Evolutionary Analysis and Functional Identification of Clock-Associated PSEUDO-RESPONSE REGULATOR (PRRs) Genes in the Flowering Regulation of Roses

Pseudo-response regulators (PRRs) are the important genes for flowering in roses. In this work, clock PRRs were genome-wide identified using Arabidopsis protein sequences as queries, and their evolutionary analyses were deliberated intensively in Rosaceae in correspondence with angiosperms species. To draw a comparative network and flow of clock PRRs in roses, a co-expression network of flowering pathway genes was drawn using a string database, and their functional analysis was studied by silencing using VIGS and protein-to-protein interaction. We revealed that the clock PRRs were significantly expanded in Rosaceae and were divided into three major clades, i.e., PRR5/9 (clade 1), PRR3/7 (clade 2), and TOC1/PRR1 (clade 3), based on their phylogeny. Within the clades, five clock PRRs were identified in Rosa chinensis. Clock PRRs had conserved RR domain and shared similar features, suggesting the duplication occurred during evolution. Divergence analysis indicated the role of duplication events in the expansion of clock PRRs. The diverse cis elements and interaction of clock PRRs with miRNAs suggested their role in plant development. Co-expression network analysis showed that the clock PRRs from Rosa chinensis had a strong association with flowering controlling genes. Further silencing of RcPRR1b and RcPRR5 in Rosa chinensis using VIGS led to earlier flowering, confirming them as negative flowering regulators. The protein-to-protein interactions between RcPRR1a/RcPRR5 and RcCO suggested that RcPRR1a/RcPRR5 may suppress flowering by interfering with the binding of RcCO to the promoter of RcFT. Collectively, these results provided an understanding of the evolutionary profiles as well as the functional role of clock PRRs in controlling flowering in roses.


Introduction
The circadian clock is a time-keeping mechanism in a wide range of organisms controlling endogenous biological rhythms to adapt to 24 h day-night cycles [1,2]. The circadian clock of land plants consists of multiple interconnected transcriptional feedback loops [3,4], wherein the sequential expression of core circadian components assist plants in predicting daily changes in zeitgebers via rhythmic expression of circadian target genes [2]. The first molecular model of circadian clock proposed in Arabidopsis thaliana is comprised of a negative transcriptional-translational feedback loop including two MYB-like transcription factors CIRCADIAN CLOCK-ASSOCIATED (CCA1) and LATE ELONGATED HYPOCOTYL (LHY), and their homologs, REVEILLE8 (RVE8/LHY-CCA1-LIKE5/LCL5) and RVE4, as well as PRR1, PRR3, PRR5, PRR7, and PRR9 [5][6][7][8][9][10]. In the circadian rhythm of Arabidopsis, the translation of CCA1 and LHY occurs in the morning time and mutually interacts to repress the expression of TOC1/PRR1 [5,11,12]. The sequential expression of clock PRR5, PRR7, and PRR9 during daytime form an additional loop showing their partial redundant role in repressing the transcription of CCA1 and LHY and are considered the homologs of TOC1/PRR1 [9]. PRR genes are very substantial in plant circadian rhythm in terms of flowering time, development of inflorescence architecture, and transition of vegetative to reproductive phase [13]. PRR genes are highly conserved in the circadian rhythm of Arabidopsis (TOC1/PRR1, PRR3, PRR5, PRR7, and PRR9) and rice (OsPRR1, OsPRR37, OsPRR73, OsPRR95, and OsPRR59), and both species have the same number of PRR gene family members with RR domain at N-terminal and CCT domain on C-terminal [8,14,15]. PRR genes have received attention in both dicots such as Arabidopsis [16,17], soybean [18], and monocots such as rice [19], wheat [20][21][22], barley [23], sorghum [24], maize [25], yet their evolutionary profiles and biological functions especially related to flowering regulation in woody plants remain largely unknown.
Roses are globally important ornamental plants having a complex and long history of domestication. Roses are commonly found in growing gardens or in vases as cut flowers. Roses are the well-known treasured flowers in the history of mankind, and it has been grown since before the Common Era. Roses that are originally grown for medicinal purposes and perfumes eventually become a valued ornamental flowering plants [26]. There are about 150 wild-growing species of roses in the northern hemisphere alone and may be tens of thousands of rose cultivars grown to date. However, only 8 to 20 of these have been involved in the breeding of four main currently cultivated lineages [27]. Nowadays, rose flower cultivars are becoming more diverse, ranging from more traditional varieties with tall, pointed petals to those with rounded, cup-shaped petals; quarter-shaped petals similar to those of the old rose; or single-petal flowers [26]. Flowering remains the key life event of roses and is greatly affected by various endogenous and environmental signals. Due to continuous flowering behavior, roses have attained special attention with respect to their commercial value and use in the landscape. The release of the genome sequences of 24 Rosaceae species covering 2 major lineages (https://www.rosaceae.org/, accessed on 30 April 2022) [28] enabled us to comprehensively study the evolutionary and functional profiles of circadian clock-associated PRR genes (clock PRRs) in roses.
In this study, we performed genome-wide identification of clock PRRs in Rosaceae in relation to basal angiosperms species, covering the major subfamilies (Rosoideae and Amygdaloideae) of the order Rosales, and initiated a systematic phylogenetic analysis to obtain the overall evolutionary picture of clock PRRs in Rosaceae. We further investigated conserved protein motifs, domains, and gene structure organization by comparative analysis of the protein sequences of clock PRRs to understand their function in roses. Moreover, we also performed co-expression network construction and virus-induced gene silencing (VIGS) of clock PRRs in Rosa chinensis. Our study provided an understanding of the evolutionary profiles as well as the functional role of clock PRRs in controlling flowering in roses.

Clock PRRs Expanded in Rosaceae (Perennial Species) as Compared with Brassicaceae (Annual Species) and Basal Angiosperm
To gain insight into the identification and evolution of clock PRRs in Rosaceae in comparison with basal angiosperms, the Arabidopsis clock PRRs protein sequences were used as queries (See Materials and Methods (Section 4)) against the whole genomes of all the representative species ( Figure 1A). We identified 10 clock PRRs in 2 species of basal angiosperms, 50 in 8 Brassicaceae species, and 157 clock PRRs in 18 Rosaceae species. Within the Rosaceae family, we identified 105 clock PRRs in 11 species of subfamily Amygdaloideae and 52 in 7 species of subfamily Rosoideae, respectively. Among them, the copy number of clock PRRs ranged from 4 to 7 with a mean value of 6.3 in Brassicaceae species, while it ranged from 5 to 19 with a significantly higher mean value of 8.7 in Rosaceae species ( Figure 1B). Furthermore, the range of copy numbers of clock PRRs in subfamilies Amygdaloideae and Rosoideae of Rosaceae were the same (5 to 19), but the mean value for copy number in Amygdaloideae (9.5) was higher than that of Rosoideae (7.4) ( Figure 1C). Moreover, clock PRRs in two representative species of basal angiosperms were 3 in Amborella trichopoda and 7 in Nymphaea colorata. These results indicated that in comparison with basal angiosperms and Brassicaceae, the clock PRRs were significantly expanded in Rosaceae during evolution. Rosaceae species ( Figure 1B). Furthermore, the range of copy numbers of clock PRRs in subfamilies Amygdaloideae and Rosoideae of Rosaceae were the same (5 to 19), but the mean value for copy number in Amygdaloideae (9.5) was higher than that of Rosoideae (7.4) ( Figure 1C). Moreover, clock PRRs in two representative species of basal angiosperms were 3 in Amborella trichopoda and 7 in Nymphaea colorata. These results indicated that in comparison with basal angiosperms and Brassicaceae, the clock PRRs were significantly expanded in Rosaceae during evolution.

All Rosaceae Clock PRRs Were Classified into Three Clades
To obtain a deeper insight into the expansion profiles of clock PRRs in Rosaceae, a maximum likelihood tree (see Materials and Methods (Section 4)) was constructed from a total of 239 identified clock PRRs proteins. Based on the topology of the phylogenetic tree, the evolutionary analysis revealed that the clock PRRs of Rosaceae had three major clades, including PRR5/9 (clade1), PRR3/7 (clade2), and TOC1/PRR1 (clade3) (Figures 2 and S1). Clade 1 was orthologous of AtPRR5 and AtPRR9, clade 2 was orthologous of AtPRR3 and AtPRR7, and clade 3 was orthologous of AtPRR1 (TOC1) of Arabidopsis thaliana. The evolutionary insight further verified that, similar to Brassicaceae, clade 1 was the major clade, followed by clade 2 and clade 3 in Rosaceae. Furthermore, in subfamily Rosoideae, clade 1 was the major clade, followed by clade 3 and clade 2, just like basal Rosids, while in subfamily Amygdaloideae, the major clade was clade 2, followed by clade 1 and clade 2 such as basal angiosperms (Figures 2 and S2). On the basis of these comparative phylogenetic relationships, it was revealed that the clock PRRs in Rosaceae were clustered in the three clades as that of Arabidopsis thaliana, supporting the previously described lineage.

All Rosaceae Clock PRRs Were Classified into Three Clades
To obtain a deeper insight into the expansion profiles of clock PRRs in Rosaceae, a maximum likelihood tree (see Materials and Methods (Section 4)) was constructed from a total of 239 identified clock PRRs proteins. Based on the topology of the phylogenetic tree, the evolutionary analysis revealed that the clock PRRs of Rosaceae had three major clades, including PRR5/9 (clade1), PRR3/7 (clade2), and TOC1/PRR1 (clade3) (Figures 2 and S1). Clade 1 was orthologous of AtPRR5 and AtPRR9, clade 2 was orthologous of AtPRR3 and AtPRR7, and clade 3 was orthologous of AtPRR1 (TOC1) of Arabidopsis thaliana. The evolutionary insight further verified that, similar to Brassicaceae, clade 1 was the major clade, followed by clade 2 and clade 3 in Rosaceae. Furthermore, in subfamily Rosoideae, clade 1 was the major clade, followed by clade 3 and clade 2, just like basal Rosids, while in subfamily Amygdaloideae, the major clade was clade 2, followed by clade 1 and clade 2 such as basal angiosperms (Figures 2 and S2). On the basis of these comparative phylogenetic relationships, it was revealed that the clock PRRs in Rosaceae were clustered in the three clades as that of Arabidopsis thaliana, supporting the previously described lineage.

Response Regulator (RR) Domains Were Conserved in Clock PRRs of Roses during Evolution
On the basis of the comparative phylogenetic relationship, another phylogenetic tree for Rosa chinensis with neighbor-joining model was constructed using the clock PRR of Arabidopsis thaliana and Rosa chinensis via MEGA 11 software ( Figure S3). The basic gene parameters of identified clock PRRs of Rosa chinensis were exhibited in Table S4, including gene ID, location on the chromosome, gene length (AA), molecular weight (Mw), isoelectric point (pI), and predicted subcellular localization. To further support the phylogenetic reconstruction, the identified five clock PRRs from Rosa chinensis, named RcPRR1a, RcPRR1b, RcPRR5, RcPRR3, and RcPRR7 ( Figure S3), were further investigated for conserved domains and motifs, as well as for gene structure organization and chromosomal localization. All the clock PRRs had both RR (response regulator) domain and REC domain at N-terminal and CCT domain at C-terminal, except for RcPRR1b, which had no CCT domain ( Figure 3B). Similarly, five conserved motifs corresponding to RR and REC domains were present in all clock PRRs, whereas motif 2 corresponding to the CCT domain was missing in RcPRR1b ( Figure 3A). Logos of the identified 5 motifs of all clock PRRs were shown in Figure 3D. Furthermore, gene structure organizations of the clock PRRs were further illustrated ( Figure 3C). The figure demonstrated the distribution of coding regions (CDS) and untranslated regions (UTRs) of clock PRRs, indicating that the exons and introns were highly diverse in all genes. RcPRR5, RcPRR3, and RcPRR7 had 8 exons, and RcPRR1a had 6 exons; however, RcPRR1b appeared to be the shortest one among all PRRs with 4 exons. Moreover, the UTRs were also detected in all clock PRRs of Rosa chinensis. The chromosomal localization further revealed that RcPRR3 and RcPRR5 were present on chromosome 1, RcPRR7 was present on chromosome 3, while RcPRR1a and RcPRR1b were present on chromosome 7 of Rosa chinensis ( Figure 3E). These results suggested that RR domains were conserved in roses and thus shared some similar features.

Response Regulator (RR) Domains Were Conserved in Clock PRRs of Roses during Evolution
On the basis of the comparative phylogenetic relationship, another phylogenetic tree for Rosa chinensis with neighbor-joining model was constructed using the clock PRR of Arabidopsis thaliana and Rosa chinensis via MEGA 11 software ( Figure S3). The basic gene parameters of identified clock PRRs of Rosa chinensis were exhibited in Table S4, including gene ID, location on the chromosome, gene length (AA), molecular weight (Mw), isoelectric point (pI), and predicted subcellular localization. To further support the phylogenetic reconstruction, the identified five clock PRRs from Rosa chinensis, named RcPRR1a, RcPRR1b, RcPRR5, RcPRR3, and RcPRR7 ( Figure S3), were further investigated for conserved domains and motifs, as well as for gene structure organization and chromosomal localization. All the clock PRRs had both RR (response regulator) domain and REC domain at N-terminal and CCT domain at C-terminal, except for RcPRR1b, which had no CCT domain ( Figure 3B). Similarly, five conserved motifs corresponding to RR and REC domains were present in all clock PRRs, whereas motif 2 corresponding to the CCT domain was missing in RcPRR1b ( Figure 3A). Logos of the identified 5 motifs of all clock PRRs were shown in Figure 3D. Furthermore, gene structure organizations of the clock PRRs were further illustrated ( Figure 3C). The figure demonstrated the distribution of coding regions (CDS) and untranslated regions (UTRs) of clock PRRs, indicating that the exons and introns were highly diverse in all genes. RcPRR5, RcPRR3, and RcPRR7 had 8 exons, and RcPRR1a had 6 exons; however, RcPRR1b appeared to be the shortest one among all PRRs with 4 exons. Moreover, the UTRs were also detected in all clock PRRs of Rosa chinensis. The chromosomal localization further revealed that RcPRR3 and RcPRR5 were present on chromosome 1, RcPRR7 was present on chromosome 3, while RcPRR1a and RcPRR1b were present on chromosome 7 of Rosa chinensis ( Figure 3E). These results suggested that RR domains were conserved in roses and thus shared some similar features.

Determination of Non-Synonymous (Ka) and Synonymous (Ks) Substitution Rate and Cis-Regulatory Elements Analysis of Clock PRRs in Roses
To find the evolutionary aspects of clock PRRs, a divergence analysis was performed. The non-synonymous substitution per non-synonymous site (Ka) and synonymous substitution per synonymous site (Ks) were determined for two paralogous clock PRRs based on the phylogenetic tree generated via Ka/Ks calculation server to see the evolutionary discretion among clock PRRs ( Table 1). The Ka/Ks value (<1) of the two pairs of genes indicated the purifying selection pressure during the evolution. The divergence time (T) for both pairs of genes ranged from 25.1 to 42.5 million years ago (MYA). The promoter region analysis of rose clock PRRs for the presence of cis-regulatory elements resulted in a diverse range of cis elements. The cis elements were classified into four major categories, including light-responsive elements, hormone-responsive elements, stress-responsive elements, and the elements involved in plant growth and development. The results revealed that light-responsive elements were G-box, GT1-motif, Sp1, and 3-AF1 binding site, and hormone-responsive elements were TGACG-motif, TGA-element, TCA-element, P-box, AuxRR-core, ABRE, GARE-motif, CGTCA-motif, and TATC-box, stress-responsive elements were LTR, MBS, and TC-rich repeats, the

Determination of Non-Synonymous (Ka) and Synonymous (Ks) Substitution Rate and Cis-Regulatory Elements Analysis of Clock PRRs in Roses
To find the evolutionary aspects of clock PRRs, a divergence analysis was performed. The non-synonymous substitution per non-synonymous site (Ka) and synonymous substitution per synonymous site (Ks) were determined for two paralogous clock PRRs based on the phylogenetic tree generated via Ka/Ks calculation server to see the evolutionary discretion among clock PRRs (Table 1). The Ka/Ks value (<1) of the two pairs of genes indicated the purifying selection pressure during the evolution. The divergence time (T) for both pairs of genes ranged from 25.1 to 42.5 million years ago (MYA). The promoter region analysis of rose clock PRRs for the presence of cis-regulatory elements resulted in a diverse range of cis elements. The cis elements were classified into four major categories, including light-responsive elements, hormone-responsive elements, stress-responsive elements, and the elements involved in plant growth and development. The results revealed that light-responsive elements were G-box, GT1-motif, Sp1, and 3-AF1 binding site, and hormone-responsive elements were TGACG-motif, TGA-element, TCA-element, P-box, AuxRR-core, ABRE, GARE-motif, CGTCA-motif, and TATC-box, stress-responsive elements were LTR, MBS, and TC-rich repeats, the responsive elements for plant growth and development were ARE, GCN4_motif, and CAT-box (Figure 4).

Interaction of PRRs with Chemical Compounds and Known miRNAs
To identify the interacting miRNAs targeting the clock PRRs, the coding sequences clock PRRs were used against the miRNAs of Rosa chinensis. It is revealed that 27 miRNAs showed interaction with all clock PRRs except RcPRR3 ( Figure 5). The results further revealed that 17 miRNAs alone showed interaction with RcPRR5, four with RcPRR7, 3 with RcPRR1a, and another 3 with RcPRR1b. Although all the five clock PRRs have a strong interaction among them, however, the miRNAs only targeted four clock PRRs, namely, RcPRR1a, RcPRR1b, RcPRR5, and RcPRR7. The Excel spreadsheet containing miRNAs ID, targeting sites, and alignment with clock PRRs are given in the supplementary file.

Interaction of PRRs with Chemical Compounds and Known miRNAs
To identify the interacting miRNAs targeting the clock PRRs, the coding sequences clock PRRs were used against the miRNAs of Rosa chinensis. It is revealed that 27 miRNAs showed interaction with all clock PRRs except RcPRR3 ( Figure 5). The results further revealed that 17 miRNAs alone showed interaction with RcPRR5, four with RcPRR7, 3 with RcPRR1a, and another 3 with RcPRR1b. Although all the five clock PRRs have a strong interaction among them, however, the miRNAs only targeted four clock PRRs, namely, RcPRR1a, RcPRR1b, RcPRR5, and RcPRR7. The Excel spreadsheet containing miRNAs ID, targeting sites, and alignment with clock PRRs are given in the supplementary file.
To further identify the interaction of some chemical compounds with clock PRRs, the network of clock PRRs with chemical compounds was generated via Cytoscape ( Figure 6). The five clock PRRs showed interaction with 5 different phyto-elements. RcPRR1a, RcPRR3, and RcPRR7 collectively interacted with auxin. RcPRR1b and RcPRR7 showed interaction with ascisic acid, salicylic acid, gibberellin, and methyl jasmonate (MeJA). RcPRR3 showed interaction with ascisic acid, salicylic acid, and gibberellin, along with auxin. RcPRR5 showed interaction with salicylic acid, gibberellin, and MeJA but strong interaction with gibberellin.

Co-Expression Network of Clock PRRs and Flowering Pathway Genes in Rosa Chinensis
To obtain insight into the association of clock PRRs with flowering control genes in roses, the identified clock PRRs were further analyzed via string software (see Materials and Methods (Section 4)) to obtain a visualized map (Figure 7). The transcripts of Rosa chinensis were recognized in the string database with known annotation based on possible co-expressions. It further confirmed that the association of clock PRRs with flowering control genes was not random and connected with a highly significant value of p < 1 × 10 −16 . The co-expression networks were further analyzed for protein-protein interactions (edges/lines) and the shared biological processes. The addition of edges for connecting nodes showed a more significant interaction, while the reducing number of edges represented less interaction among proteins. It was observed that all clock PRRs had a strong association with each other and with another clock-associated gene, CCA1. To further identify the interaction of some chemical compounds with clock PRRs, the network of clock PRRs with chemical compounds was generated via Cytoscape ( Figure  6). The five clock PRRs showed interaction with 5 different phyto-elements. RcPRR1a, RcPRR3, and RcPRR7 collectively interacted with auxin. RcPRR1b and RcPRR7 showed interaction with ascisic acid, salicylic acid, gibberellin, and methyl jasmonate (MeJA). RcPRR3 showed interaction with ascisic acid, salicylic acid, and gibberellin, along with auxin. RcPRR5 showed interaction with salicylic acid, gibberellin, and MeJA but strong interaction with gibberellin.  To further identify the interaction of some chemical compounds with clock PRRs, the network of clock PRRs with chemical compounds was generated via Cytoscape ( Figure  6). The five clock PRRs showed interaction with 5 different phyto-elements. RcPRR1a, RcPRR3, and RcPRR7 collectively interacted with auxin. RcPRR1b and RcPRR7 showed interaction with ascisic acid, salicylic acid, gibberellin, and methyl jasmonate (MeJA). RcPRR3 showed interaction with ascisic acid, salicylic acid, and gibberellin, along with auxin. RcPRR5 showed interaction with salicylic acid, gibberellin, and MeJA but strong interaction with gibberellin.  control genes was not random and connected with a highly significant value of p < 1 10 −16 . The co-expression networks were further analyzed for protein-protein interaction (edges/lines) and the shared biological processes. The addition of edges for connectin nodes showed a more significant interaction, while the reducing number of edge represented less interaction among proteins. It was observed that all clock PRRs had strong association with each other and with another clock-associated gene, CCA1. Moreover, all clock PRRs genes shared some identical biological processes wit flowering pathways genes. The results revealed that all the 13 genes (5 clock PRRs and flowering pathway genes) together mainly contributed to the regulation of biologica processes (e-value 1 × 10 −6 ). Within these 13 genes, all the 5 clock PRRs, CCA1, CO, FT1 and FT2, were closely connected in controlling flowering based on the KEGG pathway association of circadian rhythm (2.7 × 10 −20 ). CCA1 had a strong interaction with cloc PRRs genes and was involved in photoperiodic flowering (e-value 9 × 10 −5 ) and rhythmi process (e-value 3 × 10 −9 ). FT-1 and FT-2 were involved in inflorescence development, an the regulation of timing of the transition from vegetative to reproductive phase (e-valu 0.00), SOC1, FT1, FT2, and CO were mutually involved in the regulation of flowe development (e-value 5 × 10 −5 ). RcPRR1a, RcPRR1b, SOC1, FUL, FT1, and FT2 togethe affected the flower development (e-value 1 × 10 −5 ) and reproductive structur Figure 7. Co-expression network and functional annotation of clock PRRs genes with other flowering pathways genes in Rosa chinensis. Different colors of nodes represent genes sharing biological processes, while the edges/lines connecting the nodes represent the protein-to-protein interaction between genes. The addition of edges/lines between two nodes signifies a more significant interaction. Moreover, all clock PRRs genes shared some identical biological processes with flowering pathways genes. The results revealed that all the 13 genes (5 clock PRRs and 8 flowering pathway genes) together mainly contributed to the regulation of biological processes (e-value 1 × 10 −6 ). Within these 13 genes, all the 5 clock PRRs, CCA1, CO, FT1, and FT2, were closely connected in controlling flowering based on the KEGG pathways association of circadian rhythm (2.7 × 10 −20 ). CCA1 had a strong interaction with clock PRRs genes and was involved in photoperiodic flowering (e-value 9 × 10 −5 ) and rhythmic process (e-value 3 × 10 −9 ). FT-1 and FT-2 were involved in inflorescence development, and the regulation of timing of the transition from vegetative to reproductive phase (e-value 0.00), SOC1, FT1, FT2, and CO were mutually involved in the regulation of flower development (e-value 5 × 10 −5 ). RcPRR1a, RcPRR1b, SOC1, FUL, FT1, and FT2 together affected the flower development (e-value 1 × 10 −5 ) and reproductive structure development (2 × 10 −5 ). To obtain a deeper understanding of the co-expression network, the KEGG pathway of the circadian rhythm was visualized via a direct link of KEGG genome pathways from the String database (https://www.genome.jp/pathway/rcn04712) (accessed on 29 March 2022) ( Figure S4). Taken together, these results suggested that clock PRRs of Rosa chinensis work together with flowering controlling genes.

Silencing of RcPRR1a and RcPRR5 Promoted Flowering in Rosa Chinensis
To further identify the functional role of clock PRRs in controlling flowering, we silenced RcPRR1a and RcPRR5 in Rosa chinensis, followed by checking the flowering phenotype and the expression levels of flowering regulating genes (RcCO and RcFT). qRT-PCR analysis revealed that the expression levels of RcPRR1a and RcPRR5 in silenced lines were significantly reduced ( Figure 8C), and both RcPRR1a and RcPRR5 silenced lines flowered earlier as compared with the control plants ( Figure 8A,B). Consistently, the expression of flower regulating gene RcFT ( Figure 8E) was upregulated as compared with the control. These results directed that clock PRRs had influenced the floral integrator and suppressed flowering. Moreover, no significant increase in the expression level of RcCO was found in silenced lines ( Figure 8D). This alerted us that RcPRR1a and RcPRR5 might have protein interactions with RcCO to interfere with its function during flowering in Rosa chinensis.
silenced RcPRR1a and RcPRR5 in Rosa chinensis, followed by checking the flowerin phenotype and the expression levels of flowering regulating genes (RcCO and RcFT). qRT PCR analysis revealed that the expression levels of RcPRR1a and RcPRR5 in silenced line were significantly reduced ( Figure 8C), and both RcPRR1a and RcPRR5 silenced line flowered earlier as compared with the control plants ( Figure 8A,B). Consistently, th expression of flower regulating gene RcFT ( Figure 8E) was upregulated as compared wit the control. These results directed that clock PRRs had influenced the floral integrator and suppressed flowering. Moreover, no significant increase in the expression level of RcCO was found in silenced lines ( Figure 8D). This alerted us that RcPRR1a and RcPRR5 migh have protein interactions with RcCO to interfere with its function during flowering in Ros chinensis.  Rose GAPDH gene was used as a reference. Three biological replicates were performed for each experiment. Asterisks above the bars indicate significant differences between gene silenced lines and the control as determined by the LSD test, ** p < 0.01 and * p < 0.05.

RcPRR1a/RcPRR5 Physically Interacted with RcCO
To further determine the possibility of RcPRR1a and RcPRR5 genes interfering with RcCO during flowering, we examined the protein-to-protein interactions between RcPRR1a/RcPRR5 and RcCO. We performed split luciferase (LUC) complementation assays by fusing RcPRR1a and RcPRR5 separately to the N-terminal and RcCO to the C-terminal fragments of luciferase, respectively. We infiltrated Agrobacterium tumefaciens cells with these constructs into Nicotiana benthamiana leaves. Luciferase activity was detected only in N. benthamiana leaves co-infiltrated with both 35S:RcPRR1a-N:LUC and 35S:RcCO-C:LUC or 35S:RcPRR5-N:LUC and 35S:RcCO-C:LUC but not in leaves infiltrated with the N-or C-terminal fragment of RcPRR1a, RcPRR5 or RcCO alone (Figure 9). These results confirmed the interaction of RcPRR1a/RcPRR5 with RcCO and suggested the RcPRR1a/RcPRR5 may decrease the free pool of RcCO and interfere with its binding to the RcFT promoter during flowering.
RcPRR1a/RcPRR5 and RcCO. We performed split luciferase (LUC) complementat assays by fusing RcPRR1a and RcPRR5 separately to the N-terminal and RcCO to the terminal fragments of luciferase, respectively. We infiltrated Agrobacterium tumefaci cells with these constructs into Nicotiana benthamiana leaves. Luciferase activity w detected only in N. benthamiana leaves co-infiltrated with both 35S:RcPRR1a-N:LUC a 35S:RcCO-C:LUC or 35S:RcPRR5-N:LUC and 35S:RcCO-C:LUC but not in leaves infiltra with the N-or C-terminal fragment of RcPRR1a, RcPRR5 or RcCO alone (Figure 9). Th results confirmed the interaction of RcPRR1a/RcPRR5 with RcCO and suggested RcPRR1a/RcPRR5 may decrease the free pool of RcCO and interfere with its binding to RcFT promoter during flowering.

Discussion
The circadian clock of plants is referred to an endogenous oscillator regulat various plant physiological processes such as photomorphogenesis, stress responses, a
With the release of whole-genome sequences of Rosaceae, we genome-wide identified the PRR genes in Rosaceae in correspondence with angiosperms species and demonstrated their evolutionary features along with their functional identification. The clock PRRs were significantly expanded in Rosaceae during evolution (Figure 1) but had conserved RR domains, indicating their potential functional conservation in angiosperms as previously described in monocots and dicots species [13,16,42]. The phylogenetic analysis showed that the clock PRRs could be classified into three major clades, including PRR5/9 (clade1), PRR3/7 (clade2), and TOC1/PRR1 (clade3) (Figure 2), and the clock PRRs within each clade were highly correlated to each other [13,42,[44][45][46]. These results facilitated the understanding of PRRs regarding their functional conservation and evolution in Rosaceae.
To explore the evolutionary correlations of rose clock PRRs, the phylogenetic reconstruction of roses revealed the three major clads ( Figure S3) that have also been reported in various earlier studies [44,46]. The further results from domains, conserved motifs, and gene structure further confirmed the structural conservation of clock PRRs in rose plants (Figure 3), while the divergence analysis (Table 1) revealed purifying selection pressure of clock PRRs during the evolution [47]. The PRR family proteins appeared to be unique to plants [48], having an RR domain at N-terminal [8] followed by the additional C-terminal (CCT) motif that could also be found in the CO (CONSTANS) transcription factor [49]. The structure similarity implied the function resemblance, which is further supported by cis-elements analysis (Figure 4), interaction with miRNAs ( Figure 5) [50,51], and the co-expression network and functional annotation of clock PRRs with flower regulating genes ( Figure 7).
Moreover, virus-induced gene silencing (VIGS) of RcPRR1a and RcPRR5 induced early flowering in Rosa chinensis, providing the genetic evidence for the function of clock PRRs in rose flowering regulation (Figure 8). Our results were inconsistent with the previously described role of clock PRRs as negative flowering regulators in rice [52], sorghum [41], and soybean [18]. In Arabidopsis, the decreasing of AtPRR5 was diligently associated with late flowering [52][53][54], defining it as flowering activators. However, in rice, the overexpression of the AtPRR5 homolog OsPRR58 [52] delayed flowering, while silencing of the AtPRR3 homolog OsPRR37 promoted flowering [55]. The decreasing of GmPRR37 in soybean [18] also accelerated the flowering, in line with the result of OsPRR37 in rice. Similarly, the nonfunctional allele of SbPRR37 (AtPRR3 homolog) attenuated the expression of CO but upregulated the flowering activators Ehd1, FT, and ZCN8 [41], suggesting SbPRR37 as a flowering repressor in sorghum. These results indicated the functional divergence of clock PRRs in long-day and short-day plants [18]. In the present study, there was a negative correlation between PRR1a/RcPRR5 and FT [55] (Figure 8). We hypothesized that clock RcPRRs may form complexes with RcCO to control FT transcription [56]. The further results of protein-protein interaction analysis confirmed the protein interaction of clock PRRs with CO ( Figure 9), showing that RcPRR1a/RcPRR5 may decrease the free pool of RcCO and interfere its binding to RcFT promoter during flowering.

Clock PRRs Orthologous Identification in Roses and Angiosperms
The Arabidopsis clock-associated PRRs protein sequences, i.e., PRR1 (At5G61380), PRR3 (At5G60100), PRR5 (At5G24470), PRR7 (At5G02810), and PRR9 (At2G46790), were used as queries in BLASTP searches against the protein sequences of all the representative species of basal angiosperms, basal rosids (Vitals), rosaceae (Fabids), and brasicaceae (Malvids) families. The evolutionary analysis was carried out for basal angiosperms vs. Rosids. Within the Rosids, the species were selected from each subgroup to obtain profound results for Rosaceae. All the sequences with an e-value threshold of 1 × 10 −3 were extracted as candidate clock PRRs orthologs. For further assurance of the clock PRRs orthologs, another BLASTP search was performed using the candidate orthologs as queries against the whole genome of Arabidopsis set as a database. The candidate orthologs with Arabidopsis clock PRRs, as best hits, were identified as clock PRRs orthologs in all the representative plant species of basal angiosperms and Rosids. As Arabidopsis clock PRRs belong to the response regulator (RR) gene family and are among the 32 members RR family (A-type response regulator gene family having 11 members, B-type response regulator gene family having 12 members, and pseudo-response regulator gene family having 9 members) (https://arabidopsis.org/browse/genefamily/ARR.jsp) (accessed on 7 November 2021), the sequences of all the identified clock PRRs orthologs in different species were aligned using MAFFT in corresponding to Arabidopsis RR gene family members and maximum likelihood method implemented in IQ-Tree was used to construct a phylogenetic tree with 1000 bootstraps to identify clock PRRs clade. All the sequences identified as clock PRRs in all species were further confirmed by performing domain annotation using the Pfam database (http://pfam.janelia.org/), SMART database (http://smart.embl-heidelberg.de/), NCBI conserved domain database (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi), and MOTIF search (https://www.genome.jp/tools/motif/) (accessed on 20 November 2021) for CCT and RR (Response Regulator) domains. The sequences without RR (Response Regulator) domain or with partial domain sequences (Table S1) were not used for further analysis [57].

Conserved Domains and Motifs Analysis and Gene Structure Organization of Clock PRRs in Roses
For domain analysis, the clock PRRs protein sequences of Rosa chinensis were subjected to NCBI CDD online software (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) (accessed on 6 February 2022), and the predicted information was then used to visualize domain information via TBtools software (https://github.com/CJ-Chen/TBtools) (accessed on 6 February 2022). Similarly, conserved motif analysis of clock PRRs proteins was accomplished by MEME diagrams. The protein sequences of clock PRRs were submitted to MEME suite software 5.4.1 (https://meme-suite.org/meme/tools/meme) (accessed on 6 February 2022) for 5 conserved motifs identification and then visualized by TBtools software. Correspondingly, the gene structure organization of the clock PRRs were also visualized in TBtools software by submitting the gff3 files of Rosa chinensis along with the identified gene IDs.

Divergence and Cis-Elements Analysis
The server Ka/Ks calculation tool (http://services.cbu.uib.no/tools/kaks) (accessed on 26 June 2022) was used to determine the non-synonymous substitution per nonsynonymous site (Ka) and synonymous substitution per synonymous site (Ks) by inputting the protein DNA sequences of clock PRRs via using default parameters. The divergence time was calculated by the given formula [61]: Synonymous ubstitution rate (dS or Ks) 2 × Divergence rate (6.56 × 10 −9 ) × TMY (10 −6 ) Cis-element analysis was performed by selecting an upstream region of 1500 bp of each genomic sequence of the clock PRR gene and was searched for the presence of cis-regulatory elements in the PlantCARE server (https://bioinformatics.psb.ugent.be/ webtools/plantcare/html/) (accessed on 26 June 2022) [47,62].

Interaction Analysis of PRRs with Chemical Compounds and Known miRNAs
The STRING network was generated for clock PRRs using the STRING server (http: //stringdb.org) (accessed on 26 June 2022) and was further analyzed by the Cytoscape extension of STRING. The targeting miRNAs for clock PRRs were recognized by searching the CDS of clock PRRs against the published miRNAs of Rosa chinensis, downloaded from the specific genome portal (https://lipm browsers.toulouse.inra.fr/pub/RchiOBHm-V2) (accessed on 26 June 2022), through psRNATarget database (accessed on 26 June 2022) and finally visualized via the Cytoscape software (Cytoscape Consortium, USA) (http: //apps.cytoscape.org/apps/stringapp) (accessed on 26 June 2022) [51].

Co-Expression Network Organization of Clock PRRs and Flowering Pathway Genes in Roses
The co-expression network of the PRRs genes with flowering control pathways genes was generated by using the protein sequences of the clock PRRs genes along with the flowering pathways genes via String (String consortium, 2022, version 11.5) (https://cn. string-db.org/) (accessed on 29 March 2022). The protein sequences of important genes involved in the flowering pathway of Arabidopsis were used as queries to BLAST against protein sequences of the Rosa chinensis genome, and the sequences with the e-value of 1 × 10 −10 , detected as best hits, were identified as the homologous sequences. These sequences were further used along with PRRs for their co-expression analysis and functional annotation in the String database against the available transcriptomic data of Rosa chinensis, following the methods previously described by [63].

Virus-Induced Gene Silencing of RcPRR1a and RcPRR5 in Rosa chinensis
To investigate the role of clock PRRs in controlling flowering in roses, RcPRR1a and RcPRR5 from Rosa chinensis were silenced according to the previously described method [64]. Specific gene fragments of the full-length CDS, selected from RcPRR1a and RcPRR5, were amplified using the primers given in Table S2. Vectors were constructed for pTRV2:RcPRR1a and pTRV2:RcPRR5, while empty pTRV2 was used as a mock. The mixture of A. tumefaciens cultures carrying pTRV1 and TRV2:RcPRR1a, pTRV2:RcPRR5 with the ratio of 1:1 (v/v) or with pTRV1 and pTRV2 (empty used as the mock) was vacuum infiltrated into grown cuttings (stem having two nodes). The infiltrated segments of the stem were carefully washed with distilled water and planted into vermiculite for rooting and branching. Leaf samples were collected in liquid nitrogen after 6-8 weeks of infiltration, and the expression level of silenced genes and flowering-related genes were examined via qRT-PCR using the primers given Table S3. RcGADPH was used as a reference gene as described previously [65]. The phenotypic evaluation was also recorded.

Protein-Protein Interactions Analysis of RcPRR1a and RcPRR5 with RcCO
To assay protein-protein interactions, the CDS of RcPRR1a and RcPRR5 without the stop codons was cloned into pMK7-nL-WG2 (http://www.psb.ugent.be/) (accessed on 13 January 2022), while the CDSs of RcCO without the stop codons was cloned into pMK7-cL-WG2. A. tumefaciens strain GV3101 cells carrying 35S:RcPRR1a:LUC-N, 35S:RcPRR5:LUC-N, and 35S:RcCO:LUC-C constructs were co-infiltrated into N. benthamiana leaf epidermal cells to examine the reconstitution of LUC. Three biological replicates were performed for each experiment. Luciferase imaging was performed using a CCD camera (Andor Technology, Belfast, UK). At 48 h after agroinfiltration of N. benthamiana leaf epidermal cells, images were acquired every 10 min for 30 min, and luciferase activity was quantified as the mean counts per pixel per exposure time using Andor Solis image-analysis software (Andor Technology, Belfast, UK) [66,67].

Conclusions
The evolutionary analysis indicated that the clock PRRs were significantly expanded in Rosaceae and classified into three major clades PRR5/9 (clade1), PRR3/7 (clade2), and TOC1/PRR1 (clade3). It indicated that clock PRRs were conserved in Rosaceae and reflected their functional conservation for flowering. Moreover, Rosa chinensis was detected to have 5 clock PRRs (RcPRR1a, RcPRR1b, RcPRR5, RcPRR3, and RcPRR7). The domain and motif analysis further confirmed that clock PRRs had conserved RR domain, and the similar features of gene structure may be due to the duplication events during evolution. Divergence analysis indicated the role of duplication events in the expansion of clock PRRs. To anticipate the functional analysis, cis-element analysis, interaction analysis with miRNAs, and chemical compounds were performed. Co-expression network of clock PRRs showed interaction with flowering regulating genes. Moreover, the phenotypic and genetic evidence of the silenced lines of clock PRRs further confirmed the role of clock PRRs in flowering regulation in roses. Consistently, the protein interaction of RcPRR1a and RcPRR5 with RcCO further explored the involvement of RcPRR1a and RcPRR5 in interfering with RcCO binding to the promoter of RcFT during flowering. It could be concluded that clock PRRs play a crucial role as a critical factor for photoperiodic flowering time.