Clostridium manihotivorum sp. nov., a novel mesophilic anaerobic bacterium that produces cassava pulp-degrading enzymes

Preparation of samples and basal medium

Samples of cassava pulp and soil beneath the waste heap were obtained from a starch factory landfill in Chonburi Province, Thailand. The pulp was ground by an ultra centrifugal mill ZM-100 and sieved through a 0.5 mm mesh screen (Retsch, Haan, Germany). The pulp was washed several times with distilled water to remove the remaining sugar and other dirt, oven-dried at 50 °C until at a constant weight and then stored in plastic bags at 4 °C for further experiments.

The basal medium (BM7; pH 7.0) was composed of (per liter) 1.5 g KH₂PO₄, 2.9 g K₂HPO₄, 2.1 g urea, 4.5 g yeast extract, 0.5 g cysteine-HCl, 0.001 g resazurin and 200 µL mineral solution (25.0 g/L MgCl₂.6H₂O, 37.5 g/L CaCl₂.2H₂O and 0.3 g/L FeSO₄.6H₂O). The BM7 was anaerobically prepared in bottles sealed with butyl rubber stoppers, under an atmosphere of high-purity N₂, and sterilized by autoclaving at 121 °C for 15 min.

Screening and isolation of cassava pulp-degrading strains

The enrichment and isolation were performed under anaerobic conditions. Approximately 1 g of the soil sample was transferred into Hungate tubes containing 15 mL BM7 (pH 7.0) and 1% (w/v) cassava pulp. After inoculation, each test tube was flushed with N₂ and incubated at 37 °C. The culture that showed the highest degradation of pulp, as visually indicated by the remaining cassava pulp (approximately ≤ 50% residue dry weight), was selected and serially diluted into agar-cassava pulp medium that had been preliminarily melted and cooled to 55 °C. The cultures were then subjected to the roll-tube technique for isolating obligate anaerobes (Hungate, 1969), after which solidified samples were incubated at 37 °C. Single colonies were isolated with sterile needles and inoculated into BM7 broth containing cassava pulp. Afterward, the cultures were incubated to study their ability to degrade the cassava pulp. Pure cultures were obtained following repeated sub-culturing (ten times) in BM7 containing cassava pulp.

The composition of cassava pulp and residual cassava pulp digested by C. manihotivorum CT4^T, C. polyendosporum PS-1^T and C. amylolyticum SW408^T were analyzed following the National Renewable Energy Laboratory (NREL) protocol (Sluiter et al., 2008).

16S rRNA gene sequencing

Genomic DNA for 16S rRNA gene sequencing was prepared by phenol-chloroform extraction. The 16S rRNA gene was amplified by PCR using the following primers: 8F (5′–AGAGTTTGATCCTGGCTCAG–3′) and 1492R (5′–GGTTACCTTGTTACGACTT–3′). The PCR reaction conditions were as follows: 94 °C for 3 min, 35 cycles at 94 °C for 30 s, 55 °C for 40 s and 72 °C for 2 min, with a final extension time of 5 min at 72 °C. The amplified fragment was ligated into the pGEM-T Easy vector (Promega, Madison, WI, USA), and the recombinant plasmid was sequenced using T7 and SP6 primers. A sequence similarity search was performed using the BLAST program (http://blast.ncbi.nlm.nih.gov/Blast.cgi). A phylogenetic tree was generated by the neighbor-joining method with 1,000 bootstrap replications, employing the MEGA version 6.0 (Tamura et al., 2013).

Physiological and biochemical analysis

Gram staining of strain CT4^T was conducted using the conventional methodology and confirmed using the KOH test (Powers, 1995). Endospore staining was examined by Schaeffer–Fulton’s method (Schaeffer & Fulton, 1933). Cell morphology was observed by scanning electron microscope (SEM; model SU800 Hitachi, Japan). Substrate utilization was tested by growing the strain in BM7 containing 0.5% (w/v) of the following substrates: D-glucose, D-galactose, D-arabinose, D-rhamnose, D-mannose, D-xylose, D-fructose, D-trehalose, D-raffinose, lactose, sucrose, maltose, cellobiose, mannitol, soluble starch from potato (ACS reagent), birchwood xylan, cellulose powder (Type 20) and Avicel^® (PH-101); these chemicals were purchased from Sigma-Aldrich, Saint Louis, MO, USA. Raw cassava starch and cassava pulp were obtained from a starch factory landfill in Chonburi Province, Thailand. After 5 days of incubation, cell growth was assessed by determining the optical density at 600 nm. The fermentation products in the supernatant extracted from the glucose-grown culture were determined by gas chromatography equipped with a flame-ionization detector and a Carbopack B-DA/4% Carbowax 20M column (GC-14A; Shimadzu, Japan). The column, injector and detector temperatures were 170, 230 and 230 °C, respectively. Other biochemical tests were conducted according to the methods of Holdeman, Cato & Moore (1977) and Summanen et al. (1993). The isomers of diaminopimelic acid (DAP) in the cell wall were determined as described by Komagata and Suzuki (Komagata & Suzuki, 1988). Cellular fatty acids were extracted, methylated and analyzed using the standard microbial identification system (MIDI) protocol (Sherlock microbial identification system, version 6.1) while the fatty acids were identified using the TSBA6 database of the microbial identification system (Sasser, 1990). The polar lipids were analyzed from freeze-dried cells by two-dimensional TLC, as described by Minnikin et al. (1984). Appropriate detection reagents were used to visualize the spots: phosphomolybdic acid reagent 5% (w/v) solution in ethanol (Sigma-Aldrich, Saint Louis, MO, USA) was used to detect total polar lipids; ninhydrin reagent (0.2% solution; Sigma-Aldrich) was used to detect amino lipids; Dittmer and Lester reagent (molybdenum blue reagent, 1.3%; Sigma-Aldrich) was used to detect phospholipids and Dragendorff’s reagent (Sigma-Aldrich) was used to detect phosphatidylcholine.

Cultivation and enzyme production

Strain CT4^T was anaerobically cultivated in 1,000 mL of BM7 containing 1% (w/v) cassava pulp for 3 days at 37 °C, pH 7.0 under static conditions in an anaerobic chamber (Bactron II, USA). The culture supernatant was collected by centrifugation at 10,000× g for 10 min at 4 °C and subsequently concentrated using a hollow fiber cartridge with a 10-kDa-cutoff membrane (GE Healthcare, USA). The retentate (50-times more concentrated) was then used as the crude enzyme.

Enzyme assays and protein determination

All enzyme assays were performed in 50 mM sodium phosphate buffer (pH 7.0) and incubated at 55 °C for 10 min. The enzymatic activities on 1% (w/v) cassava pulp, soluble starch, pullulan, birchwood xylan, cellulose powder (Type 20) and pectin from citrus peel were assayed by determining the amount of reducing sugars liberated by the Somogyi–Nelson method (Nelson, 1944). One unit (U) of enzyme activity was defined as the amount of enzyme that released 1 µmol of reducing sugar in 1 min under assay conditions. The protein concentration was determined by the Lowry method (Lowry et al., 1951) using bovine serum albumin as a standard.

Library preparation and genome sequencing

The genomic DNA was extracted from the cultures using a blood and cell culture DNA midi kit (Qiagen, Germany) according to the manufacturer’s instructions. Strain CT4^T was sequenced using two sequencing platforms: Illumina (Hiseq2500) and ONT (MinION). Illumina sequencing paired-end DNA libraries were prepared following the Illumina DNA manufacturer’s instructions (NEBNext^® Ultra™ DNA library prep kit). The size-selected, adaptor-ligated DNA fragments were PCR-amplified using the following protocol: polymerase activation (98 °C for 30 s), followed by 10 cycles (denaturation at 98 °C for 10 s, annealing at 65 °C for 75 s and extension at 65 °C for 75 s) with a final 5 min extension at 65 °C. The DNA libraries were purified by magnetic beads, and their size distribution was checked using Agilent Bioanalyzer DNA high sensitivity chip assay. The DNA fragments were sequenced using the Illumina Hiseq2500 with 2 × 150 bp paired-end protocol (Illumina, Inc., California, USA).

The ONT library preparation and bioinformatics analysis were performed according to Jenjaroenpun et al. (2018). In brief, a total amount of 600 ng genomic DNA was used as input for a rapid sequencing kit (SQK-RAD002) to generate the DNA sequencing library. The library was then loaded onto a flow cell version FLO-MIN106 on a MinION Mk1B (released in 2014 through the MinION Access Program) (Oxford, UK) to perform DNA sequencing for 48 h. Base-calling was performed using the local-based software, Albacore version 1.2.3 (ONT, USA).

Genome assembly and annotation

The high-quality ONT reads (average quality score of >7) were first assembled using combination Minimap2 (Li, 2018) and Miniasm (Li, 2016), resulting in a circular draft genome. The draft genome was polished using Racon (Vaser et al., 2017) and Nanopolish (Loman, Quick & Simpson, 2015), based on the consensus pileup of high-quality ONT reads and additionally using Pilon (Walker et al., 2014), based on short Illumina reads.

Genome annotation was conducted with a Prokka annotation pipeline (Seemann, 2014). The rRNA and tRNA genes were identified with RNAmmer (Lagesen et al., 2007) and Aragorn software (Laslett & Canback, 2004), respectively. Functional classification of protein-coding genes in Clostridium sp. strain CT4^T was done by assigning Cluster of Orthologous Group (COG) codes to each gene using eggnog-Mapper (Huerta-Cepas et al., 2017) and eggNOG version 4.5 (Huerta-Cepas et al., 2016).

Average amino acid identity analysis

The first analysis comprised pairwise comparisons of AAIs (Konstantinidis & Tiedje, 2005) of the 665 genomes belonging to the Clostridia class (Cabal et al., 2018). For each pair of genomes, the average AAI was then calculated based on the identities of all conserved reciprocal best matches, a calculation that was not always symmetrical. In such cases, the average of the two AAI values was assigned to each pair of genomes. The AAI tree was built with BIONJ (Gascuel, 1997) to find dissimilarities of AAI values (100% minus AAI).

Comparison of glycoside hydrolase producing genes in strain CT4^T with related species

Strain CT4^T (GenBank accession number CP025746) was compared with the closely related C. amylolyticum SW408^T (NZ_FQZO00000000; NCBI) that had available genomic information in the NCBI database, using OrthoMCL (Chen et al., 2006) to characterize their specific genetic features and identify overlaps among orthologous clusters. The protein sequences were grouped into gene families encoding amylolytic-, hemicellulolytic- and cellulolytic-enzymes, using the criteria: E-value <1E-5 and sequence identity >50%. The genomic information of C. polyendosporum PS-1^T was not reported in the NCBI database, therefore, the strain PS-1^T was excluded from genome comparison.

Isolation and identification of cassava pulp-degrading bacterium

In total, 15 individual colonies were isolated by the roll-tube technique and were subcultured 10 times in BM7 separately, utilizing cassava pulp as carbon sources. Visualization of the roll-tube appearance revealed that isolate CT4^T performed best in relation to cassava pulp degradation. Moreover, approximately 60.8% (w/v) removal of dry weight was detected when cultured in BM7 broth. The isolated CT4^T thoroughly utilized the starch contains in cassava pulp after 5 days of culturing. The result showed that the isolated CT4^T could remove 99.0% starch in cassava pulp, while cellulose and hemicellulose contents were removed by 42.2% and 39.2%, respectively, when compared with the control. Besides, this strain removed starch and non-starch polysaccharide in cassava pulp better than the related species, strain PS-1^T and SW408^T (Fig. S2). Thus, it was consequently selected for further analysis. Prior to genome sequencing, the 16S rRNA gene sequence of strain CT4^T (accession number MH879026) was compared with the nucleotide sequences in NCBI. The analysis revealed that strain CT4^T shared 95% sequence identity with Anaerobacter polyendosporus PS-1^T, now reclassified as Clostridium polyendosporum comb. nov. (Duda et al., 1987; Stackebrandt et al., 1999) and 94% sequence identity with C. amylolyticum SW408^T (Song & Dong, 2008), Clostridium putrefaciens DSM 1291^T (Sturges & Drake, 1927), and Clostridium algidicarnis NCFB 2931^T (Lawson et al., 1994). Phylogenetic analysis based on 16S rRNA gene sequences and neighbor-joining method indicated that strain CT4^T belongs to the genus Clostridium (Fig. 1). Therefore, isolated CT4^T was classified as Clostridium sp. CT4^T

Physiological and biochemical characteristics of strain CT4^T

The SEM image revealed that cells of strain CT4^T were rod-shaped, and surrounded by a polysaccharide capsule (Fig. 2). Strain CT4^T was Gram-positive, single endospore-forming, non-motile and non-flagellate (Table 1). To understand the optimal growth conditions, strain CT4^T was cultivated under different pH (pH 4.0–11.0) and temperature (25−50 °C) conditions. Strain CT4^T could grow at a wide range of temperatures (25−45 °C) and pH (5.5–7.5) in BM7 medium containing 1% (w/v) cassava pulp. The optimum growth of strain CT4^T was found at 37 °C and pH 7.0. Moreover, strain CT4^T used a wide range of carbon sources, including D-glucose, D-xylose, D-galactose, D-fructose, D-mannose, D-arabinose, D-rhamnose, D-trehalose, D-raffinose, sucrose, lactose, maltose, mannitol, cellobiose, soluble starch, xylan, cellulose and Avicel^®. The main metabolic products of strain CT4^T, ranked based on quantity, were acetate, butyrate, ethanol and propionate. Butanol was not observed during growth, whereas it was produced in the closest relative, C. polyendosporum PS-1^T (Table 1). In contrast, C. putrefaciens isolated from spoiled ham (Sturges & Drake, 1927), and C. algidicarnis isolated from vacuum-packed refrigerated pork (Lawson et al., 1994) cannot hydrolyze starch despite their relatedness to strain CT4^T (Table 1). Strain CT4^T presented LL-diaminopimelic acid (LL-DAP) in their cell wall, whereas most members in the genus Clostridium contains meso-diaminopimelic acid. Thus, the strain CT4^T was different from the other related strains, except C. putrefaciens that have the same with strain CT4^T. The cellular fatty acid profiles of strain CT4^T are listed in Table S1. The major fatty acids detected from strain CT4^T were C _16:0 (37.4%), C _14:0 (15.0%), anteiso-C _15:0 (5.5%), summed feature 1 (C _13:0-3OH and/or C _15:1isoH; 4.5%), C _19:0cyclo ω8c (4.2%) and C _17:02-OH (4.0%). In terms of their polar lipid profiles, strain CT4^T contained phosphatidylethanolamine (PE) and phosphatidylglycerol (PG) as the major polar lipids, while phosphatidylcholine (PC) was found as minor polar lipid. Additionally, three unidentified phospholipids (PL1–PL3) and three unidentified amino lipids (AL1–AL3) were also detected (Fig. S1).

Based on the 16S rRNA gene sequence similarity, physiological attributes and biochemical properties, strain CT4^T was considered to be a novel species of the genus Clostridium. Thus, the strain CT4^T was introduced in the namely Clostridium manihotivorum CT4^T, which can degrade cassava pulp. The meaning of “manihotivorum” is devouring cassava. This bacterium was deposited as a type strain in the Thailand Bioresource Research Center (TBRC), and NITE Biological Resource Center (NBRC), Japan under accession numbers TBRC 11758^T and NBRC 114534^T, respectively.

Characterizations of amylolytic-, hemicellulolytic- and cellulolytic-enzymes of C. manihotivorum CT4^T

In this study, a C. manihotivorum CT4^T was discovered to degrade cassava pulp, which was able to produce the cassava pulp degrading enzymes, including amylolytic-, hemicellulolytic- and cellulolytic-enzymes. In order to characterize the properties of the crude enzyme from strain CT4^T, the isolate was cultivated in BM7 medium containing 1% (w/v) cassava pulp at pH 7.0, 37 °C. Afterwards, the culture supernatant was harvested at the early stationary phase (3 days) and concentrated by ultrafiltration technique. The crude enzyme gave the highest activity on cassava pulp (1,901.1 U/g protein), which was 1.56-fold higher than that obtained from soluble starch (1,212.7 U/g protein). In addition, a pullulanase activity of 27.5 U/g protein was detected (Table 2). C. manihotivorum CT4^T was also able to produce xylanase (43.5 U/g protein), cellulase (32.0 U/g protein) and pectinase (42.4 U/g protein) as shown in Table 2, which are involved in the degradation of xylan, cellulose and pectin contained in the cell wall structure of cassava pulp, respectively.

The complete genome of C. manihotivorum CT4^T and comparative genomics

In this study, the complete genome of C. manihotivorum CT4^T, deposited in GenBank under the accession number CP025746, was described. A complete, gapless and circular genome assembly was generated, with a total size of 6,364,326 bases and a 40-fold coverage, as shown in Fig. 3 and Table 3. The origin of replication was determined based on GC skew analyses. The average G + C content was approximately 32 mol%, and plasmid was not detected. The DNA G + C content of strain CT4 (32 mol%), was within the range of 23–37% reported for the genus Clostridium (Lawson & Rainey, 2016). Genome annotation was performed using Prokka (Seemann, 2014) and Blast2GO (Conesa et al., 2005). The genome was predicted to have 5,664 protein-coding sequences (CDS), 42 rRNA sequences, 95 tRNA sequences, 1 tmRNA sequence and 153 misc_RNA sequences. Furthermore, NCBI Prokaryotic Genome Annotation Pipeline (PGAP) version 4.11 was also employed to annotate the genome, which provided slightly different result of 5,308 CDSs and 5,654 total genes (Table 3). Hereafter, 5,664 CDSs were used for further analysis, in which the details are available in Data S1. According to the comparison of the genomes between C. manihotivorum CT4^T and C . amylolyticum SW408^T, the strain CT4^T has much larger genome size than the strain SW408^T about 2.1 Mb. Moreover, 5,664 CDSs were predicted in C. manihotivorum CT4^T whereas only 3,957 CDSs were reported in C. amylolyticum SW408^T.

Average amino acid identity and phylogenetic analysis

The whole-genome phylogeny of C. manihotivorum CT4^T was compared with a unique set of 665 Clostridia class genomes (Cabal et al., 2018), employing the average amino acid identity (AAI) analysis method. AAI has proven to have a better resolution power at the species level than 16S rRNA gene sequence-based comparison (Mahato et al., 2017). The derived phylogenetic tree based on the AAI analysis of all 666 genomes revealed several main clusters (Fig. 4). The strain CT4^T was clearly separated from the other Clostridia, in which a single branch was observed. The result aligned with the above physiological and biochemical characteristics that differ from other related type strains.

Functional category of strain CT4^T

Approximately, 75% (4,223 out of 5,664) of the protein-coding sequences in C. manihotivorum CT4^T were classified into COG functional categories (Table 4): replication, recombinant and repair (L: 548 protein-coding sequences); transcription (K: 381); carbohydrate transport and metabolism (G: 318); amino acid transport and metabolism (E: 276); cell wall/membrane/envelope biogenesis (M: 260) and translation, ribosomal structure and biogenesis (J: 197), based on Tatusov et al. (2000). The COG category (G: 318) comprised mainly protein-coding sequences that were involved in the degradation of starch and polysaccharides contained in lignocellulosic materials, and the transportation of the compounds (Tomazetto et al., 2016). These results suggest that C. manihotivorum CT4^T contains genes encoding glycoside hydrolases, related to starch, hemicellulose, cellulose and pectin degrading enzymes (Table 5).

Identification of the genes encoding amylolytic-, hemicellulolytic-, cellulolytic- and pectinolytic-enzymes in C. manihotivorum CT4^T

The genes encoding amylolytic-, hemicellulolytic-, cellulolytic- and pectinolytic-enzymes were detected in the genome of C. manihotivorum CT4^T (Table 5). Amylolytic enzymes were found in the complete genome of strain CT4^T including α-amylase, oligo-α-1,6-glucosidase, α-glucosidase, amylo-α-1,6-glucosidase, pullulanase and glucan-α-1,6-glucosidase that could be classified into endo-acting amylase, exo-acting amylase and debranching amylase. The α-amylase of C. manihotivorum CT4^T was predicted to contain starch binding domains (SBDs) of CBM20 (gene locus; CT4_05358) and CBM53 (CT4_01439) while α-glucosidase contained CBM34 (CT4_04500 and CT4_04692), as shown in Table 5. Moreover, the genes encoding hemicellulolytic- and cellulolytic-enzymes, such as endo-1,4-β-xylanase, α-galactosidase, β-galactosidase, β-glucuronidase, β-xylosidase, α-L-arabinofuranosidase, endo-β-1,4-mannanase, endo-α-1,5-L-arabinanase, endo-β-1,4-glucanase, β-glucosidase, endo-β-1,6-glucanase and 6-phospho-β-glucosidase, have been observed to be involved in the hydrolysis of hemicellulose and cellulose. In addition, the genome of C. manihotivorum CT4^T also harbors a gene (CT4_00924), which encodes a putative pectate lyase that can be hydrolyzed the internal α-1,4 linked D-galacturonic acid within the pectin.

As illustrated in Fig. 1, phylogenetic analyses showed that strain CT4^T forms a cluster with C. polyendosporum PS-1^T. The latter cluster forms a sibling group with the C. amylolyticum SW408^T, C. putrefaciens DSM 1291^T and C. algidicarnis NCFB 2931^T branch. However, only the strain SW408^T was released as a genome announcement. Thus, the C. amylolyticum SW408^T was chosen for comparative genomic exploration. Analysis of genes encoding carbohydrate-active enzymes in the genomes of strains CT4^T and SW408^T revealed differences in the distribution of genes (Table 6). C. manihotivorum CT4^T had a higher number of genes encoding amylolytic enzymes than that of C. amylolyticum SW408^T. Moreover, strain CT4^T contained more genes encoding hemicellulolytic-, cellulolytic- and pectinolytic-enzymes than C. amylolyticum SW408^T, except the debranching enzyme, α-galactosidase.

Description of Clostridium manihotivorum sp. nov.

Clostridium manihotivorum sp. nov. (ma.ni.ho.ti.vo’rum. N.L. n. manihot, a botanical genus name (cassava); L. v. voro, to eat, devour; N.L. neut. adj. manihotivorum, devouring cassava).

Cells are Gram-positive, anaerobic, rods, single-endospore forming, non-motile and non-flagellate. Cells are 4.0–8.0 µm long and 0.5–1.5 µm wide. Colonies are 0.5–1.0 mm in diameter after incubation on BM7 agar supplemented with cassava pulp (1%, w/v) at 37 °C for 5 days. Cell growth is observed at 25−45 °C (optimum, 37 °C) and at pH 5.5–7.5 (optimum, 7.0). The strain curdles milk, but is negative for catalase, H₂S, and indole. Utilizes D-glucose, D-xylose, D-galactose, D-fructose, D-mannose, D-arabinose, D-rhamnose, D-trehalose, D-raffinose, sucrose, lactose, maltose, mannitol, cellobiose, soluble starch, xylan, cellulose and Avicel^®. The end products of glucose fermentation are acetate, butyrate, ethanol, propionate, CO₂ and H₂. The diagnostic amino acid in their cell wall is LL-diaminopimelic acid (LL-DAP). The major fatty acids are C _16:0, C _14:0, anteiso-C _15:0, summed feature 1 (C _13:0-3OH and/or C _15:1isoH), C _19:0cyclo ω8c and C _17:02-OH. The major polar lipids present are phosphatidylethanolamine (PE) and phosphatidylglycerol (PG). The genome size of the type strain is around 6.3 Mb and the genomic DNA G + C content 32 mol%. The type strain, CT4^T (=TBRC 11758^T = NBRC 114534^T), was isolated from soil collected from a cassava pulp landfill located at Chonburi province, Thailand.