The materials used in this study were A. galanthum and A. fisulosum var. viviparum Makinot, which were obtained from the germplasm resource nursery of Sanping Base at Xinjiang Agricultural University. Anthers at different developmental stages were collected and preserved in FAA fixative (37% formaldehyde 10 mL, anhydrous ethanol 50 mL, glacial acetic acid 5 mL, ddH₂O 35 mL). Leaves, scapes, inflorescence meristems, and floral buds were also collected. Three replications were conducted for each treatment, and the collected materials were snap-frozen with liquid nitrogen and stored in a refrigerator at -80°C.

According to the classification period of anthers in A. thaliana (Sanders et al., 1999), the anthers at the developmental stages of S12-S14 (pollen maturation-anther dehiscence pollen release) were taken and stored in FAA fixative (37% formaldehyde 10 mL, anhydrous ethanol 50 mL, glacial acetic acid 5 mL, ddH₂O 35 mL). After gradient dehydration and paraffin embedding, the fixed sample material was sliced using a slicer with a thickness of 6 μm. The excess wax on the surface was washed off, and then the sample was dyed with safranine green dye, followed by dyeing with neutral gum sealant. The slices were dried before micro observation and imaging.

The pollen viability and stigma receptivity were determined using the methyl thiazolyl tetrazolium (MTT) staining method. A MTT sucrose solution (100 mg MTT dissolved in 5 mL of 50 g/L sucrose solution) was prepared, and the stigma was fully immersed in the dye solution. After 15 minutes at room temperature in the dark, the changes in the stigma were observed under an anatomical microscope, and the color changes of the stigma were noted post-treatment. Another MTT solution was prepared (500 mg of MTT dissolved in 100 mL of distilled water). The anthers that were about to burst were dissected to create a pollen suspension, and the solution was added drop by drop to a concave slide. After the addition of the MTT solution, the cover glass was placed over it and the slide was kept in a thermostat at 25°C in the dark for 15 minutes and observed under a microscope. Healthy pollen grains were stained blue-purple or reddish-brown, while inactive pollen grains were stained lightly or colorless (Dafni, 1992).

High throughput sequencing (Hlumina Novaseq 6000) was contracted through Novaseq Biotechnology Co. Samples were sequenced with 3 replicates for each sample and 17 gene libraries. It is noteworthy that three replicates from the flower buds of A. galanthum were contaminated with one replicate; however, all data originated from the same batch and this contamination did not compromise the integrity of the data. The raw data generated by the high-throughput sequencer were subjected to a cleaning process that involved the removal of low-quality sequences, spliced sequences, and sequences containing ambiguous bases. The cleaned data were subsequently aligned to the A. fistulosum reference genome using HISAT2 and StringTie software for sequence comparison and transcript assembly (Ewing et al., 1998; Kim et al., 2015; Pertea et al., 2015). The genomic data for A. fistulosum were obtained from the China National Sequence Archive (CNSA) at the following URL: https://ftp.cngb.org/pub/CNSA/data5/CNP0002276/CNS0461863/CNA0047403/chr.Afis.genome.fa.gz.

The number of corresponding reads for each transcript was quantified, and gene expression was measured using FPKM (fragments per kilobase of transcript per million fragments mapped (FPKM) (Trapnell et al., 2010). Differential gene expression analysis was conducted utilizing DESeq2 (version 1.20.0), with genes that satisfied the criteria of a P value (padjust) < 0.05 and |log2(fold change)| > 1 classified as differentially expressed genes (DEGs) (Love et al., 2014). Additionally, correlation analysis, principal component analysis, Gene Ontology (GO), and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed. The GO and KEGG enrichment analyses were executed using the Allium BD online software (https://allium.qau.edu.cn/Allium/tools_enrichment.php), with the Benjamini-Hochberg (BH) method employed for multiple test correction. A corrected P value (padjust) < 0.05 was deemed indicative of significant enrichment of genes associated with GO functions and KEGG pathways (Yang et al., 2024). Heat maps of RNA-seq data were generated using the HeatMap Illustrator tool within TBtools software.

In order to identify potential SKP1 genes in A. fistulosum, genomic data for A. fistulosum was retrieved from the China National Species Archive (CNSA) (https://ftp.cngb.org/pub/CNSA/data5/CNP0002276/CNS0461863/CNA0047403/chr.Afis.genome.fa.gz). Additionally, genomic data for Solanum lycopersicum (SL3.0, GCA_000188115, http://plants.ensembl.org/), A. thaliana (TAIR10, GCA_000001735, http://plants.ensembl.org/), and Oryza sativa (IRGSP-1.0, GCA_001433935, http://plants.ensembl.org/) were obtained from the Ensembl Plants database (http://plants.ensembl.org/). Subsequently, the SKP1 domain hidden Markov model (ID: PF01466) was downloaded from the Pfam database (http://pfam.xfam.org/) and the Hmmer tool was used to search and compare the SKP1 genome-wide proteins of A. fistulosum (E-value ≤ 1e−5) (Prakash et al., 2017). The retrieved proteins were then validated using the Conserved Domain Database program, accessible via NCBI (https://www.ncbi.nlm.nih.gov/), and the Search program, available through the Pfam database (http://pfam.xfam.org/).

The physicochemical properties of the AfSKP1 protein were determined using the Expasy online software (https://web.expasy.org/tools/) (Gasteiger et al., 2003). Subcellular localization was predicted using the online tool WoLF PSORT (https://wolfpsort.hgc.jp/).

The SKP1 protein sequences of A. fistulosum, S. lycopersicum, A. thaliana, and O. sativa were aligned using Muscle in MEGA 7.0 software. Family phylogenetic trees were constructed and analyzed using the Neighbor-joining (NJ) method with a bootstrap value of 1,000. The bootstrap repetition value was set at 1,000. The evolutionary tree was enhanced using the online tool EVOLVIEW (https://www.evolgenius.info/evolview-v2/) (Kumar et al., 2018). A conserved motif analysis was performed using MEME (https://meme-suite.org/tools/meme/) (Bailey et al., 2009). Gene structure maps and protein structural domain distribution maps were generated using TBtools software (Chen et al., 2020).

The chromosomal location information of the AfSKP1 gene was mapped using TBtools software, which also enabled the identification of its location distribution. MCScanX was employed for the covariance analysis of the AfSKP1 gene (Wang et al., 2012), while the Circos tool was utilized to generate chromosome covariance maps (Krzywinski et al., 2009).

The 2000 bp sequence upstream of the AfSKP1 gene was extracted using TBtools software. Cis-acting elements in the promoter region were predicted using the Plant Care online website (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) (Lescot et al., 2002). The HeatMap tool in TBtools software was then utilized to generate an expression heat map.

Protein interaction relationships were predicted using the STRING online website (https://string-db.org/), and the interaction networks were visualized with Cytoscape software (Szklarczyk et al., 2018; Su et al., 2014).

Total RNA was extracted using a Tengen Biochemical Technology (Beijing) Co. extraction kit. The RNA-prep Plant Kit (DP441) was employed for the isolation of RNA, while cDNA was obtained through reverse transcription using the PrimeScript™ RT reagent Kit (Perfect Real Time). The qRT-PCR reaction system (20 μL) comprised 10 μL of 2x TB Green qPCR Master Mix, 8.2 μL of ddH2O, 0.4 μL of each upstream and downstream specific primer, and 1 μL of cDNA template. The reaction system for quantitative real-time polymerase chain reaction (qRT-PCR) (20 μL) comprised 10 μL of 2x TB Green qPCR Master Mix, 8.2 μL of deionized water (ddH2O), 0.4 μL of each upstream and downstream specific primer, and 1 μL of cDNA template. The reaction procedure was conducted on a 7500Fast (96) real-time fluorescence quantitative PCR instrument. It commenced with pre-denaturation at 94°C for 120 s, followed by denaturation at 94°C for 5 s, annealing at 15 s, and extension at 55°C for 10 s. This was repeated for 45 cycles. The relative expression was calculated using three biological replicates and the 2-^ΔΔCT method, with onion Actin serving as the internal reference gene. The relative expression represents the ratio between the treatment and control groups (Yao et al., 2020). The fluorescent quantitative PCR primers were designed based on the conserved sequences of the AfSKP1 gene family members ( Table 1 ).

The results of anther sectioning are illustrated in Figure 1 . The anther structure of A. fisulosum var. viviparum Makinot exhibited a trend of shrinkage during the S12-S14 stages, which was in stark contrast to A. galanthum. At the S13 stage, the anther chamber could not crack, the maturation and release of pollen grains. Consequently, the anther became enveloped within itself by the S14 stage and ultimately detached along with the filaments. In contrast, at the S13 stage, the anther of A. galanthum was cracked, allowing for the release of mature pollen grains. By the S14 stage, the anther of A. galanthum had completed its function, collapsing to facilitate the successful release of pollen. At the mature stage of pollen grains (S13), the pollen grains of A. galanthum were full and deeply colored, while the pollen grains of A. fisulosum var. viviparum Makinot were deformed and lightly colored. A. fisulosum var. viviparum Makinot to develop mature pollen grains ultimately resulted in the abortion of its pollen.

The results of pollen treated with MTT solution are shown in Figure 2 . The A. galanthum pollen is stained grey or light grey to indicate inactivity ( Figure 2A ), and the A. fisulosum var. viviparum Makinot pollen is stained blue or light blue to indicate vitality ( Figure 2B ). Stigmas treated with MTT sucrose solution were stained blue-violet to indicate viable stigmas. A dark color indicates high viability, while a stigma with no change in color indicates that the stigma is not viable or has lost viability. The A. fisulosum var. viviparum Makinot is divided into 9 periods, from before the florets open to the drying of the florets. The stigma is blue-violet throughout the development period and is receptive ( Figure 2C ). The A. galanthum is divided into 14 periods from the time of floret dehiscence to the time of floret dehiscence; except for the pre-floret dehiscence and ovary expansion periods, all other developmental periods of the blue-purple stigma are fertile ( Figure 2D ). The A. fisulosum var. viviparum Makinot exhibits fertility during the pre-floret dehiscence stage when compared to A. galanthum. However, A. fisulosum var. viviparum Makinot is incapable of producing viable pollen and lacks an ovary expansion period. Due to the inability of A. fisulosum var. viviparum Makinot to mature and release pollen, the assessment of pollen activity, stigma pollination, and overall pollination is limited to the application of chemical reaction MTT staining. Future research will involve conducting pollen tube growth tests through distant hybridization to further investigate its fertility.

In order to find the important genes controlling anther development, combined with the results of paraffin section observation, RNA-seq analysis was carried out on the inflorescence meristems (before bolting), floral buds (before flowering) and leaves (before flowering) of the two materials. A total of 17 RNA libraries were prepared and sequenced. The sequencing generated 109.96 G of raw data and 106.30 G of filtered clean data, with a Q30 value exceeding 91.51%. The data obtained was of exceptional quality, making it well-suited for subsequent analysis.

After quality control data processing, a total of 28,436 differentially expressed genes were identified by analyzing the differential expression of genes in different samples ( Figure 3A ). A total of 7072 DEGs were obtained from inflorescence meristems (S-1-2) of both materials, and 7516 DEGs were obtained from floral buds (B-1-2) of both materials. A total of 2603 DEGs were obtained from inflorescence meristems and floral buds of A. fisulosum var. viviparum Makinot (1-B-S). A total of 4,248 DEGs were obtained from inflorescence meristems and floral buds of A. galanthum (2-B-S). Combined with the previous anther phenotyping study, this investigation further examines the differential gene expression between two treatment groups, S-1-2 and B-1-2, with a specific focus on genes associated with floral organ development. In the S-1-2 treatment group, 3 AfSKP1 genes—AfSKP1-1, AfSKP1-2, and AfSKP1-8—were identified, alongside 12 MADS-box genes and 38 MYB genes. Conversely, in the B-1-2 treatment group, the genes AfSKP1-1, AfSKP1-2, AfSKP1-8, AfSKP1-22, and AfSKP1-23 were detected. Collectively, these 5 AfSKP1 genes, in addition to 14 MADS-box genes and 36 MYB genes, were catalogued ( Supplementary Table 1 ). In the current study, functional analyses were conducted based on the three floral development-related differential genes identified in the preliminary overview. The results indicated that the functions of the AfSKP1 genes are primarily associated with oocyte meiosis (KO04114), the cell cycle (KO04110), and ubiquitin-mediated proteolysis (KO04120) ( Figure 3B ). Additionally, these genes were linked to male meiosis (GO:0007140), mitotic nuclear division (GO:0007067), and the regulation of the mitotic cell cycle (GO:0007346) ( Figure 3C ). Consequently, it is hypothesized that the identification of the AfSKP1 gene may play a significant role in the regulation of anther development. Further investigation into the AfSKP1 gene family is warranted to deepen our understanding of its functional implications.

In this study, 34 potential AfSKP1 genes were initially identified in A. fisulosum through genome-wide identification analysis and named AfSKP1-1 to AfSKP1-34 based on their positions in the genome chromosome. The results of protein physicochemical property analysis ( Table 2 ) showed that the number of amino acid residues of AfSKP1 family members ranged from 57 aa to 385 aa, and the molecular weights of the proteins ranged from 6556.28 to 43941.54. Among them, AfSKP1-19 were the smallest with 55 aa and 6556.28, while AfSKP1-9 were the largest with 385 aa and 43941.54. The isoelectric points ranged from 4.25 to 6.4, with AfSKP1-18 being the lowest and AfSKP1-13 the highest. All AfSKP1 gene family proteins are hydrophilic except AfSKP1-34 The secondary structures of proteins encoded by the AfSKP1 gene family mainly include α-helix, β-sheet, and random coil. The 34 proteins encoded by the AfSKP1 gene family were predominantly α-helix (42.58% - 86.21%) and random coil (10.94% - 45.38%), with less β-sheet (1.47% - 9.72%). Subcellular localization prediction ( Table 2 ) indicated that 15 AfSKP1 family members were localized in the cytosol, 8 in the mitochondria, 6 in the chloroplast, 4 in the nucleus, and only AfSKP1-3 in the cytoskeleton.

Chromosomal localization analysis showed ( Figure 4 ) that 34 members of the AfSKP1 gene family were distributed on 8 chromosomes. AfSKP1 genes were mainly distributed on Chr2 (8), Chr4 (5), Chr5 (4), Chr6 (6), Chr1, Chr3, and Chr7 each had 3 members of the gene family distributed, and Chr8 had only 2 members of the gene family distributed. AfSKP1 genes were distributed at both the upper and lower ends of chromosomes Chr1, Chr2, Chr3, and Chr4, and mainly at the lower ends of chromosomes Chr5, Chr6, and Chr7. The members of the AfSKP1-15~AfSKP1-17 gene family are more concentrated at the upper end of the Chr4 chromosome, while the members of the AfSKP1-25~AfSKP1-29 gene family are more concentrated at the lower end of the Chr6 chromosome. The formation of this distribution may be related to molecular events such as gene duplication, recombination and chromosomal remodeling.

To clarify the phylogenetic relationship of AfSKP1 genes, phylogenetic trees were constructed for the protein sequences of a total of 103 SKP1 conserved structural domains of A. fisulosum (AfSKP1, 34), A. thaliana (AtSKP1, 20), O. sativa (OsSKP1, 24), and S. lycopersicum (SlSKP1, 25) ( Figure 5 ). According to the protein sequence of AfSKP1 family members, the phylogenetic tree was constructed, and the SKP1 genes of these four species were divided into 6 groups. The SKP1 gene family members in the same group had a similar genetic relationship. There were 42 SKP1 genes in group I, including 11 AfSKP1, 13 AtPIP5K, 14 OsSKP1, and 4 SlSKP1; 20 SKP1 genes in group II, all of which were AfSKP1; and 16 SKP1 genes in group III, including 3 AfSKP1, 7 AtSKP1, 2 OsSKP1, and 4 SlSKP1. Group IV had a total of 10 SKP1 genes, all of which were SlSKP1; group V had a total of 9 SKP1 genes, containing 2 OsSKP1 and 7 SlSKP1; and group VI had a total of 6 SKP1 genes, all of which were OsSKP1. Meanwhile, group I was divided into two subgroups, containing six and five AfSKP1 family members, respectively; genes within the same group usually have similar functional characteristics, and most of the AfSKP1 genes were in groups I and II, thus suggesting that AfSKP1 gene may be involved in the regulation of floral organ development and pollen maturation.

As shown in Figure 6 , AfSKP1 proteins exhibit various motif types. Among them, AfSKP1-2, AfSKP1-5, AfSKP1-9, AfSKP1-12, AfSKP1-14 to AfSKP1-21, AfSKP1-24 to AfSKP1-32, and AfSKP1-34 contain only CDS (translated region), which may have implications for mRNA transport, stability, translational regulation, and incomplete gene annotation (Wu and Bazzini, 2023), other genes have both CDS (translated region) and UTR (untranslated region)( Figure 6C ). Gene structure analysis revealed that the number of introns in the AfSKP1 genome ranged from 1 to 11, with AfSKP1-24 containing only 1 intron, AfSKP1-9 containing 11 introns, and the remaining members of the AfSKP1 gene family containing 2-7 introns. Introns are sequences that impede the linear expression of genes, increasing gene length and recombination frequency, and are advantageous for species evolution and gene expression regulation (Rose, 2018). The results indicated that the AfSKP1-24 gene may have some function, but it did not significantly impact the evolution of A. fistulosum. AfSKP1-9 may be more conducive to the evolution of A. fistulosum. Phylogenetic tree analysis revealed that gene family members in the same group share a relatively similar genetic relationship. For instance, AfSKP1-26, AfSKP1-27, AfSKP1-28, AfSKP1-29, and AfSKP1-31 are clustered together and exhibit a close genetic relationship. Based on conserved motifs, the five AfSKP1 genes in the same group have similar gene structures, indicating analogous functions. Similarly, AfSKP1-2, AfSKP1-4, AfSKP1-7, AfSKP1-8, AfSKP1-11, AfSKP1-14, AfSKP1-22, and AfSKP1-23 are known to perform similar functions. Additionally, 34 AfSKP1 proteins were analyzed for 10 motifs, with 31 proteins sharing motif 1, suggesting its widespread distribution and high conservation in AfSKP1 protein sequences. This indicates that motif 1 is widely distributed and strongly conserved in the AfSKP1 protein sequence, and that this structural domain is essential for the binding of SKP1 to Cullin family proteins, which is a key step in the formation of the SCF complex. (Zhao et al., 2003) Most AfSKP1 members in the same subfamily also share conserved motifs, indicating functional conservation. AfSKP1-34 exclusively contains motif 4 (Wu and Bazzini, 2023), suggesting a unique function, however, its specific role remains to be verified. ( Figure 6A ). Furthermore, specific motifs are present in distinct subfamilies, highlighting functional diversity in AfSKP1 across different subfamilies.

The distribution of structural domains of the AfSKP1 protein is depicted in Figure 6B . The results indicate that all AfSKP1 members contain SKP1 structural domains or SKP1 superfamily members. Specifically, 13 SKP1 members include the BTB-POZ-SKP1 structural domain at the N-terminus. The BTB-POZ structural domain is a characteristic feature of SKP1 structure, characterized by the presence of the BTB/POZ structural domain fold, which includes an α-helix cluster flanking a short β-sheet at the amino terminus. The BTB-POZ transcriptional regulator regulates inflorescence structure in tomato (Xu et al., 2016). This structure facilitates the ubiquitination of proteins involved in cell cycle progression, signaling, and transcription. Therefore, it is plausible that this function is also present in A. fisulosum, but further studies are necessary.

In this study, a collinearity map of AfSKP1 genes was constructed ( Figure 7 ), and the intraspecific collinearity of 34 members of the AfSKP1 gene family was analyzed. The results revealed two homologous relationships among AfSKP1 genes, along with collinearity relationships among their members, specifically AfSKP1-22 and AfSKP1-14, AfSKP1-15 and AfSKP1-16. These relationships were characterized by fragment replication. The evidence presented above indicates that gene replication has occurred in the AfSKP1 gene family, suggesting that the AfSKP1 gene may have expanded the family through replication during evolution.

The cis-acting elements of the AfSKP1 gene family members were analyzed. According to their functions, they can be divided into phytohormone response elements, light response elements, stress response elements, and plant growth and development elements ( Figure 8 ). Among them, the number of cis-acting elements related to hormone response is the largest, totaling 269, including abscisic acid response element (ABRE), methyl jasmonate response element (CGTCA-motif, TGACG-motif), and salicylic acid response element (TCA-element), abscisic acid response element (GARE-motif). Methyl jasmonate can effectively induce flowering and inhibit flower bud differentiation, with the highest number of 91 response elements (Cao et al., 2016). Abscisic acid could induce the defense response of onion plants, showing strong disease resistance, with the second-highest number of 65 response elements. Salicylic acid could induce cell differentiation, with the lowest number of 29 response elements (Hu et al., 2022). This suggests that multiple hormones may be involved in regulating the expression of AfSKP1, jointly promoting A. fisulosum growth and flower development. The total number of light-responsive elements (Box 4, G-box, GT1 motif, and TCT motif) was 268, with Box 4 having the highest number (101), G-box the second highest (76). The AfSKP1-23 gene accounted for the highest number of Box 4 elements, followed by AfSKP1-14. Box 4 is part of the DNA module involved in the light response, suggesting that the main regulatory features of these genes are light-related. Meanwhile, AfSKP1-8 and AfSKP1-23 had the highest number of light-responsive elements, indicating that these genes are closely related to light response. Among the stress response elements, ARE, which determines the binding of regulatory proteins and anaerobically induced expression, had the highest number of response elements, totaling 83. ARE elements were detected in all 31 genes, suggesting that the AfSKP1 gene may be closely related to the anaerobic response of A. fisulosum. The total number of drought-associated MBS elements ranked second with 36, of which 17 promoters were able to detect MBS elements. Low-temperature response (LTR) elements totaled 16, and defense and stress response (TC-rich repeats) elements totaled 12. Additionally, some elements related to plant growth and development response were found, totaling 68, among which the promoter elements related to zeaxanthin metabolism (O2 site) were the most numerous, with a total of 21, suggesting their role in regulating A. fisulosum growth and development. Other cis-acting elements upstream of eukaryotic structural genes (CAT box), endosperm expression (GCN4 motif), and circadian-related cis-acting elements were also identified.

In order to explore the potential biological function of AfSKP1 gene in the growth and development of A. fistulosum, the expression of AfSKP1 gene in different tissues was analyzed by RNA-seq data. As shown in Figure 9 , the expression levels of different genes in flower buds, leaves and flower buds were different, and the expression levels of 28 AfSKP1 genes in different tissues of the two materials were generally low. The expression levels of AfSKP1-4 in the inflorescence meristems of A. galanthum was higher than that in the inflorescence meristems of A. fisulosum var. viviparum Makinot, and the expression levels in the floral buds and leaves of A. galanthum were lower than those in the floral buds and leaves of A. fisulosum var. viviparum Makinot, respectively. AfSKP1-7 was highly expressed in inflorescence meristems and floral buds of the two materials. It can be observed that the expression of AfSKP1-11 in inflorescence meristems and floral buds of the two materials was higher than that in leaves. The expression levels of AfSKP1-14 and AfSKP1-23 in the inflorescence meristems and floral buds of A. galanthum surpassed those in the corresponding tissues of A. fisulosum var. viviparum Makinot, with both species displaying higher expression levels in inflorescence meristems and floral buds compared to leaves. The expression levels of AfSKP1-11 and AfSKP1-14 in the floral organs of both species at various developmental stages may contribute to the failure of anther dehiscence and pollen maturation. Furthermore, AfSKP1-5, AfSKP1-15, AfSKP1-16, AfSKP1-18, AfSKP1-19, AfSKP1-21, AfSKP1-25, AfSKP1-26, AfSKP1-28, AfSKP1-29, AfSKP1-30, AfSKP1-31, AfSKP1-32, and AfSKP1-34 were not expressed in either species. These findings suggest that the AfSKP1 gene exhibits significant tissue-specific expression, indicating potential functional differentiation among the various AfSKP1 members. In light of the previous identification of the differential gene AfSKP1-8 and the expression analysis of AfSKP1-11 and AfSKP1-14, this study will focus on these three genes for further investigation.

In order to better understand the function of the AfSKP1 gene in A.fistulosum, this study predicted the interactions between AfSKP1-8, AfSKP1-11, AfSKP1-14, and other proteins based on the STRING online database. The results showed ( Figure 10 ) that AfisC4G01705 and AfisC4G01201 both interact with these three proteins. AfisC4G01705, annotated as EIF3E, was confirmed to be involved in pollen germination and tube growth processes (Wang et al., 2008). AfSKP1-7 and AfSKP1-11 interacted with the same proteins, a total of 9 proteins, of which AfisC4G02992 was annotated as CEN2 and was confirmed to be involved in the regulation of the pollination process and cell elongation (He et al., 2022). This suggests that AfSKP1 proteins may interact with CEN2 and EIF3E proteins in the regulation of floral organ development and pollination. In addition, there are some proteins that are not annotated. They have clear direct or indirect synergistic effects with AfSKP1 proteins, but their functions are not yet known.

In this study, qRT-PCR was used to detect the transcription levels of the AfSKP1 gene in inflorescence meristems, leaves, scapes, and floral buds of the two materials. The expression patterns of AfSKP1-8, AfSKP1-11, and AfSKP1-14 genes in the two materials were verified. By analyzing the gene expression levels of the three AfSKP1 gene families in different tissues and organs ( Figure 11 ), it was found that the expression levels of these three AfSKP1 genes varied across different tissues. There was no difference in the expression of the AfSKP1-8 gene in inflorescence meristems, scapes, and leaves between the two materials. However, the expression level in floral buds was significantly different between the two materials (p < 0.05), with the expression level in A.galanthum being 1.69 times higher than that in A.fistulosum var. viviparum Makino. The expression level of the AfSKP1-11 gene in floral buds, scapes, and leaves did not differ between the two materials, but the expression level in inflorescence meristems was significantly different (p < 0.05), with the expression level in A.galanthum being 1.76 times higher than that in A.fistulosum var. viviparum Makino. The expression of the AfSKP1-14 gene in inflorescence meristems, scapes, and floral buds was significantly different between the two materials (p < 0.05). The expression level in inflorescence meristems of A.galanthum was 6.05 times higher than that of A.fistulosum var. viviparum Makino. In floral buds, the expression level in A.galanthum was 3.67 times higher than that in A.fistulosum var. viviparum Makino. Additionally, the expression level in scapes of A.galanthum was 4.37 times higher than that in A.fistulosum var. viviparum Makino. The genes that are highly expressed in floral organs may play a significant role in the floral development of the two studied materials. Notably, AfSKP1-11 and AfSKP1-14 exhibited significantly different expression levels in the floral buds of these materials (p < 0.05). This differential expression may contribute to the inability of the anthers of A.fistulosum var. viviparum Makino to undergo dehiscence, as well as the failure of pollen maturation. These findings suggest that the AfSKP1 gene family may be involved in the regulation of floral organ development in A. fistulosum. Furthermore, in conjunction with the results from paraffin section analysis, it is proposed that the genes AfSKP1-8, AfSKP1-11, and AfSKP1-14 may be implicated in the failure of A. fistulosum var. viviparum Makino to produce mature pollen grains, ultimately resulting in pollen abortion.