OUP user menu

Immunomodulating potential of supplementation with probiotics: a dose-response study in healthy young adults

Hanne Risager Christensen, Charlotte Nexmann Larsen, Pernille Kæstel, Lisbeth Buus Rosholm, Claus Sternberg, Kim Fleischer Michaelsen, Hanne Frøkiær
DOI: http://dx.doi.org/10.1111/j.1574-695X.2006.00109.x 380-390 First published online: 1 August 2006


Certain probiotic microorganisms have been found beneficial in the treatment of immune-related diseases and may also affect immune function in healthy people. Intervention studies of probiotics in healthy humans are urgently required. Here, the immunomodulating potential of Bifidobacterium animalis ssp. lactis (BB-12) and Lactobacillus paracasei ssp. paracasei (CRL-431) was studied in a double-blind placebo-controlled parallel dose–response trial (n=71) based on five randomly assigned groups of young healthy adults supplemented for 3 weeks with 0, 108, 109, 1010 and 1011 CFU day−1, respectively, of a mixture of BB-12 and CRL-431. No statistically significant dose-dependent effect was found for phagocytic activity in blood leukocytes, fecal immunoglobulin A (IgA) concentrations or production of interferon-γ and interleukin-10 in blood cells. When evaluating data according to the amount of viable BB-12 recovered from faeces, the interferon-γ production in blood cells was significantly reduced. In conclusion, no solid effect on the immune function of young healthy adults supplemented with even high doses of B. animalis ssp. lactis BB-12 and L. paracasei ssp. paracasei CRL-431 was demonstrated in this study.

  • probiotic
  • immunomodulation
  • healthy human adults
  • dose—response


It is well documented that stimulating signals from gut microorganisms are critical for the development and maintenance of the immune function (Cebra, 1999; Holt & Jones, 2000). A corollary, which is collecting increasing evidential support, is that the panel of stimulating signals bestowed by the gut flora is dynamically affected by the microbial composition, with strong strain-dependent differences in activity (He et al., 2001; Kirjavainen et al., 2001; Guarner & Malagelada, 2003). In view of this, the concept of manipulating the gut flora by oral supplementation of live health-promoting microorganisms — probiotics — has emerged. Lactobacilli and bifidobacteria are components of the gut flora and are frequently used as probiotics (Tannock, 2002).

A tremendous amount of data on the effect of probiotic bacteria has accumulated in recent years. This, however, mainly consists of data from either in vitro or animal studies, leaving a need for more solid clinical studies. For the clinical studies, the most well documented effects of probiotics have been found in studies on certain diseases such as infantile diarrhoea (Guandalini et al., 2000; Rosenfeldt et al., 2002), atopic eczema in children (Isolauri et al., 2000; Kalliomaki et al., 2001; Rautava et al., 2002; Rosenfeldt et al., 2003) and inflammatory bowel disease (Fedorak & Madsen, 2004; Mimura et al., 2004). However, it has been speculated that probiotics may also be effective in healthy individuals, for example by preventing atopic diseases in predisposed individuals (Kalliomaki et al., 2001). Furthermore, the most effective probiotics are derived from the gut flora of healthy humans. Therefore, there is a rationale for studying the influence of probiotics on the healthy individual, with the motivation of long-term prevention of disease and general improvement of health of the general population, not only select populations. Moreover, there is much commercialization of probiotics for healthy individuals, prompting the need for solid data on proven effects of probiotics on these individuals.

To enhance the pertinence of such clinical studies, a dose–response design comprising relatively broad-ranging doses is needed to ensure a proper basis for firm conclusions. In those few studies focusing on effective dose, only a narrow dose-range has been included, typically involving only two different doses (Donnet-Hughes et al., 1999; Gill et al., 2001).

Here, we present data on the analysis of the effect on immune parameters from a double-blind placebo-controlled intervention study involving groups of human young adults supplemented with different doses (108–1011 CFU day−1) of the bacteria Bifidobacterium animalis ssp. lactis BB-12 (BB-12) and Lactobacillus paracasei ssp. paracasei CRL-431 (CRL-431). Besides the data described here, data on the fecal bacterial spectrum, blood lipids and parameters of well-being of the subjects were collected. These data will be reported elsewhere (C. N. Larsen et al., unpublished data).

Subjects and methods


The study cohort comprised 75 healthy volunteers recruited by advertising in local papers; 71 subjects (46 women, 25 men, mean age 25.6 years, range 18–40 years) completed the study. The volunteers gave informed written consent before the experiment. Exclusion criteria included: gastrointestinal, endocrine or immune system diseases including atopic diseases; colostomia patients; pregnant or breast-feeding women; treatment with corticosteroids, antibiotics (within the last 8 weeks), antacids (within the last 4 weeks), immune suppressive drugs or any drug potentially affecting the gastrointestinal function and flora; or any condition that would impair compliance.

Experimental design

The protocol was approved by the municipal Ethical Committee of Copenhagen and Frederiksberg, Denmark (journal number: (KF) 02-093/02), and by the Danish Medicines Agency (journal number: 2612–2178). The study was performed as a double-blind, placebo-controlled dose–response trial with a run-in period of 2 weeks, followed by an intervention period of 3 weeks and then a wash-out period of 2 weeks (Table 1). The study consisted of five groups of approximately 15 subjects each (prestudy sample-size calculations). Subjects were randomly assigned by computer to receive either two capsules per day, giving a total dose of either 108, 109, 1010 or 1011 CFU day−1 of BB-12 and CRL-431 in approximately equal amounts, or two placebo capsules containing the same amount of dextrose, starch, microcrystalline cellulose as in capsules with bacteria. Due to the size of the study, some of the analyses included were predestined only to provide effect/no effect information and thus were only performed on samples from the 1010 CFU day−1 and the placebo group. Therefore, without disclosing which treatment was given, subject numbers of the placebo and the 1010 CFU day−1 group (commonly used dose) were marked by the computer with a star for recognition. The capsules were manufactured by Chr. Hansen A/S (Hørsholm, Denmark). Fecal and blood samples were obtained at scheduled visits as outlined in Table 1. The subjects were instructed to deliver a fresh fecal sample (<24 h) collected in two containers, which were kept cold with freezer elements and freezer bags until the next visit. Upon receipt at the visit, one sample was frozen immediately at −20°C until further analysis, while the other sample was kept at 4°C for a maximum of 3 h before further analysis. Blood samples were drawn from the antecubital vein after 10 min of relaxation in a supine position. Subjects fasted for at least 10 h before each visit. The subjects were instructed to abstain from ingesting any kind of products containing lactic acid bacteria during the 7-week study period.

View this table:
Table 1

Study design and scheduled visits

Week−2 weeks−1 week0 week1 week2 weeks3 weeks4 weeks
Visit number1 (day–14)2 (day 0)3 (day 21)4 (day 35)
Treatment beginX
Treatment endX
Blood sampleXXXX
Fecal sampleXXX

Measurement of phagocytic activity

Phagocytic activity was determined by flow cytometry in heparinized freshly drawn whole blood using fluorescein isothiocyanate (FITC)-labeled opsonized Escherichia coli bacteria in accordance with the manufacturer's protocol (Phagotest®, BD Biosciences, Brondby, Denmark). Briefly, 100 µL of blood was incubated on ice for 10 min, then mixed with 20 µL ice-cold FITC-labeled E. coli and incubated for 10 min at 37°C; controls were kept on ice. After incubation, the samples were put back on ice and 100 µL of ice-cold ‘quenching solution’ was added to block fluorescence from bacteria that had not been phagocytosed. The samples were then washed twice with 3 mL ‘washing solution’ (Phagotest) and centrifuged for 5 min at 250 g at 4°C. Erythrocytes were lysed with 2 mL ‘lysis buffer’ (Phagotest) and the samples were washed another three times. After the last centrifugation, 200 µL ‘DNA staining solution’ (Phagotest) was added to the samples to enable discrimination between diploid cells and bacteria. The samples were analyzed within an hour by flow cytometry (FACS Vantage™, BD Biosciences) using 488 nm excitation wavelength. A primary gate on red fluorescent cells was set from which total phagocytic activity was determined. From the primary gated cells, secondary gates for the granulocytes and monocytes were set on the forward vs. side scatter dot plots to determine phagocytic activity in these cell subsets.

Recovery of BB-12 and CRL-431

Samples of approximately 3 g of fresh faeces were weighed in Stomacher bags and diluted 10 times in 1.5% (w/v) peptone (Oxoid, Basingstoke, UK) saline solution (Merck, Darmstadt, Germany) and 0.05% (w/v) cysteine hydrochloride (Merck). After 3 min of stomaching (1 min at high velocity and then 2 min at medium velocity), appropriate 10-fold dilutions were made and spread onto freshly prepared agar plates of de Man, Rogosa and Sharpe medium (MRS; Oxoid). For selective isolation of CRL-431, the plates contained vancomycin (50 g L−1; Sigma-Aldrich, St Louis, MO) and were incubated aerobically for 3 days at 37°C. For selective isolation of BB-12, plates contained tetracycline (4 g L−1, Sigma-Aldrich) and 0.05% (w/v) cysteine hydrochloride and were incubated anaerobically (Anaerogene, Oxoid) for 3 days at 37°C. After incubation, 10 solitary colonies were randomly picked from plates of 20–200 colonies, which were further spread on MRS agar with or without 0.05% (w/v) cysteine hydrochloride and incubated for 3 days anaerobically and aerobically, respectively. Pure CRL-431-like colonies were aerobically grown overnight at 37°C in MRS broth (Difco, BD Diagnostic System, Sparks, MD) with 10 mm threonine (Sigma-Aldrich) and further identified by pulsed-field gel electrophoresis (PFGE) performed according to published techniques (Hung & Bandziulus, 2005). Pure BB-12-like colonies were incubated anaerobically in MRS broth with 0.05% (w/v) cysteine hydrochloride and frozen at −18°C in 15% (w/v) glycerol until identification. Identification of BB-12 was done by fluorescence whole cell hybridization as published previously with some modifications (Beimfohr et al., 1993). Cell material from purified isolates was directly (fixation step omitted) smeared onto microscope slides, dried at 37°C for 30 min, covered with a 1 mg mL−1 lysozyme solution (100 mm Tris/HCl, 50 mm EDTA, pH 8.0) and then incubated again at 37 °C for 30 min. The B. animalis ssp. lactis-specific probe Bila-4 labeled with Cy3 was used as the BB-12-specific probe (E. Brockmann, in preparation). Eub338 labeled with fluorescein was used as the universal probe (Stahl et al., 1989).

Extraction of IgA from faeces

Frozen fecal samples were thawed and sample portions of approximately 1 g were weighed and suspended by mixing thoroughly with a volume of 50 times the sample weight of phosphate buffered saline (PBS; 0.01 m, pH 7.4) containing 0.1% (w/v) Tween-20. The suspension was incubated for 2 h at 4°C while shaking slowly. The suspension was then centrifuged for 15 min at 10 000 g at room temperature and then frozen at −80°C until ELISA analysis for IgA.

Measurement of antibody titre in blood and fecal extracts by ELISA

Only blood samples from the star-marked groups (placebo and the 1010 CFU day−1 group) were included in this analysis. Microtitre wells (MaxiSorp; Nunc, Roskilde, Denmark) were coated overnight at 4°C with anti-human IgA (clone G18-1; 1 µg mL−1), IgG (clone G18-145; 0.8 µg mL−1) or IgM (clone JDC-15; 1 µg mL−1) capture antibody (all from BD Biosciences) in carbonate buffer (0.05 m, pH 9.6). After washing in PBS with 0.05% (w/v) Tween-20 (PBS-T), 100 µL well−1 of serially diluted blood samples or fecal extracts was added and the plate was incubated for 1 h at room temperature. Thereafter, the plate was washed and incubated for 1 h at room temperature with biotinylated anti-human IgA (G20-359; 1 µg mL−1), IgG (G18-145; 0.5 µg mL−1) or IgM (G20-127; 0.5 µg mL−1) detection antibody (all from BD Biosciences), then washed again and incubated for another hour at room temperature with horseradish peroxidase-conjugated streptavidin (0.12 µg mL−1; DakoCytomation, Glostrup, Denmark). The plate was developed by incubating with 3,3′,5,5′-tetramethylbenzidine–peroxide solution and the reaction was stopped after 10 min by adding 2 M phosphorous acid. Optical density was measured at 450 nm with 630 nm as reference. Results were expressed as titres, defined as the reciprocal of sample dilution giving an absorbance of 0.2. A blood sample chosen as a reference and aliquoted to avoid repeated freezing/thawing was included on all plates.

Measurement of cytokines in in vitro-stimulated blood

Only blood samples from the star-marked groups were included in this analysis. Less than 1.5 h after the blood was drawn, it was diluted three times in Iscove's modified Dulbecco's medium (BioWhittaker Europe, Verviers, Belgium) containing 0.1% (v/v) fetal bovine serum (BioWhittaker), 100 µg mL−1 penicillin, 100 IU mL−1 streptomycin (Sigma-Aldrich) and 50 IU mL−1 sodium heparin (Sigma-Aldrich) and plated as 300 µL per well in 96-well round-bottomed plates (Nunc). To the wells was then added 50 µL of medium per well (unstimulated control), 1 µg mL−1 (final concentration) of lipopolysaccharide (LPS, E. coli 026 : B6, Sigma-Aldrich) or 50 µg mL−1 (final concentration) of phytohaemagglutin-P (PHA-P, Phaseolus vulgaris, Sigma-Aldrich). Samples were analyzed in quadruplicate. Noteworthy, preliminary studies showed that leaving the samples for more than 1.5 h before stimulation reduced the cytokine response substantially. After 24 h of incubation at 37°C and 5% (v/v) CO2, the plate was centrifuged at 250 g and 4°C and then the supernatants were collected and frozen at −80°C until cytokine analysis.

To eliminate the impact of atmospheric incubation during blood stimulation and to facilitate uniform and quick handling of a large number of samples, a separate blood sample from all individuals was stimulated directly in the blood collection tube. Just before drawing the blood, a freshly prepared mixture of 500 ng mL−1 (final concentration) of LPS and 50 µg mL−1 (final concentration) of PHA-P was applied to the tube (3 mL, heparinized, Venoject, Leuven, Belgium) using a syringe with a thin needle to avoid leakage of air into the vacuum tube. After 24 h of incubation, the blood was centrifuged for 10 min at 250 g at 4°C and then the serum fraction was collected and frozen at −80°C until cytokine analysis.

Interferon (IFN)-γ was determined according to manufacturer's instructions (CytoSet™ Biosource, Nivelles, Belgium). The detection limit was approximately 50 pg mL−1. Interleukin (IL)-10 was determined using paired antibodies: antihuman IL-10 antibody (clone JES3-12G8; 1 µg mL−1) for coating and biotinylated antihuman IL-10 antibody (clone JES3-12G8; 1 µg mL−1) for detection, both antibodies from BD Biosciences. The detection limit was approximately 80 pg mL−1.


All statistical analyses were performed using SPSS 12.0 (SPSS®, Chicago, IL). Normal distributed variables were described as mean±standard deviation (SD). To assess whether there was a dose–response effect of ingestion of probiotics, linear regression analyses were performed with bacterial dose coded as 0 (placebo), 1 (108), 2 (109), 3 (1010), or 4 (1011 CFU day−1). Using this strategy, a dose–response effect was defined as a slope different from unity of the line between these points. The regression coefficient thus reflects the theoretical change in the dependent variable per a 10 times increase of bacterial dose, from 108–1011 CFU day−1. We used a successive analysis strategy, with the first step being the linear regression analysis using the bacterial dose in a linear scale. If this test was significant, the different bacterial doses were added to the model as a categorical variable, whereby the deviation from the estimated line was estimated for each point. Independence and normal distribution of residuals were assessed by plotting standardized residuals against predicted values. Level of significance was defined as P<0.05 and 95% confidence interval (CI) is reported with the estimates. As the difference of baseline values reflected intraindividual variation rather than effects caused by changed behavior in the run-in period, it was decided to include both baseline values of the dependent variable (visit 1, before the run-in period, and visit 2, immediately before intervention) as covariates in all analyses.


Subject characteristics

The age and sex distribution of the five groups enrolled in the study are given in Table 2. Despite randomization, the proportion of males was unfortunately lower in the 1011 CFU day−1 group than in the other groups.

View this table:
Table 2

Characteristics of study groups

    n (F+M)1414151315
    Age (years)26 ± 528 ± 626 ± 327 ± 426 ± 5
    Sex (F/M)8/69/58/78/513/2
Ingested bacteria
    Total (CFU day−1)01.1 × 1081.1 × 1091.9 × 10102.4 × 1011
    CRL-431 (CFU day−1)00.2 × 108 ± 0.00.4 × 109 ± 0.10.3 × 1010 ± 0.10.9 × 1011 ± 0.1
    BB-12 (CFU day−1)01.0 × 108 ± 0.20.6 × 109 ± 0.11.6 × 1010 ± 0.21.5 × 1011 ± 0.2
Baseline values
    Total (%)52.0 ± 455.3 ± 754.8 ± 853.4 ± 855.1 ± 8
    Granulocytes (%)85.5 ± 686.7 ± 686.3 ± 786.0 ± 785.5 ± 6
    Monocytes (%)71.8 ± 771.4 ± 571.4 ± 772.3 ± 871.8 ± 6
    IgA, faeces17.1 ± 1.216.8 ± 2.115.9 ± 1.917.1 ± 1.816.8 ± 1.7
    IgA, blood19.0 ± 0.9NDND18.7 ± 0.5ND
    IgG, blood19.8 ± 0.9NDND19.5 ± 0.6ND
    IgM, blood19.6 ± 1.0NDND19.8 ± 1.1ND
    IFN-γ, tube (ng mL−1)18.2 ± 11.215.2 ± 8.416.2 ± 7.115.5 ± 6.112.6 ± 6.4
    IL-10, tube (ng mL−1)2.06 ± 1.01.16 ± 0.41.53 ± 0.41.64 ± 0.51.57 ± 0.6
    IFN-γ, LPS (ng mL−1)0.5 ± 0.5NDND0.80 ± 0.7ND
    IL-10, LPS (ng mL−1)1.05 ± 0.4NDND0.93 ± 0.4ND
    IFN-γ, PHA-P (ng mL−1)6.84 ± 5.1NDND4.42 ± 1.9ND
    IL-10, PHA-P (ng mL−1)1.27 ± 0.5NDND1.09 ± 0.4ND
  • Mean ± SD.

  • Mean of visit 1 sample (2 weeks before intervention) and visit 2 sample (immediately before intervention) ± SD.

  • Stated in titre units defined as the least sample dilution (Log 2) giving an absorbance of 0.2.

  • No sample at visit 1, thus the SD stated is the analytical SD and not between visit 1 and 2 results.

  • ND, not determined (only determined for star-marked groups, i.e. the placebo and the 1010 group).

  • Measured in serum from blood stimulated directly in the tube with a mixture of LPS and PHA-P; see ‘Subjects and methods’.

  • Measured in blood stimulated with LPS in 96-well plates; see ‘Subjects and methods’.

  • Measured in blood stimulated with PHA-P in 96-well plates; see ‘Subjects and methods’.

  • IFN, interferon; Ig, immunoglobulin; IL, interleukin; LPS, lipopolysaccharide; PHA-P, phytohaemagglutin.

As the numbers of bacteria in the large scale manufactured capsules may deviate somewhat from the intended concentrations, the bacterial content of the capsules used in the study was checked to keep track of the exact doses given in each group. Based on the intake of two capsules a day, the daily dose of total bacteria was in the intended range for each group (Table 2). It was found, however, that the distribution of CRL-431 and BB-12 in the capsules was not entirely equal, with a tendency for BB-12 to preponderate.

Baseline values for all of the immune parameters included in the present study are calculated as the mean of visit 1 (before the run-in period) and visit 2 (immediately before intervention) and are outlined in Table 2 as a reference for later data, which include only differences over time.

Fecal recovery

To evaluate the survival of the ingested bacteria through the gastrointestinal tract, fecal recovery analysis of the bacteria was performed. Despite ingestion of amounts up to 1011 CFU day−1 of CRL-431 for the highest dose group, viable CRL-431 bacteria could not be isolated from the fresh fecal samples from either visit 3 or visit 4. Spiking experiments using human faeces spiked with CRL-431 and stored under the same conditions as in the trial confirmed that the lack of CRL-431 recovery was not a methodological issue (data not shown).

In contrast, the fecal recovery of BB-12 exhibited a dose–response relationship (linear regression analysis: P<0.001) with 1010 CFU day−1 being the lowest dose giving a statistically significant chance of recovering viable BB-12 from the faeces (logistic regression analysis) (Fig. 1). Linear regression analysis of the BB-12 fecal recovery data gives a regression coefficient of 1.30, i.e. within the dose range tested in the present study; a 10-fold increase of ingested BB-12 caused the average number of recovered viable BB-12 to increase by a factor 20 (101.3). It is moreover evident from the scatter plot (Fig. 1) that the higher the ingested dose, the greater the number of subjects from whom BB-12 was recovered, reaching as high as 13 of the 15 subjects in the 1011 CFU day−1 group. Notably, BB-12 was recovered in low amounts from visit 3 faeces of two subjects from the placebo group. The recovery of BB-12 from these subjects was further verified by PFGE fingerprinting. Despite the subjects being asked to abstain from any product containing lactic acid bacteria, the recovery of BB-12 is most likely due to intake of commercial products containing BB-12. These subjects were excluded from further data analysis. Viable BB-12 was recovered in visit 4 samples from only three subjects: two from the 1011 CFU day−1 group and one from the 109 CFU day−1 group (data not shown), indicating no long-term colonization of the gut with BB-12.

Figure 1

Recovery of Bifidobacterium animalis ssp. lactis BB-12 (BB-12) (±95% CI) in faeces immediately after cessation of bacterial supplementation (visit 3 sample) depicted against supplementation dose. Linear regression analysis on dose–response relationship gives P<0.001. Logistic regression analysis (P<0.001) gives the P-values stated at the top of each column, indicating the statistical chance of recovering viable BB-12 in faeces from subjects supplemented with the given dose of BB-12; the 1010 CFU day−1 dose was the lowest dose providing a statistically significant chance of recovering viable bacteria.

Phagocytic activity

Granulocytes, constituting about 60% of blood leukocytes, are efficient phagocytic cells and are the major cells of phagocytic activity in blood (Bick, 1993). Monocytes, constituting only about 3–7%, contribute less to the total phagocytic activity of blood and are themselves less efficient phagocytic cells. Phagocytic activity in blood as a whole (total phagocytic activity) and in the granulocyte and the monocyte cell fractions was evaluated as the change between the baseline value (Table 2) and the value immediately after intervention (visit 3) or 2 weeks after intervention (visit 4) of each individual. Although a general increase of phagocytic cells in whole blood as well as in the granulocyte and monocyte cell fraction was observed at visit 3, no statistically significant effect was seen in any of the dose groups compared to the placebo group, nor was there a dose–response relationship (Fig. 2). This is due to an increase of the phagocytic activity of the placebo group of the same magnitude as in the intervention groups. The same pattern was evident for the data from visit 4 (data not shown). As the methodology demanded recalibration of the flow cytometer on each new day of analysis, the mean fluorescence useful for comparing phagocytic capacity on a per cell basis could not be determined.

Figure 2

Difference in phagocytic activity (±95% CI) of peripheral blood (total) (a) the blood granulocytes (b) and monocytes (c) from before supplementation with a mixture of Bifidobacterium animalis ssp. lactis BB-12 and Lactobacillus paracasei ssp. paracasei CRL-431 (baseline) to immediately after cessation of supplementation (visit 3). Baseline values (mean of visit 1 and 2) are given in Table 2. There was no statistically significant dose–response relationship (linear regression) for total, granulocyte or monocyte phagocytosis (P=0.68, 0.56 and 0.41, respectively).

Fecal IgA and blood antibody

Total IgA was determined in fecal extracts from all subjects. There was no statistically significant difference between the amount of fecal IgA found before the intervention (visit 2, Table 2) and immediately after (visit 3) (Fig. 3) or 2 weeks after (visit 4) (data not shown) for any of the doses of CRL-431 and BB-12.

Figure 3

Difference in total fecal immunoglobulin A (±95% CI) from before supplementation with a mixture of Bifidobacterium animalis ssp. lactis BB-12 and Lactobacillus paracasei ssp. paracasei CRL-431 (baseline) to immediately after cessation of supplementation (visit 3). Baseline values (mean of visit 1 and 2) are given in Table 2. There was no statistically significant dose–response relationship (linear regression, P=0.88).

In the star-marked groups, i.e. the placebo and the 1010 CFU day−1 group, the level of total IgA, IgG and IgM was determined in blood; no significant difference was found for any of the immunoglobulins (data not shown).

Cytokine production in in vitro-stimulated blood

To detect whether supplementation with CRL-431 and BB-12 affects the cytokine production pattern of peripheral blood cells, blood samples were stimulated in vitro with LPS and/or PHA-P, which in preliminary studies were found to be efficient inducers of IL-10 and IFN-γ in blood. IL-10 is an anti-inflammatory cytokine involved in the maintenance of immune homeostasis, whereas IFN-γ is a proinflammatory cytokine upregulated during inflammation (O'Garra et al., 2004). Blood stimulated by a mixture of LPS and PHA-P directly in the tube, thus avoiding interference of atmospheric incubation and variable handling time until stimulation, did not show any statistically significant dose–response relationship for the difference in IFN-γ and IL-10 production from before (baseline) to after supplementation (Fig. 4).

Figure 4

Difference in production of IFN-γ (a) and IL-10 (b) (±95% CI) in blood samples stimulated directly in the blood collection tube by a mixture of lipopolysaccharide and phytohaemagglutin from before (baseline) to immediately after cessation of supplementation (visit 3). Baseline values (mean of visit 1 and 2) are given in Table 2. There was no statistically significant dose–response relationship (linear regression, P=0.35 and 0.55, respectively).

Blood from the star-marked groups (placebo and 1010 CFU day−1 group) was also stimulated with LPS or PHA-P separately using microtitre plate wells. Again, in this system, no statistically significant effect could be found for supplementation with CRL-431 and BB-12 on the in vitro-stimulated production of IFN-γ and IL-10 (data not shown).

In the group receiving a dose of 1010 CFU day−1, BB-12 was recovered only from around half of the subjects in their visit 3 fecal samples. If survival in the gastrointestinal tract is essential for at least some of the effects of probiotics, as indicated in several studies (Dunne et al., 2001; Gill & Rutherfurd, 2001a; Galdeano & Perdigon, 2004), an alternative way of stating the results of the present study is to use the number of BB-12 recovered from faeces rather than the supplementation dose as the independent variable. With this approach, the data from the in vitro-stimulated blood of the star-marked groups showed a statistically significant decrease in IFN-γ production in LPS-stimulated blood with an increased number of recovered BB-12 bacteria, whereas no statistical effect was seen for IL-10 production (Fig. 5). No such effect was found for the PHA-P-stimulated blood (IFN-γ: P=0.86, IL-10: P=0.84; data not shown) or for cytokine production in blood stimulated directly in the tube with a mixture of LPS and PHA-P (IFN-γ: P=0.13, IL-10: P=0.18; data not shown). There was no statistically significant effect on phagocytotic activity or fecal IgA (data not shown).

Figure 5

Difference in production of interferon-γ (a) and interleukin-10 (b) in blood samples from star-marked groups (placebo and 1010 CFU day−1 group) stimulated in vitro with lipopolysaccharide depicted according to fecal recovery of viable Bifidobacterium animalis ssp. lactis BB-12. The stated P-values come from linear regression analysis and the error lines are 95% CI.


In the present study, the immunomodulating effect of CRL-431 and BB-12 was studied in healthy young adults in a double-blind placebo-controlled trial with phagocytic activity of blood cells, fecal IgA and cytokine production in in vitro-stimulated blood as the main read-outs.

Gastrointestinal survival of the supplemented bacteria was followed by bacterial recovery analysis. It was not possible to recover viable CRL-431 bacteria from the faeces. CRL-431 bacteria are resistant to bile, but have been found to be labile at low pH (2.5) (Jacobsen et al., 1999; in that article CRL-431 is referred to as CHCC3136). CRL-431 was, however, included in the study due to its good adhesion to epithelial cells (Jacobsen et al., 1999) and its wide application in probiotic products. Moreover, the polysaccharide matrix surrounding the bacteria in the capsules was presumed to enhance bacterial survival during gastric transit. In contrast to CRL-431, BB-12 was recovered from the faeces in a dose-dependent manner. At the supplementation dose of 1010 CFU day−1, BB-12 was recovered from around 40% of the subjects, which is in agreement with the results of a study with Lactobacillus rhamnosus GG using similar concentrations (Kuisma et al., 2003). By increasing the dose to 1011 CFU day−1, a comparatively high dose rarely included in clinical studies, the proportion of BB-12-positive subjects increased to 90% along with a general increase in the concentration of BB-12 in the positive individuals (all >106 CFU g−1). This indicates that there is a higher degree of bacterial survival of BB-12 with increased amounts of ingested BB-12, and that quite high amounts of orally supplemented BB-12 bacteria are necessary to obtain a high proportion of subjects from whom viable BB-12 can be recovered.

No significant effect of the bacterial supplementation on the phagocytic activity of blood cells was found when compared to the placebo group. The general increase after the intervention was also observed for the placebo group, emphasizing the necessity of a placebo group. Notably, we experienced a tendency for the phagocytic activity to shift somewhat when changing the batch of the Phagotest kit used. This shift may act as a confounding factor, contributing to the systematic increase in phagocytic activity observed over time for all groups. For future work using the Phagotest kit, it is recommended to take into account batch variation.

In contrast to the present study, enhanced phagocytic activity of blood leukocytes upon supplementation of different probiotics has been found both in studies with mice (Gill & Rutherfurd, 2001a) and in clinical studies. However, the vast majority of the clinical studies are performed in the elderly (Arunachalam et al., 2000; Chiang et al., 2000; Gill et al., 2001; Gill & Rutherfurd, 2001b), whereas only a few studies involve young adults (Schiffrin et al., 1997; Donnet-Hughes et al., 1999). A decline in immunity has been observed in the elderly, which has been linked to a decrease among other populations of the bifidobacteria of the gut flora (Hebuterne, 2003). Thus, the elderly may be more susceptible to interventions and therefore an overt target group for probiotics. In a study by Schiffrin (1997) involving supplementation of young adults with either L. acidophilus La1 (La1) or BB-12, the phagocytic activity was significantly enhanced for both probiotic strains, in contrast to the findings of the present study. However, in the study by Schiffrin et al., fermented milk (fermentation bacteria not specified) with La1 or BB-12 added was used as vehicle and no placebo group was included to eliminate a potential effect of the fermented milk per se or other confounding factors. Moreover, the treatment was not blinded and normal milk was used for the run-in and wash-out periods. A later study by the same research group also reported an increase of the phagocytic activity of blood cells from subjects (21–57 years, mean age not stated) given fermented milk (Streptococcus thermophilus) with La1 added. In that study, a control group was included, which showed a minor effect of the fermented milk per se in the oxidative burst but not on phagocytic activity (Donnet-Hughes et al., 1999). Substantial variations in the phagocytic activity were observed over time, also for the control group, and the treatment was not blinded.

In the present study, no effect of CRL-431 and BB-12 on fecal IgA secretion was seen. Augmentation of antigen-specific IgA as well as total IgA production following ingestion of probiotics has been reported from studies in mice (Fukushima et al., 1999; Tejada-Simon et al., 1999) and clinical studies (Kaila et al., 1992; Majamaa et al., 1995; Malin et al., 1996; Fukushima et al., 1998; de Vrese et al., 2004), many of which are based on diseased individuals, infants, or the use of vaccination procedures, and therefore cannot be compared with the present study. For example, in a study involving infants given BB-12, fecal levels of total IgA and poliovirus-specific IgA were enhanced (Fukushima et al., 1998). Furthermore, in a recent study, administration of CRL-431 to young human adults significantly enhanced the IgA response specific toward booster polio vaccination (de Vrese et al., 2004).

It has been reported that intestinal dendritic cells can retain gut microorganisms and selectively induce IgA locally through direct interactions with B cells, which helps protect against mucosal penetration by gut microorganisms (Macpherson & Uhr, 2004). As the bacteria employed in the present study are already commercially available, a great part of the subjects may have ingested the bacteria several times earlier in life, whereby IgA specific to the bacteria may have been induced pre-experimentally, impeding direct contact between the bacteria and the gut immune system. This raises the issue whether repeated exposure to the same probiotic or exposure to different probiotics is the most efficient method to ensure the continuous efficacy of probiotics.

No effect was found for the IL-10 or IFN-γ induction in blood upon in vitro stimulation, either when stimulated directly in the collection tube with an LPS/PHA-P mixture or when stimulated with LPS and PHA-P separately in microtitre plates. However, evaluating the data on the basis of the fecal recovery of BB-12, the IFN-γ production was significantly reduced with increased recovery of BB-12, whereas IL-10 remained unaffected, although this pattern applied only to the LPS-stimulated blood and not to stimulation with PHA-P or with a mixture of LPS/PHA-P. Importantly, such data should be interpreted with caution as it is debatable whether ignoring the influence of the nonrecovered bacteria is justified. Nevertheless, these data may indicate an importance of a high degree of gastrointestinal survival for a systemic immunomodulating effect to take place. The observation of a decreased IFN-γ production with increased BB-12 recovery is in line with the results from other studies in our lab involving bacterial stimulation of dendritic cells. Here, we found that BB-12, as well as gut flora-derived bifidobacteria in general, possess a striking capacity to inhibit induction of proinflammatory cytokines, including the INF-γ-inducing IL-12, induced by otherwise strong proinflammatory LAB. The antiinflammatory cytokine IL-10 was not inhibited (H. R. Christensen, et al., unpublished data). Intestinal dendritic cells are continually exposed to the gut flora from which they receive stimuli that affect their stimulation pattern of T, B and other immune cells. Mucosal immune cells continuously migrate out of the gut mucosa and enter into the systemic circulation and hereby connect the local mucosal immune function with the systemic immune function (Stagg et al., 2003). In a recent study by Roller (2004) based on supplementation of azoxymethane-treated rats (colon cancer model) with BB-12, the IFN-γ production in cells of Peyer's patches was significantly reduced, which is in agreement with the observations outlined above.

With regard to fecal IgA secretion signalling pathways, PHA-P is a lectin that can activate a variety of cells via complex cellular carbohydrates and thus not in a receptor-specific manner (Kjær & Frøkiær, 2005), whereas LPS binds to Toll-like receptor-4 (TLR-4), leading to activation of necrosis factor-κB and proinflammatory gene expression (Abreu et al., 2002). Probiotics, and the gut flora in general, may affect the expression level of TLR-4 and other Toll-like receptors on mucosal and systemic cells, and thereby modulate the cytokine production capacity selectively towards LPS stimulation. Downregulation of TLR-4 and other Toll-like receptors on mucosal cells is known to occur in the gut to tolerate the bacterial stimuli of the normal gut flora, whereas pathogens increase Toll-like receptors and responsiveness to bacterial stimuli (Kelly & Conway, 2005).

Even though the results obtained in the present study have a no-effect overall structure, it can not be concluded that the tested bacteria exert no health-promoting effects. Employing healthy young adults in a study of the immune modulating effects of probiotics is somewhat problematic. As long as the etiology of diseases such as allergy and inflammatory bowel disease continues to be elusive, it is difficult to establish parameters of relevance to monitor in studies of the effects of probiotics in healthy individuals from a disease-prevention point of view. To enhance the measurability of the efficacy of probiotic healthy individuals, it may be essential to identify more specific target groups of individuals with higher susceptibilities to the potential influence of probiotics, e.g. low pre-experimental counts of fecal bifidobacteria, low intake of fermented foods or sustained subnormal capacity to produce cytokines upon in vitro stimulation of blood cells. To enhance understanding and effectiveness of probiotics for the large population of healthy individuals as potential consumers of probiotics products, we consider it important to continue clinical studies of this population.


The analysis of BB-12-like colonies by E. Brockmann, the spiking and PFGE of CRL-431-like colonies by M. Bennedsen, together with fruitful discussion of design and results with B.L. Jacobsen are greatly appreciated. The work was supported by Centre for Advanced Food Studies, Denmark and Chr. Hansen A/S, Denmark.


  • Editor: Patrick Brennan


View Abstract