Growth‐regulating factor 5 (GRF5)‐mediated gene regulatory network promotes leaf growth and expansion in poplar

Summary Although polyploid plants have larger leaves than their diploid counterparts, the molecular mechanisms underlying this difference (or trait) remain elusive. Differentially expressed genes (DEGs) between triploid and full‐sib diploid poplar trees were identified from two transcriptomic data sets followed by a gene association study among DEGs to identify key leaf growth regulators. Yeast one‐hybrid system, electrophoretic mobility shift assay, and dual‐luciferase assay were employed to substantiate that PpnGRF5‐1 directly regulated PpnCKX1. The interactions between PpnGRF5‐1 and growth‐regulating factor (GRF)‐interacting factors (GIFs) were experimentally validated and a multilayered hierarchical regulatory network (ML‐hGRN)‐mediated by PpnGRF5‐1 was constructed with top‐down graphic Gaussian model (GGM) algorithm by combining RNA‐sequencing data from its overexpression lines and DAP‐sequencing data. PpnGRF5‐1 is a negative regulator of PpnCKX1. Overexpression of PpnGRF5‐1 in diploid transgenic lines resulted in larger leaves resembling those of triploids, and significantly increased zeatin and isopentenyladenine in the apical buds and third leaves. PpnGRF5‐1 also interacted with GIFs to increase its regulatory diversity and capacity. An ML‐hGRN‐mediated by PpnGRF5‐1 was obtained and could largely elucidate larger leaves. PpnGRF5‐1 and the ML‐hGRN‐mediated by PpnGRF5‐1 were underlying the leaf growth and development.

Table S1 All primer sequences used in this study. Table S2 The height, diameter and the fifth leaf area of five-month-old PpnGRF5-1 overexpression transgenic lines.

Methods S6 DNA affinity purification sequencing (DAP-seq) and data analysis
Methods S7 Dual-luciferase assay Methods S8 Yeast two-hybrid assays Methods S9 GST (glutathione-S-transferase)-fusion protein pull-down assay and western blotting Methods S10 Split luciferase complementation assay   PpnGRF5-1 protein contained one QLQ and one WRC domain, and the less conserved TQL motif was located within the C-terminal region of PpnGRF5-1. (b) Transactivation activity assays of PpnGRF5-1.
Different DNA fragments of PpnGRF5-1 were fused to the sequence encoding GAL4-DBD and introduced into AH109 yeast cells. The vector pGBKT7 was used as a negative control. Yeast were spotted onto nutritional selective medium (SD/-Trp, SD/-Trp-His, or SD/-Trp-Ade-His) and allowed to grow at 30℃ for three days.   in comparison with the 84k wild type (WT). Each node represents an enriched GO term and a bigger node size represents a larger number of DEGs. Each node color denotes a certain scale of p-value of a GO term. Each line represents overlapping DEGs between the two GO terms; a bigger edge width represents a bigger number of overlapping DEGs. The enrichment was analysed using R module named clusterProfiler (v 3.14.3).  PpnGRF5-1 binding sites in the above-mentioned target genes.

Table S1 All primer sequences used in this study
Primer ID Primer Sequence (5'->3') Purpose   The two-step procedure for construction of a TF-mediated ML-hGRN was adopted. The step-by-step procedure was thoroughly described in two of our earlier publications (Lin et al., 2013;Wei, 2019). In the first step, we identified the TF-responsive genes, the ones whose expression profiles were highly concordant with TF's. This was done by integration of Fisher's exact test and a probability-based method. In the second step, the interference frequency between TF and each TF-responsive candidate target gene was determined by a top-down GGM algorithm (Lin et al., 2013;Wei, 2019). Briefly, given a combination of TF (z) and a pair of TF-responsive genes ( and ), the significance of interference of z on x and y was determined by testing if the presence of z makes the correlation between x and y become more or less significant. This was accomplished by examining if d = -| was significant by the multivariate delta method (MacKinnon et al., 2002), where | is the Spearman's partial correlation coefficient of x and y given z while is Spearman's correlation rho of x and y. If the p-value of d was significant, we then concluded that z interfered with both and , which were recorded as z interfered with and z interfered with . After all combinations of different x and y were tested with z, the interference frequency of z with each responsive gene was determined by counting the significant interference in all combinations. PpnGRF5-1-responsive candidate target genes that were interfered by z with the highest frequencies were retained, and subsequently intersected with PpnGRF5-1 target genes derived from DAP-seq experiment, and only those that were in the interaction and also possessed PpnGRF5-1 binding motifs in their proximal promoters were kept at the second layer of GRN as the direct target of PpnGRF5-1. The next layer (namely, the third layer) is extended in a top-down fashion only from the TFs present in the second layer (direct targets of PpnGRF5-1) by recursively calling the top-down GGM algorithm. More detailed procedure of top-down GGM algorithm can be found from two of our early publications (Lin et al., 2013;Wei, 2019), and also from two recent application studies (Chen et al., 2019;Wei et al., 2020).

Methods S2 RNA isolation, RT-PCR, and RT-qPCR
Total RNA was isolated from the collected materials with TRIzol reagent (Thermo Fisher, USA, cat 15596026), and then treated with RNase-free DNase I (Thermo Fisher, USA, cat EN0521) according to the manufacturer's instructions. Full-length cDNA was then reverse transcribed using a cDNA synthesis kit (Tiangen, China, cat KR106). qPCR was performed according to the manufacturer's instructions (TRANSGEN, China, cat AQ142-21) in a total volume of 25 μL on the Applied Biosystems 7500 real-time PCR system according to the manufacturer's manual. All real-time PCR reactions were repeated at least three times. The level of Actin gene (Potri.001G309500) transcript was used as an internal control. The specific primers of these genes were obtained by directly querying the prime data qPrimerDB (https://biodb.swu.edu.cn/qprimerdb) (Lu et al., 2018). Three technical replicates and three biological replicates were performed on all reactions. The ΔΔCt algorithm was used for calculating relative gene expression. Primers are listed in Table S1.

Methods S3 Transcriptional activation analysis in yeast cells
For transcriptional activation activity assays, the different domain of PpnGRF5-1 was fused with GAL4 DNA-binding domain in the pGBKT7 vector (Clontech, USA, cat 630443) and transformed into the yeast strain AH109 with Ade2 and His reporter genes, respectively. The empty pGBKT7 vector was used as a negative control. Transformed strains were confirmed by PCR and then plated on selective synthetic dropout (SD) media without Trp (SD/-Trp), without Trp and His (SD/-Trp-His), or without Trp, His, and Ade (SD/-Trp-His-Ade) with 5 mM 3-AT, to determine their survivability. All transcriptional activation assays were performed in three replications. Primers are listed in Table S1.

Methods S4 Yeast one-hybrid assays
The full-length sequence of PpnGRF5-1 was amplified and fused into the activation domain (AD) in the pJG4-5 vector (Clontech) to generate pJG4-5-PpnGRF5-1 construct. Fragments containing the two putative PpnGRF5-1 binding sites of 'TGTCAG' or all of the substitution mutants of the two binding sites in the CKX1 promoter were independently amplified and fused into the vector pLacZi2μ (Lin et al., 2007). Briefly, the yeast strain EGY48 cells were transformed with PpnGRF5-1 fusion constructs and various LacZ reporter plasmids. Then the yeasts were plated on synthetic dropout medium containing X-gal but without tryptophan and uracil for blue color development. Primers are listed in Table S1. All assays were performed in three replications.

Methods S5 Electrophoretic Mobility Shift Assay (EMSA)
EMSA was performed to examine if PpnGRF5-1 can bind to the promoter of CKX1. To do this, two complementary 60-bp long oligonucleotides containing the binding cis-elements were separately synthesized and then labeled with biotin using EMSA Probe Biotin Labeling Kit (Beyotime, China, cat GS008). GST and GST-PpnGRF5-1 recombinant proteins were expressed in the Escherichia coli BL21 (DE3) strain and then purified using GST-tag Protein Purification Kit (Beyotime, China, cat P2262). DNA gel mobility shift assay was performed using the EMSA kit (Beyotime, China, cat GS009) following the manufacturer's protocol. Briefly, the DNA probes and proteins were co-incubated in the reaction buffer at room temperature. We added a specific competitor (non-biotin) and non-specific competitor (mutated) probes into the reaction mixture for competition reaction. After incubation, the reaction mixture was separated by 6% native polyacrylamide gel, and then, labeled DNA was detected using the Biostep Celvin S420 system (Biostep, German). All assays were performed in three replications. Primers are listed in Table   S1.

Methods S6 DNA affinity purification sequencing (DAP-seq) and data analysis
DAP-seq was performed as described previously (Bartlett et al., 2017). The coding sequence of PpnGRF5-1 was cloned into the vector pFN19K. Then PpnGRF5-1 protein expression was expressed using TnT® Coupled Wheat Germ Extract System (Promega, USA, L4130). Genomic DNA (gDNA) was extracted from apical bus (meristem + tiny unopened leaves) that harvested from three-month-old 84K populus following the procedure of the DNeasy plant mini kit (Qiagen, Germany, cat 69104). The gDNA was sonicated to a fragment size of 200 to 800 bp. Halo-PpnGRF5-1 protein was bound to anti-Halo monoclonal antibody agarose beads (Promega, USA, cat G9211) and incubated with 200 ng of fragmented gDNA for 1 hour at room temperature. After incubation, the beads were washed and DNA was recovered. Samples were pooled and sequenced on an Illumina Novaseq 6000 platform and an average of 150 bp paired-end reads were generated from each library. A total of 10-30 million reads were obtained for each sample. Two biological replicates were analyzed. Primers are listed in Table S1.
Mapped reads were filtered to obtain uniquely mapping reads using SAMtools software. Uniquely mapping reads were used for all subsequent analyses. Peak calling was done with MACS2 software. Association of DAP-seq peaks located within 2 kb upstream or downstream of the transcription start site (TSS) were analyzed using BEDtools according to the General Feature Format (GFF) files. Visualization of peaks coverage over chromosomes and profiles of peaks binding to TSS regions were analyzed using ChIPseeker software (v1.22.1). Motif discovery was performed using Homer software (v4.11).

Methods S7 Dual-luciferase assay
Firefly luciferase (LUC), under the control of the promoter of CKX1, was inserted into the reporter vector pGreenII 0800-LUC vector (Henriksson et al., 2005). To generate the effector vector, the full-length PpnGRF5-1 coding sequence was cloned into the vector pGWB17, to obtain 35S::PpnGRF5-1-nos construct. Both the effector and reporter constructs were transformed into GV3101. N. benthamiana plants were transfected with both the effector and reporter, as previously described (Wang et al., 2011). Luciferase activity was detected as in 'split luciferase complementation assay'. The luciferase activity quantification was also measured using the dual-luciferase reporter assay system (Promega, USA, cat E1910), and Renilla luciferase was used in normalization. Primers are listed in Table S1. All assays were performed in three replications.

Methods S8 Yeast two-hybrid assays
To examine if PpnGRF5-1 protein can interact with PpnGIFs proteins, the PpnGRF5-1 and PpnGIFs sequences were independently amplified and cloned into vectors pGBKT7 and pGADT7. The plasmids were co-transformed into the yeast strain AH109 mediated by PEG4000. The yeast two-hybrid was performed according to Matchmaker Gold Yeast Two-Hybrid System (Clontech, USA). All assays were performed in three replications. Primers are listed in Table S1.

Methods S9 GST (glutathione-S-transferase)-fusion protein pull-down assay and western blotting
The interaction of PpnGRF5-1 and PpnGIFs was determined by the GST-pulldown assay. The GST, GST-PpnGRF5-1, and PpnGIFs-His6 recombinant proteins, were expressed in the Escherichia coli BL21 (DE3) strain and then purified using GST-tag Protein Purification Kit (Beyotime, China, cat P2262) and His-tag Protein Purification Kit (Beyotime, China, cat P2226), respectively. The equal amounts of GST and GST-PpnGRF5-1 proteins that had been coupled to glutathione-sepharose beads were incubated with recombinant PpnGIFs-His6 with continuous rotation at 4 °C for 1 hour. Washing with phosphate-buffered saline (PBS) was followed to remove non-specific bound proteins for 5 times. After that, the proteins bound to the beads were eluted in sodium dodecyl sulfate (SDS) sample buffer with boiling. The sample mixes were separated in sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto polyvinylidene fluoride (PVDF) membrane (Merck Millipore) by using electro-blotter. The membranes were blocked with 5% (w/ v) skim milk through incubation at 37 °C for 2 hours. The blots were incubated with primary antibodies overnight at 4°C in blocking buffer (5% BSA (bovine serum albumin) add to TBST buffer). Following that, alkaline phosphatase-conjugated secondary antibody that targeted for the specific primary antibody was added. Target bands were visualized by the reaction with nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP). Several antibodies were used are anti-GST (cat ab19256), anti-His (cat ab137839), and IgG H&L (cat ab6722), which were all purchased from Abcam (Cambridge, UK). All assays were performed in three replications. Primers are listed in Table   S1.

Methods S10 Split luciferase complementation assay
The split-luciferase complementation assay was used to test if PpnGRF5-1 and PpnGIFs interact with each other. The coding sequence of all PpnGRF5-1 and PpnGIFs without the stop codon was inserted into the vector (pCAMBIA-NLuc and pCAMBIA-CLuc) to generate the 35S::PpnGRF5-1-nLUC-nos, 35S::cLUC-PpnGIF1-nos, 35S::cLUC-PpnGIF2-nos and 35S::cLUC-PpnGIF3-nos fused constructs. These constructs were transformed into GV3101. Nicotiana benthamiana leaves were co-injected with mixed bacterial cultures using a needleless syringe. The leaves were collected after 3 days under long-day white-light conditions and infiltrated with 150 μg/ml luciferin solution. Luciferase activity was detected with a 30 s exposure time, 4 × 4 binning, slow readout, and high gain using Night SHADE LB 985 system (Berthold,