Regulatory circuits involving bud dormancy factor PpeDAM6

BPC family proteins bind a regulatory intronic region of PpeDAM6 gene

PpeDAM6 was highly expressed in leaf, flower, and vegetative buds and noticeably less in embryo, whereas its expression was practically imperceptible in fruit and flower components (Fig. 1a). The fact that PpeDAM6 was appreciably expressed in tissues that display growth arrest and dormancy mechanisms evidences its patent relationship with these processes. We analyzed PpeDAM6 expression profile along floral bud development in two cultivars with different low (“early”) and medium (“late”) chilling requirements for dormancy release. PpeDAM6 was timely downregulated in these cultivars according to their different estimated dormancy release dates, when their respective chilling requirements were achieved (Fig. 1b).

A region spanning about 1.1 kb of PpeDAM6, containing the first intron, the translation start site, and part of the large second intron of the gene, was found previously enriched in the repressive histone mark H3K27me3 concomitantly with dormancy release²⁵ (Fig. 1c). In order to identify putative regulatory factors that specifically bind to this region, we performed a Y1H approach. This region was divided into two fragments of 558 bp (“Reg1”) and 575 bp (“Reg2”) that were used independently as baits against a cDNA expression library made from mixed dormant and dormancy-released flower bud samples. Reg1 and Reg2 included several CArG box elements (CC(A/T)₆GG motif recognized by MADS-box domain proteins) and two stretches with, respectively, 19 and 9 GA tandem repeats (Fig. 1c and Supplementary Fig. S1). We screened 10⁶ and 5 × 10⁵ yeast transformants with pABAi-Reg1 and pABAi-Reg2 baits, respectively. No positive candidates were obtained in Reg1 screening, whereas two positive clones corresponding to partial sequences of Prupe.1G338500 and Prupe.1G369400 transcripts bound Reg2 fragment containing the start of the second intron of PpeDAM6 (Fig. 1d and Supplementary Fig. S1). By BLASTP analysis against “Peach v2.1” genome database²⁷, we detected an additional peach transcript highly similar to Y1H positive sequences (Prupe.8G082900). The deduced proteins of these genes contain a GA-repeat binding domain, which has been previously described in the BARLEY B RECOMBINANT (BBR)/BASIC PENTACYSTEINE PROTEIN (BPC) protein family. Thus, from now on we will use the names PpeBPC1, PpeBPC2, and PpeBPC3 to designate Prupe.1G338500, Prupe.1G369400, and Prupe.8G082900 genes, respectively.

A phylogenetic tree was constructed using protein sequences of previously characterized BPC genes from Arabidopsis thaliana, Hordeum vulgare, Populus trichocarpa, and Vitis vinifera^28,29. As shown in Fig. 1e, BPC proteins clustered into three groups (I, II, and III), in agreement with previous studies³⁰. PpeBPC1 fell into group I, while PpeBPC2 and PpeBPC3 were part of group II. Within group II, PpeBPC2 clustered with AtBPC6, PtBBR/BPC6, and VvBBR/BPC6, suggesting that PpeBPC2 could structurally and functionally resemble BPC6-like proteins. PpeBPC1, PpeBPC2, and PpeBPC3 gene expression profiles were very similar, showing a slight increase along flower bud development in both early and late flowering cultivars, unlinked to dormancy release dates (Fig. 1f, showing early flowering cultivar data).

By yeast two-hybrid system (Y2H), we confirmed that PpeBPC proteins are potentially able to form heterodimers with each other, as stated in other species (Fig. 1g). However, no interaction was observed with other elements of repressive complexes described as BPC interactors in previous reports, such as the peach orthologs of LIKE HETEROCHROMATIN PROTEIN1 (LHP1), SWINGER (SWN), and SEUSS (Supplementary Fig. S2).

PpeBPC1 represses PpeDAM6 by binding to GA-repeat motifs

In order to determine the DNA-binding specificity of peach BPC factors, we used Y1H strains containing reporter constructs with serial deletions in the Reg2 fragment (Fig. 2a). As shown in Fig. 2b, PpeBPC1, PpeBPC2, and PpeBPC3 only activated reporter with constructs containing at least one of the two GA-repeat motifs found in Reg2, indicating that their interaction with the H3K27me3-enriched region of PpeDAM6 is exclusively mediated by these motifs.

For the purpose of clarifying the role of PpeBPC proteins in PpeDAM6 gene expression regulation, a dual luciferase transient expression assay was performed in Nicotiana benthamiana leaves. We designed effector vectors using the complete coding sequences of PpeBPCs. For constructing reporter vectors with the luciferase gene (LUC) we cloned a PpeDAM6 genomic fragment including the promoter (1 kb), 5’ untranslated region, translation start site, and full first and second introns (Fig. 2c). Three different versions of this vector containing none (Pro.1-LUC), one (Pro.2-LUC), and two GA-repeat motifs (Pro.3-LUC) were used (Fig. 2c). A second reporter expressing the Renilla luciferase gene (REN) under 35S promoter was employed as an internal reference. According to dual luciferase results, there was a slight reduction of LUC/REN ratio when PpeBPC1 was co-infiltrated with Pro.3-LUC vector, suggesting that GA-repeat motifs are necessary for the interaction between the PpeBPC1 protein and the PpeDAM6 regulatory region, and PpeBPC1 could act as a transcriptional repressor of PpeDAM6 (Fig. 2d).

PpeDAM6 overexpression impairs growth in plum

We transformed European plum (Prunus domestica cv. “Claudia Verde,” “CV”) with the constitutive expression vector producing PpeDAM6 with c-myc epitope in its N-terminal end. Since current transformation protocols show low efficiency in peach, European plum offers some advantages over other species for functional studies: its taxonomical proximity to peach and similar bud dormancy behavior³¹. After transformation, three independent plum lines expressing 35S::PpeDAM6 in leaves were identified by quantitative real-time RT-PCR (qRT-PCR). In the three lines, PpeDAM6 was highly expressed and contributed to most of the combined expression of DAM6 genes from both species (PpeDAM6 + PdoDAM6) (Fig. 3a). On the other hand, the expression of plum PdoDAM6 was slightly reduced in transgenic lines compared with the control “CV.” The presence of PpeDAM6 protein was detected by western blot analysis (Fig. 3b). The results showed poor correlation between mRNA and protein expression levels and protein accumulation, since leaves from line #1 showed higher PpeDAM6 transcript expression by qRT-PCR, whereas protein accumulation was higher in line #2 (Fig. 3a, b).

Transgenic lines exhibited drastic alterations in vegetative development. 35S::PpeDAM6 transformed plants were shorter, despite the fact that they developed about the same number of leaves than the control (Fig. 3c). Consequently, internodes were shorter (Fig. 3d). CV was not a true control since the genetic background of transformants differs due to seed segregation of heterozygous parents. However, these alterations were present in the three PpeDAM6 transformants and absent in the different lines produced in the same in vitro procedure with control plasmid and also without plasmidic DNA (CV lines), arguing for a transgene-dependent effect. Unfortunately, most transgenic shoot apices ceased growth few months after plant acclimatization (Fig. 3e, f). Microscopic sections showed a total extinction of the shoot apical meristem (SAM) in plants that ceased growth (Fig. 3g), and reduced SAM dimensions (width and height) in actively growing 35S::PpeDAM6 plants (Fig. 3h). Excised apices of 35S::PpeDAM6 plants showed a concomitant downregulation of SAM development and the organization genes CLAVATA1 (CLV1)-like, SHOOT MERISTEMLESS (STM)-like, and ARGONAUTE10 (AGO10)-like (Fig. 3i). These alterations in meristem proliferation precluded any attempt to obtain reproductively competent 35S::PpeDAM6 plants, and consequently no direct functional evidences about the role of PpeDAM6 on floral bud dormancy could be obtained. In this point, considering that the native expression of PpeDAM6 genes is not constrained to dormant organs and in fact it is highly expressed in leaves (Fig. 1a), we decided to continue the analysis of transgenic leaves and apices in order to achieve general mechanistic clues about the molecular activity of PpeDAM6 in this heterologous model, to be subsequently tested by expression studies in dormant tissues. However, only functional approaches performed in flower buds could confirm the relevance of these mechanisms in the dormancy process.

We analyzed the global expression pattern of leaves from 3-month-old 35S::PpeDAM6 transgenic plum lines #1 and #2 and control “CV” by RNA-seq analysis (three replicates per sample). The transcriptomic sequences were uploaded to NCBI BioProject database (ID PRJNA630876). High-throughput sequencing resulted in 84 million high-quality paired-end reads per replicate (Supplementary Table S1). Clean reads were successfully de novo assembled by Trinity, leading to the identification of 187,901 unigenes (Supplementary Table S2).

The overexpression of PpeDAM6 modified the expression of around 13,000 differentially expressed unigenes (DEUs) in both transgenic lines #1 and #2, from which 6494 were upregulated and 6640 were downregulated in 35S::PpeDAM6 plants (Supplementary Fig. S3a). Eleven Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were significantly upregulated in both lines, whereas 14 were downregulated, among which “ribosome” (ko03010) and “carbon metabolism” (map01200) accounted for the largest proportion of DEUs (Supplementary Fig. S3b, c). Several essential pathways for plant survival and development were downregulated in both transformed lines, such as “photosynthesis-antenna pathway” (map00196), “photosynthesis” (ko00195), “nitrogen metabolism” (map00910), and “carbon fixation in photosynthetic organisms” (ko00710). The analysis of KEGG pathways suggested that PpeDAM6 overexpressing plum lines had lower cellular activity, in agreement with their dwarf phenotype. KEGG enrichment analysis also revealed that “alpha-linolenic acid metabolism” (map00592), involved in jasmonic acid (JA) biosynthesis, was significantly upregulated in 35S::PpeDAM6 transgenic plum, whereas “plant hormone signal transduction” (map04075) was downregulated (Supplementary Fig. S3b, c).

PpeDAM6 overexpression modifies hormones synthesis and response

Subsequently, we evaluated the contribution of hormone-related pathways to the transcriptome of 35S::PpeDAM6 transgenic plants. We found DEUs associated with various aspects of hormone homeostasis and response, mostly related to ABA, cytokinin (CK), GA, and JA hormones (Supplementary Table S3).

The JA biosynthetic genes were found upregulated in both transgenic lines, from 13-LYPOXIGENASE1-like (LOX1-like) to 3-KETOACYL-COA THIOLASE-like (KAT2-like), with the exception of OPC-8:0 COA LIGASE (OPCL) (Fig. 4a, b). Such enhanced expression level of JA biosynthetic genes correlated well with JA and (+)-7-iso-JA-Ile (JA-Ile) hormone content, but we found no difference in the content of the precursor cis-(+)-12-oxo-phytodienoic acid (OPDA) (Fig. 4c).

The expression of CYTOKININ DEHYDROGENASE-like gene (CKX-like), which catalyzes the irreversible degradation of CKs and is thus a key regulator of CK content in plants (Fig. 5a), was highly increased by PpeDAM6 overexpression (Fig. 5b). In close agreement with these results, the content of the CK hormone isopentyl-adenine (iPA) was reduced in leaves of transformed plum plants compared with wild-type “CV” (Fig. 5c).

Likewise, genes involved in GA biosynthesis, catabolism, and signal transduction pathways were identified (Fig. 6a). In GA biosynthetic pathway, ENT-COPALYL DIPHOSPHATE SYNTHASE 1-like (CPS1-like), ENT-KAURENOIC ACID OXIDASE 2-like (KAO2-like), and GA20-OXIDASE 2-like (GA20OX2-like) were downregulated, while the GA catabolic gene GA2-OXIDASE 8-like (GA2OX8-like) was upregulated in transgenic lines. With respect to GA signaling pathway, we found the GA receptor GIBBERELLIN INSENSITIVE DWARF1b-like (GID1b-like) upregulated, while GA-STIMULATED TRANSCRIPT 1-like (GAST1-like) and the GA signaling repressor DELLA1-like were downregulated (Fig. 6b). Despite the fact that gene expression analysis in the GA pathway suggested a reduction of bioactive GA content in transformed plum plants, we could not detect consistent changes in three GAs accumulated at detectable levels (GA₁, GA₄, and GA₁₉) (Fig. 6c). However, the exogenous application of active GA₃ significantly enhanced growth of both transgenic lines, becoming similar to the control “CV” (Fig. 6d).

Within ABA biosynthesis pathway, the genes ZEP-like and VED-like encoding zeaxanthin epoxidase and violaxanthin de-epoxidase enzymes are involved in the production of violaxanthin from zeaxanthin and the reverse conversion, respectively (Fig. 7a). In 35S::PpeDAM6 plants, ZEP-like and VED-like were respectively up- and downregulated compared to CV (Fig. 7b), promoting the violaxanthin production step. However, the expression of a NCED-like gene, codifying for 9-cis-epoxycarotenoid dioxygenase was not significantly altered. Consistently with these data, ABA was over-accumulated in 35S::PpeDAM6 leaves (Fig. 7c). Interestingly, the ABA receptor gene PYL2-like was strongly repressed in PpeDAM6 overexpressing plants (Fig. 7b), suggesting a complex effect on ABA synthesis and response.

Hormone accumulation and gene expression in dormant floral buds of peach

The aforementioned genes and pathways were described in the overexpressing heterologous model of transgenic plum by analyzing transgenic leaves and apices. In spite of the growing body of knowledge about the expression and role of DAM-like genes in leaves and other vegetative tissues (Fig. 1a), since the purpose of this study focuses on the involvement of PpeDAM6 in flower bud dormancy promotion, we analyzed hormone accumulation and gene expression in flower buds of peach, a well-known model. Two cultivars with different flowering time behavior were assayed.

The hormones JA and JA-Ile decreased in floral buds of the late flowering cultivar during the progression of dormancy until dormancy release (first three samples), and also JA-Ile in the early flowering cultivar (Fig. 8a), in accordance with PpeDAM6 downregulation (Fig. 1b). This was in agreement with a higher JA and JA-Ile accumulation observed in leaves of overexpressing PpeDAM6 plum lines (Fig. 4c). However, after this initial drop, JA and JA-Ile levels sharply increased (Fig. 8a), in parallel to known flowering developmental processes occurring during the ecodormancy stage³². The expression analysis of JA biosynthetic genes matched these observations, since AOS-like, AOC1-like, and KAT2-like reduced significantly their expression in the late genotype prior to dormancy release, and LOX1-like, AOS-like, OPR1-like, OPR2-like, ACX-like, and MFP-like were noticeably upregulated after dormancy release (Supplementary Fig. S4).

On the other side, CK levels were coincidently lower in 35S::PpeDAM6 lines and dormant floral buds, in close agreement with their high PpeDAM6 levels (Fig. 8a). However, the late and sharp accumulation of iPR and iPA in ecodormant floral bud samples was not associated with an increase in the CK catabolizing gene CKX-like in floral buds (Supplementary Fig. S4), which argued for the presence of additional mechanisms for the drastic CK overproduction in floral buds prior to bud break.

GA₁ level was not changing significantly during floral bud dormancy, despite the fact that GA biosynthesis gene GA20ox2-like and GA-response gene GAST1-like were upregulated concomitantly with dormancy release and ecodormancy progression (Fig. 8b).

Regarding ABA content, the decreasing hormone level during floral bud development, reaching its lowest stable value after dormancy release in a cultivar-dependent manner (Fig. 8a), consistently matched observations obtained in 35S::PpeDAM6 plants and previous data reported by the literature. The ratio of ZEP-like to VED-like gene expression fairly confirmed that the conversion of zeaxanthin to violaxanthin was also a target of ABA synthesis regulation in floral buds (Fig. 8b), whereas PpeNCED-like expression did not match ABA levels in this tissue (Supplementary Fig. S4).

On the other side, SAM-related CLV1-like and AGO10-like genes showed lower expression values in dormant floral buds where PpeDAM6 was highly expressed, reinforcing the idea that PpeDAM6 affects CLV1-like and AGO10-like regulation in the frame of both flower bud and apical meristem developmental switches.

Most importantly, observed variations in hormone and gene expression values were in every case correspondingly earlier in the early flowering cultivar, confirming that they were dependent on the dormancy stage of floral buds, instead of temperature and other environmental inputs.