Adhesion, autoaggregation and hydrophobicity of 13 strains of Bifidobacterium longum


B. Del Re, Dipartimento di Biologia Evoluzionistica Sperimentale, via Selmi 3, 40126 Bologna, Italy (e-mail:


To identify bacterial traits related to adhesion ability in human bifidobacteria, 13 strains of Bifidobacterium longum isolated from human gastric juice and intestine were studied. Strains were tested for their capability to adhere to Caco-2 cells and classified as adhesive (Adh+) or non-adhesive (Adh–). Adh+ and Adh– strains were then investigated for their autoaggregation ability and surface hydrophobicity. Comparing the properties of Adh+ and Adh–, we observed that strains were able to adhere to cell monolayers if they autoaggregate and manifest a good degree of hydrophobicity as determined by microbial adhesion to hydrocarbons. These two traits could be used for preliminary screening to identify potentially adherent isolates.


Bifidobacteria are one of the most significant bacterial groups living in the large intestine (Scardovi 1986; Sgorbati et al. 1995). This bacterial group is thought to exert a range of biological activities related to host health. For example, bifidobacteria can significantly inhibit the adhesion of a variety of enteropathogenic bacteria (Bernet et al. 1993; Gibson and Wang 1994; Fujiwara et al. 1999), reduce serum cholesterol levels (Tahri et al. 1995) and stimulate immune system reactivity (Schiffrin et al. 1997). It is known that the abilities of bacteria to adhere to the intestinal epithelium play an important role in colonization of the gastrointestinal tract, preventing bacterial elimination by peristalsis and providing a competitive advantage in the ecosystem (Johansson et al. 1993; Crociani et al. 1995; Alander et al. 1997).

Bifidobacteria are widely used as food additives and adhesion to the intestinal mucosa is a property employed in their selection for use in commercial preparations.

Enterocyte-like Caco-2 cells (Fogh et al. 1977) have been successfully used to evaluate the adhesion ability of bifidobacteria (Bernet et al. 1993; Crociani et al. 1995). However, in vitro adhesion testing is expensive and time consuming, therefore reliable in vitro methods for the preliminary selection of potentially adherent strains are required.

A relationship between autoaggregation and adhesion ability in Bifidobacterium bifidum (Perez et al. 1998) and B. suis (Del Re et al. 1998) has been reported and a correlation between hydrophobicity and adhesion ability has been observed in some lactobacilli (Wadstrom et al. 1987).

The aim of the present study was to determine whether bacterial traits such as hydrophobicity and autoaggregation ability could be ‘predictive’ of adhesiveness of human bifidobacteria.

Materials and methods

Bacterial strains and culture conditions

All isolates studied were B. longum. Strains B1, B2, B6, B7, B8, B9 and B10 were freshly isolated from human gastric juices, while all of the others were obtained from the culture collection of the Agrarian Microbiology Institute (University of Bologna, Italy). Strains isolated from human gastric juice were from subjects who did not take fermented milk for 1 month before isolation of the organisms.

The bacteria were cultured in TPY broth in anaerobic jars at 37 °C (Anaerocult, Merck, Darmstadt, Germany). Before the adhesion test, bacteria were counted using a haemocytometer.

Cell culture

The following cell lines were tested: Caco-2 (human colon adenocarcinoma), HT29 (human colon adenocarcinoma) and KATO III (human gastric carcinoma), which were kindly provided by P. Nanni et al. (Istituto di Cancerologia, University of Bologna). Caco-2 and HT29 cells were routinely grown in Dulbecco-modified minimal essential medium (4500 mg l−1 glucose) (DMEM; Gibco BRL, Paisley, Scotland) supplemented with 20% inactivated (30 min, 56 °C) fetal calf serum (FCS) (Gibco). KATO III cells were grown in RPMI 1640 medium (Gibco) supplemented with 20% FCS. Cells were cultured in 25-cm2 flasks in an incubator with 10% CO2 at 37 °C. For adhesion assays, monolayers of Caco-2 and HT29 were prepared in NUNC tissue culture dishes (9 cm2) (NUNC, Denmark) with 104 cells cm−2; the cultures were used after 15 d growth. KATO III cells grew both as a cellular suspension and as an incomplete monolayer; for the assays they were seeded at a concentration of 2 × 104 cells cm−2 and tested after 7 d growth.

Adhesion assay

The cell monolayers were washed twice with phosphate-buffered saline (PBS) before the adhesion test. Bacterial strains were grown for 72 h at 37 °C in tryptone peptone yeast (TPY) agar plates, colonies were collected in TPY broth, centrifuged and resuspended in DMEM at a concentration of 108 bacteria ml−1. Aliquots of this suspension (2·5 ml) were added to the tissue culture dishes and incubated for 2 h at 37 °C. After incubation, monolayers were washed five times with PBS, fixed with methanol, Gram stained and counted, using 20 randomized microscopic fields per dish.

Adhesion was measured as the number of bacterial cells adhering to monolayers. Bacterial strains were scored as non adhesive when fewer than five bacteria adhered to 100 cells, adhesive with six to 40 bacteria adhering to 100 cells and strongly adhesive with more than 40 bacteria adhering to 100 cells.

Aggregation assay

Bacteria were grown at 37 °C for 24 h in TPY broth. The cells were harvested by centrifugation and suspended in PBS to 0·5 optical density (O.D.) units at 600 nm. Two ml bacterial suspension were placed in each tube and centrifuged. The cells were then resuspended in their culture supernatant fluids. After incubation at 37 °C for 2 h, 1 ml of the upper suspension was transferred to another tube and the O.D. measured. Aggregation was expressed as 1 − (O.D. upper suspension/O.D. total bacterial suspension) × 100.

Measurement of bacterial hydrophobicity

The salting out of bacterial cells was performed as described by Lindahl et al. (1981). Briefly, bacteria were agglutinated on a slide with various concentrations of ammonium sulphate (0·01–4·00 mol l−1). The lowest concentration causing bacterial agglutination was the salting-out aggregation test (SAT) value. Cell surface hydrophobicity is inversely correlated with the SAT value.

The bacterial surface charge was determined by ion exchange using chromatography on diethylaminoethyl-cellulose columns (Amersham Pharmacia Biotech, UK). Bacteria (1010 ml−1) were suspended in sterile phosphate buffer (0·05 mol l−1, pH 6·0) and inoculated after repeated washings of the column with phosphate buffer. Adherent organisms were then eluted from the ion-exchange columns with increasing concentrations of NaCl (0·01–0·5 mol l−1) in phosphate buffer (0·05 mol l−1, pH 6·8). Increasing salt concentrations required to elute bacteria retained on the columns correlated with the increasing negative surface charge (Smith et al. 1990).

Hydrophobic interaction chromatography (HIC) was performed using a phenyl-sepharose column (Pharmacia). Culture samples (0·1 ml) from bacterial cultures (1010 cells ml−1) were passed through the column by washing with 5 mol l−1 PBS. Retained bacteria were estimated by the difference in O.D.600 nm before and after passage. Increasing retention on the column correlated with increasing bacterial cell surface hydrophobicity (Biavati et al. 1992).

The procedure of Westergren and Olsson (1983) was used to measure microbial adhesion to hydrocarbons (MATH). Briefly, bacteria were washed twice in 10 ml phosphate buffer and diluted in the same buffer to 0·5 O.D. units at 600 nm. One hundred ml hexadecane (Merck) was added to 3 ml of the bacteria. The tube was vortexed for 1 min. After 15 min, the O.D. of the water phase was determined. Adsorbance to hexadecane correlates with surface hydrophobicity.


Adhesion to cell lines

Thirteen strains of B. longum were examined for adhesion to human Caco-2 cells.

The results (Fig. 1) indicate that the ability of B. longum to adhere varies considerably between strains: B1, B2, B6, B7, B9, B10, B1604 and B2352 were adhesive, particularly B1, B2 and B6 which were strongly adhesive. Strains B8, B1990, B2192, B2406 and B2577 were non adhesive. Adhesive and non-adhesive B. longum were, respectively, designated as Adh+ strains (B1, B2, B6, B7, B9, B10, B1604 and B2352) and Adh– strains (B8, B1990, B2192, B2406 and B2577).

Figure 1.

Adhesion of Bifidobacterium longum strains to Caco-2 cells. Twenty randomized microscopic fields per coverslip were counted. Each adhesion assay was conducted in duplicate with cells from three successive passages. Adhesion is expressed as mean ±s.e. of bifidobacteria adhering per 100 Caco-2 cells

Adh+ and Adh– strains were examined for adhesion to HT29 and KATO III cells (data not shown). All of the Adh+ strains were able to adhere while all of the Adh– strains were not.

Autoaggregation ability

Autoaggregation was investigated on Adh+ and Adh– on the basis of their sedimentation characteristics (Fig. 2). Three autoaggregation phenotypes were found and defined as follows. Strongly autoaggregating Agg+ strains (B7, B8, B10, B1604 and B2352) showed a high autoaggregation percentage (≥ 80%) aggregating immediately, forming a precipitate and resulting in a clear solution. Non-autoaggregating Agg– strains (B1990, B2192, B2406 and B2577) were unable to autoaggregate (autoaggregation percentage ≤ 10%) and produced constant turbidity. Mixed Agg+/– strains (B1, B2, B6 and B9) showed an autoaggregation percentage of around 50% and their suspension showed both a precipitate and constant turbidity. All Adh+ autoaggregated and all of the Adh– strains, except B8, did not show autoaggregation (Fig. 2). The most adhesive strains (B1, B2 and B6) showed an Agg+/– phenotype (autoaggregation ≈ 50%).

Figure 2.

Autoaggregation ability of Adh+ (▪) and Adh– (□) Bifidobacterium longum strains. Error bars represent standard errors of the mean (n = 12)

Surface hydrophobicity

Cell surface hydrophobicity methods do not measure the intrinsic microbial cell surface hydrophobicity, but rather the bacterial adhesion to a certain hydrophobic substratum (i.e. hydrocarbons in MATH, sepharose beads with covalently bound hydrophobic moieties in HIC, etc.) (Busscher and Van der Mei 1995). Consequently, results obtained by one technique may not be comparable with those obtained by another. For this reason we used four different methods to measure bifidobacterial hydrophobicity.

Adh+ and Adh– were compared for cell surface hydrophobicity as determined by SAT, ion exchange chromatography (IEC), HIC and MATH; great variability in hydrophobicity was observed both amongst the several strains tested and with respect to the test procedures adopted. Adh+ and Adh– strains showed a similar range of hydrophobicity values measured by SAT, IEC and HIC (data not shown). When hydrophobicity was measured by MATH, a relationship with adhesion ability was observed (Fig. 3), except for three strains that were not able to adhere in spite of their apparent hydrophobicity.

Figure 3.

Adhesion ability to Caco-2 cells as a function of surface hydrophobicity measured by microbial adhesion to hydrocarbons. *Percentage of bacteria adsorbed by hexadecane

Microbial adhesion to hydrocarbons hydrophobicity vs autoaggregation

Percentages of autoaggregation of Adh+ and Adh– strains were plotted in Fig. 4 against hydrophobicity values determined by MATH. Adhesive strains never exhibited low hydrophobicity or autoaggregation values.

Figure 4.

Relationship between autoaggregation ability and hydrophobicity measured by microbial adhesion to hydrocarbons in Adh+ (●) and Adh– (○) strains. Hydrophobicity is expressed as percentage of bacteria adsorbed by hexadecane


Adhesion is a complex trait that could be a multistep process in which both non-specific mechanisms and a specific ligand receptor play a role. For this reason we considered the relationship between the combination of different traits (hydrophobicity and autoggregation ability) and the adhesiveness of 13 Bifidobacterium strains.

A great variability in adhesion ability was observed among the strains. All fresh human isolates from gastric juice (B1, B2, B6, B7, B8, B9 and B10), except for B8, were strongly adhesive, while only two (B1604 and B2352) of seven strains from a collection of freeze-dried cultures (repeatedly transferred and freeze dried) were able to adhere (Fig. 1). An irreversible loss of surface structures may therefore occur during laboratory cultivation.

In all of the three cell lines tested (Caco-2, HT29 and KATO III), bacteria manifested the same adhesive qualities. More analysis will aid understanding of whether the adhesion observed is a specific bacteria–host cell interaction (e.g. involving a specific cellular surface structure common to the three cell types studied) or is a general mechanism such as electrostatic binding. Future studies are required to determine whether environmental factors, cell growth phase and degree of pleomorphism affect the expression of adherence traits in bifidobacteria. For example, studies with lactobacilli have shown that pH influences the binding of bacteria to Caco-2 cells (Granato et al. 1999).

In this study, we demonstrated that autoaggregation is strongly related to adhesion. Indeed, all of the Agg– strains were unable to adhere to Caco-2 cells (Fig. 2). This confirms observations made by Pérez et al. (1998) on B. bifidum and Del Re et al. (1998) on B. suis.

The fact that the most adhesive strains (B1, B2 and B6) showed the Agg+/– phenotype suggests that the simultaneous presence in a strain of two subpopulations of bacterial cells, autoaggregating and non autoaggregating, provides an advantage in adhesion ability. Further studies are required to determine how the inter-relationship between the two phenotypes takes place and how it supports the adhesion ability of the strain.

The only Agg+ strain (B8) that did not adhere showed a low degree of hydrophobicity as measured by MATH (Fig. 4). Therefore, hydrophobicity, as determined by MATH, and autoaggregation ability seem to be two independent traits, both of which are necessary for adhesion.

In conclusion, our findings indicate that the ability to autoaggregate, together with cell surface hydrophobicity as measured by MATH, could be used for preliminary screening to identify potentially adherent bacteria suitable for commercial purposes.