Carsten Strömpl, Division of Microbiology, National Research Centre for Biotechnology GBF, Mascheroder Weg 1, 38124 Braunschweig, Germany (e-mail: cst@GBF.DE).
Two obligately anaerobic bacterial strains were isolated from the contents of a pilot scale, anaerobic digester treating slaughterhouse waste with a high protein and lipid content. The isolates, LIP1 and MW8, were characterized as spore-forming, Gram-positive rods, capable of fermenting glycerol. Isolate LIP1 was also observed to be lipolytic and was able to hydrolyse tallow and olive oil. Both isolates grew optimally at 37 °C and formed either acetate and formate (LIP1), or acetate and butyrate (MW8), as major glycerol fermentation products. Both isolates produced ethanol as the major reduced fermentation end-product. Neither MW8 nor LIP1 had growth and metabolism inhibited by the addition of stearic acid at concentrations normally considered bactericidal. Analysis of the 16S rRNA gene sequences, in conjunction with the phenotypic data, confirmed that the isolates are members of the genus Clostridium (sensu lato), clustering with species of clostridial clusters I (MW8) and XIVa (LIP1).
The wool, meat, dairy and food industries produce large volumes of waste water containing significant amounts of protein and lipid. It has been estimated that 15 200 tonnes of lipid waste are produced per year in New Zealand alone ( Cohen et al. 1994 ). This waste is considered non-recoverable and disposal is both an economic and environmental concern. Problems with current aerobic treatments are compounded by uncertainties regarding the future levels of spray irrigation and landfilling that will be possible, due to the availability of land necessary for these practices ( Cohen et al. 1994 ). Therefore, anaerobic processing of these wastes is seen as a viable alternative technique compared with the conventional aerobic waste treatment processes.
Anaerobic digestion processes require bacterial consortia consisting of three trophic (feeding) groups: (i) hydrolytic fermentative micro-organisms; (ii) syntrophic acetogenic bacteria; and (iii) methanogenic bacteria ( Thiele 1991). The bacterial consortia and therefore treatment capacities and stabilities of high-rate anaerobic waste water systems containing lipid are seriously hampered by biomass encapsulation and the inhibitory effects of long-chain fatty acids ( Rinzema et al. 1994 ). Certainly, data exist which demonstrate the inhibition of pure cultures of anaerobic rumen bacteria by long-chain fatty acids (LCFA) at concentrations between 0·005 and 0·1 g l−1 ( Maczulak et al. 1981 ). However, little information exists on the LCFA inhibition thresholds for pure cultures of hydrolytic fermentative bacteria from anaerobic digesters. It has been established that a concentration of 1 g l−1 LCFA in anaerobic digester sludge is sufficient to cause severe inhibition of the digestion process ( Hanaki et al. 1981 ). This inhibitory effect is compounded further by the presence of high levels of ammonia-nitrogen (NH3-N) in high protein content slaughterhouse waste waters ( Heinrichs et al. 1990 ). It has been established, as well, that the initial and severe concentration thresholds for NH3-N toxicity are 40–80 mg l−1 and 200 mg l−1, respectively ( Heinrichs et al. 1990 ). While data exist on the anaerobic treatment of waste waters containing diluted liquid neutral lipids ( Angelidaki et al. 1990 ; Velioglu et al. 1992 ), there is a lack of both operational and microbiological data on the anaerobic digestion of slaughterhouse waste waters containing high levels of solid neutral lipids.
The present paper reports the isolation and characterization of glycerol-fermenting bacteria from a pilot scale anaerobic digester, treating high lipid- and protein-content waste water from a slaughterhouse.
Materials and methods
Source of isolates
The isolates were obtained from a pilot scale, continuously stirred tank reactor (CSTR) fed a meat waste containing 43% total solids (TS). The waste consisted of mixed mince, offal and paunch contents from sheep with a composition of lipid (49%, w/v), protein (39%, w/v) and ash (12%, w/v). The digester had a daily loading rate of 3·36 kg chemical oxygen demand (COD)/m3 and a solubilization rate for total suspended solids (TSS) of 46 g d−1. The concentration of ammonia (NH3-N) was 3 g l−1. The digester had an operational temperature of 37 °C and maintained pH values in the range of 6·4–7·8 for a period of 3 months prior to sampling.
Media and isolation procedure
Unless otherwise stated, all media components were of analytical grade and obtained from Sigma. All gases were oxygen-free, analytical grade (NZIG, Dunedin, NZ). Tallow obtained from a slaughterhouse offal rendering factory (Fortex, NZ) was analysed and prepared for use in broth culture incubations, as described previously ( Jarvis et al. 1998 ). All media preparations and anaerobic culture transfer techniques were carried out using the Hungate technique ( Hungate 1966). The medium (MBHI) used for enrichment of isolates LIP1 and MW8 was composed of (g l−1 or ml l−1): dehydrated brain heart infusion (Difco), 37; tripalmitin, 2·5; Tween-40 solution (20 g l−1), 2·5; gelatin (Difco), 2·5. All culture tubes were maintained with an anaerobic atmosphere consisting of N2:CO2 (50/50, v/v). The final pH of all autoclaved media was between 7·2 and 7·4.
Growth experiments and phenotypic characterization
The isolates MW8 and LIP1 were sub-cultured in a defined medium (GYE), which was a modified version (no sodium acetate added) of the glycerol-based GAYE medium ( Jarvis et al. 1997a ), with glycerol (100 or 150 mmol−1) as the major source of carbon and energy. All cultivations using agar media (2%, w/v) were undertaken using GYE medium. Growth in liquid culture was monitored by spectrophotometric measurement of optical density at 600 nm. Pelleted cells from liquid cultures were assayed for total protein using a modified version of the Lowry assay ( Hartree 1972). Glycerol concentration in the culture supernatant fluid was determined using a colorimetric technique ( Jarvis et al. 1997a ). Unless otherwise stated, all experiments were performed in duplicate.
The cultures were tested for their ability to utilize various organic substrates as carbon and energy sources using the bioMérieux API 20 A test kit according to the instructions of the manufacturers (bioMérieux, Marcy l’Etoile, France). Qualitative lipase activities of LIP1 and MW8 were determined using a modification of the rhodamine B conjugate technique ( Jarvis & Thiele 1997b) with olive oil or tallow as the lipid substrate. Quantitative lipase activity of isolate LIP1, incubated at 37 °C in basal GAYE medium ( Jarvis et al. 1997a ) containing tallow (10 g l−1) but no glycerol, acetate or yeast extract, was determined in duplicate, using LCFA and Lowry protein analyses.
Bacterial cell morphologies were determined initially by light microscopy using an Olympus (Auckland, New Zealand) Vannox microscope. Gram-reactions, motility, and endospore production were determined using established methods ( Doetsch 1981) with mid-, late-log and stationary-phase cultures grown in glycerol- or gelatin-containing liquid media. Thin-section electron microscopy was carried out on cells obtained from liquid cultures in the late-log phase of growth, as described previously ( Jarvis et al. 1997b ).
Analysis of fermentation products
The gas chromatographic (GC) and high-performance liquid chromatographic (HPLC) analyses of volatile fatty acids (VFA) and alcohols in culture supernatant fluids, and carbon dioxide in culture tube headspaces, were performed as described previously ( Jarvis et al. 1997a ). Long-chain fatty acids in culture sub-samples were quantified after derivatization to nitrophenylhydrazides and separation using a reversed-phase HPLC technique ( Jarvis & Thiele 1997a).
Genomic DNA isolation and PCR amplification, and nucleotide sequence determination of 16S rRNA genes
Isolation of genomic DNA ( Wilson 1987), PCR amplification, and nucleotide sequence determination and analysis of the 16S rRNA genes were carried out as described previously ( Jarvis et al. 1997b ). Evolutionary distances ( Jukes & Cantor 1969) were calculated from pair-wise sequence similarities using only homologous, unambiguously determined nucleotide positions. Dendrograms were generated using the programmes in the PHYLIP (Phylogeny Inference Package, Version 3·5c) software package ( Felsenstein 1989).
Long-chain fatty acid toxicity experiments
Stearic acid (C18 : 0) was prepared as a sterile anaerobic stock solution (70 mmol l−1) in water and the pH neutralized using equimolar levels of NaOH. The stock solution was added aseptically to Hungate tubes containing GYE broth to give final concentrations of 1, 2, 4 and 8 mmol l−1. A set of GYE broths with no stearic acid was also included. Duplicate tubes containing LCFA at each concentration were then inoculated (10% v/v) with LIP1 or MW8 grown to mid-log phase in GYE broth. All Hungate tubes were incubated with orbital agitation (130 rev min−1) at 37 °C and sub-samples (1 ml) were taken at various time points. The cells were pelleted by centrifugation at 15 000 g for 15 min. Bacterial protein in the pellet fraction, and VFA, glycerol and alcohol concentrations in the supernatant fraction, were then assayed. Stearic acid was used as the test LCFA, as it is one of the major LCFA (30% w/w basis) present in tallow/fat derived from the carcasses of sheep ( Jarvis & Thiele 1997a) and therefore the digester feed.
Isolation of bacteria
The two isolates LIP1 and MW8 were obtained from a serial dilution in MBHI of a 10% (v/v) inoculum from digester contents. The entire dilution series (10−1–10–10) was transferred three times, and 100 μl from each tube of each transfer was plated onto GYE agar. An inoculum originating from the 10–3 dilution yielded large, well separated colonies on the GYE agar, which were then picked and inoculated into GYE broth. These broths were re-plated on GYE agar, and colonies were picked again and sub-cultured in GYE broth. This process was repeated twice before obtaining a pure culture of isolate MW8.
The third MBHI broth sub-culture of the 10−1 dilution was used as an inoculum for plating onto rhodamine B-olive oil agar. Colonies exhibiting qualitative lipolytic activity (orange fluorescence at 310 nm) were then picked and inoculated into GYE broths and subsequently replated onto GYE agar. This process was repeated twice to obtain a pure culture of the isolate LIP1. The qualitative lipolytic activity of this isolate was re-confirmed on two further independent occasions using the rhodamine B-olive oil agar.
Phenotypic characterization and metabolism
The isolate LIP1 was observed to be Gram-positive and rod-shaped (1·5–3·0 μm × 0·6–0·8 μm) with sub-terminal endospore formation ( Fig. 1) during the stationary phase of growth in glycerol liquid medium. The isolate LIP1 was able to utilize glycerol as a growth substrate in defined broth medium, with a growth rate of 0·024 h−1. The colony form was circular and exhibited an umbonate elevation and entire margin. After 7 d incubation, the colonies were 1–2 mm in diameter and white in colour. This isolate was observed to be lipolytic and able to grow in defined media containing olive oil or tallow as the sole carbon source, as well as in complex media (MBHI) containing tripalmitin. Isolate LIP1 exhibited growth rates on tallow-based and MBHI media of 0·003 h−1 and 0·123 h−1, respectively. Analysis of the glycerol fermentation products indicated that LIP1 produced CO2 (14 mmol l−1), with formate (17 mmol l−1) and acetate (14 mmol l−1) as the major VFA products. Ethanol (45 mmol l−1) was the major reduced end-product, although trace amounts of lactate (2 mmol l−1) and butanol (1 mmol l−1) were also detected.
Isolate MW8 was observed to be a Gram-positive, non-motile rod (3 μm × 0·4 μm) which formed sub-terminal/ central endospores ( Fig. 2). This isolate was non-lipolytic but was capable of utilizing glycerol as a sole carbon and energy source, with a rapid growth rate of 0·798 h−1. The colony form was circular, exhibiting a raised elevation and entire margin. After 7 d incubation, the colonies had a diameter of 2 mm and were white in colour. Analysis of glycerol fermentation products indicated that acetate (14 mmol l−1) and butyrate (12 mmol l−1) were the major VFA produced, with ethanol (76 mmol l−1) produced as the major reduced end-product late in the fermentation process. Apart from CO2 (14 mmol l−1), small amounts of lactate (4 mmol l−1) and butanol (3 mmol l−1) were also detected. Both isolates utilized a wide range of saccharides as growth substrates, and isolate LIP1 could also ferment gelatin, unlike isolate MW8 ( Table 1). Neither MW8 nor LIP1 was capable of growth in media with yeast extract (0·01%, w/v) as the sole carbon and energy source. Neither isolate produced catalase, although MW8 exhibited urease activity and did not exhibit indole formation. The converse was true for isolate LIP1, which formed indole and did not exhibit urease activity. Isolate LIP1 has been placed in the German Collection of Micro-organisms and Cell Cultures (DSMZ, Braunschweig, Germany) under the accession number DSM 10643.
Table 1. Comparison of substrate utilization by anaerobic digester isolates MW8 and LIP1 with phenotypically-related members of the genus Clostridium
*Based on duplicate assyas for each tested carbon source using the API 20A test kit (bioMérieux, France). Results analysed after 24 h of anaerobic incubation at 37 °C. Data for Clostridium butyricum adapted from Holdeman & Moore (1972) and Cato et al. (1986) . Data for Cl. clostridiiforme adapted from Kaneuchi et al. (1976) . All isolates can utilize glucose, maltose and mannose.
Determined in duplicate using growth on agar plates incubated anaerobically at 37 °C for a minimum of 7 d.
+ w, Weak reaction for lipolytic activity; +/−, most strains positive; −/+, most strains negative.
PCR amplification and sequencing of the 16S rDNA of the digester isolates enabled approximately 94% of the gene to be determined. Sequence comparisons ( Table 2) indicated that the organisms clustered within the Clostridium (sensu lato) taxonomic group ( Collins et al. 1994 ). The sequence similarity values indicated that the 16S rDNA of MW8 was identical (100% over the 1432 nucleotide positions examined) to that of a pectin-degrading strain (DSM 2478) of Cl. butyricum ( Schink & Zeikus 1982), whereas the 16S rDNA of isolate LIP1 (DSM 10643) was most similar (98·3%) to that of the type strain (DSM 933T) of Cl. clostridiiforme ( Kaneuchi et al. 1976 ). The sequence for LIP1 in the region of helix 6 ( Woese et al. 1983 ) was unable to be resolved due to the secondary structure of the molecule, or micro-heterogeneity in the 16S rDNA of multiple operons. However, as this region is generally susceptible to hypervariability, it is not essential for phylogenetic analyses. A dendrogram was constructed from the 16S rRNA gene sequence data ( Fig. 3) which demonstrates that isolates MW8 and LIP1 group within clusters I and XIVa, respectively, of the genus Clostridium ( Collins et al. 1994 ) and are not closely related. The 16S rDNA sequences for isolates LIP1 and MW8 have been deposited with the EMBL nucleic acid sequence database under the accession numbers AJ002591 and AJ002592, respectively.
Table 2. 16. rRNA gene sequence similarities between the anaerobic digester isolates LIP1 and MW8 and reference species
DSM, German Collection of Micro-organisms and Cell Cultures DSMZ, Braunschweig, Germany; ATCC, American Type Culture Collection, Rockville, USA.
TDesignated type strain of species.
Similarities derived from the comparisons of unambiguous homologous nucleotide positions from optimally aligned sequences.
Cl. botulinum (type G)
Long-chain fatty acid toxicity studies
The effect of stearic acid (C18 : 0) on the growth and metabolism of LIP1 and MW8 was followed in a time course study. The cumulative results, after extended incubation (348 h) which allowed the cultures to reach the stationary phase of growth, are presented in Fig. 4. Biomass and VFA production for isolate MW8 were shown to be adversely affected by LCFA concentrations of 2 mmol l−1 or greater, although VFA production levels in the presence of 8 mmol l−1 stearic acid was observed to be similar to the levels produced by cultures incubated with 0 or 1 mmol l−1 of this LCFA ( Fig. 4b). The level of glycerol utilized by isolate MW8 was high (88–92%) and was not affected by the presence of stearic acid. Similar general biomass production and glycerol utilization trends were observed for the isolate LIP1 ( Fig. 4a). However, glycerol utilization was, on average, 13% lower in the presence of stearic acid compared with incubations without it, and VFA production was comparatively lower if more than 2 mmol l−1 stearic acid were present ( Fig. 4a). In cultures with low VFA production, solvent production (ethanol and butanol) was 10–30 mmol l−1 higher at the end of the stationary phase of growth compared with non-LCFA containing cultures ( Jarvis 1995). Interestingly, it has also been established that the metabolism of LIP1 is not adversely affected by the presence of glycerol monostearate at concentrations as high as 8 mmol l−1, and the isolate is able to cleave this monoglyceride, thereby releasing the LCFA moiety and utilizing the glycerol as a growth substrate ( Jarvis 1995).
Overall, these data indicated that both LIP1 and MW8 were able to continue to grow and metabolize glycerol in the presence of stearic acid at concentrations up to, and including, 8 mmol l−1 (2·3 g l−1). Such concentrations are considered to be above the established LCFA toxicity thresholds for anaerobic bacteria ( Maczulak et al. 1981 ).
Few data exist describing the bacteria present in anaerobic digesters treating high protein- and lipid-containing wastes. While it is well established that anaerobic digestion is essentially a microbiological process, conflicting views exist with respect to the characterization of the bacteria associated with the degradation of the organic material (carbohydrates, proteins and lipids) in anaerobic digesters ( Sahm 1984; Van Andel & Breure 1984; Thiele et al. 1988 ; Lowe et al. 1993 ).
This paper is the first published report of the isolation and characterization of a mesophilic lipolytic anaerobic bacterium (isolate LIP1) from an anaerobic digester ecosystem. It also presents the first data on the isolation and characterization of glycerol-fermenting bacteria from an anaerobic digester treating high protein- and lipid-content slaughterhouse waste.
Based on the phenotypic and phylogenetic data, isolate MW8 is almost identical to Clostridium butyricum, a member of cluster I in the clostridial taxonomic group ( Collins et al. 1994 ). Interestingly, examination of glycerol dissimilation by four reference strains of Cl. butyricum showed that 1,3 propanediol, acetate and butyrate were the major fermentation products and only trace amounts (1–2 mmol l−1) of ethanol were formed ( Biebl et al. 1992 ). These data contrast with the results of this study, where it was established that ethanol, rather than propanediol is the major reduced end-product formed late in the fermentation process. However, it is also known that some strains of Cl. butyricum do produce ethanol as a fermentation end-product ( Cato et al. 1986 ), and it has been established that the production of ethanol is a common feature of many saccharolytic, mesophilic, butyrate-producing clostridia (such as Cl. butyricum) in late stages of the fermentation process ( Jones & Wood 1989). The isolate MW8 was observed to be saccharolytic, but was unable to hydrolyse lipids. However, isolate MW8 was able to use glycerol as the sole carbon substrate for growth. While being related phylogenetically to strains of Cl. butyricum ( Fig. 3), including a pectinolytic strain (DSM 2478) isolated from lake sediment ( Schink & Zeikus 1982), isolate MW8 exhibited a number of phenotypic differences, including the ability to ferment rhamnose and sorbitol and the inability to utilize mannitol or hydrolyse esculin. Therefore, isolate MW8 probably represents a new strain of Cl. butyricum, different to those previously described.
On the basis of 16S rRNA gene sequence comparisons, isolate LIP1 was shown to be phylogenetically related (98·3%) to the type strain of Cl. clostridiiforme (DSM 933T). However, Cl. clostridiiforme ferments lactose and is not capable of hydrolysing gelatin or lipids ( Kaneuchi et al. 1976 ), whereas isolate LIP1 cannot ferment lactose but exhibits gelatinolytic and lipolytic activities. Based on these data, isolate LIP1 probably represents a lipolytic strain of a new species belonging to clostridial cluster XIVa ( Collins et al. 1994 ).
The taxonomic organization of the genus Clostridium is undergoing significant revisions ( Collins et al. 1994 ) and the majority of species currently classified within this genus will ultimately be separated and reclassified within newly described, distinct genera. The isolate LIP1 described in this paper was observed to cluster with species of cluster XIVa. This phenetically heterogeneous group comprises described species currently classified as Clostridium, as well as several other recognized genera ( Collins et al. 1994 ). It is clear that all Clostridium species within cluster XIVa will be reclassified in the future as species in newly recognized genera. Thus, it is prudent, at this time, to recognize the isolate LIP1 of the current study simply as ‘Clostridium sp. (strain LIP1)’, rather than attempting to describe a new species epithet.
As there was comparatively little carbohydrate present in the anaerobic digester feed used in the current study, it is conceivable that isolate LIP1 may be a member of the lipid utilizing/glycerol-fermenting niche, either as an opportunist or as an indigenous member. Clostridium butyricum (strain MW8) exhibited a rapid growth rate on glycerol and may therefore be able to compete successfully for this growth substrate in the anaerobic digester ecosystem. Interestingly, a strain of Cl. butyricum has been isolated from the anaerobic digester ecosystem in a previous study ( Xiuzhu et al. 1991 ), although that strain was isolated and studied based on its ability to degrade galactomannan, rather than the ability to ferment glycerol.
The results from the LCFA toxicity studies demonstrated that isolates MW8 and LIP1 were capable of active growth and metabolism at LCFA concentrations 20–460 times higher than the levels reported to be toxic for other anaerobic bacteria ( Maczulak et al. 1981 ). The lipolytic isolate, LIP1, was observed to undergo a physiological change, resulting in a different glycerol fermentation end-product profile, with decreased VFA production being accompanied by a concomitant equimolar increase in the formation of reduced solventogenic end-products. This resulted in an alteration in the ratio of oxidized and reduced fermentation products in favour of the latter. One explanation for this may be related to the inhibition of hydrogenase activity and the regeneration of reduced co-factors, which is known to lead to the carbon flow of species within the genus Clostridium being directed to the formation of more reduced end-products, such as ethanol and butanol ( Jones & Wood 1989). Whether these effects are responsible for the results observed for the clostridial isolate LIP1 were not tested. The fact remains that both isolates were metabolically and physiologically active at LCFA concentrations that were well in excess of established toxic thresholds and which could be considered to represent extreme growth conditions. It is conceivable that the results obtained in vitro may reflect the initial extremophilic conditions present in the anaerobic digester which served as a source of the two isolates. Indeed, the digester had an NH3-N concentration (3 g l−1) well in excess of established severe toxicity threshold values ( Heinrichs et al. 1990 ) and had a feed containing almost 50% lipid within the solids phase. Therefore, it is not surprising that we have been able to isolate LCFA toxicity-resistant anaerobic bacteria, which, based on the results of the LCFA toxicity studies, could be considered to be included in the diverse group of bacteria termed ‘extremophiles’ ( Lowe et al. 1993 ).
The authors thank Richard Easingwood for electron microscopy, and Dr Tico Cohen (Waste Solutions Ltd, Dunedin, NZ) for helpful comments on the work, and both the samples and operational data from the anaerobic digesters. The work was financially supported by grants from Otago Research Committee (Grant KLIB07), a grant from Waste Solutions Ltd and the German-New Zealand Science and Technology Co-operation Project (STC 93·20). Graeme Jarvis was the recipient of a German Academic Exchange Service (DAAD) scholarship (Referat 424). The authors are indebted to Dr Hans-Jürgen Hamann (GSFmbH) and Professor Ken N. Timmis (GBFmbH) for their continued support of the scientific collaboration.
Present address: Section of Microbiology, 156 Wing Hall, Cornell University, Ithaca, NY 14853, USA.
Present address: Waste Solutions Ltd, PO Box 997, Dunedin, New Zealand.