Virus-induced gene silencing shows that LATE FLOWERING plays a role in promoting flower development in soybean

Virus-induced gene silencing (VIGS) is a useful tool to investigate the role of genes particularly in hard-to-transform plant species via the Agrobacterium-mediated genetic transformation process. Soybean is one of the most important crops for the food and protein source, but its low transformation efficiency makes it hard to identify the functions of genes of interest. Here, we adopted tobacco ringspot virus (TRSV)-based VIGS systems and examined the role of a LATE FLOWERING (GmLATE) gene in soybean. Because TRSV induces symptoms affecting leaf senescence and development, we screened soybean genotypes and selected a genotype, named Aram, which shows resistance to TRSV symptoms while is susceptible to TRSV-induced gene silencing. The TRSV-based silencing of GmLATE in soybean showed suppressed flower development with decreased expression of genes related to flowering. These results suggest that GmLATE plays a role in promoting flower development in soybean, which is different from its role as a floral repressor in Arabidopsis. Our results show the novel function of GmLATE and demonstrate that TRSV-based VIGS system can be used as a tool to study genes involved in flowering in soybean.


Introduction
Soybean [Glycine max (L.)] is an annual leguminous crop whose seeds are rich in proteins, essential amino acids, and oil (Li et al. 2017;Patil et al. 2017). Because of its high nutritional value, soybean seeds are used as a major ingredient for human food and livestock feed (Lu et al. 2021). Floral transition is an important developmental process for seed production (Jung and Müller 2009), thus knowledge on the genetic factors that regulate flowering is critical to improve seed yield in soybean. To study the function of unknown genes in plants, Agrobacterium-mediated genetic transformation has been used as a major method to introduce foreign DNA into host plants (Hinchee et al. 1988). However, the transformation efficiency using this method in soybean is still low and highly variable depending on genotypes (Jia et al. 2015). The low transformation efficiency in soybean is due to the multistep method to generate transgenic plants (Meurer et al. 1998). Only the soybean genotypes which are susceptible to Agrobacterium infection and have high organ regeneration capacity can be used for the transformation.
to 10% (Li et al. 2017;Paes de Melo et al. 2020), but it is still a bottleneck to study the role of the genes involved in soybean flowering.
Virus-induced gene silencing (VIGS) has been used as a reverse genetic tool to study function of genes in crops (Ramegowda et al. 2014). For VIGS, the viral genome is modified to produce double stranded RNA (dsRNA) containing partial sequence of the target gene in the host plants. Plants activate defense systems in response to dsRNA and produce small interfering RNAs (siRNAs), which identify complementary RNA sequences and trigger degradation (Lu et al. 2003;Ramegowda et al. 2014). Using the VIGS system, a specific mRNA could be silenced by the virus infection in host plants without genetic transformation. Various virus species including Cucumber mosaic virus (CMV; Otagaki et al. 2006), Soybean mosaic virus (Seo et al. 2009), Bean pod mottle virus (BPMV; Zhang and Ghabrial 2006;Zhang et al. 2010), Soybean yellow common mosaic virus , and tobacco ringspot virus (TRSV; Zhao et al. 2016) were adopted for expression of foreign genes and VIGS in soybean. For example, BPMV-based VIGS was used to study role of WRKY transcription factors in soybean. Silencing GmWRKY58 and GmWRKY76 largely reduced plant height and suppressed leaf expansion, showing that WRKYs are required for overall plant growth (Yang et al. 2016). CMV-based VIGS was used to identify the minimal amount of mRNA encoding flavonoid 3'-hydroxylase for pigmentation in soybean pubescence (Nagamatsu et al. 2009).
Molecular mechanisms underlying the regulation of flowering have been intensively studied in the model plant, Arabidopsis thaliana. Among the various pathways, the photoperiod-dependent pathway is one of the best identified processes explaining how plants recognize environmental signals to determine timing of flower development (Song et al. 2013). The CONSTANS (CO) and FLOWERING LOCUS T (FT) have been identified as key factors involved in photoperiod flowering (Putterill et al. 1995;Suárez-López et al. 2001). After the findings, various upstream regulators controlling CO and FT abundance were uncovered, showing that the CO-FT module is a signaling hub for the floral transition Song et al. 2014;Weingartner et al. 2011;Zhang et al. 2019). Among them, a zinc-finger transcription factor LATE FLOWERING (LATE) is identified as a floral repressor in Arabidopsis (Weingartner et al. 2011). Overexpression of LATE delays flowering by inhibition of CO and FT expressions. Ectopic expression of FT in the LATE-overexpressing plants restored the late flowering phenotype (Weingartner et al. 2011), suggesting that LATE is an upstream regulator of the CO-FT module. While the function of LATE is characterized in Arabidopsis, whether its role is conserved in crops has not been examined.
The Brassica napus homolog of the Arabidopsis LATE, BnLATE, was identified as a negative regulator of lignin accumulation in the silique walls (Tao et al. 2017), but its role in the regulation of flowering is unknown.
Because floral transition is an important developmental process for seed yields in soybean, genes involved in controlling flowering have been identified by quantitative trait loci (QTLs) analysis (Wang et al. 2019). Among the QTLs, E9 and E10 have been demonstrated as homologs of Arabidopsis FT gene (Kong et al. 2014;Samanfar et al. 2017). Analysis of soybean genome sequences and the phylogenetic relationship showed that there are 6 members of FT homologs and 4 members of the upstream CO homologs in soybean (Fan et al. 2014). Transcript levels of the soybean FT and CO genes were relatively high in trifoliate leaves than those in stems and flowers, suggesting that regulation of FT transcription by CO occurs mainly in the leaves like Arabidopsis (Fan et al. 2014). The function of soybean CO and FT genes were indirectly investigated using the Arabidopsis. Complementation of CO-deficient mutant by overexpression of soybean CO-like genes, GmCOL5, restored the delayed flowering. In addition, overexpression of soybean FT-like genes, GmFTLs, promoted Arabidopsis flowering (Fan et al. 2014), indicating that the function of CO-FT module is conserved in soybean. In addition to CO and FT genes, other genes related to flowering were also reported. Quadruple mutation of four soybean APETALA1 (GmAP1) genes delayed flower development and changed flower morphology under short day conditions, proving the role of GmAP1 in the photoperiod flowering . Meanwhile, overexpression of miR156 delayed flowering and suppressed expression of flowering genes including GmFT and GmAP1 (Cao et al. 2015). These results show that the role of several floral regulators were identified in soybean, but there are still many genes whose functions are demonstrated in Arabidopsis but remain unknown in soybean.
In this work, we adopted the TRSV-based VIGS system in soybean and examined the role of soybean homolog of Arabidopsis LATE gene in flower development. We found that TRSV-based VIGS suppressed expression of GmLATE mainly in fourth trifoliate leaves and decreased flower numbers. In addition, analysis of GmCOLs and GmFTLs expression showed that GmLATE is possibly involved in GmCOL13 and GmFTL5 expression. Based on our data, we newly found that soybean LATE gene is related to induction of flower development, which is different from Arabidopsis LATE gene. Our work shows novel functions of LATE gene in soybean and demonstrates that TRSV-based VIGS can be used as a tool to study function of flowering genes in soybean.

Preparation of TRSV-based VIGS vectors.
A 209-bp fragment of soybean phytoene desaturase (GmPDS) gene was amplified via PCR using GmPDS-1627 F and GmPDS-1857 R primers which were previously described (Zhao et al. 2016). A 219-bp fragment of GmLATE gene was amplified using GmLATE-SnaBI-F and GmLATE-SnaBI-R primers. A 210-bp fragment of GmSM gene were amplified using GmSM-SnaBI-F and GmSM-SnaBI-R primers. Primer sequences were listed in Table S1. The PCR products were inserted into pT2S vector (Zhao et al. 2016) using the SnaBI site. A pTRSV1 vector and a pT2S vector containing gene fragments were separately transformed into Agrobacterium tumefaciens strain GV3101. For agroinfiltration, Agrobacterium cells carrying pTRSV1 and cloned pT2S vectors were grown at 28 o C for 12 to 16 h in the liquid Luria-Bertani medium. Cells were harvested by centrifugation and resuspended in the infiltration medium containing 10 mM MES pH 5.6, 10 mM MgCl 2 , and 200 µM acetosyringone and then incubated for 3 h at 23 o C.

Virus infection for VIGS
For VIGS, Nicotiana benthamiana (N. benthamiana) plants were grown for 6 weeks at 23 o C in soil under long day conditions. Agrobacterium cell cultures were infiltrated into N. benthamiana leaves using 1 ml needless syringes and the plants were incubated for 6 days under same growth conditions. The infiltrated leaves were then harvested and ground with 2 ml phosphate-buffered saline (PBS) containing 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 1.8 mM KH 2 PO 4 , pH 7.4 at room temperature for sap preparation.
The sap was inoculated to the carborundum-dusted soybean unifoliate leaves by rubbing with cotton swab.

Analysis of gene expression
Plant materials were ground in liquid nitrogen and total RNA was extracted using Trizol (Thermo Fisher Scientific, Waltham, MA) following the manufacturers' instructions. First strand cDNA synthesis was performed using TOPscript cDNA Synthesis Kit (Enzynomics, Daejeon, Korea). Reverse transcription-mediated PCR (RT-PCR) reactions were carried out using the prepared cDNAs and genespecific primers listed in Table S1. For analyzing expression of genes in soybean, Reverse transcription-mediated quantitative PCR (RT-qPCR) reactions were carried out in 96-well plates with a CFX Connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA) using TOPreal qPCR PreMIX (Enzynomics). The ACTIN 11 (ACT11) was used as a reference gene (Le et al. 2012). The comparative ΔΔC T method was used to determine relative quantities of each RT-qPCR results. The threshold cycle (C T ) was automatically determined for each reaction with default parameters in the system.

Analysis of soybean phenotype after VIGS
The PDS-silenced soybean plants using the VIGS method were grown in the controlled growth room at 23 o C under long day conditions for 2 weeks after the virus infection. Soybean plants that show white or pale-green color in their upper leaves were counted as PDS-silenced plants. For GmLATE or GmSM silencing, soybean plants in pots were grown in the greenhouse from 4th August 2020 to 1st September 2020 in Daejeon, Korea. The average temperature during this period is 27.6 o C. Flower numbers per plants were counted to evaluate the effects of GmLATE or GmSM silencing on soybean flowering.

Screening soybean genotypes for application of TRSV-based VIGS
We previously have developed TRSV-based vectors for VIGS in diverse plant species (Zhao et al. 2016). For practical use of this tool in soybean, we screened appropriate soybean genotypes for TRSV-based VIGS by silencing GmPDS gene. Because the virus-infected N. benthamiana leaves are required for introduction of the TRSV-based vectors in soybean, Agrobacterium cells containing the pTRSR1 were co-infiltrated into the N. benthamiana leaves with those containing pT2S-empty, or pT2S-GmPDS vectors (Fig. 1a). Analysis of transcript levels via RT-PCR showed that GmPDS gene fragments were highly expressed in the systemic leaves (Fig. 1b), indicating that N. benthamiana plants were infected by our engineered TRSV.
To find VIGS-compatible soybean genotypes, we screened 43 genotypes by silencing GmPDS gene using the virus-infected N. benthamiana leaves. By the initial screening process, we found that 40 genotypes showed apical necrosis (Fig. 2a), which is a typical phenotype of TRSVinfected plants (Fuchs et al. 2010). In addition, 16 genotypes showed leaf senescence, which made us hard to investigate GmPDS silencing. We thus excluded genotypes that showed leaf senescence and selected 5 candidate genotypes that showed high rate of GmPDS silencing and low-level apical necrosis symptoms ( Fig. 2a and b). We also selected Miso as a control genotype that showed no GmPDS silencing and highest rate of defective symptoms such as leaf senescence and apical necrosis.
We next performed second screening process using 5 candidate genotypes and 1 control genotype. In this experiment, we additionally infected plants by TRSV carrying no gene fragment (TRSV-EV) as a negative control group. Similar to the initial screening results, 5 candidate genotypes showed low-level TRSV symptoms, while all the tested Miso plants showed defective TRSV symptoms ( Fig. 3a and c). Among them, 2 genotypes, Haepum and Aram, showed no TRSV symptoms. When the selected candidate genotypes were infected by TRSV carring GmPDS gene fragment (TRSV-GmPDS), Aram and Soyeon showed no defective symptoms ( Fig. 3b and c). However, Aram showed 93.8% rate of GmPDS silencing while Soyeon showed only 78.6% silencing rate. While we obtained similar results from the first and second screening, the PDS silencing efficiency and TRSV symptoms were slightly different between two tool (https://www.ebi.ac.uk/Tools/msa/clustalo). For phylogenetic tree analysis, LATE amino acid sequences from 6 species were analyzed using the neighbor-joining method in MEGA 11 software.

Statistical analysis
For bar graphs, the mean ± standard deviation calculated from at least three independent replicates is displayed using whiskers. Standard deviation and R 2 values were calculated using Microsoft Excel 2003. For box plots, each box extends from the first quartile to the third quartile values of the data. Center lines show median values. Whiskers show 1.5 x interquartile range (IQR). Data points higher than the upper bound (third quartile values + 1.5 x IQR) and those lower than the lower bound (first quartile values -1.5 x IQR) are considered as outliers. Statistical significance between two groups were determined using 2-tailed Student's t-test (*P < 0.05). For more than two groups, one-way ANOVA with Tukey's post-hoc test was performed and statistical significance was annotated as different letters (P < 0.05).

Accession number
Gene accession numbers: (Glyma19G05170), GmFTL2 independent screening experiments. These variations were possibly because biological activity of TRSV is not exactly identical in every independent experiment. By combination of the screening results, we finally selected Aram as a target genotype for further gene silencing analysis.

Identification of GmLATE by protein sequence analysis
Previous reports have demonstrated that a C 2 H 2 zinc-finger protein LATE plays an important role as a floral repressor, thus overexpression of LATE in Arabidopsis largely delays flowering time (Weingartner et al. 2011). To find soybean homolog of LATE protein, we performed BLAST analysis using the Arabidopsis LATE amino acid sequence.
The results showed that Glyma19G162600, which is annotated as soybean LATE, has 42% of identities with 55% of positives. We thus designated the Glyma19G162600 as GmLATE. Amino acid sequences are highly conserved within C 2 H 2 type zinc finger domain between Arabidopsis and soybean LATE, suggesting that GmLATE is also a zinc finger transcription factor (Fig. 4a). Phylogenetic analysis revealed that GmLATE is an evolutionally distinct protein, while it is relatively similar to the protein from Zea mays (Fig. 4b).
To find whether there are GmLATE-like proteins in soybean, we performed BLAST analysis using the GmLATE protein sequence in soybean protein sequence database. The results showed that Glyma03G160900, which is annotated as SUPERMAN (SM), is highly similar to GmLATE and has 87.64% of identities (Fig. S1). We thus designated it as GmSM and analyzed in parallel to GmLATE in the following experiments. Fig. 3 Analysis of TRSV-mediated gene silencing in the selected genotypes using a TRSV-EV as a control a Measurement of TRSV symptoms after infection of empty vector (EV)-containing TRSV. Selected genotypes were grown for 2 weeks before infection. Symptoms were analyzed 2 weeks after TRSV-EV infection (n = 6-8). Genotypes showing no TRSV symptoms are marked in bold b Measurement of TRSV symptoms and PDS silencing efficiency in the selected genotypes. Soybean plants were grown for 2 weeks and then GmPDS vector-containing TRSV was infected. Symptoms and PDS silencing efficiency were analyzed after 2 weeks (n = 11-16). A genotype showing no TRSV symptoms with high PDS silencing efficiency is marked in bold c Phenotype of Aram after TRSV infection. The Aram was grown for 2 weeks and then EV-or GmPDS vector-containing TRSV was infected. Plants were photographed after 2 weeks Fig. 2 Screening of soybean genotypes for high efficiency of PDS silencing and low severity of TRSV symptoms a Measurement of PDS silencing efficiency and TRSV symptoms. The PDS silencing efficiency and TRSV symptoms of 43 soybean genotypes were analyzed. Soybean plants were grown for 2 weeks before infection. Symptoms and PDS silencing efficiency were measured after 2 weeks (n = 3-14). Genotypes that exhibited less than 40% of apical necrosis and more than 40% of PDS silencing efficiency with 0% leaf senescence and leaf curling are marked in bold (black arrow). Miso was selected as a susceptible cultivar showing severe TRSV symptoms (white arrowhead) b Phenotype of soybean genotypes after TRSV infection. Selected soybean genotypes were photographed at the time point when PDS silencing efficiency and TRSV symptoms were analyzed in (a) had significantly less number of flowers than control plants ( Fig. 5c and d). Note that we excluded two TRSV-GmLATE plants (Rep 2 and 3) based on the outlier calculations. GmLATE silencing reduced standard deviation more than 3 fold (13.4 to 3.45), indicating that individual variation of flower development is controlled by GmLATE. However, plants infected by TRSV-GmSM did not show any difference to those infected by TRSV-EV ( Fig. 5c and d), suggesting that GmLATE and GmSM have different functions.
To validate the silencing of GmLATE in plants, we harvested the 2nd, 3rd, 4th trifoliate leaves and flowers and analyzed the expression of innate GmLATE gene. We analyzed 3 and 4 independent plants infected by TRSV-GmLATE and TRSV-EV, respectively. Two-dimensional analysis of the gene expression and the flower number showed that GmLATE expression was highly correlated with the flower number in 4th trifoliate leaves; GmLATE expression exponentially increased with flower number (R 2 = 0.989). On the other hand, there was less or no correlation in other tissues of the control plants (Fig. 5e). Consistent with the results, TRSV-GmLATE showed low expression of GmLATE in the 4th trifoliate leaves (Fig. 5e). However, GmLATE expression levels in TRSV-EV and TRSV-GmLATE were not significantly different in this tissue (P = 0.171) because of high variation of GmLATE expression in TRSV-EV. Instead, GmLATE silencing was effective in reducing standard deviation of GmLATE expression (0.102 to 0.006). Together with the phenotype data, these results suggest that TRSV-based silencing of GmLATE possibly decreased GmLATE expression in 4th trifoliate leaves and suppresses flower development in soybean.
Effects ofGmLATEsilencing on expression of flowering genes.
The CO-FT module is conserved in soybean (Fan et al. 2014), thus we analyzed expression of GmCOLs and GmFTLs in the GmLATE-silenced plants. Before the analysis, we firstly investigated whether the expression of flowering genes shows distinct patterns in the 4th trifoliate leaves where GmLATE may function to promote flowering based on our data (Fig. 5e). Analysis of gene expressions in 2nd, 3rd, and 4th trifoliate leaves revealed that there is no particular difference between 4th trifoliate leaves and other tissues while expression of several genes were slightly decreased in 3rd and 4th trifoliate leaves than that in 2nd trifoliate leaves (Fig. 6a).
Because TRSV-based VIGS was particularly effective in 4th trifoliate leaves, we analyzed expression of flowering time genes in those tissues from the GmLATE-silenced plants. Two-dimensional analysis of the gene expression and the flower number showed that the expression of GmCOLs is proportional to the flower number. In particular, GmCOL13 and GmCOL5 expression increased exponentially with

GmLATE silencing suppresses flower development in soybean
To examine the role of GmLATE in soybean, we have inserted the GmLATE gene fragment into the pT2S vector, producing pT2S-GmLATE. Because GmSM has highly similar protein sequences to GmLATE, we also generated pT2S-GmSM to compare the function of the two proteins. The pT2S-GmPDS was used as a positive control and the pT2S-empty was used as a negative control for VIGS. We firstly infiltrated Agrobacterium cells containing each vector construct into N. benthamiana leaves for virus infection. The infection was verified by the analysis of transcript levels of each gene fragment in the systemic leaves (Fig. 5a). Next, we harvested the infected N. benthamiana leaves and extracted the sap for the virus infection of soybean genotype Aram and silencing the each gene. After 4 weeks of the virus infection, we found that the plants infected by TRSV-GmLATE developed decreased number of flowers in compared to those infected by TRSV-EV (Fig. 5b). Measurement of flower numbers revealed that GmLATE-silenced plants Amino acid sequence of soybean LATE was obtained from NCBI (https://www.ncbi.nlm.nih.gov) by BLAST analysis. Sequence alignment was performed using the Clustal Omega tool (https://www.ebi. ac.uk/Tools/msa/clustalo) b Phylogenetic analysis of the LATE protein in multiple plant species. Putative LATE amino acid sequences in Glycine soja, Gossypium hirsutum, Solanum lycopersicum, Zea mays, and Glycine max were obtained by BLAST analysis using the Arabidopsis LATE sequence. Phylogenetic tree was built using the MEGA 11 software. Numbers indicate branch lengths revealed that GmLATE is related to promoting flower development possibly by inducing expression of GmCOL13 and GmFTL5, which are conserved flowering genes in soybean.

Discussion
In this study, we adopted TRSV-based VIGS tool to investigate function of GmLATE in soybean flowering. While we have demonstrated that TRSV-based VIGS tool is useful for gene silencing in multiple plant species including N. benthamiana, Arabidopsis, soybean, melon, and cucumber, only PDS silencing was tested in the previous study (Zhao et al. 2016). Here, we built TRSV vectors carrying GmLATE gene fragment and silenced the gene for the analysis of flowering. By the VIGS tool, we could observe suppressed flower development phenotype in soybean, proving the TRSV-based VIGS as a useful tool for analyzing function of flowering genes. However, there are still limitations for the TRSV-based VIGS systems. We screened 43 soybean genotypes, but about 93% of the tested genotypes exhibited defective apical necrosis phenotype after application of TRSV-based VIGS (Fig. 2). These results suggest that flowering time (R 2 = 0.980 and 0.986, respectively; Fig. 6b). GmCOL13 and GmCOL5 showed relatively low expression levels in TRSV-GmLATE than those in TRSV-EV. However, they are not significantly different (P = 0.215 and P = 0.597, respectively) because of high variation particularly in TRSV-EV plants. Notably, GmLATE silencing was effective in reducing standard variation of GmCOL13 expression (0.47 to 0.017). Similar results were obtained by analyzing GmFTLs expression. Among the 4 GmFTL genes, GmFTL5 expression increased exponentially and showed high correlation with flower number (R 2 = 0.989; Fig. 6b). Expression of GmFTL5 in TRSV-GmLATE was relatively lower than that in TRSV-EV, but high variation precluded statistical significance (P = 0.217). Similar to GmCOL13, GmLATE silencing reduced standard deviation of GmFTL5 expression (0.45 to 0.12).
We next analyzed time-course expression of GmLATE in shoot apexes as well as leaves, because the expected function of GmLATE is inducing flower development. During the growth stages developing 2nd to 5th trifoliate leaves, GmLATE showed similar expression levels in shoot apexes and leaves (Fig. S2), suggesting that GmLATE may function both in leaves and shoot apexes. All together, our data Flower numbers were measured and displayed as a box plot (c). Box plots represent low quartile, median, and third quartile; whiskers represent 1.5 fold interquartile range; dots represent outliers (n = 9-10; Tukey's test; P < 0.05). Flower numbers for each replicate were displayed as a table (d). Asterisks represent outliers e Two-dimensional plots showing the flower number and GmLATE gene expression. Three and four different soybean plants infected with TRSV-EV and TRSV-GmLATE, respectively, were used. Expression of GmLATE gene was analyzed in different tissues. T2, T3, T4, and F represent second trifoliate, third trifoliate, forth trifoliate, and flower, respectively factor. It has been shown that a transcription factor functions both as a transcription activator and as a repressor. The Arabidopsis TGACG SEQUENCE-SPECIFIC BINDING PROTEIN 2 is a transcription factor mediating salicylic acid signaling and has dual functions (Kesarwani et al. 2007). In addition, the Sp3 transcription factor can activate or repress expression of target genes depending on protein acetylation in breast cancer cells (Ammanamanchi et al. 2003). Because Arabidopsis and soybean LATE protein sequences have a number of different sites, it is possible that AtLATE is a transcription activator but GmLATE is a transcription repressor of flowering genes.
C2H2 zinc finger proteins are classified by the types of zinc finger domains based on whether the protein contains tandem arrays of fingers (type A and B) or not (type Englbrecht et al. 2004). Type C zinc finger proteins typically contain a single or several dispersed fingers. Among the type C proteins, the C1 family is characterized by the conserved QALGGH sequence in the zinc finger domain (Englbrecht et al. 2004;Lyu and Cao 2018). In Arabidopsis, a number of C2H2 zinc finger proteins have been identified to be involved in flowering. For example, EARLY only limited soybean genotypes can be used as a subject for VIGS. Our previous study have identified that a gene encoding TOLERANCE TO TOBACCO RINGSPOT VIRUS 1 (TTR1) is responsible for the lethal systemic necrosis upon TRSV infection (Nam et al. 2011). Ectopic expression of the TTR1 gene from the TRSV-sensitive Est ecotype of Arabidopsis in the TRSV-tolerant Col-0 ecotype causes development of the lethal systemic necrosis phenotype. Because two amino acid residues, Leu956 and Lys1124, are critical to determine TRSV susceptibility (Nam et al. 2011), it is possible that protein sequences of TTR1 homologs in the resistant genotype Aram and in the sensitive genotype Miso are different. Therefore, analysis of TTR1 homologs in soybean will be important to select appropriate genotypes for application of TRSV-based VIGS system.
In Arabidopsis, LATE is a floral repressor, thus overexpression of LATE causes delayed flowering with defects in flower differentiation (Weingartner et al. 2011). However, our data showed that GmLATE is involved in inducing flower development, which is different from the Arabidopsis LATE. These results suggest that Arabidopsis and soybean LATE might have different activity as a transcription Fig. 6 Analysis of genes related to flowering time in GmLATEsilenced soybean a Expression of soybean genes related to flowering time in different tissues. Plants were grown for 4 weeks. T2, T3, and T4 were harvested for analyzing gene expressions (n = 3, Tukey's test, P < 0.05) b Two-dimensional plots showing the flower number and the expression of flowering time genes in GmLATE-silenced soybean. Soybean plants grown for 2 weeks were infected to TRSV-EV or TRSV-GmLATE and further grown for 4 weeks. T4 leaves were harvested for analyzing gene expressions Author contributions JSM and HJL conceived and designed research. SYS and HJL performed experiments and analyzed data. HJL wrote the manuscript with the contributions of JSM and MRP. All authors read and approved the manuscript.
Funding This work was supported by the KRIBB Research Initiative Program (KGM5372221) and the New Breeding Technologies Development Program (Project No. PJ0165302022) provided by the Rural Development Administration of Korea.
Data Availability All data generated in this study are included in this article and its supplementary information files. Any additional information required to reanalyze the data is available from the corresponding author upon request.

Conflict of interest
The authors declare that they have no conflict of interest.
Consent for publication All authors read this paper and confirm the context.

Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. FLOWERING 6 (ELF6) and its close homolog RELATIVE of EARLY FLOWERING 6 (REF6) are A-type zinc finger proteins that regulate flowering (Lyu and Cao 2018). While ELF6 represses flowering, REF6 promotes flowering by decreasing transcription of the FLOWERING LOCUS C floral repressor (Noh et al. 2004). Meanwhile, zinc finger proteins belonging to the subset C1 tend to be related to flower development. SM, RABBIT EARS, JAGGED, and KNUCKLES are C1-type proteins and regulate floral organ formation (Bowman et al. 1992;Jacobsen and Meyerowitz 1997;Krizek et al. 2006;Payne et al. 2004;Schiessl et al. 2014). Because GmLATE belongs to the subset C1 containing the QALGGH motif (Fig. S1), it is plausible that GmLATE controls flower development in soybean like other C1-type zinc finger proteins in Arabidopsis. Because silencing GmSM did not cause changes in flower development (Fig. 5c), GmLATE may have functional redundancy with GmSM in soybean flowering. Further study on the molecular mechanisms of GmLATE and GmSM would be required to identify relationship between these transcription factors in soybean.
In our gene expression analyses, GmLATE, GmCOL13, and GmFTL5 showed low expression levels in TRSV-GmLATE in compared to those in TRSV-EV, but there was no statistical significance between two groups (Figs. 5e and 6b). These results are due to high variation of flowering gene expression in individual TRSV-EV plants. Because expression levels of these genes increased exponentially with flower numbers (Figs. 5e and 6b), it seems that silencing GmLATE reduces flower numbers and related gene expressions to the basal levels. Indeed, analysis of standard deviation showed that individual variation of flower numbers and expression of GmLATE, GmCOL13, and GmFTL5 were largely decreased by GmLATE silencing (Fig. 5d, e, and 6b). Therefore, our data suggest that GmLATE is involved in promoting flower development and confers individual variation of flowering depending on its expression levels in soybean leaves.
In summary, we demonstrated that TRSV-based VIGS system can be applied to study functions of flowering genes in soybean by suppressing GmLATE. Through screening of soybean genotypes, we selected Aram ecotype, which showed resistance to TRSV-induced apical necrosis and efficient GmPDS silencing. Furthermore, we suggested that GmLATE plays a role in promoting flower development, and affects expression of key flowering genes such as GmCOL13 and GmFTL5. Our results will be helpful to study genes related to flowering using the TRSV-based VIGS tool in soybean.