Introduction

WD40 proteins are a group of transcriptional regulators containing multiple WD40 repeats (Smith et al. 1999; Neer et al. 1994). Each repeat contains a core region ~ 40 amino acids with conserved Glycine-Histidine (GH) dipeptide at the N-terminals and tryptophan-aspartate (WD) dipeptide at C-terminals that are separated by 11–24 amino acids in variable lengths (Stirnimann et al. 2010). Normally, a WD40 repeat contains a four-stranded anti-parallel β-sheet and five to seven such repeats form a bladed propeller scaffold where protein and protein interaction takes place (Andrade et al. 2001; Xu and Min 2011). WD40 proteins normally have additional domains to recruit other factors to form protein–protein complexes (van Nocker and Ludwig 2003; Jain and Pandey 2018; Dayebgadoh et al. 2019). WD40 proteins are involved in many developmental and physiological processes in plants (Wu et al. 2008; Park et al. 2019; Pazhouhandeh et al. 2011; Long et al. 2006; Xu et al. 2019). For example, the Arabidopsis Musashi1 (MSI1) interacts with LIKE HETEROCHROMATIN PROTEIN 1 (LHP1) to inhibit H3K27 methylation (Derkacheva et al. 2013). MSI1 interacts also with CULLIN4-DAMAGED DNA BINDING PROTEIN 1B (CUL4-DDB1) complex to regulate gene imprinting during seed development and floral transition (Dumbliauskas et al. 2011; Pazhouhandeh et al. 2011).

One important representative WD40s is Arabidopsis COP1, the core factor in photomorphogenesis and light signal transduction (Yi et al. 2002; Marine 2012). Arabidopsis COP1 owns a typical seven‐bladed β‐propeller structure and has a zinc finger motif, a coiled helix region, and seven WD40 repeats at C‐terminal. All the three domains play essential roles in protein–protein interactions (Holm et al. 2001; Yi and Deng 2005; Zhu et al. 2008). COP1 promotes ubiquitination and degradation of the Elongated Hypocotyl 5 (HY5) by interacting with SUPPRESSOR OF PHYA (SPA1) and other components of E3 ubiquitin ligase complex, and regulates blue-light signaling via protein complex with Cryptochrome 1 (CRY1) and 2 (CRY2) (Lau and Deng 2012; Liang et al. 2019; Liu et al. 2016; Yang et al. 2000). Furthermore, COP1 can regulate circadian clock and flowering time via interacting with EARLY FLOWERING 3 (ELF3) and GIGANTEA (GI) (Wang et al. 2015). COP1 also interacts with its closet homolog SPA proteins in Arabidopsis to modulate the accumulation of photomorphogenesis-promoting transcription factors in nucleus to tune photomorphogenesis (Lian et al. 2011; Liu et al. 2011; Zuo et al. 2011; Sheerin et al. 2015).

Despite of its pivotal roles in developmental and physiological processes, WD40 genes are moderately conserved with their members systematically identified and characterized in human, wheat, yeast, silkworm and a few other organisms (Zou et al. 2016; Mishra et al. 2014; Salih et al. 2018; Li et al. 2014; Zhu et al. 2015; Hu et al. 2018). However, a thorough investigation of WD40 genes has not been done in horticultural and fruiting plants, for example, Rosaceae plants. Rosaceae includes many important crops, like Fragaria (strawberry), Malus domestica (apple), Prunus persica (peach), Pyrus communis (pear) and rose, which is one of the most important ornamental plants. With the development of a set of genomic resources, roses have now becoming a model woody species for understanding specific traits that not present in current model species (Raymond et al. 2018; Hibrand Saint-Oyant et al. 2018; Li et al. 2018, 2019; Bendahmane et al. 2013; Dong et al. 2017).

In this study, we systematically identified the WD40 family genes in rose (OB). We investigated their chromosome distribution, gene structure, and tissue-specific expression. We found a duplication in COP1-like genes and evaluated their protein interaction spectrum. Our work laid a foundation for further functional exploration of WD40 family in rose and other Rosacaeae plants.

Materials and methods

Plant materials and RNA isolation

OB plants were grown in a glasshouse without additional light in Kunming Botanical Garden, Kunming Institute of Botany, Chinese Academy of Sciences. Leaf materials at just open stage were collected for total RNA extraction with a RNAprep Pure Plant Kit (Tiangen, Beijing) as described by (Li et al. 2018).

RcWD40 protein identification

To identify WD40 proteins in OB, the hmmsearch program (HMMER3.0 package https://hmmer.org/) was employed against the Rose Genome Database (https://lipm-browsers.toulouse.inra.fr/pub/RchiOBHm-V2/) by using the hidden Markov model (HMM) of the WD40 domain (PF00400) as the query file with E value ≤ 10−10 (Eddy 1998). Predicted WD40 protein were manually annotated using BLASTp in NCBI (https://www.ncbi.nlm.nih.gov/). After removing redundant sequences, all candidate proteins were evaluated via Pfam (https://pfam.xfam.org) (Finn et al. 2014) and Smart (https://smart.embl-heidelberg.de/) (Ponting et al. 1999), and rectified by TBtools (Chen et al. 2018) program to eliminate redundant sequences.

Gene and protein structure and physiological characteristics for RcWD40s

The coding sequence, protein sequence, the sequence of 1.5 Kb regions upstream of transcriptional starting site (TSS) and full-length of genomic sequence were downloaded from Rose Genome Database (https://lipm-browsers.toulouse.inra.fr/pub/RchiOBHm-V2). Exon/intron organization was predicted with Gene Structure Display Server (GSDS) (https://gsds.cbi.p-ku.edu.cn/) (Hu et al. 2015). Physicochemical properties were analyzed with EXPASY (https://www.expasy.org/tools/protparam.html) for molecular weight, predicted isoelectric point (PI), sub-cellular localization, negatively and positively charged residues, aliphatic index, grand average of hydrophobicity, and instability index (Wu et al. 2012). The conserved domain was determined by Smart (https://smart.embl-heidelberg.de/) and visualized by Illustrator for Biological Sequences (IBS) (Ren et al. 2009). The cis-regulatory-elements (CREs) for transcription factor binding sites (TFBSs) were predicted by JASPAR (https://jaspar2016.genereg.net/) (Wasserman et al. 2004) for the 1.5 kb upstream regions of TSS.

Chromosomal localization and gene duplication analysis

Physical coordinates of RcWD40s were extracted from the Rose Genome Database (https://lipm-browsers.toulouse.inra.fr/pub/RchiOBHm-V2/) and used for chromosome mapping with MapChart (Voorrips 2002). The criteria of tandem duplication and segmental duplication events were designated as the followings: 1) the physical distance for two or more genes on one chromosome was less than 100 Kb; 2) sequence identity above 80% of the full length; 3) No insertion between the duplicated genes. Genome segments longer than 1 Kb with sequence identity over 90% were considered as segmental duplication (Hanada et al. 2008; Zhao et al. 2013; Leister et al. 2004; Lynch et al. 2000).

Gene ontology (GO) enrichment assays

GO annotation was downloaded from the Rose Genome Databases (https://lipm-browsers.toulouse.inra.fr/pub/RchiOBHm-V2). The online website OmicShare (https://www.omicshare.com/tools/Home/Soft/gogsea) was used to test the statistical enrichment of RcWD40 genes with adjusted p-value less than 0.05 considered as enriched.

Expression analysis

RNA-seq data for OB was retrieved from the database of LIPM (https://www.lipm-browsers.toulouse.inra.fr/plants/R.chinensis) (Dubois et al. 2012) and the heatmap was drawn by using TBtools for visualization (https://www.github.com/CJ-Chen/TBtools). COP1-like sequences were identified with BLASTp against strawberry, apple, peach, and pear genomes and downloaded from the Genome Database for Rosaceae GDR (https://www.rosaceae.org/). RNA-seq reads of COP1-like genes were retrieved from https://www.science.umd.edu/CBMG/faculty/Liu/lab/ using ids gene14041/FvH4_5g22570.1 and gene19736/FvH4_3g01260 for strawberry, and from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE62415 with ids MDP0000245133, MDP0000241199, and MDP0000195882 for apple.

For quantitative real-time PCR (qRT-PCR) analysis, twelve rose tissues including young roots, shoot apical meristem, young shoots, old shoots, young leaves, old leaves, closed flower without sepal, petals at anthesis, stamen, axillary buds, young prickles, and young fruits were collected from four-month old cutting-propagating plants and immediately frozen in liquid nitrogen for RNA extraction. OminiPlant RNA Kit (DNase I; Transgene) was used to extract total RNA with BioPhotometerD30 (Eppendorf) used for quantification. The HiScript® II Q RT SuperMix for qPCR (+ gDNA wiper) (Vazyme, Nanjing) was used for first-strand cDNA synthesis. Gene-specific primers for RcCOP1 and RcCOP1L were designed with primer-blast (https://www.ncbi.nlm.nih.gov/tools/primer-blast) using RcPP2A as reference (Supplementary file: Table S6). PCR was carried out using ChamQ Universal SYBR qPCR Master Mix (Vazyme) on QuantStudio 7 Flex (Applied biosystems) with 384-well format. Relative expression was determined according to Hu et al. (2014) and Klie and Debener (2011).

Phylogenetic analysis of RcWD40s and AtWD40s

For phylogenetic clustering of RcWD40s, we obtained 231 Arabidopsis WD40s from TAIR10 (https://www.arabidopsis.org/) (van Nocker and Ludwig 2003), and aligned them with Clustal X version 2.1. A maximum-likelihood (ML) tree was constructed using MEGA5 with 1000 times of bootstrap replicates (Tamura et al. 2011). A similar procedure was applied for phylogenetic analysis of COP1-like genes in Rosaceae plants.

Protein interaction assays with yeast-two-hybrid (Y2H)

Known and potential protein interaction partners for AtCOP1 were predicted using STRING (https://string-db.org/) (Szklarczyk et al. 2019). The sequences of potential orthologs for these partners were identified with BLASTp using best match in the rose genome. cDNA clones of these rose orthologs were isolated from leaf RNA pools and sequenced using gene specific primers (Supplementary file: Table S6). The GAL4 activation domain (AD) of pGADT7 and the GAL4 DNA-binding domain (BD) of pGBKT7 were digested by Ndel, Xhol and EcoR1, BamH1, respectively. RcCOP1 and RcCOP1L were fused to AD and the potential interacting proteins were fused to BD using Clone ExpressIIOne Step Cloning Kit (Vazyme, Nanjing, China). Protein pairs were then co-transfected into Y2H gold yeast strains, and selected according to procedures described in manual of Matchmaker Gold Yeast Two-Hybrid system (ClonTech, Beijing, China). The positive interaction between AtAGL16 and AtSVP was used as positive control (Hu et al. 2014). Final interaction pattern was photographed after incubation at 28 °C for 3 days.

Bimolecular fluorescent complimentary assay (BiFC)

BiFC analysis was carried out according to Hu et al. (2014) using pFGC-cYFP and pFGC-nYFP vectors (Kim et al. 2008). Full-length CDSs of RcCOP1 and RcCOP1L were inserted into pFGC-nYFP vectors, while RcHY5 and RcSPA4 CDSs were cloned into pFGC-cYFP vectors. See Supplementary Table S6 for primers information. All plasmids were introduced into Agrobacterium tumefaciens strain GV3101 and later used to test the pairwise interactions in Nicotiana benthamiana (tobacco) leaves following protocols described in Hu et al. (2014). YFP fluorescence signals were observed and documented under a confocal laser scanning microscope (Olympus Fluoview Ver. 2.0c Viewer).

Results and discussion

Rose genome has 187 WD40 proteins featuring different properties

We identified 187 WD40 proteins in rose genome mainly based on sequence similarity for WD40-domain, and named them as RcWD40-1 to RcWD40-187 following their positions on rose chromosomes for convenience (Fig. 1; Supplementary file: Table S1) (Raymond et al. 2018).

Fig. 1
figure 1

Uneven distribution of RcWD40s on rose chromosomes. a RcWD40s distribution on seven chromosomes (indicated above). Ticks on each chromosome marked the RcWD40s. Tandem duplicated genes were marked in red and the two RcCOP1-like genes were labelled by blue dots. Scale bar was in Megabase (Mb). Note the uneven distribution of RcWD40s on all chromosomes. b Statistics of RcWD40s on rose chromosomes. Y-axis on the left showed the number of RcWD40s on each chromosome, while the Y-axis on the right side gave the number of RcWD40s per Mb of each chromosome

Full size image

RcWD40 proteins varied dramatically in their WD40 domain, length, size and other physicochemical properties. Their length ranged from 84 to 1756 amino acids with the Isoelectric Point (IP) values being between 4.39–9.48 and Molecular Weights (MW) being between 9,735 Da-364,711 Da. The number of WD40 domains varied between 1–13. RcWD40 proteins had numerous atypical WD40 domains, while 98 WD40 proteins were stable and the remaining 89 proteins being unstable (Wu et al. 2012). Other physical and chemical information about RcWD40 proteins were included in Supplementary file: Table S1.

Uneven distribution on rose chromosomes and gene duplication

RcWD40 genes located extensively and unevenly on the seven rose chromosomes (Fig. 1a; Supplementary file: Table S1). Chr2 harbored 35 RcWD40s with about 0.40 gene per Mb genome sequence, and Chr6 and Chr7 had 33 (0.47 gene per Mb) and 32 (0.46 gene per Mb) genes, respectively. In contrast, Chr1 featured only 18 genes, the least abundant chromosome with about 0.26 gene per Mb genome sequence (Fig. 1b).

We further observed an uneven distribution of RcWD40 genes on different chromosome arms (Fig. 1a). Except Chr3, on which RcWD40 genes were more likely present in the center, all the other six chromosomes seemed have RcWD40 genes at either upper (Chr5) or lower (Chr1, Chr4, and Chr6) or both (Chr2 and Chr7) chromosome arms. The distribution of RcWD40 genes appeared anti-collocating with the distribution of transposable elements on chromosomes (Raymond et al. 2018; Hibrand Saint-Oyant et al. 2018), but was consistent with previous observation in other species (Zou et al. 2016; Mishra et al. 2014; Salih et al. 2018; Li et al. 2014; Zhu et al. 2015; Hu et al. 2018; Ouyang et al. 2012).

We next examined whether tandem duplication and segmental duplication contributed to amplification of RcWD40 genes. We identified 11 members (5.9%) on Chr1, 5, 6 and 7 showing a signature of tandem duplication event (Fig. 1a). No RcWD40 gene was duplicated via potential segmental genome duplication (Zhao et al. 2013; Leister et al. 2004; Lynch et al. 2000). These results implied that, in addition to random duplication, tandem duplication played a role in amplification of RcWD40 genes in rose.

RcWD40 proteins feature different structure and function

RcWD40 genes varied significantly in their exon and intron organization (Supplementary Fig. S1; Supplementary file: Table S1). The gene featuring maximum number of introns was RcWD40-181, which had 38 introns, while 24 RcWD40 genes had no intron. The introns number of other genes ranged between 2 and 14.

We classified the RcWD40 proteins into 15 subfamilies based on their domain structure. There were 127 proteins only containing WD40 domain and were grouped as subfamily A. The remaining 60 RcWD40 proteins had additional domains and were classified into subfamily B to O, respectively (Fig. 2). Gene ontology (GO) annotation revealed that RcWD40 genes were involved in many aspects of biological processes like metabolism and development as well as responses to stimulus and rhythmic regulation (Supplementary Fig. S2; Supplementary file: Table S2). Interestingly, RcWD40 genes were significantly enriched for GO terms related to DNA repair or integrity, root or shoot development, and responses to shade and others stimulus (Fig. 3; Supplementary file: Table S3).

Fig. 2
figure 2

Diversified protein domain structure of RcWD40s. The WD40 repeats were shown in blue, while the other domains were marked in different colors. Texts in red above each cartoon indicated the types of domains. For each subfamily (from A to O), numbers in brackets gave the protein numbers for that subfamily, which was indicated by a representative protein shown below

Full size image
Fig. 3
figure 3

GO enrichment profile for RcWD40 genes. The size and color represented the range of the gene number and the -log10(FDR), respectively. BP, CC and MF represented biological process, cellular component and molecular function, respectively. See Supplementary File: Table S3 for detailed information

Full size image

Rose WD40 proteins experienced significant expansion and contraction along specific phylogenetic lineages

To better understand the evolution of rose WD40 proteins, we conducted a phylogenetic analysis with the 187 RcWD40 proteins and 231 Arabidopsis proteins (Fig. 4). Rose and Arabidopsis WD40 proteins could be grouped into five clusters (I-V), which contained 15, 45, 33, 40 and 54 rose and 39, 60, 34, 28 and 70 Arabidopsis proteins, respectively. In clade I, rose experienced a significant contraction (Chi-square test, p = 0.018) of WD40 genes which were involved in peptidyl-serine dephosphorylation, phagophore assembly, autophagy of nucleus and mitochondria, and trichome differentiation (Supplementary Fig. S3a). Clade IV showed a significant expansion (Chi-square test, p = 0.031) of rose WD40 genes that were involved in histone acetyltransferase activity (Supplementary Fig. S3b). These data suggested that rose had undergone a significant WD40 genes expansion and contraction along specific phylogenetic lineages.

Fig. 4
figure 4

Phylogenetic clustering of WD40s from rose and Arabidopsis revealed lineage-specific gene expansion and contraction with important function in rose. A maximum-likelihood tree clustered the 187 RcWD40s (marked by red circles) and 231 AtWD40s into five clusters (Cluster I-V), which were labelled with different colors. Note that cluster I and IV experienced a lineage-specific contraction and expansion in rose, respectively. Light blue triangles indicated the duplicated RcCOP1-like genes, while numbers on branches showed the bootstrap support values in percentage

Full size image

RcWD40 genes featured spatial and temporal expression profiles

We explored the spatial and temporal expression profile for RcWD40 genes using the expression data for 11 rose tissues (stamen, dormant axillary buds, active axillary buds, floral buds, white young roots, young leaves and stems, early floral organs, closed flower, open flower, senescent flower and fruit) (Supplementary file: Table S4). Figure 5 showed the heatmap of 186 RcWD40 genes in these tissues. Accordingly, we classified the RcWD40 genes into four main groups (I-IV).

Fig. 5
figure 5

RcWD40s featured spatial and temporal expression profiles. A heatmap showed the Log2 values for reads per million kilobases (RPKM) in 11 rose tissues indicated below. RcWD40s were grouped into four main clusters, which were indicated by I-IV, respectively. Cluster IV was further grouped into three subgroups that were labelled with IV1-3. Note that cluster I and III showed the lowest and highest expression levels, respectively. Expression profiles for the duplicated RcCOP1-like genes were boxed in red (see also Supplementary Fig. S4)

Full size image

The group I (32 genes) was the group with lowest expression in most of the tissues analyzed. Group II (44 genes) showed a relatively higher expression than group I. Group III (20 genes) had the highest expression, while group IV (90 genes) displayed a moderate but highly variable expression. Group IV was further divided into three subgroups: IV1, IV2 and IV3, which included 27, 23 and 40 genes, respectively. Those genes were more detected in special tissues, indicating that they might participate in specific physiological and biological processes under certain conditions.

Rose had two COP1-like genes with diversified expression patterns

The E3 ubiquitin ligase COP1 contained seven WD40 repeats and played essential roles in photomorphogenesis and many other biological processes (Yi et al. 2002; Marine 2012; Liu et al. 2016; Yang et al. 2000). In our phylogenetic analysis, we identified two copies of COP1-like genes showing distinct expression pattern in rose (Figs. 4, 5; Supplementary Fig. 4), in contrary to previous notion that angiosperms normally have one copy of COP1 (Artz et al. 2019). We renamed RcWD40-176 as RcCOP1 and RcWD40-96 as RcCOP1-like (RcCOP1L) based on their sequence similarity and phylogenetic relationship with AtCOP1 (Fig. 4). RcCOP1 featured the lowest expression in white young roots and dormant axillary buds and the higher expression in stamen, active axillary buds, and fruit, while RcCOP1L expression was detected in stamen at the lowest level and in the senescence flower at the highest level, indicating that these two genes had temporal and spatial expression (Fig. 5 and Supplementary Fig. 4). A further quantitative real-time PCR (qRT-PCR) assay using 12 different tissues also revealed a similar pattern (Fig. 6a). Thus, RcCOP1 and RcCOP1L might have undergone functional diversification post their duplication.

Fig. 6
figure 6

The duplicated COP1-like genes in rose and other Rosaceae plants featured distinct expression pattern. a Quantitative RT-PCR analysis revealed a differential expression of RcCOP1 (red bars) and RcCOP1L (blue bars) in twelve tissues of OB. RcPP2A was used as reference. Mean relative expression with standard deviation was shown. Significant variation was estimated with Student’s t-test, *, p < 0.05; **, p < 0.01; *** p < 0.001. b Phylogenetic clustering of COP1-like genes in Arabidopsis and five Rosaceae plants. COP1s and COP1Ls were indicated in red and blue colors, respectively. c A simplified model showing the duplication of COP1-like genes occurred prior to the split of Rosaceae plants from their common ancestors (pointed by an arrow). Numbers after each species indicated the copy numbers of COP1-like genes. d Diverged expression of strawberry COP1 (FvCOP1, red bars) and COP1L (FvCOP1L, blue bars) in 13 tissues. e Expression of apple COP1s in lateral and central seeds. Note that MdCOP1L1 showed almost no expression in the tissues examined. f Predicted cis-regulatory-elements (CREs) in the 1.5 Kb promoter regions of RcCOP1. g Predicted cis-regulatory-elements (CREs) in the 1.5 Kb promoter regions of RcCOP1L. Only CREs specific for RcCOP1 (red) and RcCOP1L (blue) were shown with names given above each tick in f and g. For d and e, RPKM values for each tissue were given

Full size image

We next asked whether the duplication of rose COP1-like genes occurred also in other Rosaceae plants. We identified the COP1-like genes from strawberry, apple, peach and pear and performed a phylogenetic reconstruction (Fig. 6b). Interestingly, all these Rosaceae species had two or three copies of COP1-like genes, and accordingly we named them as COP1s and COP1Ls (Fig. 6b). Therefore, duplication of COP1-like genes seemed occurred prior to the split of Rosaceae plants from their common ancestor (Fig. 6c). More importantly, post the duplication, strawberry and apple COP1-like genes have also experienced expression diversification in different tissues (Fig. 6d, e).

RcCOP1s had different cis-regulatory-element (CRE) profiles

We continued to ask whether the diverged expression of rose COP1-like genes correlated with the divergence in CRE profiles by identification of known transcription factor binding sites (TFBSs) in the 1.5 kb regions upstream of transcriptional starting site (TSS). We detected 20 TFBSs shared between RcCOP1 and RcCOP1L, while found 13 TFBSs only present in RcCOP1 and 7 TFBSs only present in RcCOP1L (Fig. 6f, 6g; Supplementary file: Table S5). Notably, the RcCOP1 specific CREs were likely bound by different WRKY and NAC transcription factors, while the RcCOP1L specific CREs were related to SQUAMOSA-LIKEs (SPL4/SPL5), MYB, GARP (KAN1) and AP2/B3 (RAV1) families of transcription factors (Fig. 6f, 6g), suggesting a different regulatory potential of COP1-like genes post duplication.

RcCOP1 proteins differed in their protein partners

As one of the key factors controlling photomorphogenesis and many other developmental and physiological processes, AtCOP1 was known to form protein complexes with a set of partners like HY5, CRY1, CRY2, FUSCA 6/9 (FUS6/9), SUPPRESSOR OF PHYA-105 1/4 (SPA1/4), UVB-RESISTANCE 8 (UVR8), and DDB1B (Kim et al. 2017). Furthermore, COP1 could form homodimers (Torii et al. 1998; Stacey et al. 2000; McNEllis et al. 1996; Xie et al. 2015). To investigate further the functional potential of RcCOP1s post duplication, we first examined the protein sequence divergence between RcCOP1s and AtCOP1 (Fig. 7). Like AtCOP1, both RcCOP1 and RcCOP1L had seven WD40 repeats and one RING-domain at the N-terminal and one coiled-coil region between RING and WD40 repeats. All these domains seemed relatively conserved than rest segments of the three proteins. However, RcCOP1L differed in many positions including the WD40 repeats, a domain where COP1 hetero-dimerize with HY5 (Holm et al. 2001; Osterlund et al. 2000), and the coiled-coil domain, where SPA4 interacts with COP1 (Laubinger and Hoecker 2003; Yi and Deng, 2005), thus suggesting a different interacting spectrum from that of RcCOP1 (Fig. 7).

Fig. 7
figure 7

RcCOP1 and RcCOP1L had significant sequence variation in different domains. Identical amino acids between AtCOP1, RcCOP1 and RcCOP1L were shaded in dark. Grey shadings represented positions with similar amino acids. The RING-finger domain, the coiled-coil region, and the seven WD repeats were underlined in dark lines. The different amino acids between RcCOP1 and RcCOP1L were marked with red boxes, and the amino acids variation between AtCOP1 and RcCOP1/RcCOP1L were marked with blue boxes

Full size image

To further substantiate this, we tested the interaction spectrum of RcCOP1s with yeast-two-hybrid (Y2H) approaches. We cloned the corresponding sequences of ten rose homologs to Arabidopsis partners (Fig. 8a) and tested their interaction potential with RcCOP1 and RcCOP1L (Fig. 8b). Our assays confirmed the positive interaction between AGL16 and SVP of Arabidopsis (Hu et al. 2014) and the negative interactions between AD-RcCOP1/RcCOP1L and BD-empty vector (Supplementary Fig. S5). The same to AtCOP1, both RcCOP1 and RcCOP1L could form homodimers, while at the same time they could also interact with each other to form heterodimers (RcCOP1-RcCOP1L) (Torii et al. 1998; Stacey et al. 2000; McNEllis et al. 1996; Xie et al. 2015). Both proteins formed complexes with CRY1, CRY2, FUS6, FUS9, and SPA1, indicating that both proteins could participate in the same biological processes. Corroborating with its sequence variation in the WD40-repeats and coiled-coild domain (Fig. 7), RcCOP1 could interact physically with HY5 and SPA4 proteins of rose, while RcCOP1L did not, suggesting a functional diversification between these two proteins. A further bimolecular fluorescent complimentary (BiFC) assay further substantiated this pattern (Fig. 8c). Both proteins could not form complex with UVR8 or DDB1B, a contrasting pattern to that of AtCOP1 (Fig. 8b). Taken these together, the two RcCOP1-like proteins shared most of their partners while functional divergence could be happened.

Fig. 8
figure 8

RcCOP1 and RcCOP1L featured diversified interactomes. a A protein interaction network for Arabidopsis COP1. b Interaction spectrum of RcCOP1/RcCOP1L tested with yeast-two-hybrid assays. The interactions for Arabidopsis AGL16-SVP and empty vector on BD with AD-RcCOP1s, were used as positive and negative controls, respectively. Note that RcHY5 and RcSPA4 did not interact with RcCOP1L in comparison to RcCOP1. See methods part and supplementary Fig S5 for detailed information about negative controls. c A BiFC assay confirmed the interactome variation between RcCOP1/RcCOP1L with RcHY5/RcSPA4. Empty vectors with one-side fusion proteins were used as negative controls

Full size image

Conclusion

WD40 proteins were famous for their pivotal roles in plant development and physiological processes. In this study, we systematically identified and characterized 187 WD40s from rose and detected a significant structural variation among these proteins (Figs. 1 and 2). Rose had likely undergone a lineage-specific WD40 genes expansion and contraction (Fig. 4), a similar pattern to that of MADS-box family transcription factors (Liu et al. 2018).

More importantly, we found that rose had duplicated its COP1-like genes with a strong expression and functional divergence. RcCOP1 differed significantly from RcCOP1L with their distinct expression in stamen and several other tissues/organs, a similar pattern also observed in other Rosaceae plants like strawberry and apple (Figs. 5 and 6; Supplementary Fig. S4). Interestingly, the duplication and expression diversification seemed occurred prior to the split of Rosaceae plants from their common ancestors (Fig. 6). Corroborating with the expression divergence between the two rose COP1-like genes, CREs for both genes differed obviously (Fig. 6). Furthermore, the two rose COP1-like proteins had different partners in forming protein complexes, indicating again the strong functional diversification (Fig. 8).

Taken these together, our analyses showcased a rose-specific evolution pattern for WD40 family with COP1-like genes as an example. Our efforts should pave the way to further understand the molecular regulatory mechanisms of biodiversity in roses as well as in other Rosaceae species (like apple and strawberry), which is one of the most important plant families for human being.