Figures
Abstract
Clostridium phytofermentans was isolated from forest soil and is distinguished by its capacity to directly ferment plant cell wall polysaccharides into ethanol as the primary product, suggesting that it possesses unusual catabolic pathways. The objective of the present study was to understand the molecular mechanisms of biomass conversion to ethanol in a single organism, Clostridium phytofermentans, by analyzing its complete genome and transcriptome during growth on plant carbohydrates. The saccharolytic versatility of C. phytofermentans is reflected in a diversity of genes encoding ATP-binding cassette sugar transporters and glycoside hydrolases, many of which may have been acquired through horizontal gene transfer. These genes are frequently organized as operons that may be controlled individually by the many transcriptional regulators identified in the genome. Preferential ethanol production may be due to high levels of expression of multiple ethanol dehydrogenases and additional pathways maximizing ethanol yield. The genome also encodes three different proteinaceous bacterial microcompartments with the capacity to compartmentalize pathways that divert fermentation intermediates to various products. These characteristics make C. phytofermentans an attractive resource for improving the efficiency and speed of biomass conversion to biofuels.
Citation: Petit E, Coppi MV, Hayes JC, Tolonen AC, Warnick T, Latouf WG, et al. (2015) Genome and Transcriptome of Clostridium phytofermentans, Catalyst for the Direct Conversion of Plant Feedstocks to Fuels. PLoS ONE 10(6): e0118285. https://doi.org/10.1371/journal.pone.0118285
Academic Editor: Willem J.H. van Berkel, Wageningen University, NETHERLANDS
Received: April 27, 2013; Accepted: January 12, 2015; Published: June 2, 2015
Copyright: © 2015 Petit et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Funding: This research was supported by U.S. Department of Energy grant DE-FG02-02ER15330 and Cooperative State Research, Extension, Education Service, U.S. Department of Agriculture, Massachusetts Agricultural Experiment Station Project No. MAS00923 to S.L. and Project No. MAS009582 to J. B., a Sponsored Research Agreement between Qteros and J.B. and S.L., UMass College of Natural Sciences and Mathematics Excellence Initiatives funding to E.P., a Howard Hughes Medical Institute Award to the University of Massachusetts Amherst, a National Science Foundation equipment grant NSF-MRI (0722802) and a National Science Foundation grant NSF BBS 8714235 to the University of Massachusetts Amherst Central Microscopy Facility. The work conducted by the U.S. Department of Energy Joint Genome Institute was supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Plant biomass is one of the most abundant renewable energy sources on Earth and a largely underutilized feedstock for biofuels [1]. Production of biofuels from the lignocellulose fraction of plant biomass differs from production from grains in two fundamental aspects: (1) different types of saccharolytic enzymes are required to break down lignocellulose into soluble carbohydrates; and (2) fermentation of pentose sugars, in addition to hexoses, is required to harvest the majority of energy stored in lignocellulose [2]. At present, the cost of producing saccharolytic enzymes and the complexity of the hydrolysis and fermentation processes limit the use of plant biomass as a competitive alternative to gasoline and pose key challenges in the development of a global biomass industry for manufacturing a wide range of products from agricultural and forestry wastes [3].
One potential solution is the use of microbes that produce lignocellulose-decomposing enzymes and simultaneously ferment the resulting hexose and pentose carbohydrates to products such as ethanol. Merging these processes in a single microbe could substantially reduce the costs of lignocellulosic biofuel production [4]. Such microbes, primarily members of the Clostridiales, are found in natural anoxic environments where vast quantities of cellulose and other plant cell wall components are decomposed.
Species of Clostridium have a rich tradition in the development of biofuels. Clostridium acetobutylicum is a long-standing commercially valuable bacterium that has been used to produce acetone, butanol and ethanol from starch [5]. Processes based on acetone-butanol-ethanol fermentation were industry standards until the late 1940's, when low oil prices favored processes based on hydrocarbon cracking and petroleum distillation techniques. C. acetobutylicum and its relative Clostridium beijerinckii have recently regained market interest for use in the production of butanol as a gasoline and diesel fuel replacement.
Microbial fermentation of cellulose has been studied extensively in Clostridium cellulolyticum and Clostridium thermocellum [6–9]. Carbon metabolism during growth on cellulose and cellobiose in C. cellulolyticum has been investigated using carbon isotope labeling and metabolic flux analysis [8]. To degrade cellulose, C. cellulolyticum and C. thermocellum produce extracellular enzymatic complexes (cellulosomes) that permit bacterial adhesion to insoluble substrates and promote the hydrolysis of cellulose [6,9,10]. C. acetobutylicum laboratory strains do not grow on cellulose although they contain genes for cellulosome synthesis [7] and secrete a small cellulosome [11]. Products of cellulose degradation, such as cellobiose, are transported across the cell membrane and enter into the Embden-Meyerhof-Parnas pathway. C. acetobutylicum laboratory strains do not grow on cellulose although they contain genes for cellulosome synthesis [7] and secrete a small cellulosome [11].
We isolated a new species, Clostridium phytofermentans (strain ISDg ATCC 700394) from forest soil near the Quabbin Reservoir in Massachusetts, U.S.A. that directly ferments all major components of plant biomass, including cellulose, hemicellulose, pectin and starch to yield ethanol as the primary product of fermentation [12]. The combination of carbohydrate substrate versatility and high ethanol yield in a single organism distinguishes C. phytofermentans from other described species and suggests that it possesses unusual catabolic pathways. Extensive metabolism of the complex sugars within lignocellulose (without high ethanol yield) appears to be a trait found in several hyperthermophiles [13] but C. phytofermentans stands out as one of few, if not the only mesophile with this capacity. Thus, C. phytofermentans offers opportunities to understand the molecular mechanisms of plant biomass conversion to biofuels in a single organism. These attributes have led to the adoption of C. phytofermentans as a study system by groups in the US, Japan and Europe [14–25].
Here we investigate the unique properties of C. phytofermentans through analyses of its complete genome sequence and transcriptional profiling during growth on key components of plant biomass.
Materials and Methods
Growth and DNA extraction
Clostridium phytofermentans ISDgT was cultured in anaerobic medium GS-2CB containing cellobiose (3 g/l) prepared as described previously [12]. Cultures were incubated in an atmosphere of O2-free N2 at 30°C. Genomic DNA was purified from 100 ml of mid exponential phase GS-2CB cultures using a standard DNA isolation procedure recommended by the Joint Genome institute, the Bacterial CTAB protocol [26].
Construction, isolation and sequencing of insert libraries
Genomic DNA was sequenced using an established whole genome shotgun strategy [27]. Random 2–3 kb-DNA fragments were isolated after mechanical shearing. These gel-extracted fragments were concentrated, end-repaired and cloned into pUC18. Double-ended plasmid sequencing reactions were carried out using PE BigDye Terminator chemistry (Perkin Elmer) and sequencing ladders were resolved on PE 3700 Automated DNA Sequencers.
Sequence assembly and gap closure
Sequence traces were processed with Phred [28] for base calling and assessment of data quality before assembly with Phrap [29] and visualization with Consed [30].
Sequence analysis and annotation
Gene modeling was performed with both the Critica [31] and Glimmer [32] modeling packages. The results were combined and a basic local alignment search tool for proteins (BLASTP) versus GenBank's nonredundant database (NR) was conducted. The alignment of the N terminus of each gene model versus the best NR match was used to pick a preferred gene model. If no BLAST match was returned, the Critica model was retained. Gene models that overlapped by greater than 10% of their length were flagged, giving preference to genes with a BLAST match. In addition to BLASTP versus NR, the revised gene/protein set was searched against the KEGG GENES [33–35], InterPro [36] (incorporating Pfam [37], TIGRFams [38], SMART [39], PROSITE [40], PRINTS [41] and ProDom [42]) and Clusters of Orthologous Groups of proteins (COGs) [43] databases. From these results, functional categorizations were developed using the KEGG and COGs hierarchies. Initial criteria for automated functional assignment required a minimum 50% residue identity over 80% of the length of the match for BLASTP alignments, plus concurring evidence from the above profile methods (e.g. pfam). Putative assignments were made for identities down to 30%, over 80% of the length.
To determine whether C. phytofermentans produced cellulosomes, a position-specific amino acids matrix, based on the sequence of the cellulosome domains from the Clostridium species, C. cellulolyticum, C. thermocellum and C. acetobutylicum, was constructed and searched against a local database of predicted C. phytofermentans proteins using PSI-BLAST. In addition, we searched C. phytofermentans proteins using Pfam models of cohesins and dockerins. Analyses of the theoretical subcellular localization and signal peptide cleavage sites were carried out using PSORT (http://psort.hgc.jp/form.html). Transporters were annotated by TransAAP [44]. The complete sequence of C. phytofermentans was made available in August 2007 (accession number NC_010001).
Phylogenetic analysis of 16S rRNA sequence
To elucidate the phylogenetic relationship between C. phytofermentans and other members of the class Clostridia, including non-sequenced genomes, 16S rRNA gene sequences of the isolate and closely related species were used for neighbor-joining analysis. Sequences were aligned using ClustalX Version 2 [45]. The phylogenetic tree was constructed by neighbor-joining in Phylip [46]. Bacillus subtilis was used as an outgroup. Bootstrap values were calculated using a heuristic search and 1,000 bootstrap pseudoreplications.
Detection of carbohydrate-active enzymes in bacterial proteomes
The search for carbohydrate-active modules (glycoside hydrolases, glycosyltransferases, polysaccharide lyases and carbohydrate esterases) and their associated carbohydrate-binding modules (CBMs) in C. phytofermentans was performed exactly as for the daily updates of the Carbohydrate-Active enZYme (CAZy) database (http://www.cazy.org/). Briefly, the sequences of the proteins in CAZy were cut into their constitutive modules (catalytic modules, CBMs and other noncatalytic modules or domains of unknown function). The resulting fragments were assembled and formatted as a sequence library for BLAST [47] searches. Accordingly, each protein model from C. phytofermentans (and other bacterial proteomes) was searched via BLAST against the library of approximately 100,000 individual modules using a database size parameter identical to that of the NCBI nonredundant database. All models that gave an expectation value lower than 0.1 were automatically kept and manually analyzed. Manual analysis involved examination of the alignment of the model with the various members of each family (whether of catalytic or non-catalytic modules), with a search of the conserved signatures and motifs characteristic of each family. The presence of the catalytic machinery was verified for borderline cases whenever known in the family. The models that showed the usual features that would lead to their inclusion in the CAZy database were kept for annotation and classified in the appropriate class and family.
Investigation of evolutionary origins of glycoside hydrolases
The taxonomic distribution of the BLASTP best hits (e-value< = 0.01) of the glycoside hydrolases in GenBank's nonredundant database (as of January 2011) excluding C. phytofermentans sequences was compared to that of BLASTP best hits of all the ORFs within the C. phytofermentans genome. A Pearson's Chi-squared test was used to determine the significance of the differences between the two taxonomic distributions.
Culturing conditions for the microarray expression data
C. phytofermentans was cultured in a modified form of a previously described anaerobic medium [48] containing the following (g/l): yeast extract, 6.0; urea, 2.1; KH2PO4, 4.0; Na2HPO4, 6.5; trisodium citrate dihydrate, 3.0; L-cysteine hydrochloride monohydrate, 2.0; resazurin, 1; with pH adjusted to 7.0 using KOH. This medium was supplemented with 0.3% (wt/vol) of the following substrates: glucose, cellobiose, xylose, L-arabinose, birchwood xylan, and apple pectin (Sigma-Aldrich) as well as cellulose and “plant biomass” (Brachypodium distachyon). Substrates were added as a filter-sterilized solution to the sterile medium if soluble or autoclaved with the medium if insoluble. Duplicate liquid cultures were incubated at 30°C under anaerobic conditions (in an atmosphere of N2) as described by Hungate [49]. Growth on soluble substrates was determined spectrophotometrically by monitoring changes in optical density at 660 nm. Growth on solid substrates was estimated by visually monitoring and marking the reduction in biomass levels in the test tubes.
RNA isolation
RNA was isolated from two replicates of each type of culture at mid exponential phase. Briefly, cells were flash-frozen by immersion in liquid nitrogen, harvested by centrifugation for 5 min at 8,000 rpm at 4°C and re-suspended in 100 μl in TE buffer pH 8 (EMD Chemicals) containing 2 mg/ml lysozyme (Sigma-Aldrich) and incubated at 37°C for 40 min. Total RNA was isolated using the RNeasy RNA purification kit (QIAGEN) according to the manufacturer’s instructions. Contaminating DNA in total RNA preparations was removed with RNase-free DNase I (QIAGEN).
Determination of fermentation products
Non-gaseous fermentation products were determined by high-performance liquid chromatography (HPLC) and gas chromatography (GC). Acetate, ethanol, formate and lactate concentrations in culture supernatants were measured using a BioRad Aminex HPX 87H 300 x 7.8 mm column at 55°C with 0.005 M H2SO4 as the mobile phase and a flow rate of 0.60 ml/min, in a Hitachi model L-7100 HPLC unit equipped with a Sonntek Refractive Index Detector. The concentration of ethanol was also measured by GC, using a Shimadzu GC 2014 with a Flame Ionization Detector and a Restek stabilwax-DA 30 m x 0.25 mm ID column (film thickness 0.25 μm). The carrier gas was helium at a flow rate of 1.5 ml/min. Injector and detector temperatures were both 200°C and the column temperature began at 70°C for 2 min, ramped to 175°C at 20°C/min, and was held at 175°C for 2 min.
Microarray design
A C. phytofermentans Affymetrix microarray was custom-designed for the measurement of the expression level of all open reading frames, estimation of the 5’ and 3’ untranslated regions of mRNA, operon determination, and discrimination between alternative gene models (differing primarily in the selection of the start codon). Putative protein coding sequences were identified using both GeneMark and Glimmer, and the union of these two predictions was used to design the array. Each coding sequence (CDS) was represented by eleven 24-mer probes. Standard Affymetrix array design protocols were followed to ensure each probe was unique to minimize cross hybridization. If two CDS differed only in their N-terminal region, the smaller of the two proteins was used for transcript analysis, but the extended region was also represented by probes to define the actual N-terminus. Remaining probes were used to map expression in intergenic regions. These probes represented both DNA strands and were tiled with a 1-nucleotide gap. The array design was implemented on a 49–5241 format Affymetrix GeneChip with 11-μm features. The microarray was designed prior to the final annotation of the complete genome using gene prediction methods that are slightly different from those used in the final annotation done by the Joint Genome Institute. The microarray files uploaded to NCBI's GEO reflect the later complete genome annotation done by the Joint Genome Institute.
Microarray processing
cDNA synthesis, array hybridization and imaging were performed at the Genomic Core Facility at the University of Massachusetts Medical Center. Ten μg total RNA from each sample was used as template to synthesize labeled cDNAs using Affymetrix GeneChip DNA Labeling Reagent Kits. The labeled cDNA samples were hybridized on the arrays according to Affymetrix guidelines. The hybridized arrays were scanned with a GeneChip Scanner 3000. The resulting raw spot image data files were processed into pivot, quality report, and normalized probe intensity files using Microarray Suite version 5.0 (MAS 5.0) [50]. In addition, expression values were calculated using the Custom Array Analysis Software (CAAS) package (http://www.sourceforge.net/projects/caas-microarray/) that implements the Robust Multichip Average method [51]. The individual microarray files and the normalized gene summary values for the complete data set will be deposited in the Gene Expression Omnibus (GEO) database at NCBI [52].
The quality of the microarray datasets were analyzed using probe-level modeling procedures provided by the affyPLM package in BioConductor [52]. No image artifacts due to array manufacturing or processing were observed. Microarray backgrounds were within the typical 20–100 average background values for Affymetrix GeneChip. In summary, all quality control checks indicated that the RNA purification, cDNA synthesis, labeling and hybridization procedures adapted for use in C. phytofermentans resulted in high quality data. All microarray data reported in the text and figures represent the average of expression values derived from two independent RNA preparations from duplicate cultures.
Results and Discussion
C. phytofermentans is distinct from other well-studied solventogenic and cellulolytic species found within clostridial Clusters I (Clostridiaceae), III (Ruminococcaceae), and X (Thermoanaerobacteraceae) (Fig 1). A member of Cluster XIV (Lachnospiraceae), C. phytofermentans is closely related to human commensals that have been sequenced as part of the International Human Microbiome Consortium [53], and to bacteria isolated from rice paddy soils, earthworm intestines and other anaerobic, carbon rich environments (Fig 1). As a genetically tractable [15] member of this under-explored group, and the first with a publicly available genome sequence, C. phytofermentans is an important point of reference for comparative genomic analyses.
Taxa with sequenced genomes are marked with an asterisk. Cluster numbers correspond to the cluster system of Collins et al. [68]. Bootstrap values were determined for 1,000 replicates.
Features of the C. phytofermentans genome
C. phytofermentans has a single circular 4.8 Mbp chromosome, no plasmids and a G+C content of 35%. The genome encodes 3,926 CDS, 27% of which lack a predicted function (Table 1). Genes encoding eight rRNA clusters were found in proximity to the origin of replication and 61 tRNAs were detected (Table 1). Two putative prophage regions were found (S1 File). Genes for processes typical of clostridia, such as sporulation (S2 File), motility and chemotaxis, are present. Identification of sporulation-related genes is typically based on sequence homology to those of Bacillus subtilis–the model organism for studying the sporulation cycle [54–56]. The genome of C. phytofermentans contains a homolog of the master regulator of sporulation of B. subtilis, SpoOA (Cphy_2497, 55% amino acid identity to SpoOA of B. subtilis, Table A in S2 File). Although the majority of the genes in the sporulation cascade of B. subtilis downstream of the master regulator SpoOA are present in C. phytofermentans, the ones upstream (the sensory histidine kinase and phosphorelay system) are not. C. phytofermentans is motile, moving by means of one or a few sub-terminal flagella [12]. Genes predicted to be involved in flagellar biosynthesis are found in two clusters (Cphy_0303–0316 and Cphy_2687–2720). Chemotaxis genes are found within one of these clusters (Cphy_2687–2691). C. phytofermentans also has three distinct genetic loci coding for proteinaceous bacterial microcompartments [57,58] (S3 File), two of which are predicted to be involved in choline, ethanolamine and 1,2-propanediol metabolism and one whose function cannot be inferred from sequence analysis.
Genes encoding carbohydrate-active enzymes
C. phytofermentans is capable of breaking down the recalcitrant, insoluble components of plant cell walls including cellulose, hemicellulose, pectin and starch [12] as well as switchgrass, corn stover and pulp wastes that have been minimally processed without thermo-chemical pretreatment (Fig 2). Numerous carbohydrate-active enzymes (CAZy) predicted to be involved in the degradation of various plant cell wall components are encoded throughout the C. phytofermentans genome, including glycoside hydrolases (GH), polysaccharide lyases (PL) and carbohydrate esterases (CE). The diversity of GH families in the C. phytofermentans genome is unparalleled among sequenced clostridial genomes (Fig 3). A total of 116 GHs distributed among 44 families are encoded in the genome of C. phytofermentans including but not limited to endo- and exo-cellulases, hemicellulases, chitinases, pectinases, amylases, and lichenases (Fig 3 and S4 File). Only the GH content of a distant relative in Cluster I, Clostridium cellulovorans, is comparable, with 113 GH domains distributed among 37 families (Fig 3). A closer relative of C. phytofermentans, Butyrivibrio proteoclasticus (Fig 3), has a comparable number of GH domains (113), but less diversity with only 25 families and no exo-cellulase (GH48).
(A) Fermentation products during growth on 2% (w/v) cellobiose. Data are an average of two samples; error bars represent range. (B) Ethanol produced on a variety of substrates expressed as the molar percentage of non-gaseous products. All substrates were present at a concentration of 1% (w/v) except where otherwise indicated. The particle size of insoluble substrates was reduced by grinding; the substrates were not otherwise pre-treated. Fermentation products were measured after obvious growth ceased (3–5 days) at 30°C. In most cases, substrate conversion was incomplete.
(A) A conceptual illustration of how GH (blue), ABC transporters (purple) and AraC regulators (red) may work together. (B) Number of AraC transcriptional regulators per genome. (C) Number of GH domains per genome. Organisms having both GH48 and GH9 are marked with two asterisks, and organisms having GH9 alone are marked with one asterisk. (D) Number of putative ABC transporters per genome.
To gain insight into the origin of the GHs of C. phytofermentans, we identified the closest relatives of the GHs of C. phytofermentans in the GenBank database using BLASTP and compared their distribution to that of the closest relatives of all of the protein-coding genes within the C. phytofermentans genome. The latter analysis was performed to calibrate how much similarity to other bacteria would be expected on average. In total, approximately 40% of the GHs of C. phytofermentans were most similar to GHs present in species outside the class Clostridia (Fig 4), whereas only 20% of all the genes in the C. phytofermentans genome were most similar to genes from outside the Clostridia. The higher than expected proportion of GHs with distant relatives is statistically significant (Pearson's Chi-squared test, X-squared = 77.8583, df = 9, p-value = 4.299e-13) (Fig 4). This result suggests that horizontal gene transfer from diverse origins rather than vertical divergence from an ancestral genome played a key role in the assembly of the unique set GHs present in C. phytofermentans.
In some bacteria, notably C. cellulolyticum and C. thermocellum, lignocellulose-degrading enzymes are attached to complex extracellular structures called cellulosomes that are believed to be critical for efficient plant cell wall breakdown. However, there is no genomic evidence for the production of cellulosomes by C. phytofermentans (S4 File). In fact, two critical cellulases of C. phytofermentans, the GH9 family endocellulase and GH48 family exocellulase are more similar to the soluble cellulases of C. thermocellum than to cellulosomal cellulases [15]. The majority of the GHs of C. phytofermentans are multimodular. Carbohydrate-binding modules (CBMs) are found within 17% of the GHs of C. phytofermentans, including the critical endocellulase (GH9) [15]. In the absence of a cellulosome, these CBM domains may enable GHs to adhere to plant cell wall substrates, facilitating degradation of the heterogeneous, highly cross-linked lignocellulose polysaccharides. Among the 31 GH enzymes predicted to be extracellular, 16 contain domains involved in anchoring proteins to the cell surface, including transmembrane helices and/or cell-wall binding domains, suggesting that these enzymes are cell-associated. Despite the absence of cellulosomal assembly domains, the striking multimodular nature of cellulosomal proteins, in which multiple domains from diverse families of GH, CE, PL and carbohydrate-binding modules (CBM) are found within individual proteins, is preserved in C. phytofermentans (Table A in S4 File). C. phytofermentans has 19 multimodular GH proteins, representing about 17% of all putative GH genes (Table A in S4 File). In fact, the largest protein in the proteome is the multimodular glycoside hydrolase family 10 protein Cphy_3862, with 2457 amino acids and a predicted molecular weight of 266 kD [16]. This protein contains consecutive GH10, CE15, and CBM domains. In non-cellulolytic bacteria, the corresponding GH domains are found mainly in single-domain polypeptides, which are cytosolic and act on smaller, soluble carbohydrate substrates [59]. Thus, the multi-modular organization that seems to be characteristic of enzymes from cellulolytic species, may reflect their involvement in the extracellular processing of heterogeneous insoluble substrates, such as plant cell walls [59]. Biofilm formation may also play an important role in the orchestration of the degradation of the plant cell wall polysaccharides. Cells might adhere to each other via a variety of different domains such as pfam07705 (CARDB, cell adhesion domain in bacteria) and pfam01391 (Collagen, Collagen triple helix repeat), both of which are found in the C. phytofermentans genome.
Genes potentially involved in carbohydrate transport
Further examination of the genome revealed 148 genes encoding subunits of ATP-binding cassette (ABC) transporters, more than found in other clostridia (Fig 3). These genes are typically organized in operons consisting of two permeases and one solute-binding component. The majority of the ABC transporter-encoding operons lack an ATPase, suggesting that these transporter complexes may interact with a multitasking ATPase. Cphy_3611 is similar to MsmX of Bacillus subtilis, which is proposed to be an ATPase for several oligosaccharide transporters [60]. These findings suggest that C. phytofermentans is capable of active uptake of a diverse array of metabolites, including multiple oligosaccharides and simple sugars. The presence of GH genes adjacent to 50% of the transporter loci, suggests that carbohydrate degradation and uptake are frequently coupled. C. phytofermentans may feed cytoplasmic oligosaccharides into glycolysis via cellobiose/cellodextrin phosphorylases as occurs in other cellulolytic bacteria [8]. Import of oligosaccharides followed by internal hydrolysis via phosphorolysis minimizes ATP consumption [3].
Genes potentially involved in the regulation of carbohydrate metabolism
To orchestrate the regulation of diverse metabolic pathways in response to changing growth substrates, C. phytofermentans has numerous transcriptional regulators, including 70 AraC (Fig 3) and 23 PurR family members. AraC regulators typically activate transcription of genes involved in carbon metabolism, stress responses and pathogenesis [61], whereas PurR regulators act as repressors [62]. The abundance of these regulators suggests a complex regulatory network allowing rapid adaptation to varying substrate availability. Among the ABC-transporter genes found clustered with GHs, 50% are adjacent to AraC and 25% to PurR regulator genes.
Analysis of gene expression during growth on a variety of simple and complex carbohydrates
We designed a custom Affymetrix GeneChip to identify genes expressed in C. phytofermentans during growth on monosaccharides that are common in plant cell walls (glucose, galactose, xylose, arabinose, mannose), purified polysaccharides (cellobiose, cellulose, xylan and pectin) and fibrous plant biomass (Brachypodium distachyon) (S4 File and S6 File). These microarray studies suggest that C. phytofermentans regulates the stoichiometry of the plant degradative and assimilatory machinery in response to growth substrate. When C. phytofermentans was cultured with glucose, genes involved in biomass degradation (e.g. cellulase and xylanase) were essentially off (Fig 5), and the most abundant transcript was a putative ABC monosaccharide transporter (S6 File). During growth on xylose and xylan, transcripts for enzymes involved in pentose interconversion (xylose isomerase (Cphy_0200, and Cphy_1219) and xylulokinase (Cphy_3419) were among the most highly expressed (S6 File). When C. phytofermentans was grown with cellulose as substrate, the GH9 cellulase gene was among the most abundant transcripts (Fig 5, cellulase_Cphy_3368). This cellulase gene has been shown by gene inactivation to be essential for growth on cellulose in C. phytofermentans [15]. On nearly all substrates tested, we observed specific sets of co-regulated groups of genes, often consisting of GHs, an ABC transporter and a transcriptional regulator (Tables C and D in S4 File and S5 File). The putative multitasking ABC transporter ATPase subunit Cphy_3611, was expressed during growth on all substrates (transcript abundance within the 50th percentile) (S5 File). To orchestrate the regulation of these genes, a number of transcriptional regulators, typically physically close to the transporters and CAZy, are highly expressed on a given substrate (Tables C and D in S4 File). Thus, microarray experiments facilitated identification of enzymes involved in the breakdown and transport of specific carbohydrates. In addition, expression profiling with defined substrates was useful for deciphering data from more complex fibrous substrates. When plant biomass was used as growth substrates, GH expression profiles were similar to each other and to profiles with cellulose as substrate, with the exception that a putative xylanase (Cphy_2105) and mannanase (Cphy_1071) were more highly expressed on the plant biomass than on cellulose or xylan (S4 File). Thus, gene expression analysis proved to be a useful strategy for deciphering the functions of diverse enzymes involved in lignocellulose degradation (Fig 5, S4 File).
Transcript rank abundance curves during growth on (A) glucose, (B) hemicellulose, (C) cellulose and (D) Brachypodium. ADH_Cphy_1029 refers to a putative alcohol dehydrogenase. Cellulase_Cphy_3368 denotes the putative cellulose. Xylanase_Cphy_2105 denotes the putative xylanase.
Genomics and transcriptomics investigation of C. phytofermentans central metabolism
Perhaps the most industrially relevant property of C. phytofermentans is that it produces ethanol as the major fermentation product during growth on a wide variety of substrates including, simple sugars, cellulose and minimally processed plant biomass [12] (Fig 2). The fact that C. phytofermentans produces predominantly ethanol, suggests that it can maintain its redox balance without forming equivalent levels of lactate and/or formate and that it can generate sufficient energy for growth in the absence of high levels of acetate synthesis, which yields ATP via substrate-level phosphorylation. To gain insight into the basis for high levels of ethanol production, we used a combination of transcriptional profiling and comparative genomic analysis to identify a subset of genes that were both highly expressed on all growth substrates and predicted to be involved in ethanol production, energy conservation, and/or redox balance. The results of this analysis are the basis of a simplified model of the core physiology of C. phytofermentans (Fig 6, S7 File) and indicate that high levels of ethanol production may be due to a combination of factors. Firstly, pyruvate appears to be funneled to ethanol. The levels of the transcripts of the enzymes within the ethanol biosynthesis pathway are extremely high, and exceed those of all enzymes involved in the synthesis of alternate carbon fermentation products (Fig 5, Table A in S7 File). In particular, two alcohol dehydrogenases (ADH), Cphy_3925 and Cphy_1029, were constitutively transcribed at levels that rivaled or exceeded those of many ribosomal protein genes (average transcript abundances within the 98th percentile, Fig 5 and Table A in S7 File). Examination of these genes revealed that both NADH and NADPH are likely to contribute to ethanol production, another factor that may increase ethanol production. Secondly, reduced ferredoxin generated during conversion of pyruvate to the ethanol precursor, acetyl-CoA, by pyruvate ferredoxin oxidoreductase may contribute to ethanol production both directly, through reduction of NAD and NADP and indirectly, by participating in energy conservation. Two constitutively highly expressed protein complexes are likely to play a role in enabling reduced ferredoxin to contribute to ethanol production: NfnAB, an NADH-dependent reduced ferredoxin:NADP oxidoreductase [63] and Rnf, a sodium-translocating NADH:ferredoxin oxidoreductase [64]. C. phytofermentans may be able to exploit the sodium gradient produced by Rnf for energy production by way of the highly expressed sodium-translocating F1Fo-ATPase (Cphy_3735–42). Thus, Rnf may contribute to favorable energetics of ethanol production and reduce dependence on ATP generation via acetate production. Hydrogenases may also play an important role in ferredoxin metabolism in C. phytofermentans. In other clostridia, hydrogenases dissipate excess ferredoxin-reducing equivalents [65,66]. C. phytofermentans generates free hydrogen as a product of the fermentation of cellulose and cellobiose [67], and three cytoplasmic [FeFe]-hydrogenase-encoding clusters, one encoding a putative ferredoxin-dependent hydrogenase and two encoding NAD-dependent hydrogenases were constitutively highly expressed. The simultaneous expression of a ferredoxin-oxidizing hydrogenase with NAD-dependent hydrogenases, which could catalyze hydrogen-dependent NAD reduction and feed ferredoxin reducing equivalents into ethanol production, may prevent excess hydrogen accumulation thus enabling C. phytofermentans to maintain a high rate of ferredoxin turnover. Finally, C. phytofermentans may further reduce the requirement for acetate production by utilizing pyrophosphate–dependent glycolytic enzymes, which can substantially increase the ATP yield of glycolysis.
Conclusions
Analysis of the C. phytofermentans genome revealed a diverse array of genes for metabolism of lignocellulosic biomass and production of alcohols and hydrogen that constitute a unique repertoire among sequenced clostridial genomes with relevance for the biofuels industry. Our analysis of the genome revealed the genomic basis for the generalist behavior of this microbe. The diverse CAZy likely enable complete hydrolysis of cellulosic and hemicellulosic substrates. Unexpectedly, we found no evidence of cellulosomes in C. phytofermentans, suggesting that C. phytofermentans has evolved alternative strategies to optimize degradation and uptake of plant cell wall components [16]. The absence of a cellulosome simplifies strategies for engineering levels of individual enzymes to improve the conversion of plant biomass to fermentable sugars. Many active sugar transporters are in close proximity to polysaccharide hydrolases, likely cooperating for efficient simultaneous degradation and uptake of carbohydrate growth substrates. Further investigation will be required to determine the substrate-specificity of these transporters.
Genomic analysis and transcriptional profiling also suggest that high levels of ethanol production by C. phytofermentans may be due to a combination of factors. These include: 1) increasing the energetic yield of glycolysis by utilizing pyrophosphate-dependent enzymes; 2) high levels of expression of the enzymes involved in ethanol production coupled with the ability to utilize both NADH and NADPH for ethanol biosynthesis; 3) the presence of multiple pathways for the dissipation of excess reducing equivalents; and 4) the presence of sodium-dependent energy generating pathways. Experimental studies will be required to determine if these hypotheses are indeed valid. Examination of the genomes of several well-studied cellulolytic and solventogenic clostridia indicated that very few of the central metabolic enzymes and complexes discussed above are unique to C. phytofermentans. It may therefore be the specific combination of enzymes and their transcriptional regulation that makes C. phytofermentans metabolism unique.
Efficient direct conversion of biomass to bioproducts using a microbial catalyst such as C. phytofermentans requires an increased understanding of cell growth dynamics, rate-limiting steps of biomass conversion, enzyme production and regulation. These genome-based experiments and analyses provide a blueprint for identifying bottlenecks and guiding strategies to generate novel productive strains for specific uses.
Supporting Information
S3 File. Bacterial microcompartment (BMC) loci genes.
https://doi.org/10.1371/journal.pone.0118285.s003
(PDF)
S4 File. Genes involved in complex carbohydrate metabolism: diversity, functions, and origins.
https://doi.org/10.1371/journal.pone.0118285.s004
(PDF)
S5 File. Expression data from microarray: raw expression values, averages and standard deviations, percentile ranking, ratio compared to glucose.
https://doi.org/10.1371/journal.pone.0118285.s005
(XLS)
S7 File. Investigation of C. phytofermentans central metabolism: Insights into high levels of ethanol production.
https://doi.org/10.1371/journal.pone.0118285.s007
(PDF)
Acknowledgments
We thank Phyllis Spatrick of the University of Massachusetts Medical Center for performing the cDNA synthesis, array hybridization and imaging.
Accession numbers
Platform: GPL7481 18 Samples, Accession: GSE52495, ID: 200052495.
Author Contributions
Conceived and designed the experiments: JLB JCH SBL EP. Performed the experiments: JCH BH NI NK EP TW. Analyzed the data: DA AB JLB GMC MVC LH JCH BH NI NK ML WGL JL SBL AL SM EP DJS ACT TW. Contributed reagents/materials/analysis tools: JLB SBL DJS. Wrote the paper: JLB MVC SBL EP ACT.
References
- 1. Antoni D, Zverlov VV, Schwarz WH. Biofuels from microbes. Appl Microbiol Biotechnol. 2007; 77: 23–35. pmid:17891391
- 2. Lynd LR, Cushman JH, Nichols RJ, Wyman CE. Fuel ethanol from cellulosic biomass. Science. 1991; 251: 1318–1323. pmid:17816186
- 3.
Zhang YHP, Lynd LR. New generation biomass processing: Consolidated bioprocessing. Biomass Recalcitrance: Deconstructing the Plant Cell Wall for Bioenergy. Chichester, UK: Wiley-Blackwell; 2008. pp. 480–494.
- 4. Lynd LR, van Zyl WH, McBride JE, Laser M. Consolidated bioprocessing of cellulosic biomass: an update. Curr Opin Biotechnol. 2005; 16: 577–583. pmid:16154338
- 5. Ezeji TC, Qureshi N, Blaschek HP. Butanol fermentation research: upstream and downstream manipulations. Chem Rec. 2004; 4: 305–314. pmid:15543610
- 6. Demain AL, Newcomb M, Wu JHD. Cellulase, clostridia, and ethanol. Microbiol Mol Biol Rev. 2005; 69: 124–154. pmid:15755956
- 7. Desvaux M. Clostridium cellulolyticum: model organism of mesophilic cellulolytic clostridia. FEMS Microbiol Rev. 2005; 29: 741–764. pmid:16102601
- 8. Desvaux M. Unravelling carbon metabolism in anaerobic cellulolytic bacteria. Biotechnol Prog. 2006; 22: 1229–1238. pmid:17022659
- 9. Lynd LR, Weimer PJ, van Zyl WH, Pretorius IS. Microbial cellulose utilization: fundamentals and biotechnology. Microbiol Mol Biol Rev. 2002; 66: 506–577. pmid:12209002
- 10. Schwarz WH. The cellulosome and cellulose degradation by anaerobic bacteria. Appl Microbiol Biotechnol. 2001; 56: 634–649. pmid:11601609
- 11. Mingardon F, Perret S, Belaich A, Tardif C, Belaich JP, Fierobe HP. Heterologous production, assembly, and aecretion of a minicellulosome by Clostridium acetobutylicum ATCC 824. Appl Environ Microbiol. 2005; 71: 1215–1222. pmid:15746321
- 12. Warnick TA, Methé BA, Leschine SB. Clostridium phytofermentans sp. nov., a cellulolytic mesophile from forest soil. Int J Syst Evol Microbiol. 2002; 52: 1155–1160. pmid:12148621
- 13. VanFossen AL, Verhaart MRA, Kengen SMW, Kelly RM. Carbohydrate utilization patterns for the extremely thermophilic bacterium Caldicellulosiruptor saccharolyticus reveal broad growth substrate preferences. Appl Environ Microbiol. 2009; 75: 7718–7724. pmid:19820143
- 14. Jin M, Gunawan C, Balan V, Dale BE. Consolidated bioprocessing (CBP) of AFEXTM-pretreated corn stover for ethanol production using Clostridium phytofermentans at a high solids loading. Biotechnol Bioeng. 2012; 109: 1929–1936. pmid:22359098
- 15. Tolonen AC, Chilaka AC, Church GM. Targeted gene inactivation in Clostridium phytofermentans shows that cellulose degradation requires the family 9 hydrolase Cphy3367. Mol Microbiol. 2009; 74: 1300–1313. pmid:19775243
- 16. Tolonen AC, Haas W, Chilaka AC, Aach J, Gygi SP, Church GM. Proteome-wide systems analysis of a cellulosic biofuel-producing microbe. Mol Syst Biol. 2011; 7: 461. pmid:21245846
- 17. Zhang XZ, Sathitsuksanoh N, Zhang YHP. Glycoside hydrolase family 9 processive endoglucanase from Clostridium phytofermentans: heterologous expression, characterization, and synergy with family 48 cellobiohydrolase. Bioresour Technol. 2010; 101: 5534–5538. pmid:20206499
- 18. Zhang XZ, Zhang Z, Zhu Z, Sathitsuksanoh N, Yang Y, Zhang YH. The noncellulosomal family 48 cellobiohydrolase from Clostridium phytofermentans ISDg: heterologous expression, characterization, and processivity. Appl Microbiol Biotechnol. 2010; 86: 525–533. pmid:19830421
- 19. Liu W, Zhang XZ, Zhang Z, Zhang YHP. Engineering of Clostridium phytofermentans Endoglucanase Cel5A for Improved Thermostability. Appl Environ Microbiol. 2010; 76: 4914–4917. pmid:20511418
- 20. Lee SJ, Warnick TA, Pattathil S, Alvelo-Maurosa JG, Serapiglia MJ, McCormick H, et al. Biological conversion assay using Clostridium phytofermentans to estimate plant feedstock quality. Biotechnol Biofuels. 2012; 5: 5. pmid:22316115
- 21. Nakajima M, Nishimoto M, Kitaoka M. Characterization of three beta-galactoside phosphorylases from Clostridium phytofermentans: discovery of d-galactosyl-beta1->4-l-rhamnose phosphorylase. J Biol Chem. 2009; 284: 19220–19227. pmid:19491100
- 22. Nihira T, Nakai H, Kitaoka M. 3-O-α-d-Glucopyranosyl-l-rhamnose phosphorylase from Clostridium phytofermentans. Carbohydr Res. 2012; 350: 94–97. pmid:22277537
- 23. Brat D, Boles E, Wiedemann B. Functional expression of a bacterial xylose isomerase in Saccharomyces cerevisiae. Appl Environ Microbiol. 2009; 75: 2304–2311. pmid:19218403
- 24. Demeke MM, Dietz H, Li Y, Foulquié-Moreno MR, Mutturi S, Deprez S, et al. Development of a D-xylose fermenting and inhibitor tolerant industrial Saccharomyces cerevisiae strain with high performance in lignocellulose hydrolysates using metabolic and evolutionary engineering. Biotechnol Biofuels. 2013; 6: 89. pmid:23800147
- 25. Zuroff TR, Barri Xiques S, Curtis WR. Consortia-mediated bioprocessing of cellulose to ethanol with a symbiotic Clostridium phytofermentans/yeast co-culture. Biotechnol Biofuels. 2013; 6: 59. pmid:23628342
- 26.
JGI—Protocols Provided by the JGI User Community (n.d.). Available: http://my.jgi.doe.gov/general/protocols.html. Accessed 2012 September 18.
- 27. Fleischmann RD, Adams MD, White O, Clayton RA, Kirkness EF, Kerlavage AR, et al. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science. 1995; 269: 496–512. pmid:7542800
- 28. Ewing B, Green P. Base-calling of automated sequencer traces using phred. II. Error probabilities. Genome Res. 1998; 8: 186–194. pmid:9521922
- 29.
De la Bastide M, McCombie WR. Assembling genomic DNA sequences with PHRAP. Curr Protoc Bioinforma. 2007; Chapter 11: Unit11.4. https://doi.org/10.1002/0471250953.bi1104s17
- 30. Gordon D, Abajian C, Green P. Consed: a graphical tool for sequence finishing. Genome Res. 1998; 8: 195–202. pmid:9521923
- 31. Badger JH, Olsen GJ. CRITICA: coding region identification tool invoking comparative analysis. Mol Biol Evol. 1999; 16: 512–524. pmid:10331277
- 32. Delcher AL, Bratke KA, Powers EC, Salzberg SL. Identifying bacterial genes and endosymbiont DNA with Glimmer. Bioinformatics. 2007; 23: 673–679. pmid:17237039
- 33. Kanehisa M, Goto S, Sato Y, Furumichi M, Tanabe M. KEGG for integration and interpretation of large-scale molecular data sets. Nucleic Acids Res. 2012; 40: D109–114. pmid:22080510
- 34. Kanehisa M, Goto S, Furumichi M, Tanabe M, Hirakawa M. KEGG for representation and analysis of molecular networks involving diseases and drugs. Nucleic Acids Res. 2010; 38: D355–360. pmid:19880382
- 35. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000; 28: 27–30. pmid:10592173
- 36. Hunter S, Jones P, Mitchell A, Apweiler R, Attwood TK, et al. InterPro in 2011: new developments in the family and domain prediction database. Nucleic Acids Res. 2012; 40: D306–D312. pmid:22096229
- 37. Bateman A, Bireny E, Cerruti L, Durbin R, Etwiler L, Eddy SR, et al. The Pfam protein families database. Nucleic Acids Res. 2002; 30: 276–280. pmid:11752314
- 38. Haft DH, Loftus BJ, Richardson DL, Yang F, Eisen JA, Paulsen IT, et al. TIGRFAMs: a protein family resource for the functional identification of proteins. Nucleic Acids Res. 2001; 29: 41–43. pmid:11125044
- 39. Ponting CP, Schultz J, Milpetz F, Bork P. SMART: identification and annotation of domains from signalling and extracellular protein sequences. Nucleic Acids Res. 1999; 27: 229–232. pmid:9847187
- 40. Sigrist CJ, Cerutti L, de Castro E, Langendijk-Genevaux PS, Bulliard V, Bairoch A, et al. PROSITE, a protein domain database for functional characterization and annotation. Nucleic Acids Res. 2010; 38: D161–166. pmid:19858104
- 41.
Attwood TK, Coletta A, Muirhead G, Pavlopoulou A, Philippou PB, et al. The PRINTS database: a fine-grained protein sequence annotation and analysis resource—its status in 2012. Database 2012; bas019. https://doi.org/10.1093/database/bas019
- 42. Corpet F, Servant F, Gouzy J, Kahn D. ProDom and ProDom-CG: tools for protein domain analysis and whole genome comparisons. Nucleic Acids Res. 2000; 28: 267–269. pmid:10592243
- 43.
McEntyre JO. The Clusters of Orthologous Groups (COGs) Database: Phylogenetic classification of proteins from complete genomes. 2002; Available: http://www.ncbi.nlm.nih.gov/books/NBK21090/. Accessed 2012 May 24.
- 44. Ren Q, Chen K, Paulsen IT. TransportDB: a comprehensive database resource for cytoplasmic membrane transport systems and outer membrane channels. Nucleic Acids Res. 2007; 35: D274–279. pmid:17135193
- 45. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 1997; 25: 4876–4882. pmid:9396791
- 46. Felsenstein J. PHYLIP—Phylogeny Inference Package (Version 3.2). Cladistics. 1989; 5: 164–166.
- 47. Besemer J, Borodovsky M. GeneMark: web software for gene finding in prokaryotes, eukaryotes and viruses. Nucleic Acids Res. 2005; 33: W451–454. pmid:15980510
- 48. Galtier N, Gouy M, Gautier C. SeaView and Phylo_win, two graphic tools for sequence alignment and molecular phylogeny. Comput Appl Biosci. 1996; 12: 543. pmid:9021275
- 49. Hungate RE. A roll tube method for cultivation of strict anaerobes. Methods Microbiol 1969; 3B: 117.
- 50.
The MAS 5.0 User Manual (n.d.). Available: http://www.affymetrix.com/Auth/support/downloads/manuals/mas_manual.zip.
- 51. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostat. 2003; 4: 249–264. pmid:14615633
- 52.
Gene Expression Omnibus Database (n.d.). Available: http://www.ncbi.nlm.nih.gov/geo/.
- 53. Mahowald MA, Rey FE, Seedorf H, Turnbaugh PJ, Fulton RS, Wollam A, et al. Characterizing a model human gut microbiota composed of members of its two dominant bacterial phyla. Proc Natl Acad Sci U S A. 2009; 106: 5859–5864. pmid:19321416
- 54. Errington J. Regulation of endospore formation in Bacillus subtilis. Nat Rev Microbiol 2003; 1: 117–126. pmid:15035041
- 55. Hilbert DW, Piggot PJ. Compartmentalization of gene expression during Bacillus subtilis spore formation. Microbiol Mol Biol Rev. 2004; 68: 234–262. pmid:15187183
- 56. Barák I, Wilkinson AJ. Where asymmetry in gene expression originates. Mol Microbiol. 2005; 57: 611–620. pmid:16045607
- 57. Kerfeld CA, Heinhorst S, Cannon GC. Bacterial microcompartments. Annu Rev Microbiol. 2010; 64: 391–408. pmid:20825353
- 58. Petit E, LaTouf WG, Coppi MV, Warnick TA, Currie D, et al. Involvement of a bacterial microcompartment in the metabolism of fucose and rhamnose by Clostridium phytofermentans. PLoS ONE 2013; 8: e54337. pmid:23382892
- 59. Flint HJ, Bayer EA, Rincon MT, Lamed R, White BA. Polysaccharide utilization by gut bacteria: potential for new insights from genomic analysis. Nat Rev Microbiol. 2008; 6: 121–131. pmid:18180751
- 60. Ferreira MJ, de Sá-Nogueira I. A multitask ATPase serving different ABC-type sugar importers in Bacillus subtilis. J Bacteriol. 2010; 192: 5312–5318. pmid:20693325
- 61. Gallegos MT, Schleif R, Bairoch A, Hofmann K, Ramos JL. Arac/XylS family of transcriptional regulators. Microbiol Mol Biol Rev. 1997; 61: 393–410. pmid:9409145
- 62. Weickert MJ, Adhya S. A family of bacterial regulators homologous to Gal and Lac repressors. J Biol Chem. 1992; 267: 15869–15874. pmid:1639817
- 63. Wang S, Huang H, Moll J, Thauer RK. NADP+ reduction with reduced ferredoxin and NADP+ reduction with NADH are coupled via an electron-bifurcating enzyme complex in Clostridium kluyveri. J Bacteriol. 2010; 192: 5115–5123. pmid:20675474
- 64. Biegel E, Müller V. Bacterial Na+-translocating ferredoxin:NAD+ oxidoreductase. Proc Natl Acad Sci U S A. 2010; 107: 18138–18142. pmid:20921383
- 65. Thauer RK, Jungermann K, Decker K. Energy conservation in chemotrophic anaerobic bacteria. Bacteriol. 1977; Rev 41: 100–180.
- 66. Adams MW, Mortenson LE, Chen JS. Hydrogenase. Biochim Biophys Acta. 1980; 594: 105–176. pmid:6786341
- 67. Ren Z, Ward TE, Logan BE, Regan JM. Characterization of the cellulolytic and hydrogen-producing activities of six mesophilic Clostridium species. J Appl Microbiol. 2007; 103: 2258–2266. pmid:18045409
- 68. Collins MD, Lawson PA, Willems A, Cordoba JJ, Fernandez-Garayzabal J, et al. The phylogeny of the genus Clostridium: proposal of five new genera and eleven new species combinations. Int J Syst Bacteriol. 1994; 44: 812–826. pmid:7981107