OUP user menu

Distribution of β-glucosidase and β-glucuronidase activity and of β-glucuronidase gene gus in human colonic bacteria

Marta Dabek, Sheila I. McCrae, Valerie J. Stevens, Sylvia H. Duncan, Petra Louis
DOI: http://dx.doi.org/10.1111/j.1574-6941.2008.00520.x 487-495 First published online: 1 December 2008


β-Glycosidase activities present in the human colonic microbiota act on glycosidic plant secondary compounds and xenobiotics entering the colon, with potential health implications for the human host. Information on β-glycosidases is currently limited to relatively few species of bacteria from the human colonic ecosystem. We therefore screened 40 different bacterial strains that are representative of dominant bacterial groups from human faeces for β-glucosidase and β-glucuronidase activity. More than half of the low G+C% Gram-positive firmicutes harboured β-glucosidase activity, while β-glucuronidase activity was only found in some firmicutes within clostridial clusters XIVa and IV. Most of the Bifidobacterium spp. and Bacteroides thetaiotaomicron carried β-glucosidase activity. A β-glucuronidase gene belonging to family 2 glycosyl hydrolases was detected in 10 of the 40 isolates based on degenerate PCR. These included all nine isolates that gave positive assays for β-glucuronidase activity, suggesting that the degenerate PCR could provide a useful assay for the capacity to produce β-glucuronidase in the gut community. β-Glucuronidase activity was induced by growth on d-glucuronic acid, or by addition of 4-nitrophenol-glucuronide, in Roseburia hominis A2-183, while β-glucosidase activity was induced by 4-nitrophenol-glucopyranoside. Inducibility varied between strains.

  • microbiota
  • colon
  • β-glucosidase
  • β-glucuronidase
  • enzyme activity
  • degenerate PCR


Members of the gut microbiota in the human large intestine exhibit a variety of enzymatic activities with potential impact on human health through biotransformation of secondary plant products and xenobiotic compounds (McBain & Macfarlane, 1998; Heavey & Rowland, 2004; Blaut & Clavel, 2007). β-Glucuronidases liberate toxins and mutagens that have been glucuronated in the liver and excreted into the gut with the bile. This can lead to high local concentrations of carcinogenic compounds within the gut, thus increasing the risk of carcinogenesis (Gill & Rowland, 2002). Furthermore, reuptake of the deconjugated compound from the gut and reglucuronidation in the liver leads to an enterohepatic circulation of xenobiotic compounds, which increases their retention time in the body. β-Glucosidases can exert either beneficial or harmful effects, as they form aglycones from a range of different plant glucosides, which might exhibit either toxic/mutagenic or health-promoting effects (Hill, 1995; Manachet al., 2004). Some plant glucosides are also subject to deconjugation by host β-glucosidases in the upper gut and may subsequently be glucuronated by the host, making them a substrate for bacterial β-glucuronidases when they reach the colon with the bile (Manachet al., 2004). The resulting aglycones of plant polyphenols may be subject to further degradation and biotransformation by the gut microbiota (Blautet al., 2003; Atkinsonet al., 2005).

Several studies have investigated bulk enzyme activities in faecal samples. A high interindividual variation in enzyme activities has been demonstrated in human faeces (McBain & Macfarlane, 1998), which might be due to differences in the composition of the gut microbiota or other influencing factors, such as diet. Increased β-glucosidase activity upon soy consumption has been shown in human volunteers (Wisemanet al., 2004). Faecal β-glucuronidase activity increased in rodents after consumption of a high-protein/high-fat diet (Eriyamremuet al., 1995) and decreased after consumption of diets high in carbohydrates (Shiau & Chang, 1983; Gestelet al., 1994). It has also been reported that cancer patients exhibit higher β-glucuronidase activities than healthy controls (Kim & Jin, 2001).

Relatively little information is available on the distribution of these enzyme activities between different gut bacteria, especially with regard to the low G+C% Gram-positive firmicutes, which are a dominant bacterial group within the human large intestine (Flintet al., 2007). Two groups in particular, bacteria related to Roseburia spp. and Eubacterium rectale within clostridial cluster XIVa (Collinset al., 1994) and Faecalibacterium prausnitzii within clostridial cluster IV, have been estimated by FISH to comprise c. 7% and 10% of the total microbiota in healthy human adults (Aminovet al., 2006; Muelleret al., 2006). Bacterial β-glucosidases seem to be more widespread among the colonic microbiota than β-glucuronidases (McBain & Macfarlane, 1998). Nakamuraet al. (2002) examined various enzyme activities in fresh human faecal isolates and reference strains and found the highest β-glucuronidase and β-glucosidase activities in some of the isolates related to Clostridium spp. Most of the Bacteroides spp. examined also carried high β-glucosidase activity. McBain & Macfarlane (1998) examined 20 strains from various genera and detected β-glucosidase activity in most of the strains tested, with particularly high activity in Bacteroides ovatus. β-Glucuronidase activity was low or absent in Clostridium spp., but generally present in Bacteroides spp. and most Bifidobacterium spp. Some studies have compared enzyme activities in in vitro culture with faecal activities in gnotobiotic animals and found that activities tended to be higher in vivo (Coleet al., 1985; Gadelleet al., 1985), showing that environmental conditions are important in determining those activities.

The aim of this study was to examine a range of human gut bacteria for the presence of β-glucuronidase and β-glucosidase activities, with an emphasis on the understudied but predominant bacterial group of low G+C% Gram-positive bacteria. A selection of bacterial strains was examined in more detail to gain an insight into whether growth conditions affect these enzyme activities.

Materials and methods

Bacterial strains and culture conditions

All bacterial strains used in the current study except Clostridum clostridioforme JCM 1291 were isolated from human faeces. Strains A-, L- and T-, originated from a study with two adult volunteers and a baby (Barcenilla, 1999; Barcenillaet al., 2000), and other strains from several adult volunteers (except for Eubacterium siraeum 70/3 and Ruminococcus sp. 80/3) were described by Louiset al. (2004). Clostridial cluster IV strains E. siraeum 70/3 and Ruminococcus sp. 80/3 were isolated from a faecal sample from an adult female consuming a western diet. The volunteer had not received antibiotics or any other product likely to influence the composition of the faecal microbiota in the previous 6 months. A faecal slurry (10%) was prepared in anaerobic phosphate buffer (pH 6.8). One aliquot of the slurry was heated to 70 °C and one to 80 °C in water baths for 10 min, and then quickly cooled to room temperature. Following preparation of 10-fold serial dilutions in M2 medium under anaerobic conditions, these dilutions were used to inoculate M2GSC roll tubes (Miyazakiet al., 1997) and incubated at 37 °C for 48 h. Single colony isolates were grown in M2GSC medium and repurified following a second passage through roll tubes as described. Strains 70/3 and 80/3 were isolated from the samples heated at 70 and 80 °C, respectively. The 16S rRNA gene of both strains was cloned and sequenced as described previously (Louiset al., 2004). The GenBank accession numbers are EU266550 and EU266551, respectively.

Bifidobacterium spp. and Bacteroides spp. were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ), except for Bifidobacterium adolescentis L2-32, which was isolated at the Rowett Research Institute (Barcenilla, 1999), and B. ovatus V975, which was obtained from Terry Whitehead at the US Department of Agriculture, Illinois, USA. Clostridum clostridioforme JCM 1291 was obtained from the National Collections of Industrial and Marine Bacteria (NCIMB).

For the initial screening of β-glucosidase and β-glucuronidase activity, all bacteria were grown on M2GSC medium for 24 h. All further experiments were performed in YCFA (yeast extract, casitone, fatty acid) medium (Duncanet al., 2002) containing 0.1% casitone and 0.2% glucose, or other carbohydrates and supplements as indicated in the results, except for strains F. prausnitzii L2-6 and M21/2, which were grown in M2GSC, as they exhibited poor growth in YCFA medium. 4-Nitrophenyl-β-d-glucopyranoside or 4-nitrophenyl-β-d-glucuronide were added to cultures from 10-fold concentrated stocks in 100 mM sodium phosphate buffer (pH 6.5) to achieve the final concentrations as indicated in the results.

To investigate whether glycosides might lead to induction, the 4-nitrophenyl-substrates utilized for the enzyme assays were added to cultures of Roseburia hominis A2-183 in the presence of glucose, and enzyme activities were determined in the stationary phase. Substrates were added at 0.2 and 1 mM and the enzymatic product 4-nitrophenol was included to the control for any effects it might elicit from the cells.

Preparation of bacterial cell extracts

Bacterial culture (7.5 mL) was centrifuged (3800 g, 10 min, 4 °C). The cell pellet was washed two times in 3 mL chilled 100 mM sodium phosphate buffer, resuspended in 1.25 mL buffer and kept on ice until cell disruption. For the initial screen of all strains examined, the phosphate buffer was adjusted to pH 6.0, based on a study reporting the highest enzyme activities at this pH in human faeces (Mallettet al., 1989). Subsequent analysis of strains R. hominis A2-183 and F. prausnitzii A2-165 revealed that pH 6.5 was a better compromise for measuring both activities under the same conditions (data not shown); therefore the pH was adjusted to 6.5 throughout all further experiments to enable a direct comparison of the results.

For cell disruption, 1 mL of cell suspension was added to 250 μL of glass beads (106 μm, washed several times in deionized water and autoclaved) in a 2-mL threaded tube and beaten in a mini bead beater (Stratech Scientific) at 5000 r.p.m. two times for 30 s each, with a 30-s incubation on ice in between. The tube was centrifuged (16 000 g, 10 min, 4 °C) and the supernatant, representing the cell extract, was removed and stored on ice.

To determine enzyme activities in the insoluble fraction, the pellet was washed three times with 0.95 mL sodium phosphate buffer and finally resuspended in 0.95 mL buffer. Enzyme activity determined in the washes was regarded as belonging to the soluble fraction and added to the values obtained for the cell extract in order to determine the relative proportions of enzyme activity recovered from the soluble fraction and insoluble fraction, respectively.

Determination of β-glucosidase and β-glucuronidase activity

The substrates 4-nitrophenyl-β-d-glucopyranoside or 4-nitrophenyl-β-d-glucuronide were prepared as 10 mM solutions in 100 mM sodium phosphate buffer of the same pH as was used for preparing cell extracts and prewarmed to 37 °C. Bacterial extracts, cell debris or culture supernatants were prewarmed to 37 °C before adding 75 μL of sample in duplicate to wells of a microtitre plate. Seventy-five microlitres of substrate solution was added and the development of 4-nitrophenol was recorded over 30 min in a plate reader (Tecan) at 37 °C at 405 nm. The 4-nitrophenol concentration was determined from a standard curve of 4-nitrophenol in sodium phosphate buffer of the same pH as was used for preparing cell extracts. Protein concentrations were determined with a bicinchoninic acid kit (Pierce). All data are mean and SD of at least three repeats.

Degenerate PCR, cloning, sequencing and sequence analysis

Degenerate PCR and primer design were performed with bacterial cells as template, as described previously (Louiset al., 2004). Degenerate primer GNfor (TAT TTA AAA GGI TTY GGI MRI CAY GAR GA) was designed from an alignment of deduced protein sequences from Ruminococcus gnavus (AY307023), Lactobacillus gasseri (AF305888), Staphylococcus sp. (AF354044) and Escherichia coli (S69414). Degenerate primer GNP2mod (CC TTC TGT TGT IKB RAA RTC IGC RAA RTT CCA) is a modification of primer P2 from Beaudet al. (2005). Both primers contain 5′-tags based on the sequence of the R. gnavus gene (AY307023). Amplification was performed with BIOTAQ DNA Polymerase (Bioline) according to the manufacturer's instructions with a ramped annealing approach (Skantar & Carta, 2000) using the following conditions: initial denaturation (3 min at 94 °C), then 35 cycles of denaturation (30 s at 94 °C), ramped annealing (20 s at 55 °C, 5 s at 50 °C, 5 s at 45 °C and 5 s at 40 °C), and elongation (1 min at 72 °C) and a final extension (10 min at 72 °C). PCR products obtained with Roseburia intestinalis L1-82 and B. ovatus V975 as template were cloned into pGEM-T-Easy (Promega) according to the manufacturer's instructions. Sequencing with vector primers M13F and M13R (Promega) and internal primers (L1-82bglucF, ATCCGTGAACACCATGAACAG and L1-82bglucR, GGAATGCCAGTACTCGTATGC; V975gusF, TAATGCACGGTTGGCAGTG and V975gusR, CACCCCAGATAGTAAGACTG) was performed on a Beckman capillary sequencer. blastp analysis (Altschulet al., 1990) was performed with the deduced protein sequence to identify the closest match in the GenBank database. The accession numbers for the gene fragment of R. intestinalis L1-82 and B. ovatus V975 are EU331462 and EU541480, respectively.

The phylogenetic tree was constructed using the program mega4 (Tamuraet al., 2007), using the neighbour-joining method, 1000 times bootstrap and distance model Kimura.


Presence of β-glucosidase and β-glucuronidase activity in human colonic bacteria

A range of human colonic bacteria were screened for β-glucosidase and β-glucuronidase activity after growth in M2GSC medium to facilitate growth of all bacteria under the same conditions. Two low G+C% Gram-positive strains of clostridial cluster IV and all cluster XIVa strains related to the Roseburia/E. rectale group exhibited β-glucosidase activity, but only few of the cluster XIVa strains outside that bacterial group displayed the activity (Table 1,Fig. 1). The highest β-glucosidase activity was found in cell extracts of some of the Bifidobacterium species examined, while among the Bacteroides strains tested, only Bacteroides thetaiotaomicron displayed activity (Table 1).

View this table:

β-Glucosidase and β-glucuronidase activity in human colonic bacteria and degenerate PCR screen for β-glucuronidase gene gus

Phylogenetic group Bacterial strainEnzyme activity [U (mg protein)−1]gus degenerate PCR product
Clostridial cluster XIVa
Anaerostipes caccae L1-92 (DSM 14662T)NDND
Butyrivibrio fibrisolvens 16/40.048 ± 0.028ND
Coprococcus comes A2-232NDND
Coprococcus comes SL7/1NDND
Coprococcus eutactus ART55/10.032 ± 0.001ND
Coprococcus sp. L2-500.022 ± 0.024ND
Eubacterium hallii L2-7 (DSM 17630)NDND
Eubacterium hallii SM6/1NDND
Eubacterium rectale A1-860.119 ± 0.091ND
Eubacterium rectale M104/10.022 ± 0.006ND
Eubacterium rectale T1-8150.058 ± 0.016ND
Roseburia faecis M72/1 (DSM 16840T)0.094 ± 0.018ND
Roseburia hominis A2-1810.072 ± 0.0230.070 ± 0.013+
Roseburia hominis A2-183 (DSM 16839T)0.089 ± 0.0100.059 ± 0.014+
Roseburia intestinalis L1-81510.225 ± 0.0180.038 ± 0.045+
Roseburia intestinalis L1-82 (DSM 14610T)0.155 ± 0.0790.008 ± 0.002+
Roseburia intestinalis L1-9520.157 ± 0.0350.010 ± 0.002+
Roseburia intestinalis M50/10.186 ± 0.0380.007 ± 0.002+
Roseburia inulinivorans A2-194 (DSM 16841T)0.168 ± 0.016ND
Ruminococcus obeum A2-1620.084 ± 0.009ND
Isolate M62/1NDND
Isolate SSC/2NDND
Clostridial cluster IV
Eubacterium siraeum 70/30.037 ± 0.002ND
Faecalibacterium prausnitzii A2-165 (DSM 17677)ND0.505 ± 0.013+
Faecalibacterium prausnitzii L2-6ND0.020 ± 0.006+
Faecalibacterium prausnitzii M21/2NDND+
Ruminococcus bromii L2-63NDND
Ruminococcus sp. 80/30.069 ± 0.019ND
Clostridial cluster XVI
Eubacterium cylindroides SM7/11NDND
Eubacterium cylindroides T2-87NDND
Bifidobacterium spp.
Bifidobacterium adolescentis DSM 20083T2.046 ± 0.612ND
Bifidobacterium adolescentis L2-320.535 ± 0.083ND
Bifidobacterium angulatum DSM 20098T0.859 ± 0.140ND
Bifidobacterium bifidum DSM 20456TNDND
Bifidobacterium breve DSM 20213T0.275 ± 0.103ND
Bifidobacterium longum DSM 20219TNDND
Bifidobacterium pseudocatenulatum DSM 20438T0.312 ± 0.134ND
Bacteroides spp.
Bacteroides ovatus V975NDND+
Bacteroides thetaiotaomicron DSM 2079T0.082 ± 0.033ND
Bacteroides vulgatus DSM 1447TNDND
  • * ND: enzyme activity<0.005 U mg−1.

  • ± : presence/absence of a PCR product of the expected size after amplification with primers GNfor and GNP2mod.

  • Note that low level β-glucuronidase activity was detected for Faecalibacterium prausnitzii M21/2 in subsequent experiments, see main text for details.


Phylogenetic tree of human faecal low G+C%firmicutes based on 16S rRNA gene sequences (accession numbers given in parentheses) corresponding to positions 44 to 1474 of the Escherichia coli numbering system (Brosiuset al., 1978). Clostridial clusters are indicated by roman numbers. GS, β-glucosidase activity; GN, β-glucuronidase activity; +, positive and −, negative for respective enzyme activity; (+), Strain did display activity in subsequent experiments; see main text for details.

β-Glucuronidase activity was only found in few of the bacteria examined, namely, in strains belonging to R. hominis and R. intestinalis within cluster XIVa, and two F. prausnitzii-related isolates from cluster IV (Fig. 1). Faecalibacterium prausnitzii A2-165 exhibited by far the highest activity of all strains examined (Table 1).

Growth phase-dependence and localization of enzyme activity in R. hominis A2-183 and F. prausnitzii A2-165

Two strains were selected to examine the two glycosidase activities in more detail. Roseburia hominis A2-183 was chosen as it harbours both activities and F. prausnitzii A2-165 because it exhibited the highest β-glucuronidase activity.

Further experiments were performed with the more defined YCFA medium, containing glucose or other carbon sources where indicated. The β-glucosidase activities of R. hominis A2-183 remained very similar, regardless of whether cells were harvested in exponential [0.058±0.004 U (mg protein)−1] or stationary growth phase [0.062±0.002 U (mg protein)−1]. β-Glucuronidase activity, on the other hand, increased in stationary phase in both bacteria [R. hominis A2-183 exponential phase: 0.049±0.002 U (mg protein)−1, stationary phase: 0.146±0.00 U (mg protein)−1; F. prausnitzii A2-165 exponential phase: 0.43±0.02 U (mg protein)−1, stationary phase: 0.77±0.09 U (mg protein)−1].

The localization of the enzyme activities was examined by separating culture supernatants, the soluble fraction of cell extracts and cell debris as described inMaterials and methods.

The bulk of the activity (between 72% and 100%) was found in the soluble fraction of the cells in both the exponential and the stationary phase cells for both strains and both activities, with a slight increase in the insoluble fraction in stationary phase (data not shown). The activity found in the culture supernatant was negligible. Based on these results, all subsequent measurements were performed on the soluble extracts only.

Effect of alternative carbon sources on enzyme activities

Enzyme activities were determined after growth of R. hominis A2-183 on four alternative carbon sources, glucose, cellobiose, xylose and d-glucuronic acid. Cells were passaged once through the respective medium, and activities were determined in stationary cells of the second passage. Activities for both enzymes were only marginally affected by growth on cellobiose and xylose in comparison to glucose; however, β-glucuronidase activity increased approximately ninefold in the presence of d-glucuronic acid (Fig. 2). An increase of β-glucuronidase activity to similar levels was also seen in exponentially growing cells in the presence of d-glucuronic acid (data not shown). When 0.02% of d-glucuronic acid was added to glucose-containing medium, the β-glucuronidase activity of R. hominis A2-183 remained low during exponential phase but rose to 0.282±0.002 U (mg protein)−1 compared with 0.095±0.005 U(mg protein)−1 in the glucose only control in stationary phase, indicating that the presence of glucose prevents the effect of d-glucuronic acid (data not shown).


Enzyme activities and final OD of stationary phase cells of Roseburia hominis A2-183 after growth in YCFA medium with different carbon sources at 0.2%. Grey bars, β-glucosidase activity; black bars, β-glucuronidase activity; hatched bars, OD. Significance level compared with glucose-grown cultures: *P<0.05; ****P<0.0001.

In contrast, the β-glucuronidase activity of R. intestinalis L1-82 was similar for glucose and d-glucuronic acid-grown cultures [0.0093±0.0003 and 0.010±0.0003 U (mg protein)−1, respectively]. Roseburia inulinivorans A2-194, Roseburia faecis M72/1 and E. rectale A1-86 showed no detectable β-glucuronidase activity when grown on M2GSC medium (Table 1), and failed to grow on d-glucuronic acid as the sole carbon source in YCFA medium. When grown in the presence of 0.02%d-glucuronic acid in addition to glucose, these strains did not show detectable β-glucuronidase activities.

Faecalibacterium prausnitzii A2-165 did not exhibit an altered activity after growth in the presence of 0.2%d-glucuronic acid compared with growth on glucose [0.72±0.05 and 0.77±0.09 U (mg protein)−1, respectively]. The other two F. prausnitzii strains, L2-6 and M21/2, exhibited very poor growth in YCFA medium and experiments were therefore performed in M2GSC with either 0.2% glucose or d-glucuronic acid added. We found very low β-glucuronidase activity in the strain M21/2, which was very slightly increased in the presence of d-glucuronic acid [0.0025±0.0003 U (mg protein)−1 compared with 0.0019±0.0001 U (mg protein)−1 on glucose, P=0.028]. This low level of activity was probably not picked up in the initial screen as a higher assay pH was used (seeMaterials and methods), which increases the absorbance of the assay product 4-nitrophenol and thus is slightly more sensitive. Strain L2-6 exhibited 0.025 U (mg protein)−1 regardless of the carbon source present.

Effect of glucuronide and glucoside substrates on glycosidase activites

An increase of each enzyme activity was seen with its respective substrate in stationary phase cells of R. hominis A2-183. 4-Nitrophenol alone had an inhibitory effect on growth, but did not result in induction (Fig. 3). This effect of the substrates was also found in exponentially growing cells (tested for 1 mM substrate concentration only, data not shown).


Enzyme activities and final OD of stationary phase cells of Roseburia hominis A2-183 after growth in YCFA medium with glucose and 4-nitrophenol compounds. Grey bars, β-glucosidase activity; black bars, β-glucuronidase activity; hatched bars, OD. 4-np, 4-nitrophenol; GS, 4-nitrophenyl-β-d-glucopyranoside; GN, 4-nitrophenyl-β-d-glucuronide. Significance level compared with buffer: *P<0.05; **P<0.01; ****P<0.0001.

An examination of other bacteria of the Roseburia/E. rectale group and F. prausnitzii strains revealed that the enzyme activities of most strains were not affected by the presence of 1 mM glycoside substrate (Table 2). Eubacterium rectale A1-86, however, exhibited a 12.6-fold increase in β-glucosidase activity and F. prausnitzii L2-6 a 4.5-fold increase in β-glucuronidase activity (Table 2). Inducibility therefore differed markedly between strains and species.

View this table:

Effect of glucuronide and glucoside substrates on glycosidase activites

Bacterial strainTreatmentEnzyme activity [U (mg protein)−1]/fold change
R. hominis A2-183Buffer0.054 ± 0.0050.119 ± 0.008
GS0.383 ± 0.007/7.10.146 ± 0.006/1.2
GN0.041 ± 0.004/0.80.336 ± 0.017/2.8
R. intestinalis LI-82Buffer0.085 ± 0.0100.014 ± 0.002
GS0.104 ± 0.023/1.20.014 ± 0.001/1.0
GN0.088 ± 0.017/1.00.014 ± 0.002/1.0
R. inulinivorans A2-194Buffer0.086 ± 0.001ND
GS0.090 ± 0.010/1.1ND
R. faecis M72/1Buffer0.118 ± 0.007ND
GS0.102 ± 0.009/0.9ND
E. rectale A1-86Buffer0.017 ± 0.004ND
GS0.218 ± 0.020/12.6ND
F. prausnitzii A2-165BufferND0.740 ± 0.146
GNND0.917 ± 0.328/1.2
F. prausnitzii M21/2BufferND0.002 ± 0.000
GNND0.002 ± 0.000/0.9
F. prausnitzii L2-6BufferND0.016 ± 0.001
GNND0.073 ± 0.003/4.5
  • * Compared with buffer control.

  • GS: 1 mM 4-nitrophenyl-β-d-glucopyranoside.

  • Significance compared with buffer control P<0.0001.

  • § Significance compared with buffer control P<0.05.

  • GN: 1 mM 4-nitrophenyl-β-d-glucuronide.

  • ND, not determined.

Degenerate PCR of β-glucuronidase gene gus

As some of the strains examined here possessed very low β-glucuronidase activities, it was questionable whether strains exhibiting no activity in our initial screen did indeed lack a β-glucuronidase gene or were not expressing the respective enzyme. We therefore performed degenerate PCR against a β-glucuronidase gene belonging to family 2 glycosyl hydrolases (Carbohydrate Active Enzymes database, http://www.cazy.org/; Coutinho & Henrissat, 1999) that has been described in other bacteria (Beaudet al., 2005; Russell & Klaenhammer, 2001), with the 40 bacterial strains examined in the enzyme activity screen. For isolates belonging to Firmicutes, a product of the expected size was only found for the two R. hominis strains, four R. intestinalis strains and three F. prausnitzii strains, in accordance with the enzymological results (Table 1). Of the Bifidobacterium and Bacteroides strains tested, only B. ovatus V975 exhibited a faint band of the expected size, despite not showing detectable enzyme activity.

We cloned and sequenced the degenerate PCR product of R. intestinalis L2-82 and B. ovatus V975. blastp analysis of the R. intestinalis L2-82 sequence revealed the highest similarity to an ORF of the draft genome sequence of F. prausnitzii M21/2 (ZP_02090669, 71% identity) and matches to other β-glucuronidase genes (closest match to C. cellulolyticum H10 ZP_01574828.1, 56% identity; match to R. gnavus gusAAQ76046, 38% identity). The Conserved Domains Database (Marchler-Bauer & Bryant, 2004) revealed the highest matches to the TIM barrel domain of glycosyl hydrolases family 2 (pfam02836) and β-glucuronidase (PRK10150) and lower matches to β-galactosidases (COG3250, PRK09525, PRK10340). The two glutamate residues involved in the catalytic mechanism of other members of family 2 β-glucuronidases (Sallehet al., 2006) are conserved in the R. intestinalis L1-82 gene. The B. ovatus V975 sequence also showed similarity to family 2 glycosyl hydrolases. It exhibited 100% identity to a hypothetical protein from B. ovatus ATCC 8483 (ZP_02065724) and 42–41% identity to β-galactosidases and family 2 glycosyl hydrolases from other Bacteroides spp. However, it displayed only 28% identity to the deduced protein sequence from R. intestinalis L1-82 (EU331462) and 24% to the R. gnavus sequence (AAQ76046) and is therefore unlikely to encode the same type of enzyme.


The human gut microbiota is involved in the metabolism of certain xenobiotics and secondary plant compounds; however, the bacteria responsible and their enzymatic activities mostly remain poorly characterized. Here, we screened 40 bacterial strains mainly belonging to the numerically important firmicutes for two glycosidase activities responsible for the conversion of glycosidic compounds to their respective aglycones. In accordance with McBain & Macfarlane (1998), who reported a higher prevalence of β-glucosidase producers than β-glucuronidase producers based on most probable number measurements, we found detectable levels of β-glucosidase activity in 23, but of β-glucuronidase activity in only nine, of the 40 strains screened. To our knowledge, these enzyme activities have not previously been surveyed in a wide range of low G+C% Gram-positive bacteria belonging to the main intestinal clostridial clusters XIVa and IV, although other studies have reported respective activities in strains belonging to Bifidobacterium and Bacteroides spp. A direct comparison of the data is difficult, as the experimental set-up varied greatly between studies. Here, we performed assays under aerobic conditions, as it was previously shown that this leads to similar results as anaerobic incubations, whereas 4-nitrophenol substrates might be degraded under anaerobic conditions if nitroreductase activities are present (Chadwicket al., 1995). McBain & Macfarlane (1998) determined enzyme activites in the extracellular compartment as well as in intact cells under anaerobic conditions. Glycosidase activities were found in several strains; however, the reported activities mostly were below the detection limit of our study. Bacteroides ovatus DCNC 11 exhibited by far the highest β-glucosidase activity of all strains tested, whereas we could not detect this activity in B. ovatus V975. However, the different strains used might behave differently, analogous to the differences McBain & Macfarlane (1998) have found for Bifidobacterium and Beaudet al. (2006) for R. gnavus strains. Bifidobacterium breve DSM 20213T (NCFB 2257) displayed only detectable β-glucosidase activity in our study, whereas McBain & Macfarlane (1998) found both enzyme activities with intact cells; however, β-glucuronidase activity was lower than β-glucosidase activity. In the present study, closely related strains mostly showed similar activity profiles; however, the level of β-glucuronidase activity varied dramatically in the case of the three F. prausnitzii isolates investigated. Nakamuraet al. (2002) have also tested activities of anaerobic cultures and found higher values of β-glucosidase than β-glucuronidase activity in the Bacteroides and Bifidobacterium species that were also tested in this study. Bacteroides vulgatus DSM 1447T (JCM 5826T) was also used in this study, and we also tested the rumen strain C. clostridioforme JCM 1291T for direct comparison (data not shown) but could not detect either of the enzyme activities in cell extracts. However, the detection limit of our study was higher than the values reported by Nakamuraet al. (2002), as we aimed to identify bacteria carrying high activities, which are more likely to be important contributors to those activities in the gut. Apart from differences in cell preparation and enzyme assay conditions, the results are also likely to be influenced by the media used and growth conditions, as we could show here that the enzyme activity of some strains varied dramatically under different growth conditions.

The results of the degenerate PCR screen used in the present study against a β-glucuronidase gene (gus) previously described in other bacteria (Beaudet al., 2005; Russell & Klaenhammer, 2001) correlated well with the results obtained by enzyme activity measurements. Thus, this method could be useful as a quick tool to identify novel isolates likely to exhibit β-glucuronidase activity. Only one strain, B. ovatus V975, exhibited a faint band by degenerate PCR while it did not show detectable β-glucuronidase activity; however, the cloned fragment showed only low similarity to the translated gus gene and is therefore unlikely to code of a homologous enzyme. Beaudet al. (2006) screened for the presence of the gus gene in a range of R. gnavus strains by degenerate PCR and Southern hybridization, and found strains harbouring enzyme activity that were negative in the molecular screen and vice versa. Therefore, more work is needed to resolve whether this gene is indeed responsible for the enzyme activities measured here. Two genes within the genome sequence of B. thetaiotaomicron have been annotated as β-glucuronidase genes (Xuet al., 2003). In the present study, this bacterium did not display detectable β-glucuronidase activity. The two genes (BT3292, BT4151), as well as the R. gnavus gene screened for in this study, belong to family 2 glycosyl hydrolases (Carbohydrate Active Enzymes database, http://www.cazy.org/; Coutinho & Henrissat, 1999), but share little sequence identity (BT3292: 22%, BT4151: 20%) to the R. gnavus sequence (AY307023), which has been confirmed as encoding β-glucuronidase activity by complementation (Beaudet al., 2005). Whether the B. thetaiotaomicron genes encode β-glucuronidase genes or glycosyl hydrolases with different specificities remains to be confirmed.

In conclusion, our results indicate that changes in faecal bulk glycosidase activities in response to changes in dietary intake are likely to be due to both, changes in the number of bacteria carrying those activities and regulatory changes within certain strains. The strong effect on β-glucuronidase activity seen in R. hominis A2-183 of a glucuronide substrate as well as of one of the products of the reaction, d-glucuronic acid, indicates that the level of exposure to glycosides in the colon, which is dependent on the type of diet consumed, could affect the enzyme activity levels in some members of the gut microbiota. However, this is a complex issue, as many factors could influence those enzyme activities in vivo. The in vitro results presented here may help to design future experiments in model systems to address these points. Further studies are also necessary to elucidate the underlying mechanisms of regulation.


The Rowett Research Institute is funded by the Scottish Government Rural and Environment Research and Analysis Directorate. M.D. received funding from the European Union as a fellow of the framework 5 training site Anaerobe. We would like to thank Terry Whitehead for strain B. ovatus V975, Marketa Hrdinova for help with isolating strain 80/3 and Harry Flint for critically reading the manuscript.


  • Editor: Julian Marchesi


View Abstract