|Home | About | Journals | Submit | Contact Us | Français|
The evening complex of ELF4-ELF3-LUX proteins is an integral component of a plant circadian clock. LUX ARRHYTHMO (LUX) is one of the key components of the evening complex, and that play a key role in circadian rhythms and flowering. Here, we report that diverged soybean LUX has the additional role in male reproductive development. We studied diurnal and circadian rhythms of soybean LUX (GmLUXa, GmLUXb, and GmLUXc) using qRT-PCR, and show its nuclear localisation by particle bombardment. Yeast-two hybrid (Y2H) studies indicate that both GmLUXb and GmLUXc form an evening complex with GmELF4b and GmELF3a, respectively. Ectopic expression of GmLUXb in Arabidopsis lux mutants can complement functions of AtLUX, whereas GmLUXc generates novel phenotypes of serrated leaves, stunted plants, shortened anther filament, and low seed set. Overall, our results suggest that the LUX gene has diverged in soybean where GmLUXb and GmLUXc share the role to control flowering time, but GmLUXc has evolved to regulate anther filament growth and seed set by regulating the Gibberellin hormone biosynthesis pathway.
The world’s population is projected to surpass 9 billion by 2050; feeding this enormous population requires a doubling of food production from the same amount of arable land. Moreover, looming climate change threatens our food security, as plant reproductive development like flowering and grain production are inherently sensitive to the changes in the external environment. Unlike animals, plants are sessile; it is important for plants to integrate exogenous signals with endogenous rhythms to ensure the best time for progression to reproductive stage to maximise reproductive success1, 2. The circadian clock system is well known for its role in synchronising environmental signals with endogenous rhythms to ensure proper timing of flowering. Most of our knowledge on circadian clocks and flowering comes from a model plant, Arabidopsis3. Our knowledge of molecular components of the circadian clock system controlling flowering of food legume crops is lacking4. Food legume crops such as soybean, chickpea, faba bean, pea, and lentils are important for food and feed usage around the world. Soybean is a major food crops important for its seed oil and protein content. Being a legume, soybean forms symbioses with Rhizobium bacteria to fix amtmospheric nitrogen. The palaeopolyploid genome of soybean is results of multiple rounds of genome wide duplication5, 6. These duplications of the genome may result in pseudogenization, subfunctionalization, or neofunctionalization of genes. Among the flowering genes characterised in soybean, two of the main photoperiod responsive genes - CONSTANS (CO) and FLOWERING LOCUS T (FT) – display more than one orthologue with the functions in flowering time control. CO regulated by the circadian clock perceives and integrates environmental and endogenous signals to activate florigen (FT), to evoke floral transition. Four putative soybean CO orthologues were found, but only GmCOL5 showed function in flowering time control7. Soybean genome includes 10 FT-like genes consisting of five homeologous pairs (GmFT1-GmFT5)8. Of the 10 FT-like genes, the transcripts levels of GmFT2a and GmFT5a are up-regulated by short-day and were shown to have a strong correlation with flowering, whereas their homeologues GmFT2b and GmFT5b showed no transcript in short-day-grown plants8, 9.
Soybean is well known for the discovery of photoperiodism10. Most commercial soybean cultivars have strong photoperiod requirements and are classified as maturity groups with a restricted latitudinal range. However, molecular basis underpinning maturity grouping remains unknown. In soybean, nine loci (E1 to E8, and J) are associated with photoperiod and maturity. Positional cloning and candidate gene approaches have identified that E1 encodes a B3 superfamily member11, 12, E2 encodes an ortholog of Arabidopsis circadian clock component GI11, 13, and E3 and E4 encode the photoreceptors GmPHYA3 and GmPHYA2, respectively14. Also, there has been an attempt to associate cryptochromes, that is, blue light receptors, to explain the latitudinal distribution of soybean15.
Being sessile, plants sense and adjust to the environment changes by a circadian clock. The circadian clocks of plants have three main components: inputs, circadian oscillators, and outputs. External cues for example changes in light and temperature are conveyed to the circadian oscillator by the input pathway, which links to various output processes. Our understanding of the plant circadian clock is mostly based on studies using the model plant Arabidopsis. The core circadian clock of Arabidopsis comprises genes interacting via multiple transcriptional and post-transcriptional feedback loops to generate rhythmic gene expression. The morning loop consists of two PSEUDO RESPONSE REGULATOR (PRR) 7 and 9, LATE ELONGATED HYPOCOTYL (LHY), CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1), and TIMING OF CAB EXPRESSION 1 (TOC1)) form the core loop16, while the evening loop is a complex of ELF3, ELF4, and LUX17. The circadian clock regulates multiple output pathways including growth, flowering,, photosynthesis, starch metabolism, and disease resistance18. It is well established that the evening complex (ELF3-ELF4-LUX) plays a vital role in maintaining circadian rhythms and synchronising growth and development19. Further, regulated growth controlled by the evening complex occurs by repression of the gene expression and activity of PIF4. LUX suppress its target gene via recruitment of the evening complex to their promoters20. However, it is not clear how the evening complex connects with other pathways to regulate plant development and physiology. Recent studies on Arabidopsis showed that plants have tissue-specific clocks, and evening complex components have tissue-specific expression patterns, but their functional significance is unknown. Moreover, preferential retention of circadian clock components during genome duplication events indicates positive selection21. The conserved function of the evening complex in crop plants suggests its vital role in crop adaptation to various environments. However, the molecular workings of the components of the clock and their wiring in crop plants still remains an open question in plant biology. Hence, molecular understanding of clock function in crop plants will pave the way to manipulate internal clocks for crop productivity in changing climates.
In this study, we characterise a soybean circadian clock gene in the evening complex, LUX ARRHYTHMO, GmLUXb, and GmLUXc and demonstrate conserved protein sequences, gene expression patterns, and subcellular localisation similar to Arabidopsis. The protein-protein analysis shows both GmLUXb and GmLUXc are able to forms an evening complex with GmELF3a and GmELF4b, suggesting that both of them have a role in the circadian clock of soybean. Complementation analysis of the Arabidopsis lux-4 mutant with GmLUXb restores the phenotype back to wild-type, whereas GmLUXc introduces novel phenotypes including serrated leaves, compact and stunted inflorescence, shortened filament, and low seed set, suggesting neofunctionalization of GmLUXc. Further studies on GmLUXc functions suggest that it affects MYBs (MYB21, MYB24) and GA biosynthesis gene expression (GA2ox1, GA20ox2). Treatment with exogenous GA rescues the seed set of GmLUXc transgenic plants. This indicates that the GmLUXc drives anther filament elongation control and seed set by regulating the GA hormone biosynthesis pathway.
Three soybean homologues of LUX ARRHTHMO were identified in a soybean genome sequence database; they are designated as GmLUXa, GmLUXb, and GmLUXc. Phylogenetic analysis (Fig. 1A) showed that these three soybean homologues clustered together and fall into the same clade with other legume (i.e., pea, medicago, and lotus) LUX proteins. Protein alignment revealed conserved amino acid sequences, especially on the DNA-binding MYB domain, except GmLUXa was truncated in the middle of the MYB domain, likely resulting in shortened protein than GmLUXb and GmLUXc (Supplementary Figure 1).
In Arabidopsis, LUX encodes a nuclear localised MYB domain protein with transcription factor activity22, 23. To study GmLUXb and GmLUXc subcellular localisation, we used transient assays where the epidermal cell layer of onion was bombarded with particles coated with DNA (GmLUXb::GFP and GmLUXc::GFP). The fluorescence was located in the nucleus of bombarded epidermal cells, indicating GmLUXb::GFP and GmLUXc::GFP are localised in the nucleus (Fig. 1B).
Tissue-specific expression study showed that GmLUXa was hardly expressed in any tissues examined, whereas highest expression of GmLUXb and GmLUXc were found in floral tissues (Fig. 1C). The expression level of GmLUXc in the flower was almost double of GmLUXb (Fig. 1C). Interestingly, GmLUXc has higher transcript level in all the tissue examined when compared to GmLUXb, especially at root and root tip where GmLUXb expression is low (Fig. 1C). Diurnal and circadian rhythms of GmLUXb and GmLUXc were further examined in soybean (Fig. 1D). Under long-day (LD) and short-day (SD) entrainment, both GmLUXb and GmLUXc transcripts showed clear diurnal rhythms where both genes peaked in the evening in both LD (ZT12-16) and SD (ZT12) conditions (Fig. 1D). Consistently, one evening element (AAAATATCT) was found at the promoter of GmLUXb and GmLUXc, respectively, at 461bp and 480bp upstream of the translational start site (Supplementary Table 1). Interestingly, only GmLUXc has a LUX binding site (LBS) as AtLUX but not in GmLUXb (Supplementary Table 1), which is 321bp upstream of the translational start site for GmLUXc and 545bp upstream of AtLUX. Distinct circadian rhythms were found between GmLUXb and GmLUXc under constant light (LL), showing GmLUXb transcripts gradually increased its level after released into LL condition, whereas GmLUXc displayed damped expression rhythms in LL (Fig. 1D). Previous study showed that LBS at AtLUX promoter was bound by AtLUX in vivo and ChIP assay further confirmed this result20. This binding account for self-regulation of gene expression of AtLUX by negative autoregulatory feedback loop. This may explain the difference of expression in LL between GmLUXb and GmLUXc where GmLUXc is damped after transferred to LL (Fig. 1D).
In Arabidopsis, EARLY FLOWERING 4 (ELF4), EARLY FLOWERING 3 (ELF3), and LUX proteins are known to form an evening complex (EC) that transcriptionally regulates other clock genes and output genes in the evening17. Here, we isolated soybean homologues of ELF3 and ELF4, designated as GmELF4a (Glyma11g35270), GmELF4b (Glyma18g03130), GmELF3a (Glyma04g05280), and GmELF3b (Glyma14g10530) (Supplementary Figures 2 and 3). Supplementary Figure 2 shows that both paralogous soybean pairs of ELF3 and ELF4 are highly conserved in protein sequences with other plant species. We also characterised diurnal and circadian rhythms of soybean ELF3 and ELF4 homologoues (Supplementary Figure 3). GmELF4a peaked at ZT8 in SD and ZT12 in LD, while GmELF4b peaked at ZT12 in SD and ZT12-16 in LD (Supplementary Figure 3A). On the other hand, both GmELF3a and GmELF3b peaked at ZT12 in SD and ZT12-16 in LD (Supplementary Figure 3B). Under the constant condition, like GmLUX, both GmELF4 and GmELF3 paralogous pairs showed different circadian rhythms, where GmELF4a and GmELF3a seemed to damp rapidly to a low level (Supplementary Figure 3). GmELF4b and GmELF3b maintained the transcript level after being subjected to continuous light (Supplementary Figure 3).
We used yeast-two and yeast-three hybrids to examine protein-protein interactions of soybean ELF4, ELF3, and LUX. The yeast-two hybrid showed that both ELF4a and ELF4b interacted with ELF3a and ELF3b (Fig. 2A). Both ELF4a and ELF4b showed self-interaction and interaction with each other, whereas neither of ELF4a or ELF4b interacted with LUXb or LUXc (Fig. 2A). This observation was validated by quantitation of β-galactosidase activities (Fig. 2A). Using the yeast-three hybrid assay in combination with fusion proteins of ELF4a/ELF4b-GAL4-DNA binding domain (BD) and LUXb/LUXc-GAL4-activating domain (AD), we further confirmed that ELF4a/ELF4b could not interact independently with LUXb/c (Fig. 2B). Yeast growth was only observed in selection media (SD-WLM+AbA) in two combinations: (1) ELF4b-ELF3a-LUXb; (2) ELF4b-ELF3a-LUXc (Fig. 2B). These findings suggest two potential combinations of evening complexes in soybean, which include ELF4b and ELF3a and interact with either LUXb or LUXc.
To further explore the functional divergence of GmELF4 and GmELF3, ectopic expression of the GmELF4 and GmELF3 paralogous pair in Arabidopsis elf4-1 and elf3-1 mutants were carried out to check if soybean homologs can complement the function of Arabidopsis genes.
The Arabidopsis elf4-1 (Ws) mutant was sourced from the Feldmann T-DNA insertion mutant population, resulting in plants flowering early in both LD and SD with long hypocotyls and petioles24. Overexpression of GmELF4a or GmELF4b under a CaMV 35S promoter complemented the elf4-1 Arabidopsis mutant under SD conditions, rescuing the early-flowering phenotype of elf4-1 to a flowering time similar to wild-type Ws (Fig. 3A,B). Also, the hypocotyl and petioles lengths of plants overexpressing GmELF4a or GmELF4b were greatly reduced when compared to the elf4-1 mutant (Fig. 3A). These results suggest that GmELF4a and GmELF4b can complement ELF4 functions in Arabidopsis.
The Arabidopsis elf3-1 (Col) mutant was originally isolated as a photoperiod-insensitive early-flowering mutant also showed a long-hypocotyl phenotype25. Both over-expressions of GmELF3a or GmELF3b under the CaMV 35S promoter do not fully complement the elf3-1 Arabidopsis mutant under SD conditions, resulting in transgenic plants with flowering time between elf3-1 mutant and wild-type COL and closer to elf3-1 mutant (Fig. 3C,D). However, both overexpression transgenic plants of GmELF3a or GmELF3b rescued the long-hypocotyl phenotype of the elf3-1 mutant (Fig. 3C). These results point out that neither overexpressing of GmELF3a nor GmELF3b can complement the loss-of-function in the elf3-1 mutant, suggesting functional divergence of soybean and Arabidopsis ELF3 and the complexity of the circadian clock system in the paleopolyploid species soybean.
To study if GmLUXb and/or GmLUXc carry out the same functions as Arabidopsis LUX in flowering time control, a complementation analysis in the lux-4 mutant was performed. The Arabidopsis lux-4 mutant carries a nonsense mutation and produces a protein truncated right at the start of the MYB domain, generating in plants that flower early in both LD and SD with long hypocotyls and petioles22, 23. In Arabidopsis, overexpression of AtLUX in the lux-4 Arabidopsis mutant restored the circadian rhythms and hypocotyl growth20. Overexpression of GmLUXb under the CaMV 35S promoter complemented the lux-4 Arabidopsis mutant under LD (Fig. 4A) and SD (Fig. 4B) conditions, rescuing the early-flowering phenotype of lux-4 to a flowering time similar to wild-type COL (Fig. 4F). In addition, the petioles lengths of plants overexpressing GmLUXb were greatly reduced when compared to the lux-4 mutant (Fig. 4C), and weak inflorescence stem of the lux-4 mutant were not found in plants overexpressing GmLUXb (Fig. 4D). Similar phenotypes were observed for overexpressing GmLUXb in COL wild-type plants (Supplementary Figure 4).
In contrast to the overexpression of GmLUXb, overexpression of GmLUXc in both lux-4 mutants and COL wild-type delayed the early-flowering phenotype of lux-4 but also introduced a novel phenotype that was not seen in either lux-4 or lux-4 35S::GmLUXb plants (Fig. 4, Supplementary Figure 4). The first observation is serrated rosette and cauline leaves in both LD and SD-grown plants (Fig. 4A–C, Supplementary Figure 4). The transgenic plants also produced compact and stunted plants (Fig. 4D) where this phenotype was more severe in the LD condition (Fig. 5A). The T1 lux-4 35S::GmLUXc plants were sterile and developed short siliques with aborted seeds (Fig. 5B). Hand pollination of lux-4 35S::GmLUXc carpels with lux-4 pollens yielded viable seed but not vice versa, suggesting the female fertility was normal. Observation of anthers on fully opened flowers showed that plants overexpressing GmLUXc have shorter filament length, but anthers dehisce and release pollens normally (Fig. 4E). A detailed analysis of flower development stages as assigned by Smyth et al.26 was performed (Fig. 5C). Figure 5C shows that filaments of plants overexpressing GmLUXc never elongated to the same level, as the stigma and stigmatic papillae did not mature as seen in lux-4 plants. The phenotype of shortened filaments in lux-4 35S::GmLUXc plants could explain why self-pollination resulted in short siliques with low to nil seed set but could not account for reduced seed set by hand pollinating 35S::GmLUXc pollen to lux-4 carpels.
To further examine male-sterility of overexpression of 35S::GmLUXc plants, a pollen viability test was performed where all mature pollen from WT and transgenic plants were found viable containing three nuclei, and pollen from WT and transgenic plants looked similar (Fig. 5D). To examine any developmental abnormalities during microsporogenesis, histological analyses on different pollen development stages were carried out. At each stage, pollen development of transgenic plants was indistinguishable from those of wild-type (Fig. 6A,C). Taken together, overexpression of GmLUXc does not affect pollen viability or pollen development. We further investigated pollen germination in vivo, where self-pollination and cross-pollination of wild-type and overexpression of 35S::GmLUXc pistils were performed. Analysis of Aniline blue stained pollinated pistils revealed that when 35S::GmLUXc pistils were pollinated with WT pollen, the pollen tubes growth in pistils was similar to the self-pollinated WT (Fig. 6B). Inversely, pollination of WT pistils with 35S::GmLUXc pollen grains revealed that the pollen from transgenic plants could not attach effectively to the surfaces of wild-type stigmas and form pollen tubes (Fig. 6B). A similar phenotype was also observed in self-pollinated 35S::GmLUXc pistils (Fig. 6B). These findings further confirmed that the low seed set of the 35S::GmLUXc plant is due to male but not female fertility. This in vivo pollination and pollen tube germination assay also suggested that 35S::GmLUXc pollens have difficulty attaching pollen to stigma. We further examined pollen attachment to stigma by evaluating pollen adhesion and pollen retention on the stigma. For pollen attachment, a mature stigma of an emasculated flower was lightly touched three times by freshly dehisced anther, and the pollen grains attached to the stigma surface were observed using a microscope. The pollen of 35S::GmLUXc transgenic plants failed to attach to the stigma when compared to those from wild-type plants (Fig. 5E, top panel). Pollen retention was also evaluated by washing gently pollinated pistils in sodium phosphate buffer (50mM, pH 7.0) containing 1% Tween 20. The wild-type and 35S::GmLUXc pollen retention on stigma was found to be similar (Fig. 5E, bottom panel). From these experiments we concluded that the pollen grains of overexpression of 35S::GmLUXc plants are defective in pollen adhesion. To examine whether the defect in pollen adhesion is due to any changes in the morphology of pollen exine, we performed scanning EM of pollen. Scanning EM of pollen exine of overexpression 35S::GmLUXc and WT were found to be indistinguishable (Fig. 6D).
To further investigate how GmLUXc causes the novel phenotypes, MYBs (MYB21, MYB24, MYB57) genes that are known to affect anther filament length27 and genes involved in the jasmonate (GTR1, PDF1, DAD1, OPR3), gibberellin (GA2ox1, GA20ox2), auxin (RGA, GAI) signalling pathways, were examined in inflorescences and whole 35S::GmLUXc transgenic plants (Fig. 7A, Supplementary Figure 5). MYB21, MYB24, MYB57 function redundantly in Arabidopsis to control stamen filament elongation and mutation in these genes caused reduced filament length in anther27. Figure 7A shows that MYB21 and MYB24 transcripts were down-regulated in lux-4 35S::GmLUXc plants. The GA biosynthesis genes GA2ox1 and GA20ox2 showed the presence of splice variants in the lux-4 35S::GmLUXc plants (Fig. 7A). Cis-element analysis of the promoter regions of GmLUXb and GmLUXc revealed higher numbers of GA pyrimidine box ((C/T)CTTTT(C/T)) in GmLUXc (10 elements) as compared to GmLUXb (4 elements) and AtLUX (7 elements) (Supplementary Table 1). These results suggested that GmLUXc might regulate those downstream pathways via interacting with the GA signalling pathway. To confirm this hypothesis, exogenous GA3 and GA4 (dissolved in 0.02% ethanol) were sprayed onto lux-4 35S::GmLUXc plants and compared to the lux-4 mutant and Col wild-type. lux-4 35S::GmLUXc plants were found to be rescued by GA treatment where plants elongated normally and formed normal inflorescence compared to stunted and compact inflorescence in lux-4 35S::GmLUXc plants (Fig. 7B). Examination of anthers in fully opened flowers revealed that the anther length of lux-4 35S::GmLUXc plants were rescued by GA treatment (Fig. 7C). Siliques were found to be more normal in size with some properly developed seeds (Fig. 7D). GA4 was able to rescue the seed set better than GA3 treatment, as indicated by the number and seed weight (Fig. (Fig.7D).7D). Almost double the amount of the seeds per silique and total seed weight was obtained from the plants sprayed with GA4. Resulting seeds were found to be 100% viable.
The circadian clock is a crucial regulator of different biological functions such as enhances adaptation, fitness, and survival of plants28–30. In Arabidopsis, the circadian clock system consists of three main interlocking loops: morning, central, and evening loops. The evening complex, a tripartite protein complex of ELF4-ELF3-LUX, is an integral to the plant circadian clock for maintaining circadian rhythms17, 31. The evening complex component seems to be positively selected during genome duplication events21. Soybean, a palaeopolyploid, has had its genome duplicated several times. The genome sequence of soybean indicated the presence of three soybean homologues of LUX. Among the three GmLUX genes, GmLUXa transcripts were undetectable in all of the tissues examined in this study. Further, sequence analysis indicated that GmLUXa is truncated in the middle of the MYB DNA-binding domain, resulting in loss-of-function, suggesting GmLUXa is a pseudogene. On the other hand, both GmLUXb and GmLUXc transcripts were detected in all tissues examined with a high expression in the flower. Although our study clearly shows that GmLUXb and GmLUXc transcripts are entrained by LD/SD and constant light to display diurnal and circadian rhythms, there is a clear difference in the patterns of circadian rhythms of GmLUXb and GmLUXc soon after the release of entrained plants into the constant light. Under constant light, GmLUXb mRNA was increased gradually, whereas GmLUXc mRNA was damped 3-fold of the expression level when compared with the peak under entrained condition, suggesting these might have an additional role at light gating or light sensing. This is the first indication that the three homologues of LUX might function differently in soybean where GmLUXa becomes a pseudogene, and GmLUXb and GmLUXc paralogues could undergo subfunctionalization.
In Arabidopsis, the ELF4-ELF3-LUX evening complex, ELF4 and ELF3 encode for novel nuclear proteins, and ELF3 serves as a bridge for the interaction of ELF4 and LUX, while LUX suppress gene expression by recuiting the evening complex to promoters of the target genes20 to regulate plant growth17, 31. Further, transcripts expression patterns of ELF4, ELF3, LUX are similar, peaking in the evening. To study whether this complex is also conserved in soybean, soybean homologues of ELF4 and ELF3 were isolated and characterised. Transcripts of all the four soybean genes—GmELF4a, GmELF4b, GmELF3a, and GmELF3b—showed diurnal rhythms, peaking around the same time as GmLUXb and GmLUXc in the evening under both LD and SD (Fig. 1C). Even orthologs of ELF3, ELF4, and LUX, were found in other species, soybean is the first species other than Arabidopsis shown to form evening complex. Further, our study showed that there are two possible interactions between the three paralogous pairs—ELF4b-ELF3a-LUXb and ELF4b-ELF3a-LUXc—suggesting two putative evening complexes might exist in soybean. In Arabidopsis, only one evening complex was present, which controls both flowering and hypocotyl growth; however, in soybean, ELF4b-ELF3a-LUXb seems only to control flowering in SD (Fig. 8), while ELF4b-ELF3a-LUXc not only controls flowering but also filament length and pollen adhesion (Fig. 8). Based on the different phenotypes of plants overexpressing GmLUXb and GmLUXc, it is highly likely that these two complexes regulate different downstream targets. Above all, our study unravels novel functions of GmLUXc in plant development such as control of leaf serration, internode elongation, anther filaments elongation, and pollen adhesion.
The LUX gene has been the focus of recent studies on circadian clocks and photoperiod responsiveness32–34. In garden peas, one of the earliest discovered photoperiod responsiveness gene, STERILE NODE (SN), was recently identified as a homologue of LUX by a candidate gene approach35. SN was found to play role in both the circadian clock and photoperiodic flowering control35. A lotus homologue of LUX has also been shown to display circadian rhythm under continuous light36. In cereals, homologues of LUX have been identified as a candidate gene for the early-flowering mutant in barley (early maturity 10, eam10) and wheat (earlines per se 3, eps-3A) through fine-mapping32–34. Both mutants are early-flowering, photoperiod insensitive, and have distorted diurnal and circadian rhythms for clock genes. Interesting, LUX genes in both cereal plants interact with another major determinant of photoperiod response, Ppd-H1/Ppd gene, to regulate florigen FT (Flowering Locus T) expression and regulate flowering time.
Here, we identified GmLUXb as the orthologue for Arabidopsis LUX, as the Atlux-4 mutant can be complemented by overexpression of GmLUXb to wild-type phenotype with delayed flowering in both LD and SD and reduced petiole length. Intriguingly, GmLUXc was found to be neofunctionalised where ectopic expression of GmLUXc in Arabidopsis causes novel phenotypes of serrated leaves, compact and stunted plants, shortened filament, reduced pollen adhesion, and low seed set. These phenotypes are much more severe under LD where transgenic plants are very small and hardly produce any siliques containing viable seeds followed by early abscission of petals and anthers. Moreover, these phenotypes are correlated with high expression levels of GmLUXc in the flower. In garden peas, SN showed interaction with another clock gene, GIGANTEA (GI)/LATE FLOWERING (LATE1), in regulating flowering node and a total number of reproductive nodes37, while wheat LUX (HvLUX1) sequences are highly correlated with wheat accessions from warmer climates33. The results from these studies suggest that molecular functions of the LUX gene might have diverged and resulted in novel functions during evolution and domestication of crop plants.
Overexpression of GmLUXc produced interesting phenotypes specifically on male reproduction – that is, anther filament elongation and pollen adhesion. This is the first report of circadian clock gene involvement in male reproductive development. Our results showed shortened anther filament length and reduced pollen adhesion in overexpressed 35S::GmLUXc plants. These findings are consistent with the strong expression of GmLUXc in flowers. Moreover, GmLUXc expression in mature soybean pollen has been reported previously using soybean Genechip38. The underlying mechanism of pollen adhesion has been less studied in the literature. Temperature and the pollen exine coating appeared to be the two major main causes for defective pollen adhesion39–42. High-temperature treatment in rice, cherry, and also peach was found to reduce the number of pollen grains that can adhere to stigma40–42. A genetic screen for pollen-stigma adhesion mutant in Arabidopsis has successfully identified several lap (less adherent pollen) mutants that have a defective exine layer, suggesting exine is necessary for pollen adhesion39. The circadian clock can be entrained by external cues; in addition to light, the temperature is a major input cue43, 44. The morning clock component, PSEUDORESPONSE REGULATOR 7 (PRR7) and PRR9, have been shown to be important for temperature entrainment in Arabidopsis, and the double mutant has impaired sensitivity to temperature signals45. Furthermore, a bHLH transcription factor PHYTOCHROME INTERACTING FACTOR4 (PIF4) has been shown to regulate hypocotyl growth and thermal activation of flowering17, 46. Function of the EC is essential for regulated expression of PIF4, since it represses PIF4 expression early in the night17. Based on these studies on Arabidopsis, one might speculate that overexpression of GmLUXc might have affected the temperature entrainment of the circadian clock or exine development of pollen, which indirectly affects the pollen-stigma adhesion process and produces a male sterile phenotype. A close observation of the pollen exine of GmLUXc and WT plants by scanning EM fail to show any differences.
LUX belongs to an MYB-related GARP family, and its role in anther development has not been reported, though functions of some MYB transcription factors have been reported in anther development including MYB80/103 in pollen and tapetum development47, 48 MYB21, 24 in filament elongation49 and MYB26 in anther dehiscence50. Further, transcriptome analyses of whole Arabidopsis plants have revealed circadian clock regulation of many plant hormone biosynthesis, signalling, and responsive genes51–53. In particular, abscisic acid, gibberellin, auxin, and jasmonate have been shown to be targets of circadian outputs54–56. Reciprocally, hormones including abscisic acid and cytokinin can act as an input for the circadian clock in Arabidopsis57. Moreover, phytohormones including auxin, gibberellin, and jasmonate have been shown to regulate stamen filament elongation. Short filaments were noticed in mutants defective in auxin biosynthetic genes (YUC2, YUC6), auxin response factor (ARF6, ARF8), and auxin transport (MDR1, PGP1)58. On the other hand, gibberellin and jasmonate-deficient mutants, ga1-3 and opr3, are both male sterile owing arrested filament elongation, delayed anther dehiscence, and reduced pollen viability59–61. Data on Arabidopsis also showed that gibberellin promotes jasmonate biosynthesis influencing MYB21, MYB24, and MYB57 expression to control stamen development62, 63.
Accordingly, we further investigated whether overexpression of the GmLUXc phenotype is due to perturbation of hormones biosynthetic pathways. Indeed, our study found decreased transcript levels of MYB21 and MYB24 and gibberellin biosynthesis genes (GA2ox1, GA20ox2) in 35S::GmLUXc transgenic plants, but no change on jasmonate and auxin biosynthesis and signalling genes were observed (Supplementary Figure 5). Application of exogenous GA rescued the seed set in 35S::GmLUXc transgenic plants, suggesting GA could be a downstream target of GmLUXc.
It is now well established that circadian clock gene alleles have played a vital role in the domestication of crop plants by affecting key agricultural traits such as flowering time and yield62. Recent research also emphasized the role of the circadian clock in abiotic and biotic stresses responses. Examples of clock genes that have contributed towards adaptation of crop plants include Pseudo-Response REGULATORS, GIGANTEA, and evening complex genes ELF3, ELF4, and LUX. Preuss et al., reported an increase in soybean crop productivity by modulation the expression of morning clock genes64. These investigators used AtBBX32, B-box domain gene from Arabidopsis to change the expression of morning clock genes (GmTOC1 and GmLCL2) transcripts to increase soybean grain yield64. Further, it was proposed that the increase in circadian clock genes expression during the transition from dark to light is critical for the reproductive development of soybean64.
Given the genomic context of crop plants such as soybean, it is possible that the evening complex genes can regulate multiple outputs as reported in our present study. Thus, an understanding of how clock genes function in crop plants has the potential to uncover new genetic targets for breeding crops that are resilient to climate change.
All plants were grown in a growth cabinet with controlled environments and received nutrient solution weekly. Arabidopsis were grown in a 1:3 mixture of perlite and potting mix at 20°C under either long-day (LD; 16h light, 8h dark) or short-day (SD; 10h light, 14h dark) conditions under 150µmolm−2s−1 white light from cool-white fluorescent tubes. Soybean plants (Glycine max L. Merr. cv. Bragg) were grown in potting mix topped with seed raising mix (Debco) at 24°C under LD or SD condition at 24°C, 400µmolm−2s−1 and 70% humidity. Flowering time was measured as a number of days from the day of sowing to the day when the petals of the first flower were visible and total number of rosette leaves and cauline leaves at bolting. Total seed weight (mg) was measured by weighting the harvested seeds of 10 plants, and by counting the number of seeds yielded by ten siliques we estimate a mean number of seeds per silique.
Protein sequences of different plant species were retrieved from NCBI (https://www.ncbi.nlm.nih.gov/) and soybean genome database at Phytozome (https://www.phytozome.net/). Sequences were aligned in the multiple sequences alignment tool CLUSTALX2.0, using Gonnet Protein Weight Matrix with default parameters. The phylogenetic tree was constructed by the Neighbour-Joining algorithm using the MEGA 5.0. The bootstrap consensus tree was inferred from 1000 replicates.
The expression data of Glycine max genes were extracted either from Soybean electronic fluorescent pictograph (eFP) browser at the Bio-Array Resource (BAR) (http://bar.utoronto.ca/efpsoybean/cgi-bin/efpWeb.cgi)65. All diurnal and circadian rhythms experiments were conducted under LD or SD or constant light (LL; 12h light:12h dark) for three weeks conditions. The soybean seedlings were three weeks old at harvest and samples were collected at four h intervals across a 24h or 72h period. Three biological replicates were obtained, and each replicate consists of two unifoliate leaves randomly pooled together from two different plants. Total RNAs were then extracted using SV Total RNA Isolation System (Promega) according to manufacturer’s protocol. Reverse transcription was conducted with one µg of total RNA using SuperScript™ III Reverse Transcriptase (Invitrogen) according to manufacturer’s instructions. Quantitative real-time RT-PCR (qRT-PCR) was performed with 1.5µL cDNA template in a ten µL reaction volume using Brilliant III Ultra-Fast SYBR® Green QPCR Master Mix (Agilent, USA) with the Stratagene Mx3000P™ System (Agilent, USA). Two technical replicates were performed for each sample. Standards, a no-template control, and an RT-negative sample were included in each run. Primers used in RT-PCR can be found in Supplementary Table 2. RT-PCR was performed with two µL cDNA template in a 25µL reaction volume using Taq DNA Polymerase, recombinant (Invitrogen). Three biological replicates were performed and analysed on 1.5% agarose gel. Primers for gene expression studies can be found at Supplementary Table 2.
Cells in epidermal layers of onion bulbs were transformed with pUC18/2x35S::GFP::GmLUXb, 2x35S::GFP::GmLUXc or 2x35S::GFP by particle bombardment. The full-length GmLUXb and GmLUXc cDNA fragments were amplified using XbaI-LUXb-F and LUX-KpnI-R primers (Supplementary Table 2). GFP fragment was amplified by BamHI-GFP-F and GFP-NS-XbaI-R primers (Supplementary Table 2). Both fragments were cloned into pGEMT-easy (Promega) and digested with XbaI and SpeI and re-ligated. After digestion with BamHI and KpnI, the digested fragment was cloned into pUC18 containing 2xCaMV35S promoter and Nos terminator. Bombardments were performed using the Biolistic PDS-1000/He Particle Delivery System (BioRad). Fluorescent signals were recorded with an Olympus BX60 fluorescence microscope equipped with an Olympus DP70CCD camera system. The same cells were stained with 1µgml−1 of 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) in PBS and fluorescence was recorded similarly.
Arabidopsis elf3-1, elf4-1, and lux-4 or Columbia plants were transformed with Agrobacteria harbouring the binary vector pMLBART/35S::GmLUXb or 35S::GmLUXc using the floral dip method. The full-length soybean ELF3, ELF4, or LUX cDNA fragments were amplified specific restriction enzyme overhang primers (Supplementary Table 2). The resulting PCR products were cloned into pRT101 containing CaMV35S promoter and poly-A terminator. NotI restriction sites were introduced by CaMV35S-NotI-F and polyA-NotI-R primers (Supplementary Table 2) and the PCR fragments were cloned into pGEMT easy (Promega) by TA cloning and were subsequently inserted into the binary vector of pMLBART followed by NotI digestion. Plants treated with GA were sprayed with 50µM GA3 or GA4 (Sigma, USA) every two days for two weeks started when plants have 3cM inflorescence.
Pollen viability test was performed using mature pollen grains or the anthers with Alexander’s stain or double staining with 0.5µgml−1 fluorescein diacetate (FDA) and oneµgml−1 propidium iodide (PI) (Regan and Moffatt, 1990). To visualise nuclei of mature pollen grains, open flower were placed to 500uL of DAPI staining solution (0.1M sodium phosphate. pH7.5, 1mM EDTA, 0.1% v/v Triton X-100) and vortexed briefly to release the pollen. The tube was centrifuged and resuspended in 40µL of DAPI staining solution containing oneµgml−1 DAPI (Park et al., 1998). Aniline blue staining of pollen tube was carried out according to Kho and Baer (1968)66. Pistils were cut longitudinally 12h postpollination and fixed in 3:1 ethanol: acetic acid for two h at room temperature, washed with distilled water and softened overnight in NaOH solution (8M). The pistils were washed in distilled water for one hour in the following day for three times. It was followed by staining in aniline blue solution (0.1% aniline blue in 0.1M K2HPO4-KOH buffer, pH 11) for three hours in complete darkness. The stained pistils were placed in a drop of glycerol and carefully squashed under a cover slip and observed with the fluorescence microscope.
To obtain cross and longitudinal section of developing anthers and flowers, wild-type and transgenic inflorescence were fixed in fixed in 4% w/v paraformaldehyde (Sigma, Australia), 4% w/v DMSO (Sigma, Australia) in phosphate-buffered saline (PBS) with vacuum infiltration. After overnight fixation at 4°C, samples were subjected to a serial of ethanol dehydration and stained with 0.1% w/v Eosin Y. The samples were then subjected to Histoclear (Sigma, Australia) and embedded in paraplast (Sigma, Australia). Embedded tissues were cut at eight-micrometer sections and stained with toluidine blue. Sections were observed with the light microscope.
To determine pollen adhesion, two assays were carried out based on protocol reported previously67. The two assays are used to determine pollen attachment and pollen retention. For the attachment assay, out of the six stamens of fully opened flower, one of the four longer stamens was isolated. The anther of this stamen was used to lightly touch a mature stigma of an emasculated flower for three times. For the retention assay, the same procedures of the attachment assay were carried out. 30minutes after pollination, the pistil was cut and washed gently in sodium phosphate buffer (50mM, pH 7.0) containing 1% Tween 20. The stigma with style was then removed from the pistil, mounted on a slide with 80% glycerol for microscope observation.
Yeast two-hybrid was performed using Matchmaker GAL4 yeast two-hybrid system (Clontech, USA) according to manufacturer’s instructions. Constructs were generated by cloning full-length cDNA sequences into the multiple cloning sites of pGBKT7 and PGADT7 plasmids using relevant primers (Supplementary Table 2). Baits were expressed as Gal4 DNA-BD fusion proteins in the pGBKT7 plasmid; preys were expressed as Gal4 AD fusion proteins in pGADT7 vector and both baits and preys were co-transformed into the AH109 yeast strain. Co-transformants were selected on SD-WL (synthetic drop-out media lacking the tryptophan and leucine). Interactions were tested on selective media lacking tryptophan, leucine, and histidine (SD-LWH+x α-gal). Serial 1:10 dilutions were prepared in water, and five μl of each dilution was used to yield one spot. Plates were incubated at 30°C for three days before scoring and taking photographs. β-galactosidase activity was assayed for LacZ reporter using o-Nitrophenyl-β-galactoside (ONPG) as the substrate in liquid culture according to manufacturer’s instructions (Clontech). This assay was repeated three times. For yeast three-hybrid assay, pBridge plasmid was used instead of pGBKT7 in the Matchmaker Gold Yeast Two-hybrid system (Clontech) where the third protein is expressed in the absence of methionine (M). Both baits and preys were co-transformed into the Y2Hgold yeast strain. Co‐transformants were selected on SD‐WL. Interactions were tested on SD-WL+AbA (Aureobasidin A) and SD-WLM+AbA.
At: Arabidopsis thaliana AtLUX (NP_001190022), AtBOA (NP_200765) Ps: Pisum sativum PsLUX(KJ801796) Mt: Medicago truncatula MtLUXlike (Medtr4g064730), Cr: Chlamydomonas reinhardtii He: Helianthus exilis Ee: Euphorbia esula Gm: Glycine max GmLUXa (Glyma01g36730), GmLUXb (Glyma12g06406), GmLUXc (Glyma11g14490), Le: Lycopersicon esculentum, Lj: Lotus japonicas LjLUX (chr3.CM0792.250.r2.d), Md: Malus domestica Mc: Mesembryanthemum crystallinum Os: Oryza sativa Ppp: Physcomitrella patens Pg: Picea glauca Pta: Pinus taeda Pt: Populus trichocarpa Pp: Prunus persica Rc: Ricinus communis RcLUXlike (XP_002520534.1) St: Solanum tuberosum Tp: Trifolium pratense Vv: Vitis viniferaa.
We thank Steve Kay (University of California, San Diego) for providing Arabidopsis lux-4 mutants seeds. Financial support for this work from the Australian Research Council in the form of the ARC Centre of Excellence for Integrative Legume Research (CE0348212) ARC Discovery grants ARC DP0988972 is gratefully acknowledged. This work was also supported by McKenzie Postdoctoral Fellowship (LCL) by the University of Melbourne.
P.L.B. & M.B.S. plan the experiments; L.C.L. conducted the experiments; L.C.L., M.B.S. & P.L.B., wrote the manuscript.
The authors declare that they have no competing interests.
Electronic supplementary material
Supplementary information accompanies this paper at doi:10.1038/s41598-017-10823-y
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.