Genome-wide cyclin gene evolution in Arabidopsis and Brassica reveals polyploidization-driven duplication and flowering-time associations Authors: Aldrin Y Cantila, Sheng Chen, Kadambot H M Siddique, Wallace A Cowling Published: 2026 ABSTRACT AbstractCyclin genes are plant cell cycle regulators that play essential roles in growth, development, and reproduction. However, the evolutionary dynamics and genomic organization of cyclin genes across the Brassicaceae family remain poorly understood, particularly in the context of allotetraploid genome evolution. Here, we investigated the diversity, expansion mechanisms, and potential functional diversification of cyclin genes across ten Brassicaceae genomes, including four Arabidopsis and six Brassica species. A total of 1087 cyclin genes representing 23 cyclin types were identified. Comparative genomic analyses revealed that cyclin gene expansion was strongly influenced by polyploidization in Brassica species, with 1845 duplication events involving 1063 genes. Whole-genome duplication was the predominant mechanism driving expansion, while both inter- and intra-genomic duplications contributed to gene retention in tetraploid Brassica species, with the highest duplication frequency observed in Brassica juncea. Across genomes, 120 physical gene clusters were identified, including homogeneous and heterogeneous types. Ortholog analysis between progenitor and allotetraploid species identified 852 orthologous pairs involving 366 genes, indicating extensive conservation following allotetraploid formation. Phylogenetic analysis resolved cyclins into three major clades, while expression-based clustering in Brassica napus grouped genes into four major clusters, suggesting functional diversification. Integration of pan-genomic and flowering-time QTL analyses further identified two cyclin genes, Bna21cycA2 and Bna113cycD4, which contain amino acid polymorphisms and represent putative candidate variations potentially associated with flowering-time variation across multiple genomes. These findings provide new insights into the evolutionary expansion, retention, and potential functional divergence of cyclin genes in Brassicaceae and highlight candidate loci for future functional studies and crop improvement.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s00438-026-02515-y.Keywords: Brassicaceae family, Cyclin, Duplication events, Evolutionary mechanism, Pan-genome, Physical clustering Introduction The Brassicaceae family, encompassing over 372 genera and 4060 species, is cultivated as vegetables, edible oil, herbs, spices, condiments and fodder (Warwick 2011; Anjum et al. 2018; Tamokou et al. 2017). Among its member species, the oilseed Brassica species (B. rapa, B. napus, and B. juncea) gained global economic importance, ranked as the third-largest source of vegetable oils (Asl and Niazmand 2020). Three tetraploid Brassica species and their diploid progenitors form the ‘Triangle of U’ (U 1935), in which polyploidization by interspecific hybridization occurred between the diploid species (B. rapa, AA genome with 2n = 20; B. nigra, BB genome with 2n = 16; and B. oleracea, CC genome with 2n = 18) to form allotetraploid species (B. juncea, AABB genome with 2n = 36; B. carinata, BBCC genome with 2n = 34; and B. napus, AACC genome with 2n = 38). Other species in the Brassicaceae family are significantly used for scientific studies. For instance, Arabidopsis thaliana is a model organism in plant genetics (Koornneef and Meinke 2010), Arabis alpina in ecological genomics (Wötzel et al. 2022), Boechera species in apomixis (Brukhin et al. 2019), Cardamine hirsuta in leaf structure and morphology (Gan et al. 2016), and Lepidium meyenii in floral structure (Lee et al. 2002). The rising demand for vegetable proteins, carbohydrates, and oils from Brassicaceae crops highlights the necessity of understanding their reproductive growth and development mechanisms to ensure successful cultivation and to optimize yields in various environments. Reproductive development is a critical stage for the formation of marketable vegetable Brassica (for example, in broccoli, cabbage and cauliflower) and for gaining high grain yield in oilseed rape or canola. Knowledge of the genes controlling reproductive traits is crucial for advancing our understanding of these mechanisms. Recent developments in genome sequencing technologies and bioinformatics have greatly expanded the availability of genome sequences in databases (Mohd Saad et al. 2022), which provide a valuable resource for studying reproduction-related genes in crops. Reproductive development in plants involves a complex transition from vegetative growth to the formation of floral structures. At the onset of floral initiation, several genes function across regulatory pathways and are regarded as “flowering-time regulators” in A. thaliana (Bouché et al. 2015; Putterill et al. 2004). These core regulatory genes have been widely used to identify orthologous loci in related species within the Brassicaceae family such as in B. napus, B. rapa, and B. oleracea (Li et al. 2018; Schiessl et al. 2017), Camelina sativa (Lily et al. 2021), and Raphanus sativus (Wang et al. 2017). Among the orthologs identified in B. napus, several cyclin genes were detected (Li et al. 2018). While these genes are primarily known for their role in cell cycle control, their expression in datasets associated with developmental transitions raises their possible involvement in broader reproductive processes. However, their specific contribution to reproductive regulation remains to be clarified. Numerous studies have highlighted the biological role of cyclin genes, influencing plant developmental processes that occur around the transition to reproductive growth. The loss of function of the cyclin gene CYCD3 in the cycd3;1–3 mutant in A. thaliana, for instance, resulted in noticeable delay in developmental progression, including flowering under standard growth conditions (Dewitte et al. 2007). In Antirrhinum majus, cyclin D3b gene expression was modulated during key stages of floral development, including boundary zones between primordia, petal morphogenesis, and in the dorsal anther, influenced by cycloidea activity (Gaudin et al. 2000). In addition, CYCC1, a partner of CDK8 in the Mediator complex, was shown to broadly influence transcription, and its disruption resulted in delayed developmental transitions, phenotypically resembling cdk8 and med12 mutants (Hasan 2023). Arabidopsis MED12 was reported as a modulator of both upstream and downstream components of the key reproductive development genes (Imura et al. 2012), although its role appears multifaceted. In this study, cyclin genes were identified and their distribution characterized across ten Brassicaceae species. Evolutionary patterns and relationships, including gene duplications types, gene clustering patterns, and phylogenetic groups, were analyzed to gain insights into cyclin gene evolution within the family. Orthologs between Brassica progenitors and their polyploid derivatives were also identified. The analysis explored how polyploidization and gene duplication may have shaped cyclin gene expansion and potential functional diversification, and whether specific cyclin variants might be linked to flowering-time variation in B. napus. A pan-genomic approach combined with flowering-time-associated quantitative trait loci (QTL) analysis highlighted candidate cyclin genes with potential relevance for crop improvement. In summary, these findings provide a valuable resource for understanding cyclin gene evolution and highlight loci that may contribute to reproductive timing in Brassicaceae species. Methods Identification of cyclin genes and their gene clusters The most prominent members of the Brassicaceae family were investigated in this study, including six Brassica species in U’s triangle (B. carinata, B. juncea, B. napus, B. nigra, B. oleracea, and B. rapa) (U 1935) and 4 Arabidopsis species (A. arenosa, A. lyrata, A. suecica, and A. thaliana). Their genomic information, including protein sequences and chromosomal positions for each species, was collated from various publicly available database and publications (Table S1). To identify cyclin genes, 49 well-characterized A. thaliana cyclin genes curated in a previous study (Zhang et al. 2014) were used as reference sequences. These genes have also served as a basis for identifying cyclin genes in B. rapa (Zulfiqar et al. 2022) and other plant species (Meng et al. 2020; Zhao et al. 2022a, b). The reference protein sequences were used to identify homologous genes across ten Brassicaceae genomes via BLASTp (Altschul et al. 1997) with compositionally adjusted substitution matrices (Altschul et al. 2005) using the NCBI platform (https://blast.ncbi.nlm.nih.gov/Blast.cgi; accessed 30 May 2026). Candidate homologs were initially retained using an E-value threshold of ≤ 0.001 (Pearson 2013). The resulting putative cyclin proteins were further validated by screening against the Pfam database (Finn et al. 2014) using Hidden Markov Models (HMMs) (Eddy 1998) implemented through the GenomeNet MOTIF platform (https://www.genome.jp/tools/motif/MOTIF.html; accessed 31 May 2026). To further validate homology in Brassica genomes, each Brassica cyclin gene was subjected to reciprocal BLASTp against the A. thaliana proteome, and only those sequences that reciprocally returned the original Arabidopsis cyclin genes as the top hit were retained as true homologs. In addition, this study used a standardized custom nomenclature system that assigned the identified cyclin genes based on their closest corresponding homologs among the 49 reference A. thaliana cyclin genes. Gene clusters were defined when two or more cyclin genes were located within a 200 kb region on the same chromosome. Homogeneous clusters consisted of genes belonging to the same gene type, whereas heterogeneous clusters contained genes from different types (Cantila et al. 2022; Jupe et al. 2012). Gene duplication and ortholog analyses of cyclin genes To identify and classify duplication events among cyclin genes, DupGen_finder (Qiao et al. 2019) was used. DupGen_finder categorises gene duplications into five types based on genomic locations and sequence similarity: tandem, proximal, transposed, dispersed, and whole-genome duplication (WGD). All cyclin protein-coding gene models from each species were used as input, with genomic coordinates extracted from the respective genome annotations. The output for each species included a list of duplication pairs, each with the two gene identifiers, their cyclin subtypes, genomic locations, E-value, and the assigned duplication type. Duplication pairs were further classified as same-type when the two cyclin genes belonged to the same cyclin subtype (e.g., cycA2-cycA2) or cross-type when they belonged to different subtypes (e.g., cycA2-cycB1). The total number of duplication pairs per species, the number of distinct genes involved, and the counts of each duplication type and same-type/cross-type category were calculated. Another type of duplication in the Brassica allotetraploid A, B, and C subgenomes (excluding genes located on contigs, scaffolds, or other unplaced sequences) was also recorded. Intra-genomic duplications were defined as duplications occurring within the same subgenome while inter-genomic duplications referred to duplication events occurring between the A, B, and C subgenomes. Ortholog analysis was performed using OrthoFinder v2.5.5 (Emms and Kelly 2019). Cyclin protein sequences from the diploid progenitor species and their corresponding allotetraploid species were included in the analysis with each pair of species (progenitor vs. allotetraploid), one-to-one orthologous pairs were extracted from the results. In Brassica allotetraploids, orthologs were assigned to their respective A, B, or C subgenomes based on genome annotations, enabling assessment of progenitor genome contributions and patterns of gene retention following polyploidization. Orthologs originating from a diploid progenitor were considered retained if they were located within the expected corresponding subgenomes (e.g., B. rapa genes in the A subgenome, B. nigra genes in the B subgenome, and B. oleracea genes in the C subgenome). Conversely, orthologs located in a non-corresponding subgenome were classified as exchanged. The numbers and percentages of retained and exchanged orthologs were subsequently calculated for each progenitor-allotetraploid combination. Phylogenetic and transcriptomics-based analyses of cyclin genes To examine the evolutionary relationships and expression divergence of cyclin genes, both sequence-based phylogenetic analysis and transcriptomics-based clustering were performed. For the sequence-based phylogenetic analysis, cyclin protein sequences from Arabidopsis and six Brassica species were retrieved. After removing identical sequences, the remaining sequences were retained for subsequent analyses. These sequences were aligned using Multiple Alignment using Fast Fourier Transform (Katoh and Standley 2013), and poorly aligned regions were trimmed prior to phylogenetic reconstruction. Phylogenetic analysis was conducted using IQ-TREE v3.1.2 (Minh et al. 2020), with the best-fit substitution model selected by ModelFinder (Kalyaanamoorthy et al. 2017) under the Bayesian Information Criterion. The Q.PFAM+R3 model was identified as the optimal substitution model. A maximum likelihood (ML) phylogenetic tree was reconstructed using nearest-neighbour interchange branch swapping with 1,000 ultrafast bootstrap replicates (Hoang et al. 2018) to assess branch support. The resulting phylogenetic tree was visualized and annotated using the Interactive Tree of Life (iTOL) platform (Letunic and Bork 2024). For the transcriptomics-based analysis, previously published RNA-seq datasets (in transcript per million, TPM) from reproductive development stages were utilized and were retrieved from the Sequence Read Archive (SRA) (Table S2) via BrassicaEBD (Chao et al. 2020) https://brassica.biodb.org/downloads; accessed 30 May 2026. As the Darmor-bzh v5 genome is the most widely used reference for B. napus transcriptomic studies, gene identifiers were harmonised to the corresponding models in the updated Darmor-bzh v10 assembly based on the mapping framework established by Rousseau-Gueutin et al. (2020). This framework integrated reciprocal best-hit protein alignments using DIAMOND and BLASTp, together with synteny refinement and stringent sequence similarity criteria. Unsupervised hierarchical clustering was performed to investigate expression-based relationships among cyclin genes using a TPM expression matrix. Genes with zero variance across samples were first removed, as correlation-based distance measures cannot be computed for constant expression profiles. The remaining expression values were transformed using log₂ (TPM + 1) (Harris et al. 2020) to stabilise variance and appropriately handle zero expression values, following standard RNA-seq normalisation practices. Pairwise gene distances were calculated using correlation distance (1-Pearson correlation coefficient), enabling clustering based on similarity in expression patterns rather than absolute expression levels. Hierarchical clustering was then conducted using Ward’s minimum variance linkage method to minimise within-cluster variance during cluster formation. All analyses were implemented in Python version 3.13 (Python Software Foundation, 2025) using the SciPy library (Virtanen et al. 2020). The tree was exported in Newick format using the scipy.cluster.hierarchy.to_tree function with a custom recursive traversal and was subsequently visualized using iTOL (Letunic and Bork 2024). This approach allowed direct comparison between expression-based clustering and sequence-based phylogenetic relationships of the cyclin types. Investigating cyclin genes associated with flowering time in Brassica napus Flowering time, a key determinant of environmental adaptation and yield stability, was selected as the target trait. SNPs from the Brassica 60 K Illumina Infinium array, previously reported to be significantly associated with flowering time (Raman et al. 2019; Xu et al. 2015; Helal et al. 2021) with their positions in genome B. napus Darmor-bzh v10 (Vollrath et al. 2021). A sensitivity analysis was performed using sliding window sizes of 50, 100, 200, and 500 kb, together with gene positional information in B. napus, to support subsequent QTL delineation. Cyclin genes colocalizing with these QTL were subjected to pan-genomic analysis using eight B. napus genomes with contrasting flowering phenotypes, obtained from a previous study (Song et al. 2020). These genotypes included two early-flowering (Westar and No2127), two late-flowering (Quinta and Tapidor), and four intermediate-flowering (Gangan, Shengli, ZS11, and Zheyou7). Protein sequences of QTL-colocalizing cyclins were queried against the above eight genomes using BLASTp via the BnaOmics (Cui et al. 2023). Homologous proteins were aligned using Multiple Sequence Comparison by Log-Expectation (MUSCLE) in Geneious prime (Kearse et al. 2012), and phylogenetic relationships were inferred using the Neighbor-Joining method with 10,000 bootstrap replicates. Structural variation, including presence/absence and sequence deletions, was assessed across genomes to identify polymorphisms potentially associated with flowering time. Results Identification and distribution of cyclin genes A total of 1123 putative cyclin homologs were initially identified using 49 A. thaliana cyclins as queries (Table S3). These candidate cyclin homologs were further evaluated for the presence of conserved cyclin domains using Pfam analysis, which confirmed 1087 proteins containing cyclin domains across four Arabidopsis species and six Brassica crop species (Fig. 1, Tables S3-S4). For the Brassica species, 823 cyclin proteins were further validated by reciprocal BLASTp against the A. thaliana proteome and classified as reciprocally supported homologs (Tables S3, S5). Heat map of cyclin gene counts per type in ten Brassicaceae species Among the 23 cyclin types detected, cycA (241 genes) and cycD (225 genes) were the most numerous, followed by cycB (213 genes), cycU (164 genes), and cycT (112 genes). The cycD gene type exhibited the highest subtype diversity, consisting of cycD1 (25 genes), cycD2 (22), cycD3 (78), cycD4 (46), cycD5 (15), cycD6 (18), and cycD7 (21) (Fig. 1, Table S3). Across the four Arabidopsis species, cyclin counts ranged from 49 to 114, with an average of 66 genes per species (Fig. 1). The six Brassica species showed broader variation, ranging from 90 to 189, with an overall average of 137.2 cyclins per species (Fig. 1, Table S3). The six diploid species (A. arenosa, A. lyrata, A. thaliana, B. nigra, B. oleracea, B. rapa) displayed similar totals, averaging 71.5 cyclin genes per species. In contrast, the four allotetraploid species (A. suecica, B. carinata, B. juncea, B. napus) contained substantially higher counts, ranging from 114 to 189, with an average of 164.5 cyclin genes, clearly exceeding the diploid average (Fig. 1, Table S3). Gene duplication in cyclin genes Across 10 genomes, a total of 1,845 duplication events were identified involving 1063 cyclin genes (Fig. 2, Table S6). The highest number of duplication events was observed in B. juncea, with 377 events involving 185 genes, followed by B. napus (372 events involving 186 genes), B. carinata (299 events involving 167 genes), and A. suecica (197 events involving 114 genes ). The remaining species, all of which are diploids, contained fewer than 160 duplication events and fewer than 100 cyclin genes involved in duplication events each. For example, the diploid progenitors of B. napus, B. rapa and B. oleracea, contained fewer duplication events, with 154 events involving 87 genes and 140 events involving 94 genes, respectively (Fig. 3). On average, tetraploids showed 311.3 duplication events involving 163 genes, whereas diploids showed an average of 100.0 duplication events involving 68.5 genes (Fig. 2, Table S6). When comparing genera, Brassica species displayed a higher frequency of gene duplication than Arabidopsis. On average, Brassica species had 247.3 duplication events involving 134.8 cyclin genes, while Arabidopsis species showed only 90.3 duplication events involving 63.5 genes (Table S6). Heat map of duplication event counts among cyclin types across ten Brassicaceae species Duplication of cyclin genes in the tetraploid Brassica napus (A) and its two ancestral species, B. rapa (B) and B. oleracea (C), visualized using a Circos plot. The outer track displays chromosomal positions of cyclin genes for each chromosome (middle green track). The inner track shows connecting lines between duplicated gene pairs, with each connection representing a single duplication event. Connecting line colors indicate different duplication types: cyan = whole-genome duplication, yellow = dispersed, blue = proximal, red = tandem, and black = transposed. Numbers shown on the outer track indicate base pair positions along the chromosomes In terms of cyclin gene types, cycB2 showed the highest level of duplication, with 234 duplication events, followed by cycA2 and cycT1 which exhibited 203 and 192 duplication events, respectively (Fig. 2, Table S6). Intermediate levels of duplication were observed for cycT1, cycU4, cycA3, cycD4, and cycU2, each showing at least 100 duplication (Fig. 2, Table S6). In contrast, the remaining cyclin types displayed relatively limited duplication, with fewer than 100 duplication events. Cross-type cyclin duplications were also identified (Table S6). Among these, the cycD2-cycD4 pairing exhibited the highest number of duplication events (31), followed by cycB2-cycB3 and cycU1-cycU2, each with 20 duplication events, and cycD1-cycD2 with 18 duplication events. All remaining cross-type pairings exhibited fewer than 15 duplication events each. In the duplication result, 969 events were classified as WGD, whereas 730 events were identified as dispersed, 20 as proximal, 51 as tandem, and 75 as transposed duplications (Table S6). In Brassica species, WGD duplications accounted for at least 45% of all duplication events (e.g., 68% in B. juncea), while in Arabidopsis species this proportion varied, with A. suecica showing 42% WGD and A. thaliana showing 22% WGD. The highest numbers of WGD duplication events were observed in B. juncea (254 events) and B. napus (254 events), followed by B. carinata (151 events) and A. suecica (83 events). The remaining species each exhibited fewer than 70 WGD duplication events. Tandem duplication events were relatively rare. A. suecica showed the highest number, with 19 tandem duplications, followed by B. carinata with 11 events, while all other species exhibited fewer than seven tandem duplication events. Additional duplication categories, namely intra-genomic and inter-genomic duplications as presented in Table S6, were examined exclusively in allotetraploid Brassica species due to their well-defined A, B, and C subgenomes. In B. carinata, a total of 274 duplication events were identified after excluding pairs involving unplaced scaffolds or contigs. Among these, 44 intra-genomic pairs occurred within the B subgenome and 76 intra-genomic pairs within the C subgenome, accounting for 120 intra-genomic events, while the remaining 154 pairs represented inter-genomic duplications. In B. juncea, 359 duplicated gene pairs were identified, including 66 intra-genomic pairs within the A subgenome and 81 intra-genomic pairs within the B subgenome, totalling 147 intra-genomic events and 212 inter-genomic events. Similarly, in B. napus, 363 duplicated gene pairs were detected after excluding unplaced sequences, with 83 intra-genomic pairs within the A subgenome and 66 intra-genomic pairs within the C subgenome, giving 149 intra-genomic and 214 inter-genomic duplications. Gene clustering of cyclin genes in genomes A total of 120 cyclin gene clusters were identified across the genomes, comprising 263 genes (Table S7). A. suecica exhibited the highest number of clusters (22 clusters involving 51 genes), followed by B. juncea (20 clusters involving 41 genes), B. carinata (19 clusters involving 39 genes), and B. napus (19 clusters involving 45 genes). The remaining species each contained fewer than 10 clusters, with fewer than 20 genes involved per species. Of these 120 clusters, 88 were homogeneous cyclin clusters, while the remaining 32 clusters were heterogeneous. CycA formed the largest number of homogeneous clusters (24 clusters involving 55 genes), followed by cycU (19 clusters involving 38 genes), cycC (14 clusters involving 32 genes), and cycT (13 clusters involving 28 genes). The other cyclin types formed fewer than 10 clusters each, with fewer than 25 genes involved per type. Orthologous pairs of cyclin genes across diploid progenitors and allotetraploids A total of 852 ortholog pairs were identified across all progenitor-allotetraploid comparisons, involving 366 genes (Figs. 4 and 5, Table S8). Among Brassica comparisons, the highest number of ortholog pairs from diploid progenitors to allotetraploids was observed for B. rapa to B. napus (129 pairs, 90 genes), followed by B. oleracea to B. napus (127 pairs, 94 genes), B. rapa to B. juncea (117 pairs, 88 genes), B. oleracea to B. carinata (106 pairs, 81 genes), B. nigra to B. juncea (100 pairs, 83 genes), and B. nigra to B. carinata (88 pairs, 77 genes). In the Arabidopsis lineage, 106 ortholog pairs were identified from A. thaliana to A. suecica (50 genes), followed by 79 ortholog pairs from A. arenosa to A. suecica (45 genes). Heat map of ortholog pair counts between diploid progenitors and allotetraploids across cyclin types in Brassicaceae Orthologous relationships of cyclin genes between progenitor and allotetraploid species are visualized using Circos plots: Arabidopsis arenosa and A. thaliana to A. suecica (A); Brassica rapa and B. nigra to B. juncea (B); B. rapa and B. oleracea to B. napus (C); and B. nigra and B. oleracea to B. carinata (D). The outer track displays chromosomal positions of cyclin genes for each chromosome (middle green track). The inner track shows colored connecting lines between orthologous gene pairs. Connecting line colors indicate progenitor origin: cyan = A. thaliana, orange = A. arenosa, blue = B. rapa, red = B. nigra, and yellow = B. oleracea. Numbers shown on the outer track indicate base pair positions along the chromosomes The subgenomic retention and exchange of cyclin orthologs from diploid progenitors to Brassica allotetraploids were presented in Table S9. In B. napus, orthologs from the A-genome progenitor B. rapa showed 67.7% retention on the expected A subgenome (86 out of 127 pairs) and 32.2% exchange to the C subgenome, while those from the C-genome progenitor B. oleracea exhibited 68.8% retention on the C subgenome (86 out of 125 pairs) and 31.2% exchange to the A subgenome. In B. juncea, B. rapa orthologs were retained on the A subgenome in 73.9% of cases (85 out of 115 pairs), with 26.0% exchanged to the B subgenome; B. nigra orthologs showed higher retention on the B subgenome (87.5%; 84 out of 96 pairs) and only 12.5% exchange to the A subgenome. In B. carinata, B. nigra orthologs were retained on the B subgenome in 66.6% of cases (54 out of 81 pairs) and exchanged to the C subgenome in 33.3% of cases, whereas B. oleracea orthologs showed 70.4% retention on the C subgenome (69 out of 98 pairs) and 29.5% exchange to the B subgenome. Overall, retention rates varied from 66.6% to 87.5% depending on the progenitor-allopolyploid combination, with the lowest retention observed for B. nigra in B. carinata (66.6%) and the highest for B. nigra in B. juncea (87.5%). Exchange events (orthologs found on a non-expected subgenome) occurred in all comparisons, with frequencies between 12.5% and 33.3%. In terms of cyclin types, cycT1 (83 pairs, 38 genes) and cycA2 (81 pairs, 36 progenitor genes) were the most abundant, followed by cycB2 (76 pairs, 34 genes), cycB1 (69 pairs, 32 genes), cycD3 (64 pairs, 28 genes), cycU4 (61 pairs, 24 genes), cycA3 (55 pairs, 24 genes), and cycA1 (48 pairs, 19 genes) (Fig. 4, Table S8). Among D-type cyclins, cycD3 showed the highest retention (64 pairs), while cycD4 (41 pairs), cycD2 (20 pairs), cycD1 (19 pairs), cycD6 (17 pairs), cycD7 (15 pairs) and cycD5 (14 pairs) were less frequent (Fig. 4, Table S8). The remaining types (cycC1, cycH1, cycU2, cycU1, cycSDS, cycB3, cycU3, cycJ18, cycL1) each accounted for fewer than 40 ortholog pairs and involved fewer than 20 progenitor genes (Fig. 4, Table S8). These results, obtained with a robust phylogeny-based orthology inference, confirm that most cyclin genes are retained in Brassica allotetraploids and that retention patterns differ substantially among species pairs and subgenomes. Phylogenetic analysis of the cyclin genes in Arabidopsis and Brassica species After quality filtering, 1079 out of 1087 cyclin protein sequences were used for the ML phylogenetic tree construction. The phylogenetic tree resolved three major clades with distinct cyclin-type compositions (Fig. 6). Clade 1 contained 281 members and was predominantly composed of cycD types (225 members), together with a smaller number of cycA1 (56 members). The distribution of cycD types within this clade included cycD3 (78), cycD4 (46), cycD1 (25), cycD2 (22), cycD7 (21), cycD6 (18), and cycD5 (15). Clade 2 was the largest group, comprising 401 members. It contained a mixture of cycA2 and multiple cycB types, including cycA2 (103), cycB2 (101), cycB1 (87), and cycB3 (24), as well as cycC1 (41), cycSDS (33), and cycJ18 (12). Clade 3 comprised 397 members and represented the most heterogeneous group. It included cycT1 (112), cycU4 (77), cycA3 (75), cycU2 (47), cycH1 (28), cycU1 (26), cycL1 (18), and cycU3 (14). Sequence-based phylogeny of 1079 cyclin genes from ten Brassicaceae genomes, illustrating evolutionary relationships among cyclin types Transcriptome-expression abundance dendrogram analysis of the cyclin genes in Brassica napus Of the 189 identified B. napus cyclin genes, 15 genes annotated in the Darmor v5 genome assembly had no corresponding genes in Darmor v10 and were excluded from subsequent analysis (Table S10). The remaining 174 genes were evaluated for expression across 22 RNA-seq TPM datasets, of which seven genes showed no detectable expression across all datasets and were excluded from expression pattern analyses (Table S10). Therefore, 167 B. napus cyclin genes were retained for clustering analysis. Hierarchical clustering of their expression profiles identified four distinct expression clusters (Fig. 7), with uneven cluster sizes: Cluster 1 (66 genes), Cluster 2 (38 genes), Cluster 3 (39 genes), and Cluster 4 (24 genes). Cluster 1 was the largest and most diverse, comprising a broad representation of cycA-, B-, and D-types, including 23 cycA, 16 cycB, and 22 cycD genes, along with a small number of other cyclins (one each of cycC1, cycH1, cycL1, and cycU4). This broad pattern suggests that multiple cyclin families may share coordinated transcriptional regulation during reproductive development. Cluster 2 was characterized by the presence of six cycSDS and five cycT1 genes, together with smaller numbers of cycA, cycB, and cycD3 genes. Cluster 3 was dominated by cycT1 genes (13 members), along with several cycC1, cycH1, cycL1, and cycU1 genes. In contrast, Cluster 4 was highly distinct and consisted predominantly of U-type cyclins (six cycU2, two cycU3, and ten cycU4), with only a few cycC1 and cycD7 genes. Hierarchical clustering of 167 cyclin genes based on log₂-transformed (TPM + 1) expression values across 22 RNA-seq datasets representing reproductive developmental stages in Brassica napus. Genes were grouped into four clusters (I-IV) according to similarities in their expression profiles. Clusters I, II, III, and IV are indicated in green, purple, blue, and red, respectively Linking cyclin genes to flowering time phenotypes with pan-genomic insights A total of 174 SNPs previously reported to be significantly associated with flowering time in B. napus were compiled and used to define QTL regions. These QTL intervals were delineated using a 50 kb window based on sensitivity analysis and subsequently used to identify colocalising cyclin genes (Table S12). Of the 189 cyclin genes in B. napus here, five of them, Bna21cycA2, Bna84cycC1, Bna85cycC1, Bna99cycD3 and Bna113cycD4, fell within the QTL intervals and were therefore selected for pan-genomic comparison across eight B. napus cultivars representing early, intermediate, and late flowering types (Table S12-13). Analysis of the five cyclin genes and their homologs was shown in Table S13 and Fig. 8. Some genes showed complete sequence conservation (100% query coverage and 100% identity) in specific cultivars, including Bna99cycD3 in ZS11, Westar, Tapidor, Shengli, Quinta, Gangan, and Zheyou7, Bna113cycD4 in Tapidor and Quinta, Bna21cycA2 in Quinta, Bna85cycC1 in Zheyou7, Westar, Shengli, and Gangan, highlighting strong conservation of gene structure across these genomes. On the other hand, several cyclin genes also showed high similarity across several B. napus genomes, including Bna21cycA2 in No2127, Zheyou7, ZS11, Gangan, Westar, Shengli, and Tapidor, Bna113cycD4 in Westar, No2127, Zheyou7, ZS11, Shengli, and Gangan, and Bna99cycD3 in Zheyou7. Other cyclin genes here were less similar across genomes with lower query coverage (45–100%) and identity values ranging from 70.8 to 100% like Bna84cycC1 in Gangan, Zheyou7, Shengli, Westar, and Tapidor, and Bna85cycC1 in Tapidor, Quinta, No2127, and ZS11. Also, no BLASTp hits were found in Bna84cycC1 in Quinta, ZS11, No2127. These results suggest structural differences such as truncations, copy number changes, and gene absence. These patterns provide useful candidates for exploring variation associated with flowering behaviour. Amino acid polymorphisms distinguishing flowering-time groups in Brassica napus. Gene name suffixes indicate cultivar identity (ZS = ZS11, ZY = Zheyou7, GG = Gangan, TA = Tapidor, SL = Shengli, QU = Quinta, NO = No2127, WE = Westar) and flowering phenotype (E = early, I = intermediate, L = late) for Bna21cycA2 and Bna113cycD4. Numbers above the alignment denote amino acid positions based on the multiple sequence alignment. Panels A-C correspond to Bna21cycA2, and panels D-E correspond to Bna113cycD4. Red dashed boxes highlight diagnostic amino acid polymorphisms that differentiate flowering-time phenotypes Among the five genes, Bna21cycA2 and Bna113cycD4 and their homologs showed consistent phylogenetic patterns across the eight B. napus genomes. Notably, the early-flowering genotypes Westar and No2127 were always grouped within the same clades for both genes, a pattern that was not observed from the other genomes. This suggests a shared and distinct sequence variation associated with early flowering. In Bna21cycA2, both Westar and No2127 carry a phenylalanine (F) at amino acid position 411, whereas all other genomes possess a threonine (T) at this site (Fig. 8). Similarly, in Bna113cycD4, Westar and No2127 share a valine (V) at position 100, while the remaining genomes contain a leucine (L). These consistent amino acid substitutions provide strong candidates for developing molecular markers specific to early-flowering genotypes. For the intermediate-flowering group (Zheyou7, ZS11, Shengli, and Gangan), distinct polymorphisms were also observed. In Bna21cycA2, these genomes show unique amino acid variant at position 67. In Bna113cycD4, intermediate genotypes differ at positions 19, 20, and 25 relative to both early and late groups. For the late-flowering genotypes, particularly Quinta and Tapidor, gene-specific molecular markers could also be designed. For Bna21cycA2, a forward primer may be positioned around the 20th amino acid region, with a reverse primer designed near the 400th amino acid region. Similarly, for Bna113cycD4, a marker could be developed using a forward primer around the 19th or 20th amino acid position and a reverse primer near the 100th amino acid region. Together, these patterns suggest that these genes may contain sequence variants that distinguish early, mid, and late flowering types; however, their potential role in flowering time requires further verification through genetic mapping or functional studies. Discussion This study investigated the physical occurrence and evolutionary mechanism of cell cycle protein cyclin genes in ten published genomes within the Brassicaceae family. Significant variation was observed in both the number and composition of cyclin genes, including differences in dispersed, proximal, tandem, transposed, WGD duplication types, inter- and intra-genomic duplication among Brassica allotetraploids, and the organization of cyclin genes into distinct clusters across species, genera, and ploidy levels. Cyclin genes are highly abundant due to their diverse and essential roles in plant development (Chaubet-Gigot 2000; Inzé and De Veylder 2006; Patterson et al. 2021; Kumar and Larkin 2017). The prevalence of cyclin-A, B, and D in the Brassicaceae family likely reflects both their critical roles in regulating the plant cell cycle and their preferential retention during evolution. A-type cyclins are tightly regulated during the cell cycle, with sequential expression patterns observed during cell-cycle progression, and specific members such as Nicta; CYCA3;2 playing key roles at the interface of cell division and differentiation (Reichheld et al. 1996; Yu et al. 2003). Cyclin-B is pivotal in controlling the G2-to-M phase transition during key developmental stages (Weingartner et al. 2003; Fobert et al. 1994; Giménez-Abián et al. 2002). Numerous studies have underscored the importance of A-, B-, and D-type cyclins in both mitotic cell cycle and mitotic growth (Chaubet-Gigot 2000; Ito 2000; Meijer and Murray 2000). Beyond their cell cycle roles, cycA is implicated in auxin signaling and meristem formation (Roudier et al. 2003), and these cyclins interact with CDKs to regulate additional plant processes (Roudier et al. 2000). CycB, for instance, governs embryonic development, seed size (Zhao et al. 2022a, b), and promotes seed formation (Niu et al. 2020), as well as playing a role in microtubule organization (Romeiro Motta et al. 2022). CycD is specifically linked to cell division in the stomatal lineage of the hypocotyl epidermis (Kono et al. 2007). Among the two genera studied, Brassica exhibited the highest number of cyclin genes, largely driven by the presence of three allotetraploid species, which increased overall gene counts. This expansion is primarily attributed to a more recent and extensive whole-genome triplication event in the Brassica lineage, which was not experienced by the Arabidopsis lineage (Chalhoub et al. 2014). Across species, allotetraploids consistently contained more cyclin genes than diploids, with elevated numbers of duplicated and clustered genes that likely contribute to the retention of essential cell cycle regulators and enhanced regulatory flexibility, underscoring their importance in plant growth, development, and reproduction (Cheng et al. 2018; Almeida-Silva and Van de Peer 2025). The expansion of cyclin genes in Brassicaceae has been largely driven by multiple rounds of WGD, followed by diversification through additional duplication modes. Consistent with previous studies, the majority of duplicated cyclin genes originated from WGD events, reflecting the widespread occurrence of paleopolyploidization across land plant evolution (Panchy et al. 2016; Sankoff and Zheng 2018; Shi et al. 2026). Subsequent tandem, proximal, and dispersed duplications likely contributed to the further expansion, functional diversification, and retention of cyclin family members. The observed expansion patterns of cyclin genes in the six cultivated Brassica species are intrinsically linked to the lineage-specific whole-genome triplication (WGT) event, which provided the initial genomic material for a three- to six-fold increase in gene copies in diploid and allotetraploid species, respectively (Chalhoub et al. 2014; Cheng et al. 2014; Song et al. 2021). However, gene copy numbers alone do not fully explain the current distribution of cyclin genes. Following WGT in Brassica, extensive gene fractionation and differential gene loss have played a major role in shaping gene family evolution (Cheng et al. 2018). In particular, selection acting on gene dosage is expected to influence the retention or loss of duplicated genes, especially those involved in dosage-sensitive networks and protein complexes. The gene balance hypothesis suggests that selection maintains dosage-sensitive genes within balanced interaction networks to preserve protein stoichiometry, as disruptions in copy number or expression, such as those caused by homoeologous exchanges in allopolyploids, can create genomic imbalance and constrain gene retention and expression patterns (Bird et al. 2023). Such processes are consistent with observations that certain genes are preferentially retained as single copies due to strong functional constraints, while others are more frequently lost following duplication events (De Smet et al. 2013). In these allotetraploid species B. napus, B. juncea, and B. carinata, extensive genome reshuffling and homoeologous exchange have further shaped cyclin gene distribution, leading to gene relocation and copy number variation has produced the complex and non-uniform cyclin gene distribution observed in this study. These processes can generate gene copies that are physically distant within chromosomes yet remain highly similar in sequence, thereby resembling dispersed duplication patterns (Hurgobin et al. 2018). Furthermore, inter- and intra-genomic duplications in Brassica allotetraploids reflect their evolutionary origin following hybridization between distinct diploid progenitors, forming the A and C subgenomes under the 'Triangle of U' model. Subsequent polyploidization was followed by extensive chromosome rearrangements and homoeologous exchanges, further contributing to subgenome restructuring and gene retention dynamics and changes (Bird et al. 2023; Yang et al. 2016; Yim et al. 2022). Polyploidization in Brassicaceae is well established to increase gene copy numbers compared to diploid relatives, through chromosome doubling or tripling (Chalhoub et al. 2014; Song et al. 2021; Kagale et al. 2014). Interspecific hybridization and subsequent genome duplication have contributed to fixed heterozygosity in allopolyploid species (Takahagi et al. 2018; Chen 2010, 2013). This is supported by comparative studies in other plant families, such as Arachis, where polyploid species were found to possess a greater abundance of cyclin genes than their diploid counterparts (Gokul et al. 2023). Homogeneous clusters likely arise from tandem duplication events and may provide functional redundancy or facilitate coordinated regulation of gene expression (Razin et al. 2021). Heterogeneous clusters, in contrast, arise through segmental duplication, ectopic recombination, translocations, and gene transpositions (Meyers et al. 2003; Richly et al. 2002; Perazzolli et al. 2014; Marone et al. 2013). Functional roles of clustered cyclin genes with diverse domains remain to be fully elucidated. Phylogenetic analysis revealed that cyclin genes cluster into distinct clades corresponding to functional classes, indicating early diversification and strong sequence conservation across Brassicaceae (Jia et al. 2014; Menges et al. 2007; Bulankova et al. 2013; Meskiene et al. 1995). The grouping of cycD types and cycA1 members within Clade I, cycA2 and cycB types within Clade II, and cycC, cycH, cycP (including cycU), cycSDS, cycL, cycT, and cycJ18 within Clade III has been consistently reported in phylogenetic studies of cyclin genes across plant species (Wang et al. 2017; Jia et al. 2014), suggesting that the major cyclin lineages originated early in plant evolution and have retained their functional identities despite subsequent gene duplication and diversification events. On the other hand, transcriptomics-based phylogen showed substantial divergence in expression patterns among duplicates, which indicates that regulatory changes, rather than coding sequence evolution, drive functional diversification (Casneuf et al. 2006). The phylogenetic relationships are based on multi-species sequence and structural features, whereas the expression clusters are species-specific and functionally defined; thus, a one-to-one correspondence between the two is not expected. Larger expression-based clusters likely reflect coordinated regulation during reproductive development, while smaller clusters may support specialized or condition-specific roles. These complementary analyses collectively highlight how cyclin genes are both evolutionarily conserved and functionally diversified within Brassicaceae. Linking cyclin genes to an economically important trait such as flowering time in canola, our results show that Bna21cycA2 and Bna113cycD4, which co-localize with previously reported flowering-time QTLs, were identified using a 50 kb window for QTL delineation. This threshold is supported by linkage disequilibrium decay estimates in B. napus, including a genome-wide half-decay of less than 45 kb (Rahman et al. 2022) and chromosome-specific decay ranging from 7 to 68 kb (Horvath et al. 2020). Both genes harbor sequence variations that correlate with flowering phenotypes in B. napus. Early-flowering cultivars, Westar and No2127, cluster together and share distinctive amino acid substitutions absent in other cultivars, whereas intermediate-flowering cultivars display unique variants. These observations suggest that these genes are putative candidates potentially associated with flowering-time variation in B. napus. However, their roles in flowering-time regulation require further verification through genetic mapping and functional validation. These results highlight the multifunctional roles of conserved cyclin domains. Beyond their classical involvement in cell cycle regulation, cyclins appear to participate in broader developmental and environmental networks, particularly during reproductive development in B. napus. This study lays the groundwork for functional validation and suggests that manipulating cyclin genes could offer a strategy to control key reproductive traits, such as flowering, in Brassicaceae crops. It also underscores the value of pan-genomic investigations to uncover novel regulatory elements for breeding climate-resilient and high-yielding varieties. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgments The authors acknowledge Gerry N. Ramos Jr. from the Philippine Genome Center Mindanao, University of the Philippines Mindanao, Davao City, Philippines, for his helpful insights and assistance with the computational coding related to the duplication-type analysis. Author contributions AYC did data acquisition and analysis, and contributed to writing (original draft, review, and editing), validation, methodology design and development, investigation, and conceptualization. SC contributed to writing (original draft, review, and editing), validation, methodological development, and project administration. KHMS and WAC contributed to writing (review and editing), validation, supervision, and project administration. Funding Open Access funding enabled and organized by CAUL and its Member Institutions. This research was funded by the Grains Research & Development Corporation and co-funded by The University of Western Australia (grant No. UWA2404-011RTX). Data availability All data generated or analysed during this study are included in this published article and its supplementary information files. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data generated or analysed during this study are included in this published article and its supplementary information files.