(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Systematic characterization of photoperiodic gene expression patterns reveals diverse seasonal transcriptional systems in Arabidopsis [1]

['Chun Chung Leung', 'Department Of Molecular', 'Cellular', 'Developmental Biology', 'Yale University', 'New Haven', 'Connecticut', 'United States Of America', 'Daniel A. Tarté', 'Lilijana S. Oliver']

Date: 2023-09

Photoperiod is an annual cue measured by biological systems to align growth and reproduction with the seasons. In plants, photoperiodic flowering has been intensively studied for over 100 years, but we lack a complete picture of the transcriptional networks and cellular processes that are photoperiodic. We performed a transcriptomics experiment on Arabidopsis plants grown in 3 different photoperiods and found that thousands of genes show photoperiodic alteration in gene expression. Gene clustering, daily expression integral calculations, and cis-element analysis then separate photoperiodic genes into co-expression subgroups that display 19 diverse seasonal expression patterns, opening the possibility that many photoperiod measurement systems work in parallel in Arabidopsis. Then, functional enrichment analysis predicts co-expression of important cellular pathways. To test these predictions, we generated a comprehensive catalog of genes in the phenylpropanoid biosynthesis pathway, overlaid gene expression data, and demonstrated that photoperiod intersects with 2 major phenylpropanoid pathways differentially, controlling flavonoids but not lignin. Finally, we describe the development of a new app that visualizes photoperiod transcriptomic data for the wider community.

Funding: This work was supported by the National Institutes of Health (R35 GM128670) to J.M.G., and D.A.T. was supported by the National Institutes of Health (T32GM007223-44). Q.W. was supported by the Forest BH and Elizabeth DW Brown Fund Fellowship provided by Yale University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2023 Leung et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Despite these inroads towards understanding photoperiod-responsive transcriptional systems, we still have an incomplete understanding of the genes and cellular processes regulated by photoperiod and the scope of potential photoperiod measuring systems in plants. Deficiencies in studying photoperiodic transcriptomes have been caused by variation in sampling frequency, time points, growth conditions, photoperiod length, and ease of data access. To address this, we performed RNA-sequencing on a 24-h Arabidopsis time course encompassing 3 photoperiods: 8 h light followed by 16 h dark (8L:16D), 12L:12D, and 16L:8D. We used an rDEI and pattern clustering pipeline to identify and classify photoperiod-regulated genes. Furthermore, cis-element analysis was performed to provide further evidence that co-clustered genes share known and de novo transcription factor-binding elements that point towards distinct photoperiodic transcriptional systems. Additionally, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses identified a host of cellular pathways that are potentially controlled by photoperiod in Arabidopsis. We then followed one important cellular pathway, phenylpropanoid biosynthesis, and found a complex regulatory network that differentially controls separate branches of this pathway. Finally, we present “Photo-graph,” an app for user-friendly visualization of photoperiod data. Together, this work provides a comprehensive examination of photoperiod-responsive transcriptional systems in Arabidopsis and suggests that a multitude of networks control important cellular pathways in response to daylength.

Recently, 2 studies reanalyzed older transcriptomic data and uncovered new photoperiod measurement mechanisms. A meta-analysis of Arabidopsis transcriptomics led to the discovery that phytochrome A is important for light sensing in short days [ 36 ]. Additionally, a study using relative daily expression integral (rDEI = sum of 24 h of expression in condition 1/sum of 24 h of expression in condition 2) followed by expression pattern clustering identified short day-induced genes in Arabidopsis and precipitated the discovery of the MDLM system [ 21 ].

In the last 30 years, transcriptomics has emerged as an important tool for understanding the breadth of photoperiodic gene regulation. Subtractive hybridization was first used to identify photoperiod-regulated genes involved in flowering time [ 11 ], and subsequently microarray was used to identify local and global gene expression changes in response to the floral transition [ 26 , 27 ]. Additionally, microarrays were used to track gene expression changes in Arabidopsis at dusk and dawn under many photoperiods, and time course studies provided a view of the genes that exhibited altered phasing under long- and short-day photoperiods [ 23 , 28 ]. Transcriptomics have now been implemented to study photoperiodic gene expression in Arabidopsis hallerrii [ 29 ], Panicum hallii [ 30 ], wheat [ 31 , 32 ], Medicago [ 33 ], sugarcane [ 34 ], and soybean [ 35 ]. These studies have revealed that photoperiodic gene expression changes mainly manifest as changes in phase (i.e., clock genes) or amplitude (i.e., FT or PP2-A13).

In addition to the CO-FT, PIF regulatory modules, and MDLM, it has been recognized that the circadian clock and circadian clock-controlled genes exhibit phase delays as photoperiod lengthens [ 23 ]. Models predict that the multiple interlocking feedback loops of the clock allow for clock genes to track dusk as it delays, relative to dawn [ 24 ]. Recently, EMPFINDLICHER IM DUNKELROTEN LICHT 1 (EID1) was shown to be required for photoperiodic response of the circadian clock in tomato, but detailed mechanistic understanding of this phenomenon is lacking in many plants [ 25 ].

Recently, a metabolic daylength measurement (MDLM) system was shown to support rosette fresh weight generation in long days and short days [ 19 – 22 ]. This system relies on the photoperiodic control of sucrose and starch allocation in order to control expression of the genes PHLOEM PROTEIN 2-A13 (PP2-A13) [ 21 ] and MYO-INOSTOL-1 PHOSPHATE SYNTHASE 1 (MIPS1) [ 22 ], which are required to support short- and long-day vegetative growth, respectively. Like the CO-FT and PIF regulatory modules, the MDLM system requires a functional circadian clock for photoperiod measurement, although the molecular connections between the clock and metabolism for this system have not been identified. Additionally, both the transcription factor(s) that control MDLM-regulated gene expression and the full scope of MDLM-regulated genes remain unknown.

Growth is also under the control of photoperiod in plants, and recently, 2 photoperiod-measuring mechanisms that support or promote photoperiodic growth have been discovered. Photoperiodic control of hypocotyl elongation by phytochrome-interacting factors (PIFs) relies on a coincidence mechanism, similar to the CO-FT regulon, although PIFs have a wide variety of functions apart from regulating genes in a photoperiodic manner [ 14 ]. The circadian clock phases the expression of PIF4/5 to the morning and late night, but the PIF4/5 proteins are only stabilized in the dark and thus promote nighttime expression of growth-regulating genes, namely ARABIDOPSIS THALIANA HOMEOBOX PROTEIN 2 (ATHB2) [ 15 – 18 ]. Therefore, PIF4/5-regulated hypocotyl elongation occurs in the latter portion of the long night during short-day photoperiods.

Plants have proved an influential study system for photoperiodism, mainly because the control of flowering time by photoperiod provides a readily observable and quantifiable phenotype. Photoperiodic flowering in many higher plants is regulated by the circadian clock-controlled expression of the CONSTANS (CO) gene [ 10 ]. In Arabidopsis thaliana, accumulation of CO mRNA occurs in late afternoon—a time that is lit only during the long photoperiods of summertime. Therefore, only in long photoperiods can the CO protein be stabilized by light and trigger the downstream inducers of flowering, namely FLOWERING LOCUS T (FT). This overlap between photoperiod and the rhythmic expression of CO thus defines the external coincidence mechanism. Transcriptionally, CO is proposed to control a small number of genes directly yet maintains a large indirect effect on gene expression and development by triggering the developmental switch from vegetative growth to flowering [ 11 – 13 ].

Photoperiod, or daylength, is a robust seasonal cue that is measured by organisms ranging from algae [ 1 ] and fungi [ 2 ], to higher plants [ 3 ] and vertebrates [ 4 ]. This circannual signal allows the anticipation of environmental changes and thus the coordination of long-term developmental and reproductive processes, such as tuberization in potatoes [ 5 ] and maturation of animal gonads [ 6 ]. Lengthening or shortening of photoperiod beyond a normal range seen in a 24-h day can cause a distinct stress response in plants [ 7 , 8 ]. In humans, photoperiod influences mood variation and related conditions like seasonal affective disorder [ 9 ].

Results

Circadian clock genes Lengthening photoperiod causes delayed phase of circadian clock genes [23]. Four subgroups have evidence that prompted us to classify them as clock genes associated with photoperiod: 3N, 4I, 4J, and 11J (Fig 2A). 3N, 4I, and 4J have a single expression peak phased to midday, while 11J has a single expression peak phased to dawn. Phase analysis shows that groups 3N and 4I show the hallmark phase delay associated with clock genes responding to lengthening photoperiod (S9 Fig and S9 Data). Groups 4J and 11J do not show the same change in phase but show an increase in magnitude in SD, resulting in a slight increase in the ratio of SD DEI to LD DEI (rDEI SD:LD ) (Table 1). All 4 clusters contain known clock genes. 4J and 11J are enriched in GO terms “circadian rhythm” and “rhythmic processes” (Fig 2C). 3N is enriched for the GO terms “response to cold” and “cellular polysaccharide catabolic process”. 4I is enriched for GO terms related to protein nitrosylation. 3N and 4J show statistically significant enrichment of the evening element, a well-studied clock cis-element [43] (Fig 3A). 11J shows enrichment of the bZIP binding core sequence, ACGT [44]. Our results identified 4 photoperiodic subgroups that are likely linked to the circadian clock. Two showed the hallmark change in phase associated with the clock response to photoperiod and 2 showed no change in phase but slight amplitude increases in response to photoperiod. Together, the identification of photoperiod-regulated clock genes and the clock cis-elements confirms that our dataset can identify known photoperiod responsive transcriptional networks.

Short day-induced genes In the clustering performed here, 11O is the largest of the SD-induced subgroups, as determined by rDEI (Fig 1B and Table 1). However, further dynamic tree cutting suggests that 11O contains 2 separate expression groups, which we termed 11Oa and 11Ob (Fig 2B). Both groups have biphasic expression in SD and are repressed in the light. 11Oa is distinguished by a dominant post-dusk peak and a weaker dawn-phased peak, while 11Ob is characterized by a weaker post-dusk peak and a more prominent dawn-phased peak. The 11Oa subgroup contains the MDLM-regulated gene PP2-A13, and the expression pattern of this subgroup is identical to the PP2-A13 daily expression pattern shown previously [21]. Furthermore, it contains genes shown to be important for short-day physiology (PP2-A13, EXORDIUM-LIKE 1, and HOMOGENTISATE 1,2-DIOXYGENASE) [21,45]. In support of its role in short-day plant physiology, 11Oa is enriched with genes involved in hypoxia, response to absence of light and amino acid catabolism (Fig 2C). The enrichment of hypoxia responsive and amino acid metabolism genes suggests a response to lower energy availability in SD. Breakdown of branched chain amino acids is an energy scavenging mechanism and is a major response to both hypoxia and extended darkness when energy is limited [46–48]. Conversely, 11Ob has a weaker post-dusk expression peak, but a more dominant dawn-phased expression peak (Fig 2B). 11Ob contains TEMPRANILLO1 (TEM1), a gene known to repress FT expression in short days, but 11Ob shows no enrichment of any individual cellular pathways (Table 1 and Fig 2C) [49,50]. We next inquired whether the 2 subgroups have enrichment of shared or distinct cis-elements. The entire 11O subgroup has 2 enriched cis-elements: the bZIP TFBS resembling the G-box (core sequence CACGTG) [44] and the AP2/ERF TFBS resembling the GCC-box (core sequence AGCCGCC) [51] (Fig 3A). Interestingly, 11Oa has the dominant post-dusk expression peak but lacks enrichment of the bZIP sites, only containing that of the AP2/ERF sites. 11J has genes that are dawn-phased and is enriched with the bZIP sites but not the AP2/ERF binding sites. 11Ob contains genes that have the post-dusk peak and the dawn-phased peak and is enriched with both AP2/ERF and bZIP sites. This correlation may indicate that the AP2/ERF sites are important for post-dusk phasing in short days, and the bZIP sites are important for dawn phasing. Cluster 3, which contains subgroups mostly induced in LD, also contains the outlier subgroup 3Y that is induced in SD (Fig 2A). This subgroup demonstrates monophasic peaking at ZT4 that increases in amplitude in short days. This SD-induction in the light rather than the dark makes 3Y unique. It was also enriched in genes involved in hypoxia (Fig 2C). We were unable to identify any know cis-regulatory elements that were enriched in 3Y (Fig 3A). A search for de novo motifs identified 1 strongly enriched element containing the sequence CCACAATCCTCA (Fig 3B). These results suggest that there are potentially 3 transcriptional systems controlling 3 major SD-induced gene expression programs. One is characterized by strong post-dusk induction and is enriched with an AP2/ERF binding site. A second potential program is exemplified by the dawn-phased genes enriched with the bZIP core. bZIP transcription factors (TFs) play a number of roles in plants, including control of the circadian clock and light signaling [52,53]. A third subgroup, 3Y, shows high amplitude SD expression at ZT4 and contains a de novo motif. Little is known about this smaller transcriptional system, but the enrichment of important cellular pathways, such as hypoxia and amino acid metabolism, suggests this may be important for plants grown in SD.

Long day-induced genes The majority of LD-induced genes reside in cluster 3, but in contrast to the SD-induced genes, cluster 3 contains a greater number of smaller subgroups rather than 1 large subgroup like 11O (Fig 2A). This could indicate that multiple photoperiod-measuring systems control gene expression in long days. This is supported by evidence showing that the MDLM and CO systems can cause similar photoperiodic gene expression changes (S7 Data) [22]. To determine if there are possible transcriptional systems that are driving LD-induced gene expression, we further analyzed 5 major subgroups from cluster 3 (3G, 3M, 3O, 3P, and 3R). All are expressed mainly in the light period of the day, hence their presence in cluster 3, but only 3M, 3O, and 3R are strongly repressed by the dark in all 3 photoperiods (Fig 2A). 3M is enriched in genes related to pigment metabolic process, desiccation, chlorophyll metabolic process, response to oxidative stress, response to red light, and water homeostasis (Fig 2C). 3O is enriched in genes involved in protein folding, glucosinolate metabolic process, response to heat, and protein processing in the endoplasmic reticulum. 3R is enriched in genes involved in blue light signaling, response to light intensity, and photosynthesis; cis-element analyses did not identify any single site enriched in subgroups 3G, 3O, and 3P (Fig 3A). Conversely, 3M and 3R are weakly enriched in bZIP sites. 3M and 3R have a similar expression pattern, resembling that of the MDLM-controlled gene MIPS1, which is located in 3M [22]. Because of the shared enrichment of cis-elements in the subgroups that contain the LD and SD MDLM genes, it is possible that the same families of TFs are in play to control gene expression in both photoperiods. In addition to the aforementioned subgroups that result in higher gene expression in LD and are expressed mostly in the light period, there is 1 night-phased LD-induced subgroup, 6G (Fig 2A). Also displaying higher expression in LD is the day-phased subgroup, 2C, which achieves this through a peak magnitude increase at ZT4. Similar to 3G, 6G and 2C have no enrichment of any biological pathways or known cis-elements (Figs 2C, 3A and 3B). In sum, we can identify target genes from known photoperiod measurement systems intermingling in the large C3 subgroup. The CO-regulated genes are spread across many subgroups, but the MDLM-regulated genes are clustered in 3M and 3R, based on cis-element enrichment analysis and expression pattern. Additionally, there may be photoperiod measurement systems that have not been identified that could account for other modes of expression.

Photoperiod regulation of ribosomal genes One subgroup, 3I, is defined by a ZT8-specific trough in LD that causes a biphasic expression pattern only in LDs (Fig 2A). Furthermore, this subgroup is strongly enriched with genes involved in ribosome biogenesis and translation (Fig 2C). In support of this, cis-element analysis showed enrichment of the binding site for the Myb-type TF TELOMERE REPEAT BINDING FACTOR (TRB) 2 and AT1G72740 (Fig 3A), both belonging to a TF family of evolutionarily conserved regulators of ribosome gene expression [54,55]. This subgroup is unique because it was the only major subgroup defined by an expression trough rather than an expression peak (S8 Fig). It will be worthwhile in the future to determine if the TRB site plays a role in this process.

Equinox induced-genes It is conceivable, and demonstrated in some cases, that some biological processes may be induced or repressed specifically in the equinox photoperiods in plants [3]. We included a 12L:12D equinox photoperiod in order to test this idea. We found few genes that were expressed highly in LD and SD but repressed in EQ, but we found a greater number of genes that are expressed specifically in EQ but reduced in LD and SD. These included clusters 3A (n = 82), 4C (n = 92), 4D (n = 126), 9A (n = 56), 11B (n = 41), and 11C (n = 39). These were spread across a variety of peak times, but only 4D contained more than 100 genes (S7 and S8 Figs). In 4D, we found enrichment of electron transport chain genes, suggesting it is important for photosynthetic processes (Fig 2C). We did not identify additional elements that point towards an EQ-specific mechanism, but this could be investigated further in follow-up studies. In the previous sections, we defined 19 major photoperiod expression patterns and tentatively linked 13 to biological processes or cis-elements. Notably, the 6 other patterns did not show enrichment of annotation or promoter cis-elements, and further evidence is required to suggest them as distinct photoperiodic transcriptional systems. What is clear is that photoperiod gene expression changes can manifest with a diverse array of daily expression patterns that cannot be accounted for with our current knowledge of photoperiod measurement systems in plants.

Photoperiodic control of phenylpropanoid biosynthesis We next tested if our pipeline is effective at identifying and classifying bona fide photoperiod-regulated cellular pathways. GSEA identified phenylpropanoid biosynthesis as one of the top cellular processes enriched with photoperiod-regulated genes (Fig 1C). Anthocyanin production is controlled by photoperiod in many plants [56], but in Arabidopsis it is not clear if they are induced by short or long days, nor if other byproducts of the phenylpropanoid pathway, such as other flavonoids or lignin, are also regulated by photoperiod [36,57]. To address this, we curated a catalog of genes involved in phenylpropanoid synthesis in Arabidopsis using KEGG, GO, and an extensive literature search (S10 Data). Each gene was annotated according to its predicted effect on the phenylpropanoid pathway, mode of action, and the branch of the pathway in which it acts. To determine how photoperiod regulates the transcription of positive and negative regulators of the phenylpropanoid pathway, both groups were plotted according to their rDEI SD:LD (Fig 4A). The expression of positive regulators of phenylpropanoid biosynthesis, especially that of the flavonoid branches, was found to be significantly higher in LD. To visualize the seasonal induction of phenylpropanoid genes more precisely, we mapped the rDEI SD:LD of key enzymes to the phenylpropanoid biosynthesis pathway (Figs 4B and S10). Notably, enzymes specific to the flavonoid branches are more highly LD-induced than those specific to the lignin branch, which also contains the SD-induced gene CINNAMOYL COA REDUCTASE 1 (CCR1) (S11 Fig). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Photoperiod regulates phenylpropanoid gene expression and metabolite accumulation. (A) Distribution of rDEI SD:LD in genes involved in phenylpropanoid production (n = 189) (S10 Data). Genes are grouped according to positive/negative effect on the phenylpropanoid pathway, molecular function as an EZ, TF, or PT regulator, or LIG vs. FLA branch. Red bars indicate mean. *, p ≤ 0.05, ***, p ≤ 0.0001 (1 sample Wilcoxon signed rank test). Blue shading, SD-induced genes, or compound accumulation; red shading, LD-induced genes or compound accumulation. (B) Simplified phenylpropanoid biosynthesis pathway (S10 Data). Box labeling corresponds to biosynthetic enzyme names; box shading corresponds to log 2 (rDEI SD:LD ) of the coding biosynthetic gene. (C) Precursor modifications and relative compound accumulation (S11 Data). Box labeling corresponds to compound name; box shading corresponds to SD:LD relative peak area ratios. *,†The indicated pairs of compounds could not be fully resolved from one another. EZ, enzyme; FLA, flavonoid; LD, long day; LIG, lignin; PT, post-translational; rDEI, relative daily expression integral; SD, short day; TF, transcription factor. https://doi.org/10.1371/journal.pbio.3002283.g004 Our expression analyses indicate that flavonoids are potentially induced in LDs, while the photoperiodic control of the lignin branch is weaker. To test if the observed pattern of phenylpropanoid gene expression corresponds to seasonal regulation of metabolites, we quantified various phenylpropanoid compounds in LD- and SD-grown plants (Fig 4C and S11 Data). In agreement with observed gene expression patterns, liquid chromatography–mass spectrometry (LC-MS) detection revealed higher levels of 18 flavonoid compounds in LD rather than in SD photoperiod (FDR < 0.05, Student’s t test). Again, in agreement with gene expression, quantification of acetyl bromide soluble lignin (ABSL) found lignin polymer accumulation to be unaffected by photoperiod (p > 0.1, Student’s t test) (Fig 4C and S11 Data). Together, these data provide a holistic view of the photoperiodic regulation of phenylpropanoids and suggest differential regulation of the lignin and anthocyanin/flavonol branches of the phenylpropanoid pathway with respect to photoperiod. Specifically, anthocyanins and flavonol genes are induced in LDs and the corresponding metabolites respond accordingly, while the lignin genes do not show consistent photoperiodic regulation and lignin content in cells remains constant across photoperiods.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002283

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.