NWWM.

;
:Ex .' twom I&IMMOM"M
.,l."

were washed once with buffer A [10 mM tris-HCI (pH Nde-2, CCTATTGACCATATGTCCACTTC. The Nde Cell Biol. 129,1617 (1995).
8.0), 10 puM phosphoserine, 10 ,uM phosphothreo- 1-to-Eco RI fragment of pBluescript-Plxl, encoding a 26. The Cdc2N133A-Acyclin B complex, prepared un-
nine, 10 p.M phosphotyrosine, 0.1 mM vanadate, and COOH-terminal fragment of Plxi, was ligated into der conditions that allow the phosphorylation of
0.1 % CHAPS] containing 500 mM NaCI, 5 mM EGTA, pVL1 393N-His (2). This procedure created pVL-His6- Cdc2 on Thr 61, was labeled on Thr'4 and Tyr15 by
20 mM ,-glycerolphosphate, and 1 p.M microcystin. Plxi -C. The PCR fragment was digested with Nde treatment with the Mytl kinase as described (2).
After three additional washes with the same buffer and ligated into pVL-His6-Plxl -C to yield a baculovi- 27. R. Hamanaka et al., J. Biol. Chem. 270, 21086
lacking microcystin, bound proteins were eluted from rus transfer plasmid encoding the full-length, histi- (1995); K. S. Lee, Y. L. 0. Yuan, R. Kuriyama, R. L.
the column with buffer A containing 500 mM NaCI and dine-tagged Plxl protein. Erikson, Mol. Cell. Biol. 15, 7143 (1995).
200 mM imidazole. The eluate was loaded on a Su- 22. The catalytically inactive Ni 72A mutant of Plxi was 28. A. Kumagai and W. G. Dunphy, Mol. Biol. Cell 6,199
perdex 200 column equilibrated with buffer A contain- created by PCR with the following oligonucleotides: (1995).
ing 50 mM NaCI, 10 mM EGTA, and 10 mM EDTA. p67-N1 72A-1, AACAGGGCCCCGAGCTTGAGGTC- 29. Recombinant His6-Xcdc25 or His6-Xcdc25-N was
Column fractions containing Xcdc25-specific kinase TCTGTG; and p67-N1 72A-2, TCGGGOCCCTGTTC- incubated for 20 min at 23°C with fractions in 20 pI of
activity were combined and incubated for 90 min at CTTAATGATGAAATGGAGG. 5 mM tris-HCI (pH 7.5), 10 mM MgCI2, 1 mM DTT,
4°C with a one-tenth volume of phosphocellulose Pl 1 23. For purification of various His6-Plxl proteins, Sf9 in- ovalbumin (0.1 mg/ml), and 50 ,uM ATP containing
(Whatman, Hillsboro, OR) prepared as described [Y. sect cells (2 x 107 cells) were treated for 3 hours with 1.6 pCi of [y-32P]ATP. After incubation, the samples
Wang and P. J. Roach, in Protein Phosphorylation: A okadaic acid 46 hours after infection. The cells were were mixed with gel sample buffer and subjected to
PracticalApproach, D. G. Hardie, Ed. (IRL Press, Ox- harvested, washed once with phosphate-buffered sa- SDS gel electrophoresis. In some cases, the reaction
ford, 1993), pp. 121-144]. The phosphocellulose was line, and lysed in 1 ml of buffer B [20 mM ,B-glycerol- was stopped by the addition of 10% phosphoric acid,
washed with 5 volumes of buffer A containing 50 mM phosphate, 10 mM Hepes-KOH (pH 7.5), 0.1 mM and portions (10 ,ul) of each sample were spotted
NaCI. Bound proteins were eluted in buffer A contain- vanadate, 5 mM EGTA, 5 mM ,B-mercaptoethanol, 10 onto P81 phosphocellulose paper (Whatman), which
ing 500 mM NaCI and 20 mM 3-glycerolphosphate. ,uM phosphoserine, 10 ,uM phosphothreonine, and was subsequently washed with 0.5% phosphoric
The phosphocellulose eluate fraction was dialyzed 10 ,uM phosphotyrosine] containing 150 mM NaCI, acid. In all cases, the incorporation of 32P into Xcdc25
against buffer A containing 25 mM NaCI and 20 mM 1% CHAPS, 3 ,uM microcystin, 1 mM PMSF, pepsta- was quantitated with the use of a Phosphorlmager

Downloaded from https://www.science.org at Universitts- und Landesbibliothek Bonn on August 24, 2023
,-glycerolphosphate and loaded onto a Mono 0 PC tin (10 jig/ml), chymostatin (10 pg/ml), and leupeptin with a known amount of 32P as the standard.
1.6/5 column (Pharmacia) equilibrated in the same (10 jig/ml). The clarified lysate was incubated with 30. Single-letter abbreviations for the amino acid resi-
buffer. The Mono 0 column was eluted with a linear nickel agarose, which was subsequently washed with dues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F,
gradient up to 1 M NaCI in the same buffer. buffer B (containing 500 mM NaCI and 0.1 % CHAPS) Phe; G, Gly; H, His; I, lIe; K, Lys; L, Leu; M, Met; N,
13. The Mono 0 fractions containing kinase activity were and eluted with buffer B (containing 25 mM NaCI and Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W,
pooled and loaded onto a 5 to 20% sucrose gradient 200 mM imidazole). The eluate was supplemented Trp; and Y, Tyr.
in buffer A containing 20 mM 3-glycerolphosphate. with 2 mM dithiothreitol (DTT) and 1 mM EDTA and 31. We thank members of the Dunphy lab for comments
The gradient was centrifuged at 40,000 rpm for 16 dialyzed against 10 mM Hepes-KOH (pH 7.5), 10 mM on the manuscript, P. R. Mueller for the Xenopus
hours in a Beckman SW55 rotor at 40C. The frac- NaCI, 1 mM DTT, and 1 mM EDTA for 2 hours at 40C. oocyte cDNA library, and D. S. Krapf and G. Hatha-
tions were collected and assayed for Xcdc25-specif- Portions were frozen in liquid N2 and stored at -800C. way for peptide sequencing. Supported in part by a
ic kinase activity (29). 24. W. J. Boyle, P. van der Geer, T. Hunter, Methods grant from NIH. W.G.D. is an investigator of the
14. Proteins were separated by electrophoresis and Enzymol. 201, 110 (1991). Howard Hughes Medical Institute.
blotted onto Problot (Applied Biosystems). p67 was 25. B. Fenton and D. M. Glover, Nature 363, 637 (1993);
excised and digested with sequencing-grade trypsin R. M. Golsteyn, K. E. Mundt, A. M. Fry, E. A. Nigg, J. 15 May 1996; accepted 15 July 1996
(Boehringer Mannheim) in the presence of hydroge-
nated Triton X-1 00 as described [J. Fernandez, L.
Andrews, S. M. Mische, Anal. Biochem. 218, 112
(1994)]. The tryptic peptides were separated by re-
verse phase chromatography in an ,uRPC C2/C1 8
A Model of Host-Microbial Interactions in an
2.1/10 column with the use of the SMART system
(Pharmacia). Peptide sequence analysis was per-
Open Mammalian Ecosystem
formed with an ABI 476A sequencer (Applied Bio-
systems) in the Protein/Peptide Micro Analytical Fa- Lynn Bry, Per G. Falk, Tore Midtvedt, Jeffrey 1. Gordon*
cility at the California Institute of Technology.
15. Three degenerate primers corresponding to the peptide The maintenance and significance of the complex populations of microbes present in
sequences KKXLX(T/G)TPNYIAPEVL, (A/S)GANTTP,
and XX(D/I)APSTIDO (30) were designed: p67-1, the mammalian intestine are poorly understood. Comparison of conventionally housed
AA(A/G)AA(A/G)AA(T/C)(T/C)TITG(T/C)(G/A)(G/C)- and germ-free NMRI mice revealed that production of fucosylated glycoconjugates
IACICC; p67-2, CCAIGGIGTIGT(A/G)TTIGCICC; and and an o1t,2-fucosyltransferase messenger RNA in the small-intestinal epithelium
p67-3, GCICCI(A/T)(G/C)IA(C/T)IAT(T/C/A)GA(T/C)-
CA. X denotes an unreadable amino acid residue. requires the normal microflora. Colonization of germ-free mice with Bacteroides the-
PCR was done with AmpliTaq DNA polymerase (Per- taiotaomicron, a component of this flora, restored the fucosylation program, whereas
kin-Elmer), the p67-1 and p67-2 primers, and Xeno- an isogenic strain carrying a transposon insertion that disrupts its ability to use
pus oocyte cDNA for 35 cycles at 94°C for 1 min,
45°C for 2 min, and 72°C for 3 min. The reaction L-fucose as a carbon source did not. Simplified models such as this should aid the
yielded a single 850-bp product. A PCR reaction with study of open microbial ecosystems.
primer p67-3 indicated that the third peptide se-
quence was also encoded in the 850-bp fragment.
16. A Xenopus oocyte cDNA library in the pAX-NMT
vector (2) was screened by colony hybridization with
the 850-bp PCR fragment that had been labeled in A complex and dynamic microbial ecosys- The factors that allow components of
the presence of o-32P-labeled deoxycytidine tem inhabits the human intestine. Establish- this indigenous microbiota to establish and
triphosphate by the random primer method. A Xho
l-to-Xho fragment from the cDNA containing the ment of a microflora begins at birth and maintain their regional habitats are largely
full coding sequence of Plxi was subcloned into progresses through a series of colonizations unknown. The stability of the ecosystem is
pBluescript SK (pBluescript-Plxl). Plasmids con- involving more than 400 bacterial species all the more remarkable given that the in-
taining nested deletions of the cDNA generated by
the Erase-a-base system (Promega Biotech) were (1). These colonizations result in a metabol- testinal epithelium is rapidly renewed
sequenced with an ABI 373A automated DNA se- ically active entity in which anaerobes pre- throughout life (5). Integration of the eco-
quencer (Applied Biosystems). dominate (2). A specific region of the intes- system into the host may be achieved, at
17. R. M. Golsteyn et a/., J. Cell Sci. 107, 1509 (1994); R.
Hamanaka et al., Cell Growth Differ. 5, 249 (1994). tine's duodenal-colonic axis will at any giv- least in part, through dynamic interactions
18. S. Llamazares et al., Genes Dev. 5, 2153 (1991). en time contain resident (autochthonous) that allow microbes to modify cellular dif-
19. H. Ohkura, I. M. Hagan, D. M. Glover, ibid. 9, 1059 bacterial species and a variable set of tran- ferentiation programs and thus create favor-
(1995). sient (allochthonous) species that tempo- able niches. One way of identifying such
20. K. Kitada, A. L. Johnson, L. H. Johnston, A. Sugino,
Mol. Cell. Biol. 13, 4445 (1993). rarily occupy a functionally "empty" niche programs is by comparing genetically iden-
21. An Nde site was created at the initiation codon of (3). Disruption of this open ecosystem-for tical mice that (i) have never been exposed
PDxI in a PCR reaction containing Pfu DNA polymer- example, with antibiotics-has revealed the to any microorganisms (germ-free, GF), (ii)
ase, pBluescript-Plxl (which contains an internal Nde
site), and the following oligonucleotides: p67-Nde-1, essential role played by the microbiota in have been raised with a "normal" function-
GGAATTCCATATGGCTCAAGTGGCCGG; and p67- preventing infectious diseases (4). al microbiota (conventional, CONV), and
1380 SCIENCE * VOL. 273 * 6 SEPTEMBER 1996
(iii) have been raised in a germ-free state are lost from the middle and upper portion increasing levels and extent of fucosylation
for a period of time and then exposed to an of crypts and their associated villi (Fig. 1, I (Fig. 1J). In contrast, adult GF mice express
intact microbiota or a selected group of its and K). The rate of clearance of fucose+ fucosylated glycoconjugates in their Paneth
component organisms (ex-germ-free, XGF) cells from P21 to P28 GF crypt-villus units cells only (Fig. 1K). The cecal and colonic
(6). Here we have used XGF mice to iden- equals the rate of cell migration as measured epithelium of CONV and GF mice have an
tify a single, genetically manipulatable res- by pulse labeling with 5-bromo-2'-deoxyuri- identical fucose+ phenotype.
ident bacterial species that reproduces an dine, indicating that loss of fucosylated gly- These findings indicate that initiation of
epithelial differentiation program normally coconjugates reflects cellular replacement fucosylated glycoconjugate production in the
specified at the completion of gut morpho- rather than suppression of production with- small intestine occurs independently of its
genesis by the indigenous flora. in an existing population. Adult CONV microbiota but that the microbiota is required
A panel of lectins (7) was used to define animals express fucosylated epitopes in their to complete the fucosylation program be-
cell lineage-specific, spatial, and temporal enterocytic, goblet, and Paneth cell lineages tween P21 and P28. The patterns of lectin
patterns of glycoconjugate accumulation in with a prominent duodenal-ileal gradient of binding in GF and CONV epithelium suggest
the intestinal epithelium of immunocompe-
tent inbred NMRI/Kl mice raised under GF
or CONV states (8-10). Glycoconjugates Fig. 1. The microbiota is required to sus- __B
were examined because their production tain production of o1 ,2-linked fucosylated
provides a sensitive marker of gut epithelial

Downloaded from https://www.science.org at Universitts- und Landesbibliothek Bonn on August 24, 2023
glycoconjugates in the ileal epithelium. (A
differentiation (7, 11) and can be modified and B) Wholemounts showing villi from the
by exogenous factors (12). Glycoconjugates distal third of the small intestine (ileum) ofJ
P17 GF and CONV animals, stained with 4-
are also used by bacteria as a source of
nutrients (13) and can serve as receptors for peroxidase-conjugated UEA1 (17). At this:
their adhesins (14). developmental stage, there are no differ-
ences in the patterns of fucosylation be-
A comparison of postnatal day 1 (P1) to tween CONV (A) and GF (B) animals. Fu- Jr
P90 GF and CONV mice (15) revealed that cose+goblet cells (brown, arrow) are more
sustained production of fucosylated intesti- numerous in the distal than in the proximal
nal glycoconjugates after P21 requires com- half of the small intestine. Villus enterocytes
ponents of the microbiota. Proliferation in (white) and crypt Paneth cells (not visible)
the intestinal epithelium is confined to the are fucose- throughout the duodenal-ileal
crypts of Lieberkuhn (5). Each villus is sup- axis. (C and D) By P21, isolated fucosel
plied by several crypts that surround its base. villi (detected with UEA1) are apparent in
Four principal lineages arise from a multipo- CONV and GF mice. In CONV animals (C),
fucosylated glycoconjugates are present in
tent crypt stem cell. Paneth cells differenti- crypts surrounding the base of the positive
ate as they move down to the base of the villus (arrow). This is not the case in GF _
crypt where they produce antimicrobial pep- mice (D), where fucosylation is being extin-
tides (16). Absorptive enterocytes, en- guished in the crypts. (E) Side view of an
teroendocrine cells, and mucus-producing ileal wholemount from a P23 CONV
goblet cells differentiate as they migrate up mouse, stained with peroxidase-UEA1.
from each crypt to the apex of an associated Acquisition of the fucose+ phenotype in-
villus where they undergo apoptosis and ex- volves a progressive upward migration of a F
foliation, a sequence that takes 3 to 5 days band of positive cells that encircles the vil-
lus. (F and G) Section from a fucose+
(5). At P21, lineage-specific changes in fu- patch similar to that shown in (E), incubat-
cosylation begin to occur in both GF and ed with tetramethylrhodamine isothiocya-
CONV mice (17) (Fig. 1, A to D). Paneth nate (TRITC)-UEA1 and digoxigenin-con-
cells acquire fucosylated glycoconjugates as jugated OPA (detected with FITC-conju-
do cells in the upper half of some crypts. gated sheep antibody to digoxigenin). (F)
These fucose+ crypts occur as multicrypt UEA1 staining alone and (G) double expo-
patches. Typically, only a small subset of sure showing colocalization of the two lec-
villi/patch have fucose+ enterocytes and tins, establishing that there is an induction
goblet cells (Fig. 1, C and D). The number of axl,2- and al 6-linked fucosylated gly-
and size of patches expand between P21 and coconjugates in villus enterocytes [closed
arrow in (F)] and goblet cells [open arrow in
P28 in CONV mice (Fig. 1, E to G). In GF (F)]. Arrowheads in (G) point to the leading
animals, Paneth cell staining expands to edges of columns of upwardly migrating
include all crypts, but fucosylated epitopes fucose+ epithelial cells in adjacent villi. (H j
and 1) Ileal wholemount from P25 CONV
L. Bry and J. I. Gordon, Department of Molecular Biology (H) and GF (I) mice. Without the microbiota,
and Pharmacology, Washington University School of fucosylation is extinguished. (J and K) Ileal
Medicine, St. Louis, MO 631 10, USA. wholemounts from P28 CONV and GF
P. G. Falk, Department of Molecular Biology and Phar- mice. The fucosylation program is fully ex-
macology, Washington University School of Medicine, St. pressed in CONV mice (J) where all of the
Louis, MO 631 10, USA, and Department of Medicine and
Laboratory of Medical Microbial Ecology, Karolinska In- patches of UEA1 + ileal crypt-villus units
stitute, S-171 77 Stockholm, Sweden. have coalesced. In GF animals (K), only
T. Midtvedt, Laboratory of Medical Microbial Ecology, Paneth cells located at the base of crypts
Karolinska Institute, S-1 7177 Stockholm, Sweden. are UEA1 +. Inset in (J) shows UEA1 staining of scattered duodenal villus epithelial cells in the same P28
*To whom correspondence should be addressed. CONV wholemount, emphasizing the marked duodenal-ileal gradient of fucosylation. All wholemounts
E-mail: jgordon@pharmdec.wustl.edu were photographed under a 6.3x objective. Bar in (F), 25 VLm.

SCIENCE * VOL. 273 * 6 SEPTEMBER 1996 1381

 tlhat the microhiota affects expression of host established that 48 hours after inoculation, tern was fully recpitullated Within 14 days
fucosyltransferases (FT) that add L-fucose in the density of viable organisms in the feces (Fig. 2, A to C). Reverse transcriptace-poly-
cx1,2 and (1,6 linkages (18) (Fig. IG). of XGF mice w!as eqUivalent to that of iden- merase chain reaction (RT-PCR) of total
XGF experiments (n = 3) demonstrated tically aged CONV mice. Fucosylation pro- cellular duoLdenal, ileal, and colonic RNA
that the fucosylation program CouIld he reini- gressed in the samne manner as observed in from GF and XGF animnalls (21 ) confirmed
tiated and com-pleted in the adult intestine P21 to P28 CONV mice wvhether GF mice that production of fuLcosvlated glCOCO(njuL-
by inoculation of GF mice with gut flora were conventionalized at P28 or P70; fu- gates is associated Withl IccTLmulation of a
from CONV animals (19). QLantitation of cose + patches first appeared 7 days after host (x1,2-FT mRNA (Fig. 2D).
anaerobe colony-forming uinits (CFUL) (20) conventionalization, and the CONV pat- To identify a comiAponent of the micro-
flora capable of inducing the CONV pat-
cxl 2-FT tern of fucosylation in XG7jF micc-, we inlOC-
ulalted P28 GF animals w ith Bacteroides the-
taiotaomicron, a member of the normcal
mouse and human small-intestinal and co-
lonic flora (22, 23). Ex ViVo studies have
shown that this anaerobe scavenges a vari-
ety of carbohydrates, including L-fucose,

Downloaded from https://www.science.org at Universitts- und Landesbibliothek Bonn on August 24, 2023
from more complex host and dietary glyco-
conjugates (13). Intestinal wholemounts
were prepared 2, 5, 7, 14, and 21 days after
inoculation. Within 2 days, the bacteria
were distributed along thc entire duodenai-
colonic axis of XCF mnicc (23). Bacteroides
thetaiotaomicron indLuced the CONV pattern
Fig. 2. The ileal epithelial fucosylation program can be reinitiated and completed in the adult intestine by of ileal fucosylation within 5 to 7 dclays.
inoculation of GF NMRI mice with a CONV microflora. (A to C) Sections of ileal crypt-villus junctions from Differences in CFLU per milliliter of ileal
P42 CONV and GF mice, and a P42 XGF mouse exposed to a CONV flora at P28. Sections were stained
with TRITC-UEA1 and a FITC-conjugated goat antibody to mouse immunoglobulin A (IgA). (A) CONV mice contents were detected among individual
express fucosylated glycoconjugates (red) in Paneth cells at the crypt base (open arrow) and in villus mice. We took advantage of this variability
enterocytes (closed arrow; note Golgi and brush border staining) and goblet cells (closed arrowhead). The to show that the extenit of fucosylation at 7
lamina propria contains numerous IgA+ cells (green). (B). GF mice express fucosylated glycoconjugates in to 21 days correlated with the density of
their Paneth cells only (open arrow). IgA- cells are rare. (C) XGF mice recapitulate the CONV pattern of colonizing organismns (Fig. 3). A minimum
fucosylation within 7 days after initial exposure to bacteria. (D) Representative RT-PCR analysis of duo- of 104 bacteria per milliliter was required to
denal (D), ileal (I), and colonic (C) RNA prepared from a P42 GF mouse and a P42 XGF animal inoculated elicit a "patchy" fucoseC_ phenotype; 107
7 days earlier with a CONV microflora. A 230-bp fragment derived from a mouse of ,2-FT mRNA (21) is bacteria per milliliter fully recapitulated the
present in XGF ileum but not duodenum. The band is absent in GF ileum but present in GF colon. The CONV pattern.
presence of this mRNA correlates with the distribution of fucosylated glycoconjugates in GF and XGF
animals. Actin mRNA levels were shown by RT-PCR to be equivalent in the RNA samples. Bar, 25 p.m. Colonization of the dIuodenum Was less
efficient than in the ileum (CFU = two
orders of magnitude lower). Nonetheless,
CFU/ml: 107 levels of bacteria sufficieIlt to induce fuco-
sylation in the ileum (10" to 10' CFU/ml)
were unable to do so in the duodenum,
demonstrating the organisms capacity to re-
produce the region-specific CONV pattern
WT of fucosylated glycoconjLugate production.
We examined whether the bacteria's abil-
ity to metabolize L-fucose correlates with its
capacity to induce epithelial fucosylation.
Fu-4 is an isogenic B. thetaiotaomicron strain
that contains an uncharacterized insertion of
Tn4351 rendering it unllable to use L-fucose
or D-arabinose as carbon sources ( 13).
Southern (DNA) blots established that this
transposon was inserted into a single site
Fu-4 within the genome ( 13). The insertion does
not block the organism's abilitv to take up
3H]L-fucose or to colonize and compete
with the isogenic wild-type strain (VPI
5482) in adult GF Balb/c cecum (13). Fu-4
colonizes the NMRI ileum and cecum at
Fig. 3. Wild-type B. thetaiotaomicron, but not an isogenic strain that lacks the ability to utilize L-fucose, levels equivalent to those of wild-type bac-
can signal ileal epithelial fucosylation. Ileal wholemounts, prepared 7 days after monocontamination of teria (<107 CFU per milliliter of ileal con-
P28 male GF mice with VPI-5482 (wild-type) or Fu-4 strains and stained with peroxidase-UEA1. Results tents) but is much less efficient at inducing
obtained from mice with 103 to 1 07CFU per milliliter of ileal contents illustrate the correlation between epithelial fucosylation: 10' CFU of Fu-4 pcr
ileal fucosylation and the density of colonizing organisms. A similar correlation was observed at 14 and milliliter produced the scattered pattern of
21 days (data not shown). fucose+ villi seen wvith only i04 CFU of the
1 382 SCIENCE * VOL. 273 * 6 SEPTEMBER 1996
 wild-type strain per milliliter (Fig. 3) (24). in the bacteria's metabolic properties. The simplicifolia type 11 agglutinin (GlcNAc(x1,3Gal/Glc).
Two results suggest that B. thetaiotaomi- integrity of the locus disrupted by the trans- Only the fucose-specific markers showed differcinc-
es in binding. These differences were limited tc the
cron affects fucosylation without directly poson appears to be necessary for the bac- small intestinal epithelium.
binding to the epithelium and that soluble teria's contribution to the cross-talk, al- 19. The intact cecum and colon from CONV P42 nice
mediators may be involved. Sections of Lin- though this hypothesis will have to be con- were homogenized in 10 ml of sterile 0.9% NaCI.
Samples (0.5 ml) were spread on the fur and inocu-
perfused CONV or XGF intestine were firmed by further genetic analyses, as will its lated into the mouth and rectum of P28 or P70 GF
treated with Gram or Warthin-Starry silver contribution to fucose utilization. The in- animals (n = 5 mice per time point per experiment).
stains. In CONV mice, or GF mice ren- duction of an (1,2-FT provides both a XGF mice were housed with CONV animals.
20. Three samples of luminal contents, recovered Trom
dered XGF with a CONV flora, compo- marker and a mediator of the fucosylation duodenum, jejunum, ileum, and cecum of each ani-
nents of the indigenous microflora were program. This bacterial locus and mamma- mal, were inoculated individually into thioglycollate
readily seen hound to ileal villus epithelial lian gene constitute the starting point for medium. Serial dilutions were made from this invcu-
cells. No binding was evident throughout further dissection of a dialogue symbolic of lum, samples plated onto blood agar, and CFU count-
ed after a 3-day incubation at 37°C under anaerobic
the duodenal-ileal axis at any time point the ongoing interaction between indige- conditions.
after inoculation of GF mice with the wild- nous organisms and their host. 21. Total cellular RNA was extracted with RNAzol B Tel-
type or mutant strains, even when their Test, Houston, TX). RT-PCR reactions were carried
out with 250 ng of RNA, and primers designed from
CFU per milliliter of ileal contents was REFERENCES AND NOTES a mouse o1,2-FT cDNA [P50 (5'-GCGMTATGC-
equivalent to that in XGF animals colo- CACGCTGTTTGC-3' and 5'-GCACGGGTATCCT-
nized with a CONV flora (n = 5 per group 1. D.C. Savage, in The Regulatory and Protective Role GTGAAGCGC-3')], under conditions suggested by

Downloaded from https://www.science.org at Universitts- und Landesbibliothek Bonn on August 24, 2023
of the Normal Microflora, R. Grubb, T. Midtvedt, K. E. the manufacturer of the GeneAmpThermostable rTth
per time point). Wild-type and Fu-4 strains Norin, Eds. (Macmillan, London, 1989), pp. 3-18. kit (Perkin-Elmer). Actin primers were used as inter-
were also labeled with fluorescein isothio- 2. P. L. Stark, and A. Lee, J. Med. Microbiol. 15, 189 nal controls to detect actin mRNA and to exclude
cyanate (FITC) after growth to mid-log (1982); K. E. Norin et al., Acta Ped. Scand. 74, 207 any genomic DNA contamination [5'-TGGAATC-
CTGTGGCATCCATGAAAC-3' and 5'-TAAAACG-
phase in thioglycollate medium or after (1985).
3. D. C. Savage, Annu. Rev. Microbiol. 31, 107 (1977). CAGCTCAGTMCAGTCCG-3'; actin mRNA, 320-
they were physically removed from the 4. D. van der Waaij, ibid. 43, 69 (1989). base pair (bp) product; actin gene, 470-bp product].
cecum of XGF mice (in vivo selection) 5. J. I. Gordon and M. L. Hermiston, Curr. Opin. Cell 22. K. Shindo and K. Fukushima, Gastroenterol. Jap 11,
and then applied to sections of duodenum, Biol. 6, 795 (1994). 167 (1976); T. Ushijima et al., Microbiol. Immunol.
6. T. Midtvedt, Microecol. Ther. 16, 121 (1986). 27, 985 (1983).
jejunum, ileum, and cecum. Neither 7. P. Falk, K. A. Roth, J. I. Gordon, Am. J. Physiol. 266, 23. P28 GF animals were inoculated with B. thetaiotaomi-
strain, harvested under these two condi- G987 (1994); L. Bry, P. G. Falk, J. I. Gordon, Proc. cron strain VPI-5482 (wild-type) or Fu-4 (isogenic with
tions, bound to CONV, GF, or XGF gut Natl. Acad. Sci. U.S.A. 93, 1161 (1996) These ref- VPI-5482) (13) and housed in GF isolators until they
erences describe the lectins and protocols used for were killed 2, 5, 7, 14, and 21 days later (four tc five
epithelium (25, 26). immunohistochemical surveys of the intestine. mice per time point per experiment; n = 3 indeDen-
Fucosylated glycoconjugates mediate at- 8. Germ-free NMRI/KI mice (9) were housed in gnoto- dent experiments). The gastrointestinal tract was re-
tachment of pathogens to epithelial surfaces biotic isolators (10) under a strict 12-hour light cycle moved en bloc under sterile conditions. Luminal con-
tents were recovered from a 1 -cm segment located in
(14) and also provide a source of nutrients and fed an autoclaved chow diet (Lactamin, Vad-
the mid-jejunum, a 1 -cm segment located 1 cm prox-
for members of the indigenous flora (13). stena, Sweden). Weekly fecal samples were imal to the ileal-cecal valve, and from a 0.5-cm seg-
checked for the presence of organisms by Gram
The ability of B. thetaiotaomicron to regulate stain and by incubation in thioglycollate medium un- ment of the proximal cecum. CFU were determined
(20). The identity of the recovered organisms was
fucosylation programs in the distal intestine der aerobic and anaerobic conditions. CONV mice
verified by Gram stain, by their failure to grow aerobi-
could affect the ability of other components were housed under the same conditions in a speci-
fied pathogen-free environment. GF animals were cally, and by their antibiotic resistance phenotype (VPI
of the normal flora to establish a stable removed from the isolators and crossed to CONV 5482 is KanRErmS; Fu-4 is KanRErmR) (13).
niche or could affect the vulnerability of mice every second generation. 24. Chi-square analysis was used to test the following
9. T. Granholm et al., Cytokine 4, 545 (1992). hypotheses: (i) animals with 1 04 to 106 CFU/mIl will
the intestine to colonization by pathogens. 10. B. E. Gustafsson, Ann. N. Y. Acad. Sci. 78,17 (1959). exhibit an intermediate fucosylation phenotype de-
Identifying molecular interactions be- 11. A. Varki, Glycobiology 3, 97 (1993).
fined as 5 to 75% of ileal villi being fucose I; and (ii)
tween indigenous microbes and host cells animals with -107 CFU/ml will produce the CONV
12. D. Lenoir et al., Biochim. Biophys. Acta 1234, 29 phenotype, defined as >95% of ileal villi being fu-
that confer stability to their ecosystems (1995); Y. Umesaki et al., Microbiol. Immunol. 39, cose+. x2 test value = 3.841 (( = 0.05; df = 1 Six
should help our appreciation of what goes 555 (1995); A. Pusztai et al., Glyconj. J. 12, 22 out of six mice with 104 to 106 CFU of wild-type
(1995); M. C. Biol, A. Martin, P. Louisot, Biochemie bacteria per milliliter of ileal contents had an interme-
wrong when pathogens gain control and 14, 13 (1994). diate phenotype (X2 = 0)-accept hypothesis (i);
may provide new strategies for preventing 13. A. A. Salyers and M. Pajeau, Appl. Environ. Micro- 0/13 mice with 1 04 to 1 06 CFU of Fu-4 per milliliter
or treating infections. However, members of biol. 55, 2572 (1989). had an intermediate phenotype (X2 = 13)-reject
14. S. J. Hultgren et al., Cell 73, 887 (1993). hypothesis (i). Six out of six mice with -107 CFU of
the microflora rarely produce the dramatic 15. GF and CONV mice were killed between 1600 and wild-type bacteria per milliliter exhibited a CONV
and distinctive phenotypes seen in models 1800hoursatPl,5,7, 10,14,17,21,23,25,28,42, phenotype (X2 = 0)-accept hypothesis (ii); 0/10
of pathogenesis. We have reduced the in- and 90 (n = 4 to 10 mice per time point per experi- mice with >107 CFU of Fu-4 per milliliter had a
testine's ecosystem to a simplified model in ment; n = 3 independent experiments). CONV phenotype (X2 = 1 0)-reject hypothesis (ii).
16. H. Cheng, Am. J. Anat. 141, 521 (1974); M. E. Sel- 25. Bacteria were labeled with FITC as described (26' with
which a single, genetically manipulatable, sted et al., J. Cell Biol. 118, 929 (1992); L. Bry et al., the following modification: cecal contents were
nonpathogenic member of the normal flora Proc. Natl. Acad. Sci. U.S.A. 91, 10335 (1994). washed five times in 0.2 M sodium carbonate buffer
induces a specific and easily identifiable 17. Surveys of wholemounts [M. L. Hermiston and J. I. (pH 9) and allowed to settle for 5 min between wash-
response. Features of this model include an Gordon, J. Cell Biol. 129, 489 (1995)] and sections es. The supernatant was taken for labeling. Metliods
indicated that from P1 to P17, the lineage-specific, for incubating Bouin's- or formalin-fixed or frozen sec-
apparent association between B. thetaio- crypt-villus, and duodenal-ileal patterns of Ulex eu- tions of intestinal Swiss rolls (17) with labeled bacteria
taomicron's capacity to utilize L-fucose and ropaeus type 1 agglutinin (UEA1), Anguilla anguilla for in situ binding assays are described in (26).
its ability to induce fucosylated glycoconju- agglutinin (AAA), Aleuria aurantia (orange peel) ag- 26. P. Falk, T. Borbn, D. Haslam, M. Caparon, Methods
glutinin (OPA), and H type 2 monoclonal antibody Cell Biol. 45, 165 (1994).
gate production in its host and the fact that (mAb 92FRA2, DAKO) binding were indistinguish- 27. We thank J. Lowe (University of Michigan) tor a
a wild-type fucose-utilizing strain must able in GF and CONV small intestine. mouse (x 1,2-FT cDNA (P50), A. Salyers (Univers!ty of
achieve a critical density in the distal intes- 18. Sections prepared along the duodenal-colonic axis Illinois, Urbana) for B. thetaiotaomicron strains, and
of P28 to P90 CONV and GF NMRI/KI gut were J. Bergstedt and A.-K. Persson for technical assist-
tine to trigger epithelial fucosylation. The stained with UEAl (specificity: Fucul,2Gal)3), OPA ance. Supported by grants from the National Insti-
density dependence may reflect either a [Fucoxl,6/3N-acetylglucosamine (GIcNAc)], AAA (oi- tutes of Health (DK30292 and 37960), Swedish Can-
microbial signal that must reach a critical L-fucose), the H2 mAb (Fucax1,2GalP1,4GIcNAc), cer Society (3523-B95-02XBB), and Medical Re-
Arachis hypogae agglutinin (Gal,B1,3GaINAc), Doli- search Council (B96-16X-1 1595-01).
threshold before an epithelial response can chos biflorus agglutinin (GalNAc(xl,3Gal), Maakia
be induced or a density-dependent change amurensis agglutinin (NeuAco2,3Gal), and Griffonia 29 April 1996; accepted 1 July 1996

SCIENCE * VOL. 273 * 6 SEPTEMBER 1996 1 383

 A Model of Host-Microbial Interactions in an Open Mammalian Ecosystem
Lynn Bry, Per G. Falk, Tore Midtvedt, and Jeffrey I. Gordon

Science,Science, 273 (5280), .

DOI: 10.1126/science.273.5280.1380

Downloaded from https://www.science.org at Universitts- und Landesbibliothek Bonn on August 24, 2023
View the article online
https://www.science.org/doi/10.1126/science.273.5280.1380
Permissions
https://www.science.org/help/reprints-and-permissions

Use of this article is subject to the Terms of service

Science (ISSN 1095-9203) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW,
Washington, DC 20005. The title Science is a registered trademark of AAAS.
© 1996 American Association for the Advancement of Science