Introduction

Glucosylceramide (GluCer) is comprised of a long-chain sphingoid base backbone including sphingosine, an amide-linked fatty acid (FA), and a glucose headgroup. The FA-linked sphingoid is called ceramide. Ceramides and sphingoids are bioactive substances that have been shown to exert various effects on animals (Hannun et al. 1986; Okazaki et al. 1989; Hannun and Linardic 1993; Hannun 1994; Kolesnick and Golde 1994; Ballou et al. 1996; Spiegel and Merrill 1996; Tokura et al. 1996). GluCer and sphingomyelin (SM), precursors of ceramides found in foods (Vesper et al. 1999), are suggested to have both preventive and suppressive effects on colon cancer (Dillehay et al. 1994; Schmelz et al. 1996, 2000; Lemonnier et al. 2003). Ceramides and sphingoids have also been reported to induce apoptosis in cell lines derived from human colon cancer (Ahn and Schroeder 2002). Dietary GluCer and SM are likely to be degraded to ceramides and sphingoids in the intestine, which are then absorbed (Nilsson 1968, 1969; Schmelz et al. 1994). Therefore, prevention and suppression of colon cancer by GluCer and SM may in part be due to apoptosis of colon cells mediated by ceramides and sphingoids.

We reported previously that dietary SM alleviated inflammatory bowel disease (IBD) in mice (Furuya et al. 2008), suggesting that ceramides and/or sphingoids derived from dietary SM alleviate IBD. A similar effect can be expected when GluCer is consumed, but dietary GluCer is barely hydrolyzed to ceramide or sphingoid in the intestinal tract of rats (Nilsson 1969). When various GluCer-containing plant foods were fed to mice, 50–90% of GluCer was excreted in feces (unpublished observations). In vitro and in situ experiments with the ligated intestines of rats showed that virtually no GluCer is absorbed unless it is hydrolyzed to ceramide (unpublished observations). Therefore, enhancement of GluCer hydrolysis by utilizing intestinal bacteria, e.g., as a probiotic, might improve the effect of GluCer.

So far, two GluCer-hydrolyzing bacteria have been isolated from soil. One (Paenibacillus sp.) hydrolyzed synthetic GluCer having a short acyl moiety, but the ability to hydrolyze GluCer having a long acyl moiety was low (Sumida et al. 2002). These results suggest that the activity of this bacterium to hydrolyze plant GluCer is low, because the acyl group in plant GluCer is generally long (Bohn et al. 2001). The other (Rhodococcus sp.) was also able to hydrolyze GluCer, but the activity was extremely low (Ito and Yamagata 1986, 1989; Ishibashi et al. 2007). Although the traditional culture techniques and the recent development of rRNA sequencing analysis have revealed that intestinal microbiota is highly diverse, no GluCer-hydrolyzing bacterium has been isolated from gut or feces. Thus, it is desirable to find intestinal bacteria that are capable of hydrolyzing plant GluCer, because many plant foodstuffs contain GluCer (Vesper et al. 1999).

This paper reports that a bacterial strain hydrolyzing plant GluCer at a high rate was isolated from dog feces. Based on the morphological, physiological, and biochemical characteristics, and the phylogenetic analysis of the gene sequence from 16S rRNA, the isolate was assigned to a new species in the genus Blautia. Recently, the genus Blautia was newly proposed, and several species belonging to the genus Ruminococcus and Clostridium were reclassified into the genus Blautia (Liu et al. 2008). The identification of the bacterial isolate and the description of a new species are reported here.

Materials and methods

Bacterial strains and culture conditions

A newly isolated bacterial strain described below (Blautia glucerasei sp. nov. HFTH-1T = DSM 22028T = NBRC 104932T) was used for this study. For the isolation of bacteria, a growth medium consisting of clarified ruminal fluid (Ogimoto and Imai 1981) and a basal medium (1:3, v/v) was used. The basal medium contained 0.45 (g/l) of K2HPO4, 0.45 of KH2PO4, 0.9 of (NH4)2SO4, 0.9 of NaCl, 0.12 of CaCl2·2H2O, 0.19 of MgSO4·7H2O, 1.0 of Trypticase-peptone (BBL, Becton–Dickinson, Cockeysville, MD), 1.0 of yeast extract (Difco Laboratories Inc., Detroit, MI), 1.0 of glucose, 1.0 of cellobiose, 1.0 of maltose, and 0.6 of cysteine-HCl (pH 7.0). Growth was estimated by optical density at 600 nm (OD600) of diluted cultures. Peptone-yeast extract-glucose (PYG) medium was used for routine culture of the isolated bacterium.

Isolation of bacteria

As a source of intestinal bacteria, freshly voided feces were collected from 10 animals each of mice, rats, rabbits, pigs, goats, cats, and dogs. The fecal samples were immediately put into vials on ice under a stream of CO2, frozen at −80°C within a few hours, and subjected to experiments within a month. Feces were suspended in 50 mM potassium phosphate (KPi) buffer (pH 7.0) prepared anaerobically under a stream of CO2 and inoculated into the growth medium described above containing GluCer (250 mg/l). After overnight incubation, GluCer hydrolysis was estimated. Then, GluCer-hydrolyzing bacteria were isolated from the feces having high GluCer-hydrolyzing activity by the Hungate’s role-tube method (Miyazaki et al. 1992) using the above-mentioned growth medium consisting of clarified ruminal fluid and a basal medium (1:3, v/v).

Evaluation of GluCer-hydrolyzing activity

Partially purified GluCer from maize germ was provided by Tsuji Oil Mill co., Ltd. (Matsuzaka, Mie, Japan) and purified to more than 80% purity (most of the remaining 20% consisted of sterylglycoside) by acetone precipitation after hydrolysis with 0.4 N methanolic KOH (38°C, 2 h). Isolated bacteria were cultured at 37°C for 24 h in the growth medium described above, supplemented with 250 mg/l of the purified GluCer. Then, thin-layer chromatography (TLC) was employed for sphingolipid analysis to deal with many samples; i.e., ceramide and sphingoid bases produced from GluCer were separated by TLC and quantified by colorimetry of the sphingoid base as follows. Each culture (3 ml) was shaken vigorously with 6 ml of chloroform–methanol (1:1, v/v; C-M). The chloroform layer separated clearly by centrifugation (1,500×g, 10 min) was harvested, and then concentrated under a stream of N2. The concentrated lipid fraction was separated by TLC (Silica gel 60, Merck, Darmstadt, Germany) with a developing solvent system of chloroform-acetone (7:3, v/v). The spots corresponding to ceramide and sphingoid (Sigma–Aldrich), judged from the Rf values and sphingolipid-specific coloration (Beiss 1964), were scraped, and put together into a tube. Lipids were extracted with 6 ml of C-M (2:1), dissolved in 1 ml of a solution containing methanol and 16.5% HCl (82:18, v/v), and then hydrolyzed by heating at 100°C for 2 h. After adjustment of the pH to 10–12 with 4 ml of 0.5 N NaOH, the reaction mixture was vigorously shaken with 5 ml of ethyl acetate, and the ethyl acetate layer separated by centrifugation (1,500×g, 10 min) was collected. To the ethyl acetate solution, 2 ml of 10 mM acetate buffer (pH 3.7) and 0.1 ml of 0.5% (w/v) methyl orange were added, and then the mixture was shaken and centrifuged again. The sphingoid–methyl orange complex in the supernatant ethyl acetate phase was measured spectrophotometrically at 415 nm against a blank carried through from the neutralization step (Lauter and Trams 1962).

Hydrolysis of GluCer by a bacterial isolate

A bacterial isolate (strain HFTH-1T) was grown in 30 ml of PYG medium containing 570 μM of purified maize GluCer (Purity: more than 99%; presented by Tsuji Oil Mill co., Ltd.) at 37°C. Two milliliters of cultures were harvested at different time after inoculation, and GluCer hydrolysis was examined by quantifying sphingolipids. In this case, high-performance liquid chromatography (HPLC) was used to quantify trace amounts of GluCer, ceramide, and sphingoid. After extraction of total lipids with C-M (2:1), the total lipids were dissolved in C-M (4:1, HPLC grade) and filtered through a membrane filter (0.45 μm, Millipore, Billerica, MA). The filtrate (10 μl) was injected into the HPLC system (Shimadzu Co. Ltd, Kyoto, Japan) equipped with a silica gel column (250 mm × 4.6 mm i.d., 5 μm spherical particle size; Shimadzu) and an Evaporative Light Scattering Detector, and eluted with a linear gradient of solvent A (chloroform) and solvent B (methanol + 2% aqueous ammonia, 95:5, v/v). The gradient was from solvents A plus B (9:1) to solvent B.

The activity of GluCer-hydrolyzing enzyme (glucosylceramidase; GluCerase) in cultures was assayed as follows. Strain HFTH-1T was grown in PYG medium for 8 h (OD600 value of 6.8), and cultures were centrifuged (13,000×g, 15 min). The supernatant was adjusted to pH 6.5 with 8% (w/v) Na2CO3 and subjected to enzyme assay. The cell pellet was resuspended in the same volume of 50 mM KPi (pH 6.5) before enzyme assay. The assay mixture consisted of 1 ml of 50 mM KPi (pH 6.5) containing 100 μM of GluCer (above 99% purity) and 0.2% (w/v) sodium taurocholate and 1 ml of enzyme sample; i.e., culture supernatant or cell suspension. The assay mixture was incubated at 45°C for 5–10 min, and then ceramide was determined by HPLC as described above. Enzyme activity was expressed as the activity per bacterial cell dry weight, which represents the amount of enzyme produced per cell.

Hydrolysis of different glycosphingolipids by strain HFTH-1T

Strain HFTH-1T was grown at 37°C for 10 h in 3 ml of PYG medium containing 200 nmol of soybean GluCer (Over 99% purity, Avanti Polar Lipids, Inc., Alabaster, AL), glucosylsphingosine (GluSph; Avanti Polar Lipids, Inc.), lactosylceramide (LacCer; Avanti Polar Lipids, Inc.), and monosialoganglioside GM3 (GM3; HyTest, Ltd., Turku, Finland), respectively. After incubation, ceramide, GluCer, GluSph, LacCer and GM3 in cultures were quantified by HPLC as described above.

Phylogenetic analysis of the gene sequence from 16S rRNA

The gene from 16S rDNA of strain HFTH-1T was amplified from the purified genomic DNA by PCR with the bacterial universal primers 27f (5′-AGAGTTTGATCCTGGCTCAG-3′, corresponding to nucleotide position 8 to 27 of E. coli 16S rRNA) and 1525r (5′-AAAGGAGGTGATCCAGCC-3′, corresponding to nucleotide position 1543 to 1525 of E. coli 16S rRNA), which had been reported by Lane (1991). The amplified product was purified with QIAquick spin PCR purification kit (Qiagen GmbH, Hilden, Germany). The gene sequence was determined with a BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA) and the ABI PRISM 310 Genetic Analyzer (Applied Biosystems). Sequences were aligned using CLUSTAL X (Version 1.83) software (Thompson et al. 1997) and adjusted manually. Atopobium minutum NCFB 2751T (X67148) was used as the outgroup. Evolutionary distances were calculated with Kimura’s two parameter model, and tree was constructed in NEIGHBOR by the neighbor-joining method (Saitou and Nei 1987). The confidence values of branches were determined by performing a bootstrap analysis (Felsenstein 1985) with 1,000 replicates.

The sequence of the gene from 16S rRNA of strain HFTH-1T was deposited in the GenBank/DDBJ under an accession number of AB439724.

Morphological, physiological, and biochemical characteristics of strain HFTH-1T

Strain HFTH-1T was grown in PYG medium at 37°C for 72 h and stored at 4°C for 1 week to check for cellular morphology by Gram staining according to the procedure with Hucker’s modifications (Gerhardt 1994). Spore formation was examined by the Wirtz method and tolerance to heat (85°C, 10 min) (Bergey et al. 1984). To examine fermentation products from glucose, strain HFTH-1T was grown at 37°C for 8 h in the growth medium described above in which maltose and cellobiose were excluded and glucose was increased to 5.0 g/l (Medium A), and then organic acids were quantified by HPLC as described previously (Hino et al. 1994). The capacity to utilize various carbohydrates was evaluated by growth (OD600) in Medium A in which glucose was replaced by either one of aesculin, arabinose, avicel, cellobiose, fructo-oligosaccharide, fructose, galactose, inulin, lactose, maltose, mannitol, mannose, pectin, potato starch, raffinose, sorbitol, sucrose, xylose, and xylan (3 g/l).

The G + C content in DNA was determined by HPLC after digestion of isolated DNA with P1 nuclease and alkaline phosphatase (Kamlage et al. 1997). The amino acid compositions in cell wall peptidoglycan were analyzed by HPLC (Waters, MA) as described previously (Takahashi et al. 1989), and the peptidoglycan type was determined as described by Schleifer and Kandler (1972). Cellular fatty acids were extracted and analyzed according to the standard protocol of the Sherlock Microbial Identification System Version 6 (MIDI, DE).

Activities of nineteen enzymes were assayed by the API ZYM (bioMérieux, Lyon, France) according to the manufacturer’s instructions. Lecithinase and lipase activities were assayed as follows. Cells were grown in PYG medium supplemented with lecithin (phosphatidylcholine; PC) or triolein (as triglyceride; TG) at 37°C for 24 h, and total lipids were extracted from each culture as described above. The lipids were separated by TLC using the developing solvent system of chloroform–methanol-H2O-acetate (65:25:4:1, v/v) for PC analysis, or C-M (95:5, v/v) for TG analysis. Spots were detected by spraying 50% sulfuric acid and heating at 120°C for 15 min. The TLC plates were photographed by Personal Density Scanning Imager (Molecular Dynamics, Sunnyvale, CA), and the amounts of PC and TG were estimated by densitometry using ImageQuaNT (Molecular Dynamics, Sunnyvale, CA).

Results

Isolation of GluCer-hydrolyzing bacteria

When feces from 10 animals each of mice, rats, rabbits, pigs, goats, cats, and dogs were incubated overnight with GluCer, high GluCer-hydrolyzing activity was found only in one canine feces. Two different GluCer-hydrolyzing bacterial strains were isolated from more than 600 colonies prepared from the canine feces. Of the two isolates, one showed nearly three times higher GluCer-hydrolyzing activity than the other for both maize germ GluCer and soybean GluCer (data not shown). The isolate having higher activity was designated as strain HFTH-1T and used for the experiments described below.

Hydrolysis of glycosphingolipids in growing cultures of strain HFTH-1T

When strain HFTH-1T was grown in GluCer-containing PYG medium, GluCer was readily hydrolyzed, and virtually 100% of GluCer was converted to ceramide in 11 h (Fig. 1). No sphingoid, a degradation product of ceramide, was detected, indicating that strain HFTH-1T does not degrade ceramide. Approximately 80% of GluCerase activity was found in the culture supernatant (2.3 nmol/min/mg of cell dry weight), whereas nearly 20% activity was in the cell fraction (0.6 nmol/min/mg of cell dry weight). Strain HFTH-1T was unable to hydrolyze GluSph, LacCer, or GM3 (data not shown).

Fig. 1
figure 1

Hydrolysis of GluCer by strain HFTH-1T. Bars indicate SD (n = 2). Symbols indicate growth (open circle), GluCer (filled circle), and ceramide

Full size image

Identification of strain HFTH-1T

To establish the phylogenetic position of strain HFTH-1T, comparative 16S rRNA gene sequence analysis was performed, and almost the entire sequence of the 16S rRNA gene (1,468 bp) was determined. The similarity values of the 16S rRNA gene sequence between strain HFTH-1T and its closely related species (Ruminococcus obeum and seven species of the genus Blautia) within the rRNA cluster XIVa were 92–95%. The highest level of 16S rRNA gene similarity with strain HFTH-1T was 95.3% of the type strain of R. obeum ATCC 29174T. The sequence diversity between strain HFTH-1T and the nearest relative, the type strain R. obeum was 3.0%. The sequence data for strain HFTH-1T was aligned with 32 sets of published data. The 16S rRNA gene sequence (1,160 nt; excluding incomparable nucleotides and spaces) was used for analysis. A phylogenetic tree showed that strain HFTH-1T formed a coherent cluster with R. obeum, Blautia luti, Blautia wexlerae, Blautia schinkii, Blautia hansenii, Blautia producta, Blautia coccoides, and Blautia hydrogenotrophica in the clostridial rRNA cluster XIVa (Collins et al. 1994), that was statistically well supported (bootstrap = 89%) (Fig. 2). The major branching orders were confirmed by using the maximum composite likelihood method (Online Resource 1) (Tamura et al. 2007).

Fig. 2
figure 2

Phylogenetic tree showing the inter-relationships of Blautia glucerasei sp. nov. HFTH-1T with some of its nearest relatives within the rRNA cluster XIVa. The tree was constructed by the neighbor-joining method as described in the text. The numbers at nodes indicate bootstrap values greater than 50% (1000 replicates). Bar 1e+003 substitutions per nucleotide position.

Full size image

The 16S rRNA gene sequence analysis showed that the closely phylogenetic relative to strain HFTH-1T is R. obeum. However, strain HFTH-1T was easily distinguished from R. obeum based on several phenotypic characteristics. In contrast to R. obeum, strain HFTH-1T formed spores and produced β-N-acetylglucosaminidase, α-fructosidase, and α-mannosidase. The characteristics that differentiate strain HFTH-1T from its three most closely related species in the clostridial rRNA cluster XIVa (Fig. 2 and Online Resource 1) are summarized in Table 1.

Table 1 Salient characteristics of an isolated bacterium (HFTH-1T) and the closely related species in the clostridial rRNA cluster XIVa
Full size table

Discussion

A bacterial strain (HFTH-1T) that is capable of hydrolyzing GluCer to ceramide at high rate was newly isolated from canine feces. Strain HFTH-1T hydrolyzed GluCer, but not GluSph, LacCer, or GM3, indicating that the strain has GluCerase that specifically hydrolyzes GluCer. Nearly 80% of GluCerase activity was present in the culture supernatant, and the remaining 20% of GluCerase was considered to be bound to the bacterial cell surface. This result indicates that GluCerase is secreted out of cells. It is conceivable that GluCer is hydrolyzed by extracellular GluCerase in strain HFTH-1T. However, physiological implications of GluCer-hydrolyzing activity in strain HFTH-1T are unknown at present.

As mentioned in the introduction, Paenibacillus sp. strain TS12 isolated from soil produced β-glucosidase, which hydrolyzed synthetic GluCer (the FA moiety was C12:0, and the sphingoid moiety was sphingosine), but the hydrolysis rate decreased with the increase in the length of the acyl moiety (Sumida et al. 2002). In contrast, strain HFTH-1T readily hydrolyzed GluCer extracted from soybean and maize germ, which contain C14–C24 FAs as the acyl moiety and various sphingoids as the sphingoid moiety; e.g., sphingosine, 4-hydroxy-trans-8-sphingenine, trans-4, trans-8-Sphingadienine, and trans-4, cis-8-Sphingadienine. Since all the plant GluCer known so far contain similar FAs and sphingoids (Bohn et al. 2001), it is conceivable that strain HFTH-1T hydrolyzes most plant GluCer at high rates.

It has also been reported that Rhodococcus sp. isolated from soil was able to hydrolyze GluCer, but the reported activity is inferred to be far lower than the activity of strain HFTH-1T (Ito and Yamagata 1986). This implies that strain HFTH-1T is the first example of a GluCer-hydrolyzing microbe that can be put to practical use. Because strain HFTH-1T is an intestinal bacterium and showed no acute toxicity in mouse tests (unpublished observations), strain HFTH-1T could be used as a probiotic to augment the conversion of dietary GluCer to ceramide that is much more readily absorbed from the intestine (Nilsson 1968, 1969; Schmelz et al. 1994). When strain HFTH-1T is orally administered, its spores may survive in the stomach, germinate in the small intestine, and grow in the large intestine, as reported for Clostridium butyricum (Sato and Tanaka 1997). Thus, administration of strain HFTH-1T as a probiotic might be effective for health.

On the other hand, ceramide is the major component of the stratum corneum (SC) (Lampe et al. 1983a, b) and hence has attracted attention to its effect on skin disorders; e.g., dry skin, psoriasis, and atopic dermatitis. It has been reported that the ceramide content in SC is decreased in patients with atopic dermatitis (Melnik et al. 1988; Imokawa et al. 1991; Macheleidt et al. 2002) and psoriasis (Motta et al. 1993, 1995), suggesting that a deficiency of ceramide is involved in these diseases. Thus, to prevent and alleviate skin lesions, it may be desirable to increase ceramide in intercellular lipids among SC corneocytes by percutaneous ceramide administration. However, because no means is available at present to produce ceramide at low cost, GluCer is used as a ceramide source for cosmetics and supplements. Strain HFTH-1T hydrolyzes GluCer to ceramide exclusively, and does not degrade ceramide further. It is presumed that ceramide is more effective than GluCer, and therefore, strain HFTH-1T or its GluCerase may be useful for the industrial production of ceramide from plant GluCer at low cost. GluCer can be obtained from inedible parts or residues of grains and fruits.

The phylogenetic tree inferred from the gene sequence of 16S rRNA suggested that strain HFTH-1T is related to the members in the genus of Blautia (Fig. 2 and Online Resource 1). Because the genus Blautia is proposed and reclassified on the basis of comparative 16S rRNA sequencing and phylogenetic studies (Liu et al. 2008), these data indicate that strain HFTH-1T is affiliated with the genus Blautia. Strain HFTH-1T has the highest 16S rRNA similarity to the type strain of R. obeum (95.3%), but R. obeum does not form spores. Morphological, physiological, and biochemical characteristics obviously discriminate strain HFTH-1T from the related species in the genus of Blautia. In conclusion, these data confirmed that strain HFTH-1T represents a novel species in the genus Blautia described as Blautia glucerasei.

Description of Blautia glucerasei sp. nov

Blautiaglucerasei (glu.cera.sei. N. L. gen. n. glucerasei of GluCerase, having GluCerase). Cells are Gram-positive, motile, strictly anaerobic, lecithinase- and lipase-negative, oval-spore-forming rods, 0.8–1.0 μm in diameter and 2.3–3.0 μm in length. Colonies on PYG agar appear white and circular with a smooth surface and edges (1–1.5 mm in diameter after 1 day of growth). Cells grow well in liquid or solid PYG medium under anaerobic conditions. The temperature range for growth is 25–45°C with optimum growth at 37°C. The optimal initial pH is 7.0. Cells possess a DNA G + C content of 40.7%. Acid is produced from glucose, maltose, galactose, fructose, lactose, inulin, raffinose, arabinose, xylose, xylan, cellobiose, starch, pectin, and fructo-oligosaccharide. The products of glucose fermentation are acetate, formate, and lactate. Cell wall peptidoglycan type of Schleifer and Kandler is A1γ (Schleifer and Kandler 1972). Cellular long-chain fatty acids are of the straight-chain saturated and monounsaturated types, with C16:0 dimethyl acetal acids (ca. 26%), C12:0 (ca. 19%) and C16:0 (ca. 12%) predominating, and C14:0 (ca. 9%), C18:0 dimethyl acetal acids (ca. 7%), C16:0 aldehyde (ca. 5%), C11:0 dimethyl acetal acids (ca. 5%), C14:0 dimethyl acetal acids (ca. 4%), C18:0 (ca. 3%), and C18:1 w9c dimethyl acetal acids (ca. 1%) as the minor components (representing more than 1.00%). Cells possess naphthol-AS-BI-phosphohydrolase, α-galactosidase, β-galactosidase, leucine allyl amidase, acid phosphatase, β-glucosidases, N-acetyl-β-glucosaminidase, α-mannosidase, and α-fucosidase, but do not possess alkaline phosphatase, esterase (C4), esterase lipase (C8), lipase (C14), valine allyl amidase, cystine allyl amidase, trypsin, α-chymotrypsin, β-glucuronidase, or α-glucosidases. GluCer is hydrolyzed to ceramide.

The type strain is HFTH-1T (= DSM 22028T = NBRC 104932T), which was isolated from feces of a healthy dog, in Japan.