CLINICAL AND DIAGNOSTIC LABORATORY IMMUNOLOGY,1071-412X/01/$04.0010 DOI: 10.1128/CDLI.8.2.320–324.2001

Mar. 2001, p. 320–324 Vol. 8, No. 2

Copyright © 2001, American Society for Microbiology. All Rights Reserved.

Association of Lactobacillus spp. with Peyer’s Patches in MiceLAURA PLANT AND PATRICIA CONWAY*

School of Microbiology and Immunology, University of New South Wales, Sydney, Australia, 2052

Received 28 February 2000/Returned for modification 17 October 2000/Accepted 21 December 2000

Sixteen strains of Lactobacillus isolated from humans, mice, and food products were screened for theircapacity to associate with Peyer’s patches in mice. In preliminary experiments, in vitro binding to tissue pieceswas assessed by scanning electron microscopy, and it was demonstrated qualitatively that 5 of the 16 strainsshowed some affinity for the Peyer’s patches, irrespective of their association with the nonlymphoid intestinaltissue. Lactobacillus fermentum KLD was selected for further study, since, in addition to its intrinsically highadhesion rate, this organism was found to exhibit a preferential binding to the follicle-associated epitheliumof the Peyer’s patches compared with its level of binding to the mucus-secreting regions of the small intestine.Quantitative assessment of scanning electron micrographs of tissue sections which had been incubated with L.fermentum KLD or a nonbinding control strain, Lactobacillus delbruckii subsp. bulgaricus, supported theseobservations, since a marked difference in adhesion was noted (P < 0.05). This preferential association ofstrain KLD with the Peyer’s patches was also confirmed with radiolabeled lactobacilli incubated with intestinaltissue in the in vitro adhesion assay. Direct recovery of L. fermentum KLD from washed tissue following oraldosing of mice revealed a distinct association (P < 0.05) between this organism and the Peyer’s patch tissue.In contrast, L. delbruckii subsp. bulgaricus showed negligible binding to both tissue types in both in vitro andin vivo adhesion assays. It was concluded that L. fermentum KLD bound preferentially to Peyer’s patches ofBALB/c mice.

Lactobacilli comprise a large percentage of the indigenousflora of the gastrointestinal tract. It has been well documentedthat lactobacilli exert various beneficial effects on the host,which has led to their classification as a probiotic organism.Although specific health-related claims are generally notmade, probiotic bacteria have been shown to possess the abilityto inhibit various intestinal pathogens (5, 7), provide a barriereffect, and also modulate the immune function of the host (6,10, 14), as well as to have a variety of other effects.

Luminal antigens gain access to the mucosal lymphoid tis-sues via the Peyer’s patches in the small intestine. The deliveryof vaccines directly to this site could enhance the probabilitythat the host will encounter the immunizing antigen. Althoughthe currently used vaccines are effective, they make use ofattenuated pathogenic bacteria such as mycobacteria, salmo-nellae, and clostridia, many of which have been shown toassociate with the Peyer’s patches. Lactic acid bacteria areorganisms that are generally regarded as safe and are beingevaluated for use as a live-vector antigen delivery system (12).

It has been shown that lactobacilli associate with the gastro-intestinal tract in a number of ways. Both specific protein-aceous (4, 13) and carbohydrate (8) structures are involved inthe adhesion of lactobacilli to various regions within the gas-trointestinal tract. The interaction of gram-negative patho-genic bacteria with M cells has been extensively studied (3, 9,11), and more recently, the interaction of gram-positive bac-teria with the surface of M cells has been examined (2). Thestudy described here aims to determine whether lactobacilliassociate with the Peyer’s patches, in preference to nonlym-

phoid intestinal tissue, and to examine this adhesion both invitro and in vivo.

MATERIALS AND METHODS

Bacterial strains and growth conditions. The Lactobacillus strains used in thestudy (Table 1) were obtained from the Cooperative Research Centre (CRC) forFood Industry Innovation Culture Collection at the University of New SouthWales and were maintained as glycerol stocks stored at 270°C. Primary culturesfor each experiment were grown from the glycerol stocks by inoculation (1%)into Mann Rogosa Sharpe (MRS) broth (Difco) and incubation for 18 h anaer-obically (Don Whitley Scientific Mark 3 Anaerobic Chamber). The bacterialcultures were centrifuged at 3,000 3 g for 10 min, and the pellets were washedtwice in 0.1 M phosphate-buffered saline (PBS; pH 7.2). The pellets were ad-justed to give an optical density at l600 of 0.5 for use in the in vitro adhesionassay. Bacteria used in the in vivo adhesion assay were concentrated by resus-pension in a smaller volume of PBS to yield an optical density of 1.2, whichcorresponded to approximately 109 CFU ml21, as determined by serial dilutionand plating of the suspension on MRS agar. For the radiolabeling of the bacteriaused in the in vitro adhesion assay, the medium was supplemented with [methy-l,19,29-3H]thymidine (124 Ci mmol21) to give a final concentration of 10 mCiml21.

Preparation of tissue pieces. Peyer’s patch and nonlymphoid intestinal tissuesamples were taken from 6-week-old specific-pathogen-free female BALB/c micewhich had been obtained from CULAS, Little Bay, Australia. The tissue pieceswere washed so that they were visibly clear of debris and were placed into wellsof a 24-well tissue culture plate (Nunc) with the villi facing upwards and with sixpieces per well. The tissue pieces obtained were roughly 2 mm2. The tissue pieceswere kept on ice for no more than 2 h prior to use.

In vitro adhesion. The in vitro adhesion assay was conducted as described byHenriksson and Conway (8). Briefly, the radioactively labeled Lactobacillus cellswere incubated with the tissue pieces (n 5 6 per Lactobacillus suspension) at37°C for 20 min with constant gentle agitation with an orbital shaker. After theincubation, the tissue pieces were washed three times in 1 ml of PBS with gentleagitation at room temperature for 5 min per wash. The tissue pieces wereweighed and then digested with perchloric acid (150 ml) and hydrogen peroxide(300 ml) at 70°C for 12 h in glass scintillation vials. Scintillation fluid (9 ml) wasadded to each vial, and the radioactivity of the samples was measured after 10min with a Packard Tricarb 2100TR liquid scintillation counter. Statistical cal-culations were carried out by the Student t test.

* Corresponding author. Mailing address: School of Microbiologyand Immunology, University of New South Wales, UNSW, Sydney,Australia, 2052. Phone: 61 2 9385 1593. Fax: 61 2 9385 1591. E-mail: P.Conway@unsw.edu.au.

320

on January 5, 2020 by guesthttp://cvi.asm

.org/D

ownloaded from

Scanning electron microscopy. The adhesion assay described above was alsoconducted with nonradiolabeled lactobacilli. Tissue pieces were prepared forexamination by scanning electron microscopy instead of being weighed anddigested. Briefly, following the adhesion assay tissue pieces were fixed in glutar-aldehyde (3% in PBS) and were dehydrated with a graded ethanol series and100% dry acetone. The tissue pieces were dried with the E3100 Jumbo Series IICritical Point Drier apparatus (Polaron, Watford, United Kingdom). The tissuepieces were gold coated with a Polaron sputter coater, according to the manu-facturer’s instructions. The mounted sections were examined with a scanningelectron microscope (S360; Cambridge Instrument Co., Cambridge, UnitedKingdom). Fifty randomly selected fields from more than six tissue pieces fromat least six individual mice were examined at 33,000 magnification.

In vivo adhesion. Specific-pathogen-free female BALB/c mice, as describedabove, were orally dosed with approximately 109 lactobacilli by orogastric intu-bation. Each group contained six mice. At 2 h postdosing, the mice were killedby CO2 asphyxiation and the Peyer’s patches and control nonlymphoid intestinaltissue were examined by enumeration of the associated lactobacilli. This wasperformed by homogenizing the tissue with an Ultra Turrax homogenizer (Jankeand Kunkel) and serially diluting and plating aliquots on Rogosa agar (Oxoid).The numbers of lactobacilli were enumerated by counting according to knowncolony morphologies. The results were analyzed by the Mann-Whitney rank sumtest. Isolates were confirmed by protein profiling (n 5 5), carbohydrate fermen-tation with an API 30 (BioMerieux) (n 5 4), and immunodetection (n 5 4) withLactobacillus fermentum KLD-specific antiserum to positively identify L. fermen-tum isolates from indigenous organisms.

RESULTS

Examination of the in vitro association of Lactobacillus spp.with the FAE of the Peyer’s patches by scanning electronmicroscopy. Initial screening of the strains of Lactobacilluswhich associate with the follicle-associated epithelium (FAE)of the Peyer’s patches was performed by adhesion assays. Lowlevels of association were seen for most of the 16 differentstrains examined, with negligible Lactobacillus cells detected inmost fields when the samples were examined by scanning elec-tron microscopy (Table 2). It can been seen that L. fermentumKLD associated with the FAE in large numbers, while, incontrast, small numbers were observed on the nonlymphoidvillous intestinal tissue pieces (Fig. 1). Of the five other strainsof Lactobacillus seen to associate with the Peyer’s patch tissue,a strong association with the nonlymphoid villous intestinaltissue was also evident. No correlation between the source ofthe lactobacilli and the pattern of Peyer’s patch assocation wasevident, with some strains isolated from all sources having

affinity for the FAE. L. fermentum KLD and Lactobacillusdelbruckii subsp. bulgaricus were selected from the initialscreen for further quantification by scanning electron micros-copy and using radiolabeled bacterial cells in the adhesionassay. L. delbruckii subsp. bulgaricus was chosen as a negativecontrol strain because it associated with neither the Peyer’spatches nor the nonlymphoid villous intestinal tissue.

It was shown that L. fermentum KLD bound in large num-bers to the FAE domes within the Peyer’s patches, with largenumbers of bacteria at the surface (Fig. 1a). The level ofassociation with the nonlymphoid villous intestine was usuallynegligible, irrespective of the presence of mucus on the tissuesurface seen in Fig. 1b. By scanning electron microscopy, noassociation of L. delbruckii subsp. bulgaricus with either thePeyer’s patch tissue or the nonlymphoid villous intestine wasdemonstrated (data not shown). When this association wasquantified by culturing the viable lactobacilli associated withtissue pieces, a similar pattern was noted for L. delbruckiisubsp. bulgaricus (Fig. 2). As can be seen in Fig. 2, L. fermen-tum KLD did associate to some extent with the villous intes-tine, but to a lesser degree than was noted for the Peyer’s patchtissue.

The differences in association of the two Lactobacillusstrains examined in 50 randomly selected fields of view with thescanning electron microscope can be seen in Fig. 3. The asso-ciation of L. fermentum KLD with the FAE of the Peyer’spatches was significantly different from the association ofL. delbruckii subsp. bulgaricus with the same regions (P ,0.05), according to the Mann-Whitney rank sum test. No as-sociation with the nonlymphoid villous intestinal tissue wasdemonstrable for the two strains.

Examination of the in vivo association of Lactobacillus spp.with the Peyer’s patches. In order to determine whether theassociation of L. fermentum KLD with the Peyer’s patches ofmice in vitro was demonstrable in vivo, viable lactobacilli wereisolated from small intestinal tissue following oral dosing (Fig.4). L. fermentum KLD was recovered from both the Peyer’s

TABLE 1. Strains of Lactobacillus screened for association withmouse Peyer’s patch tissue

Strain Reference number Source

Lactobacillus sp. strain 003 FII 532700 Mouse stomachLactobacillus sp. strain 004 FII 532800 Mouse stomachLactobacillus sp. strain 005 FII 532900 Mouse stomachLactobacillus sp. strain 006 FII 533000 Mouse colonLactobacillus sp. strain 008 FII 533200 Mouse colonL. fermentum PC1 FII 511400 Human fecesL. acidophilus FII 504400 UnknownL. fermentum LMN FII 511100 Human fecesL. fermentum 104S FII 511200 Pig stomachLactobacillus sp. strain HBL FII 511300 Human fecesL. paracasei 43338 FII 530300 Human fecesL. salivarius 43321 FII 530400 Human fecesL. paracasei 42319 FII 530500 Human fecesLactobacillus sp. strain 433121 FII 530600 Human fecesL. bulgaricus UNSW 046900 Dairy productsL. salivarius subsp. salivarius ATCC 11741 Unknown

TABLE 2. Association of Lactobacillus strains with mouse Peyer’spatches and nonlymphoid intestinal tissue

Strain Peyer’s patchassociationa

Nonlymphoidintestine associationa

Lactobacillus sp. strain 003 2 2Lactobacillus sp. strain 004 2 2Lactobacillus sp. strain 005 11 1Lactobacillus sp. strain 006 11 1Lactobacillus sp. strain 008 2 1L. fermentum KLD 11 2L. acidophilus 1 2L. fermentum LMN 2 2L. fermentum 104S 1 2Lactobacillus sp. strain HBL 1 11L. paracasei 43338 11 11L. salivarius 43321 2 2L. paracasei 42319 1 1Lactobacillus sp. strain 433121 2 2L. bulgaricus 2 2L. salivarius subsp. salivarius 1 2

a The association of Lactobacillus spp. to small intestinal tissue, as visualizedby scanning electron microscopy: 2, no association; 1, low level at association (5to 50 bacteria/field); 11, high level of association (.50 bacteria/field).

VOL. 8, 2001 ASSOCIATION OF LACTOBACILLUS SPP. WITH PEYER’S PATCHES 321

on January 5, 2020 by guesthttp://cvi.asm

.org/D

ownloaded from

patches and the nonlymphoid villous intestinal regions in num-bers significantly greater than those of L. delbruckii subsp.bulgaricus (P , 0.05), according to the Mann-Whitney ranksum test. As with the in vitro result, there were significantdifferences between the association of L. fermentum KLD withthe Peyer’s patches and the association of L. fermentum KLDwith the villous intestine (P , 0.05). The association of L. del-bruckii subsp. bulgaricus with both tissue types showed no sig-nificant difference, as there were generally no associated viablecells recovered from the tissue homogenates.

DISCUSSION

Most of the current knowledge of the adhesion to Peyer’spatches is based on the results of studies with gram-negativeorganisms and pathogens (3, 9, 11), and consequently, thisinteraction is comparatively well understood. The study de-scribed here sought to identify a strain of Lactobacillus thatassociated directly with the surfaces of the Peyer’s patches. Theimportance of this phenomenon is reflected in the need forimproved vectors for foreign antigens in vaccines against en-teric diseases, such that mucosal protection can be provided, as

well as the need for the exclusion of pathogens such as Esch-erichia coli from one of the important areas of invasion (1).

The selection of L. fermentum KLD was based on its highdegree of association with the FAE of the Peyer’s patch and itspoor association with the nonlymphoid villous intestine (Table2), since this would allow the ingested Lactobacillus cells totarget the Peyer’s patches and not be spread over the entirevillous surface. Five other strains also associated with the Pey-er’s patches but also showed a high degree of association withthe nonlymphoid villous intestine (Table 2). L. fermentum 104Swas more adhesive to the Peyer’s patches than the nonlym-phoid intestinal tissue, which reflects results of previous studieswhich have shown that this strain adheres better to nonsecret-ing squamous tissue than to secreting gastric epithelium (8).The Peyer’s patches are generally considered to be nonsecret-ing regions of the gastrointestinal epithelium due to the de-creased numbers of goblet cells. Hence, the mechanism ofadhesion to the squamous nonsecreting regions of the gastro-intestinal tract and to the intestinal mucosa could be importantwhen considering adhesion to the Peyer’s patches, as theycontain columnar cells but no overlying mucus.

The association of L. fermentum KLD with the surface of thePeyer’s patches has been shown by an in vitro adhesion assay,as well as by in vivo recovery from the Peyer’s patches follow-ing orogastric dosing of the strain. Although in most fields ofview 30 to 100% of the field was covered by L. fermentum KLD(Fig. 1), this association does not necessarily imply a directassociation with the M cells within these regions. Although thein vitro scanning electron microscope analysis reveals that L. fer-mentum KLD shows a preference for the FAE of the Peyer’spatch tissue (Fig. 3), low levels of association were observed insome domes of the Peyer’s patches, with some fields showingsmall numbers of associated bacteria or noncharacteristic ad-hesion patterns (data not shown). This variation in binding hasbeen reported for other organisms. Salmonella enterica serovarTyphimurium SL 1344 shows variation in binding to M cells (3)and differential binding to domes. The nonuniform bacterialadhesion to the domes suggests that there are M-cell subtypespresent in the domes of the Peyer’s patches.

The large degree of association of L. fermentum KLD withthe Peyer’s patches was particularly evident compared withthat for a nonassociating strain of L. delbruckii subsp. bulgari-cus. This strain showed levels of association which were signif-

FIG. 1. Scanning electron micrograph of mouse tissue after incubation in L. fermentum KLD cell suspension. (a) Typical aggregates of L.fermentum KLD-like cells on the FAE of the Peyer’s patch; (b) absence of cells on nonlymphoid villous intestine. Bars, 10 mm.

FIG. 2. Recovery of viable Lactobacillus cells following a 20-minexposure of L. fermentum KLD and L. delbruckii subsp. bulgaricus withPeyer’s patch and nonlymphoid intestinal tissue. More than 10 tissuepieces from at least six animals were examined. Results are expressedas the the numbers of CFU per milligram (wet weight) of tissue.

322 PLANT AND CONWAY CLIN. DIAGN. LAB. IMMUNOL.

on January 5, 2020 by guesthttp://cvi.asm

.org/D

ownloaded from

icantly lower than the level of association of L. fermentum KLDwith the small intestinal tissue used in this study (Fig. 2).Bacterial cells were rarely seen in scanning electron micro-graphs of tissue incubated with L. delbruckii subsp. bulgaricus(Fig. 3). The larger numbers of organisms associated with thetissue following direct recovery from the tissue may be due toa less rigorous washing procedure compared to that used forscanning electron microscope analysis. L. fermentum KLD washighly autoaggregative in most of the fields examined (Fig. 2).This suggests not only that there is an association of the lac-tobacilli with the Peyer’s patch surface but also that there areinteractions between the bacterial cells. This clumping may bebeneficial, as it could further enhance binding to the Peyer’spatches and may account for the large numbers of L. fermen-tum KLD associated with the Peyer’s patch surface.

Following administration of an orogastric dose either L.

fermentum KLD or L. delbruckii subsp. bulgaricus to mice, itwas observed that L. fermentum KLD associated with tissue insignificantly larger numbers than L. delbruckii subsp. bulgaricuswhen measured by direct enumeration of viable cells from thetissue surface (Fig. 4). In this study, the tissue was sampled 2 hafter orogastric intubation. It has previously been shown thatthe transit time through the entire gastrointestinal tract ofmice is approximately 3.5 h (Xin Wang, personal communica-tion). It is therefore assumed that 2 h after incubation, theLactobacillus cells would have reached the terminal ileum. Thesame trend was observed for the association with the nonlym-phoid villous intestine. As observed in the in vitro adhesionassay, large numbers of L. fermentum KLD were recoveredfrom the nonlymphoid villous intestine sections (Fig. 2). How-ever, the association with the Peyer’s patches was statisticallymore significant (P , 0.05) than the association with the non-lymphoid villous intestine, suggesting that L. fermentum KLDdoes preferentially associate with the Peyer’s patches in vivo.

The results of this investigation indicate that L. fermentumKLD is able to associate directly with Peyer’s patches in mice,both in vitro and in vivo. The tissue association of L. fermentumKLD was determined by comparison with that of a nonassoci-ating strain of Lactobacillus. The association is further sup-ported by the persistence of this organism within the gastroin-testinal tract for greater than 10 h and its high survival rate inthis system (L. Plant, unpublished observation). This study hasprovided novel evidence that lactobacilli associate with thePeyer’s patches in the murine intestine.

ACKNOWLEDGMENT

We acknowledge financial support from the CRC for Food IndustryInnovation.

REFERENCES

1. Berg, R. D., and A. W. Garlington. 1979. Translocation of certain indigenousbacteria from the gastrointestinal tract to the mesenteric lymph nodes andother organs in a gnotobiotic mouse model. Infect. Immun. 23:403–411.

2. Borghesi, C., M. Regoli, E. Bertelli, and C. Nicoletti. 1996. Modifications ofthe follicle-associated epithelium by short-term exposure to a non-intestinalbacterium. J. Pathol. 180:326–332.

3. Clark, M. A. M. A. Jepson, N. L. Simmons, and B. H. Hirst. 1994. Prefer-

FIG. 3. Association of bacterial cells with Peyer’s patches (a) and nonlymphoid intestinal tissue (b) following a 20-min exposure to L. fermentumKLD and L. delbruckii subsp. bulgaricus and analysis by scanning electron microscopy. Results of the extent of bacterial coverage of the tissuesurface when viewed by scanning electron microscopy at 33,000 magnification are represented as a percentage. Fifty fields with tissue from sixanimals were examined per treatment.

FIG. 4. Association of L. fermentum and L. delbruckii subsp. bul-garicus with the Peyer’s patches and nonlymphoid villous intestine 2 hfollowing administration of an oral dose of 109 viable Lactobacilluscells. Results are expressed as numbers of CFU per milligram (wetweight) of tissue (n 5 35 per treatment) as the mean value. Error barsindicate standard deviations.

VOL. 8, 2001 ASSOCIATION OF LACTOBACILLUS SPP. WITH PEYER’S PATCHES 323

on January 5, 2020 by guesthttp://cvi.asm

.org/D

ownloaded from

ential interaction of Salmonella typhimurium with mouse Peyer’s patch Mcells. Res. Microbiol. 145:543–552.

4. Conway, P. L., and S. Kjelleberg. 1989. Protein-mediated adhesion of Lac-tobacillus fermentum strain 737 to mouse stomach squamous epithelium.J. Gen. Microbiol. 135:1175–1186.

5. De Petrino, S. F., M. B. B. De Jorrat, O. Meson, and G. Perdigon. 1995.Protective ability of certain lactic acid bacteria against an infection withCandida albicans in a mouse immunosuppression model by corticoid. FoodAgric. Immunol. 7:365–373.

6. De Simone, C., R. Vesely, B. Bianchi Salvadori, and E. Jirillo. 1993. The roleof probiotics in modulation of the immune system in man and animals. Int.J. Immunother. 9:23–28.

7. Drago, L., M. R. Gismondo, A. Lombardi, C. De Haen, and L. Gozzini. 1997.Inhibition of in vitro growth of enteropathogens by new Lactobacillus iso-lates of human intestinal origin. FEMS Microbiol. Lett. 153:455–463.

8. Henriksson, A., and P. L. Conway. 1996. Adhesion of Lactobacillus fermen-tum 104S to porcine stomach mucus. Curr. Microbiol. 32:1–4.

9. Inman, L. R., and J. R. Cantey. 1983. Specific adherence of Escherichia coli(strain RDEC-1) to membranous (M) cells of the Peyer’s patches in Esche-richia coli induced diarrhea in rabbits. J. Clin. Investig. 71:1–8.

10. Link, H., F. Rochat, K. Y. Soudan, and E. Schiffrin. 1995. Immunomodula-tion of the gnotobiotic mouse through colonisation with lactic acid bacteria.In J. E. A. Mestecky (ed.), Advances in mucosal immunology. Plenum Press,New York, N.Y.

11. Owen, R. L., N. F. Pierce, and R. T. Apple. 1986. M cell transport of Vibriocholerae from the intestinal lumen into Peyer’s patches: a mechanism forantigen sampling and for microbial transepithelial migration. J. Infect. Dis.153:1108–1118.

12. Pouwels, P. H., R. J. Leer, M. Shaw, M.-J. H. den Bak-Glashower, F. D.Tielen, E. Smit, B. Martinez, J. Jore, and P. L. Conway. 1998. Lactic acidbacteria as antigen delivery vehicles for oral immunisation purposes. Int. J.Food Microbiol. 41:155–167.

13. Rojas, M. 1996. Studies on an adhesion promoting protein from Lactobacil-lus and its role in the colonization of the gastrointestinal tract. Thesis.Department of General and Marine Microbiology, Goteborg University,Goteborg, Sweden.

14. Schiffrin, E. J., F. Rochat, H. Link-Amster, J. M. Aeschlimann, and A.Donnet-Hughes. 1995. Immunomodulation of human blood cells followingthe ingestion of lactic acid bacteria. J. Dairy Sci. 78:491–497.

324 PLANT AND CONWAY CLIN. DIAGN. LAB. IMMUNOL.

on January 5, 2020 by guesthttp://cvi.asm

.org/D

ownloaded from