7107

Sup lied by U.S. ept. of Agriculture I ationQt Center for Agricultural Utilization es arch, Peoria, Illinois 912 Characterization of the of Butyrivibrio

R.B. HESPELL 1 Biochemistry Research, National Center for Agricultural Utilization Research, Agricultural Research Service, United States Department of Agriculture, Peoria, IL 61604, U.S.A. K. KATO Department of Oral Microbiology, Okayama University Dental School, Okayama, Japan AND J.W. COSTERTON Department of Biological Sciences, The University of Calgary, Calgary, AB T2N 1N4, Canada Received March 31, 1993 Revision received June 1, 1993 Accepted June 7, 1993

HESPELL, R.B., KATO, K., and COSTERTON, J.W. 1993. Characterization of the cell wall of Butyrivibrio species. Can. J. Microbiol. 39: 912-921. Most Butyrivibrio strains have been isolated from the of animals and have been classified as Butyrivibrio fibrisolvens. A few strains isolated from human feces are designated as Butyrivibrio crossatus, the other species in this genus. Butyrivibrio fibrisolvens strains are anaerobic, curved rods that produce butyrate, but numerous studies have shown that these strains display considerable variations in phenotypic properties and heterogeneity in DNA relatedness. Although over 60 strains have been characterized in these respects, the cell wall structure of only a few strains has been studied. In this study, cell wall related properties of 12 strains representative of five DNA relatedness groups were examined. All strains were very sensitive to penicillin and other antibiotics that interfere with cell wall synthesis. Although an occasional resistant strain was found, most strains were sensitive to a variety of protein synthesis antibiotics that included aminoglycosides and tetracycline. In contrast, all strains were highly resistant to nalidixic acid. Peptidoglycans were isolated from seven B. fibrisolvens strains and Lachnospira multiparus. Compositional analyses indicated molar ratios of 0.7:2:2: 1:0.8 for muramic acid, glucosamine, alanine, glutamic acid, and diaminopimelic acid, respectively, in all peptidoglycans, which also showed a low degree of cross-linking. A trichloroacetic acid extractable galactosamine-containing polysaccharide copurified with the Butyrivibrio peptidoglycans. Electron microscopy of thin sections showed all strains to possess a Gram-positive type of cell wall that was atypically thin (12-18 nm). Most strains also displayed external (surface) polysaccharide layers. Cytoplasmic inclusions and granules were evident in many strains and were composed of polysaccharides, on the basis of cell composition analyses. The findings that Butyrivibrio strains have overall similarities in cell wall properties, but differences in DNA relatedness, suggest that these organisms should be classified as several more species in the same genus or family. Key words: Butyrivibrio fibrisolvens, Butyrivibrio crossatus, cell wall, peptidoglycan, ruminal .

HESPELL, R.B., KATO, K., et COSTERTON, J.W. 1993. Characterization of the cell wall of Butyrivibrio species. Can. J. Microbiol. 39 : 912-921. La plupart des souches de Butyrivibrio ont ete isolees du tractus digestif des animaux et ont ete classees comme Butyrivibrio fibrisolvens. Quelques autres souches isolees de matieres fecales humaines ont ete classees Butyrivibrio crossatus, l'autre espece de ce genre. Les souches de B. fibrisolvens sont des batonnets incurves anaerobies producteurs de butyrate mais plusieurs etudes ont montre que ces souches varient considerablement dans leurs proprietes phenotypiques et sont heterogenes au niveau de la parente du DNA. Meme si plus de 60 souches ont ete caracterisees selon ces parametres, la structure de la paroi cellulaire n'a ete etudiee que chez quelques souches. La presente etude s'est interessee aux proprietes reliees a la paroi cellulaire chez 12 souches appartenant acinq groupes apparentes en DNA. Toutes les souches etaient tres sensibles a la penicilline et aux autres antibiotiques qui interferent avec la synthese de la paroi cellulaire. Meme si une souche resistante peut etre retrouvee al'occasion, la plupart des souches sont sensibles aux antibiotiques agissant sur la synthese proteique incluant les aminosides et la tetracycline. Toutes les souches sont par contre fortement resistantes al'acide nalidixique. Les peptidoglycans ont ete isoles chez sept souches de B. fibrisolvens et chez Lachnospira multiparus. La composition indique des ratios molaires de 0.7:2:2: 1:0.8 pour l'acide muramique, Ie glucosamine, l'analine, l'acide glutamique et l'acide diaminopimelique et ce pour tous les peptidoglycans qui presentent d'ailleurs un faible degre de reactions croisees. Un polysaccharide contenant du glucosamine pouvant etre extrait par l'acide trichloroacetique a ete copurifie avec les peptidoglycans de Butyrivibrio. L'observation en microscopie electronique sur des coupes minces a confirme la presence chez toutes les souches d'une paroi de type Gram-positif atypiquement mince (12 a 18 nm). La plupart des souches possedent aussi des couches polysaccharidiques externes dites de surface. Des inclusions ou granules cytoplasmiques se retrouvaient dans plusieurs souches et etaient de nature polysaccharidique selon les analyses de la composition cellulaire. Globalement malgre les similitudes des proprietes des parois cellulaires des souches de Butyrivibrio, les differences au niveau de la parente du DNA suggerent l'ajout dans la classification de quelques autres especes dans Ie meme genre ou la meme famille. Mots cMs : Butyrivibrio fibrisolvens, Butyrivibrio crossatus, paroi cellulaire, peptidoglycan, bacteries du . [Traduit par la redaction]

JAuthor to whom all correspondence should be sent at the following address: USDA-ARS-NCAUR-FBR, 1815 North University Street, Peoria, IL 61604, U.S.A. Printed in Canada I Imprime au Canada HESPELL ET AL. 913

Introduction ...----t~ IH17cl Members of the genus Butyrivibrio can be found in the 12 gastrointestinal tract of mammals and in habitats such as CE51 66% anaerobic digestors. Strains isolated from human feces and J UC12492 81-99%~ Other VC Strains possessing lophotrichous flagellation have been designated as Butyrivibrio crossatus (Moore et al. 1976). The other recognized species is Butyrivibrio jibrisolvens, which has 22%+ monotrichous flagellation, and is one of the more numerous species found in the rumen and cecum of a variety of r--.... ICF3 I 94-99%. Other CF Strains animals. Strains of this species are curved rods that occur singly or in chains and commonly produce butyrate as the 25%+ major fermentation acid with lesser amounts of acetate and lactate. As shown in the initial description of the genus by VV 1 L.----t~ Bryant and Small (1956) and in many subsequent studies, L....._-.... Other VV Strains all B. jibrisolvens strains ferment a wide range of substrates, PI7 but vary considerably in other morphological and physio­ I---.,;;..~~ IT9-40a I logical characteristics. Because of these phenotypic strain differences, proposals have been made to divide B. jibrisolvens strains into groups (Shane et al. 1969) or separate species (Hungate 1966). The cells of all B. jibrisolvens strains characteristically stain Gram-negatively by means of conventional methods. On the other hand, some strains have been shown to contain lipoteichoic acid and glycerol teichoic acids, components typically found in Gram-positive bacterial cell walls (Sharpe et al. 1975). However, the first ultrastructural study of AcTF2 J700/< ~~~--I~GS114 B. jibrisolvens by Cheng and Costerton (1977) showed that 0 strains D 1 (type strain) and C3 possessed a cell wall that was FIG. 1. Grouping of Butyrivibrio strains into DNA relatedness structurally similar to a Gram-positive type, but was very clusters. The grouping is based on DNA-DNA hybridization data thin (12-18 nm), which might account for the loss of the (Mannarelli 1988; Mannarelli et al. 1990) with percent DNA Gram stain complex during decoloration. Large, Buty­ reassociations indicated above the arrows. The boxed strains rivibrio-like bacteria such as strains 2-33 and B385-1 also indicate possible species type strains or currently designated species have been found to have a similar thin peptidoglycan cell type strains (T9-40a, B. crossatus; D 1, B. fibrisolvens). wall structure (Cheng et al. 1989). Recent taxonomic studies have shown that most' 'true" B. jibrisolvens strains have source. For solid media, agar was included to a final concentration DNAs with G + C base contents of 41-42 mole percent and of 1.5%. Plating and related procedures were done at room tem­ can be grouped into about five or more clusters (Fig. 1) on perature, using a glove box having a 75% nitrogen - 20% carbon the basis of DNA-DNA hybridization data (Mannarelli dioxide - 5% hydrogen atmosphere. 1988; Mannarelli et al. 1990). The DNA relatedness between these clusters is less than 60010, suggesting each could repre­ Antibiotic sensitivities sent strains of different Butyrivibrio species. In addition, The appropriate strain was grown to late logarithmic growth phase on RGM medium and the culture (100 mL) was harvested the clustering of strains on the basis of the B. jibrisolvens by centrifugation (6000 x g, 15 min, 15°C). The cell pellet was carbohydrate compositions of the extracellular polysac­ suspended with 4 mL of RGM lacking added carbohydrate. Plates charides made by these strains agrees well with the DNA having bacterial lawns were prepared by adding 0.2 mL of analysis (Stack 1988; Mannarelli et al. 1990). resuspended cells to 4 mL of RGM agar overlay medium containing In view of the fact that limited cell wall information was 1% glucose (melted and cooled to 50°C), briefly mixed, and available for most B. fibrisolvens strains, we undertook stud­ immediately poured onto a Petri dish of RGM agar medium. After ies with strains that were representative of the various DNA solidification and incubation of the plate at 37°C for 2-3 h, groupings. Each strain was examined for sensitivities to antibiotic-impregnated disks (7 mm) were placed on the surface various antibiotics, for the chemical composition of its of the plate. After further incubation (24-36 h), the diameter of isolated peptidoglycan layer, and for cell ultrastructure by the clearing zone in the lawn produced by the disks was measured. For studies of sensitivities in broth media, cultures (4 mL) were electron microscopy of thin sections of whole cells. 'supplemented with the appropriate dissolved antibiotic (5-10 J-LL, 50% (v Iv) ethanol or methanol) and inoculated with 40 J-LL of a Materials and methods culture at mid to late logarithmic growth phase. Bacterial strains and cultural conditions Isolation of peptidoglycans All strains were from R. B. Hespell's culture collection (National Cultures at late logarithmic growth phase (3-4 L) were harvested Center for Agricultural Utilization Research) and were collected by centrifugation (10 000 x g, 20 min, 4°C). The cell pellets were from various sources that have been listed in other publications freed of extracellular polysaccharide material by three successive (Mannarelli 1988; Mannarelli et ai. 1990). Cells of B. fibrisolvens washes by centrifugation using 50 mM triethanolamine, 0.5 M strains and Lachnospira multiparus were routinely grown sodium chloride, 1.0 mM dithiothreitol (TSD) buffer (pH 7.5). The anaerobically at 37°C on a complex trypticase - yeast extract washed cell pellets were rapidly frozen using a dry ice - ethanol medium (RGM, Hespell et ai. 1988) or a chemically defined medium bath and stored anaerobically at - 20°C until needed. The frozen (DM, Cotta and Hespell 1986) with 0.4070 glucose as the energy cells were thawed in four to five pellet volumes of TSD buffer 914 CA . J. MICROBIaL. VOL. 39, 1993

TABLE 1. Antibiotic sensitivities of B. fibrisolvens strains

Diameter (mm) of zones of growth inhibition produced with plate overlays containing bacterial lawns

Antibiotic Amount(J 49 H17c 12 CE51 CF3 CFlb CF2d IL6-31 NOR37 PI7 01 AcTF2 2-33

Ampicillin 40 ILg 26 36 38 20 28 20 30 39 29 25 29 29 21 Penicillin IOU 24 35 40 26 29 26 25 33 26 25 44 26 30 Cephalothin 30 ILg 31 20 27 24 40 14 35 33 28 30 35 23 26 Bacitracin 20 U 22 20 27 32 32 27 30 21 14 36 30 21 15 Polymyxin B 300 U 10 9 13 15 13 11 15 12 9 12 11 8 14 Erythromycin 15 ILg 19 27 30 30 32 18 30 28 21 25 37 29 23 Gentamycin 10 ILg 19 19 18 20 19 15 22 26 9 17 18 18 11 Kanamycin 30 ILg 17 15 19 24 23 16 30 24 8 17 13 14 12 Streptomycin 10 ILg 9 19 12 12 12 10 28 18 0 12 11 11 8 Tetracycline 30 ILg 17 23 19 12 38 24 32 0 9 30 39 26 13 Chloramphenicol 30 ILg 7 17 22 30 32 21 30 14 9 22 26 13 8 Rifampicin 15 ILg 16 18 0 24 45 32 28 45 34 30 42 26 21 Novobiocin 30 ILg 0 8 20 28 28 16 30 18 24 32 9 0 20 Metronidazole 80 ILg 27 38 35 32 43 30 30 30 27 38 38 23 23

U Amount of antibiotic contained in disc (7 mm diameter). containing DNase (10 ILg/mL) and RNase (5 ILg/mL). Whole cells ture. The treated PG was recovered by high-speed centrifugation. were disrupted by two passages of the cell suspension through a The resulting pellet was washed with distilled water three successive French pressure cell (18 000 psi (1 psi = 6.895 kPa), room tem­ times and lyophilized to dryness. The treated PG was analyzed for perature). Residual whole cells were removed by low-speed cen­ amino acids and amino as described above. trifugation (3600 x g, 10 min, 4°C). The cell pellet was Preparation of cells for electron microscopy resuspended in an equal volume of TSD buffer and recentrifuged Cells were grown to early to mid logarithmic growth phase in at low speed. The two supernatant fluids were combined and cen­ DM medium containing 0.4% glucose as the energy source. The trifuged at high speed (81 800 x g, 30 min, 4°C) to obtain a crude cells were harvested by centrifugation (10000 g, 10 min, 4°C). cell wall pellet that was suspended in 0.5 volume of TSD buffer. x The cell pellet was gently resuspended in 20% culture volume of The wall preparation was heated for 5 min in a 75°C water bath 0.1 M cacodylate buffer (pH 6.5) containing 5% (v/v) and an equal volume of hot (75°C) SDS solution (100/0) was added glutaraldehyde and stored overnight at 4°C with slow mixing. The slowly with stirring. The mixture was incubated at 75°C for 60 min, cell suspension was centrifuged as before and the pellet was gently cooled to room temperature, and centrifuged at low speed (3600 x resuspended in the 0.1 M cacodylate buffer containing 0.05% (w/v) g) to remove insoluble materials, and the resulting supernatant fluid ruthenium red. After three successive washes by centrifugation in was centrifuged at high speed to obtain a pellet of crude peptido­ this buffer, the cells were enrobed in agar and prepared for ultrathin glycan (PG). The pellet was suspended in 15 mL of 1 mM sodium sections, using previously described procedures (Cheng et al. 1989). chloride and subjected to five successive washes by centrifugation Embedded preparations were sectioned using a Reichert model (83 500 x g, 30 min, 24°C) to obtain a washed PG preparation. OM U2 ultramicrotome and the sections were stained using stan­ The washed PG was suspended in 3 mL of 50 mM Tris buffer dard lead citrate procedures before stabilization with evaporated (pH 8.3) containing 10 mM calcium chloride, 10 ILg/mL DNase, carbon. All preparations for electron microscopy were examined and 5 ILg/mL RNase and incubated at 37°C for 30 min. After addi­ using a Hitachi 5400 electron microscope operating at a 6O-kV accel­ tion of 1 mL of trypsin solution (100 ILg/mL), the mixture was erating voltage. incubated for another 12 h. The enzyme-treated PG mixture was heated to 75°C, treated with the SDS solution, and washed by Other centrifugation as described above. The final, purified PG pellet The presence of poly-,B-hydroxybutyrate (PHB) in cells was was suspended in distilled water, frozen, and stored or lyophilized examined by digesting cell pellets with alkaline hypochlorite and to dryness. The final yields of PG from 4-L cultures were about purifying the extracted materials by the method of Herron et al. 15-20 mg. (1978). The amount of PHB was estimated by the change in absor­ bance at 235 nm after heating in sulfuric acid (Law and Slepecky Analyses of peptidoglycans 1961). Intracellular glycogen was obtained by heating cell pellets The amino acid and amino composition of PG was deter­ in 30% (w/v) potassium hydroxide for 3 h at 100°C. After cooling mined by suspending about 500 ILg of purified PG in 0.5 mL of to room temperature, polysaccharide material was precipitated by 6 M HCl and hydrolyzing under nitrogen gas for 14 h at 100°C. the addition of two volumes of ethanol. The carbohydrate content After cooling, the mixture was taken to dryness in a vacuum of the precipitated material was estimated by the phenol sulfuric dessicator containing solid sodium hydroxide. The amino acid and acid assay (Ashwell 1957). amino sugar analysis was done by high-performance liquid chro­ matography as described by Kato et al. (1976). The relative abun­ Results dance of free amino groups in the PG was estimated by Antibiotic sensitivities dinitrophenylation prior to acid hydrolysis. About 500 ILg of Butyrivibrio fibrisolvens strains from a variety of isolation purified PG was suspended" in 250 ILL of 1% (w/v) Na2B407 and sources were screened for antibiotic sensitivities as measured 25 ILL of 0.1 M flurodinitrobenzoate (FDNB, in ethanol) and incubated for 30 min at 60°C. After drying under nitrogen, the by clearing zones in bacterial lawns surrounding drug­ residue was suspended in 6 M HCl and treated as above. To remove impregnated disks and by inhibition of growth in broth polysaccharide material that copurified with the PG, about cultures. Representative data from the disk method (Table 1) 20-25 mg of purified PG material was suspended in 1 mL of 10% indicated that all strains were very sensitive to antibiotics (w/v) trichloroacetic acid and incubated for 6 h at room tempera- acting at the cell wall, such as ,B-Iactams, cephalothin, and TABLE 2. Compositions of peptidoglycans isolated from B. jibrisolvens strains and L. multiparus 032

Muramic Glutamic Oiaminopimelic Aspartic Strain Treatment acid Glucosamine Alanine acid acid Lysine Glycine Galactosamine acid Histidine Ammonia

49 - 0.84 2.48 2.38 1.00 1.09 0.0 0.28 0.63 0.28 0.12 2.61 + 1.07 3.35 1.98 1.00 0.55 0.0 0.30 0.91 0.21 0.17 2.98 T 0.73 1.61 1.95 1.00 0.82 0.0 0.27 0.12 0.28 0.04 2.33 CE51 0.51 2.15 1.87 1.00 0.63 0.01 0.21 0.42 0.23 0.13 2.87 + 0.77 1.95 1.71 1.00 0.54 0.0 0.24 0.58 0.17 0.10 4.03 T 0.62 1.78 1.97 1.00 0.89 0.0 0.17 0.18 0.27 0.10 3.09 CF4c 0.64 2.73 1. 71 1.00 1.04 0.0 0.20 0.43 0.19 0.14 4.36 + 0.84 2.80 1.69 1.00 0.50 0.0 0.17 0.46 0.12 0.18 3.65 :t m T 0.57 2.27 1.54 1.00 0.90 0.0 0.19 0.16 0.17 0.10 4.78 (/) i:l GS113 - 0.69 2.01 2.41 1.00 0.90 0.03 0.18 1.00 0.40 0.11 2.54 m r + 0.76 2.37 2.20 1.00 0.61 0.0 0.17 1.14 0.24 0.16 2.46 r m T 0.48 1.69 1.97 1.00 0.99 0.06 0.16 0.35 0.39 0.03 3.57 -l NOR37 - 0.66 1.52 1.46 1.00 0.89 0.01 0.82 0.56 0.52 0.10 4.15 > + 0.81 1.86 1.54 1.00 0.68 0.0 0.21 0.42 0.33 0.09 4.24 ~ T 0.64 1.61 1.42 1.00 0.95 0.02 0.68 0.28 0.46 0.06 4.51 VV-l 0.75 1.42 2.30 1.00 0.93 0.01 0.19 0.54 0.20 0.11 2.36 + 0.78 1.82 2.08 1.00 0.54 0.02 0.19 0.49 0.12 0.17 2.71 T 0.61 1.44 2.06 1.00 0.77 0.0 0.12 0.25 0.16 0.03 2.41 Dl - 0.91 2.76 2.66 1.00 0.84 0.0 0.24 0.87 0.02 0.10 4.49 + 0.89 2.55 1.62 1.00 0.37 0.0 0.17 0.89 0.02 0.17 3.32 T 0.54 1.20 1.89 1.00 0.83 0.0 0.12 0.55 0 0.04 3.31 032 0.66 2.55 1.68 1.00 0.92 0.01 0.04 0.0 0.02 0.08 1.76 + 0.90 2.87 1.76 1.00 0.42 0.0 0.05 0.0 0.02 0.10 3.75 T 0.59 2.62 1.51 1.00 0.69 0.01 0.03 0.0 0.02 0.03 3.14

NOTE: Values are expressed as molar ratios relative to glutamic acid. -, sample not treated prior to acid hydrolysis; +, sample treated with FDNB prior to acid hydrolysis; T, sample treated with 10070 TeA prior to acid hydrolysis.

\0 VI 916 CA . J. MICROBIaL. VOL. 39, 1993 bacitracin. In broth cultures, ampicillin levels of 5 to in the PGs of most Butyrivibrio strains. The analyses of the 15 jLg/mL inhibited growth and often caused cell lysis. In amino sugar and amino acid compositions of the PGs before contrast, the growth of most strains was only slightly and after dinitrophenylation showed a substantial reduction affected by polymyxin B, which generally disrupts the outer of only the diaminopimelic acid component. membrane of Gram-negative bacteria. All of the B. jibrisolvens PGs contained large amounts Most strains were sensitive to a variety of protein synthesis of galactosamine, which was absent in the PG of inhibitors. Large clearing zones were observed with L. multiparus. Treatment at room temperature of the erythromycin, and in broth cultures, the minimal inhibitory B. jibrisolvens PGs with trichloroacetic acid resulted in loss concentrations (MIC) ranged from 1 to 10 jLg/mL for almost of 30-50% of the PG dry mass. Analysis of the extracted all strains tested. Other aminoglycosides such as kanamycin materials indicated the absence of hexuronic acids, but sub­ or gentamicin produced smaller clearing zones and the MICs stantial amounts of sugars by phenol sulfuric acid assay were higher (10 to 15 jLg/mL), and a few strains such as (Ashwell 1957) were present and could account for most of NOR37 (Table 1) were resistant to these antibiotics. Smaller the dry mass losses. Analysis of the extracted PGs showed clearing zones were also seen with streptomycin and in 11 that considerable loss of the galactosamine component had of 20 strains tested, the MICs were 50 jLg/mL or greater. occurred, but the levels of the other components remained Most strains were quite sensitive to tetracycline, having relatively constant. MICs of 10 jLg/mL or less; almost all of the strains repre­ Electron microscopy sented by strain 49 and strain CF3 clusters (Fig. 1) had MICs The cell walls of all of the strains of Butyrivibrio were of 1 jLg/mL or less. examined by transmission electron microscopy. Represen­ There was considerable strain variation with respect to tative pictures (Figs. 2-13) of sectioned material showed a several nucleic acid synthesis inhibitors. While all of the CF relatively homogeneous structure of a Gram-positive cell strains and most of the strains in the strain D 1 cluster were wall. However, these Gram-positive cell walls were unusually very sensitive to the RNA polymerase inhibitor rifampicin, thin (12-18 nm) and this may account for their negative strains of the strain 49 cluster were more resistant (Table 1). reaction in the Gram stain procedure. The thinnest cell walls However, spontaneous rifampicin mutants could be (12 nm) were seen in strains 49 (Fig. 3) and UC12492 (Fig. 4) obtained with relative ease by plating strains representative in the 49 cluster, and the thickest cell walls (18 nm) were of each strain cluster onto media containing 20 to 40 jLg/mL seen in strains GSl13 (Fig. 10) and NOR37 (Fig. 12), in the of rifampicin. Most strains were sensitive to novobicin, as GS-l13 cluster, but all of these cell walls were very thin when indicated by clearing zones, but a few were resistant compared with those of most Gram-positive bacteria. While (Table 1). This resistance was often variable and when in some instances the cytoplasmic membrane was not always examined in broth cultures some putative resistant strains clearly defined, no indications of an outer membrane displayed an MIC of 10 to 20 jLg/mL on occasion and less characteristic of the Gram-negative cell wall could be than 1 jLg/mL at other times. Strains Dl and GS113 were detected in any strains. always resistant to 20 jLg/mL novobicin. A striking feature The cells of all of these strains of Butyrivibrio were seen of all B. jibrisolvens strains was their consistently high resis­ (Figs. 2-13) to be surrounded by an electron-dense residue tance to nalidixic acid, a DNA gyrase inhibitor; all strains of extracellular polysaccharides that had been produced by grew in the presence of at least 50 jLg/mL and many of the the condensation during dehydration of the extensive strains displayed MICs of 100 to greater than 500 jLg/mL. glycocalyces that surrounded these cells in their living, fully Chemical compositions oj PGs hydrated state. This was particularly evident in strains that The PGs of seven B. jibrisolvens strains representing six produced substantial amounts of extracellular polysac­ DNA relatedness groups were isolated. In addition, the PG charide (Figs. 3,4, 11-13). The cytoplasm of cells of most of L. multiparus, which may be related to B. jibrisolvens of these strains of Butyrivibrio was seen to contain inclusions strains, was isolated. The recovered PGs represented only that varied in numbers, from cell to cell and from strain to small amounts (generally 1.4-3.5070) of the total cell dry strain. Some of these electron-translucent inclusions were mass. Thin-layer qualitative chromatographic analyses of well defined (Figs. 4 and 13), while others were amorphous the hydrolyzed PGs indicated the presence of glucosamine, (Figs. 8 and 12) or even reduced to irregular spaces (Fig. 9) muramic acid, alanine, and meso-diaminopimelic acid. No in the cytoplasmic matrix. These morphological data are putrescine, cadaverine, F, or menaquinones were consistent with the presence of cytoplasmic inclusions of detected. Quantitative analyses of the PGs confirmed these reserve carbohydrate that vary in number, and in morphol­ preliminary results and showed the PGs of different ogical definition, with changes in cellular metabolism. B. jibrisolvens strains to be relatively similar in amino sugar and amino acid compositions (Table 2). The PGs displayed Discussion average molar ratios of 0.71 :2.15:2.11: 1.0:0.81 for muramic As shown by electron microscopy in this study of acid, glucosamine, alanine, glutamic acid, and diamino­ 12 strains representing five separate DNA relatedness pimelic acid, respectively. The PG of L. multiparus strain clusters, all Butyrivibrio strains lack the trilamellar outer D32 also had similar molar ratios. None of the PGs con­ membrane structure commonly associated with Gram­ tained lysine or amino acids that are not normally associated negative bacteria. Some previous studies indicated by thin­ with PGs. In most B. jibrisolvens strains the PGs also had section electron microscopy that some strains such as 49, a glycine molar ratio of about 0.2, except in strain NOR37, NOR37, and LM8/l B might possess an outer membrane in which the ratio was about 0.8. The PG of this strain also (Dibbayawan et al. 1985; Sharpe et al. 1975), but freeze­ contained substantially greater amounts of aspartate etch preparations did not show a typical image for a Gram­ (Table 2). Neither aspartate nor glycine was present in the negative envelope. Likewise, our results with thin sections PG of L. multiparus, but these amino acids were present did not show an outer membrane structure for strains 49 HESPELL ET AL. 917

FIG. 2-5. Electron micrographs of sections of ruthenium red stained cell preparations of Butyrivibrio strains. Fig. 2. Strain 49. Fig. 3. Strain 12. Fig. 4. Strain UC12492. Fig. 5. Strain CF3. Note the densely stained residue of extracellular polysaccharides that formed knoblike accretions on the cells in Figs. 3 and 4, as they condensed during dehydration. Scale bars = 0.1 Jlm.

(Fig. 2) and NOR37 (Fig. 12). Thus, it would appear that not be expected to be retained by this wall (Beveridge and all strains possess a Gram-positive cell wall structure, but Davies 1983). Consistent with this thin cell wall is the high this structure is atypical in several respects. The wall is thin, sensitivity of Butyrivibrio strains to a variety of antibiotics being only about 12-20 nm in thickness compared with the (Table 1) that affect PG synthesis. In addition, the high sen­ normal 30- to 50-nm Gram-positive wall. Presumably, this sitivity of Butyrivibrio strains to a wide variety of antibiotics thinness accounts for the Gram-negative staining reaction suggests that the cell envelope poses a minimal barrier to displayed by all strains, as the Gram stain complex would many molecules. 918 CA . J. MICROBiaL. VOL. 39, 1993

FIG. 6-9. Electron micrographs of sections of ruthenium red stained cell preparations of Butyrivibrio strains. Fig. 6. Strain CF4c. Fig. 7. Strain VV-l. Fig. 8. Strain PI-7. Fig. 9. Strain T9-40a. Scale bars = 0.1 JLm.

The PG component of the Butyrivibrio cell wall also involve these amino acids. The L. multiparus PG lacks these displays some unusual characteristics. The dinitrophenyla­ amino acids, but does contain diaminopimelic acid as the tion data (Table 2) show that diaminopimelic acid is a major cross-linking amino acid. A puzzling aspect of the PG for component involved in cross-linking. However, cross-linking both species is the presence of high amounts of glucosamine. is not extensive, probably being only about 40-60% for The levels are about twice those one would expect on the many strains as suggested by differences in composition basis of the muramic acid contents. Initially, it was thought owing to FDNB treatment of the PGs (Table 2). Small that since high amounts of galactosamine were also present amounts of glycine and aspartate are present in the (Table 2), the PGs were contaminated with a copurified Butyrivibrio PGs, suggesting some of the cross-links may heteropolysaccharide composed of this amino sugar and glu- HESPELL ET AL. 919

10

FIG. 10-13. Electron micrographs of sections of ruthenium red stained cell preparations of Butyrivibrio strains. Fig. 10. Strain OSl13. Fig. 11. Strain OSl12. Fig. 12. Strain NOR3? Fig. 13. Strain AcTF2. Scale bars = 0.1 ILm. cosamine. This does not appear to be the case, as treatment (Cheng et al. 1979) and its PG lacks galactosamine (Table 2). of the PGs with trichloroacetic acid preferentially removed The presence of a galactosamine-containing polysaccharide the galactosamine, but had little effect on the glucosamine that copurified with the Butyrivibrio PGs suggests that this content of the PGs. It is clear that further analysis of the polysaccharide may be a component of the Butyrivibrio cell PGs is needed to determine the chemical linkages in detail. wall structure. Thin sections of all strains display a dark­ The present data are consistent with the PGs being a staining layer, presumably the PG, but with some strains modified meso-diaminopimelic acid direct PG of Schleifer a lighter staining external layer is present. This external layer and Kandler's (1972) group A type. is more evident with some strains such as 12 (Fig. 3), GS113 Lachnospira multiparus has a thick, Gram-positive wall (Fig. 10), GSl12 (Fig. 11), or NOR37 (Fig. 12). It is possible 920 CA . J. MICROBIOL. VOL. 39, 1993 that the external layer could be this galactosamine-containing Acknowledgements polysaccharide, which must be different from the excreted We thank O. Kandler for the initial chromatographic anal­ polysaccharides as these lack galactosamine (Stack 1988). yses of the PG. We thank P. O'Bryan, S. Kokeguchi, and A galactosamine-containing polysaccharide has been K. Lam for their excellent technical assistance in preparation detected in the cell walls of several other microorganisms of PGs, in analysis of the PG components, and in electron (Konig and Kandler 1979; Stackebrandt et al. 1978) and is microscopy, respectively. the major component of the cell wall of Methanosarcina barkeri, which does not contain a PG layer (Kandler and Ashwell, G. 1957. Colorimetric analysis of sugars. Methods Hippe 1977). Enzymol. 3: 73-103. Thin sections of Butyrivibrio strains also show some other Beveridge, T.J., and Davies, J .A. 1983. Cellular responses of distinct features. Many strains appear to have a dark­ subtilis and Escherichia coli to the gram stain. staining outermost layer that tends to form blebs or knoblike J. Bacteriol. 156: 846-858. structures. These were predominant in strains 12 (Fig. 3), Bryant, M.P., and Small, N. 1956. The anaerobic monotrichous UC 12492 (Fig. 4), and AcTF2 (Fig. 13), but were present -producing curved rod-shaped bacteria of the rumen. in ruthenium red stained preparations of all Butyrivibrio J. Bacteriol. 72: 16-21. Cheng, K.-J., and Costerton, J.W. 1977. Ultrastructure of strains examined in this study. As suggested previously in Butyrivibrio Jibrisolvens: a Gram-positive bacterium? studies with strain C3 (Cheng and Costerson 1977), this layer J. Bacteriol. 129: 1506-1512. might be involved in the adhesion of cells to plant cell walls. Cheng, K.-J., Dinsdale, D., and Stewart, C.S. 1979. Maceration Most Butyrivibrio strains excrete an extracellular polysac­ of clover and grass leaves by Lachnospira multiparus. Appl. charide that has a complex sugar composition that varies Environ. Microbiol. 38: 723-729. with strain (Stack 1988). Most likely, this outermost bleblike Cheng, K.-J., Phillippe, R.C., McLean, R.J.C., and Costerton, layer represents cell-associated remnants of polysaccharide. J.W. 1989. The characterization and ultrastructure of two new The blebs reflect areas of polysaccharide that have con­ strains of Butyrivibrio. Can. J. Microbiol. 35: 274-282. densed during fixation procedures and (or) surface polysac­ Cotta, M.A., and Hespell, R.B. 1986. Proteolytic activity of the charide secretion points. A second major feature of most ruminal bacterium Butyrivibrio Jibrisolvens. Appl. Environ. Microbiol. 52: 51-58. strains is the presence of large amounts of cytoplasmic inclu­ Dibbayawan, T., Cox, G., Cho, K.Y., and Dwarte, D.M. 1985. sions. Staining of cells or chemical analysis of cell contents Cell wall and plasma membrane architecture of Butyrivibrio spp. indicated that no PHB or lipidlike materials were present. J. Ultrastruct. Res. 90: 286-293. Strains 12 and CF3 contained about 5-70/0 of their cell dry Herron, J.S., King, J.D., and White, D.C. 1978. Recovery of poly­ mass in the form of glycogenlike material that could be ,6-hydroxybutyrate from environmental sludge. Appl. Environ. isolated from base-hydrolyzed cells. However, direct carbo­ Microbiol. 35: 251-257. hydrate analysis of whole cells indicated the presence of Hespell, R.B., Wolf, R., and Bothast, R.J. 1988. Fermentation about 20% carbohydrate. These carbohydrate-glycogen of by Butyrivibrio Jibrisolvens and other ruminal bacterial levels would seem sufficient to account for the relatively species. Appl. Environ. Microbiol. 53: 2849-2853. large numbers of cytoplasmic inclusions observed. Further Hungate, R.E. 1966. In The rumen and its microbes. Academic Press, Inc., New York. pp. 8-90. work is needed to determine the exact chemical nature and Kandler, 0., and Hippe, H. 1977. Lack of peptidoglycan in the roles of these inclusions. cell walls of Methanosarcina barkeri. Arch. Microbiol. 113: All characterized Butyrivibrio strains possess a number 57-60. of features that indicate these strains should be placed in Kato, K., Iwata, S., Suginaka, H., Namba, K., Kotani, S., and a single large grouping or genus. These features include (i) a Tamura, T. 1976. Chemical structure of the peptidoglycan of homogenous, very thin Gram-positive cell wall structure Vibrio parahaemolyticus A55 with special reference to the extent (Figs. 2-13; Cheng et al. 1989; Cheng and Costerton 1977), of interpeptide cross-linking. Biken J. 19: 139-150. (ii) a high similarity in the chemical composition of the PG Konig, H., and Kandler, O. 1979. N-Acetyltalosaminuronic acid and the presence of a galactosamine-containing polymer a constituent of the pseudomurein of the genus (Table 2); (iii) a very high resistance to nalidixic acid and Methanobacterium. Arch. Microbiol. 123: 295-299. Law, J.H., and Slepecky, R.A. 1961. Assay of poly-,6­ high sensitivity to penicillin and bacitracin (Table 1), and hydroxybutyric acid. J. Bacteriol. 82: 33-36. (iv) a DNA base composition of 41-42 mol% G + C Mannarelli, B.M. 1988. Deoxyribonucleic acid relatedness among (Mannarelli 1988; Mannarelli et al. 1990). However, most strains of the species Butyrivibrio Jibrisolvens. Int. J. Syst. strains can be placed into groups (Fig. 1) that differ from Bacteriol. 38: 340-347. one another on the basis of the neutral sugar compositions Mannarelli, B.M., Stack, R.J., Lee, D., and Ericsson, L. 1990. of excreted polysaccharides (Stack 1988) and DNA-DNA Taxonomic relatedness of Butyrivibrio, Lachnospira, Roseburia, hybridizations (Mannarelli 1988; Mannarelli et al. 1990). and Eubacterium species as determined by DNA hybridization Taken together, all of these morphological, chemical, and and extracellular polysaccharide analysis. Int. J. Syst. Bacteriol. genetic data indicate that the various Butyrivibrio groups 40: 370-378. represent distinct species. While strain D 1, the type strain Moore, W.E.C., Johnson, J.L., and Holdeman, L.V. 1976. Emendation of Bacteroidaceae and Butyrivibrio and descriptions of B. jibrisolvens, and strain T9-40a, the type strain of of Desulfomonasa gen. nov. and ten new species in the genera B. crossatus, fall into separate groups, at least four other DesulJomonas, Butyrivibrio, Eubacterium, , and distinct groups remain. These groups need to be differen­ Ruminococcus. Int. J. Syst. Bacteriol. 26: 238-252. tiated on phenotypic traits (e.g., sugar fermentation pat­ Schleifer, K.H., and Kandler, O. 1972. Peptidoglycan types of terns) and probably should be designated as new Butyrivibrio bacterial cell walls and their taxonomic implications. Bacteriol. species. Rev. 36: 407-477. HESPELL ET AL. 921

Shane, B.S., Gouws, L., and Kistner, A. 1969. Cellulolytic bacteria Stack, R.J. 1988. Neutral sugar composItIOn of extracellular occurring in the rumen of conditioned to low-protein teff polysaccharides produced by strains of Butyrivibrio jibrisolvens. hay. J. Gen. Microbiol. 55: 445-457. Appl. Environ. Microbiol. 54: 878-889. Sharpe, M.E., Brock, J.H., and Phillips, B.A. 1975. Glycerol Stackebrandt, E., Fiedler, F., and Kandler, O. 1978. Peptido­ as an antigenic determinant in a gram-negative glycantyp und zusammensetzung der zellwandpolysaccharide von bacterium Butyrivibrio jibrisolvens. J. Gen. Microbiol. 88: Cellulomonas cartalyticum and einigen coryneformen 355-363. organismen. Arch. Microbiol. 117: 115-118.