Bifidobacterium β-Glucosidase Activity and Fermentation of Dietary Plant Glucosides Is Species and Strain Specific

Dietary plant glucosides are phytochemicals whose bioactivity and bioavailability can be modified by glucoside hydrolase activity of intestinal microbiota through the release of acylglycones. Bifidobacteria are gut commensals whose genomic potential indicates host-adaption as they possess a diverse set of glycosyl hydrolases giving access to a variety of dietary glycans. We hypothesized bifidobacteria with β-glucosidase activity could use plant glucosides as fermentation substrate and tested 115 strains assigned to eight different species and from different hosts for their potential to express β-glucosidases and ability to grow in the presence of esculin, amygdalin, and arbutin. Concurrently, the antibacterial activity of arbutin and its acylglycone hydroquinone was investigated. Beta-glucosidase activity of bifidobacteria was species specific and most prevalent in species occurring in human adults and animal hosts. Utilization and fermentation profiles of plant glucosides differed between strains and might provide a competitive benefit enabling the intestinal use of dietary plant glucosides as energy sources. Bifidobacterial β-glucosidase activity can increase the bioactivity of plant glucosides through the release of acylglycone.


Introduction
Phytochemicals are found in leaves, fruits, vegetables, grains, and beans. Some glycosidic phytochemicals have been used in traditional medicine for centuries [1]. For example, the phenolic β-glucoside arbutin is a component of Arctostaphylos uva-ursi (bearberry leaf), which has been used in urinary tract infections [2]. Other plant derived β-glucosides include amygdalin (naturally occurring in almonds), esculin (dandelion coffee), fraxin (kiwi), polydatin (grapes), sinigrin (broccoli), and vanillin (vanilla) [3]. The biological effects of many glycosides are not attributed to their glycoside forms but to the corresponding aglycones ( Figure 1). Aglycones are bioactive compounds that have lower molecular weight and hydrophilicity. After consumption, plant glucosides can be either taken up in the small intestine and undergo enterohepatic circulation, or they can be hydrolyzed by glycosidic activity of the gut microbiota [2]. Bacterial β-glucosidases, which have been classified within glycoside hydrolases (GH) families GH1, GH3, GH5, GH9, GH30, and GH116, cleave β-D-glucosidic linkages liberating glucose and the corresponding acylglycones. Some acylglycones have been shown to be antimicrobially active [4]. GH116, cleave β-D-glucosidic linkages liberating glucose and the corresponding acylglycones. Some acylglycones have been shown to be antimicrobially active [4]. Several taxa of the major gut colonizers Firmicutes, Bacteroidetes and Actinobacteria possess βglucosidase activity [5]. Bifidobacterium spp. (Actinobacteria) represent an important group of human commensals, being among the first microbial colonizers with considerable relevance for health in later life [6,7]. Intestinal competitiveness of bifidobacteria is attributed to their ability to degrade and metabolize a diversity of carbohydrates, and to carbohydrate resource sharing and cross-feeding [8,9]. Host adaptation seems to be linked to the ability to use dietary or host-derived glycans in a glycan-rich gut environment and differs between species and strains [10,11]. Bifidobacterium spp. frequently possess the potential to encode a wide array of glycosyl hydrolases, including βglucosidases [12].
In humans, prevalence and diversity of members of the genus Bifidobacterium change in succession during life, with the species Bifidobacterium longum subsp. infantis and Bifidobacterium bifidum representing about 80% of the intestinal microbiota of infants, while Bifidobacterium adolescentis, Bifidobacterium longum subsp. longum, Bifidobacterium breve, Bifidobacterium catenulatum, and Bifidobacterium pseudocatenulatum prevail in adults representing about 1% of gut microbes [13][14][15][16][17][18]. Colonization of B. bifidum and B. breve seems to be limited to the human intestinal tract [12,19], whereas B. adolescentis, B. longum subsp. longum, Bifidobacterium animalis subsp. animalis and lactis, B. catenulatum, and B. pseudocatenulatum are considered multi-host species that were isolated from other mammals such as dogs, primates, and young ruminants on the milk diet [11,20]. Bifidobacterium dentium likely colonizes the oral cavity, and might only transiently pass the intestine [21,22]. Several taxa of the major gut colonizers Firmicutes, Bacteroidetes and Actinobacteria possess β-glucosidase activity [5]. Bifidobacterium spp. (Actinobacteria) represent an important group of human commensals, being among the first microbial colonizers with considerable relevance for health in later life [6,7]. Intestinal competitiveness of bifidobacteria is attributed to their ability to degrade and metabolize a diversity of carbohydrates, and to carbohydrate resource sharing and cross-feeding [8,9]. Host adaptation seems to be linked to the ability to use dietary or host-derived glycans in a glycan-rich gut environment and differs between species and strains [10,11]. Bifidobacterium spp. frequently possess the potential to encode a wide array of glycosyl hydrolases, including β-glucosidases [12].
Bifidobacterium spp. are common inhabitants of the human intestinal tract throughout life, and intestinal bifidobacterial β-glucosidase activity might modify the bioactivity and bioavailability Table 1. Bifidobacterial utilization of β-glucosylated substrates during growth, and β-glucosidase activity. The ability to utilize plant glucosides was tested in API 50 CHL media. The color change detection of cultivation media after 72 h of incubation at 37 • C under anaerobic conditions was determined spectrophotometrically (434 nm/588 nm). A ratio of <2 was considered no growth (-); 2.1-2.4: poor growth (+); 2.5-3.4: growth (++); and >3.5: very good growth (+++). Beta-glucosidase activity was tested by enzymatic assay with 4-nitrophenyl β-D-glucopyranoside (β-GLU) as substrate with spectrophotometric measurement at 405 nm; the difference of >0.1 was considered positive (+). The release of the acylglycone from esculin was determined visually using ammonium iron citrate as scavenger (ESN rel.).
Overnight cultures were centrifuged and re-suspended in the same volume of API medium. Strains were added (20 µL) to 180 µL API medium with or without added substrate. Plates were incubated under anaerobic conditions (GENbag anaer, bioMérieux, France) at 37 • C for 72 h. The color change from purple to yellow indicated a positive reaction. We measured absorbance at 434 nm and 588 nm using a Tecan Infinite M200 spectrometer (Tecan Group, Männedorf, Switzerland) and calculated the ration 434/588 [26], which was categorized as: no growth (<2); poor growth (2.1-2.4); growth (2.5-3.4); and very good growth (>3.5). Every strain was tested two or three times.

Determination of Whole Cell β-Glucosidase Activity, and Release of the Acylglycone Esculetin
Beta-glucosidase activity of whole cells was assessed by enzymatic assay with 4-nitrophenyl β-D-glucopyranoside (PNP-G; Sigma-Aldrich, USA) as substrate. All samples were tested at least twice. Overnight cultures (1 mL) were centrifuged, supernatant was discarded and cell pellets were frozen at -20 • C. Frozen cells were re-suspended in 20 µL BifiBuffer (1.2 g L −1 K 2 HPO 4 , 0.333 g L −1 KH 2 PO 4 ; Lachner, Czechia), 1 µL of this suspension was added to 99 µL of PNP-G solution (20 mM in Bifibuffer). Absorbance at 405 nm was measured before and after 4 h of incubation at 37 • C using a Tecan Infinite M200 spectrometer: reactions with a difference of absorbance >0.1 units were considered positive.
The release of esculetin from esculin was tested using ammonium iron citrate (Sigma Aldrich, USA) as scavenger, the reaction of esculetin with ferric ions changes the color from purple to opaque black. Bifidobacteria were inoculated in API medium supplied with 10.2 g L −1 esculin (corresponding to 28 mM solution of glucose) and 1 g L −1 ammonium iron citrate in microtiter plates as described above; and were incubated under anaerobic conditions (GENbag anaer) at 37 • C for 72 h. The color change was assessed visually. Every strain was tested at least twice.

Antibacterial Activity of Arbutin and Hydroquinone against Selected Bifidobacterium Strains
The antibacterial activity of arbutin and its acylglycone hydroquinone (Sigma-Aldrich, St. Louis, MI, USA) was tested using two-fold broth dilution assay in 96-well sterile microtiter plates. Overnight  (Table 2), grown in WSP broth, were centrifuged, and the cell pellet was resuspended in the same volume of API medium supplied with 14 mM glucose. A two-fold dilution series was prepared in microtiter plates using API stock solutions containing 14 mM glucose, and 28 mM arbutin or hydroquinone. Cultures (10%) were added, and plates were incubated under anaerobic conditions (GENbag anaer) at 37 • C for 24 h. Absorbance at 434 and 588 nm was determined using a spectrophotometer and the ratio of 434 nm/588 nm was calculated as described above. The minimal inhibitory concentration (MIC) was defined as the concentration that prevented growth, metabolite formation and thereby color change of the API medium. Every strain was analyzed at least three times.
To test whether the presence of arbutin in API medium impacted growth, we additionally conducted growth kinetics of selected strains that were able or lacked the ability to grow with arbutin in API medium supplied with 28 mM glucose, 14 mM glucose and 14 mM arbutin, or 28 mM arbutin to ensure the availability of the same concentration of glucose. Strains were grown in microtiter plates as described before, and absorbance was measured at 0, 3, 6, 9, 12, 24, 30, 36, and 48 h at 434 and 588 nm, to calculate the ratio of 434 nm/588 nm as described above. Table 2. Minimal inhibitory concentrations of hydroquinone and arbutin. Minimal inhibitory concentrations (MIC) were determined using a two-fold dilution assay in microtiter plates and API medium supplied with glucose (14 mM), and hydroquinone or arbutin. The MIC was defined as the concentration that completely inhibited growth of strains determined using the absorbance ration 434 nm/588 nm. MIC were tested in 3-5 independent replicates.   1) and B. dentium DSM 20436 (FNSE01000001.1) using blastP. To identify additional β-glucosidases, genomic data were obtained from NCBI and were annotated with RAST using default settings [29]. Glycosyl hydrolases of family 1 and 3 were identified using the dbCAN database based on a search for signature domains of every CAZyme family [30].

Growth in the Presence of β-Glucosides
We observed that β-glucosidase activity is a common yet species dependent feature of Bifidobacterium spp. To investigate whether β-glucosidase activity relates to the ability to use plant glucosides, we grew strains with esculin, amygdalin, and arbutin as a sole carbohydrate source in API 50 CHL medium ( Table 1). All strains were able to grow in the presence of glucose, verifying the suitability of the assay.
Strains belonging to B. dentium were most versatile in the utilization of the provided β-glucosides as all strains grew in the presence of amygdalin and esculin. Only three B. dentium strains (DSM 20436, TH1, and VOK II) were not able to use arbutin. All strains of B. breve used amygdalin and esculin (with one exception, B43), while the utilization of arbutin was less frequent (46%). Within B. catenulatum/pseudocatenulatum, 83% strains grew in the presence of amygdalin, while 67% utilized esculin and 38% arbutin. The majority of the B. adolescentis strains (69%) was capable of using amygdalin, strains B35 and B39 grew in presence of arbutin and esculin. For B. animalis, we observed subspecies dependent differences in substrate utilization. The majority of B. animalis subsp. lactis strains, except three isolates from dog feces (P2N1, 43/7nb, and 11/6a), utilized amygdalin (80%), 67% used esculin, and only PEG084 grew in the presence of arbutin. In contrast, B. animalis subsp. animalis strains were not capable to grow with amygdalin and arbutin, while 57% utilized esculin. None of the B. longum strains was able to use esculin and arbutin, amygdalin utilization was detected for B. longum DSM 20211 and PEG080, while B. bifidum strains were not able to utilize any of the provided plant β-glucosides.

Release of the Acylglycone Esculetin
We qualitatively determined whether β-glucoside hydrolysis of esculin would lead to the release of the acylglycone esculetin using ammonium iron citrate as a scavenger ( Table 1). The majority (94%) of the strains that were positive in the PNP-G enzymatic assay released esculetin confirming β-glucosidase activity. Few strains (namely DSM 20211, B1, 32/3na, 33/5nb, and 33/4nc), all from B. longum, were PNP-G positive, while esculetin release was not detected. More than half (66%) of the strains that were able to release esculetin grew when esculin was present as substrate.

Metabolite Formation during Growth of Bifidobacteria in the Presence of Plant Glucosides
For representative strains from different species, we investigated metabolite formation as an indicator of fermentation activity. Lactate, acetate, and formate concentrations were measured by capillary ion-exchange chromatography with suppressed conductivity detection ( Figure 2). All strains grew in the presence of glucose, producing mainly acetate metabolite (44%-83% of metabolites formed), followed by lactate (3%-51%) and formate (0%-22%). In negative controls (API medium without added glycan source), and samples without visible growth, acetate was formed as the major metabolite, otherwise the acetate:lactate:formate ratios differed between strains, and compared to growth in glucose.

Figure 2.
Metabolites formed during growth in the presence of glucose or plant glucosides. Formation of fermentation metabolites, lactate (dark grey), acetate (light grey), and formate (black), by selected strains grown in API 50 CHL medium supplied with plant glucosides (equivalent to 28 mM glucose) was analyzed after 72 h incubation by capillary ion-exchange chromatography with suppressed conductivity. The proportion of major metabolites lactate, acetate, and formate for each growth condition are shown above the bar. Strains were tested two or three times. glc, glucose; esc, esculin; escF, esculin and ammonium iron citrate; amy, amygdalin; arb, arbutin; and con, negative control.

Antibacterial Activity of Hydroquinone and Arbutin and Impact on Growth Kinetics
We tested the antibacterial and growth affecting activity of arbutin and its acylglycone hydroquinone on representative strains using two-fold dilution assay in microtiter plates and growth kinetics, respectively. Glucose (14 mM) was added to API medium to avoid growth inhibition due to substrate limitation. None of the strains were affected by the presence of arbutin even at the maximum concentration tested (25.5 mM) ( Table 2)  Metabolites formed during growth in the presence of glucose or plant glucosides. Formation of fermentation metabolites, lactate (dark grey), acetate (light grey), and formate (black), by selected strains grown in API 50 CHL medium supplied with plant glucosides (equivalent to 28 mM glucose) was analyzed after 72 h incubation by capillary ion-exchange chromatography with suppressed conductivity. The proportion of major metabolites lactate, acetate, and formate for each growth condition are shown above the bar. Strains were tested two or three times. glc, glucose; esc, esculin; escF, esculin and ammonium iron citrate; amy, amygdalin; arb, arbutin; and con, negative control.

Antibacterial Activity of Hydroquinone and Arbutin and Impact on Growth Kinetics
We tested the antibacterial and growth affecting activity of arbutin and its acylglycone hydroquinone on representative strains using two-fold dilution assay in microtiter plates and growth kinetics, respectively. Glucose (14 mM) was added to API medium to avoid growth inhibition due to substrate limitation. None of the strains were affected by the presence of arbutin even at the maximum concentration tested (25.5 mM) ( Table 2)

Discussion
Plant derived β-glucosides encompass structurally diverse compounds, which, when ingested, can reach the colon and be enzymatically modified by gut microbes. Beta-glucosidases are widely distributed in gut microbes and play important roles in biological processes. Here, we demonstrate species and strain dependent variability of Bifidobacterium spp. in the utilization of dietary plant glucosides linked to aromatic aglycones.

Genomic Potential for β-Glucosidase Activity Partly Predicts Activity
Based on genomic data, strains of the phylogenetically closely related species B. adolescentis, B. catenulatum, B. pseudocatenulatum, B. dentium, and the more distant B. breve [10], harbored a core of four β-glucosidases. In agreement, all strains of these species possessed β-glucosidase activity but showed preference for different substrates. Indeed, the purified β-glucosidases of B. pseudocatenulatum IPLA 36007 varied in their ability to release of the aglycones daidzein and genistein from isoflavone glycosides daidzin and genistin, indicating enzyme dependent substrate preference [28].
Strains of B. longum did not show β-glucosidase activity despite the presence of genes encoding β-glucosidases with high homology to a purified β-glucosidases of B. longum H1, which hydrolyzed arbutin. Lack of β-glucosidase activity of whole cells might suggest that enzymes were either not expressed or expressed at concentrations not sufficient to confer activity at test conditions.

Discussion
Plant derived β-glucosides encompass structurally diverse compounds, which, when ingested, can reach the colon and be enzymatically modified by gut microbes. Beta-glucosidases are widely distributed in gut microbes and play important roles in biological processes. Here, we demonstrate species and strain dependent variability of Bifidobacterium spp. in the utilization of dietary plant glucosides linked to aromatic aglycones.

Genomic Potential for β-Glucosidase Activity Partly Predicts Activity
Based on genomic data, strains of the phylogenetically closely related species B. adolescentis, B. catenulatum, B. pseudocatenulatum, B. dentium, and the more distant B. breve [10], harbored a core of four β-glucosidases. In agreement, all strains of these species possessed β-glucosidase activity but showed preference for different substrates. Indeed, the purified β-glucosidases of B. pseudocatenulatum IPLA 36007 varied in their ability to release of the aglycones daidzein and genistein from isoflavone glycosides daidzin and genistin, indicating enzyme dependent substrate preference [28].
Strains of B. longum did not show β-glucosidase activity despite the presence of genes encoding β-glucosidases with high homology to a purified β-glucosidases of B. longum H1, which hydrolyzed arbutin. Lack of β-glucosidase activity of whole cells might suggest that enzymes were either not expressed or expressed at concentrations not sufficient to confer activity at test conditions. B. animalis subsp. lactis and B. animalis subsp. animalis harbored highly similar GH1 and GH3 β-glucosidases and possessed β-glucosidase activity. The similar GH1 β-glucosidase Bgl572 hydrolyzed PNP-G and arbutin when purified [31]. However, growth in the presence of plant glucosides differed between B. animalis subspecies, possibly due different transport mechanism or sensitive towards the released acylglycones. Indeed, B. animalis subsp. animalis DSM 20104 was more sensitive towards hydroquinone than B. animalis subsp. lactis DSM 10140.

Bifidobacterium β-Glucosidase Increases Bioactivity and Bioavailability of Plant Glucosidases and Acylglycones
Plants are used for antibacterial properties in medical applications, and bacterial β-glucosidase activity leads to the release of bioactive acylglycones. Indeed, we confirmed the release of esculetin from esculin, by almost all strains with β-glucosidase activity, modifying bioactivity and bioavailability. Despite the ability to hydrolyze esculin, not all strains were able to grow when esculin was supplied as substrate, which might be due to the antimicrobial activity of esculetin [33].
Bioactivity of plant glucosides is likely lower than of acylglycones due to larger molecular weight and higher hydrophobicity. In confirmation, arbutin at the concentrations tested did not show inhibition while hydroquinone conferred strong antibacterial activity against the Bifidobacterium strains tested. Hydroquinone MIC values of bifidobacteria ranging from ≤0.05-0.2 mM were lower than reported for Staphylococcus aureus (1-11 mM) [34,35] and various aerobic Gram-positive and -negative strains [4] including Enterococcus faecalis, Escherichia coli, Pseudomonas aeruginosa, and Klebsiella pneumonia (1.5-6 mM).
MIC were up to 10-fold lower than the levels of hydroquinone theoretically released during growth. However, in the MIC assay, a low concentration of bacterial cells is exposed to high concentrations of the antibacterial at the beginning of the growth phase, while in the growth assay, increasing amounts of hydroquinone face an increasing number of bacterial cells. Sensitivity of bifidobacteria to hydroquinone, which, if released in the intestinal tract, could reduce the growth potential of β-glucosidase active strains, but also of neighboring cells.
In addition, β-glucosidase activity of Bifidobacterium spp. increased the bioavailability of acylglycones. Hydroquinone has been linked to anticancer activity in vitro and in vivo [36] and has been shown to confer antimycobacterial and antileishmanial activity in vitro [37].

Niche Adaption of Bifidobacteria Seems Related to β-Glucosidase Activity
Host adaptation of Bifidobacterium spp. was suggested to be linked to the ability to use dietary or host-derived glycans and differs between species and strains [10,11]. Indeed, strains of B. bifidum and B. longum subsp. infantis, which are among the most abundant microbes during the first months of life, lacked β-glucosidase activity in agreement to previous observations [5], and with the absence of β-glucosidase encoding genes in the genomes. Both species occur in the infant gut when glycan substrates are limited to human breast milk, endogenous mucin, or infant formula. Indeed, a previous cohort of studies observed that fecal β-glucosidase activity was low after birth and gradually increased with the introduction of a more diverse diet [38]. Interestingly, B. animalis subsp. animalis showed only little growth in the presence of plant glucosides despite possessing β-glucosidase activity. Most isolates were obtained from lamb and calf feces early when animals received a milk diet.
B. breve colonizes infant and adults, and similar to B. adolescentis and other adult or multi-host species, possesses multiple β-glucosidases. In the adult gut, Bifidobacterium spp. contribute a minor share of the population, and compete with other β-glucosidase positive taxa for substrate [5]. Beta-glucosidase activity of Bifidobacterium strains colonizing adults might enhance ecological competitiveness.
For the multi-host subspecies of B. animalis subsp. lactis, plant glucoside utilization profiles likewise suggested host adaption in agreement with genetic and phenotypic host-specific differences previously observed [11]. Strains from adult ruminant hosts (Cameroon sheep, Barbary sheep, and an okapi), that naturally receive a plant-based diet, used plant glucosides in contrast to strains isolated from omnivorous dogs.

Substrate Source Impacted Fermentation Profiles
Bifidobacteria metabolize hexoses via the "bifid shunt" with fructose-6-phosphoketolase as the key enzyme. Glucose (1 mol) theoretically yields 1.5 mol acetate, 1 mol lactate, and 2.5 ATP [39]. Whether the intermediate pyruvate is cleaved to acetyl phosphate and formate, or reduced to lactate, depends on type and supply of the substrate carbohydrate [40], possibly in the presence of oxygen [8] and on different rates in substrate consumption. A previous study reported that with a decrease of consumption rate, relatively more acetic and formic acid and less lactic acid was produced by Bifidobacterium spp. [41]. Indeed, the proportion of lactate was reduced for most strains grown in the presence of esculetin, and to a lesser extent with arbutin, which might indicate that hydrolysis activity reduced the consumption rate.

Conclusions
Our data shows that bifidobacterial β-glucosidase activity is preserved among species which might be related to niche adaption. Structural homology of a core set of β-glucosidases of species associated with adult humans could suggest that these enzymes evolved together. Beta-glucosidase activity may provide a competitive advantage in the mammalian gut proving access to energy sources, but they might also have environmental impact due to the release of bioactive antibacterial acylglycones; thus, increasing bioavailability due to the formation of fermentation metabolites.