For immobilization, pure cultures of B. longum ATCC 15707 (American Type Collection Culture, Rockville, MD, USA) or L. diacetylactis MD (Rhône Poulenc, Brampton, ON, Canada) from frozen stock (-80ºC) were grown twice in MRS broth (Rosell Institute Inc., Montreal, QC, Canada) supplemented with 0.5 g/l cysteine, 0.2 g/l Na2CO3 and 0.1 g/l CaCl2, for 18 h at 30°C for L. diacetylactis or 37°C for B. longum with a 2% (v/v) inoculum.

B. longum and L. diacetylactis cultures were immobilized separately or co-immobilized in κ-carrageenan/locust bean gum gel beads, as described previously (Lamboley et al. 1997). For gel bead production, a total inoculum of 2 % (v/v) was used for pure and mixed cultures. Co-immobilization was carried out with two strain ratios (1:1 and 9:1) for B. longum and L. diacetylactis . Beads with diameters in the range 1-2 mm were selected by wet sieving and used for subsequent fermentation experiments.

Fermentations were carried out in a stirred tank bioreactor (Bioflo model C30, New Brunswick Scientific Co., Edison, NJ, USA) purged with CO2 and inoculated with 100 ml of beads, giving a total culture volume of 500 ml. The pH-controlled batch fermentations were performed at 37°C in supplemented MRS broth containing KCl 0.1 M to keep the bead structure, with mixing set at 200 rpm and pH controlled at 5.8 or 6.2 by addition of 5 N NaOH. The fermented broth was replaced with fresh medium when the base consumption stopped. Three successive batch fermentations (pH=5.8) of 14.5 h, 11.5 h and 2 h or 14.5 h, 11.5 h and 6.5 h were carried out for pure cultures of B. longum or L. diacetylactis , respectively. Two experiments were conducted with co-immobilized cells with different strain ratios. For the first experiment, three successive batch fermentations (pH=5.8) of 14 h, 8 h and 4 h were carried out with a 1:1 initial strain ratio. For the second one, four successive batch fermentations (pH=6.2) of 12 h, 9 h , 2.5 h and 1.5 h were carried out with a 9:1 initial B. longum / L. diacetylactis ratio. Bead and broth samples were taken from the reactor at different time intervals for cell enumeration using plate counts and enzyme-linked immunosorbent assay (ELISA), and for immunofluorescence microscopy.

L. diacetylactis was enumerated by plating diluted samples on solid M17 agar (Difco Laboratories, Detroit, MI, USA) supplemented with 0.5% lactose and incubating aerobically at 30ºC for 48 h. Columbia agar base (Difco Laboratories) supplemented with 0.5 g/l cysteine, 10 g/l lactose, 2 g/l LiCl, 3 g/l sodium propionate and 20 mg/l kanamycin was used to selectively determine B. longum population, after incubation at 37ºC for 48 h in anaerobic jars. For bead population analysis, approximately 0.5 g of accurately weighed beads in 4.5 ml peptonized water was treated with an Ultra-Turax (Janke and Kunkel, Staufen, Germany) on ice for 45 s at 13500 rpm to break up the beads and resuspend the cells. Reported data are means for duplicate analyses.

A sandwich-type ELISA using anti- B. longum and anti- L. diacetylactis polyclonal antibodies labelled with horseradish peroxidase was used to specifically determine the total biomass concentration of each strain in beads, as described by Prioult et al. (2000). Reported data are means for triplicate analyses.

Bead preparation and confocal microscopy observation, using specific fluorescent polyclonal antibodies for the two strains, were performed as previously described (Prioult et al. 2000). The biomass volumetric fraction for each strain as a function of the gel bead radius was derived from the position and surface area of the colonies in the labelled sections (Hunik et al. 1993). For this determination, bead sections were divided into 2.5 μm thickness annular regions and the number and fraction of fluorescent pixels containing biomass were determined in each region as a function of radius (Figure 3.1). Total biomass concentration in the beads determined by ELISA was used to calculate cell concentration (cells/g) from the fraction of biomass in the annular regions. For this calculation, a numerical integration of occupation profiles as a function of depth in the bead, which was assumed to be perfectly spherical, was carried out for each strain. The mean profiles from one, two or five beads during L. diacetylactis fermentation were used to determine the effect of the number of beads analyzed for a given fermentation time on biomass profile accuracy. Five beads were used to determine biomass profiles during subsequent fermentations.

Bead colonization was first studied during single-strain repeated-batch fermentations. In L. diacetylactis cultures, immobilized cell concentration increased exponentially during the first 8-12 hours of the first batch fermentation to reach 5.4±1.3x109 CFU/g beads and only doubled to 1.1±0.2x1010 CFU/g at the end of the third culture (32 h; Figure 3.2a). Viable-cell counts in the broth at the end of batch cultures were very similar, and remained low between 5.7±0.2x107 CFU/ml and 1.7±0.1x108 CFU/ml. The total cell concentration in beads measured by ELISA was two- to threefold higher than viable-cell counts, at the beginning of the stationary phase of the first batch culture (12 h). Confocal microscopy images and analysis of spatial biomass distribution showed that L. diacetylactis growth was almost nil at the bead centre but mainly occurred in a 200 μm peripheral layer of the beads (Figure 3.3a). Cell concentration estimated by reference to the ELISA test reached 5.0x109 cells/g at the bead centre at the end of the first batch culture (14.5 h) and did not increase thereafter. However cell concentration in the peripheral layer increased during the three successive batch cultures and reached 2.7x1010 cells/g near the bead surface at the end of the third batch (31.5 h). Increasing the number of beads analyzed from 1 to 5 produced smoother profiles but did not change cell concentrations as a function of bead depth (Figure 3.4).

In B. longum cultures , viable-cell counts in beads increased during the first two batch cultures to a very high value of 9.6±1.0x1010 CFU/g after 26 h culture (Figure 3.2b). Viable-cell counts in the broth continuously increased with the number of batch cultures, from 8.2±0.6x108 CFU/ml at the end of the first batch to 2.5±0.2x109 CFU/ml for the third batch. As observed with L. diacetylactis , B. longum cell concentration determined by ELISA was similar to viable cell counts in beads during the exponential growth of the first batch culture but was higher during the stationary phase, with a maximal ratio of 3.2 after 14.5 h culture. Growth occurred mainly in a 300 μm peripheral layer but did not stop at the bead centre during the experiment. Cell concentration at the bead centre and periphery reached 8.0x1010 and 1.9x1011 cells/g at the end of the third culture (Figure 3.3b).

Microbial dynamics in beads and broth was then studied during fermentations with co-immobilized cells to observe possible interactions between the two strains. Figure 3.5 shows the strain concentrations in beads and broth during repeated-batch fermentations carried out at pH=6.2 with an inoculum composition of 9:1 B. longum / L. diacetylactis . Viable-cell counts of L. diacetylactis in beads increased until 9-12 h of culture during the first batch and remained constant during the subsequent batches at 2.1±0.1x1010 CFU/g. For B. longum , viable-cell counts in beads increased during repeated-batch fermentations up to 9.9±0.1x1010 CFU/g at the end of the fourth batch (26 h). Lactococci free-cell counts in the broth at the end of the batch cultures did not change from the first to the fourth batch, averaging 2.2±0.7x108 CFU/ml. On the other hand, B. longum viable-cell counts at the end of batch cultures progressively increased from 1.1±0.1x108 to 1.3±0.1x109 CFU/ml from the first to the fourth batch culture. L. diacetylactis and B. longum concentrations in beads determined by ELISA were four- to fivefold higher than the viable-cell counts after an initial culture period (Figure 3.5). L. diacetylactis concentration at the bead centre reached 1.2x1010 cells/g at the end of the second batch culture (21 h) and did not change thereafter (Figure 3.6a). Growth was concentrated in a 200-250 μm peripheral layer and peaked at 5.3x1010 cells/g near the bead surface (80 μm) at the end of the experiment. B. longum cell concentration near the bead centre and the bead surface (60-120 μm) increased gradually during the repeated cultures, to maxima of approximately 7.3x1010 and 1.5x1011 cells/g (Figure 3.6b).

A second experiment with co-immobilized cells was conducted under the same conditions except for pH control (5.8) and strain ratio (1:1) in the inoculum used for bead preparation, to tentatively favor the growth of L. diacetylactis and decrease the dominant behaviour of B. longum observed in the first experiment. Very similar data were obtained for both co-immobilized culture experiments, for both immobilized and free-cell concentrations and biomass profiles in gel beads (data not shown).