All plant materials were cultivated in the Spathiphyllum germplasm repository at Sanming Academy of Agricultural Sciences (Fujian, China). Two cultivars, Meijiu’ and its natural dwarf mutant ‘Meibian’, were sampled at the initial flowering stage under standardized greenhouse conditions (Figure 1A; Supplementary Figure S1). Four representative tissues—leaf (L), petiole (P), spathe (bract, B), and spadix (S)—were collected from each plant, with three biological replicates per tissue. The samples were immediately frozen in liquid nitrogen and stored at –80 °C for further experiments. Total RNA was extracted using TRIzol reagent (Invitrogen, USA), and RNA quality was assessed using NanoDrop spectrophotometry (Thermo Scientific, Waltham, MA, United States), and Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, United States). To conduct full-length transcriptome sequencing, RNA samples were mixed together from the four tissues of both cultivars.

A total of 2 µg RNA was used to construct SMRTbell libraries with the SMRTbell^® Express Template Prep Kit 2.0 (PacBio), following the Iso-Seq Express protocol. Sequencing was performed on the PacBio Sequel II platform (Berry Genomics Co., Ltd., Beijing, China) (Gonzalez-Garay, 2015). Sequence data were processed using the SMRT Analysis software (ISOseq, version 4.2.0). Raw reads were processed using SMRTLink v10.1 to generate HiFi reads (parameters: --min-passes=3, --min-rq=0.99). Full-length non-chimeric (FLNC) reads were defined as sequences containing both 5′ and 3′ primers and poly(A) tails, excluding internal primers. FLNC reads were clustered to generate high-quality consensus isoforms. Redundancy was removed using CD-HIT (version 4.8.1; -c 0.90 -s 0.85), resulting in a non-redundant transcript set. Coding sequences (CDSs) were predicted using TransDecoder, with an ORF length cutoff of 100 amino acids. Functional annotation was conducted against the NR (NCBI non-redundant protein database), Swiss-Prot (manually curated protein database), KOG (EuKaryotic Orthologous Groups), KEGG, InterPro, and Gene Ontology (GO) databases using DIAMOND BLASTX (E-value < 1e–10).

Transcriptome profiling of the four tissues from both genotypes was performed using Illumina NovaSeq 6000 (Genedenovo Biotechnology Co., Ltd., Guangzhou, China), yielding 24 paired-end libraries (150 bp). Raw sequencing reads were processed with fastp (version 0.23.4) for quality control and adaptor trimming to generate high-quality clean reads (Chen, 2023). PacBio full-length transcript sequences were used as reference transcripts. Gene expression levels were quantified with Bowtie2 (version 2.5.1) for alignment Clean reads were mapped to the PacBio-derived reference transcriptome using Bowtie2 (version 2.5.1), and expression levels were calculated with RSEM, normalized as fragments per kilobase of transcript per million mapped reads (FPKM) (Langmead and Salzberg, 2012). Differentially expressed genes (DEGs) were identified using DESeq2 with thresholds of |log₂(fold change)| ≥ 1 and false discovery rate (FDR) < 0.05 (Love et al., 2014). Functional enrichment analyses were performed using the clusterProfiler and topGO packages (Alexa and Rahnenführer, 2009; Wu et al., 2021). The final visualization was generated in R (version 4.4.0). WGCNA was applied to the expression matrix of all DEGs using the WGCNA R package (Langfelder and Horvath, 2008). A soft threshold power was selected based on scale-free topology criteria (R² > 0.85). Module–trait relationships were calculated via Pearson correlation between module eigengenes and phenotype traits. Hub genes were defined by high module membership (MM > 0.8) and gene significance (GS > 0.6).

Poly(A)+ RNAs were isolated and subjected to 5′ and 3′ RACE using the GeneRacer™ Kit (Invitrogen) to obtain the full-length coding sequence of SpRCC1. For 3′ RACE, poly(A)+ RNAs were reverse transcribed with the GeneRacer oligo(dT) primer, and the resulting cDNAs were amplified using the GeneRacer 3′ outer/inner primers and gene-specific primers. The 5′ end of SpRCC1 was amplified using the 5′ RACE System. Amplified fragments were purified, cloned, and sequenced. Primer sequences used in RACE are listed in Supplementary Table 1. The CDS of SpRCC1 was cloned into the pBWA(V)H2STMVΩ-egfp vector, and the constructed plasmid was transformed into Agrobacterium tumefaciens GV3101 and transiently expressed in Nicotiana benthamiana. The fluorescence signal was observed using a LSM 880 microscope (Zeiss, Germany) for subcellular localization analysis, with NLS-RFP used as a nuclear marker.

First-strand cDNA was synthesized using the HiScript^® Reverse Transcriptase kit according to the manufacturer’s protocol. The resulting cDNA was diluted 10-fold and used as a template for quantitative real-time PCR (qRT-PCR) with ChamQ Universal SYBR^® qPCR Master Mix. Actin was used as an internal reference gene. Relative expression was calculated using the 2^–ΔΔCt method. Primer sequences are listed in Supplementary Table 1.

The full-length SpRCC1 CDS was inserted into the binary vector pBWA(V)BS and transformed into Arabidopsis thaliana Col-0 via the floral dip method. Transformed plants were grown under a 16 h light/8 h dark photoperiod at 25 °C, and mature T1 seeds were harvested and dried with silica gel. Following surface sterilization, seeds were selected on ½ MS medium with 15 mg L^-1 basta. Resistant seedlings were transferred to soil and grown under the same photoperiod and temperature conditions. Genomic DNA was extracted from young leaf tissues using the CTAB method. Transgene integration was verified by PCR using gene-specific and vector-specific primers (Supplementary Table 1). Segregation analysis of T1 progeny confirmed a 3:1 ratio of resistant to sensitive seedlings, indicative of single-locus insertion. Independent T2 lines with single-copy insertions were further propagated, and homozygous T2 lines were identified by PCR-based genotyping. At eight weeks post-germination, leaf morphology was examined, and differences in leaf area were recorded between over-expression lines and wild-type (WT, Col-0). For expression validation, rosette leaves were harvested from both genotypes, flash-frozen in liquid nitrogen, and subjected to qRT-PCR to assess SpRCC1 transcript abundance.

The Spathiphyllum cultivar ‘Meibian’ (MBZ) is a naturally occurring mutant derived from the standard cultivar ‘Meijiu’ (BZ). Compared to ‘Meijiu’, Meibian’ exhibits markedly smaller vegetative and reproductive organs, resulting in a compact overall growth habit (Figure 1A). To elucidate the molecular basis of this architectural variation, we performed full-length transcriptome sequencing of S. kochii. A total of 10,081,745 circular consensus sequencing (CCS) reads were generated, yielding approximately 15.65 Gb of high-quality data. Most of the CCS reads ranged from 1,000 to 2,000 base pairs (bp) in length, consistent with expected full-length cDNA distributions (Figure 1B; Supplementary Table 2). From these data, 523,557 high-confidence transcripts were identified. Transcript length distribution analysis revealed that the majority also fell within the 1,000–2,000 bp range (Figure 1C), further supporting the integrity of the sequencing data. After removing redundancy, a total of 122,346 non-redundant transcript isoforms were retained to establish the first full-length reference transcriptome of S. kochii.

Among the assembled transcripts, 94,809 transcripts (77.5%) were predicted to possess protein-coding potential, as determined by CDS analysis (Figure 1D). Notably, 71,992 transcripts (75.9% of coding transcripts) contained complete open reading frames (ORFs), including both canonical start and stop codons, highlighting the high structural accuracy of the assembled transcriptome. The analysis revealed that 92.5% of conserved core plant genes were completely recovered, including 139 single-copy and 255 duplicated orthologs, with only 1.2% of orthologs classified as missing (Figure 1E). These results confirm that the full-length transcriptome of S. kochii is highly complete, accurate, and suitable as a reference resource for future functional, regulatory, and comparative genomic analyses in this species.

Functional annotation was performed on the 94,809 non-redundant protein-coding transcripts derived from the full-length transcriptome assembly. A total of 85,055 transcripts (89.71%) were successfully annotated in at least one public database, underscoring the reliability and completeness of the assembled transcriptome. Notably, 48,594 transcripts were simultaneously annotated across all five databases used, reflecting the robustness of the annotation strategy and the comprehensive functional representation of the S. kochii transcriptome (Figure 2A; Supplementary Table 3). Gene Ontology (GO) classification assigned functional terms to 58,856 transcripts (Figure 2B). Within the biological process category, 40,662 transcripts were annotated, with the majority involved in metabolic processes (34,142 transcripts), followed by localization-related processes (9,228 transcripts) and regulation of biological processes (8,737 transcripts). In the cellular component category, 32,890 transcripts were annotated, predominantly associated with organelles (14,521 transcripts) and membranes (10,462 transcripts), indicating broad transcript coverage across subcellular compartments. For the molecular function category, 49,087 transcripts were annotated, with the most enriched functions including transferase activity (15,190 transcripts) and small molecule binding (14,788 transcripts), reflecting the transcriptome’s strong representation of enzymatic and regulatory functions.

A total of 64,354 transcripts were successfully annotated with KEGG pathway information and mapped to five major functional categories (Figure 2C). The largest category was ‘Metabolism’, encompassing a broad range of primary metabolic processes. Within this group, transcripts were particularly enriched in pathways related to carbohydrate metabolism (6,090 transcripts), amino acid metabolism (3,528 transcripts), and lipid metabolism (2,981 transcripts), highlighting the essential role of primary metabolism in sustaining cellular growth and developmental processes in S. kochii. The Genetic Information Processing category included transcripts involved in translation (3,861 transcripts), transcription (2,575 transcripts), and replication and repair (1,890 transcripts), indicating an active landscape of gene expression and genomic maintenance. The Environmental Information Processing category featured genes associated with signal transduction (2,827 transcripts) and membrane transport (918 transcripts), reflecting the plant’s capacity to perceive and respond to environmental stimuli. In the ‘Organismal Systems’ category, a notable number of transcripts (3,727) were assigned to environmental adaptation, suggesting that S. kochii possesses transcriptional programs responsive to abiotic or biotic environmental cues.

To elucidate the molecular mechanisms underlying the dwarf phenotype in S. kochii, we performed comparative transcriptomic profiling of four representative tissues from two cultivars (Figure 1A). Each RNA-seq library produced approximately 62.3 million raw paired-end reads, yielding a total of 191.22 Gb of raw data across all samples (Supplementary Table 4). After quality control and filtering, 188.82 Gb of high-quality clean data were retained for downstream analysis. Transcript abundance was quantified using the previously constructed full-length reference transcriptome.

Pearson correlation analysis demonstrated strong consistency among biological replicates for each tissue and genotype (Figure 3A). Differential expression analysis revealed a total of 2,660 significantly differentially expressed genes (DEGs) between ‘Meijiu’ and ‘Meibian’ across all tissues, reflecting extensive transcriptional reprogramming associated with the dwarf phenotype. Specifically, we identified 761, 1,105, 1,210, and 958 DEGs in the spathe, leaf, petiole, and spadix, respectively (Figure 3B). In all examined tissues, the number of downregulated genes exceeded upregulated genes, indicating a global trend of transcriptional repression in the dwarf mutant. This widespread reduction in gene expression likely contributes to its smaller organ size and compact growth morphology.

DEG overlap analysis revealed both tissue-specific and shared transcriptional expression changes between ‘Meijiu’ and ‘Meibian’. Among the upregulated DEGs, most were uniquely expressed in specific tissues: 324 in the leaf, 298 in the petiole, 214 in the spadix, and 210 in the spathe. Only 18 genes (1%) were consistently upregulated across all tissues (Figure 3C). In contrast, downregulated genes exhibited greater overlap, with 208 transcripts consistently repressed in all tissues examined (Figure 3D), suggesting the presence of a core transcriptional repression module associated with the dwarf phenotype of ‘Meibian’.

GO enrichment analysis of DEGs revealed organ-specific functional disruptions linked to growth inhibition. In the leaf, downregulated DEGs were significantly enriched in photosynthesis (GO:0015979) and response to BR (GO:0009741), suggesting that reductions in energy production and hormone signaling may be associated with leaf expansion. In the petiole, DEGs were enriched in BR homeostasis (GO:0010268), BR biosynthetic process (GO:0016132), and phytosteroid biosynthetic process (GO:0016129), indicating altered expression of genes involved in hormone-related pathways associated with cell elongation. A similar pattern was observed in the spadix, with suppression of genes involved in BR biosynthesis and response pathways, further supporting the hypothesis that impaired BR metabolism is a major factor contributing to reduced reproductive organ size. In the spathe, downregulated genes were predominantly associated with flavonoid biosynthesis (GO:0009813) and response to red or far-red light (GO:0009639), implying altered secondary metabolism and photomorphogenic signaling.

To further explore the functions of the 208 genes consistently downregulated across all tissues, GO enrichment analysis was performed (Figure 3E). In the biological process category, enriched terms included starch catabolic process, glucan catabolic process, response to BR, BR-mediated signaling pathway, and response to steroid hormone, representing transcriptional reprogramming related to developmental regulation. In the cellular component category, significant enrichment was observed for chloroplast, plastid, and thylakoid membrane-related terms, indicating impaired photosynthetic capacity and plastidial function in the dwarf mutant. In the molecular function category, enriched terms included carbohydrate binding, hydrolase activity, oxidoreductase activity, and notably, histone demethylase activity—implying potential involvement of epigenetic modifications in the transcriptional repression of growth-related genes.

Collectively, these findings collectively indicate that the dwarf phenotype in S. kochii arises from coordinated repression of key pathways controlling vegetative and reproductive development. The systemic downregulation of genes related to BR signaling, energy metabolism, and chloroplast function, coupled with reduced activity of regulatory enzymes such as histone demethylases, suggests a multi-layered regulatory mechanism underlying the compact phenotype of ‘Meibian’. This convergent repression likely defines a core transcriptional program that restricts cell expansion and organ growth, giving rise to the compact architecture of ‘Meibian’.

To further explore the regulatory networks associated with the dwarf phenotype in S. kochii, a weighted gene co-expression network analysis (WGCNA) was performed based on the expression profiles of the 2,660 DEGs identified across the four tissues. The analysis identified 19 distinct co-expression modules, each representing a cluster of genes with highly correlated expression patterns (Figure 4A). Among these, module M3 exhibited a strong negative correlation with the dwarf genotype (Pearson’s r = –0.998, p = 1.81 × 10^-28), indicating that genes within this module are tightly associated with plant size and growth regulation (Figure 4B). Module M3 contained 843 genes showing highly coordinated expression patterns, with markedly different expression levels between the standard cultivar ‘Meijiu’ and the dwarf mutant ‘Meibian’ (Figure 4C). Hub genes within M3 were identified based on high module membership and gene significance, reflecting their prominent positions within the co-expression network. Heatmap analysis revealed that the majority of these hub genes were consistently downregulated across all examined tissues in ‘Meibian’ (Supplementary Figure S2), suggesting a coordinated shift in the expression of genes linked to growth- and development-related processes. This widespread downregulation highlights transcriptional differences in co-expressed gene networks associated with the compact growth phenotype of the dwarf mutant.

Several representative hub genes in M3 have been previously implicated in plant growth–related pathways. SpBSK3 (BR Signaling Kinase 3) is a component of the brassinosteroid signaling pathway and has been reported to participate in cell elongation and organ growth (Zolkiewicz and Gruszka, 2022). SpHTH (HOT-HEAD-like protein) has been implicated in lipid metabolic processes associated with epidermal integrity and organ development (Krolikowski et al., 2003). SpJMJ25 (Jumonji domain-containing protein 25) encodes a histone demethylase implicated in the epigenetic regulation of growth-related genes (Yamaguchi, 2021), while SpBBX19 (B-box zinc finger protein 19) functions as a light-responsive transcription factor that modulates photomorphogenesis and internode elongation (Gao et al., 2024). Notably, SpRCC1 (Spathiphyllum kochii Regulator of Chromosome Condensation 1) is predicted to be nucleus-localized and involved in chromatin organization and cell cycle regulation. he interaction between RCC1 and small GTPases has been shown to play an essential role in plant development and stress responses, suggesting that SpRCC1 may act as a key integrator of chromatin remodeling and signal transduction during growth regulation. Furthermore, GO enrichment analysis of all genes within module M3 revealed significant overrepresentation of biological processes such as response to cytokinin (GO:0009735), negative regulation of catalytic activity (GO:0043086), and negative regulation of molecular function (GO:0044092) (Figure 4D). These results indicate that M3 integrates hormone-responsive signaling with transcriptional repression mechanisms. Together, these transcriptional enrichments suggest coordinated hormone-responsive pathways involving cytokinins, BR, and auxin that may contribute to growth and morphological variation in the dwarf mutant.

To elucidate the functional role of SpRCC1 in regulating plant architecture, the full-length cDNA sequence was obtained using RACE. The obtained cDNA sequence was 2,303 bp in length and encoded a 556–amino acid protein containing a conserved RCC1 domain (Pfam: PF00415), a characteristic feature of the Regulator of Chromosome Condensation 1 (RCC1) protein family in plants (Figures 5A–C). Structural prediction using AlphaFold2 revealed that SpRCC1 adopts a canonical seven-bladed β-propeller fold, a hallmark structural feature of RCC1 proteins known to function in Ran GTPase signaling and nucleocytoplasmic transport (Figure 5D). The conservation of this structural motif suggests that SpRCC1 likely retains biochemical functions related to chromatin organization and cell cycle regulation.

Subcellular localization analysis demonstrated that SpRCC1 is predominantly localized in the nucleus (Figure 5E). qRT-PCR analysis revealed that SpRCC1 expression was significantly lower in the dwarf mutant ‘Meibian’ compared to the standard cultivar ‘Meijiu’ across all tested tissues and developmental stages (Figure 5F). This consistent downregulation supports the functional association of SpRCC1 with growth regulation.

Functional characterization through heterologous overexpression in Arabidopsis thaliana demonstrated a marked enhancement of vegetative growth, with transgenic plants exhibiting significantly increased leaf area compared to wild-type controls (Figures 5G, H; Supplementary Figure S3). These morphological changes support a positive regulatory role of SpRCC1 in promoting plant growth. The phenotypic concordance between Arabidopsis overexpression lines and the reduced expression of SpRCC1 in the dwarf Spathiphyllum mutant further validates its functional role in determining plant size and architecture.