All of the Arabidopsis stocks used in this study were developed in a Columbia (Col) genetic background. The mutant plants spl2 (SALK_022235), spl10 (SALK_122018) and spl11 (SALK_112209), SPL10 overexpression lines (6mSPL10 and pSPL10-SPL10-GFP) (Nodine and Bartel, 2010), MIM156 seeds (Franco-Zorrilla et al., 2007) were obtained from the Arabidopsis Biological Resource Center (Ohio State University, Columbus, OH). Seeds of the 35S:miR156 were kindly provided by Dr. Detlef Weifel (Wang et al., 2008, 2009). All the Arabidopsis seeds were incubated at 4°C for 3 days in the dark for stratification, and then transferred to a growth room with long day conditions (16 h light, 8 h dark) and set at 23°C, 70% humidity, and a light intensity of 130–150 μmol/m²/s. Plasmid constructs were transformed individually into Arabidopsis ecotype Col-0 using the floral dip method (Zhang et al., 2006).

RNA was extracted from the roots of both WT and 35S:miR156 Arabidopsis plants that were at the 20-day post germination stage. Four biological replicates (independent RNA preparations) were used for each genotype. NGS of the root RNA was performed by PlantBiosis (University of Lethbridge, Canada) under a fee-for-service contract. Using the Arabidopsis genome (TAIR10, http://www.arabidopsis.org/) as a reference, differential gene expression analysis was carried out based on published protocols (Trapnell et al., 2012). Briefly, raw sequencing data were first evaluated with the FastQC program. All filtered and properly paired reads were then mapped to the Arabidopsis genome using TopHat. The fragment alignments generated by TopHat were then used as input files to be further analyzed through the recommended Cufflinks packages to detect the differentially expressed genes between WT and 35S:miR156 Arabidopsis plants.

Plant tissues were collected at specific time points as indicated in results for gene expression analysis. Total RNA was extracted using TRIzol reagent (Invitrogen) and 1 μg was used to generate cDNAs through reverse transcription, using oligo(dT)₁₅ or gene specific reverse primers with a SuperScript® III Reverse Transcriptase kit (Invitrogen™. Expression levels of the selected transcripts were analyzed via qRT-PCR in a total volume of 10 μl and carried out in a 96-well plate on the CX96™ Real-Time PCR Detection System (Bio-Rad, California, United States). Each reaction consisted of 2 μl of cDNA template, 0.4 μl each of both gene-specific forward and reverse primers (10 μM) (Supplementary Table 2), and topped up to 10 μl with water. CBP20 and Tubulin genes were used as internal controls for all qRT-PCR in Arabidopsis (Supplementary Table 2). Each test consisted of three biological sample repeats and each biological sample contained two technical replicates. Finally, transcript levels of the respective genes were analyzed using a relative quantification 2^−ΔCt method (Livak and Schmittgen, 2001).

Leaves from WT and SPL10-GFP transgenic Arabidopsis plants were used as materials for ChIP assays, which were performed according to a previously described protocol using the Chromatin Immunoprecipitation Assay kit (Lot:2382621, Millipore, Billerica, MS, United States) (Gendrel et al., 2005). Briefly, nuclei were isolated from leaves that were cross-linked with 1% formaldehyde under vacuum for 20–30 min and ground in liquid nitrogen. The chromatin solution was then sonicated 3 × 15 s into 500–1,000 bp fragments using a Sonic Dismembrator (Fisher Scientific, Hampton, New Hampshire, United States) set at power 3. Chromatin complexes were incubated with an anti-GFP antibody (Abcam, Cambridge, United Kingdom), and immune complexes were precipitated using Protein A beads. The precipitated DNA was purified and dissolved in water for further qPCR analysis using primers q_nAtAGL79 as listed in Supplementary Table 2. SPL10 occupancy on AGL79 was estimated by comparing the percentage of input (%input) in pSPL10-SPL10-GFP and WT plants (Yamaguchi et al., 2009). The consensus sequence “GTAC” was identified as the core binding motif of SPL proteins (Klein et al., 1996; Birkenbihl et al., 2005). Primers flanking the SPL10 binding core motif GTAC in the promoter region of AGL79 were used to test for SPL10 occupancy. A DNA fragment containing a SBP binding consensus was amplified from an EIF4A1 gene (Shuai et al., 2002) to serve as a negative control. All the primers used for ChIP-qPCR are listed in Supplementary Table 2.

Fresh Arabidopsis leaves (0.1 g) were homogenized in 0.2 ml of protein extraction buffer (0.125 mM Tris, pH6.8, 4% w/v SDS, 18% glycerol, 0.024% w/v bromophenol-blue, 1.43 M β-mercaptoethanol, 0.2% protease inhibitor). After boiling for 10 min, the insoluble fraction was removed by centrifugation, and the supernatant (denatured protein) was separated on a 12% SDS PAGE gel and transferred onto a nitrocellulose membrane, followed by incubation with primary anti-GFP antibody (Abcam, ab290, Cambridge, MA, USA) and secondary goat anti-rabbit IgG HRP (Abcam) antibody. The membrane was developed with Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific, Waltham, MA, USA). The expression of SPL10-GFP fusion protein was also investigated using a Biological Confocal Laser Scanning Microscope FV10-ASW, and the emission wavelength for GFP and DAPI channels are 488 nm and 405 nm, respectively (OLYMPUS, Tokyo, Japan).

SPL10 was reported earlier to play a dominant role in repressing lateral root development (Yu et al., 2015), so we investigated the root phenotypes in WT, pSPL10-SPL10-GFP (SPL10 overexpression under native promoter), miR156OE (miR156 overexpression) and MIM156 (miR156 repression) plants. As early as 10 days after seed germination, some differences could be observed among different Arabidopsis lines. Compared to WT plants, roots of pSPL10-SPL10-GFP plants were shorter with no obvious primary roots (Figures 1A,B), and have fewer lateral root branches (Figures 1A,C). MIM156 plants (where miR156 gene transcripts were suppressed) also had less lateral roots (Figures 1A,C). In contrast, the miR156OE plants showed relatively more lateral roots compared to WT control (Figures 1A,C). These results showed that the expression level of SPL10 is negatively correlated to root development in Arabidopsis.

In order to further identify genes that are involved in the miR156-SPL regulatory network in Arabidopsis, Next Generation Sequencing (NGS)-based transcriptome analysis was carried out on the root tissues of WT and miR156OE Arabidopsis plants. This analysis revealed a range of differentially expressed genes (DEG) between WT and miR156OE roots (Supplementary Table 1). Among all 10 miR156-targeted SPL genes, only SPL10 and its homolog SPL2 were significantly downregulated. Furthermore, a root gene encoding an uncharacterized transcription factor, AGL79 (AT3G30260), was also significantly downregulated with the most prominent fold change (−5.59). Further expression analysis of AGL79 revealed that it was nearly undetectable in leaf tissues of all Arabidopsis lines (WT, pSPL10-SPL10, miR156OE, and MIM156) (Figure 2A). This low leaf expression of AGL79 is consistent with previous reports (Parenicova et al., 2003).

AGL79 gene expression was initially characterized by us in several different Arabidopsis genotypes (WT, pSPL10-SPL10, miR156OE, and MIM156). In roots, AGL79 levels were the highest in MIM156, followed by lower expression levels in pSPL10-SPL10-GFP and even lower levels in WT and miR156OE (Figure 2A). The transcript level of SPL10 was also investigated in the above-mentioned genotypes. SPL10 transcript was detected in both the leaf and root tissues (Figure 2B), and was highly expressed in both tissues of pSPL10-SPL10 and MIM156 plants, with much lower transcript levels in WT and miR156OE (Figure 2B). This expression trend is somewhat similar to that of AGL79 (Figure 2A). The correlation between expression levels of SPL10 and AGL79 suggests that AGL79 may be regulated by SPL10 through the miR156-SPL regulatory pathway.

As the afore-mentioned expression patterns suggested that AGL79 might be regulated by SPL10, further characterization was carried out using ChIP-qPCR to determine if AGL79 is a direct target of SPL10. For that, we characterized transgenic plants expressing the SPL10-GFP fusion protein (pSPL10-SPL10-GFP). Since SPL10 is a known transcription factor, we confirmed its nuclear localization in both the leaf and root tissues using confocal microscopy (Figure 2C). In addition, the SPL10-GFP fusion protein was also detected using western blot analysis (Figure 2D).

The upstream promoter region (2000 bp) of Arabidopsis AGL79 revealed 4 core GTAC sequences that are distributed in three regions (I, II, and III), with all three regions possessing the typical NNGTACR SPL binding consensus (where N = any nucleotide, R = A or G) (Figure 3A, Supplementary Document 1). Strong binding capacity of SPL10 to regions I, II and III was detected by ChIP-qPCR in the pSPL10-SPL10-GFP transgenic Arabidopsis plants (Figures 3B–D). Compared to the WT control, occupancy in these three regions was substantially higher than that in the negative control eukaryotic translation initiation factor 4A1 (EIF4A1) (Figure 3E). Of the three putative SPL binding regions, region III showed a higher binding capacity (Figure 3D). These results show that the SPL10 protein could bind to multiple regions in the AGL79 promoter. Therefore, AGL79 appears to be regulated through the miR156-SPL network to affect plant development in Arabidopsis.

To further investigate the role of AGL79 in Arabidopsis development, we generated transgenic plants with either enhanced or silenced expression of AGL79. AGL79 overexpression Arabidopsis plants (Group 1, see next paragraph), on the other hand, had fewer and smaller rosette leaves, as well as earlier flowering time compared to WT plants at the same developmental stage [Figure 4B, 5B (WT and Group 1)]. SPL10 overexpression plants (6mSPL10) also showed a phenotype similar to that of AGL79OE (Group 1) plants (fewer and smaller rosette leaves) (Figure 4C, Supplementary Figure 2). The phenotypical similarities between AGL79OE and 6mSPL10 plants suggest a potential linear regulatory relationship between AGL79 and SPL10. CRISPR-Cas9 was used to generate mutations in AGL79. Mutated plants were analyzed by Sanger sequencing, which detected mutations or deletions within the 20 bp sgRNA2 sequence regions (Supplementary Figure 1B), resulting in reduced gene expression (Supplementary Figure 1C). A phenotypic comparison between the four lines was carried out when the WT plants reached the bolting stage. Compared to WT (Figure 4A), CRISPR-Cas9-AGL79 mutant plants had more lateral shoot branches (Figure 4D, Supplementary Figure 1A).

To investigate the role of AGL79 in Arabidopsis development, we generated transgenic plants with enhanced expression of AGL79. Compared to WT, the highest AGL79 overexpression plants flowered early and had fewer and smaller rosette leaves [Figures 4B, 5B (WT and Group 1)] much like the SPL10 overexpression plants (6mSLP10) (Figure 4C, Supplementary Figure 1). Transgenic Arabidopsis plants harboring the AGL79 overexpression construct were divided into three groups depending on AGL79 expression. Group 1 (lines L1, L2, and L3) had the highest AGL79 transcript levels in both the leaf and root tissues (Figure 5A), with lower expression in roots relative to leaves. Group 2 (lines L11, L12, and L20) had intermediate AGL79 expression, with variable expression levels between leaf and root (Figure 5A). Group 3 (lines L16, L18, and L27) displayed the lowest AGL79 gene transcripts, and there were no obvious differences in AGL79 transcript levels between the leaf and root (Figure 5A). Different phenotypes could be observed in these AGL79 overexpression plants depending on AGL79 expression levels (Figure 5B). Compared to WT (3 weeks after seed germination), Group 1 plants displayed fewer rosette leaves and early flowering time (Figure 5B). Group 2 plants displayed a phenotype similar to WT (Figure 5B). Group 3 plants showed more lateral shoot branches and a higher number of rosette leaves, as well as a significant delay in flowering (Figure 5B). In addition, the transcript level of SPL10 gene was also investigated in both the leaves and roots of the above-mentioned plants. Although changes in SPL10 expression could be detected in three groups of AGL79OE plants (Figure 5C), these changes did not follow any consistent trend, as found for AGL79 (Figure 5C), suggesting that AGL79 could be a downstream gene regulated by SPL10, and hence fluctuations in AGL79 expression would not affect the expression of the upstream SPL10 gene.

As all the evidence derived from molecular and biological analysis (Figures 2A,B, 4C) revealed that AGL79 is likely regulated through the miR156-SPL pathway, we investigated whether a linear regulatory relationship exists between SPL10 and AGL79. Crossing AGL79OE plants and spl2spl10 double mutant produced F1 progeny showing WT-like phenotype (Figure 8B). The selected genotyping results of the double mutant (spl2spl10) and AGL79 OE plants are shown in Supplementary Figure 3. These results suggest a direct linear relationship between AGL79 and SPL10 genes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.