Fruits of ‘Hongmi’ (HM), ‘Xiangshui’ (XS), ‘Mangguo’ (MG), ‘Weiduoliya’ (WD), and ‘HongXiangshui’ (HXS) were harvested from the pineapple germplasm orchard of the Tropical Crops Genetic Resources Research Institute, Chinese Academy of Tropical Agricultural Sciences, located at Baodao Xincun, Danzhou, Hainan, China (19°29′17″N, 109°29′4″E; elevation 130 m; mean annual temperature 23 °C; relative humidity 85%; annual precipitation ~1,500 mm). Fruits were sampled at three aroma development stages: the non-aromatic stage (approximately 56 days after flowering, with no detectable aroma by sensory evaluation and a total soluble solids (TSS) content of about 12–13°Brix), the initial aroma stage (approximately 63 days after flowering, with a faint fruity aroma detected and a TSS content of about 14–15°Brix), and the strong aroma stage (approximately 68 days after flowering, characterized by a pronounced sweet aroma and a TSS content of about 15–18°Brix). The stage classification was based on sensory evaluation in combination with days after flowering and total soluble solids content. After harvest, fruits were held in the laboratory for 24 h to equilibrate volatiles and minimize field temperature/humidity effects. Samples were then immediately frozen in liquid nitrogen and stored at -80 °C until analysis.

The pineapple reference genome sequences and gene structural annotation were downloaded from the pineapple genome database (https://ananas.watchbio.cn). CXE protein sequences from Arabidopsis retrieved from UniProt were used as queries for BLASTP (v2.16.0) searches against the pineapple proteome (E-value < 1 × 10^-5). Candidate AcCXEs were further screened by homology against the UniProtKB/Swiss-Prot database to remove redundant entries. Conserved domains were predicted with InterPro (https://www.ebi.ac.uk/interpro/, v107.0), and proteins containing the α/β-hydrolase fold (Pfam: PF07859) were retained as CXE candidates. All gene structures of AcCXEs were further curated by GSAman (https://tbtools.cowtransfer.com/s/a11146181df14f, v0.9.53). Physicochemical properties of AcCXE proteins were computed using TBtools-II (v2.363) (Chen et al., 2023), and subcellular localizations were predicted with WoLF PSORT (https://wolfpsort.hgc.jp/).

CXE protein sequences from Arabidopsis, peach, Nanguo pear, tomato, and apple were retrieved from NCBI, GDR (https://www.rosaceae.org/), and TAIR. Homologs were identified using two approaches hmmsearch and BLASTP (v2.16.0), and redundant entries were removed. The filtered CXE sets from these species were combined with pineapple AcCXEs to infer a maximum-likelihood phylogeny using the “One Step Build a ML Tree” tool in TBtools-II (v2.363). The resulting tree was formatted and annotated in Evolview (https://www.evolgenius.info/evolview/#/treeview).

AcCXE loci and their annotations were processed in TBtools-II (v2.363) (“One Step MCScanX—Super Fast”) to generate chromosome-level gene-distribution files and map the physical positions of AcCXE genes. Intra-genomic duplication relationships (tandem and segmental) among CXEs were then identified with MCScanX (v1.0.0). The results were visualized using the “Advanced Circos” module in TBtools-II (v2.363).

For gene structure and conserved-domain analyses, conserved domains of AcCXE proteins were predicted using NCBI CDD (Batch) (Marchler-Bauer and Bryant, 2004) and Pfam with default parameters. Conserved motifs were identified across the 20 AcCXE proteins using MEME Suite (https://meme-suite.org/meme/, v5.5.8) (Bailey et al., 2009), with the maximum number of motifs set to 10 to capture motif types and counts across subfamilies. Gene structure, conserved motifs, conserved domains, and sequence identifiers were then integrated and visualized in TBtools-II (v2.363).

Genomic coordinates of CXE loci were obtained from the GP genome GFF using TBtools-II (v2.363), and the 2,000-bp sequences upstream of the translation start codon (ATG) were extracted as putative promoter regions. These sequences were submitted to PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) for prediction of cis-acting regulatory elements (Lescot et al., 2002). Detected element types were then enumerated, classified, and summarized.

Transcriptome datasets for root, stem, leaf, petal, ovule, and fruit core were obtained from NCBI BioProject PRJNA483249 (Mao et al., 2018). RNA-seq data for different cultivars and developmental stages were generated by our laboratory; three biological replicates per cultivar were sequenced, and expression values were averaged. All reads were quantified with Kallisto (v0.51.1) and normalized as TPM (Bray et al., 2016). Fruits of ‘Hongmi’ (HM), ‘Xiangshui’ (XS), and ‘HongXiangshui’ (HXS) were sampled at three aroma stages (aroma-absent, aroma-onset, aroma-intense). Total RNA was extracted and reverse-transcribed for RT–qPCR. Primer specificity was assessed using Primer5, and primers were synthesized by Wuhan Zhuandao Biotechnology Co., Ltd. Actin served as the reference gene (Yi et al., 2023; Zhang et al., 2024b). Relative expression was calculated with the 2^−ΔΔCt method (Rao et al., 2013), with three technical replicates per sample. Statistical analyses were performed in IBM SPSS Statistics 27 using two-tailed t-tests, and figures were prepared in GraphPad Prism 8.0.

From the pineapple genome, we identified 20 CXE family members and named them AcCXE1–AcCXE20 in ascending order of chromosomal position. Protein physicochemical analysis showed lengths of 169–463 amino acids and predicted isoelectric points (pI) of 4.73–8.93 (Table 1). Subcellular localization prediction indicated 12 proteins in the cytosol and 6 in chloroplasts; AcCXE10 was predicted to localize to the endoplasmic reticulum, and AcCXE9 to the nucleus.

To elucidate the evolution of the pineapple CXE family, we constructed a maximum-likelihood (ML) phylogeny comprising 147 CXE proteins from six species: 20 from pineapple, 20 from Arabidopsis thaliana, 35 from Nanguo pear (Pyrus ussuriensis), 33 from peach (Prunus persica), 23 from tomato (Solanum lycopersicum), and 16 from apple (Malus domestica), and grouped them accordingly (Figure 1A). The tree resolved five major clades (Group I–V), with Group I and Group II containing the largest numbers of members (Figure 1D), suggesting these clades dominated family expansion. Pineapple CXEs were distributed across all five clades. Notably, AcCXE4 clustered with PuCXE15, a gene implicated in ester degradation in Nanguo pear (Qi et al., 2023); AcCXE7 clustered with tomato SlCXE1 (Goulet et al., 2012), and AcCXE3 with apple MdCXE1 (Souleyre et al., 2011), indicating that AcCXE4, AcCXE7, and AcCXE3 may participate in ester hydrolysis in pineapple. In terms of family size, Nanguo pear and peach harbored the most CXEs (35 and 33, respectively), whereas pineapple possessed 20. Overall, the phylogeny and gene counts indicate clade-level conservation with lineage-specific diversification.

To assess the chromosome distribution of AcCXEs, we visualized their loci with a Circos plot (Figure 2A). The genes are dispersed across 12 chromosomes. Single-copy loci occur on contig04, contig06, contig10, contig16, contig18, and contig23, indicating an overall scattered pattern. Several genes co-localize on the same chromosome, for example, AcCXE1 and AcCXE2 on contig03, and AcCXE17 and AcCXE19 on contig21. Inspection of collinearity links revealed both tandem and segmental duplications that likely contributed to family expansion: AcCXE7-AcCXE8-AcCXE4-AcCXE5 form a tandem array, whereas AcCXE6 and AcCXE16 represent a segmental duplicate pair.

Comparative synteny showed conserved collinearity between pineapple CXE loci and those in banana, tomato, rice, and Arabidopsis, albeit with different counts. We detected 14 and 6 syntenic pairs between pineapple and banana or tomato, respectively (Figure 2B), and 15 and 2 pairs between pineapple and rice or Arabidopsis, respectively (Figure 2C). The higher numbers for banana and rice relative to tomato and Arabidopsis suggest substantial divergence in the chromosomal neighborhoods harboring CXE genes between monocots and dicots. Given that banana and tomato are well-studied for fruit aroma, these syntenies provide a useful reference for inferring the roles of pineapple CXEs in volatile ester metabolism and aroma formation.

To further explore conserved features of pineapple CXEs, we analyzed their protein domains and motifs. Ten conserved motifs were identified across AcCXE sequences (Figure 3B). Motifs 1, 2, 3, 6, and 10 are highly conserved and present in all members. Motif number and order are broadly consistent within subclades, whereas several genes carry fewer motifs-e.g., AcCXE15-suggesting functional specialization. Gene-structure analysis (Figure 3D) showed that most AcCXEs contain 3–4 exons; members within the same phylogenetic branch share similar architectures (Figure 3A), indicating lineage-specific structural diversification. AcCXE13 and AcCXE20 are conserved at the gene-structure and sequence levels, yet their exon counts and motif compositions differ from most other AcCXEs, implying possible neofunctionalization. All AcCXEs harbor the conserved α/β-hydrolase superfamily domain, and most also contain the Abhydrolase_3 domain.

To explore potential regulatory events of AcCXE genes in abiotic stress and development, we surveyed cis-acting elements within the 2,000-bp upstream promoter regions (Figure 4). After excluding ubiquitous core elements such as the CAAT-box and TATA-box, a total of 452 cis-elements were identified and classified into four major categories: light-responsive, stress-responsive (abiotic), development-related, and hormone-responsive elements.

Among them, light-responsive cis-elements were abundant—including G-box, Box4, GT1-motif, and TCT-motif. Box4 occurred most frequently in AcCXE10, whereas G-box was most frequent in AcCXE9, suggesting important roles for these genes in light signal transduction. Hormone-responsive elements were enriched for ABRE (ABA-responsive), TGACG-motif (MeJA-responsive), TGA-element (auxin-responsive), and TCA-element (salicylic acid–responsive); the prevalence of MeJA-related motifs indicates extensive involvement of the family in jasmonate-mediated regulation. Stress-associated elements (ARE, MBS, LTR, TC-rich repeats) were common, implicating AcCXEs in responses to low temperature, drought, and anaerobic stress. Development-related elements—including CAT-box, A-box, and O2-site—were also frequent, consistent with regulation of tissue-specific expression and developmental processes.

To characterize the expression profile of AcCXEs across cultivars, tissues, and fruit development, we analyzed RNA-seq datasets. By cultivar (Figure 5A), AcCXE5 and AcCXE7 were highly expressed in the light-aroma cultivar ‘HongXiangshui’ (HXS), consistent with a putative negative role in aroma formation. By tissue (Figure 5B), AcCXE4, AcCXE5, and AcCXE9 showed elevated expression in the fruit core, which exhibits weak aroma, again aligning with negative regulation. Across developmental stages (Figure 5C), AcCXE3, AcCXE4, and AcCXE13 were down-regulated as fruit aroma intensified, suggesting repression of ester accumulation. Collectively, these patterns nominate six candidates, AcCXE3, AcCXE4, AcCXE5, AcCXE7, AcCXE9, and AcCXE13, as key genes associated with pineapple aroma metabolism.

Six candidates, AcCXE3, AcCXE4, AcCXE5, AcCXE7, AcCXE9, and AcCXE13, were selected for RT-qPCR validation (Figure 6). AcCXE4 and AcCXE7 showed higher expression in the light-aroma cultivar ‘HongXiangshui’ (HXS) than in the sweet/fruit-aroma cultivars ‘Hongmi’ (HM) and ‘Xiangshui’ (XS). Notably, AcCXE4 expression decreased with ripening in XS but increased in HXS, consistent with a role for CXEs as negative regulators of ester accumulation. These results implicate AcCXE4 and AcCXE7 as key genes closely associated with pineapple aroma formation.