Comparative transcriptome analysis and identification of candidate adaptive evolution genes of Miscanthus lutarioriparius and Miscanthus sacchariflorus

Abstract

Miscanthus species are perennial C4 grasses that are considered promising energy crops because of their high biomass yields, excellent adaptability and low management costs. Miscanthus lutarioriparius and Miscanthus sacchariflorus are closely related subspecies that are distributed in different habitats. However, there are only a few reports on the mechanisms by which Miscanthus adapts to different environments. Here, comparative transcriptomic and morphological analyses were used to study the evolutionary adaptation of M. lutarioriparius and M. sacchariflorus to different habitats. In total, among 7586 identified orthologs, 2060 orthologs involved in phenylpropanoid biosynthesis and plant hormones were differentially expressed between the two species. Through an analysis of the Ka/Ks ratios of the orthologs, we estimated that the divergence time between the two species was approximately 4.37 Mya. In addition, 37 candidate positively selected orthologs (PSGs) that played important roles in the adaptation of these species to different habitats were identified. Then, the expression levels of 20 PSGs in response to flooding and drought stress were analyzed, and the analysis revealed significant changes in their expression levels. These results facilitate our understanding of the evolutionary adaptation to habitats and the speciation of M. lutarioriparius and M. sacchariflorus. We hypothesise that lignin synthesis genes are the main cause of the morphological differences between the two species. In summary, the plant nonspecific phospholipase C gene family and the receptor-like protein kinase gene family played important roles in the evolution of these two species.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12298-021-01030-1.

Introduction

Miscanthus species, perennial C4 grasses in the Poaceae family, have been deemed promising energy crops due to their high biomass yields, excellent adaptability, and low management requirements (Clifton-Brown et al. 2017; Heaton et al. 2008). More than 20 wild species of Miscanthus are widespread in East Asia, Southeast Asia, and the Pacific Islands (Lee and Kuan 2015; Li et al. 2016; Sun et al. 2010). In China, there are at least 14 wild species of Miscanthus, and they are mainly used in the paper industry and for soil and water conservation, landscape gardening, and wetland ecosystem regulation (Li et al. 2019, 2016; Sun et al. 2010). In recent decades, studies have reported that the best biomass crops are derived from hybrids. For example, Miscanthus × giganteus, a natural triploid hybrid of mountain grass, has been tested for yield and production efficiency through large-scale planting in Europe and the United States (Clifton-Brown et al. 2017; Dohleman et al. 2009; Heaton et al. 2008; Lee and Kuan 2015). Therefore, understanding the biological origins and genetic relationships among different species of Miscanthus will be helpful in breeding high-yielding energy crops.

M. lutarioriparius and M. sacchariflorus are closely related subspecies belonging to the Triarrhena Honda section of Miscanthus. The morphology of the two species is quite similar, with only a few differences, such as plant height, stem diameter and branch number (Sun et al. 2010). In wild habitats, M. lutarioriparius is tall, with many stem nodes and branches in the middle or upper part of the stem. In contrast, M. sacchariflorus is small, with few or no branches on the stem (Sun et al. 2010). M. lutarioriparius is an endemic species in China that is widespread in river, lake, and shoal areas of the middle and lower reaches of the Yangtze River. It shows strong salt and flood tolerance (Li et al. 2019; Song et al. 2017). However, M. sacchariflorus is distributed in arid hillside areas in northern China and has strong cold and drought resistance (Gao et al. 2014; Li et al. 2019). This geographical distribution pattern suggests that the two species probably respond to drought and flooding stress in different ways at the genomic level (Stroud and Losos 2016). The differences in temperature and water supply in their environments may have promoted the formation and differentiation of the Triarrhena Honda subspecies. Thus, it is important to elucidate the molecular basis of plant adaptation to different environmental conditions. However, the current research has mainly focused on the genetic diversity and geographic differentiation of M. lutarioriparius and M. sacchariflorus (Xu et al. 2015, 2016). There are few reports on the mechanisms of Miscanthus adaptation to different environments.

Recently, genome and transcriptome sequencing have emerged as effective and rapid methods for identifying genes that have differed in expression during plant adaptive evolution (Nabholz et al. 2014; Zhao et al. 2019). RNA sequencing is beneficial for studying plant population divergence, demographic history, and local adaptation. For example, Nabholz et al. (2014) compared the transcripts of 9 Oryza glaberrima, 10 wild Oryza barthii and one Oryza meridionalis individual using RNA-seq technology. The results revealed severe bottlenecks and domestication costs in African rice. Through a comparative analysis of the transcriptomes of 90 materials from 5 groups of M. lutarioriparius, 59 candidate genes associated with salt tolerance were identified (Song et al. 2017). With RNA-seq from 80 individuals in 14 Miscanthus lutarioriparius populations, which were transplanted into a harsh environment from a native habitat, researchers found that the environment and genetic diversity were the main factors determining gene expression variation in populations (Xu et al. 2016). Recently, evolution patterns were revealed in the Salix phylogeny through comparative genomics and transcriptomics analysis (Zhao et al. 2019). Moreover, analyzing the coding sequences of orthologs in the transcriptome of related species explained the variation in the genome and the evolution of species and was also helpful in discovering rapidly evolving genes. For example, 15 rapidly evolving genes that are likely undergoing positive selection were identified through transcriptome analysis of Ammopiptanthus nanus (Gao et al. 2017). The mechanism of adaptive evolution in two high-elevation alpine herbal species endemic to China was revealed by comparative transcriptome analysis (Jia et al. 2017). The molecular basis for the freshwater adaptability of prawns was revealed by comparing the transcriptomes of three Macrobrachium species (Rahi et al. 2019).

At present, comparative transcriptome analysis between M. lutarioriparius and M. sacchariflorus has not been conducted, and the adaptive differentiation mechanisms of the two species in natural environments are unclear. In this study, we analyzed the evolutionary adaptation of the two species to their different habitats and their speciation by comparing the transcriptome and morphology of M. lutarioriparius and M. sacchariflorus. Differentially expressed candidate orthologous genes related to the morphological differences in the two species were selected; then, the candidate positively selected orthologous genes were screened by Ka/Ks analysis, and the divergence time between them was estimated. These results will facilitate the understanding of the evolutionary adaptation of these two species to their different habitats and speciation. In addition, the expression levels of 20 PSGs were analyzed in response to flooding and drought stress in M. sacchariflorus and M. lutarioriparius, which laid a foundation for screening related flooding and drought stress varieties

Materials and methods

Plant materials and treatments

M. sacchariflorus were collected from an arid hillside in Shanxi, China. M. lutarioriparius were collected from a wetland lake in Hunan, China. We collected the two species in the wild and used their rhizomes to cultivate the plants in pots of the same size. The plants were grown for 2 months in a greenhouse (with a 16 h day and 8 h night photoperiod at a temperature of 23 ± 3 °C). The leaves of similarly sized plants of the two species were selected and stored in a − 80 °C refrigerator for RNA-Seq. In addition, to analyze possible relationships between the candidate genes and abiotic stress, an average of 20 plants of each species were selected and divided into 2 groups. The two groups were separately subjected to submersion and drought treatments (treatment time: 3 and 6 days, respectively). The leaves in each treatment group were stored in − 80 °C refrigerator for qRT-PCR. Three biological replicates were performed for each sample.

Morphological observation and statistics

M. sacchariflorus and M. lutarioriparius were planted in the Miscanthus Resource Garden of Wuhan University. Their morphological characteristics were observed and recorded at two and three years after planting, including plant height, weight per plant, stem diameter, number of stem nodes, number of branches and branch locations. At least 18 plants were observed at each time point.

RNA sequencing and de novo assembly

Total RNA was extracted from each sample using a plant RNA extraction kit (HiPure Plant RNA Kits, Magen, China). The specific steps were performed according to the kit instructions. The integrity, concentration and purity of the extracted RNA were then detected by agarose gel electrophoresis and Nanodrop 2000 spectrophotometry, respectively. After that, a cDNA library was generated using the purified mRNA, which was subsequently sequenced on the Illumina Hiseq platform (HiSeq 4000 SBS Kit (300 cycles)).

To obtain clean reads, SeqPrep (https://github.com/jstjohn/SeqPrep) and Sickle software (https://github.com/najoshi/sickle) were used to filter the raw data (Grabherr et al. 2011; Wang et al. 2019). After that, all clean reads were de novo assembled into contigs and singletons. In this study, clean reads from two Miscanthus species were assembled using Trinity software (https://github.com/trinityrnaseq/trinityrnaseq) to generate unigenes (Altenhoff and Dessimoz 2009).

Ortholog identification and annotation

The homologous sequences were divided into two types: orthologous and paralogous sequences. In this study, we used BLASTx and ESTScan software to predict the coding sequences (CDSs) (Iseli et al. 1999; Altenhoff and Dessimoz 2009; Gao et al. 2016). We used OrthoMCL software (p value Cutoff: 1e-05, Inflation: 2.5) to cluster homologous genes, which resulted in a subset of genes containing considerable numbers of paralogous genes (Li et al. 2003). However, it was necessary to remove paralogous genes to analyze single genes. Therefore, each group of homologous genes was aligned using MUSCLE software. After that, we built phylogenetic trees for these homologous gene clusters using RAxML software (version: 8.0.20) and corrected the tree with Agalma software to remove paralogous genes (Rahi et al. 2019).

For ortholog and protein annotation, the sequences and predicted protein-coding sequences were compared against the Nr (nonredundant) database, Pfam database, KEGG (Kyoto Encyclopedia of Genes and Genomes pathway) database, STRING database and Swiss-Prot database using BlastX (E value < 1e-5) (Tao et al. 2017; Wang et al. 2017a, b, c).

Identification of differentially expressed orthologs

To test whether the variation in biological duplication between samples agreed with the expectations of the experimental design, sample correlation was performed. Hierarchical clustering was performed to assess the transcriptome similarity among samples and to evaluate sampling differences between biological replicates. In addition, we used RSEM software (http://deweylab.biostat.wisc.edu/rsem/) to obtain the expression level of each orthologous gene (Li and Dewey 2011). The FPKM (transcripts per million reads) value of each sample was calculated as the expression of each homologous gene. We used edgeR software (http://www.bioconductor.org/packages/2.12/bioc/html/edgeR.html) to analyze the differentially expressed homologous genes between the two species (Robinson et al. 2010). The filter parameters were as follows: FDR < 0.05 and |log2FC|≥ 1. In addition, GO and KEGG analyses were performed on the differentially expressed orthologs using the software Goatools (https://github.com/tanghaibao/Goatools) and KOBAS (http://kobas.cbi.pku.edu.cn/kobas3) ((-m f (statistical method, Fisher exacttest), (-n BH (false discovery yrate (FDR) correction method)) (Lu et al. 2012).

Estimation of substitution rate and divergence time

In evolutionary analysis, it is meaningful to determine the synonymous mutation frequency (Ks), nonsynonymous mutation frequency (Ka) and ratio of the nonsynonymous mutation rate to the synonymous mutation rate (Ka/Ks). In genetics, Ka/Ks > 1 is considered to indicate a positive selection effect; Ka/Ks = 1 indicates neutral selection; and Ka/Ks < 1 indicates purifying selection (Elmer et al. 2010). In this study, KaKs_calculator software was used to calculate the mutation rate between the two species using the YN method (Zhang et al. 2006; Jia et al. 2017). Based on the formula T = K/2r, the divergence time between M. lutarioriparius and M. sacchariflorus was estimated. “r” is the mean rate of synonymous substitution and is 1.5E-8 substitutions/synonymous site/year for all dicots. “K” is the kimura value, which was determined with the distmat program in Emboss software.

Analysis of selective evolutionary pressure (positively selected genes)

To ensure the accurate screening of divergent genes (positively selected genes, PSGs) and conserved genes, we removed unrealistic Ka/Ks ratio data. Therefore, a more precise screening threshold was adopted: Ka/Ks < 0.01 and p < 0.05 were considered to represent qualified conserved orthologs, and 1 < Ka/Ks < 10 and p < 0.05 were considered the qualifications for positively selected genes (Jia et al. 2017). In addition, GO and KEGG enrichment analyses were performed for the PSGs. The results clarify the functions of these genes at the pathway level.

Expression analysis for the candidate genes of M. sacchariflorus and M. lutarioriparius under abiotic stress

To verify the differential expression of candidate divergent genes of M. sacchariflorus and M. lutarioriparius under flooding and drought stress, we performed expression level verification of 20 divergent genes found in all samples through qRT-PCR. Meanwhile, 10 orthologous genes were used to verify the transcriptome data. All primers were designed using Primer 6.0 software (Table S1). The eIF4a gene was selected as the internal reference gene. qRT-PCR was performed using the StepOne plusTM Real-Time PCR System. The procedure was as follows: 95 °C for 10 min, 40 cycles at 95 °C for 10 s, 60 °C for 20 s and 72 °C for 20 s. Three biological replicates were performed for each experiment. The results were analyzed using the 2- ΔΔCT method.

Results

Morphological observation

M. lutarioriparius and M. sacchariflorus are very similar in morphology, and Triarrhena lutarioriparia is similar to an enlarged version of M. sacchariflorus (Fig S1). For example, the plant height, yield, stem node number and diameter of M. lutarioriparius were significantly larger than those of M. sacchariflorus. In particular, the single plant weight of M. lutarioriparius was 5.9 times that of M. sacchariflorus (Table 1). In addition, M. lutarioriparius generally has more branches, and they are located on the upper part of the stem, while M. sacchariflorus has fewer branches, and they are generally located on the lower part of the stem (Table 1).

Table 1.

Morphological observation of M. lutarioriparius and M. sacchariflorus

Type	M. lutarioriparius	M. sacchariflorus
Single-plant weight(g)	109.87 ± 9.68	18.67 ± 3.34
Plant height(m)	4.56 ± 0.31	1.42 ± 0.1
Stem diameter(mm)	13.24 ± 1.07	3.98 ± 0.36
Internodes number	28.35 ± 2.14	12.24 ± 1.26
Branch number	4.6 ± 1.36	0.9 ± 0.07
The branch part	The upper stem	The lower stem

The results represent the mean standard error (mean ± SE; n = 18)

De novo assembly and annotation of unigenes

A total of 15.11 and 14.59 GB raw reads were obtained for M. sacchariflorus and M. lutarioriparius, respectively (Table S2). After removing the low-quality reads with SeqPrep and Sickle software, more than 14.96 and 14.25 GB clean reads remained, and all the reads were assembled using Trinity software. For M. sacchariflorus, we obtained 58,897 unigenes, with a mean length of 1400.59 bp and an N50 value of 1862 bp. For M. lutarioriparius, 83,021 unigenes were obtained, with a mean length of 1525.49 bp and an N50 value of 2156 bp (Table 2).

Table 2.

Summary statistics for M. sacchariflorus and M. lutarioriparius transcriptomes

Type	M. sacchariflorus	M. lutarioriparius
Total transcripts num	199,109	207,950
Total unigenes num	58,897	83,021
Total sequence base	278,870,431	317,226,591
Largest(bp)	15,671	15,493
Smallest(bp)	201	201
Average length(bp)	1400.59	1525.49
N50	1862	2156
E90N50	1782	2204
GC percent	47.93	47.67
Mean mapped reads	682.9800213	5471.42928
TransRate score	0.17757	0.19963
BUSCO score	82.7% (61.9%)	88.0% (57.0%)

Identification of orthologs in M. sacchariflorus and M. lutarioriparius

A total of 60,634 homologous gene clusters were obtained by using OrthoMCL software. To obtain the orthologous genes for each species, we built phylogenetic trees for these homologous gene clusters using RAxML software and corrected the trees with Agalma software to remove the paralogous genes. After that, a total of 7586 orthologs (OGs) were obtained from M. sacchariflorus and M. lutarioriparius (Table S3). In addition, for gene annotation, these orthologs were aligned with the NR database, STRING database, SwissProt database, KEGG database and KOG database using BlastX (E value < 1e-5). A total of 6980 orthologs were annotated in these databases, with 6970 OGs in the NR database, 4107 OGs in the SwissProt database, 4214 OGs in the STRING database, 2417 OGs in the KOG database, 2368 OGs in the KEGG database and 4277 OGs in the Pfam database. Furthermore, 604 OGs did not show significant match sequences in these databases and therefore likely represent unknown genes (Fig. 1a, Table S4).

Fig. 1.

Analysis of orthologs in M. lutarioriparius and M. sacchariflorus. a Statistics for the results of ortholog annotation in M. sacchariflorus and M. lutarioriparius. b Scatter map of the differential orthologs. c Gene ontology (GO) annotation of upregulated and downregulated DEGs (the right y-axis represents the number of genes annotated to a certain GO term, the left y-axis represents the proportion of the total number of annotated GO genes annotated to a certain GO term, and the x-axis represents the detailed classification of each GO term. The red bars represent upregulated genes, and the blue bars represent downregulated genes). d Gene ontology (GO) functional enrichment analysis of DEGs between the two species. e KEGG pathway enrichment analysis of DEGs between the two species

Differentially expressed orthologs between M. sacchariflorus and M. lutarioriparius

Based on correlation analysis, the hierarchical clustering revealed that the six samples clustered into two discrete corresponding groups, which were used for subsequent experiments (Table S5). A total of 2060 differentially expressed genes (DEGs) were identified in the two species, of which 648 DEGs were upregulated and 1412 DEGs were downregulated (Fig. 1b, Table S6). To investigate the biological pathways of these DEGs, all of the sequences were subjected to GO and KEGG analyses. The GO analysis revealed 14, 22 and 16 categories assigned to molecular function, biological processes, and cellular components, respectively. Moreover, the GO terms of upregulated and downregulated genes were different (Fig. 1c, Table S7). For example, there were many more upregulated genes than downregulated genes in categories such as biological adhesion, extracellular region part, and translation regulator activity. In contrast, the number of downregulated genes associated with the rhythmic process, extracellular matrix, enzyme regulator activity and nutrient reservoir activity categories was much greater than the corresponding numbers of upregulated genes. In addition, in the GO enrichment analysis, these DEGs were enriched in 105 GO terms (p < 0.05), including heme binding, protein kinase activity, protein phosphorylation, cell wall organization or biogenesis, and response to stress (Fig. 1d, Table S8). On the other hand, 1182 DEGs were annotated into 226 KEGG pathways (Table S9). Among them, metabolic pathways contained the highest number of DEGs, followed by biosynthesis of secondary metabolites, plant hormone signal transduction, phenylpropanoid biosynthesis, and others. However, only one metabolic pathway, phenylpropanoid biosynthesis, was obtained through the enrichment analysis (Fig. 1e), and the expression levels of most genes were higher in M. lutarioriparius than in M. sacchariflorus. To further verify the expression levels of these genes and the reliability of the RNA-seq data, 10 DEGs were randomly selected for further investigation by qRT-PCR. The qRT-PCR results for these DEGs were highly consistent with the RNA-seq data (Table 3).

Table 3.

Representative orthologs expression validation using quantitative real-time PCR (qRT-PCR

ID	Description	M. lutarioriparius		M. sacchariflorus		Regulate
		RNA-Seq	qRT-PCR	RNA-Seq	qRT-PCR
ORTHOMCL37464	Os9bglu30	23.16	16.67 ± 0.71	3	6.34 ± 0.14	down
ORTHOMCL39292	OsCAD1	67.48	40.34 ± 1.24	30.63	16.58	down
ORTHOMCL41243	Peroxidase 51; Atperox P51	62.02	59.67 ± 2.09	12.17	21.46 ± 3.16	down
ORTHOMCL41245	Peroxidase 3; ATPRC	56.47	62.87 ± 3.18	5.47	3.57 ± 0.06	down
ORTHOMCL42180	Peroxidase 1; pmPOX1	5.52	8.62 ± 0.61	0.88	4.21 ± 0.33	down
ORTHOMCL42789	Atperox P65	219.75	153.41 ± 12.46	34.97	28.94 ± 2.84	down
ORTHOMCL43274	Shikimate O-hydroxycinnamoyltransferase	8.8	10.34 ± 1.09	2.25	3.56 ± 0.61	down
ORTHOMCL46059	4-coumarate-CoA ligase-like 1	45.23	74.95 ± 3.98	1.74	8.86 ± 0.37	down
ORTHOMCL47034	Peroxidase 70; pmPOX2b	17.11	10.64 ± 1.61	2.89	7.91 ± 0.87	down
ORTHOMCL52007	Cationic peroxidase 1; PNPC1	1.05	5.96 ± 0.94	37.54	29.68 ± 2.10	up

Adaptive evolution analysis of homologous genes

For the Ka/Ks analysis of all the orthologs in M. sacchariflorus and M. lutarioriparius, Ka, Ks and Ka/Ks were calculated for a total of 5905 orthologs (77.84% of all orthologs). The average Ka and Ks values were 0.0133 and 0.0399, respectively (Table S10). More than half of the orthologs (4058) had Ka/Ks values < 1, indicating that most of the orthologs were subjected to purification selection. In addition, 977 orthologs with Ka/Ks values < 0.1 (p < 0.05) were identified; these were considered conserved orthologs. Moreover, 35 orthologs with 1 < Ka/Ks values < 10 (p < 0.05) were identified (Table 4), and these divergent genes were considered candidate positively selected genes (PSGs). Based on the previous descriptive method, we determined that the age of speciation events between the two species was approximately 4.37 Mya.

Table 4.

The candidate positively selected orthologous genes between the two species

ID	Ka/Ks	p Value (Fisher)	Dscription
ORTHOMCL40596	1.30	0.036106	Hypothetical protein SORBIDRAFT_01g036580
ORTHOMCL40603	1.43	0.013619	Probable WRKY transcription factor 41
ORTHOMCL55439	1.43	0.046512	Hypothetical protein SETIT_033026mg, partial
ORTHOMCL54959	1.45	0.015968	Transposon Tf2-7 polyprotein
ORTHOMCL50297	1.45	0.000016	Laccase-19; Benzenediol:oxygen oxidoreductase 19
ORTHOMCL47238	1.51	0.000023	WD repeat and FYVE domain-containing protein 3
ORTHOMCL54525	1.53	0.001506	Receptor-like protein kinase HAIKU2
ORTHOMCL43001	1.64	0.022681	Deacetoxyvindoline 4-hydroxylase
ORTHOMCL50904	1.65	0.002414	Retrovirus-related Pol polyprotein from transposon TNT 1-94
ORTHOMCL52636	1.96	0.006468	Retrovirus-related Pol polyprotein from transposon TNT 1-94
ORTHOMCL38294	2.10	0.000029	Transposon Ty3-G Gag-Pol polyprotein
ORTHOMCL49142	2.11	0.027765	Uncharacterized protein At3g27210
ORTHOMCL46269	2.15	0.043263	FRIGIDA-like protein 4a
ORTHOMCL43916	2.33	0.000176	–
ORTHOMCL52385	2.51	0.044066	AF466203_12 putative gag-pol precursor -orf2
ORTHOMCL42269	2.71	0.006417	Probable Pectinesterase/pectinesterase inhibitor 51
ORTHOMCL47047	2.85	0.046361	FT-interacting protein 1
ORTHOMCL38107	3.03	0.008666	Amino acid permease 2,Amino acid transporter AAP2;
ORTHOMCL38650	3.22	0.001363	Cytochrome P450 81D11;
ORTHOMCL54486	3.33	0.000476	Receptor-like protein kinase HSL1
ORTHOMCL54555	3.64	0.024206	AAA-ATPase ASD, mitochondrial
ORTHOMCL46317	4.13	0.027900	Non-specific phospholipase C6
ORTHOMCL39740	4.21	0.024707	Magnesium protoporphyrin IX methyltransferase
ORTHOMCL52706	4.45	0.029265	hypothetical protein SORBI_004G006500
ORTHOMCL49859	4.53	0.006600	Divinyl chlorophyllide a 8-vinyl-reductase
ORTHOMCL40197	4.71	0.001527	Thiosulfate/3-mercaptopyruvate sulfurtransferase 1
ORTHOMCL41908	4.73	0.044648	Fasciclin-like arabinogalactan protein 4
ORTHOMCL42889	4.82	0.044104	PI-PLC X domain-containing protein At5g67130
ORTHOMCL52600	5.22	0.035319	Bidirectional sugar transporter SWEET16
ORTHOMCL52149	5.98	0.040751	Histone H2B.2
ORTHOMCL37463	6.00	0.039295	Sphingolipid delt
ORTHOMCL46316	6.38	0.001365	Non-specific phospholipase C2
ORTHOMCL40273	6.56	0.014779	ATP-dependent 6-phosphofructokinase 2
ORTHOMCL40732	6.73	0.033102	Probable auxin efflux carrier component 1c
ORTHOMCL42802	9.25	0.000089	Probable glucuronosyltransferase Os04g0650300
ORTHOMCL45303	9.17	0.001154	Aspartyl protease family protein At5g10770
ORTHOMCL48990	9.23	0.002394	GATA transcription factor 8

GO and KEGG enrichment analyses were performed for the PSGs separately (Fig. 2a–b). In the GO analysis, 35 PSGs were mainly enriched in single-organism process, oxidoreductase activity, chlorophyll metabolic process, nitrate import, DNA integration, and hydrolase activity. In the KEGG pathway analysis, they were enriched in ether lipid metabolism, porphyrin and chlorophyll metabolism, inositol phosphate metabolism, glycerophospholipid metabolism, and sulfur metabolism.

Fig. 2.

a Gene ontology enrichment analysis of the candidate positively selected genes (PSGs). b KEGG pathway enrichment analysis of the candidate positively selected genes (PSGs)

Expression of candidate genes under flooding and drought stress

According to relevant studies and the gene expression differences in the PSGs between the two species, 20 genes were selected for analysis of their responses to flooding and drought stress to explore the possible roles of PSGs in abiotic stress responses (Yi et al. 2018). The differential expression levels of the PSG transcripts were observed in the leaves of the two species through qRT-PCR analysis (Fig. 3, Fig S2). The basal expression levels of 12 genes in M. sacchariflorus were higher than those in M. lutarioriparius, whereas 6 genes had higher expression levels in M. lutarioriparius. In addition, two genes had similar basal expression levels in the two species. Moreover, during the abiotic stress treatments, the expression levels of these genes changed significantly in the two species. For example, the gene expression of thiosulfate/3-mercaptopyruvate sulfurtransferase 1, WRKY transcription factor 41, PI-PLC X domain-containing proteins (such as At5g67130) and histone H2B.2 was significantly upregulated. Putative disease resistance RPP13, nonspecific phospholipase C2 and divinyl chlorophyllide (an 8-vinyl-reductase) gene expression levels were significantly downregulated. The amino acid transporter AAP2, aspartyl protease family protein, SWEET16 and HSL1 were upregulated under drought stress and downregulated under flooding stress in M. lutarioriparius and M. sacchariflorus. In addition, the cytochrome p45081d11 and FRIGIDA-like protein 4A genes were upregulated only in M. sacchariflorus. Meanwhile, the magnesium protoporphyrin IX, GATA transcription factor 8 L, HAIKU2, and AAA-ATPase ASD genes were upregulated in M. lutarioriparius (Fig. 3, Fig. S2). These results suggest that these genes may play important roles in the response to abiotic stress in the two species.

Fig. 3.

qRT-PCR analysis of 20 PSGs under flooding and drought stress. Green indicates downregulated genes, and red indicates upregulated genes

Discussion

The geographical environment is a major contributor to plant evolution (Stroud and Losos 2016). Studying the evolution of related gene families is helpful in understanding the adaptation strategies of plants to different habitats (Chen et al., 2019). M. sacchariflorus is distributed mainly in the cold and arid regions of the north, while M. lutarioriparius is distributed along the lakes, shoals and rivers of the middle and lower reaches of the Yangtze River (Sun et al. 2010). This geographical distribution pattern indicates that these two species may respond to different ecological traits at the genome level. In this study, we found that there were differences in plant height, branch number and yield between M. sacchariflorus and M. lutarioriparius. The results were similar to those of previous studies (Sun et al. 2010). However, the genetic basis of the adaptive mechanisms in these two species is still unknown. Here, we analyzed the transcriptome of these two species and made a first attempt to unravel their adaptation mechanisms.

DEGs associated with these two species

Plants respond to different external stimuli by up- or downregulating the amount of mRNA expression (Sharma et al. 2018). These external stimuli include temperature, moisture, location, and altitude. In general, divergence in transcriptional regulation and gene expression is considered a main driver of phenotypic variation across species (Jia et al. 2017; Rahi et al. 2019; Tao et al. 2017; Zhao et al. 2019). In the present study, a total of 2060 (648 upregulated, 1412 downregulated) differentially expressed homologous genes (DEGs) were identified in the two species. These DEGs provided not only an important basis for understanding the mechanism of the adaptation of these two species to the environment but also a molecular basis for studying their diversity. In addition, these DEGs may also be the main reason for the differences in the morphology of the two plants.

Through GO analysis, we found that the upregulated and downregulated DEGs had different GO annotations. In different living environments, DEGs with unique GO terms may be involved in the regulation of plant-specific morphological development or adaptation to the environment. Moreover, in this study, these genes were mainly enriched in GO terms such as heme binding, protein kinase activity, protein phosphorylation, cell wall organization or biogenesis, and response to stress. These results are consistent with those of previous studies. For example, studies have demonstrated that MEKK1 (MAPKKK) mRNA accumulates to cope with high salinity stress in Arabidopsis (Sinha et al. 2011). The plant protein kinase superfamily is involved in plant signal transduction and abiotic stress (Sinha et al. 2011). In addition, studies have shown that plant kinase families often play roles in plant-specific processes and that some may contribute to adaptive evolution (Lehti-Shiu and Shiu 2012). Protein phosphorylation is a ubiquitous mechanism of the temporal and spatial regulation of proteins that is involved in almost every cellular process. Recently, a deepening understanding of protein phosphorylation has sparked greater interest in the role of this regulatory mechanism in evolution and organismal phenotype development (Levy et al. 2012).

It is generally recognized that cell walls played a crucial role in the process of plants colonizing the land and their subsequent radiation and diversification (Sorensen et al. 2010). Lignin is particularly important in the formation of cell walls (Weng and Chapple 2010). For example, the presence of lignin greatly enhances the mechanical strength of plant cells. Lignin makes plant cells less permeable, which is beneficial to the cells of vascular plants that transfer water over long distances. Moreover, phenylpropane analogs have maximum absorption peaks in the UV-B range (280–320 nm), which helps plants block the sun’s ultraviolet rays (Weng and Chapple 2010). In addition, lignin is also involved in various biological and abiotic stress responses of plants. Here, 24 differentially expressed genes were enriched in the phenylpropanoid biosynthesis pathway. The expression levels of most genes (19) in M. lutarioriparius were significantly higher than those in M. sacchariflorus, which seems to explain the substantial difference in morphology between the two species. In addition, Divinyl chlorophyllide, an 8-vinyl-reductase gene, had higher expression levels in M. lutarioriparius than in M. sacchariflorus during flooding and drought stress, and its expression level rose at the beginning and then decreased with the extension of the flooding time. This result also provides a clear explanation for why photosynthetic activity in M. lutarioriparius is greater than that in M. sacchariflorus in the wild, which directly leads to M. lutarioriparius having much higher yields than M. sacchariflorus. However, the upregulation of genes does not necessarily lead to better performance and adaptation. The actual impact of this upregulation must be verified through future functional studies.

Adaptive evolution of orthologs

In adaptive evolution analysis, Ka/Ks values have been widely used to identify protein-coding genes undergoing positive or purification selection (Rahi et al. 2019). In the Ka/Ks analysis in this study, we obtained 37 positively selected genes related to adaptive evolution. These PSGs were involved in many biological processes, such as single-organism process, oxidoreductase activity, chlorophyll metabolic process, nitrate import, DNA integration, ether lipid metabolism, inositol phosphate metabolism and sulfur metabolism. These results indicate that these biological processes were under evolutionary pressure during the speciation of the two species. In addition, the values of the Ks distribution orthologs between species are often used to calculate species divergence events (Jia et al. 2017). Based on the formula: T = K/2r, we estimated that the time of the species divergence between M. sacchariflorus and M. lutarioriparius was approximately 4.37 Mya (Pliocene). This was during the C4 plant expansion period, which involved declining atmospheric CO2, increased seasonality and summer monsoon precipitation, and enhanced long-term aridity in East Asia (Shen et al. 2018). Of course, this divergence time is only a rough estimate based on homologous genes, and more evidence would be required to more precisely determine the divergence time.

In general, to adapt to changing habitats and tolerate abiotic and biotic stresses, plants have developed many adaptation strategies at the genetic, physiological, and morphological levels (Sharma et al. 2018). In this study, multiple orthologs of PSGs related to resistance to biotic or abiotic stress were obtained. These genes included FLA4 (ORTHOMCL41908), SWEET16 (ORTHOMCL52600), AP (ORTHOMCL45303), CYP450 81 (ORTHOMCL38650) and Pfk2 (ORTHOMCL40273). This result is similar to those of recent studies showing that stress-resistance genes are positively selected during plant evolution (Cheng et al. 2018; Schuler et al. 2006; Timotijevic et al. 2010; Wang et al. 2018; Xue and Seifert 2015; Yao and Wu 2016). The analysis of expression levels showed that CsSWEET16, AP and Pfk2 were upregulated under drought stress and were downregulated or unchanged under flooding stress in the two species. In addition, the expression levels of FLA4 were upregulated under drought and flood stresses in the two species. The expression levels of CYP450 81 were significantly upregulated during flooding stress in M. sacchariflorus but were significantly downregulated in M. lutarioriparius. These results are consistent with recent reports. In Arabidopsis, the CsSWEET16 gene contributed to modifying cold tolerance by regulating sugar compartmentation across the vacuole (Wang et al. 2018). FeAP9 expression is upregulated in buckwheat leaves under the influence of dark and drought stresses (Timotijevic et al. 2010). Yao et al. found that phosphofructokinase is involved in drought stress at the transcriptional level in mulberry trees (Yao and Wu 2016). The At-FLA4 gene locus plays a nonredundant role in the root growth and salt tolerance of A. thaliana (Xue and Seifert 2015; Turupcu et al. 2018). These results suggest that the PSGs certainly correlate with the wild habitats of the two species—M. sacchariflorus is mainly distributed in the arid areas of the loess hillsides in northern China, and M. lutarioriparius is mainly distributed in areas with abundant water resources in the Yangtze River basin.

Prior studies have documented the effectiveness of the plant nonspecific phospholipase C (NPC) gene family, which plays an important role in plant evolution and resistance to abiotic stress (Krckova et al. 2015; Peters et al. 2014; Pokotylo et al. 2013). For example, the overexpression of NPC5 enhances salt tolerance in Arabidopsis by promoting lateral root development (Peters et al. 2014). In Arabidopsis, an NPC1 knockout T-DNA insertion line exhibited a significant decrease in survival rate under heat stress. However, plants overexpressing NPC1 were more resistant to heat stress than wild-type plants (Krckova et al. 2015). Here, we found that the expression levels of NPC2 and NPC6 were significantly downregulated under flooding and drought stress in the two species. This result suggests that these genes play negative regulatory roles in flooding and drought stress. In addition, from an evolutionary perspective, the history of the NPC family indicates that the common ancestor of land plants contained NPC1, NPC2 and NPC6 subfamilies but that NPC2 and NPC6 gradually disappeared during adaptive evolution (Pokotylo et al. 2013). Thus, we conjectured that NPC2 (ORTHOMCL46316) and NPC6 (ORTHOMCL46317) have played important roles in facilitating the adaptation of M. sacchariflorus and M. lutarioriparius to different habitats. Meanwhile, the auxin efflux carrier component 1c gene, which is involved in the regulation of the emergence of lateral roots of Arabidopsis (Peret et al. 2013), was found in the PSGs. The expression level of this gene was significantly increased during flooding and drought stress in the two species.

Recently, plant transcription factors have been demonstrated to efficiently regulate gene expression to allow plants to cope with various stresses (Ayadi et al. 2016; Chen and Yin 2017; Chen et al. 2019; El-Esawi et al. 2019; Liu et al. 2016; Song et al. 2018). They also play important roles in plant evolution and differentiation (Chen et al. 2019). In this study, one WRKY transcription factor gene (ORTHOMCL40603) and one GAGA transcription factor gene (ORTHOMCL48990) were obtained from 37 PSGs. WRKY and GATA transcription factors are involved not only in the regulation of plant growth and development but also in biotic and abiotic stress responses such as responses to salinity, drought, and pathogen infection (Gupta et al. 2017). For example, the overexpression of TaWRKY1, TaWRKY2, TaWRKY19 and TaWRKY33 increased tolerance to drought and heat stress in Arabidopsis (He et al. 2016; Niu et al. 2012), and the overexpression of GhWRKY25 reduced tolerance to drought stress but enhanced tolerance to salt stress in N. benthamiana (Liu et al. 2016). In addition, the expression of OsGATA gene family members differs under abiotic stress in rice (Gupta et al. 2017). In particular, OsGATA23a exhibits high levels of transcriptional induction under salinity and drought. Here, we found that the expression level of WRKY41 was upregulated under drought stress but downregulated under flooding stress in M. sacchariflorus. However, its expression level was upregulated under both drought and flooding stress in M. lutarioriparius. In addition, the expression level of GATA8 increased during drought stress and decreased during flooding stress in M. sacchariflorus, but it was upregulated during flooding and drought stress in M. lutarioriparius. This result is consistent with those of a recent study in which OsGATA8a and OsGATA8b maintained high transcription levels in response to drought in rice (Gupta et al. 2017).

Receptor-like protein kinases (RLKs) are the largest gene family of transmembrane signaling proteins in plants and play important roles in the regulation of plant growth and development, signaling networks, resistance to various pathogens, and responses to abiotic stresses (Li et al. 2018; Wang et al. 2017a; Xu et al. 2018). For example, the overexpression of NtRLK5 conferred enhanced drought tolerance in transgenic Arabidopsis plants by increasing antioxidant enzyme activities (Xu et al. 2018). PnLRR-RLK2 plays a positive role in the Antarctic moss response to salt and drought stress by regulating the antioxidative system and the ABA-mediated signaling network (Wang et al. 2017a). StLRR-RLK008 and StLRR-RLK04, which are LRR-RLK family genes in potato, are involved in peptide signaling under biotic and abiotic stresses such as salt, drought, and heat stress (Li et al. 2018). In this study, the RLK-HSL1 (ORTHOMCL54486) and RLK-HAIKU2 (ORTHOMCL54525) genes were identified as PSGs; these genes have been shown to play vital roles in regulating embryogenesis and seed maturation (Xiao et al. 2016). During flooding stress, the expression level of the HSL1 gene was downregulated, and the expression of the HAIKU2 gene showed opposite trends in the two species: its expression level first dropped and then rose in M. sacchariflorus, and it first rose and then fell in M. lutarioriparius. During drought stress, the expression levels of the HSL1 and HAIKU2 genes were upregulated in the two species. This result indicates that the two genes have positive regulatory effects on drought responses but may play negative regulatory roles under flooding stress.

Conclusion

In this study, we analyzed the transcriptomes of and identified differentially expressed homologous genes in Miscanthus lutarioriparius and Miscanthus sacchariflorus. In addition, based on the analysis of the Ka/Ks ratios of these homologous genes the age of speciation events between the two species were calculated, futhermore 37 positively selected homologous gene candidates were identified. We hypothesized that genes related to lignin synthesis were the main cause of the morphological differences between the two species. The plant nonspecific phospholipase C gene family, receptor-like protein kinase gene family, and the GATA8 and WRKY41 transcription factors presumably played important roles in the evolution of the two species. Moreover, the expression levels of 20 PSGs in response to flooding and drought stress were analyzed in the two species. These results facilitate the understanding of the evolutionary adaptation of these two species to their different habitats as well as of their speciation and lay a foundation for the screening of genes related to flooding and drought stress.

Supplementary Information

Below is the link to the electronic supplementary material.

Author’s contributions

Jia Wang performed and analyzed the experiments, and wrote the manuscript; Jianyong Zhu and Jiajing Sheng contributed experiments materials and provided assistance to the experiments. Ying Diao and Zhongli Hu offered scientific advice, guided the experiments and revised the manuscript. All authors read and approved the final manuscript revision.

Funding

This work was financially supported by the National Natural Science Foundation of China (No. 31571740), and the Anhui University of Science and Technology launched a research fund to attract talents (13200389).

Availability of data and material

The sequencing raw data from 6 RNA-Seq libraries were deposited on the Sequence Read Archive from NCBI under SRA accession: SRP158951 and SRP190160. Data are available here:

Declarations

Conflict of interest

The authors declare that the submitted work was not performed in the presence of any personal, professional or financial relationships that could be constructed as a conflict of interest.