Plants of H. scandens were cultivated in a garden field at Henan Normal University. The genders were determined by observation of the morphology of flowers. Whole genomic DNA was extracted from three male and three female individuals using the cetyl trimethylammonium bromide method (Doyle and Doyle, 1987).

At least 3 μg genomic DNA of each sample was fragmented, end repaired, phosphorylated, and ligated with adapters. Then 300−400 bp fragments were selected and PCR amplified to construct paired-end library. The libraries were sequenced on Illumina NovaSeq 6000 platform in paired-end, 150-bp mode. The raw reads (DRR024400, DRR024402, DRR024404, DRR024405, DRR024456, and DRR024457) of H. lupulus, a close relative of H. scandens, were downloaded from the NCBI SRA database. For the preprocessing of raw sequence data, the FASTQ files were quality-controlled, filtered using HTQC with default parameters (v1.92.1) (Yang et al., 2013), then a custom perl script was used to filter out the reads with N and transform into FASTA format. Then, for each sample, a randomly chosen dataset comprising 2,000,000 paired-end reads, which represented approximately 0.34× of the genome, was used for further analysis. We clustered, assembled, and annotated all selected reads with the help of RepeatExplorer platform (http://www.repeatexplorer.org, Novák et al., 2020) using the Green Plants (Viridiplantae) database (Neumann et al., 2019). To categorize LTR retrotransposons into different lineages, their RT sequences were analyzed using DANTE within the RepeatExplorer platform. After eliminating duplicated sequences using CD-hit (Li and Godzik, 2006), these RT sequences were aligned with MUSCLE (Edgar, 2004), and phylogenetic trees were generated using FastTree (Price et al., 2010). FigTree software was used to draw and modify the trees.

The TAREAN tool embedded in RepeatExplorer was adopted to detect satellite DNA using the same samples described above (Novák et al., 2017). The high-confidence satellites were identified. The logo for the satellites was drawn by Web-Logo (Crooks et al., 2004).

For LTR retrotranposons, the RT domains of each lineage were amplified using specific primers ( Table S1 ). The gel electrophoresis, cloning, and sequence validation were performed following a previous study (Li et al., 2019). Finally, clones with high sequence similarity to the respective contigs were amplified and labeled with Texas-red-dCTP (PerkinElmer, Waltham, Massachusetts, USA) utilizing the nick translation approach. For satellite DNA, the monomers were identified, and 50 bp of the monomer was randomly chosen and tagged directly with Texas Red-X (Invitrogen, Shanghai, China) in the synthesis process. 45S rDNA was also labeled with Chroma Tide Alexa Fluor 488-5-dUTP (Invitrogen) to aid in chromosome identification.

For mitotic chromosome spread preparation, the branches of both male and female plants were cut and placed in water until the new roots were grown. When the roots were grown to 1–1.5 cm, they were cut and treated in nitrous oxide at 10.9 atm pressure, fixed in a 90% acetic acid solution for 10 min. For meiotic chromosome spread preparation, male inflorescences were fixed in Carnoy’s fixative and stored in 70% ethanol. The immature flowers with a diameter of 1.3–1.5 mm were selected, and the anthers were picked out for further analysis. The enzyme digestion and drop spreading were completed according to earlier instructions (Li et al., 2019).

FISH analysis was carried out as previously described (Li et al., 2019). First, the slides with well-spread chromosomes or desired meiotic stages were UV cross-linked. Then the hybridization mixture (2×SSC, 1×TE, 200 ng labeled probe) was added to the slides, followed by denaturation in boiled water for 5 min. After that, the denatured chromosome slides with probes were incubated overnight at 37°C. Next, the slides were washed in 2×SSC for 5 min, and counterstained with DAPI solution. FISH signals were detected under an Olympus BX 63 fluorescence microscope with an ANDOR CCD.

Pairwise comparisons were used to assess how the data groups differed from one another using Excel software. Least significance difference (LSD) analysis was performed on the data with a significance threshold of 0.05.

Next-generation sequencing produced a total of approximately 20 Gb of raw sequencing data for three male and three female H. scandens. For each individual, a subset of 2,000,000 clean paired reads, representing about 34% of the genome, was chosen for repetitive sequence analysis. We also used sequencing reads of the H. lupulus genome for comparison analysis. Table 1 shows the comparative proportions of different repeat types in the three genomes. The findings revealed a high composition of repetitive sequences in the H. scandens genome, with 68.30% in the female genome and 66.78% in the male genome. These values were significantly higher than those of the H. lupulus genome (59.45%) ( Figure 1A ). Similar to most other plant species, LTR-retrotransposons were the most prevalent group of H. scandens repeats, accounting for nearly 60% of the genome (when male and female values were averaged), followed by tandem repeats (3.06%), DNA transposons (1.16%), and long interspersed nuclear elements (LINEs) (<0.01%). Within the LTR retrotransposon superfamily, the proportion of Ty3/Gypsy elements was more than 30-fold greater than Ty1/Copia elements ( Table 1 ; Figure 1B ). In addition, about 3.35% of organelle DNAs were identified in the H. scandens genome.

For the comparison of different groups of repetitive sequences, the proportions of LTR-retrotransposons and tandem repeats of the H. scandens genome were all significantly higher than those of the H. lupulus genome ( Table 1 ; Figure 1B ). However, the DNA transposons showed a little higher proportions in the H. lupulus genome than in the H. scandens genome. Out of the LTR retrotransposons, the proportions of Ty3/Gypsy elements in the H. scandens genome were significantly higher than those in H. lupulus. By contrast, the H. lupulus genome showed higher proportions of Ty1/Copia elements than the H. scandens genome ( Table 1 ; Figure 1B ). Thus, the ratio of Ty3/Gypsy to Ty1/Copia in the H. lupulus genome was 6:1, compared with 30:1 in the H. scandens genome ( Table 1 ). We also compared the proportions of various lineages of LTR retrotransposons ( Figure 2 ). In the H. scandens genome, Tekay and Retand lineages belonging to the Ty3/Gypsy superfamily were prevalent, together accounting for nearly 80% of the LTR retrotransposable elements. The Athila lineage belonging to the Ty1/Copia superfamily ranked third ( Figures 2A, B ). In the H. lupulus genome, the Tekay and Retand lineages were also the more abundant elements, similar to the H. scandens genome; however, the third-ranked lineage was Angela ( Figure 2C ).

Generally, a conserved repeat proportion pattern was observed between the male and female H. scandens genomes. However, slight but significant differences were still observed in the proportions of several repeat groups between the two genomes with different genders, such as SIRE and Retand lineages, which all showed a higher proportion in female than in male genomes. By contrast, two other lineages, including Athila and Tekay, showed a higher proportion of males than females ( Figure 3 ).

Two phylogenetic trees based on the RT sequences of the Ty1/Copia and Ty3/Gypsy LTR-RTs were created to ascertain their evolutionary relationships. As shown in the evolutionary dendrograms ( Figure 4 ), the Ty1/Copia elements could be classified into three clades: one included elements belonging to Angela and Ikeros lineages, another consisted of TAR and Tork lineages, whereas SIRE, Ale, Alesia, and Ivana lineages grouped together to form the third clade. Among the Ty3/Gypsy elements, there were also three clades: Athila and Retand/Ogre each formed one clade, while the other three lineages, including CRM, Tekay, and Galadriel, formed the third clade. The Retand/Ogre lineage showed a high degree of variability and could be further segregated into two distinct groups.

We examined the chromosomal distribution patterns of each lineage of LTR-RTs by using mitotic-FISH analysis. The findings demonstrated that different lineages has varied patterns of chromosomal distribution. Most of the lineage-based elements were dispersed over all of the chromosomes; these included all eight lineages of the Copia superfamily and four lineages of the Gypsy superfamily (Ogre, Retand, Galadriel, and Tekay) ( Figures 5 , 6 ). The other two lineages (Athila and CRM) mainly occupied the pericentromeric regions of all chromosomes. In addition, the Athila elements showed clearly higher signal intensity in the Y₁ and Y₂ chromosomes than in the X and autosomes.

By using TAREAN, three clusters (CL33, CL37, and CL111) were identified as highly reliable satellite DNAs. Each of them was given the designations Hssat1, Hssat2, and Hssat3, respectively. The three clusters were all characterized by star-like graph topologies ( Figure S1 ). The monomer of Hssat1 was about 312 bp. A search of GenBank yielded no matches to other known sequences. The other two satellites, Hssat2 and Hssat3, demonstrated ~380 bp and ~122 bp, respectively. Like Hssat1, the clusters were all unknown or H. scandens-specific sequences.

The results showed that the signal of Hssat1 is distributed at the centromere of all chromosomes except for Y₁ and Y₂ chromosomes. Hssat2 was distributed specifically at the terminal position on metaphase chromosomes, but there is a large variation between male and female chromosomes. In the female plant, there were 4 pairs of chromosomes with signals at both ends, while in the male plant, there were 5 pairs. Even the chromosomes with signals at one end are inconsistent; for example, there were significant differences in the signals at the ends of chromosomes 6 and 7. In addition, it is worth noting that the signals of the two Y chromosomes were obviously inconsistent, one of which had a signal at one end while the other Y chromosome had no hybridization signal. The signal of Hssat3 showed more complex distribution characteristics, both at the end position and at the centromere position. In female H. scandens, the end positions of all chromosomes except one pair of autosomes could be seen, and obvious signals could be seen at the centromere positions of this pair of autosomes. There was no signal distribution on one Y chromosome of male H. scandens, and the signal of the other Y chromosome was also distributed at the end of one side, like the X chromosome ( Figures 7 , 8 ).

Repetitive sequences represent approximately 68% of the H. scandens genome, and this value is slightly but significantly higher than that of its close relative, the H. lupulus genome. Generally, the amount of repetitive sequences, particularly TEs, is positively correlated with the genome size of plants (Li et al., 2017). H. scandens (1.8 Gb) and H. lupulus (2.5 Gb) possess relatively large genomes, and the TE fraction proportions are generally in line with the trend. However, the genome of H. scandens is smaller than that of H. lupulus, whereas the repetitive sequence fraction of H. scandens is higher than that of H. lupulus. This suggests that TEs were amplified more extensively in the H. scandens genome. Similar to other plant genomes, in the H. scandens genome, LTR-RTs were more abundant than DNA transposons, LINEs, and tandem repeats. The diversity and proliferation ability of LTR-RTs make them an important contributor to the structure, function, and evolution of the H. scandens genome. In particular, two Ty3/Gypsy lineages, Retand and Tekay, proliferated massively. This is similar to the phenomenon in many other plant genomes, in which one or several TE groups are significantly amplified. For example, differential lineage-specific proliferation of distinct families of transposable elements contributes greatly to genome size differences in Gossypium (Hawkins et al., 2006).

According to the general model of sex chromosome evolution, X and Y sex chromosomes are gradually differentiated from a pair of autosomes due to the suppression of recombination around the sex-determining locus. The accumulation of repetitive sequences can facilitate the recombination suppression of sex-determining loci and adjacent regions between X and Y chromosomes, eventually forming the sex-specific region. Furthermore, the formation of sex-specific regions can further recruit more repetitive sequences. Thus, repetitive sequence accumulation was a conspicuous feature of the sex chromosomes in both plants and animals (reviewed in Li et al., 2016). For instance, these findings were observed in papaya (VanBuren and Ming, 2013; VanBuren et al., 2015), Rumex acetosa (Mariotti et al., 2009), Salix viminalis (Almeida et al., 2020), Salix dunnii (He et al., 2021), so on and so forth. The TEs and TE-derived repetitive sequences are thought to be involved in almost all of the major evolutionary phases of sex chromosome evolution (reviewed in Li et al., 2016). Furthermore, TEs and associated repetitive sequences may influence plant sex determination and differentiation. For example, in Populus deltoids, a Ty3/Gypsy transposable element family member within the Y-linked region can generate long non-coding RNAs and act as a male promoter (Xue et al., 2020). In H. scandens, FISH analysis showed that the Athila elements accumulated more intensively in the Y₁ and Y₂ chromosomes. The Y chromosome-biased TEs may be involved in sex chromosome evolution in H. scandens.

However, the satellites did not show such an accumulation pattern. The signals of the three satellites are absent from one or two Y chromosomes. Such a phenomenon was also observed in other dioecious plant species. For instance, one family of the Ogre/Tat lineage is found on all autosomes and the X chromosome but not on the Y chromosome in Silene latifolia (Kubat et al., 2014). Due to this, it can be challenging to comprehend the way repetitive sequences and the evolution of plant sex chromosomes are related. According to the few reports that are currently available, the accumulation or depletion of some repetitive sequences appears to be species-specific, which is in keeping with the fact that sex chromosomes have evolved multiple times independently in different lineages of plants.

The satellite DNA probes showed typical signals located at the centromeric and/or telomeric regions. The X and Y chromosomes showed different signal distribution patterns, so the satellite DNA can be used as cytogenetic markers for identifying the X and Y chromosomes. Sex chromosome identification is crucial for cytological examinations and subsequent studies of sex chromosome evolution. Based on the cytogenetic markers, at the meiotic diakinesis stage, we observed obvious trivalents, showing Y₁-X-Y₂ connection mode. The synaptic region was the telomere position of two adjacent sex chromosomes connected end-to-end Figure 9 . These results suggested that the sex-linked regions of H. scandens are large, which is in accordance with a recent study showing that the MSY covers the majority of the Y chromosomes (Razumova et al., 2023). These findings suggest advanced phases of sex chromosome evolution in H. scandens. In addition, the two Y chromosomes showed large differences because of the fact that the two Y chromosomes only pair at one telomeric end, and the majority of Y chromosomes are chromosome-specific regions. The FISH analysis using satellite probes also supports this perspective. Furthermore, combined with the fact that Hssat1 was specifically distributed on the centromeric regions of all the chromosomes except for the two Y chromosomes, we speculated that the XX-XY₁Y₂ sex chromosomes of H. scandens might have originated from a centric fission event. The centromere-specific satellite DNA might be lost during the centric fission event. However, we still need more evidence to support this speculation.