Soybean ( Glycine max ) is a vital legume with multipurpose functionality as human food, animal feed, and biodiesel (Liu et al., 2020) . Soybeans synchronize the timing of reproductive transition (flowering) to a seasonal change in daylength (photoperiod), from long day (LD) to short day (SD) during the summer solstice (Cao et al., 2017; Lin et al., 2021) . Therefore, photoperiodism has a vast influence on flowering control in the SD-plant soybean. However, the molecular basis of the photoperiodic flowering response is poorly understood in soybean.

In the long-day flowering plant Arabidopsis thaliana (Arabidopsis), photoperiod-regulated flowering is mediated through the B-box transcription factor CONSTANS (CO) where endogenous rhythms of the circadian clock and external seasonal light/dark lengths coincide (Takagi et al., 2023; Turck et al., 2008) . Among the large number of genes participating in the transcriptional and post-transcriptional regulation of CO, the CO mRNA transcription is in part controlled by a component of the circadian clock, GIGANTEA (GI) (Sawa et al., 2007) , as well as by FLAVIN-BINDING KELCH REPEAT, F-BOX 1 (FKF1) and CYCLING DOF FACTOR (CDF) proteins that are controlled by the circadian clock (Imaizumi et al., 2005; Sawa et al., 2007; Song et al., 2012) . Under blue light in LD conditions, GI and FKF1 proteins form a complex in the late afternoon and degrade CDF1 proteins, a family of Dof-type zinc finger transcription factors that repress CO transcription (Fornara et al., 2009; Imaizumi et al., 2005) , resulting in the activation of CO transcription towards the end of the day in LD. In addition, PSEUDO RESPONSE REGULATOR (PRR) proteins, components of the circadian clock’s central oscillator, are known to stabilize CO proteins at different times of the day (Hayama et al., 2017) . PRR9 is suggested to contribute to the morning peak of CO accumulation, whereas PRR5, PRR7, and TIMING OF CAB2 EXPRESSION 1 (TOC1) stabilize CO in late afternoon in LD conditions. The CO peak exposed to light at the end of the day promotes flowering through the activation of the floral integrator gene FT (Song et al., 2012; Valverde et al., 2004) . At night, EARLY FLOWERING 3 (ELF3), a component of the evening complex of the circadian clock, promotes degradation of CO proteins (Song et al., 2018) likely by making a complex with the E3 ubiquitin ligase CONSTITUTIVE PHOTOMORPHOGENESIS 1 (COP1) (Yu et al., 2008) . In a sharp contrast to the extensively studied, elaborate regulatory mechanisms of CO at the transcriptional and post-transcriptional levels, CO’s regulatory roles in the control of the flowering gene network other than its well-known target, FT , are largely unclear.

In the short-day flowering plant soybean, the CO-FT module in photoperiodic flowering regulation may be conserved, but with certain modifications (Cao et al., 2015; Wu et al., 2019) . The CO homologs CONSTANS-LIKE 1a ( GmCOL1a ) and the CONSTANS-LIKE 1b ( GmCOL1b ) appear to act as flowering repressors under LD conditions rather than inducers, by downregulating the FT homologs GmFT2a and GmFT5a that activate flowering under SD conditions, resembling the action of the CO-FT module in rice, another short-day flowering crop (Komiya et al., 2008; Yano et al., 2000) . However, further regulatory roles of GmCOL1a / GmCOL1b are yet to be discovered.

Similar to Arabidopsis, recent studies have reported that soybean’s circadian clock influences photoperiodic flowering by modulating the E1 gene, a legume specific flowering repressor (Xia et al., 2012) in twofold. Firstly, the soybean LHY homolog LHY1a directly binds to the promoter of E1 and inhibits E1 expression (Lu et al., 2020) . Secondly, components of the evening complex, GmELF3 and GmLUX homologs, directly inhibit E1 expression (Bu et al., 2021; Lu et al., 2017) . Regulatory interactions of GmCOL genes and the circadian clock genes, however, are unknown.

To clarify the regulatory roles of the CO gene family in photoperiodic regulation of flowering in soybean, we sought for genes that were controlled by GmCOL1b using a network inference approach followed by experimental verifications. The purpose of this study is to gain a clearer picture on the function of GmCOL1 genes in the flowering gene network that controls soybean’s SD flowering habit by identifying and characterizing genes that may be regulated by GmCOL1b.

We inferred regulatory interactions among soybean’s circadian clock and flowering genes using in-house time series RNA-seq data and the network inference package CausNet (Alcantara et al., 2022; Wu et al., 2019) . Inferred regulatory interactions were visualized on Cytoscape 3.9.1 (Shannon et al., 2003) . We identified a small number of candidate genes downstream of GmCOL1b that carried strong reliability weights: LHY/CCA1-LIKE 1 ( GmLCL1 ) , and LHY/CCA1-LIKE 4 ( GmLCL4 ), and TEMPRANILLO 1a ( GmTEM1a ). These candidate genes were predicted to be upregulated by GmCOL1b at 25 ^o C in LD with respective confidence weights of 0.79, 0.6, and 0.5.

To experimentally verify these predicted regulatory interactions, GmCOL1b was expressed in soybean protoplasts and mRNA levels of predicted downstream genes were measured against a non-transfected cell line by RT-qPCR. Protoplasts were isolated and transfected with the 35S:EGFP-GmCOL1b plasmid DNA (Table 2) as described previously (Wu & Hanzawa, 2018) with minor modifications. Transfected protoplasts were harvested after 24 hours at Zeitgeber time points ZT4-ZT24 with 4-hour intervals. Protoplast viability and GFP expression were examined prior to harvest ( Figure 1B ). Approximately 90% of cells were viable, while about 70% of transfected cells were expressing GFP signals, indicating successful transient expression of GFP-GmCOL1b. GFP-GmCOL1b mRNA expression was upregulated by a 50,000-100,000-fold compared to the non-transfected cell line across a 24-hour period. Overexpression of GFP-GmCOL1b upregulated mRNA expression of GmLCL1, GmLCL4, and GmTEM1a at differing time points . GmLCL1 exhibited peak expression at ZT8 with a 600-fold increase compared to the negative control, and GmLCL4 and GmTEM1a showed peak expression at ZT12 with a 100-fold and a 400-fold increase compared to the control, respectively.

These results shown above demonstrate that GmCOL1b controls GmLCL1 , GmLCL4 , and GmTEM1a. Arabidopsis LHY and CCA1 genes encode closely related MYB transcription factors and have a peak expression at dawn (Alabadí et al., 2002; Mizoguchi et al., 2002). In the daytime, LHY/CCA1 are repressed by PRR genes carrying a CONSTANTS, CONSTANS-like and TOC1 (CCT) DNA-binding domain that directly binds to LHY/CCA1 ’s promoters, creating an essential core oscillator of the circadian clock (Nakamichi et al., 2010; Nakamichi et al., 2012) . Similarly, in soybean, GmPRR3 genes are shown to directly repress LHY/CCA1 homologs via their CCT domain (Li et al., 2020; Lu et al., 2020) . CO proteins also contain a CCT domain in their C-terminus (Valverde, 2011; Wu et al., 2014) , thus it is possible that GmCOL1b directly activates GmLCL1 and GmLCL4 through its CCT domain. Our observation may be indicative of a feedback regulation between the circadian clock and the flowering gene network via GmCOL genes.

Arabidopsis TEMRANILLO 1 (TEM1) and TEM2 belong to the RELATED TO ABI3/VP1 (RAV) subfamily in the AP2 transcription factor family (Castillejo & Pelaz, 2008; Hu et al., 2021) . TEM1 and TEM2 are transcriptional suppressors of FT and delay flowering by antagonizing CO, the activator of FT (Castillejo & Pelaz, 2008) . In soybean, the TEM homolog’s role in photoperiodic flowering appears to be conserved. GmTEM1b directly interacts with the promoter of the flowering inducer GmFT5a and represses its expression, resulting in delayed flowering (Wang et al., 2021) . Our observation may indicate that GmCOL1b may delay flowering through upregulation of GmTEM genes.

Our results identified novel target genes of CO in soybean’s photoperiodic flowering and indicated a synergistic function in controlling the circadian clock and flowering transition. Limitations of this study include experimental verification of direct regulation of GmCOL1b on selected candidate genes. However, we cannot exclude indirect regulatory effects on these target genes from unknown factors in the flowering gene network. Nonetheless, this data provides a strategy to further clarify the regulatory roles of CO and the gene regulatory networks controlling the circadian rhythms and flowering transition in crops.

Plant Growth Condition and Sampling

Williams 82 (PI 518671) seeds were provided by the USDA Soybean Germplasm Collection. Seeds were sown in 6-inch pots containing Sunshine Mix #4 Professional Growing mix with Mycorrhizae and Vermiculite in a 6:1 ratio. The pots were then placed in a growth chamber with 14-hour light exposure per day at 30 °C. The pH and moisture conditions of the soil in each pot were regularly monitored, ensuring the soil moisture range was kept within 40%-50% and the pH at around 7. Fully expanded unifoliate leaves were used for protoplast isolation.

Plasmid Construction

The GmCOL1b full length CDS ( https://phytozome-next.jgi.doe.gov/report/gene/Gmax_Wm82_a2_v1/Glyma.18G278100 ) was cloned after amplification into the pCR8 vector (Invitrogen), and sequentially sequenced for verification . The GmCOL1b cDNA was then transferred to the p2FGW7 vector (VIB-UGENT Center for Plant Systems Biology; https://gatewayvectors.vib.be/collection/p2fgw7 by the LR Gateway recombination adhering to the manufacturer’s instructions (Invitrogen). Further information about p2FGW7-35S:EGFP-GmCOL1b plasmid are available upon request by contacting the corresponding author Yoshie Hanzawa (yoshie.hanzawa@csun.edu).

Protoplast Isolation of Glycine max

The protoplast extraction and transfection were conducted according to the procedures described previously with minor modifications (Wu & Hanzawa, 2018) . An enzymatic digestion solution (0.02M of MES pH 5.7, 1.50% w/v Cellulase (R10), 0.50% w/v Macerozyme, 0.20% w/v Pectolyase Y-23, 0.4M D-Mannitol, 0.1M CaCl ₂ , and 7.5% v/v of BSA) was filtered by a 0.45-μ filter. The primary vein and bottom epidermis layer of an 8-day-old soybean unifoliate leaf were removed using the leaf-tape method. The cells were incubated at 22 ^o C in low light, with gentle agitation of 100 rpm until digested. The W5 solution (154mM NaCl, 125 mM CaCl ₂ , 5 mM KCl, 2mM MES pH 5.7) was added, and protoplasts were resuspended in MMg solution (4 mM MES pH 5.7, 400 mM D-Mannitol, 15 mM MgCl ₂ ) to a final concentration of 2 x 10 ⁵ mL ^-1 .

Protoplast Transfection

To an aliquot of protoplasts containing 100,000 cells, 20 μg of plasmid DNA was added for transfection, omitting the negative control, mediated by PEG (20% w/v PEG4000, 400 mM D-Mannitol, 100 mM CaCl ₂ ). The cells were resuspended with WI solution (4 mM MES pH 5.7, 500 mM Mannitol, 20mM KCl). To a 6 well tissue culture plate, the wells were coated with 1 mL of 50% v/v sterile calf serum, and protoplasts were placed in the wells. The plate was covered and incubated at 22 ^o C overnight in the dark. The protoplasts were visualized with a confocal laser scanning microscope (Leica, USA) to verify transfection efficiency and viability at the time of harvest. Protoplast cells were harvested at ZT4-ZT24 at 4-hour intervals with three biological replications.

RNA Isolation, cDNA preparation and Real-time Quantitative RT-PCR

Total RNA was extracted using the Invitrogen RNAqueous Micro-kit for RNA isolation (Invitrogen, CA, USA) following the manufacturer’s instructions. Invasive genomic DNA was removed with the DNAse kit (Invitrogen, CA, USA). First-strand cDNA was synthesized using the iScript cDNA synthesis Kit (BIO-RAD) following the manufacturer’s instructions. The cDNA product was diluted to 1:20 before use. RT-qPCR reaction was performed using QuantStudio3 (Applied Biosystems) with three technical replications. The settings for amplification are a hold at 50 ̊C for 2 minutes, 95 ̊C for 10 minutes, a PCR of 95 ̊C at 15 seconds, ending at 60 ̊C for 1 minute where amplification was captured with a SYBR target. The results were analyzed on Thermo Fisher’s Connect Data Analysis Apps for real-time qPCR.

Table 1. Soybean accession used for protoplast isolation.

Table 2. Plasmid name with vector backbone and its respective Gene ID.

Table 3. Primers used in RT-qPCR. Primers were designed targeting exon-exon junctions of targeted genes using the CDS sequences in Williams82.a2.v1 .