Genome size, chromosome number determination, and analysis of the repetitive elements in Cissus quadrangularis

Plant materials

Stem cuttings of C. quadrangularis were collected from the roadside in Namango, Kenya, (S02°32′, E36°49′). Duplicate voucher specimens (SAJIT 002306) were deposited in the Wuhan Botanical Garden herbarium (HIB) and the Herbarium of Jomo Kenyatta University of Agriculture and Technology (JKUAT). Young leaves and petioles were collected for genome size evaluation while root tips were collected for chromosome number determination. All materials for the genome sequencing and karyotyping of C.quadrangularis were obtained from a single individual.

Genome size estimation

Flow cytometry was used to determine nuclear DNA content with minor modifications using hand chopped material as originally described by Galbraith et al. (1983). In order to isolate the nuclei, a woody plant buffer (WPB) was used, which contained 0.2 M Tris-HCl, 4 mM MgCl₂⋅6H₂O, 2 mM EDTA Na₂⋅2H₂O, 86 mM NaCl, 10 mM sodium metabisulfite, 1% PVP-10, 1% (v/v) Triton X-100, with a pH 7.5 (Loureiro et al., 2007a; Loureiro et al., 2007b). Raphanus sativus cv. Saxa (radish) seeds provided by the Institute of Experimental Botany, Czech Republic, were germinated and the plantlets were used as reference standards. The plantlets of C. quadrangularis were pre-treated for 4 days in the dark. Petioles from the young leaves of C. quadrangularis and young leaves from radishes were collected (approximately 50 mg for each sample) and hand chopped using a sharp razor on ice in 1.5 mL WPB as described by Pfosser et al. (1995). The suspension was filtered through a 40-µm-nylon mesh (Cat. 352340, Falcon, USA) to eliminate the excess debris. RNase A was added to 100 ng/mL in the nuclei homogenate solution. Propidium Iodide (50 mg/mL) was used to stain the nuclei for at least 2 min on ice before the samples were run in the flow cytometer (BD Accuri C6). Three independent samples were run on the cytometer and the genome size was calculated using the following formula: sample 2C DNA content = [(sample G1 peak mean)/(internal standard peak mean)] × Internal standard DNA content.

Chromosome count

Root tips were obtained from C. quadrangularis plantlets propagated from the same individual and collected from the field. Samples were treated with a saturated solution of 1-Bromonaphthalene for 3 h at room temperature to halt cell division (Mirzaghaderi, 2010). Microscopic slides were then prepared from the treated root tips using the protocol as developed by Kirov et al. (2014) with minor modifications to obtain the chromosomes at metaphase stage. To digest the cell wall, the root tips were incubated for 60 min at 37 °C in a 1% enzyme mix (1% pectinase and 1% cellulase in freshly prepared 0.1 M citrate buffer). Relative humidity was maintained between 40 and 60% during the dropping step. The prepared microscopic slides were stained with 4′, 6-diamidino-2-phenylindole (DAPI) and chromosome fluorescent images were captured using a fluorescence microscope (Leica DMi8) fitted with a camera (Leica DFC 550).

DNA isolation, sequencing, and data analysis

Genomic DNA was extracted from the leaves of C. quadrangularis using the plant genomic DNA kit (Tiangen, China) following the manufacturer’s protocol. DNA libraries were constructed the using NEBNext® Ultra™ DNA Library Prep Kit for Illumina. Paired-end sequences (2 × 150 bp and 400–450 bp insert size) were generated on the Illumina HiSeq X Ten platform produced by Novogene (Tianjing, China). The sequencing data were deposited in the SRA database under accession number SRR8573652.

To analyze the repetitive components of this species, graph-based clustering was performed using the RepeatExplorer platform (Novák, Neumann & Macas, 2010). Short-read sequencing data was subjected to the Galaxy-based RepeatExplorer platform as described by Novak et al. (2013). The quality of the reads was determined by the FastQC tool and poor-quality reads were discarded. Clean reads were converted into the FASTA format Using the FASTQ to FASTA converter. It was possible to infer the repeat composition for a species with typically 0.1–0.5× genome coverage using the RepeatExplorer (Macas et al., 2015). A total of 700,000 paired-end reads (150 bp) were randomly selected, which were equal to approximately 6.56% of the predicted genome for clustering. An all-to-all comparison was carried out, grouping similar sequences into respective clusters in RepeatExplorer. The genome proportions for each cluster were calculated based on the read percentages.

Repeat clusters contributing no less than 0.01% of the genome proportions were considered for further annotation while those with smaller contributions were ignored. Repeat clusters with known protein domains were annotated directly on the RepeatExplorer platform, while similarity searches against GenBank databases (nt and nr) using BLASTn and BLASTx (Altschul et al., 1990) were carried out manually with the E-value at 1e⁻⁵ to classify other repeat clusters.

Comparison of repeat content in C. quadrangularis, grape, and 4 other Cissus species

To investigate variations in repeat elements among the species and their roles during the evolution of the Cissus genus, we collected short-read sequencing data for other Cissus species, which is publicly available at NCBI. Four Cissus species whose genome size and sequencing data are available (C. tuberosa, C. trifoliata, C. discolor, and C. microcarpa), were selected for co-clustering (last accessed May 20, 2019). Sequencing and genome size data are essential in determining the proper number of reads representing similar genome proportions. Co-clustering analysis was performed using RepeatExplorer (Novak et al., 2013). Reads from C. quadrangularis, four other Cissus species and grape were simultaneously clustered. The species were randomly grouped and considered to have equal probabilities for common repeats and frequencies among the 6 species. Co-clustering allowed the grouping of similar reads from individual species suggesting similar ancestral origin. In order to avoid a sensitivity bias, the number of reads that were analyzed for all species are proportional to the genome sizes of the corresponding species. Randomly selected paired-end reads from C. quadrangularis, grape, Cissus microcarpa, Cissus trifoliata, Cissus discolor, and Cissus tuberosa, (ca. 0.06× coverage for each genome) were combined for RepeatExplorer co-clustering (Novak et al., 2013). The sources for clustering data are provided in Table 2. The genome size data used for annotation of repeats for Cissus species has been reported by Chu et al. (2018). Plastid repeats are phylogenetically uninformative due to their abundance, which is linked to the photosynthetic dynamics in plant tissues (Dodsworth et al., 2016) and were manually excluded in further analysis.

C-value determination in Cissus quadrangularis

The nuclear DNA content of C. quadrangularis was evaluated by flow cytometry using radish (2C = 1.11 pg) as the reference standard. The genome size for the C. quadrangularis individual considered in this study was estimated as 2C = 1.410 pg. (Fig. 1) representing 689 Mbp/1C (1 pg. = 978 Mbps) (Dolezel et al., 2003). According to the classification by Soltis et al. (2003) the C. quadrangularis genome falls within the group of plants with very small genome. The size of the Cissus quadrangularis genome is approximately in the same range as Cissus antarctica and Cissus trifoliata which are tetraploid species (Chu et al., 2018). In addition, its genome is roughly double that of Cissus javana and Cissus rotundifolia which have diploid genomes (Chu et al., 2018).

The mucilage of the chopped plant tissues had a viscous texture and is composed of complex polysaccharides of various concentrations, which include galacturonic acids, rhamnose, galactose and others (Ovodov, 1998). Their sticky nature causes the aggregation of the nuclei making it difficult to isolate them for cytometric analysis and in majority of species, leaves are commonly used for flow cytometry analysis. However, in our experiment, young leaves yielded unsatisfying results, which may have been caused by high amounts of polysaccharides. Following the suggestion by Suda (2004), petioles were used yielding acceptable peaks. Nuclei isolation buffers, including the Tris⋅MgCl₂ buffer and Galbraiths’ buffer (Dolezel, Greilhuber & Suda, 2007), yielded unsatisfactory results (data not shown). Loureiro et al. (2006) modified the constituents of the Tris⋅MgCl₂ buffer to develop WPB, which counters the negative effects of tannic acid. The inclusion of sodium metabisulfite and PVP-10 in the buffer enhanced its efficacy by reducing the impact of phenol and other secondary metabolites. Dolezel, Greilhuber & Suda (2007), Loureiro et al. (2007a) and Loureiro et al. (2007b) used higher concentrations of a detergent (Triton-X), which had the effect of reducing the mucilage viscosity levels and minimizing their negative impacts. WPB has been employed in genome size determination for other members in genus the Cissus genus (Chu et al., 2018) and other stubborn woody plants (Loureiro et al., 2007a; Loureiro et al., 2007b). Dark-treatment of the samples improved our results and is a method that can be applied in genome size determination for species with higher levels of polysaccharides and other secondary metabolites.

Chromosome counts

The slides prepared with root tips were stained with a fluorescent dye, DAPI, to visualize the chromosomes. C. quadrangularis chromosomes are tiny and their numbers were quantified as 2n = 48 (Fig. 2). Additional chromosomal images have been supplied in Data S1. Observations revealed small chromosomes, making it difficult to identify the centromeres except for two pairs. To confirm the chromosome numbers, fluorescent in situ hybridization (FISH) using telomere labeled DNA probes was conducted. We used nick translation method to label DNA telomere probes (Ma et al., 2010). The sensitivity of FISH is to detect 3.5 kb target sequence on chromosome (Ma et al., 2010). Additionally, we used barley as the control for this experiment. We could not obtain clear telomere signals on all chromosome ends simultaneously in Cissus quadrangularis, while the telomere signals were clear in all barley chromosomes (data not shown). During FISH procedure, the DNA on the chromosomes could be lost/degraded if the fixation step is not optimized (Schwarzacher & Heslop-Harrison, 2000). In addition, high efficiencies FISH was achieved from mitotic metaphase chromosomes prepared from floral tissues of Arabidopsis thaliana, while almost no signals were detected on the chromosomes of root meristematic tissues with the same clones as probes. This is possibly due to the differences in chromatin structure between root meristematic tissues and floral tissues (Murata & Motoyoshi, 1995). In future, optimization of FISH protocol and use of different tissues to prepare mitotic metaphase chromosomes is therefore essential.

The basic chromosome number for members in the Cissus genus can be inferred to be 10, 11, or 12 (Chu et al., 2018). Cissus quadrangularis chromosomes have previously been reported (2n = 24, 28) for diploids and 2n = 45 for a specimen from India which was thought to be a tetraploid (Raghavan, 1957). Variations in chromosome numbers observed in individuals from different localities could be due to epigenetic factors such as DNA methylation and histone modification, which are thought to contribute to the silencing of transposable elements triggering chromosomal evolution in different biogeographical regions (Li et al., 2017). The limited sampling size prohibits a conclusive determination of the chromosome numbers for C. quadrangularis. In future studies, the sample size could be expanded to include more samples from different biogeographic locations.

Based on the deduced chromosome duplication model in the Cissus genus, previous reports for C. quadrangularis and other members in the genus (Karkamkar, Patil & Misra, 2010; http://www.tropicos.org/Project/IPCN; Chu et al., 2018), C. quadrangularis individual considered in our study, with 2n = 48 can be said to be a tetraploid (Table 1). However, more information, which may include molecular data, will be required to confirm this assumption. This would include carrying out both meiotic and mitotic studies to test the possibility of having accessory chromosomes, the existence of aneusomatic division, or the presence of accessory chromosomes in roots (Gibbs & Semir, 1988). Obtaining such information will improve our understanding of chromosomal evolution in the Cissus genus. This will further assist in the interpretation of phylogenetic and biogeographic models such as Cladistic, Migrationist and Panbiogeography models (Wilson, 1991), rather than just relying on potentially out-of-date information. Differences in diploid somatic chromosomes may be indicative of intraspecific polyploidization (Husband, Baldwin & Suda, 2013). Additionally, the occurrence of species complexes with variable karyotypes have been reported in other species and it has been suggested that they are indicative of a series of dysploidy in diploid species followed by hybridization and polyploidization (Choi et al., 2008; Haga & Noda, 1976). Such species complexes could suggest the presence of cryptic lineages in this species and therefore more tests should be carried out to include morphological traits examinations and molecular phylogenetics.

The genome composition of C. quadrangularis

The complexity and tediousness of traditional molecular techniques have made it difficult to analyze genome components, especially in non-model organisms. In this study, short-read sequencing in combination with bioinformatics tools were used to analyze the C. quadrangularis genome. Genome analysis is efficient and economical using these techniques, despite the plant not being well-studied.

50,767 clusters were obtained from the RepeatExplorer results indicating that about 52% of the C. quadrangularis genome is composed of repetitive elements (Fig. 3). The output of clustering analysis forms a foundation for comprehensive study in the future to determine the repeat family structures and their variations. The raw RepeatExplorer output data has been provided in Data S2. The repeat composition was within the range of the estimates for plants with small genome sizes, which range between 25.04–66.42% in the Oryza genus (Zuccolo et al., 2007), 41.4% in grapes (Jaillon et al., 2007) and 61% in sorghum (Paterson et al., 2009).

The top 324 clusters constituting no less than 0.01% of the genome proportions were considered for further annotation. The relatively higher proportions of the single or low-copy repeat families were not considered. However, the variation of these elements may impact the phenotypic characteristics of plants (Barghini et al., 2015). To analyze the variations of the single copy retroelements, a reference genome in combination with genotype re-sequencing is needed, as in many model species. The singletons in this study represent the low-copy fraction of the genome that could not be assembled into clusters using RepeatExplorer (Fig. 3).

Characterization of LTR-retrotransposons of Cissus quadrangularis

Repeat cluster annotation and characterization were performed by a sequence similarities search against the NCBI database (Altschul et al., 1990) and on the RepeatExplorer platform for clusters with known protein domains. LTR-retrotransposons formed the major part of the identified repeats with two subfamilies: Gypsy and Copia. The Copia family comprised 24% of the LTR components while Gypsy made up of 15% of the LTR components (Fig. 4A). According to Wicker et al. (2007), the two main superfamilies, Gypsy and Copia, can further be classified into 6 and 7 lineages, respectively. The majority of the Copia lineages are of Angela type (Fig. 4B) (28% of the Copia elements) while the majority of the Gypsy lineages are of the chromovirus type (Fig. 4C) (29% of the Gypsy elements). A higher percentage of the analyzed clusters had unidentified components that represented clusters whose components could not be annotated (Fig. 4A). This is attributed to absence of protein domains in clusters and/or the limited repeat sequences of closely related species in public databases. Considerably higher numbers of unidentified clusters have been observed in the sunflower and Eragrostis tef (Gebre et al., 2016; Mascagni et al., 2015), while in sea grass (Posidonia oceanica) (Barghini et al., 2015), the proportions of unidentified clusters were considerably lower. Small clusters that were not analyzed (<0.01% genome proportion) also represented a significant proportion of the genome that was not well-characterized. Other components that were identified included LINEs, DNA transposons, organellar DNA, satellite repeats, and rDNA among others (Fig. 4A).

The possibility of chloroplast DNA insertions into the nuclear genome has been infrequently reported (Kejnovsky et al., 2006). In maize, the presence of chloroplast DNA in the nuclear genome has been reported (Roark et al., 2010), which may explain the presence of some levels of organellar DNA in our sample, while the majority could have been as a result of contamination during DNA extraction. The presence of rDNA can be due to biases generated during sequencing (Macas et al., 2011). Low efficiency during sequencing has been identified in GC-enriched segments when using Illumina technology (Nakamura et al., 2011; Aird et al., 2011) and leads to vulnerability of GC-rich templates to biases during Illumina sequencing. It is likely that the quality of the sequencing sample caused the bias in the representation of rDNA.

In this study, only 6 of the Copia subfamilies were found, which may be due to the limited consideration of clusters whose genome compositions were no less than 0.01% each were considered and therefore a large number of elements with lower representations were not annotated. In addition, part of the plant genome is not yet annotated into the main subfamilies as inferred from the analysis by RepeatExplorer. Therefore, more extensive sequencing techniques and coverage are required to decisively annotate the whole genome. We carried out repeat composition comparison analysis for C. quadrangularis and other well-studied species. From our comparison, a linear relationship between the genome size and LTR contents was observed but with exceptions (Table 3). For example, potato (Solanum tuberosum) has a genome size smaller than C. quadrangularis but a higher LTR composition. Our observations are in agreement with Wang et al. (2014) and Li et al. (2017) who identified a positive correlation between genome size and quantity of repetitive sequences in plants.

Variation in genome sizes for angiosperms have been attributed to differences in transposable element content, especially the LTR (Feng et al., 2017). This has been demonstrated in Spirodela polyrhiza (158 Mbp) with chromosome numbers 2n = 40 and Lemma minor which has the same chromosome number but a 481 Mbp genome size (Feng et al., 2017). Genomic repeats serve as vital indices in phylogenetic studies (Dodsworth et al., 2015). Therefore, changes in the LTR composition and its impact on genome evolution could be applied in evolutionary studies for C. quadrangularis and its relatives as new technologies are developed.

Comparing the repeat contents for Cissus quadrangularis, grape, and 4 other Cissus species

The alignment of homologous DNA sequences has been the basis for molecular systematics. The differences in the sequence alignment patterns are used for the construction of phylogenetic trees. Insufficiency in divergences for homologous protein domains for repetitive elements between taxa makes repetitive elements unsuitable for phylogenetic analysis. However, variations in the number of repeat types and specific retrotransposons can be used quantitatively for phylogeny reconstruction (Dodsworth et al., 2015). Distantly related species exhibit divergences in the structure of their repetitive elements, whereas there is a level of uniformity noted in closely related species (Dodsworth et al., 2015).

Based on our comparison, 45S rDNA was the major element present in all 6 species (Fig. 5A) which agreed with the high homology of 45S rDNA in angiosperm (Roa & Guerra, 2012). The application of repeat elements for phylogenetic inferences is defined up to the genus level due to a limited number of common repeats beyond genus level (Dodsworth et al., 2016). Therefore, phylogenetic inferences were restrained using repeats from the Cissus genus and grape. Other major elements represented in all of the six species studies include 5S rDNA, Ty_copia, Ty3_gypsy, and some elements whose protein domains could not be identified in the annotated clusters and which are identified as unknown in Figs. 5A & 5B. It was further noted that the most abundant repeats were shared by the 5 Cissus species (Fig. 5B). These elements include Ty3_gypsy, Ty1_copia, 45S rDNA, hAt:hAt, pararetrovirus:PARA and elements whose protein domains could not be identified (Fig. 5B). The noticeable conservation of the Copia elements may explain their homology for the species considered. Gypsy elements were shared among some species but the degree was lower compared to Copia elements. This indicates the less conservative nature of these elements as observed by Barghini et al. (2015). Other elements such as satellites and LINE: LINE were shared by some Cissus species, which may indicate that they are newly formed. These elements display a lower uniformity in the considered species and reflect variations in proliferation rates among different and related species (Hawkins et al., 2006).

Comparative analysis conducted on Musaceae family indicated quantitative differences in the repeat elements and the classified elements at different taxonomic levels (Novak et al., 2014). In our comparison, as a result of limited homology for majority of the repeat elements in Cissus, it was not possible to infer phylogenetic relationships. Dodsworth et al. (2015) noted a similar problem in constructing bifurcating phylogenetic trees while evaluating relationships in legume tribe Fabeae with homoploid and polyploid hybridization. In addition, comparative analysis of repetitive elements involving several species have unveiled variation in the sequences for probed repeat families and their abundances (Kelly et al., 2015; Macas et al., 2015). Novak et al. (2014) deduced that these differences may be due to the incomplete assembly and the unclear constituents that are encountered in the assembly of the genome. Considering the huge number of species in the Cissus genus, our sample size for co-clustering could contribute to the limited phylogenetic information in this genus. However, the information from this study is a foundation for future work in this genus.