Characterization of bifidobacteria by random DNA amplification

International Journal of Food Microbiology 23 (1994) 55-70

ELSEVIER

International Journal of Food Microbiology

Characterization of dairy-related Bifi'dobacterium spp. based on their fl-galactosidase electrophoretic patterns D e n i s R o y a,., J e a n - L u c B e r g e r b, G e r h a r d

Reuter

b

a Food Research and Development Centre, Agriculture Canada, 3600 Casavant Boulevard West, Saint-Hyacinthe, Que. J2S 8E3, Canada b Institute of Food and Meat Hygiene and Technology, Free University of Berlin, Brummerstr. 10, D-IO00 Berlin 33, Germany

Received 1 November 1993; revision received 23 December 1993; accepted 15 March 1994

Abstract

Numerical analysis of phenotypic characteristics based on enzymatic activity and carbohydrate fermentation allowed the discrimination of most strains of bifidobacteria of animal origin from those of human origin. Strains of bifidobacteria studied were separated into nine groups based on numerical analysis. Three groups contained most strains of animal origin, three groups comprised both strains of animal and human origin, and three groups were strictly composed of strains of human origin. The results indicate that one group of animal origin (group II) contained all reference strains of Bifidobacterium animalis and 10 strains isolated from fermented milks or commercial preparations. Although numerical analysis of enzymatic activities and carbohydrate fermentation patterns allowed the differentiation of 'wild' strains of B. animalis and B. longum isolated from commercial preparations, this method failed to confirm the species. In the present study, the determination of electrophoretic patterns of fl-galactosidases resulted in the development of a new technique for the differentiation of Bifidobacterium species. Several isoenzymes of fl-galactosidase were detected among strains of bifidobacteria. Each species had a specific electrophoretic pattern. The detection of fl-galactosidase by electrophoresis is a new tool for distinguishing between dairy- and non-dairy-related bifidobacteria. Dairy-related bifidobacteria (B. bifidum, B. breve, B. infantis and B. longum) as well as B. animalis could be better differentiated from other bifidobacteria by comparison of their /3-galactosidase electrophoretic patterns, rather than by numerical analysis of their phenotypic characteristics. Keywords: Bifidobacterium; fl-Galactosidase; Dairy industry; Electrophoresis

* Corresponding author. 0168-1605/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0168-1605(94)00026-3

56

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

1. Introduction

Bifidobacteria are of increasing interest to the dairy industry. The incorporation of new bacterial cultures such as bifidobacteria into the human diet corresponds to the emergence of a new generation of fermented dairy products, which use the beneficial effect of bacteria of human origin on intestinal metabolism. Of the 29 known Bifidobacterium species, B. longum, B. bifidurn, B. breL,e and B. infantis of human origin are of industrial interest. Reuter (1990) mentioned that B. anirnalis is also used for fermented milk products. However, Biavati et al. (1992) noted that the bifidobacterial strains used in the preparation of milk products shou|d be from the human habitat. These authors concluded that their identification is, therefore, essential. Carbohydrate fermentation patterns and enzymatic activity profiles are useful for routine characterization and identification of bifidobacteria (Roy and Ward, 1992; Yaeshima et al., 1992). However, the phenotypic characteristics of dairy-related bifidobacteria such as B. longum and B. animalis are often similar. Additional methods that correlate best with genotype must be developed for distinguishing between these species. In addition to D N A / D N A hybridization tests, enzymatic systems and soluble protein profiles have been considered to determine, at the macro-molecular level, the genetic relatedness between Bifidobacterium species. Sgorbati and Scardovi (1974) studied the electrophoretic behaviour of transaldolase, transketolase and 6-phosphogluconate dehydrogenase (6PDG) in order to find correlations between enzymatic activities of the different species of bifidobacteria and their known DNA homology relationships. Several isoenzymes were detected, but no clear correlation with the habitats was found. Each strain tested exhibited only one band for either transaldolase or 6PGD (Scardovi et al., 1979; Sgorbati and Scardovi, 1974). On the other hand, Biavati et al. (1982) observed that there was excellent correlation between the 25 electrophoretic patterns of soluble cellular proteins and 24 DNAD N A homology groups in the genus Bifidobacterium. However, Gavini et al. (1991) noted that the results based on electrophoretic patterns of soluble proteins or transaldolase and 6PGD did not correlate well with phenotypic or genomic descriptions of the species. In the present study, the /3-galactosidases produced by 65 strains of Bifidobacterium spp. were examined to improve the phenotypic characterization of these organisms. Final identification of 16 strains of bifidobacteria isolated from fermented milk or commercial preparation previously distinguished by numerical analysis of their phenotypic characteristics was confirmed by comparison of their /3-galactosidase electrophoretic patterns with those of reference strains of human and animal origin. 2. Materials and methods

Strains and cultivation. The species of bifidobacteria used in this study are listed in Table 1. These strains were freeze-dried in skim milk (20% w / w ) and sucrose (5%

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70 Table 1 Strains used in this study Group

Reference

Name as received

Culture collection or other a

Isolated from

Ungrouped

Bar

B. asteroides

ATCC 25910

Hindgut of honeybee

I

Bcy Bbm Bth

B. coryneforme B. boum B. thermophilum

ATCC 25911 ATTC 27917 ATCC 25525

Hindgut of honeybee Bovine rumen Swine feces

II

Anl An2 An4 An5 An6 An7 Bip BiF BiV R56 RW5 H6 B15 B17 B18 B19 B22 Bch Bpu BpL

B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. infantis B. longum B. infantis B. longum B. longum B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. choerinum B. pullorum B. pseudolongum

ATCC 25527 ATCC 27536 ATCC 27672 ATCC 27673 ATCC 27674 DSM 20104 ATCC 27920P FRDC FRDC FRDC FRDC G. Reuter G. Reuter G. Reuter G. Reuter G. Reuter G. Reuter ATCC 27686 ATCC 27685 ATCC 25526

Rat feces Chicken feces Rat feces

III

Bci Bgb Bmg

B. cuniculi B. globosum B. magnum

ATCC 27916 ATCC 25865 ATCC 27540

Rabbit feces Bovine rumen Rabbit feces

IV

Bil Bi4 Bi3 Bi2 Bid

B. B. B. B. B.

ATCC ATCC ATCC ATCC ATCC

15697 25962 17930 15702 25912

Infant intestine Infant intestine

V

Ba2 Ba3 Ba4 Ba5 Bcl Bpl Bgl

B. adolescentis B. adolescentis B. adolescentis B. adolescentis B. catenulatum B. pseudocatenulatum B. gallinarum

ATCC ATCC ATCC ATCC ATCC ATCC ATCC

15703 15704 15705 15706 27539 27919 33777

Adult intestine Adult intestine Adult intestine Adult intestine Adult intestine Infant feces Chicken cecum

VI

BI1 BI2 BIV BIN R46 R69 Bi5 Bnl Bss

B. longum B. longum B. longum B. longum B. longum B. longum B. infantis B. angulatum B. suis

ATCC ATCC FRDC FRDC FRDC FRDC ATCC ATCC ATCC

15707 15708

Adult intestine Infant intestine Commercial preparation Commercial preparation Commercial preparation Commercial preparation Large colonies Human feces Pig feces

infantis infantis infantis infantis indicum

27920G 27535 27533

Sewage Rabbit feces small colonies Fermented milk Commercial preparation Fermented milk Commercial preparation Fermented milk Fermented milk Fermented milk Fermented milk Fermented milk Fermented milk Pig feces Chicken feces Pig feces

Infant intestine Hindgut of honeybee

57

58

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55 70

Table 1 (continued) Group

Reference

Name as received

Culture collection

Isolated from

or other " VII

Brl Br3 Br4 s17 R70

B. B. B. B. B.

breve brece brece brece brece

ATCC 15698 ATCC 15700 ATCC 15701 G. Reuter FRDC

Infant intestine Infant intestine Infant intestine Infant intestine Commercial preparation

VIII

Bdl Bd2 Bd3 Bd4 Bd5 Bd6

B. B. B. B. B. B.

dentium dentium dentium dentium dentium dentium

ATCC ATCC ATCC ATCC ATCC ATCC

Lung abcess in adult man Pleural fluid from adult Dental caries Human feces Human vagina Human dental caries

Ungrouped

Bmi

B. minimum

ATCC 27538

Sewage

IX

Bb2 Bb3 Bb4 s28 R75

B. B. B. B. B.

ATCC 11863 ATCC 15696 ATCC 29521 G. Reuter FRDC

Human Infant intestine lnfant feces Infant intestine

bifidum bifidum bifi'dum bifidum bifidum

15423 15424 27534 27678 27679 27680

Commercial preparation

a ATCC, American Type Culture Collection, Rockville, MD; G. Reuter, Freie Universit~it Berlin, FRG; FRDC, Food Research and Development Centre, Agriculture Canada, St. Hyacinthe Quebec, Canada.

w / w ) . Lactobacilli MRS broth (Institut Rosell Inc., Montreal, Canada) supplemented with 0.05% L-cysteine-HCl, 0.1% Tween 80 and 2.0% glucose (filter-sterilized) was used to rehydrate the freeze-dried microorganisms and recovered strains were subcultured twice. Active cultures were incubated for 15 h at 37°C in an anaerobic chamber (Anaerobic system, Forma Scientific, Marietta, OH, USA) with 5% CO 2, 10% H 2 and 85% N 2 gas atmosphere.

Biochemical tests'. Enzymatic activity profiles and carbohydrate fermentation experiments were determined according to Roy and Ward (1992). The presence of fructose-6-phosphate phosphoketolase was detected according to Chevalier et al. (1991). The results of 20 enzymatic tests and 25 carbohydrate tests were treated as follows: a positive result was scored 1 and a negative result 0. The data were analysed using the simple matching coefficient (Ssm) of Sokal and Michener (1958). Strains were grouped at a 84% similarity level by unweighted pair-group average linkage analysis. All computations were performed on a Digital Equipment Corporation V A X computer using the IMSL package (IMSL Math/Library- 1MSL Inc, Texas, USA, 1982, Vol. 2, Chap. III).

Preparation of cell-free extract. Each strain of bifidobacteria tested was grown in 50 ml of MRS medium (125-ml Erlenmeyer flasks) supplemented with 2.0% lactose instead of glucose (filter-sterilized), except for lactose-negative strains. Cells were

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

59

harvested at the end of the exponential phase of growth (15 h), centrifuged at 10 000 g for 15 rain at 4°C and washed twice with McIlvaine buffer (Citrate-phosphate buffer, 100 mM, pH 6.0; Gomori, 1955). The pellet was resuspended in 3.0 ml of the same buffer. This mixture (0.75 ml) was mixed with 0.5 ml of glass beads (0.25-0.50 mm diameter) and poured into a 1.5-ml Eppendorf tube. After 25 rain treatment at 4°C on a vortex adaptator for microfuge tubes (Micro-tube inserts, Scientific Industries Inc., Bohemia, NY, USA) at 100% intensity, crude extract was separated from the beads by centrifugation using a 15-ml filtration column with a porous bottom filter (Poly-Prep column containing a porous polymer bed support, Cat. No. 731-1550, Bio-Rad Laboratories Ltd., Mississauga, ON, Canada). The filtrate was centrifuged at 16000 g at 4°C for 20 min to remove cell debris. This cell-free extract was assayed for protein and enzyme activity and frozen at -40° C for subsequent gel electrophoresis analysis. Gel electrophoresis. Native polyacrylamide gel electrophoresis (Native-PAGE) was carried out at pH 7.0 in a 7.5% polyacrylamide slab gel (16 × 20 cm) (Hames, 1987). A Bio-Rad Protean-II Slab Cell was used as the electrophoresis system (Bio-Rad Laboratories Ltd., Mississauga, ON, Canada). Samples (30/xl) of crude enzyme extracts (approximately 3-8 mg of protein per ml) containing 10/zl of 20% glycerol and 0.002% bromophenol blue as migration reference were applied to stacking gel wells. Gel staining. Following electrophoresis (100 V/cm2), the gel was stained for the detection of /3-galactosidase activity according to Bhowmik and Marth (1989). /3-Naphthyl-/3-D-galactopyranoside was used as substrate with fast garnet GBC (Sigma Chemicals Co., St-Louis, MO, USA) as the dye coupler in McIlvaine buffer pH 6.0. Gel analysis. /3-Galactosidases from Escherichia coli (No. cat 5531LA, Gibco BRL Life Technologies, Inc., Burlington, ON, Canada) and B. bifidum var. pennsylvanicus ATCC 11863 (crude extract) were chosen as internal standards. The relative front of migration value (Rf) was determined for each enzyme band as follow: Rf = ( ( X - A ) / C ) x 100. Internal standard (C) was determined by C = migration distance of/~-galactosidase of B. bifidum (B) - migration distance of /3-galactosidase of E. coli (A) (C = B-A). The migration distance of the enzyme band of interest was denoted as X. Distance was expressed in cm. Calculations were done for each gel electrophoresis performed. Enzyme assay, fl-Galactosidase was determined by the release of O-nitrophenol from the substrate o-nitrophenyl-/3-D-galactopyranoside (ONPG) (Sigma Chemicals Co., St-Louis, MO, USA). The ONPG was dissolved in McIlvaine buffer pH 6.0 at a concentration of 10 mM. Substrate (1.0 ml) was preincubated at 37°C and the reaction was initiated by addition of 100 pA enzyme solution. The reaction was stopped by addition of 1 ml cold NazCO 3 (0.5M). The absorbance of the mixture was measured at 420 nm and the amount of O-nitrophenol (ONP) was determined

60

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

from a standard curve. One unit of/3-galactosidase activity is defined as 1 nmole of O N P liberated from O N P G per min under the conditions described. Protein assay. Protein content was determined using a modified Biuret reaction. The BCA (bicinchoninic acid) protein reagent supplied with the system (Pierce Chemical Co., Illinois, USA) was used with bovine serum albumin as standard (Smith et al., 1985). Chemicals. All chemicals used were of analytical reagent grade.

3. Results

Fig. 1 shows that strains of bifidobacteria studied were separated into nine groups at an 84% similarity level based on numerical analysis of phenotypic characters. Three groups contained most strains of animal origin; three groups

Similarity Level S

i[~:.i( ~

Groups 100

~~

VI VIIi

:~ s , ~ m f , T t , m

IX

Fig. 1. Dendrogram obtained by numerical analysis of enzyme profiles and carbohydrate fermentation patterns of 65 Bifidobacterium strains using the Ssm by the unweighted average pair-group linkage analysis.

D. Roy et al. / InternationalJournalof FoodMicrobiology23 (1994)55- 70

61

comprised both strains of animal and human origin; and three groups were strictly composed of strains of human origin (Table 1). Only Bifidobacterium asteroides and B. minimum were not included into any group (Fig. 1). Among groups containing strains of animal origin, group II included all reference strains of B. animalis and 10 'wild' strains isolated from fermented milk or commercial preparations (Table 1). Only three reference strains of animal original were included in groups of human origin: B. indicum (group IV), B. gallinarum (group V) and B. suis (group VI). Group IV included four reference strains of B. infantis. All reference strains of B. adolescentis were included into group V which also contained type strains of B. catenulatum and B. pseudocatenulatum. Group VI was composed of reference strains of B. longum plus four strains isolated from commercial preparations. B. infantis ATCC 27920G, B. suis and B. angulatum were also placed into Group VI as subgroups (melezitose-negative). Reference strains of B. breLre (group VII), B. dentium (group VIII) and B. bifidum (group IX) were separated into well-defined groups. Two 'wild' strains isolated from commercial preparations, R70 and R75, were identified as B. brece (group VII) and B. bifidum (group IX), respectively. Table 2 lists the main phenotypic characteristics of the 9 groups of bifidobacteria separated by numerical analysis. Strains included in groups of animal origin (groups I, II and III) did not possess alkaline phosphatase or N-acetyl-fl-glucosaminidase activity. All strains of infant origin (group IV: B. infantis; group VI: B. longum; group VII: B. breve and; group IX: B. bifidum) had N-acetyl-fi-glucosaminidase activity. Most strains of animal origin exhibited cystine aminopeptidase activity. Strains of group I did not ferment lactose although/3-galactosidase activity was detected for all strains, suggesting that these strains could lack a transport system (permease) for lactose. Among dairy-related bifidobacteria, strains of B. animalis (group II) isolated from fermented milks or commercial preparations could be differentiated from strains of B. longum (group VI) by the presence of cystine aminopeptidase and /3-glucosidase activities. Reference strains of B.longum did not possess N-acetyl/3-glucosaminidase activity. This enzymatic activity was not adequate to distinguish B. longum from B. animalis. However, all strains isolated from commercial preparations identified as B. longum possessed N-acetyl-fl-glucosaminidase activity. In addition, all strains of B. longum were melezitose-positive whereas B. infantis ATCC 27920G, B. angulatum and B. suis did not ferment melezitose. Although numerical analysis of enzymatic activities and carbohydrate fermentation patterns allowed the differentiation of 'wild' strains of B. animalis and B. longum isolated from commercial preparations, this method failed to confirm the species. Electrophoretic patterns of /3-galactosidase activity were generated in order to identify species at the macro-molecular level. Table 3 shows the distribution of/3-galactosidase bands as estimated by Rf values. Several isoenzymes were detected in all groups of bifidobacteria, except groups I, III and IX. No fi-galactosidase band was observed for strains belonging to groups I and III (except B. magnum) and ungrouped strains B. asteroides and B. minimum although weak

0 100 95 100 95 0 19 48 33 0 100 100 5 0 100 10 0 5

0 100 33 100 100 0

33 0 0 0 0 100 0 0 100 66 0 33

Group II (n = 20)

% of positive strains

Group I (n = 3)

66 66 33 0 66 0 0 0 0 66 0 0

0 66 100 100 100 0

Group III (n = 3)

0 60 40 60 100 100 0 0 100 0 20 0

20 0 0 100 60 100

Group IV (n = 5)

28 70 42 0 100 100 0 0 100 14 0 56

14 14 14 100 100 0

Group V (n = 7)

100 44 100 11 100 100 0 66 100 22 0 0

55 0 0 100 0 66

Group VI (n = 9)

0 80 0 80 100 100 80 20 100 80 60 20

20 40 0 100 100 100

Group VII (n = 5)

100 100 i00 83 100 100 34 50 100 83 100 100

0 0 17 100 100 17

Group VIII (n = 6)

0 0 0 0 100 0 0 0 0 0 0 0

80 100 80 100 0 100

Group IX (n = 5)

n is the number of strains in the group. The type strains of the following species belongs to the groups: Group 1, B. coryneforme, B. boum and B. thermophilum; Group I1, B. animalis, B. pseudolongum, B. choerinum and B. pullorum; Group III, B. cuniculi, B. globosum and B. magnum; Group IV, B. infantis and B. indicum; Group V, B. adolescentis, B. gallinarum, B. catenulatum and B. pseudocatenulatum; Group VI, B. longum, B. angulatum and B. suis; Group VII, B. bret,e; Group VIII, B. dentium; and Group IX, B. bifidum.

Enzymatic tests Alkaline phosphatase Cystine aminopeptidase Phosphohydrolase 13-Oalactosidase /3-Glucosidase N-Acetyl-13-glucosaminidase Fermentation patterns L-Arabinose Ribose D-Xylose D-Mannose Lactose Melibiose Cellobiose Melezitose Raffinose Glycogen Mannitol Salicin

Characteristics

Table 2 Phenotypic characteristics for the differentiation of the nine groups of bifidobacteria

t~ I

D. Roy et al. / International Journal o f Food Microbiology 23 (1994) 55-70

63

Table 3 The relative front of migration (Rf) values of /3-galactosidase bands and /3-galactosidase activity of strains in the nine groups of bifidobacteria Group

Strain

Ungrouped

Bar Bmi Bbm Bcy

I

Bth

Source

No. isoR~standard-err°r) enzymes a

B. asteroides B. minimum

25910 27538

ND " ND

3

B. bourn B. coryne forme B. thermo philum

27917 25911

ND ND

3 7

25525

ND

9

Activity ~ (U/mg) 0

II

Anl An2 An4 An5 An6 An7 BiF RW5 BiV R56 H6 B15 BI7 B18 B19 B22 Bip Bch Bpu BpL

B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. animalis B. choerinum B. pullorum B. pseudo longum

25527 27536 27672 27673 27674 20104 FRDC FRDC FRDC FRDC Reuter Reuter Reuter Reuter Reuter Reuter 27920P 27686 27685 25526

(3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) (3) ND ND ND

III

Bci Bgb Bmg

B. cuniculi B. globosum B. magnum

27916 25865 27540

ND ND (1)

101.79 ~178)

IV

Bil Bi2 Bi3 Bi4 Bid

B. B. B. B. B.

infantis infantis infantis infantis indicum

15697 15702 17930 25962 25912

(3) (3) (3) (3) (3)

25.64~139);97.71°2°);130.44{167) 34 20.72{°'72);97.71(1"2°);131.19{452) 42 24.17{°~4);100.12{345);131.19 ~4~2) 42 20.72{°72);97.71{L2°);134.65 {464) 40 25.47{°'47),101.85{~s5},138.23 {25~) 52

V

Ba2 Ba3 Ba4 Ba5

B. B. B. B.

adolescentis adolescentis adolescentis adolescentis

15703 15704 15705 15706

(3) (3) (3) (5)

Bcl Bpl

B. catenulatum B. pseudocate nulatum B. gallinarum

27539 27919

(2) (2)

27.35~2°3);114.73{2s5);123.19(289} 2 30.80(226);111.76~277};118.96 ~2s3) 31 20.15~1"~2);101.25075);123.19 {2"89) 12 28.60~24~); 102.53~4-53);112.23~3-4a); 120.69{3°8); 126.25 °.75). 37 8.97(°83);105.86 (8.6`9) 30 7.55{°'~);101.75 (s-78) 3

33777

(1)

24.57 wa3)

Bgl

63.48{°4s);82.81{°63};130.7006°) 66.99(36°);83.63052);136.00 65.58~c~'°5);80.28{°'97};130.70~°'6°) 66.22{215);85.65{294);130.70(°6°) 68.05{2t4);85.65(294);130.70 (0'60) 62.47{1 °2);80.210"~4); 130.70{° 6°) 65.35{~)'61);81.19{°'58);130.70 {°6°) 65.35{°6t);81.19{°58);130.70 {°6°) 63.33{L36);81.50{183);130 65.35{°~1);81.19(°58);130.70{°6°) 65.35{°611;81.19(°58);130.70 {°6°) 65.35~°61);81.19{°581;130.70 {°6°) 65.35{~~'t~;81.19(°58);130.70 {°'6°) 65.35(°6~);81.19(°58);130.70 {°6°) 65.35~°6~;81.19(°58);130.70 {°6°) 65.35c°61);81.19~°58);130.70(°6°) 69.14{°5~);85.98{°96);133.33

8 8 2 5 5 0 13 11 12 3 7 12 8 5 8 12 13 3 17 1 1 0 6

5

64

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

Table 3 (continued)

Group

Strain

V1

BI1 BI2 BIV BIN R46

B. B. B. B. B.

longum h)ngum longum longum longurn

R69 Bi5 Bnl Bss

B. B. B. B.

longum FRDC infantis 27920G angulaturn 27535 suis 27533

Brl

B. brece

15698

(4)

Br3 Br4

B. brete B. brece

15700 15701

(3) (4)

S17 R70

B. brece B. brel,e

Reuter FRDC

(3) (4)

Bdl Bd2

B. dentium B. demium

15423 15424

(2) (4)

Bd3

B. dentium

27534

(5)

Bd4 Bd5 Bd6

B. dentium B. dentium B. dentium

27678 27679 27680

Bb2 Bb3 Bb4 s28 R75

B. B. B. B. B.

11863 15696 29521 Reuter FRDC

VII

VIII

IX

bifidum bifidum bifidum bifidum bifidum

Source

No. isoR}slandarderror) enzymes ~'

Activity b (U/rag)

15707 15708 FRDC FRDC FRDC

(2) (2) (2) (3) (4)

2 5 1 12

(3) (3) (2) (2)

42.67(159);76.61 (H°) 43.01(1'6°1;80.53 (°77) 38.24(2"95);73.53 (2'94) 23.42(z26);38.24(2°5):137.09076) 24.89(2"72);79.63(~s51; 89.11 (141 I;97.22(177) 24.89(272);100.93(°93);148,53 tI'14) 22.29(l°m;94.87(z49~;130.03(3lS~ 125.86(15~-);185.67 (~-'°s) 22.65(°'43);98.15 ~1~5~ 9.49(1 °4);90.47(t22);107.10(z°3); 133.37 ~e'34) 16.15~°Av~;93.30(~'ss~;106.95~zt~ 9.49(1°4);90.47(122); 104.76(I"19); 140.40 (2.47) 12.16(I2~);92.01(192);105.84 (zt°) 5.88;15.69(1s3);93.51~149);106.06(2°2)

11 46 35 7 72 60 5 40 21 29

(3) (3) (3)

23.33tI32);96.61 (°°6) 19.85(165~;66.38~u4);83.29~~'3); 96.40 (°-~s) 13.71(tm~;19.98(~29~;92.37~I~4); 110.67(~146);134.87 (254) 19.85°65~;57.49°'3a);96.40 (~'55) 21.12(1"~9);57.49(~'32);96.40(°'55) 21.12(H')~;55.800's~);95.13 (l'°l)

4

10 25 20 8

(1) (1) (1) (1) (1)

100.86 (?'~6) 100.86 (~'zm 100.86 ~-~6) 100 97.06 (~~4)

21 19 15 3 29

5

Number of /3-galactosidase bands detected. b One unit of ~-galactosidase activity is defined as 1 nmole of ONP liberated from ONPG per min under the conditions described in materials and methods. c not-detected.

/3-galactosidase activity was found in cell extracts (Table 3). These strains were grown in MRS medium supplemented with glucose instead of lactose since they were lactose-negative according to their phenotypic characterization (Table 2). Among strains of group II, all strains identified by numerical analysis as B. anirnalis possessed identical isoenzymes of /3-galactosidase whereas B. choeorinure, B. pullorum and B. pseudolongum did not exhibit any/3-galactosidase band. Fig. 2 shows that dairy-related bifidobacteria ( B. bifidum, B. brec,e, B. infantis and B. longum) and B. animalis could be differentiated from other species of human origin as B. adolescentis (group V) and B. dentium (group VIII) based on their/3-galactosidase electrophoretic patterns. All strains of B. bifidum (group IX)

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

1

2

3

4

5

6

7

8

65

9 10 11 12 13

Fig. 2. /3-Galactosidase electrophoretic patterns of nine Bifidobacterium species of human origin and one of animal origin. Lanes 1, 7 and 13: band a, /3-galactosidase band of Escherichia coli; band b, /3-galactosidase band of Bifidobacterium bifidum ATCC 11863 (crude extract); Lane 2: B. bifidum ATCC 29521; Lane 3: B. breve ATCC 15700; Lane 4: B. infantis ATCC 15697; Lane 5: B. longum ATCC 15707; Lane 6: B. adolescentis ATCC 15703; Lane 8: B. angulatum ATCC 27535; Lane 9: B. catenulatum ATCC 27539; Lane 10: B. pseudocatenulatum ATCC 27919; Lane 11: B. dentium ATCC 27534; Lane 12: B. animalis ATCC 25527.

exhibited a single coloured band of/3-galactosidase (lane 1, band b; lane 2). Strains of B. breve (group VII) possessed at least three isoenzymes (lane 3) (Table 3). Three clearly discernible bands of/3-galactosidase were found for the type strain of B. infantis (lane 4). B. longum ATCC 15707 (lane 5) had a different /3-galactosidase electrophoretic pattern from those of other dairy-related bifidobacteria (from lane 2 to lane 5) and B. animalis (lane 12). In addition, B. longum could be differentiated from B. angulatum (lane 8; Fig. 2). The type strain of B. animalis (lane 12) and other reference strains of group II shared a similar profile of three /3-galactosidase isoenzymes (Table 3). Among strains of group V (Table 3), the type strains of B. catenulatum (lane 9) and B. pseudocatenulatum (lane 10) could be differentiated from the type strain of B. adolescentis (lane 6; Fig. 2). Other strains of B. adolescentis had three coloured bands in common (Table 3). Strains of B. dentium (Table 3; group VIII) could be separated into three subgroups based on their /3-galactosidase electrophoretic patterns (data not shown). In addition, it was clear that a comparison of/3-galactosidase profiles could be used to differentiate strains isolated from the culture collection (Fig. 3). Two strains were isolated from the original culture of B. infantis ATCC 27920 which could be separated on the basis of small and large colonies. The strain 'Bip' (ATCC 27920P) which gave small colonies was grouped with B. animalis strains (Table 1). Comparison of/3-galactosidase electrophoretic patterns (lane 6; Fig. 3) confirmed that this 'Bip' strain belonged to the species B. animalis (lanes 2 and 3).

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

66

1

2

3

4

5

6

7

8

9

10 11 12 13

a

Fig. 3./3-Galactosidase electrophoretic patterns of various Bifidobacterium species of human or animal origin. Lanes 1, 7 and 13: for explanations, see Fig. 2; Lane 2: B. animalis DSM 20104; Lane 3: B. animalis ATCC 25527; Lane 4: B. longurn ATCC 15708; Lane 5: B. longum ATCC 15707; Lane 6: Bip strain (ATCC 27920P); Lane 8: B. infantis ATCC 27920G; Lane 9: B. infantis ATCC 15697; Lane 10: Bifidobacteriurn sp. ATCC 15699; Lane 1 I: B. brace S17 (Reuter sl7c); Lane 12: B. bret,e ATCC 15700.

The strain Bi5 (ATCC 27920G) which gave large colonies was grouped with strains of B. longum (Table 1) based on phenotypic characteristics. However, comparison of electrophoretic patterns revealed that this strain (lane 8; Fig. 3) belonged to B. infantis (lane 9). Finally, comparison of/3-galactosidase profiles also indicated that

a

Fig. 4. /3-Galactosidase electrophoretic patterns of Bifidobacterium animalis and strains isolated from fermented milks and commercial preparations (Group II, Table 3). Lanes 1, 7 and 13: for explanations, see Fig. 2; Lane 2: B. animalis ATCC 25527; Lane 3:H6 (see Table 3); Lane 4: BI5; Lane 5: B17; Lane 6: B18; Lane 8: B19; Lane 9: B22; Lane 10: Biv; Lane 11: R56; Lane 12: BiF.

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

67

Fig. 5. fl-Galactosidase electrophoretic patterns of Bifidobacterium longum, B. infantis and strains isolated from fermented milks and commercial preparations (Groups VI, VII and IX; Table 3). Lanes 1, 7 and 11: for explanations, see Fig. 2; Lane 2: B. longum ATCC 15707; Lane 3: B1V strain (see Table 3); Lane 4:R75 strain; Lane 5:R70 strain; Lane 6: BIN strain; Lane 8:R46 strain; Lane 9:R69 strain; Lane 10: B. infantis ATCC 15697. This figure is a composite of two gels put together at lanes 7 and 8.

the $17 strain (lane 11) which was first isolated by Reuter shared an identical enzyme profile with the type strain of B. breve (lane 12), whereas B. breve ATCC 15699, previously identified as S17 (lane 10) did not. Fig. 4 confirmed that all 'wild' strains of group II (Table 3) exhibited identical /3-galactosidase electrophoretic patterns with the type strain of B. animalis. Finally,/3-galactosidase profiles of other strains of commercial origin were compared with those of type strains of B. longum and B. infantis (Fig. 5). Only one commercial strain (lane 3) was identical to B. longum ATCC 15707 (lane 2). The R75 strain (lane 4; group IX)) had a single band as did the reference strain B. bifidum ATCC 11863 (band b: lanes 1, 7 and 11). Fig. 4 shows that R70 strain (lane 5) was identical to B. breve (lane 3; Fig. 2). Three commercial strains (lanes 6, 8 and 9) previously identified as B. longum based on fermentation of melezitose (group VI; Table 1) shared some /3-galactosidase bands with the type strain of B. infantis (lane 10) instead of B. longum (lane 2).

4. Discussion

Numerical analysis of phenotypic characteristics based on enzymatic activity and carbohydrate fermentation allowed the discrimination of most strains of bifidobacteria of animal origin from those of human origin. Gavini et al. (1991) also found that a clear separation between strains of animal origin and those of human origin was observed by performing numerical analysis based on a large number of enzymatic and carbohydrate tests. The results indicate that one group of animal origin (group II) contained all

68

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55- 70

reference strains of B. anirnalis and 10 strains isolated from fermented milks or commercial preparations. These results are in agreement with Biavati et al. (1992) who found that B. animalis was the only species present in the fermented milk preparations examined by these authors. Only three groups that contained strains of human origin were well defined (B. breL,e, B. dentiurn and B. bifidurn). The other groups included both strains of human and animal origin or strain of different species. B. indicum ATCC 25912 shared enzymatic profiles and fermentation patterns identical to those of B. infantis strains (Group IV). On the other hand, B. suis ATCC 27533 was grouped with strains of B. longurn. Lauer and Kandler (1983) found B. suis ATCC 27533 was also highly related to B. longum and B. infantis based on DNA-DNA homology and transaldolase isoenzyme profiles. These authors concluded that a comparative study of type strains of B. suis and the original strains isolated by Scardovi should be performed to clarify the differences between their results and those of Scardovi et al. (1979). The same re-investigation is also necessary for B. indicum since this strain gave similar results to those of B. infantis although the former species was isolated from the intestine of bees (Scardovi and Trovatelli, 1969). Numerical analysis allowed us to differentiate B. longum from B. anirnalis. Yaeshima et al. (1991) observed that all strains of B. longum possessed N-acetyl/3-glucosaminidase activity, but none of them possessed /3-glucosidase and phosphohydrolase. Strains of B. animalis we examined possessed /3-glucosidase and phosphohydrolase, but none of them possessed N-acetyl-/3-glucosaminidase. However, we found that the use of N-acetyl-/3-glucosaminidase activity as a differential characteristic between these two species was not adequate, since reference strains of B. longum (ATCC 15707 and 15708) did not possess this enzymatic activity. The presence of cystine aminopeptidase activity could be used instead of N-acetyl-/3glucosaminidase activity to differentiate B. animalis from B. longum since the former species gave a positive reaction and the latter none. The use of major phenotypic characteristics was useful for the presumptive differentiation of 'wild' strains isolated from commercial preparations. However, they cannot be exactly identified without D N A / D N A hybridization experiments. Yaeshima et al. (1992) noted that the D N A / D N A hybridization test is difficult to perform routinely in clinical and ecological studies. These authors suggest that the determination of enzyme activities are preferable for differentiation of bifidobacteria. In the present study, the determination of electrophoretic patterns of /3-galactosidases resulted in development of a new technique for the differentiation of Bifidobacteriurn species. Several isoenzymes of /3-galactosidase were detected among strains of bifidobacteria. Each species has a specific electrophoretic pattern. The detection of /3-galactosidase by electrophoresis is a new tool for distinguishing dairy- and non-dairy-related bifidobacteria. Dairy-related bifidobacteria and B. animalis could be better differentiated from other bifidobacteria by comparison of their /3-galactosidase electrophoretic patterns, than by numerical analysis of their phenotypic characteristics.

D. Roy et al. / InternationalJournal of Food Microbiology 23 (1994) 55-70

69

No /3-galactosidase band was detected among strains of animal or honeybee origin except for B. animalis strains (Groups I, II and III). There might be a problem of substrate specificity for the detection of/3-galactosidase activity, which could explain the discrepancies between electrophoretic patterns and enzyme profiles. The hydrolysis of beta-galactosidic linkage by /3-galactosidase is highly influenced by the variation of the aglycone moiety of the substrate (Shukla, 1975; Wallenfels and Weil, 1972). In our study, the hydrolysis of /3-naphtyl-CJ-D-galactopyranoside by /3-galactosidase belonging to strains of animal origin could be reduced as compared to the hydrolysis of O-nitrophenol-/3-D-galactopyranoside. 'Wild' strains of B. bifidum, B. breve and B. animalis isolated form fermented milk or commercial preparations used in the dairy industry were easily identified by comparing their /3-galactosidase electrophoretic patterns with those of the corresponding type strains. All 'wild' strains of B. animalis possessed /3-galactosidase profiles identical to that of the reference strains of B. animalis. In addition, the 'Bip' (ATCC 27920P) strain which was isolated from a culture of B. infantis ATCC 27920 also exhibited a similar pattern to B. animalis. These results suggest that the presence of B. animalis in commercial products might be due to the use of an erroneously identified isolate. Moreover, Biavati et al. (1992) noted that the difficulties of distinguishing B. longum from B. animalis on the basis of common phenotypic properties are probably the main reason for the industrial use of B. animalis. The determination of/3-galactosidase electrophoretic patterns may confirm the animal or human origin of commercial strains used in dairy products. Finally, three commercial strains (BIN, R46 and R69) identified as B. longum by numerical analysis did not exhibit any/3-galactosidase band similar to that of the type strain ATCC 15707. However, these strains of B. longum shared many /3-galactosidase bands with strains of B. infantis and possessed N-acetyl-/3-glucosaminidase activity as did B. infantis; however, both species were melezitose-positive. These results suggest the presence of new biovars of B. longum. Yaeshima et al. (1992) found that strains identified as B. longum by D N A - D N A homology were divided into six phenotypic groups. The determination of the pulsed-field gel electrophoretic patterns of the genomic DNA of commercial strains of B. longum will be carried out to determine the diversity of biovars of this species.

Acknowledgements The authors wish to thank Luc Savoie for his scientific and technical assistance and Pierre Ward for his collaboration. Food Research and Development Centre contribution no. 311.

References Biavati, B., Mattarelli, P. and Crociani, F. (1992) Identification of bifidobacteria from fermented milk products. Microbiologica 15, 7-14.

70

D. Roy et al. / International Journal of Food Microbiology 23 (1994) 55-70

Biavati, B., Scardovi, V. and Moore, W.E.C. (1982) Electrophoretic patterns of proteins in the genus Bifidobacterium and proposal of four new species. Int. J. System. Bact. 32, 358-373. Bhowmik, T. and Marth, E.H. (1989) Simple method to detect /3-galactosidase. Appl. Microbiol. 55, 3240-3242. Chevalier, P., Roy, D. and Savoie, L. (1991) X-o~-Gal-based medium for simultaneous enumeration of bifidobacteria and lactic acid bacteria in milk. J. Microbiol. Meth. 13, 75-83. Lauer, E. and Kandler, O. (1983) DNA-DNA homology, murein types and enzyme patterns in the type strains of the genus Bifidobacterium. System. Appl. Microbiol. 4, 42-64. Gavini, F., Pourcher, A.M., Neut, C., Monget, D., Romond, C., Oger, C. and Izard, D. (1991) Phenotypic differentiation of bifidobacteria of human and animal origins. Int. J. System. Bact. 41, 548-557. Gomori, G. (1955) Preparation of buffers for use in enzyme studies. In: S.P Colowick and N.O. Kaplan (editors) Methods in Enzymology, I. Academic Press, New York, pp. 138-146. Hames, B.D. (1987) An introduction to polyacrylamide gel electrophoresis. In: B.D. Hames and R. Rickwood (editors) Gel Electrophoresis of Proteins. A Practical Approach. IRE Press, Washington, DC, pp. 1-92. Reuter, G. (1990) Bifidobacteria cultures as components of yoghurt-like products. Bifidobacteria and Microflora 9, 107-118. Roy, D. and Ward, P. (1992) Rapid detection of Bifidobacterium dentium by enzymatic hydrolysis of /3-glucuronidase substrates. J. Food Prot. 55, 291-295. Scardovi, V. and Trovatelli, L.D. (1969) New species of bifid bacteria from Apis mellifica L. and Apis indica F. A contribution to the taxonomy and biochemistry of the genus Bifidobacterium. Zbl. Bakt., II. Abt. 123, 64-88. Scardovi, V., Casalicchio, F. and Vincenzi, N. (1979) Multiple electrophoretic forms of transaldolase and 6-phosphogluconate dehydrogenase and their relationships to the taxonomy and ecology of bifidobacteria. Int. J. System. Bact. 29, 312-327. Sgorbati, B. and Scardovi, V. (1974) Electrophoretic types of transaldolase, transketolase, and other enzymes in bifidobacteria. Antonie van Leeuwenhoek 40, 427-440. Shukla, T.P. (1975) Beta-galactosidase technology: a solution to the lactose problem. Crit. Rev. Food Technol. 5, 325-356. Smith, P.K., Krohan, R.I, Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano, M.D., Fujimoto, E.K., Goeke, N.M., Olson, B.J. and Klenl, D.C. (1985) Measurement of protein using bicinchoninic acid. Anal. Biochem. 150, 76-85. Sokal, R.R. and Michener, C.D. (1958) A statistical method for evaluating systematic relationships. Kansas State Univ. Bull. 38, 1409-1438. Wallenfels, K. and Weil, R. (1972)/3-galactosidase. In: P.D. Boyer (editor), The Enzymes, 7. Academic Press, New York, pp. 617-663. Yaeshima, T., Fujisawa, T. and Mitsuoka, T. (1991) Differential characteristics of Bifidobacterium longum and Bifidobacterium animalis. Syst. Appl. Microbiol. 14, 169-172. Yaeshima, T., Fujisawa, T. and Mitsuoka., T. (1992) The diversity in phenotypic characteristics of Bifidobacteriurn longum. Milchwissenschaft 47, 212-214.