Circadian and photoperiodic regulation of the vegetative to reproductive transition in plants

In addition to the negative interactions, multiple transcriptional activators are found to complement the plant circadian regulation. LIGHT-REGULATED WD1 (LWD1) and LWD2 are part of a larger family of proteins known as WD-repeat proteins, which are involved in transmitting light signals to the circadian clock to promote expression of CCA1, TOC1, PRR5, and PRR9 in the morning^42–44. The midday-phased components REVEILLEs (RVE4, RVE6, and RVE8) are considered to be homologs of CCA1 and LHY due to their shared structural and functional similarities⁴⁵. Interestingly, RVEs antagonize the action of CCA1 and LHY and bind to the same evening element (EE) during the midday period. This competition results in the activation of clock genes such as PRR9, PRR5, TOC1, GI, ELF4, and LUX. In turn, PRR5, PRR7, and PRR9 repress RVE8 expression, indicating that RVEs play an important role in maintaining the balance and robustness of the circadian regulation^45,46. Two other activating components NIGHT LIGHT INDUCIBLE AND CLOCK-REGULATED1 (LNK1) and LNK2 are transcriptional coactivators of RVE8 to regulate expression of PRR5 and TOC1, both of which repress LNK1 and LNK2 expression in return, forming another loop⁴⁷.

In Arabidopsis, the transcription of CDFs is regulated by various clock components. Specifically, CDFs are induced by the morning clock genes CCA1/LHY and repressed by the evening clock genes PRR5, PRR7, and PRR9³². This leads to a typical diurnal expression pattern of CDFs with a peak at dawn and this regulation is essential for plants to optimize their reproductive success by flowering at the appropriate time of year^53–56. Loss-of-function mutations in CCA1 and LHY lead to early flowering ²⁹, while constitutive expression of CCA1 leads to circadian arrhythmicity and late flowering ²⁸. Additionally, the evening clock genes PRR9, PRR7, and PRR5 function antagonistically with CCA1/LHY and coordinately and positively regulate CO dependent photoperiodic flowering process⁵³. CCA1/LHY also represses the transcription of CO by adjusting the rhythmic expression of the clock component GI and the F-box protein FKF1, and this leads to the appropriate timing of CO expression and photoperiodic flowering³⁵.

The photoperiodic flowering pathway is regulated by the interaction of various proteins. GI together with the blue-light receptors ZTL, LKP2, and FKF1 regulates the degradation of the flowering repressor CDFs to control the clock-mediated photoperiodic flowering pathway ⁵⁷. Under long-day conditions, the timing of GI and FKF1 expression is synchronized, peaking in the late afternoon, and the blue-light absorption enables FKF1 to form a sufficient protein complex with GI to target CDF proteins for degradation, which alleviates the transcriptional repression of the CO gene and promotes flowering⁵⁰. In contrast, diurnal expression of FKF1 and GI proteins is out of phase during short days, leading to low levels of FKF1-GI complex and CO expression under light, and consequently, the flowering is delayed. A computational model predicts that FKF1 may control FT expression directly in addition to its role in CO transcriptional activation⁵⁸. ZTL and LKP2 also play important roles in regulation of flowering time and the circadian clock by ubiquitin-mediated degradation of clock components. Introducing both ztl and lkp2 mutations into the fkf1 mutant further enhances the late-flowering phenotype of the fkf1 mutant and reduces the CO expression level, due to the increased abundance of CDF2 protein⁵⁹. In contrast to FKF1, overexpression of ZTL or LKP2 leads to an unexpected late-flowering phenotype accompanied with low levels of CO expression, similar to the ztl fkf1 lkp2 triple mutant and the gi mutant⁶⁰. One explanation for this late-flowering phenotype of these overexpressing lines is that a high level of the ZTL/LKP2 sequester FKF1 in the cytosol by forming the ZTL/LKP2-FKF1 complex. This reduces the amount of FKF1-GI-CDF1 complex in the nucleus, causing repression of CO expression⁶¹. Another possibility is that a high level of ZTL/LKP2 destabilizes core clock components TOC1 and PRR5, which interact with and stabilize CO protein to promote flowering in response to the day length^40,41,55.

Transcript levels of CO, CO/FKBP12, and FT, respectively, are diurnally oscillated (Fig. 2, bottom panel). At the post-transcriptional level, the stability of CO protein is compromised by the action of post-translational modification. The degradation of CO protein is mainly controlled by two RING-finger E3 ubiquitin ligases, HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENE1 (HOS1)⁶² and CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1)⁶³, which act in a light-dependent manner. HOS1 directly targets to CO protein to promote its degradation in the early morning, and mutation in HOS1 causes extreme early flowering in both long- and short-day conditions⁶². However, COP1 promotes CO degradation during night, leading to suppression of the floral integrator FT and delayed flowering⁶⁴, and SUPPRESSOR OF PHYA (SPA1) can enhance the E3 ubiquitin ligase activity of COP1⁶⁵. In contrast, the small immunophilin FK506 BINDING PROTEIN 12 KD (FKBP12) physically interacts with CO to prevent its degradation by COP1, and this interaction can also modulate its phosphorylation and nuclear localization, enabling it to trigger FT expression and flowering⁶⁶.

Photoreceptors also play critical roles in regulating CO stability and act antagonistically to generate daily rhythms in CO abundance¹⁴. The blue-light photoreceptors cryptochrome1 (CRY1) and CRY2 physically interact with SPA1 to suppress the activity of COP1/SPA1 ubiquitin ligase and stabilize CO protein in a blue-light-dependent manner⁶⁷. The red/far-red responsive phytochromes (PhyA–PhyE in Arabidopsis) also control flowering time⁶⁸. Among the five phytochromes, PhyA and PhyB play the most predominant function in CO protein stability^8,69. Under the far-red light, PhyA stabilizes CO protein and promotes flowering by inhibiting the COP1-SPA1 complex, while red light-activated PhyB interacts with HOS1 to promote CO protein degradation^14,62. The light-dependent control of CO protein stability is a complex and tightly regulated process that controls photoperiodic flowering. The knowledge gained from Arabidopsis research has provided fundamental insights into the mechanisms controlling photoperiodic flowering in other plant species including agricultural crops.

The function of core circadian regulators is conserved among Arabidopsis and crops. For example, the function of CCA1, LHY, and TOC1 is conserved in rice^70,71, maize⁷², and duckweed⁷³. Overexpressing maize ZmCCA1b and rice OsPRR homologs, respectively, can complement mutant phenotypes in Arabidopsis^71,72. In rice, OsCCA1 upregulates expression of strigolactone receptor and signaling and responsive genes to repress tiller-bud outgrowth⁷⁰. Down-regulating and overexpressing OsCCA1 increases and reduces tiller numbers, respectively, while altering OsPRR1 expression leads to the opposite effects. Moreover, both exogenous and endogenous sugars negatively regulate expression of OsCCA1, which in turn mediates expression of strigolactone receptor and signaling pathway genes to control tillering, flowering, and panicle development.

Unlike CO in Arabidopsis, the Hd1 gene in rice is regulated by the circadian component OsGI and displays bifunctional responses to day length: promoting flowering by activating the expression of Hd3a under short-day conditions, but repressing Hd3a expression to delay flowering under long-day conditions^77,79. This photoperiod-dependent conversion of Hd1 function from activation to repression is modulated by phytochromes and circadian clock. Photoperiodic Sensitivity5 (SE5) encodes a putative heme oxygenase involved in the biosynthesis of the phytochrome chromophore. In the se5 mutant, Hd3a expression is strongly induced by Hd1 under long-day conditions, resulting in early flowering phenotype similar to phytochrome-deficient mutants, such as OsphyB and OsphyC^80,81. Furthermore, overexpression of Hd1 can repress Hd3a expression and delay flowering under short-day conditions, and this effect requires functional phytochromes⁸². Therefore, phytochrome signals are essential for the conversion of Hd1 from an activator to a repressor of Hd3a transcription.

The rice-specific regulator Ghd7 encoding a CCT (CO, CO-LIKE, and TIMING OF CAB1) domain transcriptional regulator was identified as a major target of phytochrome signals in flowering control and acts as a repressor of Ehd1^78,83,84. Ehd1 is an important B-type response regulator that activates expression of Hd3a and RFT1, which are responsible for the transition from vegetative to reproductive growth in rice^78,84. Ghd7 acting as a repressor of flowering is acutely induced by red light in morning through the PhyA homodimer or the PhyBPhyC heterodimer, and the repressor activity of Ghd7 is modulated post-transcriptionally by PhyB. Specifically, when PhyB is activated by red light, it undergoes a conformational change that allows it to interact with Ghd7 protein, and this interaction leads to the ubiquitination and subsequent degradation of Ghd7 through the 26 S proteasome pathway, which relieves its repressor activity and promotes flowering⁸⁵. Days to heading8 (DTH8) encodes a putative HAP3 subunit of NF-YB transcription factor that is capable of binding to Hd3a promoter to enhance H3K27 trimethylation and repress Hd3a expression^86,87. Under long-day conditions, both Ghd7 and DTH8 can interact with Hd1 to form a strong repressor complex that inhibits Ehd1 expression, and this effect of the Hd1-Ghd7-DTH8 complex on Ehd1 expression can also partly explain the photoperiod-dependent conversion of Hd1 from activation to repression^86,88,89.

In sorghum, SbPRR37 activates expression of several downstream genes that repress flowering in long days⁹⁰. SbPRR37 expression is dependent on light and regulated by the circadian clock. In short days, SbPRR37 is not expressed during the evening phase, allowing sorghum to flower. Similarly, clock components such as OsPRR37/DTH7 and the EC components, also regulate rice flowering by enhancing photoperiod sensitivity. OsPRR37 is predicted to act downstream of the OsPhyB to switch the genetic effects of Hd1 on Hd3a expression and delay flowering in long-day conditions through the formation of a transcriptional repressor complex^91–93.

The EC is a crucial flowering repressor³³. OsELF3-1 and OsELF3-2, two rice ELF3 paralogs, physically interact with OsELF4a in the nucleus, whilst OsELF3-1 shows a stronger interaction with OsLUX compared to OsELF3-2⁹⁴. Recent studies have shown that functional OsELF3 proteins accumulate during the evening in short-day conditions, forming the EC to induce flowering by inhibiting the expression of floral repressors OsPRR37 and Ghd7⁹³. However, in long-day conditions, OsELF3 protein levels are low and insufficient to suppress floral repressors, resulting in delayed flowering⁹⁵. Taken together, these data indicate that rice possesses several complex genetic pathways to affect flowering time and photoperiod sensitivity.

The flowering repressor gene E1 encodes a legume-specific transcription factor that induces GmFT1a and GmFT4a expression but represses GmFT2a and GmFT5a expression. This gene is directly repressed by four soybean CCA1/LHY genes (GmLHY1a, GmLHY1b, GmLHY2a and GmLHY2b) at the transcriptional levels^98,99,105. Indeed, the soybean quadruple knockout mutant lhy1a lhy1b lhy2a lhy2b generated by CRISPR-Cas9 displayed late flowering and high E1 transcript levels in long days¹⁰⁵. Meanwhile, GmPRR3a and GmPRR3b, orthologs of Arabidopsis PRR3, induce E1 expression to delay photoperiodic flowering through downregulation of GmLHY genes^105,106. This pathway provides an additional layer of regulation for soybean flowering. The E2 locus in soybean encodes a homolog (GmGIa) of the Arabidopsis GI, which acts upstream of CO and FT in the photoperiodic flowering pathway. Like the role of GI in Arabidopsis, GmGIa plays a crucial role in integrating light and circadian signals in response to photoperiodic changes, ultimately leading to the transition from vegetative growth to reproductive development¹⁰¹. GmGIb and GmGIc are two homologs of GmGIa and can interact with two Arabidopsis FKF1 orthologs (GmFKF1 and GmFKF2) and one Arabidopsis CDF1 ortholog (GmCDF1) proteins, but GmGIa cannot interact with these three proteins, suggesting that GmGIa may play a unique role in regulating soybean flowering¹⁰⁷. GmELF3, an ortholog of clock component ELF3, physically interacts with GmLUX2 to directly repress E1 expression and promotes soybean flowering¹⁰⁸.

In addition to these genes, two maturity loci, E3 and E4, encode the phytochrome A proteins GmPhyA3 and GmPhyA2, respectively. These proteins induce expression of E1 and which, in turn, suppresses expression of GmFT2a and GmFT5a, resulting in delayed flowering under both natural day length and artificially long-day conditions^109,110. The PhyA-regulated E1-GmFT pathway is a key determinant for soybean adaption to different latitude environments. The blue-light photoreceptor protein cryptochromes also play an essential role in regulating circadian rhythm and flowering time. Soybean contains two cryptochrome genes, GmCRY1a and GmCRY2a. However, only GmCRY1a has been found to strongly promote floral initiation in soybean, and the circadian rhythm of GmCRY1a protein has different phase characteristics in different photoperiods, which suggests that GmCRY1a plays a more predominant role in regulating flowering time in response to changes in day length¹¹¹. The role of GmCRY2a in soybean flowering remains unclear and to be further investigated to elucidate its function.

Epigenetic regulation of circadian and flowering-time pathway genes plays a broader role in determining photoperiodic flowering in plants. Epigenetics refers to heritable changes in gene expression that do not involve alterations to the DNA sequence. Epigenetic mechanisms, involving DNA methylation, chromatin modifications, long-noncoding RNAs, and small interferring RNAs, have been shown to regulate expression of key photoperiodic flowering pathway genes and flowering time.

One of the best-studied epigenetic flowering events is known as vernalization, which refers to the induction of flowering after prolonged exposure to cold temperatures in winter, mediated by changes in the chromatin status of FLC¹¹². COLD INDUCED LONG ANTISENSE INTRAGENIC RNA (COOLAIR)¹¹³ and COLD ASSISTED INTRONIC NONCODING RNA (COLDAIR)¹¹⁴ originated from the first intron and 3’ end of FLC, respectively. They have been proposed to facilitate FLC silencing by removing H3K36me3 from chromatin during vernalization, particularly under low-temperature conditions^113,114. However, recent research suggests that the COOLAIR is not required for vernalization. Support for this notion comes from the normal vernalization response observed in the CRT/DRE-binding factors (CBFs) cbfs mutants with reduced levels of COOLAIR induction; moreover, both cbfs and FLC_ΔCOOLAIR mutants show a normal vernalization response despite their inability to activate COOLAIR expression during cold¹¹⁵. This work also highlights the importance of CBFs in initiating COOLAIR expression during the early stage of vernalization. CBFs bind to CRT/DREs at the 3’-end of FLC and increase progressively during vernalization, promoting COOLAIR expression. As vernalization progresses, FLC chromatin shifts to an inactive state, prompting CBF proteins to detach from the CRT/DREs in the COOLAIR promoter, thereby diminishing COOLAIR levels.

Epialleles or epigenes can also result from cross-fertilization such as imprinting^117,118 and polyploidy ^25,26. In Arabidopsis intraspecific hybrids^119,120 and allotetraploids¹²¹, expression waveforms of circadian clock genes such as CCA1 and LHY are altered by epigenetic and chromatin modifications to increase photosynthesis, chlorophyll biosynthesis, and starch metabolism¹²¹. The more starch accumulates during the day, the more it can be degraded at night to promote growth vigor, in a widespread phenomenon known as hybrid vigor or heterosis^24,27. Stress-responsive gene expression is gated by the circadian clock¹²². Altered circadian gene expression in the hybrids also regulates expression of abiotic and biotic stress-responsive genes as a trade-off to balance the energy used for growth and defense¹¹⁹, as well as mediates ethylene biosynthesis that in turn regulates growth vigor¹²³. Mechanistically, CCA1 expression changes are related to parent-of-origin effect on DNA methylation¹²⁰, suggesting an epigenetic cause. Further analysis showed that circadian clock genes mediate diurnal regulation of histone H3K4 methylation reader and eraser genes, which in turn regulate rhythmic histone modification dynamics for the clock and its output genes¹²⁴. This reciprocal regulatory module between chromatin modifiers and circadian clock oscillators orchestrates diurnal gene expression that governs plant growth and development.

All flowering plants are polyploids or of polyploid origin^125,126, which result in genetic and epigenetic changes^25,26. Duplication of flowering pathway genes in polyploids may increase the range of flowering time variation or induce other changes including epigenetic modifications and epistasis or trans-acting effects^25,127,128. Natural variation of flowering time in Arabidopsis suecica allotetraploids is largely controlled by two epistatically acting loci, namely FRIGIDA (FRI) and FLC^129,130. FRI upregulates FLC expression that represses flowering¹³¹. In Arabidopsis allotetraploids that are formed by pollinating A. arenosa with A. thaliana^132,133, there are two sets of FLC and FRI genes and they flower late¹²⁸. Inhibition of early flowering is caused by upregulation of A. thaliana FLC (AtFLC) that is trans-activated by A. arenosa FRI (AaFRI) (Fig. 5b). Two duplicate FLCs (AaFLC1 and AaFLC2) originating from A. arenosa are expressed in some allotetraploids but silenced in other lines. The expression variation of FLCs in the allotetraploids is associated with deletions in the promoter regions and first introns of A. arenosa FLCs. The strong AtFLC and AaFLC loci are maintained in natural Arabidopsis allotetraploids, leading to extremely late flowering¹²⁸. Furthermore, FLC expression correlates positively with histone H3K4me2 and H3K9ac marks and negatively with the H3K9me2 mark. This combination of epistasis between loci and chromatin regulation within a locus may provide a complexity of flowering time variation in polyploid plants in response to environmental cues and adaptive niches.

Cotton allotetraploids were formed ~1–1.5 million years ago (Mya)¹³⁴ (Fig. 5c), followed by natural diversification and crop domestication¹³⁵. Polyploidization between an A-genome African species (Gossypium arboreum-like) and a D-genome American species (G. raimondii-like) in the New World created a new allotetraploid or amphidiploid (AD-genome) cotton clade¹³⁴, which has diversified into five polyploid lineages, Gossypium hirsutum (Gh) (AD)₁, G. barbadense (Gb) (AD)₂, G. tomentosum (AD)₃, G. mustelinum (AD)₄, and G. darwinii (AD)₅. Gossypium ekmanianum and G. stephensii are recently characterized and closely related to G. hirsutum¹³⁶. Gh and Gb were separately domesticated from perennial shrubs to become annualized Upland and Pima cottons¹³⁵.

Interspecific hybridization induces genome shock, including nonadditive gene expression and epigenetic changes in polyploid plants and crops^133,137. The epigenetic changes produce epigenes or epialleles, which are selected during evolution and domestication. In allotetraploid cotton, over 500 epigenes have been identified and maintained over 600,000 years of genomic diversification, some of which are predicted to originate during polyploidization over 1–1.5 million years ago²². Many of these epigenes are related to domestication traits including flowering time, stress response, seed development, and seed dormancy.

Cotton, being originated in tropical and subtropical areas, flowers in short-day conditions¹³⁸. Loss of photoperiod sensitivity is a major “domestication syndrome” trait¹³⁹ of Upland or American cotton (G. hirsutum L.) and Pima or Egyptian cotton (G. barbadense L.) that accounts for >95% and ~5% of annual cotton crop worldwide, respectively¹⁴⁰. This process is associated with cotton CO-liked (COL) genes, which are Arabidopsis CO homolog¹⁴¹. Among 23 COLs identified in cotton¹⁴², eight G. hirsutum COLs (GhCOLs) are in the same subgroup. Among them, only GhCOL2 exhibited similar expression rhythms with GhFT, indicating that GhCOL2 is a major regulator of GhFT.

Allotetraploid cotton has two COL2 homoeologs, GhCOL2A and GhCOL2D (Fig. 5c). COL2D is heavily methylated and silenced in G. raimondii that is photoperiod sensitive, while COL2A is hypomethylated and highly expressed in cultivated G. arboreum, which is photoperiod insensitive¹³⁸. In allotetraploid cotton, the COL2A homoeolog was hypermethylated and repressed in cultivated Upland and Pima cottons, while the COL2D homoeolog was highly expressed. This suggests that COL2A in the allotetraploid cottons is likely silenced after polyploid formation since it is expressed in the extant G. arboreum species. The high-level expression of COL2D is likely associated with positive selection and loss of DNA methylation of COL2D during domestication of Upland and Pima cottons¹⁴², which could lead to the loss of photoperiod sensitivity for global cotton cultivation. Indeed, methylation levels of COL2D are lower and their expression levels are higher in cultivated and photoperiod-insensitive G. hirsutum and G. barbadense than in their photoperiod-sensitive wild relatives. Removal of DNA methylation in the wild G. hirsutum (TX2095) seedlings using 5’-aza-2’-deoxycytidine (5-aza-dC), a chemical inhibitor for DNA methylation¹⁴³, activates COL2D expression. Moreover, using virus induced gene silencing (VIGS)¹⁴⁴, GhCOL2 is down-regulated, which is consistent with the repression of GhFT. Down-regulating COL2 and GhFT expression has delayed flowering time for 9 days, compared to the control plants and wild cottons that do not flower in the LD conditions¹³⁸. This example has demonstrated important roles of epigenes and epialleles in photoperiodic flowering during natural selection and crop domestication. These epigenes and epialleles can be candidate targets for gene-editing and biological breeding to improve crop production and resilience.

Growth, development, and reproduction during appropriate times of the year is essential for plants to adapt to seasonal changes. The photoperiodic response to flowering is a key mechanism for plants to optimize utilization of natural resources, adapt crop varieties in regional environments, and significantly impact crop yield stability and potential. Insights gained from studying the interplay between light signaling, the circadian clock, and photoperiodism in the model plant Arabidopsis have facilitated the discovery of complex molecular circuitry networks underlying photoperiodic flowering pathways in both short- and long-day crops, many of which are polyploids and have genomic redundancy that is subject to epigenetic modifications. Indeed, some key regulators, including circadian clock genes, are epigenetically regulated, presumably resulting from natural selection, crop domestication, and/or modern breeding.

Genome sequencing and genome-wide association studies in crop plants have supported that allele variants of several key circadian genes have been selected by breeding and under agricultural culture of adapting crop flowering time to certain seasons. While most current studies are focused on light and circadian control of photoperiodic flowering, temperature changes can drastically affect plant flowering and have been grossly under-investigated. Further investigations are needed to maintain agricultural and ecological sustainability in response to changes in extreme weather and climate such as drought, heat, and flood.

There are good examples of the genes in the circadian clock and photoperiodic flowering pathways that are associated with agronomic traits in crops. Rice Hd1⁷⁷, an ortholog of the Arabidopsis CO, controls expression of two florigen genes Hd3a and RFT1^74–76 during the floral transition. ZmCCT9 in maize is diurnally regulated and negatively regulates the expression of the florigen ZCN8 through a CACTA-like TE insertion in ZmCCT promoter region¹⁴⁵, resulting in reduction of photoperiod sensitivity, thus accelerating maize spread to long-day environments¹⁴⁶. The circadian clock of cultivated tomatoes has slowed during domestication. EMPFINDLICHER IM DUNKELROTEN LICHT1 (EID1) is an F-box protein that targets phytochrome A for degradation in Arabidopsis¹⁴⁷. The EID1 allele in cultivated tomatoes is under selection sweep and enhances plant performance under long-day photoperiods presumably resulting from moving away from the equator¹⁴⁸. Fruit yield in hybrid tomato is related to SINGLE FLOWER TRUSS (SFT)¹⁴⁹, a homolog of FT in Arabidopsis. Moreover, epigenetic alteration of circadian clock genes including CCA1 and LHY in plant hybrids are related to growth vigor in Arabidopsis^119–121 and maize^72,150. In cotton, GhCOL2 is an epiallele, which may be selected during domestication to increase its expression and thereby reduce photoperiod sensitivity, helping spread worldwide cotton cultivation²². These examples will help us design strategies to optimize circadian input and output signals and improve flowering time and crop yield using genome-editing tools. Likewise, discovery and utilization of new epigenes and epialleles in combination of targeted-gene editing will improve flowering time and crop yield and resilience.

Peer review file