The coding sequence (CDS), protein sequence, genome sequences, and gene sets of T. aestivum were retrieved from Ensembl Plants website (http://plants.ensembl.org/info/data/ftp/index.html). Subsequently, a total of 79 CONSTANS-like amino acid sequences, comprising 17 from A. thaliana, 16 from O. sativa, 18 from Z. mays, 14 from Sorghum bicolor, and 14 from Phyllostachys edulis, were employed as queries to perform a local BLASTP search (E-value < 1e-5) in the T. aestivum protein database (Griffiths et al., 2003). The amino acid sequences corresponding to the resulting identity names were subsequently submitted to the NCBI Conserved Domain Database (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) to verify the presence of both CCT and B-box domains (Huang et al., 2021). Following chromosomal order, each CONSTANS-like gene was assigned a unique name. Sequence statistics were computed with the Fasta Stats function of TBtools, and the molecular weight and theoretical isoelectric point of all CONSTANS-like proteins were batch-calculated via ExPASy (https://web.expasy.org/compute_pi/) (Wilkins et al., 1999). Subcellular localizations were predicted using Plant-mPLoc (http://www.csbio.sjtu.edu.cn/bioinf/plant-multi/) (Chou and Shen, 2010).

Multiple sequence alignment of CONSTANS-like proteins was performed with ClustalW under default parameters (Thompson et al., 1994), and a neighbor-joining phylogenetic tree was constructed in MEGA11 with 1,000 bootstrap replicates (Tamura et al., 2021). Gene structures were visualized by uploading coding and genomic sequences to GSDS 2.0 (http://gsds.gao-lab.org/) (Hu et al., 2015). Conserved motifs were identified using MEME (Bailey et al., 2015) with the following settings: motif width 6–100 aa, maximum number of motifs = 10. HMMER was employed for additional domain verification (Potter et al., 2018).

Synteny analysis was performed with MCScanX (TBtools) using the wheat genome and GFF file (BLASTP E-value ≤ 1e-5, max five hits per locus) (Wang et al., 2012), and the resulting blocks were visualized with Advanced Circos. Genome assemblies and annotations for Triticum dicoccoides, Aegilops tauschii, H. vulgare, Secale cale, and S. italica were downloaded from Ensembl Plants (http://plants.ensembl.org); those for S. bicolor and O. sativa were obtained from Phytozome (https://phytozome-next.jgi.doe.gov/), and the Z. mays was retrieved from MaizeGDB (https://www.maizegdb.org/). Collinearity plots were generated with the Multiple Synteny Plot tool in TBtools (Chen et al., 2020). Non-synonymous (Ka) and synonymous (Ka) substitution rates of duplicated gene pairs were calculated with the Simple Ka/Ks Calculator (NG) implemented in TBtools (Chen et al., 2020; Gao et al., 2021a) and visualized in GraphPad Prism 8.3.0. Ka/Ks > 1, = 1 and < 1 indicate positive, neutral and purifying selection, respectively. The species timetree was constructed using divergence time data from TimeTree (http://timetree.org/) (Kan et al., 2025).

The 2000bp sequence upstream of the CDS of TaCOLs was extracted from the wheat genome and defined as the corresponding gene promoter. Subsequently, the promoter sequences were submitted to the PlantCARE website (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) to identify various cis-acting regulatory elements (Lescot et al., 2001). These elements were classified into three main categories: plant growth and development, phytohormone response, and biotic and abiotic stress response (Gao et al., 2018).

To obtain and analyze wheat gene expression profiles under diverse conditions, multiple datasets were accessed and integrated. Gene expression data across five tissues (root, stem, leaf, spike, and grain), four developmental stages (booting, heading, anthesis, and grain-filling), two temperature conditions (4°C and 23°C) were retrieved from the WheatOmics platform (http://wheatomics.sdau.edu.cn/) (Ma et al., 2021). Additionally, RNA-seq data under various abiotic stresses, including heat (H), drought (D), combined heat and drought (HD), salt (S), salt-drought (SD), salt-heat (SH), and salt-heat-drought (SHD) stresses, were downloaded from the GEO dataset GSE183007 and aligned to the Chinese Spring wheat reference genome (Da Ros et al., 2023). Data for polyethylene glycol (PEG) and abscisic acid (ABA) treatments were also acquired from the NCBI Sequence Read Archive (SRA). All raw sequencing reads were uniformly pre-processed and aligned to the Chinese Spring wheat genome with TBtools (Chen et al., 2020). Expression values were normalized and subjected to hierarchical clustering analysis, with results visualized as heatmaps. The corresponding SRA accession numbers were provided in Supplementary Table S5 .

A co-expression network was constructed using the WGCNA R package on the RStudio platform based on transcriptome data from 32 wheat samples subjected to abiotic stress conditions (Langfelder and Horvath, 2008; Yu et al., 2020). The dataset comprised eight stress treatment groups, each with four biological replicates. The resulting network was visualized using Cytoscape. Gene Ontology (GO) enrichment analysis was performed utilizing the Triticeae-Gene Tribe database (http://wheat.cau.edu.cn/TGT/), with results presented as bubble charts; detailed characteristics of each co-expression module were summarized in Supplementary Table S7 . Additionally, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was carried out via eggNOG-mapper (http://eggnogmapper.embl.de/) for functional annotation of protein sequences, and the outcomes were visualized using bar charts.

T. aestivum cv.Chinese Spring wheat seeds were surface-sterilized with 5% sodium hypochlorite and sown in nutrient soil in a greenhouse. The greenhouse conditions were set at 23 ± 1°C, with a 12 h light/12 h dark photoperiod, 60-70% relative humidity, and 6000 lx light intensity. After two weeks, seedlings were transferred to two different photoperiod regimes: 16 h light/8 h dark and 8 h light/16 h dark (Zhao et al., 2022; Huang et al., 2022; Yang et al., 2020). Leaves were collected at 4-hour intervals over a 24-hour period under each photoperiodic condition, immediately frozen in liquid nitrogen and stored at -80°C.

Total RNA was extracted using the RNA Rapid Extraction Kit (Mei5bio, MF610-01). First-strand cDNA synthesis was conducted with the MonscriptTM RTIII All-in-One Mix with dsDNase. Primers were designed using Primer Premier 5.0 software and validated via NCBI-Primer-BLAST. TaGAPDH (GI: 7579063) served as the internal reference gene (Tang et al., 2020). qRT-PCR was performed using the 2× Quantinova SYBR Green PCR Master Mix under the following conditions: 95°C for 5 min, 45 cycles of 72°C for 20 s, and a melting curve. Relative gene expression was calculated using the 2^–ΔΔCt method and presented using GraphPad Prism 8.3.0 (Lei et al., 2021). One-way analysis of variance (ANOVA) was used via Duncan’s test at p < 0.05 in IBM SPSS v25.0 to compare significant differences in different time point within each treatment (Gao et al., 2021b). The relevant genes and their primer sequences were listed in Supplementary Table S8 .

The CDS of TaCOL genes, excluding stop codons, were amplified from a cDNA library derived from Chinese Spring wheat and subsequently cloned into the Simple vector (TransGen, Beijing, China) for sequencing. The PCR products were fused with linearized vectors via homologous arms. We digested the pCAMBIAI1305 and pGBKT7 vectors with the restriction enzyme pairs XbaI/BamHI and EcoRI/BamHI, respectively. All primers used in this study are listed in Supplementary Table S9 .

The pCAMBIA1305 vector contains a 35S-driven GFP sequence. The TaCOL-GFP and control vectors were introduced into Agrobacterium tumefaciens GV3101 (Weidi, Shanghai, China). These Agrobacterium strains were used to infiltrate Nicotiana benthamiana leaves. After 40 hours, leaves were stained with DAPI and observed under an Olympus SpinSR10 microscope (Japan) (Kan et al., 2025). The pGBKT7 vector, which expresses proteins as fusions to the GAL4 DNA-binding domain (BD), was used in this study. The recombinant plasmid pGBKT7-TaCOL, along with negative control (empty pGBKT7 vector) and positive control (pGBKT7-53 + pGADT7-T), was transformed into Y2HGold competent yeast cells (Weidi, Shanghai, China) using the PEG/LiAc method. Following transformation, yeast cells were selected on synthetic dropout (SD) medium lacking tryptophan (SD/-Trp) to confirm bait plasmid retention. To assess possible autoactivation of the bait protein, transformed yeast were also plated on a higher stringency medium lacking tryptophan, histidine, and adenine, and supplemented with X-α-Gal (SD/-Trp/-His/-Ade/X-α-Gal). Growth and blue coloration on this medium would indicate transcriptional activation activity of the MEL1 reporter via α-galactosidase (Kan et al., 2025).

Fifty-one candidate members were identified in the T. aestivum protein database via BLASTP. It was confirmed that they all contained at least one B-box domain and a complete CCT domain. Seventeen of these proteins featured two B-box domains: the B-box1 domain had the sequence C-C-C-L-C-C-D-H-A-H-R, while B-box2 was C-C-P-A-C-L-C-C-D-H-A-A-H-R. The CCT domain, rich in arginine (R): R-RY-EK-R-F-K-RY-RK-A-R-R-KGRF-K ( Figure 1 ), was also conserved and aligns with prior studies (Robson et al., 2001). The largest protein was Ta-7A-COL37, with 490 amino acids. It also had the longest CDS (1473 bp) and the highest molecular weight (52005.04 Da). The theoretical isoelectric points of the 51 proteins ranged from 4.79 (Ta-6D-COL30) to 7.65 (Ta-7A-COL33). And all TaCOL proteins were predicted to be nuclear-localized. The specific and detailed information was shown in Supplementary Table S1 .

To classify these TaCOLs more accurately, a comprehensive phylogenetic tree was constructed using 116 protein sequences from A. thaliana (17), O. sativa (16), Z. mays (18), P. edulis (14), and T. aestivum (51). Consistent with Hu et al. (2018), these CONSTANS-like members were divided into three distinct subfamilies ( Supplementary Figure S1 ). Meanwhile, the 51 full-length TaCOL proteins were also classified into three subfamilies (I, II, and III) based on the phylogenetic tree ( Figure 2A ). Subfamily I and subfamily II each comprised 18 members, and subfamily III consisted of 15 members. Motif 1 corresponded to the CCT domain; motif 2 and motif 3 were linked to B-box domains ( Supplementary Table S2 ). This explained why all members contained motif 1, along with either motif 2 or motif 3, or all three motifs collectively. In subfamily I, three TaCOL proteins (Ta-6D-COL28, Ta-6A-COL18, Ta-6B-COL23) harbored motif 2 but lacked motif 3, whereas the remaining 15 members possessed both motifs in a variable arrangement ( Figure 2B ). Motif 9 occurred in this subfamily and six members of subfamily III, invariably residing at the C-terminus. In subfamily II, five TaCOL proteins retained both motif 2 and motif 3, whereas the remaining 13 members possessed only motif 2, signifying a single B-box domain. Motif 4 was ubiquitous throughout the subfamily, whereas motif 8 was restricted to a subset of members; these diagnostic motifs likely underlay the delineation of distinct subfamilies ( Figure 2A, B ). In terms of genetic structure, the number of exons among the 51 TaCOL genes ranged from one to four. Notably, six TaCOLs within subfamily III were intronless, while the remaining nine members each contained one intron. Members of subfamily I typically contained one to two introns, whereas those in subfamily II generally harbored three introns, with the exceptions of Ta-4A-COL7 and Ta-4D-COL12. Overall, each subfamily exhibited a largely conserved motif compositions and gene architecture, alongside subtle variations.

The common hexaploid bread wheat (Triticum aestivum) possesses a heterohexaploid genome (AABBDD, 2n = 6x = 42) consisting of three distinct subgenomes A, B, and D; each subgenome derived from different wild grass progenitors. The A subgenome originated from the diploid wheat Triticum urartu (AA). Tetraploid wheat (Triticum turgidum, AABB) subsequently arose from a hybridization event between T. urartu (AA) and Aegilops speltoides (BB). Finally, hexaploid bread wheat was formed through a second hybridization between tetraploid wheat and the diploid wild relative A. tauschii (DD) (null et al., 2014). Among the 51 TaCOL genes, the majority were mapped to 15 chromosomes, with a subset located at unknown chromosomal positions (Un). None of the TaCOL genes were present on chromosomes 1 and 3 of the A, B, and D subgenomes. The high number of TaCOL genes was observed on chromosomes 6A (5 TaCOLs), 6B (5 TaCOLs), 6D (5 TaCOLs), 7A (6 TaCOLs), 7B (5 TaCOLs), and 7D (7 TaCOLs) ( Figure 3A ). And we identified 64 pairwise duplications among the 51 TaCOLs, indicating that most genes exist as two or more homologs. Specifically, 17 TaCOLs were present as pairs; six (Ta-6A-COL19, Ta-6D-COL28, Ta-7A-COL38, Ta-7B-COL43, Ta-7D-COL48 and Ta-7D-COL50) had three paralogous genes; seven (Ta-2A-COL2, Ta-2B-COL4, Ta-6A-COL21, Ta-6B-COL26, Ta-6D-COL31, Ta-7A-COL35 and Ta-7B-COL41) possessed four homologous genes; and nine (Ta-2D-COL6, Ta-6A-COL22, Ta-6B-COL24, Ta-6B-COL27, Ta-6D-COL29, Ta-6D-COL32, Ta-7A-COL34, Ta-7B-COL40 and Ta-7D-COL46) retained five homologous genes ( Supplementary Table S3 ). The maximum Ka and Ks values for all homologous pairs were 0.275 and 0.904, respectively ( Figure 3B ; Supplementary Table S3 ). And the Ka/Ks value was significantly below 1, indicating that the TaCOL genes were under strong purifying selection.

To further explore the evolutionary relationships of CONSTANS-like genes, we conducted a collinearity analysis of T. aestivum and eight other Poaceae plants (T. dicoccoides, A. tauschii, H. vulgare, Secale cereale, S. italica, S. bicolor, O. sativa, and Z. mays) ( Figure 4A ). Despite considerable differences in the number of orthologous gene pairs, the number of associated TaCOLs remained relatively comparable. The orthologous pair distributions were as follows: Td-Ta (96 pairs involving 38 TaCOLs), Ae-Ta (46 pairs involving 33 TaCOLs), Hv-Ta (63 pairs involving 43 TaCOLs), Sc-Ta (58 pairs involving 40 TaCOLs), Sb-Ta (47 pairs involving 32 TaCOLs), Zm-Ta (71 pairs involving 34 TaCOLs), Si-Ta (40 pairs involving 32 TaCOLs), and Os-Ta (65 pairs involving 40 TaCOLs). And the average Ks values of these orthologous pairs were as follows: Td-Ta (0.203016013), Ae-Ta (0.26228061), Hv-Ta (0.281602287), Sc-Ta (0.293103976), Si-Ta (0.527560353), Os-Ta (0.595051996), Sb-Ta (0.602337798) and Zm-Ta (0.652450481) ( Figure 4B ; Supplementary Table S3 ). The Ka/Ks ratios were less than 1, except for a maximum value of 1.2 in S. cereale ( Figure 4C ; Supplementary Table S3 ). Additionally, the time trees of the nine Poaceae plants revealed that wheat diverged most recently from T. dicoccoides, followed by A. tauschii, S. cereale, H. vulgare, O. sativa, and finally S. italica, S. bicolor, and Z. mays ( Figure 4D ). This divergence order largely mirrored the trend of mean Ks values, the sole exception being the average Ks observed in rice.

To elucidate the potential regulatory functions of the 51 TaCOLs, we investigated the cis-regulatory elements in their promoter regions and systematically classified them into three main categories: plant growth and development, phytohormone response, and biotic and abiotic stresses ( Figure 5 ; Supplementary Table S4 ). Light-responsive cis-acting elements encompass a wide range of types, such as Sp1, AE-box, TCT-motif, Box 4, GATA-motif, G-box, GT1-motif, I-box, ACE, and TCCC-motif. Notably, the G-box element was present in every TaCOL gene and was the predominant element in the light-responsive category. In the phytohormone-responsive category, we identified cis-acting elements associated with specific hormone responses: abscisic acid (ABA) response (ABRE), auxin responsiveness (TGACG/CGTCA-motif), gibberellin (GA) response (TATC-box, P-box, and GARE-motif), methyl jasmonate (MeJA) responsiveness (TGACG/CGTCA-motif), and salicylic acid (SA) response (TCA-element). Among these, ABRE and TGACG/CGTCA-motif were the most frequent, with 255 and 128 occurrences, respectively. GA-responsive and SA-responsive elements were detected in 34 TaCOL members (49 total occurrences) and 16 TaCOL members (19 total occurrences), respectively. Stress-responsive elements were also abundant, including those associated with anaerobic conditions (ARE and GC-motif), drought stress (MYB, MYC, MBS), low-temperature stress (LTR), and defense/stress responses (TC-rich repeats). Seven TaCOL genes (Ta-2B-COL4, Ta-6A-COL20, Ta-6B-COL26, Ta-6D-COL30, Ta-7D-COL44, Ta-7D-COL46, and Ta-U-COL51) contained no fewer than 20 stress-responsive elements ( Figure 5B ). Drought-related MYB and MYC elements were particularly prominent: MYB elements occurred 358 times across 50 TaCOLs, while MYC elements occurred 174 times across 47 TaCOLs. Low-temperature-responsive LTR elements were identified 48 times in 29 TaCOLs. These findings suggested that TaCOL genes might play crucial regulatory roles in plant growth, development, and stress adaptation.

To evaluate the expression patterns of TaCOL genes across five tissues (root, stem, leaf, spike, and grain) and four developmental stages (booting, heading, anthesis, and grain-filling), transcriptome data revealed that over half of the TaCOL genes showed no detectable expression levels ( Figure 6 ), a common characteristic of transcription factors. Notably, seven TaCOL genes, such as Ta-7A-COL36, Ta-7D-COL48, Ta-4B-COL10, Ta-4D-COL12, Ta-7A-COL37, Ta-7B-COL42, and Ta-7D-COL49. showed high expression across all five tissues. However, some genes displayed tissue-specific expression. For instance, Ta-4B-COL9, Ta-4A-COL8, Ta-4D-COL11, Ta-2A-COL2, Ta-5A-COL13, Ta-2B-COL4, Ta-5B-COL14, Ta-2D-COL6, and Ta-5D-COL16 had relatively higher expression in leaves than the other four tissues ( Figure 6A ). Additionally, 13 TaCOL genes, including Ta-7A-COL36, Ta-7A-COL38, Ta-7D-COL50, Ta-4B-COL9, Ta-4D-COL11, Ta-2A-COL2, Ta-4A-COL8, Ta-6A-COL18, Ta-6D-COL28, Ta-7D-COL48, Ta-7B-COL43, Ta-2B-COL4, and Ta-2D-COL6, showed particularly high expression levels across all four developmental stages. Notably, the expression level of Ta-7B-COL43 exceeded 100. Furthermore, the expression levels of Ta-7D-COL46, Ta-5A-COL13, Ta-5B-COL14, and Ta-5D-COL16 showed higher expression during the booting, heading, and anthesis stages than during the grain-filling stage ( Figure 6B ). Based on these findings, Ta-2A-COL2, Ta-2B-COL4, Ta-2D-COL6, Ta-4A-COL8, Ta-4B-COL9, Ta-4D-COL11, Ta-5A-COL13, Ta-5D-COL16, Ta-7A-COL36, and Ta-7D-COL48 might play crucial roles in wheat growth and development. We also observed that Ta-6A-COL20, Ta-6B-COL25, and Ta-6D-COL30 showed higher expression levels during anthesis and grain-filling stages, being three times higher than those during the booting and heading stages. This suggested that these genes might play important roles in the later stages of wheat reproductive development.

To assess the expression patterns of TaCOL genes under diverse abiotic stress conditions, we conducted cluster analysis and visualized the results via heatmaps. The analysis revealed distinct and varied expression profiles across different treatments ( Figure 7A ). Notably, 12 TaCOL genes, namely Ta-7D-COL50, Ta-7A-COL38, Ta-7B-COL43, Ta-2B-COL4, Ta-4D-COL11, Ta-4A-COL8, Ta-4B-COL9, Ta-2A-COL2, Ta-2D-COL6, Ta-5D-COL16, Ta-5A-COL13, and Ta-5B-COL14, maintained sustained high transcription levels across all treatments, including the control. Furthermore, some TaCOL genes displayed coordinated transcriptional responses both individual and combined stress treatments. For example, Ta-6A-COL18, Ta-6D-COL2, Ta-6B-COL23, Ta-4B-COL10, Ta-4A-COL7, and Ta-4D-COL12 were moderately upregulated under heat and drought stress compared to the control, with similar trends observed under combined stress. Other genes, such as Ta-6D-COL32, Ta-6A-COL22, and Ta-6B-COL27, showed comparable transcription patterns.

In an additional transcriptome analysis, we investigated the transcription levels of TaCOL genes under PEG-induced osmotic stress and ABA treatments ( Figure 7B ). Among these, Ta-4B-COL10 was significantly induced, showing more than a threefold increase in expression following both ABA and PEG treatments at 3 h and 24 h. Conversely, Ta-7B-COL39 and Ta-7D-COL45 were markedly suppressed under the same conditions. Ta-7A-COL36 and Ta-7D-COL48 were upregulated in response to PEG but remained unresponsive to ABA. Additionally, Ta-7D-COL49, Ta-6D-COL28, and Ta-7D-COL50 exhibited moderate expression levels that were unaffected by either ABA or PEG treatment.

Temperature stress was also found to markedly influence the expression of TaCOL genes, consistent with its critical role in plant growth, development, and flowering time regulation. Ta-2A-COL1 and Ta-2B-COL3 were significantly induced under low-temperature conditions, while Ta-4A-COL7, Ta-4D-COL12, and Ta-7B-COL43 showed moderate upregulation at 4°C. Interestingly, several TaCOL genes exhibited higher expression levels at 23°C, including Ta-6D-COL32, Ta-7A-COL34, Ta-7B-COL40, Ta-7D-COL46, Ta-6B-COL23, Ta-6D-COL29, Ta-6B-COL27, Ta-6D-COL30, Ta-7A-COL37, and Ta-7B-COL42 ( Figure 8C ). These findings underscored the potential functional diversity of TaCOL genes in mediating plant developmental processes and adaptive responses to environmental stresses.

To dissect the transcriptional regulation of TaCOLs under individual and combinatorial drought, heat, and salt stresses, we constructed a weighted gene co-expression network from high-quality transcriptome data. After stringent filtering (mean TPM ≥ 1 and row sum ≥ 28), 35574 genes were retained. A scale-free topology was achieved with a soft-thresholding power of β = 11 ( Supplementary Figure S2B ). Weighted Gene Co-expression Network Analysis (WGCNA) resolved 10 modules of highly co-expressed genes ( Supplementary Figure S2D ), among which 25 TaCOLs were distributed across five distinct modules. To visualize the co-expression network with TaCOLs as central nodes, we extracted the top 100 (blue), 200 (black, red, green-yellow), and 300 (turquoise) genes exhibiting the highest topological overlap matrix values from each module, retaining a final set of 16 TaCOL genes. The blue module, which contains Ta-4B-COL10, was enriched for protein transport and fatty-acid metabolism ( Figure 8A ). Three TaCOLs (Ta-5A-COL13, Ta-5B-COL14, and Ta-5D-COL16), in the greenyellow module were associated with Mg²⁺-dependent serine/threonine phosphatase activity and ABA signaling ( Figure 8B ). The red module, harboring Ta-7A-COL38, Ta-7B-COL43, and Ta-7D-COL50, was linked to amino-acid transmembrane transport and floral development ( Figure 8C ). In the black module, five TaCOLs (Ta-6A-COL18/19, Ta-6B-COL23/24, Ta-6D-COL28) converged on flower development, starch biosynthesis, and chloroplast-membrane organization ( Figure 8D ). The turquoise module, comprising Ta-2D-COL6, Ta-4A-COL7, Ta-4D-COL12, and Ta-6D-COL29, was enriched for ubiquitin-dependent protein catabolism, floral development, and mRNA processing ( Figure 8E ). KEGG pathway annotations for each module were provided in Supplementary Figure S3 .

Based on the consistently high expression profiles in available transcriptome data across multiple tissues, developmental stages, and stress treatments; while also ensuring representation from all three phylogenetic subfamilies, we monitored the expression profiles of 12 TaCOL genes under short-day (SD) and long-day (LD) conditions to explore their potential roles in photoperiodic regulation. Notably, Ta-4B-COL10, Ta-6B-COL23 and Ta-7D-COL48 were sharply up-regulated under both photoperiods, whereas the remaining nine genes were globally repressed by light ( Figure 9 ). Among the latter, Ta-2B-COL4, Ta-2D-COL6, Ta-4B-COL9, Ta-4D-COL11, Ta-5B-COL14 and Ta-5D-COL16 maintained low, albeit fluctuating, transcript levels throughout the time-course. Ta-2A-COL2, Ta-4A-COL8 and Ta-5A-COL13 declined progressively under SD, but exhibited discrete peaks under LD. Interestingly, seven TaCOLs (Ta-2B-COL4, Ta-4B-COL9, Ta-4B-COL10, Ta-5B-COL14, Ta-5D-COL16, Ta-6B-COL23 and Ta-7D-COL48) displayed near-identical expression trajectories in SD and LD, indicating a photoperiod-independent mode of regulation.

Expression analysis under temperatures of 4°C and 23°C, as well as across five tissues and four developmental stages, revealed that Ta-2B-COL4, Ta-5D-COL16, and Ta-7D-COL48 exhibited stable and high expression levels ( Figures 6 , 7 ; Supplementary Table S6 ). Based on these findings, these three genes were selected for sub-cellular-localization and trans-activation assays. The coding sequences without stop codons, were fused to GFP constructs. These constructs were expressed in Nicotiana benthamiana leaves alongside a 35S promoter-driven GFP control. Whereas 35S::GFP fluorescence was distributed throughout the cell, the three CONSTANS-like proteins were distinctly localized in the nucleus ( Figure 10A ), in agreement with the predictions provided on the website ( Supplementary Table S1 ). In addition, to evaluate transcriptional activity, each pGBKT7-TaCOL construct was individually introduced into Y2HGold yeast, together with the positive control (pGBKT7-53 + pGADT7-T) and the negative control (pGBKT7). All transformants produced white colonies on SD/-Trp medium. On SD/-Ade/-His/-Trp/X-a-gal medium, only yeast cells with Ta-5D-COL16 grew well and turned blue, similar to the positive control, whereas Ta-2B-COL4 and Ta-7D-COL48 failed to grow on this medium, indistinguishable from the negative control ( Figure 10B ). Thus, Ta-5D-COL16 activated both the HIS3 and LacZ reporters, demonstrating its self-transcriptional activity in yeast.