Comparative transcriptome proling provides insights into different photosynthetic responses in two tall fescue cultivars under low-light stress

Background Tall fescue (Festuca arundinacea Schreb) is a primary cool-season forage and turfgrass, low-light (LL) stress is a primary limiting factor for turfgrass growth. Studies on two tall fescue cultivars, Arid 3 and Airlie, showed that Airlie is more susceptible to LL stress than Arid 3. However, the underlying susceptibility mechanism is obscure. In this study, we investigated the physiological and transcriptional changes of two tall fescue cultivars under LL stress. The two cultivars differed in growth characteristics, chlorophyll contents, photosynthetic gas exchange, and chlorophyll uorescence. A total of 136925 unigenes were obtained from our RNA-sequencing. There were 24944 and 30816 differentially expressed genes (DEGs) between LL stress and the control in Arid 3 and Airlie, respectively. Meanwhile, there were 20599 DEGs detected between Arid 3 and Airlie under the control, and under LL stress the DEGs between two cultivars were 20783. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis suggested that the DEGs are mainly involved in ‘cell prat’ and ‘glutathione metabolism’. These results indicated that LL stress affects the expression of phytochrome-interacting factor 5 (PIF5), constitutively photomorphogenic 1 (COP1), phytochrome A (PhyA), and phytochrome B (PhyB) in tall fescue, which are vital transcription factors. Genes related to chlorophyll metabolism and photosynthesis were also affected by LL stress, and were signicantly differentially expressed between the two cultivars. This study reveals two tall fescue cultivars develop different physiological and transcriptional changes to cope with LL stress, and provides an insight into different photosynthetic responses in two tall fescue cultivars responds to LL stress.


Background
Turfgrass is the primary vegetative ground cover in the landscape. It effectively reduces the environmental problems and improves the quality of our life. However, low-light (LL) stress is a common environmental factor for turf deterioration [1]. It is estimated that as much as 50% of the turfgrass grown in China is subjected to varying degrees of LL stress [2]. LL stress causes numerous morphological and physiological changes to turfgrass. Leaves and leaf cuticles become thinner in shady areas, stems become taller and slender, the root biomass decreases, and the leaf color becomes lighter [3,4]. In addition, LL stress restricts the rate of photosynthesis, resulting in decreased turfgrass growth [5,6]. LL stress also destroys the chloroplast ultrastructure, and reduces the activity of Rubisco [7]. However, plants have developed intricate mechanisms for adapting to biotic and abiotic stresses. The physiological growth of plants changes under LL stress. In order to adapt to LL stress, leaves adjust the shape of epidermal cells and palisade tissue to increase the ability of light to passage through leaf epidermis to the mesophyll [8]. Besides, LL stress increases the ratio of Chl b to Chl a, and plants produce more LHC to obtain su cient light [9]. Interestingly, turfgrass species and cultivars are variedly adapted to shade. Thus, a better understanding of the gene expressing pro les of turfgrass under LL stress is imperative to elucidate the mechanisms of turfgrass resistance to LL stress. So far, the LL stress-response mechanism in tall fescue remains unknown.
Tall fescue (Festuca arundinacea Schreb) is a primary cool-season forage and turfgrass, which has been widely used in temperate zones. As a hexaploidy outcrossing grass, tall fescue has better tolerance to various abiotic stresses. Notably, certain turfgrass species and cultivars have excellent growth and development under LL stress due to their higher light energy utilization or less light needs than intolerant species and cultivars [1]. In our preliminary experiments, two cultivars of tall fescue, Arid 3 and Airlie, exhibited speci c LL stress tolerance. Under LL stress, Arid 3 had an excellent growth and a higher photosynthetic rate than Airlie, suggesting that Arid 3 is better adapted to LL than Airlie [10,11]. However, the molecular mechanism of LL stress response in the two cultivars (Arid 3 and Airlie) remains unknown. This study aimed to elucidate the physiological and molecular mechanisms by which two tall fescue cultivars respond to LL stress. Pro ling LL tolerant cultivars and identifying genes involved in LL stress responses greatly contributes to the study of stress tolerance mechanisms.
RNA-Seq technology, a powerful high-throughput method for investigating gene expression and regulatory networks in various plant species [12,13], has been widely applied in plants lacking a genome sequence.
Several RNA-Seq studies on Kentucky Bluegrass (Poa pratensis) [14], Manila Grass (Zoysia matrella) [15], and Japanese Lawn grass (Zoysia japonica) [16] focused on transcriptional changes in response to abiotic stress and associated pathways. In tall fescue, RNA-Seq was used to reveal resistance responses to cadmium (Cd) induced stress [17], investigate the transcriptome under water stress [18], explore two cultivars in response to plumbum (Pb)-induced stress [19], and analyze for thermotolerance [20]. However, no such analysis has addressed the effect of LL stress on tall fescue growth. So, this study aims to unravel the molecular mechanisms of tall fescue's response to LL stress. An RNA-seq study was performed to reveal the potential genes and networks involved in LL stress responses of two tall fescue cultivars.

Morphology and physiology parameters under LL stress
Measurements of growth, chlorophyll contents, and gas exchange parameters were performed in the two cultivars under LL stress and the control. The tiller number and total dry weight of the two cultivars were signi cantly reduced after LL treament. Notably, obvious difference in tiller number between Arid 3 and Airlie were also observed, as demonstrated by that Arid 3 produced more tillers than Airlie under LL stress (Fig. 1).
Contents of Chl a, Chl b, Chl a + b, and carotenoid decreased remarkably in the two cultivars under LL stress compared to the control. As expected, the content of detected photosynthetic pigments was more abundant in Arid 3 compared to Airlie under LL stress. In particular, the content of Chl b in Airlie was signi cantly decreased (39.1%) in Airlie under LL stress compared to Arid 3 (Fig. 1). LL treatment also reduced net photosynthetic rate (Pn), stomatal conductance (Gs), intercellular CO 2 concentration (Ci), and transpiration rate (Tr) in Arid 3 and Airlie signi cantly compared with respective the control. Under LL stress, the value of Pn, Gs, and Tr values were 30.6%, 40.4%, and 29.5% higher in Arid 3 than them in Airlie, respectively (Fig. 2).
Chlorophyll uorescence characteristics were measured to elucidate changes in the photosynthetic capacity of the two cultivars under LL stress. Values of F v /F m and Φ PSII in Arid 3 decreased under LL stress compared to the control, but the difference between LL stress and the control was not signi cant. However, the chlorophyll uorescence traits (F v /F m, Φ PSII, q P , and ETR) in Airlie remarkably reduced under LL stress compared with the control. Moreover, Arid 3 displayed a better chlorophyll urescence than Airlie under LL stress (Fig. 2). Collectively, this results suggested that Arid 3 possesses higher photosynthetic capacity under LL stress than Airlie, which may explain Arid 3 is more tolerant to LL than Airlie.

The sequencing data from transcriptome analysis
To understand the molecular mechanism underlying LL responses of tall fescue, RNA-sequencing was performed on two cultivars under LL stress and the control using Illumina HiSeq 2500 platform. Statistical analysis and counts of raw reads cleaned reads (after removing low-quality reads) and reads containing poly-N or adapter sequences are listed in Additional le 1. Phred quality score (Q-score) was used to assess the accuracy of sequencing. The clean Q30 base rate of all samples was above 93%, to ensure that a high quality of transcriptome sequencing data for further analysis (Additional le 1).
De novo assembly was performed to generate transcripts from our RNA-Seq data using the Trinity tool, since the tall fescue genome is not yet sequenced. A total of 423,008 transcripts and 136,925 unigenes were obtained from the sequencing data. The minimum and maximum lengths of transcripts were were 201 bp and 13,340 bp, consistent with the unigenes (Additional le 2). The length of most transcripts and unigenes were concentrated between 200 to 400 bp (Additional le 3). The KOG analysis identi ed 25 categories and functions of unigenes. The 'general function prediction only' was the largest category (8105 genes), followed by 'signal transduction mechanisms' (3192 genes) and 'posttranslational modi cation, protein turnover, chaperones' (3182 genes) (Fig. 3A). Altogether, the unigenes identi ed in this study contribute to the data of transcriptomes responsive to LL stress in tall fescue, and provide useful information for other turfgrass and forage grass.

DEGs identi cation in two tall fescue cultivars under LL stress and the control
A total number of 24944 DEGs were identi ed in Arid 3 between LL stress and the control, whereas more DEGs were found in Arilie. For Arid 3, the number of up-regulated genes was higher than that of downregulated genes. However, more than half of DEGs in Airlie were down-regulated (Table 2; Fig. 4). A total of 20783 DEGs were identi ed between the two cultivars under LL stress; over half were up-regulated. To understand which DEGs account for the LL response in tall fescue, we performed a composition analysis with DEGs identi ed in the two cultivars under LL stress and the control. There were considerable overlaps and distinctions in DEGs between two cultivars, and LL stress and the control. Airlie had 9269, and Arid 3 had 6403 speci c DEGs after LL treatment, while the two cultivars shared 5254 DEGs ( Fig. 3B; Fig. 3C). The overlapping differences between the two cultivars under LL stress suggest that they may have shared, but different molecular mechanisms to cope with LL stress. The genes differentially expressed among samples were also shown in the heatmap (Fig. 5).

Gene Ontology categorization and KEGG pathway analysis
GO enrichment analysis of DEGs was performed under three categories, including cellular components, biological processes, and molecular functions ( Fig. 6A; Fig. 6B). The DEGs in both cultivars under LL stress and the control displayed similar GO enrichment patterns, but different gene numbers. Most of the DEGs in both cultivars were involved in 'cellular process', 'metabolic process', 'cell part', 'organelle', 'binding', and 'catalytic activity'. For Arid 3, the proportion of up-regulated DEGs was higher than that of down-regulated DEGs in these abundant categories. However, The DEGs of Airlie distributed in these abundant categories showed an opposite trend.
The KEGG pathway analysis was performed to unravel the signaling pathway involved in LL response of two cultivars. It showed that 'glutathione metabolism', 'plant-pathogen interaction', 'alpha-Linolenic acid metabolism', and 'phenylpropanoid biosynthesis' were signi cantly enriched in Arid 3 and Airlie after LL treatment (P < 0.05). Apart from the co-enriched pathways, other pathways related to metabolism were also signi cantly enriched in Airlie. These include 'phenylalanine', 'tyrosine and tryptophan biosynthesis', 'linoleic

Differential expression of genes involved in photosynthetic procession
The expression of genes associated with chlorophyll metabolic pathways and photosynthetic pathway was obviously in uenced by LL stress in two cultivars. In Arid 3, genes that encode for four important enzymes, In photosystem II, four genes encoding for photosystem II oxygen-evolving enhancer protein 2 (PsbP), photosystem II oxygen-evolving enhancer protein 3 (PsbQ), photosystem II PsbW protein (PsbW) and photosystem II 13 kDa protein (Psb28) had a decreased expression level in Arid 3 after LL treatment. Arid 3 had four down-regulated genes of photosystem II, but almost all genes in Airlie's photosystem II were down regulated after LL treatment, which involved in ten gens ( Fig. 8; Additional le 5). Similar results are reported for genes involved in electron transport chain and F-type ATPase, which were down-regulated in LL-treated Airlie. By contrast, only two genes in LL-treated Arid 3 were down regulated, included ferredoxin (PetF) and ferredoxin-NADP + reductase (PetH). Expression of the chlorophyll a/b-binding protein, well known as an important component of the light-harvesting chlorophyll protein complex I and II, was examed in Arid 3 and Airlie. There were 18 up-regulated and 15 down-regulated genes in Arid 3 after LL treatment, whereas Airlie had only one up-regulated and 26 genes down-regulated (Additional le 6). These results indicate that in Airlie, photosynthesis process is strongly inhibited by LL treatment, while Arid 3 has better photosynthetic capacity under LL stress. This is an agreement with the better photosynthetic capacity in Arid 3 than in Airlie under LL stress.
The expression of several genes changed after LL treatment. The expression of FaPhyA remarkably increased in LL treated Arid 3 and Airlie compared with the control. Under LL stress, the expression level of FaPhyA in Arid 3 was almost twice as much as it in Airlie. For gene FaPhyB, under LL stress, the expression increased in Arid 3, while it decreased in Airlie when compared with the control. And the expression level in FaPhyB was much higher than in Airlie under LL stress. FaPIF5 was up-regulated in both Arid 3 and Airlie, but its expression was more substantial in Airlie. The expression of FaCOP1 in Airlie remained stable with LL treatment, but remarkably decreased in Arid 3 (Fig. 9).
2.6. Validation of RNA-seq data using qRT-PCR Ten DEGs were randomly selected for expression analysis using qRT-PCR. Each cDNA preparation was normalized using FaActin and Faα-Tubulin as internal controls. The qRT-PCR results of the ten selected genes showed consistent expression patterns with the RNA-seq data, con rming the reliability of the Illumina sequencing results. Moreover, expression pro ling of the ten DGEs through qRT-PCR and RNA-seq results showed high correlation coe cients value of 0.8835 (Fig. 10A) and 0.8812 (Fig. 10B) in Arid 3 and Airlie, respectively.

Discussion
When subjected to LL stress, plants ultilize two different adaptation strategies: shade avoidance and shade tolerance [21,22]. In this study, LL tolerant cultivar Arid 3 displayed better performance under LL stress than LL sensitive cultivar Airlie. The most notable features are that Arid 3 had relatively higher biomass and produced more tillers than Airlie under LL stress. Tiller is one of the important characteristics of turfgrass, which directly affects trampling resistance and the turf quality. Moreover, more tillers mean higher stem density and biomass content of fescue [23]. Chlorophyll has a signi cant response to LL stress; it traps light energy and drives electron transfer process, playing a central role in plant photosynthesis [24]. The content of Chl b in Arid 3 was observably higher than in Airlie under LL stress (Fig. 1). Under LL stress, the content of chlorophyll in cells decreases signi cantly, especially chlorophyll b of Scenedesmus dimorphus [25]. However, some plant species produce high levels of Chl b to improve light-catching ability under LL stress, as well as increase the content of light-catching pigment complexes and the activity of photophosphorylation [26]. The synthesis of chlorophyll is regulated by many factors, including the stimulation of photoreceptors by external light [27]. The transformation from glutamic acid (Glu) to chlorophyll is catalyzed by many enzymes, such as GluTR, uroporphyrinogen III synthase (UROS), uroporphyrinogen III decarboxylase (UROD), chlorophyll speci c synthase (MgCh), DVR, POR, CAO, and CHLG. The expression levels of genes encoding these key enzymes differed between Arid 3 and Airlie (Fig. 5). GluTR is the key enzyme in the biosynthesis of porphyrin compounds, regulated by biological rhythm, light, and end products [28]. As one of the three known lightdependent enzymes in the chlorophyll biosynthesis pathway, POR catalyses the reduction of photosensitizer and substrate protochlorophyllide to form the pigment chlorophyllide. The active site of POR promotes the light-driven reduction of protochlorophyllide through the transfer of local hydride and long-distance proton [29]. The different expression levels of genes encoding for these key enzymes in Arid 3 and Airlie under LL stress may explain the varied chlorophyll contents between the two cultivars. The gene encoding for chlorophyll degradation enzyme pheophorbide, a monooxygenase (PAO), was down-regulated after LL treatment. The down-regulation may be a self-protection mechanism of plants when subjected to LL stress.
The expression level of FaCHLI, the gene related to chlorophyll biosynthesis, was higher in Arid 3 than Airlie under LL stress, which is an important evidence that Arid 3 produced more Chl than Airlie under LL stress ( Fig. 7).
Photosynthesis and light are two closely related, primary processes in all plants [30]. LL stress decreases photosynthetic and transpiration rates, as reported in seashore paspalum and bermudagrass [31]. In this study, GO and KEGG enrichment analysis of Arid 3 and Airlie suggested obvious changes in photosynthesis in the two cultivars under LL stress (Fig. 6). Under the LL stress, the values of F v /F m and Φ PSII remarkably decreased in Airlie, but the decrease in Arid 3 was not substantial. Consistent with other stresses, the degree of opening of reaction centers of PS also declined under LL stress [32] Compared to control, tall fescue under LL stress broke the maintenance of photochemical health due to lower Φ PSII activity. Due to LL stress, several proteins of PS , including the PsbP ,PsbQ, PsbR, PsbW, Psb27, and Psb28, were down-regulated in Arid 3, more than in Airlie. The different effects of LL on PSII proteins of Arid 3 and Airlie imply a stronger photosynthetic capacity of Arid 3 (Fig. 6). Chlorophyll uorescence provides valuable information for the study of light in the PS photochemical reaction. The results of this study showed that LL stress inhibited F v /F m and Φ PSII in tall fescue, suggesting that electronic produced from PS shift from chlorophyll was limited by PQ [33]. The F v /F m , Φ PSII , q p , and ETR indicators were higher in Arid 3 than in Airlie under LL stress, suggesting that Arid 3 had high PS maximal quantum yield and electron transfer e ciency.
Interestingly, the value of F v /F m and Φ PSII in Arid 3 only changed slightly under LL stress compared with control, while in Airlie, these two indicators were substantially reduced. The q P value represents the open ratio of PSII reaction centers. A rise in q P value accounts for the increased capacity for electron transport, but q P value decreases under abiotic stress [34,35]. This study showed that q P signi cantly decreased under LL stress, although the q P value in Arid 3 maintained a remarkably higher level than in Airlie (Fig. 1), demonstrating that Arid 3 mitigates the damage of the PSII reaction center and maintains a higher ETR between electron acceptors and providers compared with Airlie. LL stress also decreased the expression of genes encoding for PetF, PetH, and PetJ, which are related to photosynthetic electron transport. Moreover, the decreased levels of these genes in Arid 3 were lower than in Airlie. These results could explain why Arid 3 maintained a higher ETR than Airlie under LL.
Phytochrome is a widely studied family of photoreceptors that respond to red and far-red light in Arabidopsis thaliana [36,37], which involves in many physiological processes, such as seed germination, seedling development, photosynthesis, and shading reaction [38]. LL stress, de ned as the reduced photosynthetically active radiation (PAR) and changed red/far-red light ratio, involves phytochrome A and phytochrome B [39,40]. Photochrome A (PhyA) is the only photoreceptor capable of sensing far-red light and responding to this signal in plants [41]. In this study, the expression levels of FaPhyA were signi cantly up-regulated in both cultivars, and at higher levels in Arid 3 than in Airlie (Fig. 7). Studies have shown that PhyA plays an important role under LL stress, inhibiting the negative regulatory factor phytochrome-interacting factors (PIFs) in the photomorphogenesis process [40]. PIFs, a branch of the superfamily of basic helic-loop helix transcription factors, are key inhibitors of photomorphogenesis [42]. PIF4 plays a role in the phytochrome signaling pathway [43]. PIFs interact with abscisic acid (ABA) receptors to regulate ABA signal transduction under LL and inhibits plant photomorphogenesis [44]. FaPIF5 was up-regulated in two cultivars under LL stress. However, the expression level in Airlie was higher than that in Arid 3, indicating that Arid 3 inclined to photomorphogenesis, while Airlie was inclined to skotomorphogenesis under LL stress. PhyB functions as red-light receptors and accumulates in large amounts under the induction of red light [45].
Phytochrome A (PhyB) degrades PIF5 through phosphorylation and then promotes photomorphogenesis; On the one hand, MYB30 inhibits this degradation process, on the other hand, it directly induces the expression of PIF5 [46]. This study showed that the expression of FaPhyB was signi cantly decreased in Arid 3, but upregulated in Airlie. The differential expression of the positive regulatory factor FaPhyB in photomorphogenesis may explain shade avoidance and shade tolerance strategies utilized by the two cultivars. Constitutively photomorphogenic 1 (COP1), an E3 ubiquitin ligase, is a protein complex that acts as a central inhibitor in photomorphogenesis [47]. COP1 is believed to be a central switch of global lightresponsive gene expression in Arabidopsis thaliana by disrupting HY5 and other transcriptional regulators critical to light morphogenesis in the dark [48,49]. It was found that (Fig. 7) the expression of FaCOP1 decreased remarkably in Arid 3 under LL stress compared to control, but little change was observed in Airlie.
Under red light, red receptor PhyB enters the nucleus, where it inhibits the activity of COP1 and promotes its degration. PhyB also promotes the degradation of PIFs, thus promoting photomorphogenesis [50]. Synthesis of COP1 is induced by far-red light and inhibited by PhyA, indicating the positive regulatory role of COP1 in photomorphogenesis [51].
Except for FaPIF5 and FaCOP1, both FaPhyA and FaPhyB played positive roles in regulating photomorphogenesis of tall fescue cultivars Arid 3 and Airlie under LL stress and the control. FaPIF5, FaCOP1, FaPhyA, and FaPhyB inter-regulated with each other to jointly promote the photomorphogenesis of Arid 3 and the skotomorphogenesis of Airlie, which explains the better performance of Arid 3 under LL stress.
However, the functional mechanisms of these transcription factors in tall fescue remains to be further explored.

Conclusion
This study revealed gene numbers and expression levels caused by cultivar and LL stress-dependent differences in Arid 3 and Airlie. Photosynthetic physiology of tall fescue under LL stress is associated with several genes and pathways, such as phytochrome genes, chlorophyll metabolism, photosystem , and photosynthetic electron transport. Additionally, the two cultivars had contrasting photosynthetic capacity under LL stress, re ected in the transcriptome pro le. This study rstly provides a understanding of LL responses in tall fescue at the photosynthetic physiological and molecular level and highlights the importance of utilizing RNA-seq technology to determine the genes responsible for LL responses.

Low-light phenotyping and measurement of photosynthetic parameters
The tiller numbers of each plant were counted individually. Total dry weight contains aboveground dry weight and belowground dry weight, were determined after drying at 80°C for 72 h in drying.
Chlorophyll (Chl) and carotenoid contents were determined as previously using the modi ed method of Arnon [52]. In brief, approximately 0.1 g of fresh leaves were soaked in a solutions of 10 mL acetone (80%) and 5 mL ethyl alcohol (95%) for 24 h in the dark. The absorbance of the supernatant was measured at 440, 663, and 645 nm using a spectrophotometer. The contents were calculated using the formula described by Arnon [52].
Net photosynthetic rate (Pn), stomatal conductance (Gs), intercellular CO 2 concentration (Ci), and transpiration rate (Tr) of the second fully expanded leaf from the top were monitored by open gas exchange system (Li-6400, LICOR, Inc, Lincon, NE). All samples were illuminated for 30 min with 800 µmol m − 2 s − 1 of saturated photosynthetic photon ux density using a 6400-02B LED light source [53]. High-quality clean reads were obtain after ltering the low-quality and primer/adapter contaminated reads with NGSQC ToolKit. The constructed library was sequenced with Illumina platform, and the sequencing strategy was PE150. The original o ine sequences obtained by Illumina platform sequencing were processed by removing low-quality sequences, removing joints and other processes to obtain high-quality sequences.

Analysis of RNA sequencing data
The Trinity software was used for de novo assembly of tall fescue transcriptome. Busco evaluation was performed on all transcripts, and for integrity assessment, Unigene was assembled by Trinity. Meanwhile, other assessments, including accuracy and uniformity, were performed to verify the assembly results. The Trinotate (Trinotate Release v3.0.2) integration was used to obtain comprehensive functional annotation results. Functional annotations from the metabolic pathway and KOG annotations [55] were obtained to classify the functional characteristics.

Gene expression pro ling
Differentially expressed genes (DEGs) between two cultivars, and the LL stress and the control were determined as log 2 Tables   Table.1 The primers for qRT-PCR.  control), ALL (Arid 3 treated by LL stress), VCK (Airlie control), and VLL (Airlie treated by LL stress). Each value is the mean±SD of ve independent experiments (n=5). Different letters indicate signi cant differences at P < 0.05, according to Duncan's multiple range tests.    Heatmap of genes differentially expressed. The Heatmap between ALL and ACK (A), VLL and VCK (B). Blue color represents low expressed genes and red color represents highly expressed genes. Cluster analysis was done using the log2 (FPKM + 1) value from larger to smaller.  The expression of genes encoding important enzymes in chlorophyll metabolic pathway between ALL-ACK and VLL-VCK. Blue color represents down-regulate and red color represents up-regulate. The darker the gradient, the larger the value of the gradient.

Figure 8
Photosynthesis-related gene expression based on the KEGG pathway analysis. Blue represents the downregulation. The darker the gradient, the larger the value of the gradient.