Changes of cellulose, hemicellulose and lignin content in elephant grass stem at different development stages
The content of cellulose, hemicellulose and lignin in different stem segments of T1, T2, T3 phase were measured. It was found that the content of cellulose and hemicellulose increased first and then decreased. For example, the content of cellulose and hemicellulose in T1-S2 was higher than that in T1-S1, but the content of T1-S3 was lower than that of T1-S2, whereas the content of lignin fell gradually. The similar changes also appeared in different stem nodes at T2 and T3 stages. Meanwhile, by analyzing the content change of the same stem node in different development times, it was also found that with the increase of growth time of elephant grass, the content of cellulose and hemicellulose decreased, while the content of lignin increased (Fig. 1b).
By analyzing the ratio of the primary cell wall to the secondary cell wall, it was found that the secondary cell wall increases as the cell development time increases， while the rate of the cell wall to the primary cell wall keeps rising, which might be the reason of the gradually increases of lignin content to maintain and support the strength of elephant grass stems (Fig. S1).
Characteristics of the cell wall in different developmental stages of the elephant grass stem
The changes of cell wall morphology at different development stages of elephant grass, especially the primary and secondary cell wall were observed (Fig. 2a). The ratio of secondary cell wall (sw) thickness to primary cell wall (pw) thickness in T1-S1 and T1-S2 was 1.18 and 0.85, the ratio of sw/pw thickness in T2-S1 and T2-S2 was 1.67 and 0.92, and the ratio of sw/pw thickness in T3-S1 and T3-S3 was 2.15 and 1.25, respectively (Fig. 2b). The above data showed that the change trend of sw/pw thickness of T1, T2, and T3 was consistent, that is, with the increase of development time, the development speed of secondary cell wall (sw) was faster than that of primary cell wall (pw). The ratio of sw/pw in S1 node of three different development stages was also analyzed. It was found that with the increase of development time, the thickness ratio of sw/pw in T1-S1 T2-S1, T3-S1 increased subsequently. The change of sw/pw thickness ratio in S2 and S3 stages of three different development stages showed the same trends.
To further understand the morphological changes of cell wall during the development of elephant grass, we selected S1 and S5 stem segments with a longer development time span at T3 stage as samples for micro-CT observation. The stem of elephant grass is composed of epidermis, parenchyma cells and vascular bundles. Vascular bundles, which scattered in parenchyma cells and cannot be thickened, are composed of phloem and xylem without cambium. With the development of stem tissue, S1 vascular bundle were arranged regularly and compactly (Fig. S1). The content of cellulose and hemicellulose in vascular bundle decreased gradually, while the content of lignin increased gradually (Fig. 1b) to meet the need of mechanical support during the maturation of elephant grass stems need in the process of stem maturity of elephant grass.
Differential gene expression during the development of elephant grass stems
36 cDNA libraries were constructed from different stages of elephant grass stems (three biological replicates for each tissue). Totally, 1.32 billion raw reads (396.96 Gb) were obtained, 1.29 billion cleaned reads (388.57Gb) were acquired after filtering with 6.56-21.07 Gb in each sample. The error rate was 0.03% (Q20 and Q30 values were more than 93% and 90%, respectively), which met the requirements of gene discovery (Table S1). De novo assembly generated 77,435 cluster sequences from 12 representative samples with the largest sequencing depth. Finally, we got a non-redundant transcript clusters include 230572 unique genes with the average length of 961.35bp and an N50 of 1435bp, an N90 of 423bp (Fig S2). Transcriptome de novo assembly was carried out with short reads assembling program-Trinity . The Pearson correlation coefficient based on the expression value of each library indicated that there was a high correlation between sample replicates (Fig. S3). Cluster analysis among samples showed that the development time of elephant grass was the main factor affecting the clustering. The DEGs in three developmental stages of elephant grass stems were analyzed, a total of 15611, 10235 and 27389 DEGs were identified in T1, T2, and T3, respectively (Fig. 3). The DEGs of different stem segments at three development stages were also analyzed, it was found that 147 genes were expressed in three segments of T1, 54 in four segments of T2, and 91 in five segments of T3 (Fig. 3). The intersection of all DEGs at three different developmental stages was compared to determine the shared core set. It was found that 3852 genes were differentially expressed at three developmental stages (Fig. 3d).
3852 DEGs were then subjected to enrichment analysis of GO functions and KEGG pathways. 4 of the top 10 enriched GO annotation functions were related to cell composition such as apoplast, cell wall, extracellular region, plant-type cell wall, four were related to molecular functions such as peroxidase activity, xyloglucan, and xyloglucosyl transferase activity, heme-binding, xyloglucan−specific endo−beta−1,4−glucanase activity, and two were related to biological processes such as cell wall macromolecule catabolic process, hydrogen peroxide catabolic process (Fig. 4a) (Table S2). Based on KEGG pathway analysis, all DEGs were enriched to 9 pathways (Table S3), of which the two most significant pathways were phenylpropane metabolism (23 DEGs) and starch and sucrose metabolism (23 DEGs) (Fig. 4b).
Genes highly correlated with the synthesis of cellulose, hemicellulose and lignin by WGCNA analysis
Weighted gene co-expression network analysis (WGCNA) was performed on 3852 DEGs at three stages, and the network was divided into three modules. The analysis of module-trait relationship showed that the ‘blue’ module was highly correlated with the synthesis of cellulose (r=0.67, P=6.0×10-6) and hemicellulose (r=0.51, P=0.001), whereas the ‘turquoise’ module was related to lignin synthesis (r=0.68, P=5.0×10–6) (Fig. S4).
The ‘blue’ module was filtered according to Module membership > 0.9, the absolute value of the correlation coefficient between the ‘turquoise’ module and lignin was greater than 0.75, and 20 and 23 genes remained in the ‘blue’ and ‘turquoise’ module respectively. The WGCNA gene significance (GS) (i.e., related to traits) showed that the genes with highest GS in ‘blue’ and ‘turquoise’ modules were Cluster-55067.0 (0.659) and Cluster-17353.3 (0.798), respectively. Six of 20 genes in ‘blue’ module were known to be functional, such as GTL1 transcription factor, O-methyl transferase (OMT), expansin-likeA2, alpha-humulene synthase, probable galactinol sucrose, GhGalT1. Meanwhile, 14 genes with unknown functions were also covered by ‘blue’ module, which needs further study. The GO function annotations of these 20 genes in ‘blue’ module included the genes which were related to extracellular region, C-4 methyl sterol oxidase activity, terpene biosynthesis, etc. The expression levels of these genes were shown in Fig. 5.
Among 23 genes in ‘turquoise’ module, there were eight coding genes such as GTL1, MYB2 transcription factors, threonine-protein kinase ERECTA, probable methionine-tRNA ligase, alcohol holding hydrogenase-like2, zinc finger protein GIS3, protein slr0074, proline-rich receptor-like protein kinase PERK8. The GO function annotations of these 23 genes in ‘turquoise’ module included genes that were related to cell growth regulation, secondary cell wall formation regulation, cell wall composition regulation, etc. 15 genes remained with unknown functions (Table S4).
Lignin and cellulose synthesis pathway during stem development of elephant grass
Cellulose synthase (CesA) is the most important enzyme in the cellulose synthesis pathway. It can directly utilize UDPG produced by starch and sucrose metabolism to synthesize cellulose. In elephant grass, 27 CesA genes were identified. At T1 stage, the expression level of CesA1 - CesA6 was higher in the whole stem, but decreased in T2 and T3 stages, while the expression level of CesA7 - CesA27 in tender stems was much higher than that in mature stems at T1, T2, and T3 stages. This indicated that CesA gene mainly synthesizes cellulose in the tender stem tissue. When the cellulose accumulated to a certain level, its expression level gradually decreased (Fig. 6a).
Lignin synthesis is one of the most important pathways of phenylpropane metabolism. In the lignin metabolism pathway, CAD (21 unigenes), 4CL9 (21 unigenes), C4H (8 unigenes), PAL (16 unigenes), CCR (25 unigenes), F5H (6 unigenes), CCoAOMT (4 unigenes) were identified to be related. Expression analysis found that most members of these gene families showed higher expression levels in the mature stem. In contrast, a few gene members were continuously expressed in the whole tissue (Fig. 6b).
The results of qRT-PCR demonstrated that the expression trends of these genes were consistent with that of RNA-seq data (Fig. S5, Table S6). Overall, as the stem development, the expression level of cellulose synthesis genes and lignin synthesis related genes showed the opposite trend. On the other hand, the changes of stem cellulose, hemicellulose and lignin content, as well as the changes of primary stem wall and secondary cell wall thickness, were positively correlated with the expression of these two types of genes.