Entête

Chapitre IV: Production et caractérisation d’anticorps monoclonaux anti-bifidobactéries et leur application dans le développement d’une méthode de détection en immuno-culture. Production and characterization of anti-bifidobacteria monoclonal antibodies and their application in the development of an immuno-culture detection method

Table des matières

T. Amrouche a,b,c, Y. Boutin b,d, O. Moroni a,b, E. Kheadr a,e, and I. Fliss a,b*

a Dairy Research Center STELA, Pavillon Paul-Comtois, Université Laval,

Québec (Qc) G1K 7P4,

b Institute of Nutraceutical and Functional Foods INAF, Université Laval,

Québec (Qc) G1K 7P4,

c Department of Food Technology, Faculty of Agronomy and Biological Sciences, University of Tizi-Wezzu, Algeria,

d TransBiotech, CEGEP Lévis Lauzon, Lévis (Qc) G6V 9V6,

e Department of Dairy Science and Technology, Faculty of Agriculture, University of Alexandria, Alexandria, Egypt.

Journal of Microbiological Methods: sous presse

Les travaux rapportés dans ce chapitre représentent la continuité de ceux présentés au premier chapitre et qui démontrent la mitogénicité de la paroi des bifidobactéries. L’objectif visé par ce chapitre est de produire et caractériser des anticorps monoclonaux spécifiques aux bifidobactéries afin de développer une méthode de détection et de quantification des bifidobactéries dans les aliments et l’environnement. Des souches standard ATCC de bifidobactéries d’origine humaine et animale ont été utilisées.

Les bifidobactéries, très connues pour leur potentiel probiotique, sont de plus en plus utilisées dans l’industrie alimentaire. Cependant, l’un des problèmes liés à leur utilisation est la détection et l’identification des espèces de bifidobactéries qui survivent dans les produits alimentaires. Cette étude a pour but de développer un anticorps monoclonal permettant de détecter spécifiquement les bifidobactéries vivantes dans les aliments. Dans une étude antérieure on a isolé localement des souches de bifidobactéries et clairement démontré le pouvoir immunogène de la paroi de bifidobactéries. Lors de la présente étude, des protéines de paroi ont été utilisées comme antigènes de surface pour produire des anticorps monoclonaux contre les bactéries du genre Bifidobacterium . Les protéines ont été extraites et purifiées à partir de six espèces différentes de bifidobactérie ( B. animalis, B. breve, B. longum , B. infantis , B. bifidum et B. pseudolongum) cultivées sur milieu MRS en anaérobiose. L’analyse par SDS-PAGE des extraits protéiques obtenus a révélé des différences au niveau du profil protéique des parois isolées. Cependant, des bandes similaires (58 et 34 kDa) indiquant probablement des protéines communes aux espèces étudiées ont été identifiées. Testées pour immunogénicité, les protéines pariétales issues de B. bifidum et B. longum ont généré une plus forte production d’immunoglobulines spécifiques (titre en anticorps de 53333) chez la souris Balb/C. Des anticorps monoclonaux spécifiques aux bifidobactéries ont été ensuite produits par fusion cellulaire. Les clones sélectionnés pour leur forte production d’IgG (anti- B. longum ) ont montré une réactivité croisée avec l’ensemble des espèces de bifidobactérie indiquant une forte homologie entre les différentes espèces. L’antigénicité partagée par les bifidobactéries a été confirmée par le western-blot révélant un épitope (segment protéique) supporté par une protéine commune (58 kDa) et reconnu par l’anticorps anti- B. longum . En outre, la spécificité de cet anticorps a été déterminée en le testant sur d’autres bactéries ( Propionibacterium freudenreichii, Pediococcus acidilacticii, Lactococcus lactis subsp. lactis, Enterococcus faecium et Listeria innocua) répondant négativement au test . A l’exception de Lactobacillus rhamnosus qui a donné un signal relativement élevé indiquant une similitude antigénique avec les bifidobactéries. Par ailleurs, une observation au microscope électronique après un traitement immunochimique des bifidobactéries a montré clairement l’interaction anticorps-antigène bactérien. La sensibilité de l’anticorps produit a été estimée à 10 (5) cfu/ml. L’anticorps ainsi développé a montré son efficacité dans la détection des bifidobactéries vivantes par des tests d’immuno-culture et laisse suggérer une possibilité de son utilisation dans la quantification de ces bactéries dans divers aliments et matrices.

Mots clés : Anticorps monoclonaux, Bifidobactéries, Immuno-culture, Protéines pariétales, Quantification.

An immuno-culture method has been developed by combination of specific monoclonal antibodies and plate culture to allow detection of viable bifidobacteria. Cell wall proteins were selected as surface antigen to produce antibodies against bifidobacteria. The cell wall proteins were extracted and purified from six ATCC strains of bifidobacteria grown in MRS broth using an anaerobic system. To compare the profile of the protein extracts, all the protein solutions obtained were analyzed by SDS-PAGE. Similar bands corresponding to the major proteins of each species of bifidobacteria were observed. The proteins were tested for their immunogenicity in Balb/c mice after immunization and subsequent analysis using ELISA procedures. Profound immune responses were generated in mice immunized by proteins from Bifidobacterium bifidum and Bifidobacterium longum . Monoclonal antibodies were produced against B. longum and tested for their specificity, sensitivity and cross reactivity with other bifidobacteria species. All the hybridoma cells selected produced anti- B. longum antibodies cross reacting with native and purified proteins from five other bifidobacteria species. An epitope supported by a cross-reacting protein of 58 kDa shared by bifidobacteria was revealed by western blot. This was confirmed by immune-transmission electron microscopy observations which showed the specific interaction of these antibodies with bifidobacterial cell wall proteins. Also, the antibody obtained was found to be specific for the genus Bifidobacterium and sensitive, allowing the detection of at least 105 target cells/ml. An immuno-culture detection approach was then developed using the selected anti- B. longum antibodies. This method was shown to be very efficient for the detection of viable cells of bifidobacteria suggesting the possibility of its use to quantify this bacteria in various food products and matrices.

Key words: Bifidobacteria, cell wall proteins, monoclonal antibodies, immuno-culture, quantification.

Bifidobacteria were reported to be the main genus in the gut bacterial population of infants (Ouwehand et al., 2004). These bacteria are considered essential for maintaining a healthy equilibrium between the population of beneficial and potentially harmful micro-organisms in the gastrointestinal tract (Fuller, 1989, Postnikova et al., 2004). The health and nutritional benefits attributed to bifidobacteria include maintenance of a healthy intestinal microflora, improvement of lactose digestibility and tolerance, antitumor activity, reduction of serum cholesterol levels, synthesis of vitamins, and increasing immunity in host animals (Noda et al., 1994; Jiang et al., 1996; Pereira and Gibson, 2002; You et al., 2004; Vinderola et al., 2004).

Therefore, there is an increasing interest in using these bacteria as probiotics, either in fermented dairy products or formulated as tablets. Indeed, several bifidobacteria-containing products such as yogurt, fermented milk and health foods are currently sold in the world particularly in USA, Japan and Europe (Mercenier et al., 2002). Unfortunately, numerous probiotic products are incorrectly labeled and yielded low bacterial counts, which may decrease their probiotic potential (Drouault et al., 1999; Bunthof et al., 2001). Accurate identification of bifidobacterial strains in products must then be implemented so that consumers can be reliably informed of the content of probiotic products (Tannock, 1999). The use of a microscopic technique is a more direct approach; however, this approach does not allow a differentiation of live and dead bacteria. Auty et al. (2001) assessed the viability of the human probiotic strains Lactobacillus paracasei NFBC 338 and Bifidobacterium sp. strain UCC 35612 in reconstituted skim milk by confocal scanning laser microscopy using the LIVE/DEAD BacLight viability stain. However, the viability staining was limited by environmental factors such as pH, ionic profile, and water activity. Currently, the approach used to analyze probiotic products is still based on culture-dependent methods using specific isolation media and allowing identification of a limited number of isolates. These methods are relatively insensitive, laborious, and time-consuming. The detection or enumeration of surviving bifidobacteria in some products such as yogurts, cheese, butter, and tissues has been most difficult because of lack of a truly selective medium which can discriminate between the genus Bifidobacterium and other genera (Lim et al., 1995; Roy, 2001).

However, in the last few years numerous culture-independent approaches based on molecular methods have emerged for the identification and a rapid detection of bifidobacteria in mixed probiotic cultures (Matsuki et al., 2004; Takada et al., 2004). This approach involved extraction of total bacterial DNA directly from the product, polymerase chain reaction (PCR) amplification of 16S ribosomal DNA, and separation of the amplicons on a denaturing gradient gel. These methods allowed direct identification of the amplicons at the species level (Temmerman et al., 2003, Fasoli et al., 2003). For example, culture independent fluorescent in situ hybridization (FISH) technique, and PCR were recently used to analyze the development of intestinal flora in infants, and bifidobacteria in human feces. These methods offer several advantages over cultural methods in terms of sensitivity and specificity (Kok et al., 1996; Harmsen et al., 2000; Takada et al., 2004). However, the molecular methods reported are not able to distinguish between dead and live bifidobacteria in probiotic products.

Comprehensive studies should be initiated to develop new methods able to detect and quantify viable bifidobacteria in fermented products. The development of antibodies against a specific cell compound could provide an attractive and reliable method for detecting micro-organisms in biological samples (Porsch-Ozcurumez et al., 2004). Bacterial-detection immunoassays (ELISA, immunobloting, etc...) and immunomagnetic separation offer several advantages compared to conventional cultural methods: easy to use, inexpensive, specific to the target host, and rapid (Varadaraj, 1993).

However, newly prepared antibodies are not tested for their reaction towards live cells to determine their affinity to viable bacteria (Brovko et al., 2004). A method based on a combination of semiselective cultivation and colony immunoblotting techniques allowing simple and quantitative detection of Bifidobacterium animalis strains in human feces has been developed by Duez et al. (2000) using rabbit polyclonal antibodies. However, the polyclonal antibodies produced were specific to strains belonging to B. animalis, thus it can not be used to detect other bifidobacteria i.e. the species or strains most frequently found in commercial fermented products. Specific monoclonal antibodies allowing the detection of viable bifidobacterial species and strains in biological media and food products are not available yet.

The overall aim of this study was to produce and characterize different monoclonal antibodies against different species of bifidobacteria and to evaluate their effectiveness in developing a new immuno-culture approach allowing the detection and quantification of various viable bifidobacteria.

The cells were harvested from MRS broth by centrifugation at 5,500 g for 10 min and washed in 100 ml of phosphate buffered saline (PBS, 0.01M, pH 7.4) to be finally re-suspended in 150 ml deionised water. The cells were disturbed mechanically by dynamic high pressure using High-Pressure Homogenizer (Avestin, Ottawa, ON, Canada). The suspension was treated five times at 200 Megapascal (MPa) to obtain a maximal cellular lysis. The suspension was centrifuged three times at 30,000 g for 30 min. The supernatant was removed and pellets washed with PBS between centrifugations. The pellets containing cell wall were suspended together in water at the final volume of 1 ml. Sodium dodecyl sulfate (SDS) was added to the samples at final concentration 20% (wt/vol) and they were heated at 100°C for 10 min to precipitate other portions of cellular wall away from the proteins. The proteins were separated by centrifugation at 15,800 g for 25 min. The residual SDS was removed by maintaining the protein solution overnight at 4°C and then, centrifuged at 15800 g for 25 min.

The protein concentration of all samples was determined by Lowry method (1951) using bovine serum albumin (BSA) (Sigma, MO, USA) as standard. The cell wall protein profile of each bifidobacterial strain was obtained by 10% polyacrylamide gel electrophoresis (PAGE) performed in the presence of sodium dodecyl sulfate (SDS) according to Laemmli (1970). Molecular markers (Bio-Rad Laboratories, Hercule CA, USA) with molecular mass ranging from 21.5 to 97.5 kDa were used to compare the molecular weight of cell wall proteins.

Six to eight week-old male Balb/c (Charles Rivers, St. Constant, Quebec, Canada) mice were immunized with cell wall proteins. In brief, 100 µl of protein solution (1 mg/ml) were emulsified in an equal volume of complete Freund adjuvant (Eskenasy, 1968) for the first injection and incomplete Freund adjuvant for the subsequent injections. The immunization protocol consisted of four subcutaneous injections of 100 µl emulsion (protein plus adjuvant) at three-week interval. Serum samples were obtained just before the first immunization by retro-orbital sinus method to be used as negative control in specific antibody evaluation. Serum samples were also obtained ten days after each antigen injection and subsequently analyzed separately.

All sera were analyzed by Enzyme-Linked Immunosorbent Assay (ELISA) procedures to determine specific antibody titer. Briefly, wells of flat-bottomed, polystyrene plates (Immulon II, Dynatech Laboratories, Alexandria, VA, USA) were coated overnight at 4°C with a solution (100 µl/well) containing cell wall proteins at optimal concentration of 1 µg/ml in 0.05 M sodium carbonate buffer (pH 9.6). Microplates were washed twice with 200 µl Tris-buffered saline (TBS 1X, pH 7.5) - 0.1% (vol/vol) Tween 20 with an automated ELISA plate washer (Dynex, Chantilly, VA, USA) and then blocked for 1 h at 25°C with PBS – 1% (vol/vol) blocking reagent (Roche Diagnostic GmbH, Mannheim, Germany). All serum samples were initially diluted 1/100 in PBS – containing 0.5% (vol/vol) blocking reagent. If positive, sera were tested at higher dilutions: 1/1000, 1/2000, 1/4000, 1/8000, 1/16000, 1/32000, 1/64000, 1/128000 as needed to determine the antibody titer. After four washes with TBS–0.1% Tween 20, 100 µl of sample were added in duplicate in wells and incubated at 37°C for 2 h. Wells were then washed and horseradish peroxidase-conjugated anti-mouse IgG and IgM (KPL, Maryland, USA) diluted to 1:1000 in PBS – 0.5% BR were added to each well (100 µl/well). Wells incubated with sera collected before the first antigen injection were used as negative control. The wells were washed six times with TBS- 0.1% Tween 20 and added with 100 µl of orthophenylenediamine solution (Sigma, Saint Louis, MO) at 0.4 mg/ml of 0.05 M phosphate-citrate buffer (pH 5.0) as substrate. The absorbance was measured at 450 nm on an ELISA plate reader (Molecular Devices, Sunnyvale, CA, USA).

Prior to cell fusion, mouse with the highest antibody titer was boosted with 50 µg of cell wall proteins from B. longum ATCC 15708 without adjuvant for three days successively . The mouse was then sacrificed by cervical dislocation and splenectomised. Spleens were removed aseptically from euthanized Balb/C mice and single cell suspensions were prepared by mechanically disrupting the tissues through a sterile nylon cloth sieve (100 µm pore size) into Iscove Modified Dulbecco’s media (IMDM) containing penicillin/streptomycin (1 % vol/vol). The splenocytes were fused with SP2/O-Ag14 mouse myeloma cells using polyethylene glycol. The fused cells were suspended in IMDM added with 20% (vol/vol) fetal calf serum, antibiotics and hypoxanthine-aminopterin-thymidine (HAT). The suspension was dispensed (200 µl/well) into 96-well microplates (Costar, Cambridge, MA, USA). The microplates were incubated at 37°C in the presence of 4.5% CO2 for 12 days. In order to screen cultures for antibodies production, 100 µl of culture supernatant were added to wells of 96-well Immulon II microplates previously coated with cell wall proteins. The supernatants were also tested on bacterial cells to evaluate the ability of antibodies to detect bifidobacteria. Briefly, suspension of each bifidobacterial strain was prepared in PBS (O.D 650 nm = 0.1) and added (100 µl) to wells and centrifuged at 2800 rpm for 5 min. The cells (pellet) were then immobilized by glutaraldehyde (0.25%, vol/vol) for 15 min at room temperature. The plates were washed and blocked with PBS - 3% (vol/vol) BSA for 1 hour at 37°C. The ELISA test was completed as described above. Hybridomas from cultures showing significant antibody production were selected and cloned by limiting dilution culture.

Cells from 18 h MRS culture of B. longum ATCC 15708 were washed and encapsulated in 3% (wt/vol) agarose. Cubes of agarose (0.8 to 1.0 mm) were fixed overnight at 4°C in 4% (wt/vol) paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.2). After being washed four more times with sodium cacodylate buffer, the samples were dehydrated in a graded ethanol series, embedded in Quetol (Marivac, Halifax, NS, Canada), and polymerized at 55°C overnight (Abad et al. 1987). Ultrathin sections (0.1 µm) of samples were cut with an ultramicrotome (Reichert-Jung, Vienna, Austria) and collected on Formvar-coated nickel grids (JBEM, Dorval, PQ, Canada). For the immunological reaction, the grids were incubated at 37°C for 1 h in 0.3% (wt/vol) BSA and washed with PBS. The grids were then incubated at 37°C for 1 h with Mab anti- B. longum (5 µg protein /ml). The grids were then washed five times (10 min each) in PBS. Gold labeling was carried out by incubating the grids for 15 min at room temperature with protein A-colloidal gold (10 nm in diameter; Sigma) diluted 1/10 in PBS containing 0.2% (wt/vol) polyethylene glycol 8000 (Sigma). The grids were washed again six times, dried, stained with uranyl acetate and lead citrate, and examined with a JEOL 1200 EX transmission electron microscope (Tokyo, Japan) at 80 kV. Eight to twelve fields from five to seven ultrathin sections resulting from four grids were examined, and photographs were taken through the observed grids.

To develop an Mab able to be used as molecular tool to detect specifically bifidobacteria species in foods and environment, the cell wall proteins of bifidobacteria were selected as antigens for the present study. These cell components are more advantageous because they should be detected on surface of whole cells. Also, the cell wall proteins should have several amino-acid sequences common to the bifidobacterial strains. However, the composition and the conformation of these proteins are not known. These antigens were purified, analyzed and tested for their immunogenicity by immunization of mice. The profiles of cell wall proteins from six species of bifidobacteria used in this study are represented in Figure 4.1. Major proteins with similar molecular weight around 58, 48, 37 and 34 kDa were visualized after coomassie blue coloration in each strain. The similar bands indicate the probable common proteins shared by the species of bifidobacteria studied.

However, other bands showing difference between bifidobacteria were also observed. Indeed, the proteins from B. animalis ATCC 27536, B. breve ATCC 15700 and B. longum ATCC 15708 show some difference with those from B. infantis ATCC 15697, B. bifidum ATCC 15696 and B. pseudolongum ATCC 25526. B. animalis ATCC 27536, B. breve ATCC 15700 and B. longum ATCC 15708 show similar profiles, with major bands of 59, 48, 37 34 and 25 kDa. Whereas, B. infantis has an electrophoretic profile different from the other with a major band of 83 kDa. Protein profiles of B. bifidum and B. pseudolongum are similar but different from the other species.

No earlier studies have reported characterization of cell wall proteins from bifidobacteria with antibodies. To study the immunogenicity of proteins used for immunization, antibody titers in sera were evaluated by ELISA after the mice received four injections of the same antigen. The antibody titer is defined by the dilution giving twice the signal (OD450 nm) compared to the serum from a non immunized mouse. The variations of polyclonal antibody generation are shown in Figure 4.2. Sera of mice receiving cell wall proteins from bifidobacteria strains showed an antibody titer increasing with protein administration (data not shown). The maximal antibody titers ranging from 6666 to 53333 were obtained after three or four immunizations of mice by proteins.

However, significant differences in antibody responses against the proteins were observed in mice. Indeed, optimal antibody titers were observed with mice immunized by proteins from B. longum ATCC 15708 and B. bifidum ATCC 15696 . In contrast, the cell wall proteins from B. animalis ATCC 27536 generated the lowest antibody production in mice (6666). Our results are not in accordance with those reported by Duez et al. (2000) revealing high immunogenic protein in cell wall of B. animalis strain DN-173 010. They demonstrated that a protein of about 45 kDa was detected by specific rabbit polyclonal antibodies at 1: 64,000 dilution of the serum. The contrasting results on the generated antibodies capacity against proteins tested could depend on several factors including variations in composition and/or structure of proteins administrated, purity of protein and immunization procedure and genetic characteristics of mice. The results obtained are in accordance to those reported by Magnarelli et al. (2002). These authors observed a cross reactivity between human sera and several antigens (membrane protein) from Borrelia burgdorferi . Regardless of the antibody levels obtained, the proteins used in our study were able to induce a significant immune response.

To further study the reactivity between polyclonal antibody produced and the native total protein exposed on cell surfaces of bifidobacteria, sera raised against purified proteins were tested on all bifidobacteria used in this study. All the sera obtained in 9 weeks were tested against cell wall proteins and whole cells of six bifidobacteria immobilized by glutaraldehyde (data not shown). The sera showed a high reactivity with bifidobacteria species tested compared to the negative control (non immunized mouse). They gave similar or high signals with purified proteins suggesting the exposure of the proteins on cell surface of bifidobacteria. According to Li and Magee (1993), the antibodies react strongly with purified antigen but weakly with native total protein antigen. In addition, the sera cross reacted with all bifidobacteria used due probably to the common proteins mentioned above.

As mice immunized with proteins from B. longum ATCC 15708 and B. bifidum ATCC 15696 gave the highest antibody levels, they were selected to produce monoclonal antibodies. Twenty two clones selected for their significant antibody production compared to the positive control (sera from immunized mice) were tested for their cross reactivity with the cell wall protein of bifidobacteria. Interestingly, all supernatants tested reacted with the proteins from the six bifidobacteria species. These results confirm the cross reactivity observed previously with polyclonal antibodies. In order to produce a Mab that will recognize most members of the bifidobacterial genus, only one clone producing antibody potentially specific for B. longum ATCC 15708 was selected and used in our study. This clone was found to produce an antibody cross-reacting with all bifidobacteria used. Indeed, when tested on bacterial cells immobilized by glutaraldehyde it recognized the bifidobacterial species as well as observed in sera from mice immunized (data not shown).

The isotype of the monoclonal antibody obtained was found to be Ig G2bκ or Ig Mκ. A western blot was done to determine if the cell wall protein detected by SDS-PAGE matched the cross reacting protein found in ELISA assays. Western blotting carried out using cell wall proteins (Figure 4.3) validate the ELISA results mentioned above and revealed one cross reacting protein at a mass of approximately 58 kDa.

Presence of a common epitope in bifodobacterial species indicate antigenic similarities between different species tested. The same results were obtained by the cells of bifidobacteria (data not shown) suggesting the ability of the antibody to bind its targeted protein on the cell wall. A reasonable explanation of results is that the monoclonal antibody produced was raised against an epitope inherent in the common protein (58 kDa) shown by SDS-PAGE analysis. On the other hand, B. longum cells were observed by TEM after immunochemical treatment. Figure 4.4 indicated that the monoclonal antibody is fixed on cells of bifidobacteria compared to the negative control.

In order to determine the specificity of the Mab, the reactivity of the antibody was tested comparatively against bifidobacteria species and other bacteria : Propionibacterium freudenreichii P36, Pediococcus acidilacticii R1001, Lactobacillus rhamnosus R0011, Lactococcus lactis subsplactis R0058, Enterococcus faecium R0026 and Listeria innocua ( strain number). The comparison of the reactivity of Mab between bifidobacteria and other bacteria immobilized by glutaraldehyde is given in Figure 4.5 a,b. The results show a distinction between bifidobacteria and other bacteria according to signals obtained by cells immobilized (OD650 nm = 0.1) and Mab (0.5 µg/ml). Compared to bifidobacteria, Listeria innocua like Propionibacterium freudenreichii P36, Pediococcus acidilacticii R1001, Lactococcus lactis subsp lactis R0058, Enterococcus faecium R0026 shown insignificant signal.

The bifidobacterial species showed high signals compared to other bacterial species; this result indicate revealed the specificity of the antibody produced to bifidobacteria tested. However, Lactobacillus rhamnosus R0011 was found to give signals nearly similar to those shown by the bifidobacteria suggesting antigenic similarities between the two bacterial genera. According to Orla-Jensen (1924), bifidobacteria were first grouped in the genus Bacillus and then the genus Bifidobacterium was proposed in the 1920’s. However, there was not a taxonomic consensus for this new genus and it was classified in the genus Lactobacillus , due to their rod-like shapes and obligate fermentative characteristics. The Bifidobacterium genus was characterized by a unique hexose metabolism that occurs via a phosphoketolase (Fructose-6-phosphate phosphoketolase or F6PPK) pathway often termed the ‘bifid shunt’s (Grill et al., 1995). These results suggested that the bacteria not belonging to bifidobacteria genus are not recognized by the antibody anti- B. longum ATCC 15708, probably because of the difference in the characteristics of protein surface. These results agree with those reported by Bolin et al. (1995) demonstrating the high specificity of monoclonal antibodies raised against membrane preparations of Helicobater pylori . To study the effect of the concentration of the Mab on the signal intensity distinguishing between bifidobacteria and other bacterial species, a serial dilution of the Mab ranging from 0.5 to 0.00005 µg/ml was carried out and tested comparatively on bifidobacteria, Lactobacillus rhamnosus and Listeria innocua. According to the results shown in Figure 4.6, the signal decreased significantly when the Mab concentration diminished from 0.5 to 0.05 µg/ml giving an optimal signal for all the bifidobacterial species.

The sensitivity of the Mab obtained was evaluated by studying the effect of the concentration of bacterial cells on the signal. Serial dilution of bacterial suspension from 103 to 107 cfu/ml were prepared from overnight culture and tested by ELISA using an optimal antibody concentration of 0.5 µg/ml. The results shown in Figure 4.7 indicate that the purified Mab anti- B. longum ATCC 15708 is more reactive with the antigen (cells) at concentration of 105 cfu/ml . Auty et al. (2001) mentioned that a minimum detection limit for in situ viability staining in conjunction with confocal scanning laser microscopy enumeration was around 108 bacteria/ml (equivalent to around 107 cfu/ml), based on Bifidobacterium sp. strain UCC 35612 counts .

Immune-TEM observations showed that bacterial cell envelopes were uniformly recovered by protein-A colloidal gold. The cell wall protein appears to be target of the monoclonal antibody produced. Use of the anti- B. longum for direct immunofluorescence labeling of bacteria in fixed food samples and epifluorescence microscopy could represent an alternative method to detect bifidobacterial species in fermented product.

Detection and quantification of bifidobacteria remains a major limitation in the field of probiotic research. Until recently, there is no efficient method allowing a specific detection of viable bifidobacteria even in non complex food or environmental matrices. Moreover there is no efficient selective medium for the specific culture of bifidobacteria. In this study, a powerful system based on the use of a combination of a specific Mab and a culture medium was developed for the specific detection and quantification of viable bifidobacteria. Using the developed IC detection method and as reported in Figure 4.8, panel A1, bifidobacteria colonies were detected only when the anti-bifidobacteria Mab was added indicating that these colonies from bifidobacterial cells were being captured by the antibody. There was no bacterial colonies in wells that do not contain specific anti-bifodobacteria Mabs (Figure 4.8, panel A2). The specificity of the IC detection system was further confirmed by using three different concentrations of bifidobacteria. As shown in Figure 4.8, panel B the number of colonies detected in the microplate wells is proportional to the initial number of bacteria added.

The IC detection system offers several advantages over the traditional culture methods. Even though a similar sensitivity was obtained, higher specificity was observed with IC detection system because of the addition of the anti-bifidobacteria Mab as a capture probe prior to the addition of MRS-cysteine. Moreover, this method was simple to perform and no special equipment (anaerobic jar containing carbon dioxide generating sachet) was needed to create such as an anaerobic conditions for the growth of bifidobacteria. Only viable bifidobacteria are still detected.

This study was performed to produce specific monoclonal antibody able to detect live bifidobacteria using proteins present on the bacterial cell surface. Cell wall of bifidobacteria was found to have some similar proteins with molecular size ranging from 58 to 34 kDa. Furthermore, proteins extracted and purified from bifidobacteria are shown to be immunogenic in Balb/c mice. Monoclonal antibody was then produced against cell wall protein of B. longum ATCC 15708 and found to be more immunogenic . Interestingly, the antibody produced showed a cross reactivity with purified cell wall proteins and whole cells from other bifidobacterial species studied ( B. breve ATCC 15700 , B. infantis ATCC 15697, B. bifidum ATCC 15696 B. pseudolongum ATCC 25526 and B. animalis ATCC 27536) sharing antigenicity with B. longum ATCC 15708. The antigenicity shared by bifidobacteria was confirmed by western blot revealing an epitope supported by common protein of 58 kDa recognized by the monoclonal antibody obtained. The antibody produced was also shown to be specific to members of bifidobacterial genus. The sensitivity of the antibody produced was estimated at 105 cfu/ml. Immuno-culture test developed with best detection of viable cells of bifidobacteria using specific monoclonal antibody would be a good tool to detect and quantify viable species of human and animal origin bifidobacteria in bifidus food and tissues.

ACKNOWLEDGMENTS

This work was supported by grants from FCAR, NOVALAIT, MAPAQ, NSERC. T. Amrouche was recipient of fellowship from World Bank. Stephen Davids is thanked for reading the manuscript.

© Tahar Amrouche, 2005