Genome‑wide analysis of the CaWRKY transcription factors reveals CaWRKY28 as a candidate gene for cold tolerance breeding in Coffea arabica L. Authors: Tianci Hu, Zhenwei Zhang, Yu Lan, Tianxiang Yang, Feifei He, Jing Gao, Xiangshu Dong Published: 2026 ABSTRACT BackgroundIn China, over 98% of coffee is produced in Yunnan Province, where Coffea arabica L. predominates. Cold stress represents a major genetic constraint limiting coffee production, especially during the fruiting stage. WRKY transcription factors (TFs) are key regulators of stress responses, yet their systematic characterization and breeding value in C. arabica L. are remain largely unexplored.ResultsHere, we conducted a genome-wide analysis of WRKY genes across four C. arabica L. genomes, identifying 400 CaWRKY members phylogenetically classified into three groups. Segmental duplication was identified as the primary driver for CaWRKY family expansion. Physiological assessment of fruiting trees under cold stress revealed significant alterations in key traits, including increased electrolyte leakage and soluble sugars, and decreased photosynthesis. Transcriptomic analysis identified 47 cold-responsive CaWRKYs, which can be classified into five clusters, with CaWRKY28, 82, 85, 86, and 89 showing up‑regulation under cold treatment. Crucially, heterologous expression of CaWRKY28 in yeast conferred significantly enhanced cold tolerance, functionally validating its role in stress adaptation.ConclusionsOur study provides the first comprehensive genetic overview of the CaWRKY family in C. arabica L. and pinpoints CaWRKY28 as a high-value candidate for molecular breeding aimed at enhancing cold tolerance in this vital crop.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12870-026-09284-1. Introduction Coffee is a globally vital agricultural commodity and provides a livelihood for over 25 million smallholder farmers across more than 60 tropical countries, exerting substantial economic influence [1]. Coffee cultivation is highly dependent on the environment, primarily restricting coffee production to the region known as the “Coffee Belt”, which lies between 25°N and 30°S latitude. In this zone, optimal average temperatures for coffee cultivation vary by species, with Coffea arabica L. (arabica coffee) thriving at 18–22 °C and C. canephora (robusta coffee) preferring slightly warmer conditions of 22–28 °C [2]. Of these coffee species, C. arabica L. is the most widely cultivated and economically significant, accounting for over 70% of global coffee bean production [3]. In China, C. arabica L. constitutes the majority of coffee cultivation, with more than 98% of both planted area and yield concentrated in Yunnan Province [4]. In Yunnan, C. arabica L. is predominantly grown between 21° and 26°N latitude. This region is within the marginal edges of the Coffee Belt and is characterized by vertical climatic gradients due to its high elevation and relatively northern latitude [4]. Although these geographical features contribute to the distinctive quality attributes of Yunnan-grown coffee beans, they also expose the crop to abiotic stressors [4, 5]. Among the climatic stressors impacting C. arabica L. in Yunnan, cold stress has historically been the predominant factor. For example, in 2017, widespread frost events damaged approximately 10,000 ha of coffee plantations, equivalent to 9.03% of Yunnan’s total coffee-growing area, and resulted in estimated economic losses of USD 10 million [4]. The adverse effects of cold stress are not confined to Yunnan. Other major coffee-producing regions within the Coffee Belt have also experienced significant impacts due to chilling and frost events [2]. In 2021, severe frosts occurred in southeastern Brazil—the world’s largest coffee-producing region—causing damage to nearly 8–10% of the C. arabica L. crop [2]. Considering these recurring challenges, illustrating the responses and improving the cold tolerance of C. arabica L. has become a priority for coffee breeding programs worldwide, aiming to ensure the resilience and sustainability of coffee production in the face of climate variability. WRKY transcription factors (TFs) were first identified as DNA-binding proteins in sweet potato, where they typically bind to the W-box cis-element [(C/T) TGAC(T/C)] in the promoters of target genes [6]. They play crucial roles in regulating plant growth and development, as well as in mediating responses to both biotic and abiotic stresses [7, 8]. Structurally, WRKY proteins are defined by the presence of one or two highly conserved WRKY domains, each comprising approximately 60 amino acids and containing the invariant heptapeptide motif WRKYGQK [9]. This domain is responsible for specific recognition and binding of the W-box sequence. Additionally, most WRKY proteins harbor a zinc-finger motif at the C‑terminus, which contributes to DNA-binding specificity and facilitates protein–protein interactions. Based on the number of WRKY domains and the type of zinc-finger motif, WRKY proteins are classified into three major groups (I, II, and III). Group I members contain two WRKY domains and a C2H2‑type zinc‑finger motif with the consensus sequence CX4−5CX22−23HXH. In contrast, Groups II and III each contain a single WRKY domain. Group II proteins possess either a C2H2 or C2HC‑type zinc‑finger motif (CX7CX23HXC), whereas Group III proteins typically carry a C2HC‑type zinc‑finger motif [10]. Group II is further divided into five subgroups (IIa–IIe) based on sequence similarity and structural features. The conserved architecture of the WRKY domain provides a reliable basis for genome‑wide identification and classification of WRKY family members, forming the foundation for the subsequent bioinformatic analyses in this study. Due to their critical roles in plant stress adaptation and development, WRKY genes have been extensively studied and characterized at the genomic level among plant species, including Arabidopsis thaliana [11], Zea mays [12], Populus trichocarpa [13], Cannabis sativa [14], Prunus sibirica [15], Citrus ichangensis [16], Panax notoginseng [17], and Pyrus betulifolia [18]. Numerous WRKY genes have been isolated and functionally validated, particularly those related to abiotic stress responses, including cold stress. For example, CsWRKY46 is upregulated under cold stress and abscisic acid (ABA) treatment. Moreover, CsWRKY46 overexpression has been shown to improve tolerance to freezing and chilling conditions in cucumber [19]. CdWRKY2 plays a dual role in cold tolerance in Cynodon dactylon by mediating the sucrose biosynthesis pathway and coordinating with the CBF signaling pathway, thus contributing to the plant’s ability to withstand cold stress [20]. CsWRKY21 enhances the cold resistance of tea plants by regulating the expression of CsABA8H and CsUDP, which control the ABA content [21]. SlWRKY33 enhances cold tolerance in tomato by directly targeting and inducing the expression of multiple genes, including those encoding kinases (e.g., CDPK11), TFs (e.g., MYBS3), and molecular chaperones (e.g., BAG6) [22]. Additionally, cold stress triggers the production of SlWRKY50 in tomatoes. SlWRKY50, a Group II WRKY TF, regulates jasmonic acid (JA) biosynthesis, thereby increasing the plant’s tolerance to cold stress [23]. Considerable progress has been made in identifying WRKY genes in coffee, with reports from C. canephora (49 CcWRKYs) and from EST-based surveys in C. arabica L [24, 25]. However, a significant knowledge gap remains: the lack of a systematic, genome-wide characterization of the WRKY family in the allotetraploid C. arabica L. genome, and an incomplete understanding of its role in cold stress tolerance. This study was designed to address these limitations. First, we performed a comprehensive identification and evolutionary analysis of the WRKY family using four reference genomes of C. arabica L., examining their phylogenetic relationships, structural diversity, and duplication history. Second, to explore their functional relevance, we integrated physiological measurements with transcriptome profiling (RNA-seq) of fruiting C. arabica L. plants under cold stress to identify differentially expressed CaWRKYs. Finally, the function of a key cold-responsive candidate, CaWRKY28, was validated through heterologous expression in yeast. This multi-pronged strategy reveals the evolutionary dynamics and functional diversity of CaWRKYs, providing a valuable foundation for improving cold stress tolerance in coffee breeding programs. Materials and methods Plant materials, cold treatment, and phenotyping In this study, we used fruiting-stage plants (five-year-old C. arabica L. ‘Catimor, CIFC 7963’), which is the predominant cultivar grown in Yunnan, China. Based on findings from previous research, coffee plants at the cherry maturity stage (collected in December) were selected for experimental treatment and grown at 8/4°C (day/night) for 2 and 4 days, which was considered cold stress conditions [4]. Plants were maintained in a walk-in plant growth chamber (Conviron Ltd., Winnipeg, Canada) at a diurnal temperature of 24 °C during the day and 20 °C at night for 7 days. Subsequently, the temperature was adjusted to 8/4°C (day/night) to induce cold stress. After treatment, plants were returned to normal conditions (24/20°C). Phenotypes were observed, beans were collected, and the abortion ratio was recorded at one week, one month, and six months after cold treatment, respectively. All experiments were performed with three biological replicates. Immediately after 0 (control), 2, and 4 days post-treatment initiation, five plants were randomly selected, and the two most recently matured pairs of leaves were harvested from the apical portion of each plant. Leaf samples were flash-frozen in liquid nitrogen and stored at − 80 °C for subsequent analyses. To detect changes in photosynthesis and gas exchange, the Pn and gs were recorded on the second fully expanded leaf of each plant between 09:00 and 11:00 h using a LI-6400XT Portable Photosynthesis System (LI-COR Biosciences, Lincoln, NE, USA) following cold treatment. Instrument parameters were set as follows: CO2 reference concentration of 400 µmol mol− 1, photosynthetic photon flux density of 300 µmol m− 2 s− 1 (red–blue light source), and flow rate of 500 µmol s− 1. Following cold treatment, electrolyte leakage was assessed in both cold-treated and control plants, as previously described [5]. Briefly, five leaf discs (6-mm diameter) were randomly selected from a pooled sample of 10 discs per plant, derived from the five most recently fully expanded leaves. These discs were immediately transferred to a glass test tube containing 10 mL of deionized water and incubated on an orbital shaker (150 rpm) at ambient room temperature (22 ± 1 °C) for 30 min. The initial electrical conductivity (I) was measured using a CON110 conductivity meter (Oakton Instruments, Vernon Hills, IL, USA). Leaf discs were subsequently heat-treated in boiling deionized water for 10 min and cooled to room temperature, and the final conductivity (F) was recorded. The relative conductivity was calculated using the following formula: Relative conductivity (%) = (I / F) × 100. To determine the physical changes in C. arabica L. leaves, all biochemical assays were conducted with kits, following the manufacturer’s protocols and using the leaf samples collected above. The total soluble sugar and starch contents were quantified using a Plant Soluble Sugar Content Assay Kit and Starch Content Assay Kit, respectively, and peroxidase (POD, EC 1.11.1.7) and catalase (CAT, EC 1.11.1.6) activities were determined using a Peroxidase (POD) Activity Assay Kit and Catalase (CAT) Activity Assay Kit (all kits were from Nanjing Jiancheng Bioengineering Institute, Nanjing, China). All experiments were performed with three biological replicates. To determine the biochemical content changes in green coffee beans, fully mature coffee cherries from the treated plants were harvested in early January. Following harvest, green coffee beans were isolated, flash-frozen in liquid nitrogen, and stored at − 80 °C until further processing. Green coffee beans were sent to Yunnan Sanbiao Agro-forestry Technology Co., Ltd. (Kunming, China) for quantitative determination of the caffeine, trigonelline, crude protein, crude fat, reducing sugar, and chlorogenic acid contents. The chlorogenic acid, trigonelline, and caffeine contents were quantified using high-performance liquid chromatography (HPLC) (Waters 2695, Waters Corporation, Shanghai, China) in accordance with NY/T 3514 − 2019, NY/T 3012 − 2016, and GB 5009.139–2014, respectively. The crude protein content was determined using the Semimicro Kjeldahl method (K9840, Haineng Future Technology Group Co., Ltd., Jian, China) in accordance with NY/T 3-1982, and crude fat was extracted using Soxhlet extraction in accordance with GB/T 6433 − 2025. To determine the phenotypes of cold-treated plants, the floral buds were observed six months after cold treatment. Based on descriptions from the literature, the number of normally developing floral buds and aborted floral buds was counted for each plant [26]. The flower bud abortion ratio was calculated as the number of aborted floral buds divided by the total number of floral buds. The cold treatment experiments were independently replicated three times using different sets of plants. Identification of CaWRKY genes The genomic sequences of four C. arabica L. cultivars—Bourbon, ET-39, Geisha, and Kona—were downloaded from the National Center for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov/datasets/genome/) based on previously published studies [27–30]. All protein sequences from these four cultivars were used as queries to search against the profile HMM using HMMER v3.4 (http://hmmer.org/) with the Pfam 34.0 HMM library. The WRKY DNA-binding domain (PF03106, E-value < 0.1) was identified and isolated based on the HMM searching results [31]. Subsequently, all candidate members that did not contain the conserved WRKYGQK heptapeptide sequence were discarded. Based on the C. arabica L. genome annotation file (gff3), putative CaWRKY genes were mapped to chromosomes, renamed systematically by location, and their distribution was visualized with TBtools software (version 1.0987663) [32]. Phylogenetic, gene duplication, and synteny analysis of WRKY family members Protein sequences from C. canephora L., A. thaliana, and C. arabica L. were subjected to multiple sequence alignment utilizing MAFFT software (version 7.526), with all parameters set to their default values [33]. The protein sequences of CcWRKYs and AtWRKYs were retrieved from previously published studies [11, 24]. Subsequently, phylogenetic trees were generated using MEGA6 software using a Poisson model with the following options: homogeneous pattern, pairwise deletion, and 1000 replicate bootstrap [34]. MCScanX with default parameters was used to identify gene duplication in CaWRKYs and synteny between CcWRKY and CaWRKY genomes [35]. RNA sequencing and assembly Total RNA was extracted from nine samples (three biological replicates per condition) using TRIzol reagent (Invitrogen, USA) according to the manufacturer’s instructions. RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA), and only samples with RNA integrity number (RIN) ≥ 8.0 were used for subsequent library construction. RNA-seq libraries were constructed using the Truseq™ Stranded mRNA Library Prep Kit (Illumina, USA) following the manufacturer’s protocol. Briefly, mRNA was enriched from 1 µg of total RNA using oligo (dT) magnetic beads, fragmented to approximately 300 bp, and reverse transcribed into double-stranded cDNA. After end repair, A-tailing, and adapter ligation, the libraries were amplified by PCR and purified. Library quality and concentration were assessed using an Agilent 2100 Bioanalyzer and Qubit 3.0 Fluorometer (Thermo Fisher Scientific, USA). Sequencing was performed on the Illumina NovaSeq 6000 platform (Illumina, USA) at Gene Denovo Biotechnology Co., Ltd. (Guangzhou, China), generating 150 bp paired-end reads. Raw reads were subjected to quality control using fastp (version 0.20.0) [36] with default parameters, including removal of adapter sequences, filtering of reads containing > 10% ambiguous bases (N), and trimming of low-quality bases (Phred score < 20). Clean reads were then aligned to the C. arabica L. reference genome (ET-39, NCBI accession: GCA_036785885.1) using HISAT2 (version 2.2.1) [37] with default parameters. Gene expression levels were quantified as transcripts per million (TPM) using StringTie (version 2.2.0) [38]. Differential expression analysis between cold-treated and control samples was performed using DESeq2 (version 1.38.0) [39] in R (version 4.2.0). Genes with |log2 (fold change) | ≥ 1 and an adjusted P-value < 0.05 were considered differentially expressed. The raw sequencing data have been deposited in the China National Center for Bioinformation (CNCB) under the BioProject accession number PRJCA050185. RNA isolation, cDNA synthesis, and quantitative real-time PCR Total RNA was extracted from leaf samples using RNAiso Plus Reagent (Takara Biomedical Technology Co., Ltd., Beijing, China) following the manufacturer’s instructions. To eliminate potential genomic DNA contamination, an on‑column DNase digestion was performed. Briefly, 20 µg of total RNA was treated with 5 µL of 10 × DNase I Buffer, 20 U of RNase Inhibitor, and 2 µL (2 U) of RNase‑free DNase I in a total reaction volume adjusted with diethyl pyrocarbonate (DEPC)‑treated water. The reaction was incubated at 37 °C for 20 min, followed by heat inactivation of DNase I according to the manufacturer’s protocol. Subsequently, 1 µg of the DNase‑treated total RNA was used as a template for first‑strand cDNA synthesis using a PrimeScript™ RT Reagent Kit (Takara Biomedical Technology Co., Ltd., Beijing, China). The resulting cDNA was diluted to a final concentration of 10 ng/µL for downstream PCR applications. qRT-PCR was performed using the QuantStudio™ 7 Flex Real-Time PCR System (Applied Biosystems®, Foster City, CA, USA). The reaction mixture was carefully prepared at a total volume of 20 µL. This mixture contained 2 µL of the diluted cDNA, 1.0 µL of each primer at a concentration of 10 µM, 10 µL of SYBR® Premix Ex Taq™ II (Tli RNaseH Plus) (TAKARA BIO INC., Shiga, Japan), and 6 µL of distilled water. The thermal cycling conditions for qRT-PCR were as follows: initial denaturation step at 95 °C for 30 s, followed by 40 cycles of denaturation at 95 °C for 5 s and extension at 56 °C for 30 s. The 2−∆∆CT method was applied to analyze the qRT-PCR results. All primer sequences employed in this study are listed in Table S1. Heterologous expression of CaWRKY28 in yeast The CaWRKY28 coding sequence (CDS) was cloned into the yeast expression vector pYES2 using In-Fusion cloning to generate the recombinant plasmid pYES2-CaWRKY28. Both the pYES2-CaWRKY28 construct and the empty pYES2 vector (control) were transformed into Saccharomyces cerevisiae strain BY4741 via the lithium acetate/single-stranded carrier DNA/polyethylene glycol (LiAc/SS carrier DNA/PEG) method. Transformants were selected on SD/-Ura agar plates (SD/-Ura + 2% glucose + 2% agar) and incubated at 29 °C for 4 days. For cold tolerance assays, a standard galactose induction protocol was adopted. Single colonies of positive transformants were inoculated into SD/-Ura liquid medium containing 2% (w/v) glucose and grown overnight at 29 °C with shaking (200 rpm). Cells were then harvested by centrifugation, washed twice with sterile water, and resuspended in induction medium (SG/-Ura liquid medium containing 2% (w/v) galactose) at an initial OD₆₀₀ of 0.2. After 4 h of induction at 29 °C, cultures were transferred to low temperatures (4 °C, 0 °C, and − 20 °C) for 24, 48, and 72 h. Following each cold treatment, cultures were serially diluted, spotted onto SG/-Ura solid medium containing 2% (w/v) galactose, and incubated at 29 °C for 3–4 days to monitor colony growth. This experiment was performed with three independent biological replicates. Results Identification of WRKY genes in C. arabica L. Based on the Hidden Markov Model (HMM) search, 101 WRKY genes were identified in the C. arabica L. ‘ET-39’ genome and named CaWRKY01 to CaWRKY101 according to their positions on chromosomes. The locus ID, gene name, genome location, CDS length, and protein length of the CaWRKYs are listed in Table S2. The genomic DNA sequences of the CaWRKYs ranged from 949 (CaWRKY73) to 10,583 base pairs (bp) (CaWRKY57). The CDS length of the CaWRKYs ranged from 357 (CaWRKY62) to 3066 bp (CaWRKY64), with an average length of 1202 bp (Table S2). To investigate copy number variation of CaWRKY genes among C. arabica L. cultivars, we also isolated these genes from the ‘Bourbon’, ‘Geisha’, and ‘Kona’ varieties. A total of 102, 98, and 99 CaWRKY genes were isolated from Bourbon, Geisha, and Kona respectively (Table S3). The CaWRKY genes from different varieties were named and classified based on their best BLASTp hit to CaWRKYs in the ET-39 genome (Table S3). In addition, these 400 CaWRKYs constituted 49 orthologous gene groups (OGGs), which were classified as follows: 47 OGGs were present in all genomes, and two OGGs were absent in at least one genome (Table S4). Phylogenetic analysis of CaWRKY genes To investigate the evolutionary relationships among CaWRKY proteins, we performed a phylogenetic analysis of the 49 OGGs using a pangenome-based gene family analysis method [40]. For each OGG, the CaWRKY gene from ET-39 was used as the representative sequence. For the one OGG absent in ET-39, the longest CaWRKY sequence (CaWRKY37_Bourbon) in the remaining three genomes was used. C. canephora (CcWRKY) and A. thaliana (AtWRKY) proteins were used as references for subgroup classification. Finally, the phylogenetic analysis showed that CaWRKY proteins were categorized into three major groups (Fig. 1). Among them, 20 CaWRKY proteins, each containing two intact WRKY domains, were assigned to Group I. Based on the classification scheme established for WRKY genes in Arabidopsis [10], the 72 CaWRKY proteins within Group II were further subdivided into five subgroups. Specifically, subgroups II a, II b, II c, II d, and II e comprised six, 12, 29, 12, and 13 CaWRKY proteins, respectively. Group III consisted of 10 CaWRKY proteins, which were characterized by the presence of a single WRKY domain and a C2HC-type zinc finger motif (Fig. 1 and Table S3). Fig. 1Phylogenetic tree of CaWRKY proteins. Groups (or subgroups) are color-coded by branch lines. The red circles highlight gene IDs in C. canephora, the blue triangle denotes a gene from Arabidopsis, and the red diamond marks a gene from C. arabica L. CaWRKY37_Bourbon belonging to the OGG absent in ET-39 is indicated in red font Phylogenetic tree of CaWRKY proteins. Groups (or subgroups) are color-coded by branch lines. The red circles highlight gene IDs in C. canephora, the blue triangle denotes a gene from Arabidopsis, and the red diamond marks a gene from C. arabica L. CaWRKY37_Bourbon belonging to the OGG absent in ET-39 is indicated in red font WRKY proteins are uniquely characterized by the presence of a WRKY domain, which contains a conserved WRKYGQK heptapeptide motif [41]. In C. arabica L. variety ET-39, a variant of this canonical WRKY domain, specifically WRKYGKK, was identified in eight CaWRKY proteins belonging to subgroup II c: CaWRKY59, CaWRKY62, CaWRKY71, CaWRKY78, CaWRKY91, CaWRKY92, CaWRKY93, and CaWRKY94 (Table S2). In addition to the WRKYGKK variant, other WRKY domain sequence variations were also discovered among different C. arabica L. cultivars. For instance, the WIKYGEK variant was detected in both the Bourbon and Geisha cultivars, specifically in CaWRKY37_Bourbon, CaWRKY43_Geisha, CaWRKY37_Geisha, and CaWRKY43_Geisha. The WRKSYYR variant was identified in the Geisha cultivar, specifically in CaWRKY54_Geisha (Table S3). CaWRKY chromosomal location and gene duplication analysis The CaWRKY genes from the ET-39 genome were used for chromosomal localization and gene duplication analysis. A total of 101 CaWRKYs were mapped to the 22 chromosomes of C. arabica L., with an uneven distribution across the genome. A total of 101 CaWRKYs were present on 22 chromosomes in C. arabica L., and they were non-uniformly distributed (Table S2 and Figure S1). Chromosome ChrB_sgEE had the most CaWRKYs (11 members), followed by chromosome ChrB_sgCC, which contained 10 CaWRKYs. Chromosomes ChrA_sgCC and ChrA_sgEE had the fewest CaWRKYs (two members). BLAST and MCScanX analysis showed that CaWRKY gene duplication events occurred in C. arabica. Briefly, 14 tandem duplicated genes (13.9%) belonging to 7 clusters were identified (Figure S1). These results indicate that tandem duplication is one of the reasons related to CaWRKY expansion in the C. arabica L. genome. Considering that C. canephora is one of the parents of C. arabica L., the collinear relationships between CaWRKYs and CcWRKYs were identified using MCScanX software. A total of 43 gene pairs were identified between C. arabica L. and C. canephora (Table S5 and Fig. 2), and 26 segmental duplication events occurred in the C. arabica L. genome, indicating that the greater number of WRKYs in the C. arabica L. genome could be due to segmental gene duplication events (Table S5 and Fig. 2). Fig. 2Syntenic relationships between CaWRKY genes of C. arabica L. and C. canephora. Chromosomes of C. arabica L. (ChrA_sgCC to ChrK_sgEE_A) and C. canephora (Chr1 to Chr11) are shown in rose red and yellow, respectively. The collinear gene pairs were mapped onto the chromosomes and linked to each other. The collinear gene pairs from C. arabica L. and C. canephora are linked by blue lines. The segmental duplication genes in C. arabica L. are connected with red lines. The locations of tandem duplicated genes are marked with black lines. Only the names of collinear gene pairs are presented Syntenic relationships between CaWRKY genes of C. arabica L. and C. canephora. Chromosomes of C. arabica L. (ChrA_sgCC to ChrK_sgEE_A) and C. canephora (Chr1 to Chr11) are shown in rose red and yellow, respectively. The collinear gene pairs were mapped onto the chromosomes and linked to each other. The collinear gene pairs from C. arabica L. and C. canephora are linked by blue lines. The segmental duplication genes in C. arabica L. are connected with red lines. The locations of tandem duplicated genes are marked with black lines. Only the names of collinear gene pairs are presented Identification of CaWRKYs responsive to cold treatment In response to pertinent industrial concerns regarding cold stress, we evaluated the potential roles of CaWRKYs by exposing C. arabica L. plants to cold treatment during the fruit maturation phase in late December. After cold treatment, the injury area on the first pair of fully developed mature leaves increased (Fig. 3A). The effects of the cold treatment duration on key physiological parameters were evaluated (Fig. 3). Leaf electrolyte leakage (relative conductivity), total soluble sugar content, and CAT activity exhibited a time-dependent increase, with elevations observed after 2 days of treatment (Fig. 3B and C, and 3H). In contrast, the starch content and POD activity initially increased after 2 days but decreased after 4 days (Fig. 3D and G). The net photosynthetic rate (Pn) and stomatal conductance (Gs) consistently decreased with prolonged cold exposure (Fig. 3E and F). The biochemical composition of green coffee beans harvested during the current season was analyzed. The caffeine content significantly decreased, and the chlorogenic acid content increased in response to extended cold treatment (Table 1). Monitoring of the floral bud development status in the subsequent season indicated that the abortion rate of floral buds increased after prolonged cold stress (Fig. 3I and J). These findings indicate that cold stress induces severe physiological damage in coffee leaves and leads to a reduction in coffee bean yield in the subsequent harvest season. Fig. 3Physiological responses of C. arabica L. during late December fruit maturation under cold stress. A Phenotype of the first fully expanded top leaves one week after cold stress conditions. B–H. Quantitative physiological parameters measured during cold stress: (B) relative conductivity; (C) total soluble sugar content; (D) starch content; (E) net photosynthetic rate; (F) stomatal conductance; (G) peroxidase (POD) activity; (H) catalase (CAT) activity; (I) post-cold stress abortion of next-season flower buds; (J) proportion of aborted flower buds following cold stress exposure. Control, normal growth conditions. Two and 4 days, refer to the duration of cold treatment. NB, normally developing floral buds. AB, aborted floral buds during development. The figures are represented based on the means ± SD (n = 3) Physiological responses of C. arabica L. during late December fruit maturation under cold stress. A Phenotype of the first fully expanded top leaves one week after cold stress conditions. B–H. Quantitative physiological parameters measured during cold stress: (B) relative conductivity; (C) total soluble sugar content; (D) starch content; (E) net photosynthetic rate; (F) stomatal conductance; (G) peroxidase (POD) activity; (H) catalase (CAT) activity; (I) post-cold stress abortion of next-season flower buds; (J) proportion of aborted flower buds following cold stress exposure. Control, normal growth conditions. Two and 4 days, refer to the duration of cold treatment. NB, normally developing floral buds. AB, aborted floral buds during development. The figures are represented based on the means ± SD (n = 3) To determine the expression profiles of CaWRKYs under cold stress, the first pair of fully expanded mature leaves was sampled and subjected to RNA-seq analysis. A total of 62.94 Gb of clean bases were obtained, with an average mapping rate of 95.10% (Table S6). Using the criteria of transcripts per kilobase of exon model per million mapped reads (TPM) ≥ 1, |log2(fold change) | ≥ 1, and adjusted p-value < 0.05, a total of 47 CaWRKY genes were identified as responsive to cold treatment and were classified into five clusters (Fig. 4A and Table S7). Thirteen genes (CaWRKY26, -07, -23, -27, -98, -99, -34, -28, -66, -82, -85, -89, and − 86) were upregulated at both 2 and 4 days, with peak expression at 4 days; another 15 genes (CaWRKY29, -68, -95, -81, -25, -58, -14, -101, -46, -90, -55, -19, -63, -04, and − 01) were also upregulated at both time points but showed maximal expression at 2 days. Ten genes (CaWRKY11, -15, -10, -75, -05, -80, -59, -74, -70, and − 79) were transiently induced at 2 days and then downregulated by 4 days. The remaining nine genes exhibited downregulation under cold stress, six (CaWRKY49, -57, -69, -17, -62, and − 12) showed progressively decreasing expression from 2 to 4 days, whereas three (CaWRKY16, -13, and − 92) reached their lowest levels at 2 days. Fig. 4Expression profiles of CaWRKY genes in response to cold stress. A Heatmap depicting the expression of cold-responsive CaWRKYs in C. arabica L., as determined by RNA-seq. (a) Genes were upregulated after 2 and 4 days of cold treatment, with peak expression occurring at 4 days. (b) Genes were upregulated after 2 and 4 days of cold treatment, with peak expression occurring at 2 days. (c) Genes were upregulated after 2 days of cold treatment and downregulated after 4 days. (d) Genes were downregulated after 2 and 4 days of cold treatment, with the lowest expression occurring at 4 days. (e) Genes were downregulated after 2 and 4 days of cold treatment, with the lowest expression occurring at 2 days. B qRT-PCR validation of the RNA-seq results, showing the expression patterns of four selected genes. Samples include a control (normal conditions) and plants subjected to 2 or 4 days of cold treatment.Figure 5. Overexpression of CaWRKY28 enhances cold stress tolerance in yeast. Growth of yeast cells expressing CaWRKY28 and the empty-vector control following exposure to 29 °C, 4 °C, 0 °C, and − 20 °C for the indicated durations (24 h, 48 h, and 72 h). Cell suspensions were serially diluted (from left to right) and spotted onto agar plates to assess viability. Expression profiles of CaWRKY genes in response to cold stress. A Heatmap depicting the expression of cold-responsive CaWRKYs in C. arabica L., as determined by RNA-seq. (a) Genes were upregulated after 2 and 4 days of cold treatment, with peak expression occurring at 4 days. (b) Genes were upregulated after 2 and 4 days of cold treatment, with peak expression occurring at 2 days. (c) Genes were upregulated after 2 days of cold treatment and downregulated after 4 days. (d) Genes were downregulated after 2 and 4 days of cold treatment, with the lowest expression occurring at 4 days. (e) Genes were downregulated after 2 and 4 days of cold treatment, with the lowest expression occurring at 2 days. B qRT-PCR validation of the RNA-seq results, showing the expression patterns of four selected genes. Samples include a control (normal conditions) and plants subjected to 2 or 4 days of cold treatment.Figure 5. Overexpression of CaWRKY28 enhances cold stress tolerance in yeast. Growth of yeast cells expressing CaWRKY28 and the empty-vector control following exposure to 29 °C, 4 °C, 0 °C, and − 20 °C for the indicated durations (24 h, 48 h, and 72 h). Cell suspensions were serially diluted (from left to right) and spotted onto agar plates to assess viability. Correlation analysis between CaWRKY expression and physiological indices revealed distinct association patterns (Figure S2). Specifically, relative conductivity, total soluble sugars, and CAT activity were strongly positively correlated with CaWRKYs that were upregulated at both 2 and 4 days (peaking at 4 days), but negatively correlated with those showing progressively decreasing expression from 2 to 4 days. In contrast, starch content and POD activity were strongly positively correlated with CaWRKYs that were transiently induced at 2 days and downregulated by 4 days. These results suggest that cold-responsive CaWRKYs may function through modulating the corresponding physiological indices. To determine the accuracy and reliability of our RNA-seq dataset, qRT-PCR was performed on a subset of CaWRKY genes that exhibited differential expression patterns under cold stress conditions, particularly those with comparatively high transcriptional abundance. The ten genes subjected to qRT-PCR validation were CaWRKY01, CaWRKY12, CaWRKY17, CaWRKY28, CaWRKY34, CaWRKY57, CaWRKY64, CaWRKY69, CaWRKY85, and CaWRKY86. qRT-PCR demonstrated a significant upregulation/downregulation of these genes in cold-affected leaves, which agreed with the expression trends observed using RNA-seq (Fig. 4B). This consistency between the two independent experimental approaches provides strong evidence supporting the robustness and validity of our transcriptomic findings. Expression of CaWRKY28 enhanced the tolerance of yeast to cold stress Given that its counterpart in C. canephora, CcWRKY14, is a known key cold‑responsive gene, and its ortholog in Arabidopsis, AtWRKY40, functions as an important stress‑responsive transcription factor, CaWRKY28 was prioritized for functional validation [24, 42]. The coding sequence of CaWRKY28 was therefore cloned into the pYES2 vector and heterologously expressed in the yeast strain BY4741. Transformed yeast expressing CaWRKY28 and the empty‑vector control were subjected to cold stress at 4 °C, 0 °C, and − 20 °C for 24 h, 48 h, and 72 h, respectively. Under each condition, yeast expressing CaWRKY28 exhibited markedly better growth than the control (Fig. 5), indicating that CaWRKY28 enhances cold tolerance in this heterologous system. Fig. 5Overexpression of CaWRKY28 enhances cold stress tolerance in yeast. Growth of yeast cells expressing CaWRKY28 and the empty-vector control following exposure to 29 °C, 4 °C, 0 °C, and −20 °C for the indicated durations (24 h, 48 h, and 72 h). Cell suspensions were serially diluted (from left to right) and spotted onto agar plates to assess viability Overexpression of CaWRKY28 enhances cold stress tolerance in yeast. Growth of yeast cells expressing CaWRKY28 and the empty-vector control following exposure to 29 °C, 4 °C, 0 °C, and −20 °C for the indicated durations (24 h, 48 h, and 72 h). Cell suspensions were serially diluted (from left to right) and spotted onto agar plates to assess viability Discussion CaWRKY genes in C. arabica L Considering the important roles of WRKY genes in plant development and stress responses, genome-wide analyses of WRKY family genes have been conducted in many plant species, including Arabidopsis, rice, poplar, and banana [11, 13, 43, 44]. In coffee, 49 CcWRKY genes were systematically identified based on the genome sequence of C. canephora, which is recognized as a natural diploid progenitor of the allopolyploid species C. arabica L [24]. Seventeen CaWRKY genes have been identified based on coffee expressed sequence tags; however, a systematic genome-wide analysis of the CaWRKY gene family remains to be conducted [25]. In the present study, we identified 101 CaWRKY genes in C. arabica L. based on newly published genome sequencing results [28]. This was approximately twice the number found in C. canephora (Table S2 and Fig. 1). Pan-genomic analyses capture a broader genetic diversity in a species than single reference genomes, offering a more comprehensive foundation for gene family characterization and functional investigation [45]. Despite the absence of published pan-genomic data for C. arabica L., we systematically identified the CaWRKY gene family in three cultivars: Bourbon, Geisha, and Kona. In total, 400 CaWRKYs were identified and classified into 49 OGGs (Table S2 and Table S3). Among these, 47 OGGs were present in all genomes. These results, together with prior studies on gene family evolution in other plant species, suggest that C. arabica L. varieties underwent divergent evolutionary trajectories to adapt to distinct environmental pressures. Phylogenetic and gene duplication analysis of CaWRKYs In agreement with the classification frameworks established for Arabidopsis and other crop species, the CaWRKY gene family in C. arabica was systematically categorized into three principal groups (Fig. 1) [11]. Among these, Group II was further partitioned into five distinct subgroups (Fig. 1). Phylogenetic scrutiny, underpinned by comprehensive multiple sequence alignment, showed that the predominant proportion of CaWRKY proteins possess the highly conserved heptapeptide motif WRKYGQK in their N-terminal regions. This motif serves as a fundamental hallmark for DNA-binding functionality. A commonly observed variant, WRKYGKK, was identified in this study, echoing analogous modifications previously reported in WRKY proteins in Arabidopsis, maize, and Bergenia purpurascens (Table S2) [11, 46, 47]. Additionally, cultivar-specific motifs, WIKYGEK and WRKSYYR, were uniquely identified in C. arabica L. cultivars Bourbon, Kona, and Geisha (Table S3). Mutations in WRKY structural domains may alter their binding specificity, enabling the recognition of DNA sequences beyond the canonical W-box motif [48]. Variations in WRKY domains can drive functional divergence, highlighting the extensive functional diversity present in the CaWRKY gene family across cultivars. Responses of CaWRKY genes under cold stress. In China, over 98% of coffee production is concentrated in Yunnan Province, where C. arabica L. variety Catimor, CIFC7963, is predominantly cultivated [4, 49]. To enhance bean quality, farmers grow C. arabica L. at higher altitudes, where pest and disease pressure is lower. However, these elevations prolong exposure to low temperatures, which is a major climatic stressor contributing to yield losses in Yunnan [4]. Beyond China, cold stress significantly impacts global coffee production, particularly in Brazil, which is the largest coffee producer globally [50]. To characterize C. arabica’s phenotypic responses to cold stress, numerous studies have employed seedlings [2, 5]. Cold treatment induced visible leaf damage, significantly impaired net photosynthesis and chlorophyll a fluorescence, and increased the ascorbate peroxidase activity and starch, sucrose, and total soluble sugar contents, indicating the effects of cold stress on the photosynthetic apparatus, antioxidant defense, and membrane systems [2, 50, 51]. In this study, we first employed fruiting-stage plants to investigate the cold response of C. arabica L. (Fig. 3). The cold-induced changes in leaf parameters followed trends consistent with previous studies. Notably, cold stress not only impaired bean quality in the current season but also increased the floral bud abortion rate for the subsequent season, a key finding that merits further research (Fig. 3 and Table 1). Table 1Caffeine, trigonelline, crude protein, crude fat, reducing sugar, and chlorogenic acid contents determined in green coffee beans harvested after cold treatment. The results are expressed as the means ± SD (standard deviations, n = 3). Different lowercase letters within a column indicate significant differences between values, according to Tukey’s test (p ≤ 0.05)Cold treatmentsCaffeine (%)Trigonelline (%)Crude protein (%)Crude fat (g/kg)Reducing sugars (%)Chlorogenic acid (mg/g)Control0.96 ± 0.03a0.88 ± 0.06a9.52 ± 0.80a75.08 ± 2.01a1.02 ± 0.04a36.32 ± 2.67a2 days0.82 ± 0.02ab0.84 ± 0.01a8.42 ± 0.52a72.86 ± 5.21ab1.22 ± 0.12a40.48 ± 2.66ab4 days0.79 ± 0.10b0.86 ± 0.02a8.60 ± 0.52a66.45 ± 1.80b1.12 ± 0.13a42.89 ± 1.81b Table 1Caffeine, trigonelline, crude protein, crude fat, reducing sugar, and chlorogenic acid contents determined in green coffee beans harvested after cold treatment. The results are expressed as the means ± SD (standard deviations, n = 3). Different lowercase letters within a column indicate significant differences between values, according to Tukey’s test (p ≤ 0.05)Cold treatmentsCaffeine (%)Trigonelline (%)Crude protein (%)Crude fat (g/kg)Reducing sugars (%)Chlorogenic acid (mg/g)Control0.96 ± 0.03a0.88 ± 0.06a9.52 ± 0.80a75.08 ± 2.01a1.02 ± 0.04a36.32 ± 2.67a2 days0.82 ± 0.02ab0.84 ± 0.01a8.42 ± 0.52a72.86 ± 5.21ab1.22 ± 0.12a40.48 ± 2.66ab4 days0.79 ± 0.10b0.86 ± 0.02a8.60 ± 0.52a66.45 ± 1.80b1.12 ± 0.13a42.89 ± 1.81b Caffeine, trigonelline, crude protein, crude fat, reducing sugar, and chlorogenic acid contents determined in green coffee beans harvested after cold treatment. The results are expressed as the means ± SD (standard deviations, n = 3). Different lowercase letters within a column indicate significant differences between values, according to Tukey’s test (p ≤ 0.05) Given the crucial roles of WRKY TFs in various physiological and stress-related processes across the plant kingdom, genome-wide identification and characterization of cold-responsive WRKY genes have been extensively conducted in several plant species, including Prunus davidiana, B. purpurascens, Acer fabri, elephant grass, and eggplant [46, 52–55]. In C. arabica L., our transcriptome analysis identified 47 CaWRKY genes that respond to cold stress (Fig. 4). Among these, CaWRKY15, CaWRKY14, CaWRKY74, and CaWRKY75 were predominantly upregulated at two days post‑treatment. Interestingly, their Arabidopsis homolog AtWRKY41 negatively regulates cold acclimation by suppressing CBF1 expression [56], suggesting a potential functional divergence of this clade in coffee. CaWRKY28 and CaWRKY34 were also upregulated under cold stress (Fig. 4). Heterologous expression of CaWRKY28 enhanced cold tolerance in yeast (Fig. 5), providing direct evidence for its role in this stress response. Notably, while the Arabidopsis homologs AtWRKY40 and AtWRKY60 are known to modulate stress‑responsive organellar genes [42], and AtWRKY40 functions upstream of several ABA signaling transcription factors such as ABI4, ABI5, and DREB1/2A [57], direct evidence for their involvement in cold tolerance remains limited. In this study, CaWRKY28 expression showed a strong positive correlation with total soluble sugar content, CAT activity, and relative conductivity (Figure S2). These correlations point to a potentially conserved role for CaWRKY28 in cold adaptation in coffee, possibly through the regulation of sugar metabolism or the activation of antioxidant systems. Supporting this, the C. canephora homolog of CaWRKY28, CcWRKY14, is also cold‑responsive, further reinforcing its status as a key candidate in the cold stress response. A comparative analysis of additional cold-responsive genes revealed further insights. AtWRKY21, a cold-responsive gene in Arabidopsis [58], and its putative coffee homolog, CaWRKY49, was downregulated in our study. Conversely, cold-upregulated Arabidopsis genes such as AtWRKY6, AtWRKY28, AtWRKY29, and AtWRKY23 [58] have orthologs in C. arabica L. with conserved cold-induction patterns. For instance, CaWRKY85 and CaWRKY86 (putative ortholog of AtWRKY6) showed peak expression after four days of cold treatment, while CaWRKY82 and CaWRKY89 (putative ortholog of AtWRKY28) exhibited a more delayed upregulation (Fig. 4). Although the precise functions of most cold-responsive CaWRKYs are yet to be determined, candidates such as CaWRKY28, CaWRKY82, CaWRKY85, CaWRKY86, and CaWRKY89 show significant promise for breeding applications. Their distinct expression dynamics and inferred functional relationships with characterized Arabidopsis homologs provide a strong foundation for future mechanistic studies, which are necessary to fully elucidate their specific roles in cold tolerance. Conclusions Taken together, a total of 400 CaWRKY genes were identified from the four C. arabica genomes and were classified into 49 orthogroups (OGGs), 47 of which were present in all four genomes. Phylogenetic analysis divided the CaWRKYs into three major groups based on their conserved WRKY domains and zinc‑finger structures. Analyses of conserved protein motifs, exon–intron structures, and gene duplication events indicated that the expansion of the CaWRKY family was primarily driven by segmental duplications. In response to the needs of coffee cultivation, we characterized the physiological responses of fruiting coffee trees under cold stress and identified 47 cold‑responsive CaWRKYs. These were classified into five clusters, including up‑regulated genes such as CaWRKY28, 82, 85, 86, 89, and others. Notably, heterologous expression of CaWRKY28 enhanced cold tolerance in yeast, suggesting that it represents a promising candidate gene for improving cold tolerance in C. arabica L. and related coffee species through genetic engineering. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2.