen, S.-I. Hakomori, and E. Dabelsteen. 1988.Blood group ABO-related antigens in fetal and normal adult bladder
urothe-lium: immunohistochemical study of Type 2 chain structures with a panel of mousemonoclonal antibodies. Lab. Invest. 58:576-583.
55.Abe, K., S. B. Levery, and S.-I. Hakomori. 1984. The antibody specific to type 1 chainblood group A determinant. J. Immunol. 132:1951-1954.