Homozygous activation-tagged pap1-D (ABRC stock CS3884) (Borevitz et al., 2000) and control Col-0 seeds (CS70000) were obtained from the Arabidopsis Biological Resource Center. Seed stocks for uncoupling protein UCP single mutants (ucp1, ucp2, and ucp3) and doubly heterozygous ucp12, ucp13, and ucp23 genotypes were the gift of Dr. Ivan Godoy Maia, São Paulo State University, Botucato, Brazil; genotyping and molecular/physiological/phenotypic characterization of homozygous stocks derived from ABRC lines CS874648, SALK_037074, and SALK_123501C (Alonso et al., 2003), respectively, and triple ucp mutant ucp123 genotypes, will be described elsewhere. We used the ucp genotypes as additional biological replicates of the sucrose induction effects on MIRNA expressions; we did not detect any ucp genotype effects on MIRNAs per se (see below; Azad, 2022).

For sucrose induction experiments, 3-day-old Arabidopsis mutants (PAP1-D, ucp1, ucp2, ucp3, ucp12, ucp23, ucp13, and ucp123) and control (Col-0) seedlings were germinated and grown as described (Luo et al., 2012). Following stratification and germination for 3 days, seedlings (~150 seedlings per biological sample) on filter papers in Petri plates containing Murashige and Skoog standard medium (MS medium, one-half strength, control) were transferred to Petri plates containing ½ MS medium plus 200 mM sucrose (6.8% w/v) and allowed to grow at room temperature on a bench under continuous light for 3 days (72h). The other half of seedlings was moved aseptically to ½ MS medium without sucrose, which served as environmental treatment control. After 72h, sucrose-treated and untreated seedlings were harvested by freezing in liquid nitrogen and/or used for various downstream experiments such as ROS assays, anthocyanin quantification, RNA and sRNA extraction, and deep sequencing library preparation.

Anthocyanin quantification of sucrose-treated and untreated control genotypes was done according to the pH differential protocol (Lee et al., 2019). Approximately 100 mg of frozen seedling tissue was pulverized to powder with mortar and pestle in liquid nitrogen, added to 1 mL of extraction buffer (1% [v/v] hydrochloric acid in methanol), and incubated at 4°C overnight. Extracts were centrifuged for 15 min at 15,000 rpm, and supernatant was transferred to a new tube. From there, two solutions were made by adding equal volumes of supernatant to pH 1.0 and 4.5 buffer volume, and absorbance readings were taken at 520 nm and 700 nm for both solutions. Notably, before taking the final readings, several dilutions using pH 1.0 and pH 4.5 buffers were made until absorbance at 520 nm was within the linear range of the spectrophotometer (ThermoScientific Biomate 5). Anthocyanin quantity was expressed as cyanidin-3-glucoside equivalents (mg/L); see Supplementary Methods File for formula).

3,3′-diaminobenzidine (DAB) and nitroblue tetrazolium (NBT) reagents were used to detect hydrogen peroxide (H₂O₂) and superoxide anion (O₂^−.) species, respectively (Kumar et al., 2014). For DAB staining, Thermo Scientific™ Pierce™ DAB Substrate Kit (#34002) was used according to the manufacturer’s protocol. For NBT staining, Invitrogen™ Nitro blue Tetrazolium Chloride (#N6495) was used. For DAB staining, a 1X solution was made according to the manufacturer protocol and for NBT staining, a 0.2% NBT solution was made in an amber-colored bottle by dissolving 0.1 g NBT in 50 mM sodium phosphate buffer (pH 7.5). Ten to fifteen 6-day olds freshly collected sucrose-treated and untreated seedlings were immersed in 1X DAB or NBT solution and vacuum infiltrated for 5 min and subsequently incubated on a shaker for 6h–8h in the dark. Following incubation, the DAB and NBT solutions were drained off and replaced with a bleaching solution (ethanol:acetic acid:glycerol = 3:1:1) and placed in a boiling water bath (~90°C–95°C setting) for 15 min. The bleaching procedure was repeated once more with fresh bleaching solution and samples were allowed to stand for 30 min. Samples at this stage were stored at 4°C or immediately photographed.

Total RNA and sRNAs were extracted from the sucrose-treated and untreated samples (biological duplicates, except sucrose treatment of pap1-D and Col-0 had biological triplicates). Approximately 100 mg of the tissue was used for sRNA extraction with miRPremier microRNA isolation kit (Sigma-Aldrich, Saint Louis, MO), or total RNA from 100-mg aliquots extracted with Spectrum Plant Total RNA extraction kit (Sigma-Aldrich) per the manufacturer’s protocols and quantified using Nanodrop (ND-1000 Spectrophotometer; Thermo Fisher Scientific, Waltham, MA). The isolated sRNA was further quantified on Agilent 2100 Bioanalyzer instrument using small-RNA kit (catalogue #-5067-1548), and total RNA was quantified and qualified (RIN > 6.0) using Agilent RNA 6000 Nano Total RNA analysis kit (catalogue #5067-1511) according to manufacturer’s protocols. See Supplementary Methods File for library sequencing details.

Bioinformatics methods for sequence data analysis were as described previously (Sunitha et al., 2019; Mittal et al., 2023), and scripts were provided in Supplementary Methods File. In brief, the RNA-seq and sRNA-seq libraries were quality assessed using FastQC v0.11.5 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The sRNA libraries were trimming using fastx_clipper tool of the FASTX toolkit (http://hannonlab.cshl.edu/fastx_toolkit/index.html) and reads with length greater than 18 bp were retained, whereas for RNA-seq data, adapter clipping was performed using Trimmomatic (Bolger et al., 2014). Quality-assured sRNA-seq and RNA-seq reads were subjected to sequential filtration steps to remove structural RNAs mapping to Arabidopsis thaliana ribosomal RNAs (rRNAs), transfer RNAs (tRNAs), small nucleolar RNAs (snRNAs), and transposable elements (TEs) (https://ftp.ebi.ac.uk/ensemblgenomes/pub/release-56/plants/fasta/arabidopsis_thaliana/ncrna/, https://www.arabidopsis.org/download_files/Genes/TAIR10_genome_release/TAIR10_transposable_elements/TAIR10_TE.fas) using bowtie-1.1.2 (Langmead and Salzberg, 2012) (Supplementary Datasets S1, S2). Additional publicly available sRNA data were downloaded from National Center for Biotechnology Information (NCBI), processed and quality assured in order to increase the confidence of ShortStack miRNA identification steps (PRJNA110625, PRJNA634468, PRJNA300285, PRJNA251351, PRJNA316991, PRJNA389307, PRJNA413472, PRJNA415623, PRJNA522058, PRJNA190673, and PRJNA560782). The filtered clean RNA-seq reads were mapped using Kallisto (Bray et al., 2016) to a custom Arabidopsis cDNA reference created by adding all of the representative protein coding genes downloaded from TAIR10 and pri-MIRNA sequences downloaded from miRBase (version 22) (Kozomara et al., 2019). The filtered clean sRNA-seq reads were mapped to Arabidopsis thaliana TAIR10 genome using ShortStack version 3.8.5 (Johnson et al., 2016) for de-novo characterization and quantification of Arabidopsis MIRNA loci (Kozomara et al., 2019), explained in detail in Supplementary Methods File.

The raw counts generated by Kallisto and ShortStack were utilized as an input for respective differential expression analysis in DESeq2 R package (release 3.14) (Love et al., 2014). A false discovery rate (FDR) multiple-testing approach was applied (Benjamini and Hochberg, 1995) with default 5% FDR as cutoff. Technical replicates were tested for significantly differential effects, and none were found (data not shown). Principal component analysis (PCA) plots were generated using a web platform iDEP version 1.1 (Ge, 2021). A heatmap of the differentially expressed sucrose responsive sRNAs (miRNAs and tasiRNAs; p-adjusted < 0.007) was generated using TBtools (Chen et al., 2020). For the heatmap analysis, only the high-confidence miRNAs were taken into consideration as per (Taylor et al., 2014; Taylor et al., 2017; Supplementary Dataset S3). In addition, UpSet plots displaying the number of significantly differentially expressed loci (p-adjusted < 0.05) in response to sucrose treatments in control and pap1-D genotypes were generated using UpSetR package (Lex et al., 2014). The sequencing runs have been submitted as raw fastq files to NCBI Sequence Read Archive with BioProject accession PRJNA995345.

To represent the metabolic processes and pathways differentially regulated in response to treatment and genotype(s) as represented by RNA-seq data analysis, genome-wide output from DESeq2 was subjected to MapMan analysis (Schwacke et al., 2019) at p-adjusted < 0.05 cutoff. PageMan is an embedded feature of Mapman that uses over-representation analysis to identify functional categories of biological/metabolic processes and pathways that are significantly over- or under-represented (Usadel et al., 2006). Wilcoxon test was performed on the DESeq2 results using PageMan and applying the most stringent parameter setting (3) and sub-setting the differentially expressed genes based on p-adjusted cutoff of ≤ 0.05, and the results were displayed as interactive heatmaps for the enriched and depleted functional categories and pathways.

700 ng of sRNA from the sucrose-treated and untreated seedling samples were used for sRNA Northern blotting as described in (Mittal et al., 2023). Synthetic DNA oligonucleotides (Supplementary Table S1) (Sigma-Aldrich, St. Louis, MO) complimentary to specific miRNAs were used as probes after 5´-end-labeling using ³²P- γATP, 6,000 Ci/mmol, (PerkinElmer). A 22-nt anti–vvi-miR828 probe was a locked nucleic acid oligonucleotide (Exiqon Inc., Woburn, MA). The sRNA blot band relative intensities were quantified using ImageJ and normalized to loading per lane SYBR Gold-stained bands as validation of reproducibility and linear response of signal strengths. 5S rRNA probe signal and SYBR Gold-stained (Thermo Fisher Scientific) 5S and tRNA abundances visualized on gels used for blots were quantified by ImageJ (Schneider et al., 2012) and used as an internal/loading control for normalization of blot test signal strengths. Numerical signal values for sucrose treatment effects are expressed as the ratio of miRNA probe signal to the SYBR Gold-stained total 5S/tRNA band slices, relative to the loading-normalized signal of control untreated signals set to unity.

Degradome (Gregory et al., 2008; Addo-Quaye et al., 2009; German et al., 2009) datasets for the Arabidopsis thaliana were downloaded from NCBI (SRR6041117, SRR6041069, SRR7652712, and SRR7652709) and were subjected to quality control with FastQC v0.11.5 and, if necessary, adaptor removal and trimming with fastx_clipper. The processed reads were then mapped to structural RNAs in Arabidopsis thaliana, that is, rRNAs, tRNAs, snRNAs, and TEs and the reads mapping to these structural RNAs were filtered out using bowtie2 (Langmead and Salzberg, 2012). Phased, small-interfering RNA-generating loci (PHAS loci) and their candidate miRNA triggers were identified using filtered degradome, ath-miRNAs (from miRBase22), and sRNA library inputs to PhaseTank software (Guo et al., 2015). Filtered reads were then subjected to CleaveLand4 to predict and identify potential AGO cleavage sites (Addo-Quaye et al., 2009). CleaveLand4 implements Generic Small RNA-Transcriptome Aligner (GSTAr) (https://github.com/MikeAxtell/GSTAr) to calculate duplex parameters on RNA-RNA thermodynamics in addition to sequence-based alignment. Outputs generated by CleaveLand4 for all the publicly available degradome datasets were compiled, and miRNA:target interactions with greater than two independent slicing T-plot evidence sources were subjects for further analysis.

In plants, excess sugar accumulation is associated with the accumulation of anthocyanin pigment, which is a stress biomarker (Couée et al., 2006; Sami et al., 2016; Jeandet et al., 2022) mediated in part by crosstalk with ABA stress hormone (Finkelstein et al., 2002; Luo et al., 2012; Rodrigues et al., 2013), but the molecular mechanisms are not well understood. Exogenous 6.8% (w/v) 200 mM sucrose treatment caused accumulation of anthocyanin in both genotypes, with pap1-D genotype having significantly higher anthocyanin accumulation than control Col-0 genotype with or without exogenous sucrose (Supplementary Figure S1), confirming the reproducibility of our earlier study (Luo et al., 2012). pap1-D genotype constitutively overexpresses PRODUCTION OF ANTHOCYANIN PIGMENT 1 (PAP1)/MYELOBLASTOSIS PROTEIN 75 (MYB75) transcription factor (Borevitz et al., 2000) and, thus, serves as a validated check for exploring genetic interactions with hypothesized sRNA effectors of sugar signaling and homeostasis, because it is targeted by miR828:TAS4 module (Rajagopalan et al., 2006), and both MIR828 and TAS4 locus expressions are regulated by a nutrient stress-response MYB75 feedback loop (Hsieh et al., 2009; Luo et al., 2012). Anthocyanins, by virtue of their antioxidant properties function as ROS scavengers and facilitate cellular ROS homeostasis during abiotic stresses (Naing and Kim, 2021). Previous studies have shown that production of ROS can affect various metabolic as well as physiological processes such as photosynthesis, cell differentiation, cell growth and signaling pathways (Miller et al., 2010; Kärkönen and Kuchitsu, 2015; Ribeiro et al., 2017; Zeng et al., 2017). In addition, superoxide functions as a metabolic signal associated with sugar levels (Román et al., 2021). We measured ROS by chromogenic staining of seedlings subjected to high exogenous sucrose stress. Figure 1 shows that sucrose stress treatment resulted in increased production ROS in both Col-0 and pap1-D, as manifested by DAB and NBT staining for H₂O₂ and O₂⁻, respectively, compared to control untreated seedlings. Thus, the expected increase in ROS in response to high sucrose stress treatments that increase anthocyanin accumulation establishes our experimental system as appropriate for assaying mRNA, miRNA, and tasi-RNA abundance changes in response to exogenous sugar stress.

Given the pivotal role of sRNAs in the regulatory networks that govern plant responses to abiotic stresses (Sunkar and Zhu, 2004; Liu et al., 2020; Islam et al., 2022; Mittal et al., 2023) and low concentration (1% w/v) of sugar (Dugas and Bartel, 2008), it is of interest to investigate the impact of high sucrose (200 mM, 6.8% w/v) stress treatment on the seedling sRNAome. The expression of stress-related sRNAs may impact through post-transcriptional gene silencing the capacity to tolerate and adapt to adverse environmental conditions. PCA, as shown in Supplementary Figure S2A, of all sRNA-generating clusters across genotypes and sucrose treatment revealed that treatment was the major variable affecting sRNA abundance. PC1 and PC2 pseudo-dimensions correlated significantly with treatment at p < 0.001 and accounted for ~35% and ~12% of observed expression variation across samples, respectively. 21 nt sRNA species were more abundant than 24 nt sRNA species in both treated and untreated libraries (Supplementary Figure S3), as expected for Arabidopsis at seedling stage (Jung et al., 2015; Tripathi et al., 2019; Aslam et al., 2020; Wu et al., 2021).

Because most miRNAs and phasiRNAs are 21 nt in size and constitute a high proportion of 21 nt sRNA abundance, but not diversity, associated with biotic stress (Källman et al., 2013), we characterized the differential accumulation of miRNAs and phasiRNAs annotated and quantified by ShortStack (Johnson et al., 2016) from seedlings subjected to high exogenous sucrose treatment. The overall sucrose treatment effect was determined using DESeq2 Wald-Log test (Love et al., 2014) on cluster counts for all genotypes (Col-0, PAP1-D, and ucp1/2/3/1,2/1,3/2,3/1,2,3) in the design matrix. Because we found no effect of ucp genotype or ucp genotype-by-sucrose treatment interaction effects on miRNA differential expression (Azad, 2022; Supplementary Dataset S3), the above design matrix provided a high degree of biological replication (n = 18) for a pure sucrose treatment effect on miRNA differential expressions, paired to genotype samples across biological replicate treatments.

A comprehensive analysis revealed that miRNAs from 36 high-confidence (Taylor et al., 2014) miRNA families exhibited significant differential expression in response to exogenous sucrose treatment (overall sucrose treatment effect, p-adjusted < 0.007), as shown in Figure 2A. Several miRNAs have previously been identified as responsive to abiotic stress, specifically nutrient stresses (Liang et al., 2015), and are involved in sucrose signaling (miR156, miR398, and miR408), the phenylpropanoid pathway (miR156, miR828, miR858), response to inorganic phosphate (Pi) (miR399 and miR827), nitrogen homeostasis (miR169), ABA signaling (miR842 and miR169), gibberellin (GA) signaling (miR159), and copper homeostasis (miR398 and miR408) (Dugas and Bartel, 2008; Ren and Tang, 2012; Ma et al., 2015; Meng et al., 2021; Azad et al., 2023). The majority of differentially expressed miRNAs exhibited an upregulation trend, except for the miR156 family, miR163, miR776, and miR8170 (Figure 2A). Although the overall trend in expression among members of the miRNA family was consistent, certain miRNA families such as miR164, miR169, and miR397 exhibited notable exceptions as the members of these miRNA families displayed contrasting trends in expression compared to one another (Figure 2A).

Previous studies have established the importance of PAP1/MYB75 in mediating sugar (glucose and sucrose) signaling and juvenile to adult phase transition. We also observed a significant (p < 0.05) effect of genotype-by-sucrose interaction on the expression of various miRNAs in PAP1-D genotype background (Figure 2A; Supplementary Dataset S3, columns P–R). For example, a positive genotype-by-sucrose interaction was observed for miR156abde, miR162ab, miR169b, miR397ab, and miR399f as accumulation of these miRNA clusters were increased because of sucrose effect in PAP1-D genotypic background as compared to overall sucrose treatment and effect of sucrose in Col-0 (control genotype) (Supplementary Dataset S3, columns D and G). On the other hand, a negative genotype-by-sucrose interaction was observed for miR391 and miR776 as accumulation of these miRNA clusters were downregulated because of sucrose effect in pap1-D genotypic background as compared to overall sucrose treatment and effect of sucrose in Col-0 control genotype (Supplementary Dataset S3).

Apart from miRNA, tasiRNAs generated from TAS loci are also a major constituent of 21 nt sRNAs. An evolutionarily conserved autoregulatory feedback loop affecting miR828-TAS4-PAP1/MYB75 has been shown to play a significant role in the regulation of anthocyanin biosynthesis pathway in response to sucrose treatment (Luo et al., 2012). We also observed that generation of tasiRNAs from TAS4, TAS4-3′D4(-) being the most abundant tasiRNA species, were significantly upregulated (p-adjusted < 0.001) in response to sucrose treatment (Figure 2A, Supplementary Dataset S4).

We independently validated our miRNA and siRNA differential expression claims by small-RNA Northern blotting analysis. We selected small RNAs that exhibited significant differential expression in our study and were previously identified as sucrose responsive (Dugas and Bartel, 2008; Luo et al., 2012; Ren and Tang, 2012), namely, miR398, miR408, miR828, and TAS4-3′ D4 (-). As expected, based on prior claims and in concordance with our sRNA-seq results (Supplementary Dataset S3) we found an over-accumulation for miR408, miR398, miR828, and TAS4-3′ D4(-) showed by increased band intensities in the sucrose-treated samples as compared to untreated samples (Figure 2B). In addition, a relatively higher expression for miR408 (~6×) and miR398 (~1.5×) was found in pap1-D–untreated samples as compared to Col-0 untreated samples (Figure 2B). Although non-significant statistically, our sRNA-seq results reflect the blot-manifested positive genotype-by-sucrose interaction for miR398 and miR408 in the pap1-D genotype (Supplementary Dataset S3). Interestingly, miR828 and TAS4-3′ D4 (-) remained undetectable in untreated control blots, which was consistent with our sRNA-seq findings, as sRNA reads mapping to sRNA clusters corresponding to miR828 and TAS4 were significantly lower in untreated samples compared to sucrose-treated samples (Supplementary Dataset S3, columns BY-DI); Supplementary Dataset S4, columns CG–DQ).

To elucidate how high exogenous sucrose treatment influences the transcriptomic landscape of seedlings, we analyzed RNA-seq data generated from the sucrose-treated and untreated samples. PCA, as shown in Supplementary Figure S2B, of all expressed transcripts across all nine genotypes and treatment, revealed that sucrose treatment was again the major variable significantly represented in PC1 pseudo-dimension accounting for ~50% of variation, as seen for sRNA-seq data from the same sample. It was apparent there was a batch effect across biological replicates represented in PC2 pseudo-dimension due to technical differences, in particular wide variability in rRNA contamination, and two different sequencing lengths (50 bp and 150 bp) for interrogated biological replicate libraries (Supplementary Dataset S1), which directly affected transcriptome read depth and coverage. Notwithstanding, the biological replicates for the genotypes of interest (Col-0 and pap1-D) and treatment were paired factors included in the DESeq2 design matrix to address the batch effect in the differential expression by sucrose treatment analysis. Once again ucp genotypes were not included in the RNA-seq data analysis since no effect of ucp knockout genotypes was observed on miRNA expression across all ucp genotypes in the sRNA-seq data (Azad, 2022; Supplementary Dataset S3) and RNAseq analysis of ucp mutants will be described elsewhere.

RNA-seq data analysis revealed that 8,106 genes were significantly differentially expressed in response to sucrose treatment for the overall sucrose effect paired across Col-0 and pap1-D. Of those 8,106 loci, the sucrose effect in Col-0 manifested as 3,934 DE genes, and the sucrose effect in pap1-D genetic background was 4,449 genes, respectively (Supplementary Figure S4, Supplementary Dataset S5, columns I and L). Supplementary Figure S4 shows an overlap (3,250) in the significantly differentially expressed genes in response to sucrose treatment across all three abovementioned tests, and a ~40% increase in the intersection of pap1-D sucrose effect versus overall sucrose effect (1,189) compared to Col-0 versus overall sucrose effect (675) (Supplementary Figure S4). The difference in gene numbers may reflect in part a genotype-by-treatment interaction of 187 genes (Supplementary Dataset S5, column N) with an observed p < 0.05 for pap1-D as a check on a hypothesized genotype effect, since PAP1 is a TF regulating secondary metabolite anthocyanin biosynthesis (Borevitz et al., 2000). Approximately 50% of the mentioned 187 loci showed a significant genotype by sucrose interaction at p < 0.05 in pap1-D genotypic background (Supplementary Dataset S5, column Q) and as expected, the effectors of flavonoid biosynthesis pathway, members of MBW (MYBs, bHLHs, and WD40s) complex (see below), were found to be affected by both sucrose treatment as well as PAP1 overexpression (pap1-D genotypic effect).

To identify the differentially regulated metabolic pathways in response to high sugar stress in Col-0 and pap1-D genotypes, output from DESeq2 for RNAseq was subjected to genome-wide enrichment analysis using PageMan, an embedded MapMan feature (Usadel et al., 2009; Schwacke et al., 2019). For overall sucrose treatment, the over-represented differential expression bins corresponding to photosynthesis, co-enzyme metabolism, chromatin organization, DNA damage response, RNA processing, protein homeostasis, and translocation significantly over-represented for the genes downregulated in response to high sucrose treatment as shown in Figure 3. Conversely, the bins corresponding to cellular respiration, carbohydrate metabolism, secondary metabolism, phytohormone action, cell wall organization, solute transport, and nutrient uptake were significantly over-represented for the genes upregulated in response to sucrose treatment (Figure 3).

All of the aforementioned bins showed comparable trends for under- or over-representation in response to sucrose treatment in Col-0 and pap1-D (Figure 3). Foyer et al. (2012) have shown that application of exogenous sucrose represses photosynthetic genes via retrograde signaling from plastids to nucleus. Since oxygen and glucose are both needed for cellular respiration, an increase in sugar concentration could result in a decrease in oxygen concentration resulting in inhibition of cellular respiration. An over-representation of upregulated genes mapped to cellular respiration bin can be speculated as a homeostatic response involving mitochondrial retrograde signaling (Barreto et al., 2022).

Previous studies reported the effect of varying sucrose concentrations on the modification of specific flavonoids in various plant species (Loreti et al., 2008; Morkunas et al., 2011; Luo et al., 2012; Kim et al., 2020; Li et al., 2020a; Peng et al., 2020; Lv et al., 2022). Along the same lines, our MapMan results showed that high sucrose treatment activated the whole phenylpropanoid pathway genes leading to enhanced flavonoid and anthocyanin biosynthesis (Figure 4A), which is consistent with metabolic flux regulation described for Arabidopsis as a synergy between the anthocyanin biosynthetic and RDR6/SGS3/DCL4 tasiRNA pathways (Jiang et al., 2020). Upon further examination of the flavonoid biosynthesis bin, we found that expression of several known activators of anthocyanin biosynthesis, namely, MYB75, MYB82, MYB90, MYB113, MYB114, GL3, TT8, MYB11, MYB12, and MYB111 except TTG1 were found to be upregulated, whereas expression of the known repressors of anthocyanin biosynthesis, that is, MYBL2, LBD37, LBD38, LBD39, SPA1, SPA2, SPA3, SPA4, COP1, HY5, and SMXL6 were found to be downregulated in response to high sucrose stress (Figure 5). Exogenous sucrose can have an impact on the processes of starch synthesis, mobilization, and distribution. Furthermore, it may also influence the equilibrium between the biosynthesis and degradation of starch (Fernie et al., 2002). Our MapMan results also indicated high sucrose resulted in increased expression of genes involved in starch biosynthesis. Expression of genes like sucrose synthases (SUS1-6), glucose-1-phosphate adenylytransferases (APL3/4), starch synthases, and starch branching enzymes involved in starch biosynthesis pathway were upregulated in response to high sucrose treatment (Figure 4B; Supplementary Dataset S6). In addition, genes involved in starch degradation and mobilization like β-amylases (BAM2/5) and α-amylases (AMY1/2) were upregulated (Figure 4B; Supplementary Dataset S6).

To investigate the functional effects of differentially expressed miRNAs in response to high sucrose, it is necessary to first identify their targets, as plant miRNAs primarily suppress the expression of their target genes by programming AGO-mediated slicing of mRNAs rather than translational repression (Arribas-Hernández et al., 2016). In addition to a thorough literature review, publicly available Arabidopsis degradome datasets from the seedling stage of development (the tissue studied here, unless otherwise noted) were analyzed using CleaveLand4 (Addo-Quaye et al., 2009) to identify known canonical as well as novel miRNA slicing events. Validated targets were quantified by RNAseq data from the same samples for which miRNA differential expression was characterized (Figure 2) to test for correlations of inferred miRNA activities on target mRNAs (Table 1). We observed miR163:SABATH family (PXMT1, FAMT, FAMT-L, and AT5G38100), miR164:NACs (NAC1L, NAC80, and NAC92), miR167:ARF8, miR169:NF-YA family (NF-YA8, NF-YA5, NF-YA2, and NF-YA10; non-canonical JAZ4), miR393:TIR1, miR396:GRFs (GRF4 and GRF7), miR398:targets (AtCCS, AtBCB, AT3G15640, and AT5G14550), miR408:targets (UCC2 and PAA2), miR827:BAH1 and miR858:MYBL2 modules were significantly mis-regulated by high sucrose treatment in an anti-concordant manner to the DE miRNAs, supporting the hypothesis that these DE miRNAs in response to high sucrose cause DE of cognate miRNA target mRNAs.

In contrast, miR156:SPLs (SPL2, SPL3, and SPL10), miR158:targets (AT1G64100 and AtFUT1), miR159:MYB97, miR397:AtLAC2, miR398:CSD2, miR399:IPS1, miR408:targets (LAC12, LAC13, ARPN, and AT2G47020), miR827:VPT1, miR828:targets (TAS4, MYB82, and MYB113), miR858:MYBs (MYB11, MYB12, and MYB111), and TAS4 tasiRNA TAS4-3′D4(-):MYBs (MYB75, MYB90, and MYB113) were mis-regulated in a concordant manner with their miRNA effector DE (Table 1), suggesting molecular mechanisms may be involved other than observed miRNA abundances as proxy for inferred AGO slicing activities. We also show clear degradome evidences for miR828 and miR858 directing MYB82 and MYBL2 slicing in seedling roots, or flowers, respectively (Figures 6C, D). MYB82 is a predicted yet unvalidated target for miR828 known to play a significant role in the anthocyanin biosynthesis pathway in Arabidopsis. Yang et al. (2013b) showed a decrease of MYB82 transcript level in miR828 overexpression line. Recently, miR828:MYB82 module has been shown to play a significant role in anthocyanin biosynthesis pathway in response to light stress in B. rapa (Zhou et al., 2020). Similarly, we also show that miR828:MYB82 module may also be involved in anthocyanin biosynthesis in response to sucrose treatment (Figure 6C). Previous study shows that miR858a enhances anthocyanin biosynthesis in Arabidopsis seedlings via translational repression of MYBL2, a negative regulator of anthocyanin biosynthetic pathway (Wang et al., 2016). However, our degradome analysis also found miR858-mediated post-transcriptional slicing evidences for MYBL2 target transcripts, suggesting a canonical mechanism, at least in flowers of post-transcriptional control of MYBL2 expression via miR858 (Figure 6D), as is known for the majority of miRNA:target modules in plants (Arribas-Hernández et al., 2016).

The majority of the miRNAs in the aforementioned sugar-deranged miRNA:target modules are also known to be differentially expressed in response to macro nutrient starvation (carbon, nitrogen, and Pi) for miR399 and miR827 and micro-nutrient starvation for elements sulfur (miR395) and copper (miR397, miR398, and miR408) (Table 1). Furthermore, previous studies in Arabidopsis have shown that alamethicin treatment induces miR163 accumulation, implying a role in defense response pathways because alamethicin is a channel-forming fungal peptide antibiotic (Chow and Ng, 2017). This study also claimed increased resistance to Pseudomonas syringae in mir163 mutants, demonstrating a role for miR163 in defense response. miR164: NAC and miR169:NF-YA modules are known to play important roles in biotic and abiotic stress responses in both monocots and dicots (Fang et al., 2014; Luan et al., 2014; Kushawaha et al., 2019; Li J. et al., 2021). All of these key miRNA:target modules, which have been shown to be mis-regulated in response to abiotic stresses such as nutrient stress (Table 1) (Liang et al., 2015), are mis-regulated in response to sucrose treatment. All of the miRNA/siRNA-target modules known to function in phenylpropanoid flavonoid secondary metabolism and anthocyanin biosynthesis showed significant concordant upregulations of the target MYB TFs (except MYBL2). With exogenously supplied high sucrose, it is hypothesized signal transduction results in increased activity of phenylpropanoid pathway leading to increased flavonol and anthocyanin formation (Figure 4A; Supplementary Figure S4). A concordant increase in the miR828 and possibly miR858 accumulation initiates a homeostatic feedback response (Hsieh et al., 2009; Luo et al., 2012) to keep the enhanced production of flavonol and anthocyanins in check.

In addition to the known “canonical” miRNA:target modules, we also uncovered two non-canonical miRNA modules in seedlings that target carbon metabolism-relevant novel mRNAs subject to AGO slicing unrelated structurally to the miRNA cognate family genes, specifically miR408:F3′H and miR398b/c*:OR (Figures 6A, B). With an Allen score of 5 and maximum free energy (MFE) ratio of 0.72, miR408 was shown to slice the early pathway anthocyanin biosynthetic gene F3′H at nt position 583 for splice variant AT3G51240.1 and nt 421 for AT3G51240.2. For miR398b/c*: OR/At5g61670 module, an Allen score of 8 and low MFE ratio of 0.64 is shown functional for the slicing of the OR mRNA at nt 1031. F3’H is coordinately expressed with chalcone synthase and chalcone isomerases and expressed in concordant fashion as miR408 (upregulated), as observed for other known canonical target MYBs involved in flavonol and anthocyanin biosynthesis pathways (Table 1). Arabidopsis OR encodes a close homolog of the cauliflower OR (Orange) DnaJ cysteine-rich zinc-binding domain protein which functions as a molecular chaperone by interacting directly with the Phytoene Synthase protein and is a positive post-translational regulator of Phytoene Synthase expression (Lu et al., 2006; Zhou et al., 2015; Sun et al., 2022). The PageMan bin for carotenoid biosynthesis, a sub-bin of the secondary metabolism bin, was found to be over-represented for differentially downregulated genes in response to sucrose treatment (Figure 3). Increased sucrose availability and abiotic stresses could potentially induce the carotenoid biosynthetic pathways, leading to higher carotenoid production (Uarrota et al., 2018; Choi et al., 2019). Contrary to that, we found that the expression of AtOR significantly downregulated in response to sucrose treatment. Our observation supports the idea that, in addition to presence and absence of a carbon source, other factors, like stage of plant development (Klepikova et al., 2016), or participation of sRNAs including antisense transcript target mimics of MIR398 (Li et al., 2020b), may indirectly regulate the carotenoid biosynthesis pathway. Furthermore, evidence that OR is a bona fide target of miR398*, whose significant accumulation in response to high sucrose treatment (Figure 2, Table 1) likely causes the observed novel slicing and thus reduction in gene expression is one piece of the puzzle that may mediate carbon flux shift from primary to secondary metabolism.