The complexity of hydrolysates increased during the cellulose-degrading process
The microbiota FP was obtained from woodland soil and cultivated using filter paper as the solo carbon source. Different stages of filter paper degradation were analyzed and compared. When FP was cultured for 3 days (the early stage, FP.3), a light yellow strain lawn appeared in the central areas of the filter paper. Compared to 3 days of cultivation, the strain lawn deepened in color and expanded after 8 days of culture (the intermediate stage, FP.8). As the fermentation proceeded to 15 days (the final stage, FP.15), some areas of the filter paper appeared pasty or even transparent, suggesting that it was further degraded (Fig. 1). However, as fermentation continued, no further degradation of the filter paper was detected (data not shown). The results indicated that FP was able to grow on filter paper and efficiently utilize cellulose as the sole carbon source. This observation was similar to that reported by Dumova et al. [10], who observed a slight degradation area in filter paper after the microbiota sample was cultured for 7 days.
Previous studies suggest that cellulose is first degraded to cellodextrin, which then continues to be hydrolyzed into short-chain cello-oligosaccharides and cellobiose. Cello-oligosaccharides and cellobiose are eventually broken down into glucose [4]. To further analyze the filter paper hydrolysates at different degradation stages, IC analysis was performed. As shown in Fig. 2, a large number of short-chain saccharide molecules with a degree of polymerization (DP) greater than 15 were detected after the microbiota FP was cultivated in filter paper medium for 3 days. After 8 days of continuous degradation, the short-chain molecules were broken down into cello-oligosaccharides and cellobiose with DP values of 2-6. At the final degradation stage (15 days), the cello-oligosaccharides and cellobiose contents decreased, and the end product, glucose, was observed. All the results suggest that the complexity of the hydrolysates increased during the early-intermediate period of cellulose degradation, which was consistent with a previous report by Haruta et al. [11]. They found that the maximum content of various metabolites occurred when cellulose was degraded by microbiota for 2-5 days and then became stable after 7 days of hydrolysis.
Sequencing data
After quality filtering, ~ 59.43 gigabases (Gb) high-quality data were analyzed in the present study (an average of ~ 6.60 Gb per sample). These clean reads were assembled using SOAPdenovo, which resulted in 82,261 scaftigs with a minimum length of 500 bp, an N50 length of 37.35 kilobases (Kb), and a maximum length of 50.88 Kb. To evaluate the quality of the assembly, we aligned the reads to the scaftigs of each sample and obtained an average alignment rate of 94.78%. This finding indicates that most of the Illumina reads were assembled into scaftigs. In addition, we mixed all unmapped reads to assemble them and obtained a total of 3,473 scaftigs with an average of ~757 bp in length. The average numbers of 24,991 (FP.3), 29,773 (FP.8), and 33,544 (FP.15) open reading frames (ORFs) were predicted by MetaGeneMark and showed an increasing trend as cellulose degradation proceeded, implying that the biodiversity of the FP community might also increase with the degradation process of cellulose. The ORF sequences for all samples were subjected to clustering, and a nonredundant gene catalog was constructed. A total of 44,829 nonredundant ORFs with ~810 bp average length and ~36.31 megabases (Mb) total length were obtained. Among them, a total of 27,604 (~61.58%) genes had complete ORFs. The statistics of sequence data for the different treatments is given in Table 1.
Dynamic shifts in the structure and composition of cellulolytic microbiota FP
Theoretically, complex carbon sources provide a wide range of nutrition for microorganisms. Cellulose degradation produces a variety of metabolites, such as cellodextrin, cello-oligosaccharides, cellobiose, and glucose. Changes in metabolite composition are likely to cause disturbances in the diversity and structure of a microbiota. To assess such disturbances during the cellulose degradation process, Shannon-Wiener and Simpson indices based on species abundance were calculated for FP samples at different degradation stages, and the Wilcox test was employed to identify significant differences in comparisons. As shown in Table 2, both the Shannon-Wiener and Simpson indices revealed that the species diversity of the microbiota FP significantly increased in the intermediate and final stages compared with the early stage (p<0.05). In addition, the increasing rate of species diversity in the early-intermediate period was significantly faster than that in the intermediate-final period (p<0.05). The 2-dimensional nonmetric multidimensional scaling (NMDS) plot showed that the microbiota FP.3 was well separated from FP.8 and FP.15, which were relatively close to each other (Fig. 3a), suggesting that the structure of FP was disturbed in the early-intermediate period but relatively stable in the intermediate-final period, which coincided well with the alpha diversity results. Considering the changing complexity of hydrolysates during the cellulose-degrading process, it could be deduced that the increased species diversity of FP was probably due to the enhanced metabolite complexity.
As shown in Fig. 3b, Bacteroidetes, Firmicutes and Proteobacteria were the main phyla; they accounted for over 96% of the abundance in FP samples from different cellulose degradation stages, which was consistent with previous studies [13, 14]. At the genus level, the genera with a total relative abundance greater than 0.5% included Sporocytophaga (40.82%), Cohnella (34.85%), Achromobacter (4.99%), Herbaspirillum (5.87%), Paenibacillus (2.00%), and Microbacterium (0.54%). At the species level, Sporocytophaga myxococcoide and Cohnella sp. CIP 111063 were the most dominant species, and the relative abundance of the two species reached 40.82% and 34.27%, respectively.
Previous studies have shown that natural cellulolytic strains degrade cellulose synergistically together with other microorganisms that utilize hydrolysates (such as glucose and cellobiose) as carbon and energy sources [2, 10]. Furthermore, there might be a "jungle rule" in micro-ecosystems due to competition for carbon sources in the process of cellulose degradation, resulting in the shift of relative abundance among species [15]. As illustrated in Fig. 4, the relative abundance of microbial taxa from FP.3 and FP.8 exhibited significant differences from each other (p<0.05). Here, microbial taxa in FP could be roughly divided into two types: downregulated and upregulated. The downregulated microorganisms were more likely to be involved in the early stage of cellulose degradation (cellulose to cellodextrin). Notably, almost all downregulated taxa belonged to Bacteroidetes, with Sporocytophaga myxococcoide as the representative species. Bacteroidetes are considered to be highly efficient in carbohydrate metabolism, and a few members in this phylum are reported to have cellulose degrading ability [16]. Previous studies have proven that S. myxococcoides can glide rapidly on solid surfaces of cellulose substrates and conduct cellulose degradation [17]. Logically, other downregulated genera, such as Flavobacterium and Chitinophaga, are believed to have functions similar to those of S. myxococcoides. On the other hand, upregulated taxa might play important roles in catabolizing metabolites such as cello-oligosaccharides, cellobiose and glucose that are produced in the intermediate degradation stage. Most of the upregulated species were derived from Firmicutes and Proteobacteria. Many members of Firmicutes can produce proteases, cellulases, lipases and other extracellular enzymes, which might contribute to cellulose catabolism [18]. Cohnella and Paenibacillus, two dominant genera from Firmicutes, possess the ability to utilize cellulose, hemicellulose, and cellobiose [19]. Cohnella sp. CIP 111063, which was most dominant in the upregulated species, contains a set of genes encoding cellulases. Proteobacteria are rich in the organic layer of forest soil [20]. The two dominant genera from Proteobacteria were Herbaspirillum and Achromobacter. Herbaspirillum is a typical nitrogen-fixing bacterium in soil [21]. Previous studies identified eight gene clusters from this genus that are involved in cellulose biosynthesis and degradation [22]. As members of Achromobacter, Achromobacter xylosoxidans and denitrificans with a low abundance cannot ferment cellobiose and glucose but are positive in oxidase and catalase activity, they probably play a role in removing other gradually accumulated metabolites that are not directly relevant to cellulose degradation [23]. Additionally, these low-abundance bacteria could serve as a species reservoir, together with other low-abundance species, to enhance the environmental adaptability of the microbiota FP.
Core genes in cellulose catabolism
The gene annotation results for microbiota FP showed that 1956 (4.36%) genes belong to carbohydrate-active enzymes (CAZymes), including 1074 glycoside hydrolase (GH) coding genes, 522 carbohydrate-binding module (CBM) coding genes, 399 glycosyltransferase (GT) coding genes, 144 carbohydrate esterase (CE) coding genes, 50 polysaccharide lyase (PL) coding genes and 19 auxiliary active enzyme (AA) coding genes [24]. Their relative abundances with different treatments were compared using Metastats analysis, and a heat map of the top 35 CAZymes and CBM families was plotted (Fig. 5a). The results revealed that all the proteins only exhibited significant abundance shifts (p<0.05) in the early-intermediate period of filter paper degradation, which was consistent with the observation of taxa disturbance. Similarly, these annotated protein families could be classified into upregulated and downregulated types. The upregulated families mainly included GH13, GH51, CE7, AA10, GH55 and CE14, of which only GH51 was related to cellulose hydrolysis [25]. A large proportion of the downregulated families were cellulose degradation-related enzymes and functional modules, including GH3, GH5, GH8-10, GH16, GH26, GH74, CBM2-4, CBM6, CBM9 and CBM32 [24]. As shown in Fig. 5b, 17,543 (39.14%) genes in the nonredundant gene catalog were annotated to 3472 Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologies (KOs). A total of 76 KOs showed significant differences across FP.3, FP.8 and FP.15 (p<0.05), among which four KOs constituted a cellulose degradation pathway: endo-1,4-beta-D-glucanase (Enzyme Commission (EC) 3.2.1.4), beta-D-glucosidase (EC 3.2.1.21), exo-beta-1,4-glucosidase (EC 3.2.1.74) and cellulose 1,4-beta-cellobiosidase (nonreducing end) (EC 3.2.1.91).
However, 7,177 genes were still unannotated, among which a total of 4,031 genes showed high abundance (≥0.01%) and species annotation information. To further explore the unannotated genes (UGs) potentially related to cellulose catabolism, cluster analysis was performed using the Short Time-series Expression Miner (STEM) algorithm. In total, 432 cellulose degradation-related genes were clustered with UGs based on different abundance variation patterns in the early, intermediate and final cellulose degradation stages. Therefore, UGs that have roles similar to those of annotated genes could probably be selected. As shown in Table 3, four typical patterns with significant abundance variation, including pattern 0, pattern 4, pattern 11 and pattern 14, were selected. Annotated genes clustered to pattern 0 and pattern 4, with downregulated trends, usually encode cellulases, including GH3, GH5, GH8, GH9, GH10 and other CAZymes families and are derived from cellulolytic bacteria, such as Sporocytophaga myxococcoides. Therefore, it could be speculated that 363 UGs from pattern 0 and pattern 4 might be related to cellulose and could be named putative cellulolytic gene 1 (PCG1). Most of the genes in pattern 11 and pattern 14, with upregulated trends, were mainly from Cohnella sp. CIP 111063 and Paenibacillus. Nearly all the annotated genes encoded beta-D-glucosidase and exo-beta-1,4-glucosidase, suggesting that the clustered 231 UGs, such as FP.3.1_13197 and FP.3.3_15298, might code for enzymes with similar functions. This type of UGs was named putative cellulolytic gene 2 (PCG2).
Predicted cellulolytic catabolism pathway
Based on the dynamic analysis of the filter paper hydrolysates, the community structure and composition, and the clustered functional genes of microbiota FP, we constructed a complete cellulose catabolism pathway. As shown in Fig. 5c, cellulose was first hydrolyzed into short-chain cellulose, cellodextrin and cellobiose by enzymes including endo-1,4-beta-D-glucanase, cellulose 1,4-beta-cellobiosidase (nonreducing end) and a small amount of cellulose 1,4-beta-cellobiosidase (reducing end) (EC 3.2.1.176). Second, short-chain cellulose and cellodextrin (DP>15) continued to be digested into shorter cello-oligosaccharides and cellobiose. This stage required the synergistic action of a variety of enzymes with different functions. Endo-1,4-beta-D-glucanase and 1,4-beta-cellobiosidases continued to participate in this stage, while beta-D-glucosidase and exo-beta-1,4-glucosidase began to hydrolyze short-chain dextran to continuously produce shorter cello-oligosaccharides and cellobiose [27]. Then, cello-oligosaccharides and cellobiose were completely hydrolyzed to glucose and glucose 6-phosphate by beta-D-glucosidase, exo-beta-1,4-glucosidase and phosphocellobiase (EC 3.2.1.86). In addition, a small number of cellobiose dehydrogenases (EC 1.1.99.18) and cellobiose phosphotransferases (EC 2.7.1.205) were also involved in cellobiose metabolism, which catalyzed the biochemical reaction of cellobiose to cellobionolactone and glucose 6-phosphate, respectively. Moreover, the PCGs were also located in this pathway. A similar cellulose biodegradation pathway at the molecular level was previously described by Juturu and Wu [4]. However, they provided no information about the dynamically synergistic action at the gene and species levels. Furthermore, two typical strains, Sporocytophaga myxococcoides and Cohnella sp. CIP 111063 dynamically participated in the synergistic degradation process, and had different functional tendencies in the early-intermediate and intermediate-final stages.