Eleocharis chromosomes are holocentric and possess a longitudinal centromere groove
To analyze whether the holocentromeres of Eleocharis are characterized by a longitudinal centromere groove as identified in other holocentric plants (e.g. Luzula elegans, L. nivea (Nagaki et al. 2005; Wanner et al. 2015)d pubera (Marques et al. 2015), we employed spatial structured illumination microscopy (3D-SIM) of DAPI-stained chromosomes. All E. parodii, E. montana and E. maculosa mitotic metaphase chromosomes revealed such a centromeric groove, and the application of pericentromere-specific antibodies confirmed holocentricity (Figs. 1a and m, and Supplementary Fig. 1). The immunostaining against H2AT121ph histone, which is pericentromere-specific (Demidov et al. 2014; Dong and Han 2012) showed line-like signals along the groove in all chromosomes of E. parodii and E. montana (Figs. 1b-g and Supplementary Fig. 2). The KNL1 antibody used to detect kinetochores, and the antibody against H3 histone phosphorylated at serine 10 used to detect pericentromeres (Houben et al. 2007; Gernand et al. 2003), displayed line-like signals along the chromatids of E. geniculata and E. maculosa, which is typical of holocentromeres (Fig. 1).
Eleocharis species are characterized by differentially accumulated repetitive families
To characterize the repeatome of Eleocharis species, the high-copy DNA fractions were characterized for five species from two clades with diverse karyotypes. The first group comprised E. parodii (2n = 10 / 1C = 0.78 pg), E. elegans (2n = 20 / 1C = 1.06 pg) and E. montana (2n = 40 / 1C = 2.35 pg), and the second clade with E. maculosa (2n = 10 / 1C = 0.50 pg) and E. geniculata (2n = 10 / 1C = 0.50 pg). The general comparative analysis of the repeatome showed that the genomes of E. parodii, E. elegans and E. montana accumulated more TEs (Classes I and II) than those of E. maculosa and E. geniculata. In contrast, satDNA repeats were found in higher abundance in the two later species. Eleocharis maculosa and E. geniculata showed 3.52% and 1.69% of their genomes occupied by satDNA, while E. parodii, E. elegans and E. montana, had 0.14%, 0.54% and 0.41%, respectively (Supplementary Fig. 3).
The six most abundant satellite DNA were characterized in more detail (Fig. 2 and Supplementary Table 5). The main satDNA found in the genome of E. maculosa, named EmaSAT14, presented 112-bp long monomers. The remaining satellites were EgeSAT112 with 427 bp from E. geniculata, EleSAT99 with 214 bp in E. elegans, EpaSAT248 with 179 bp from E. parodii, and EmoSAT234 with 91 bp and EmoSAT273 with 183 bp from E. montana. Comparing the sequences of the monomers (all against all) using the Dotter b/l tool revealed that satellites EleSAT99, EmoSAT273, EmoSAT234 and EpaSAT248 showed no similarity to other monomers, while EmaSAT14 share some regions of homology with the larger EgeSAT112 monomer (Figs. 2, 3a and Supplementary Fig. 4). By analyzing 250 conserved monomers of EmaSAT14 obtained from E. maculosa, we discovered a conserved region (> 82% identity) with 40 bp (Fig. 3a and Supplementary Fig. 5), which we refer to as Oligo/EmaSAT14 and we used to produce biotin-oligo labeled probe.
To investigate the distribution of EmaSAT14 and EgeSAT112 along the chromosomes, we searched for monomers of these two satellites in E. geniculata scaffolds that were obtained through PacBio HiFi sequencing (see Methods). EgeSAT112 has two insertions of EmaSAT14 monomer fragments, separated by about 100 bases (Fig. 3a). The alignment of these monomers against scaffolds 12 and 13 revealed that EgeSAT112 is organized in large arrays, while EmaSAT14 appears irregularly spaced along E. geniculata scaffolds, including small arrays within the EgeSAT112 blocks (Figs. 3b-c). The FISH assay using a probe of a small portion of EmaSAT14 (the Oligo/EmaSAT14 with 40 nt, which represents roughly 10% of the EgeSAT112 sequence), and also a probe for the complete sequence of EgeSAT112, showed two distal signals for EgeSAT112 (Fig. 3d-e). These signals were displayed in blocks, similar to the FISH signals from the 35S and 5S rDNA probes used as FISH controls. Remarkably, the EmaSAT14 probe showed multiple and small FISH signals dispersely distributed in interphase nuclei (Fig. 3f). In sharp contrast, EmaSAT14 occurred as line-like signals in metaphase chromosomes (Fig. 3g), suggesting that this repeat is a strong holocentromeric-specific sequence and that E. geniculata may possess repeat-based holocentromeres.
The conserved portion of EmaSAT14 was tracked in 22 genomes using a consensus sequence to determine its conservation across the Eleocharis phylogenetic tree. RepeatMasker showed traces of EmaSAT14 monomers (> 70% identity) in contigs from 11 genomes of Eleocharis. In nine of these, no more than 10 contigs presented single copies (Supplementary Table 6). EmaSAT14 was absent from the other 11 genomes. In light of the group's phylogenetic relationships, EmaSAT14 monomers have been found in all clades of subgenus Eleocharis, and in one species each of the subgenuera Zinserlingia (E. quinqueflora) and Scirpidium (E. acicularis). EmaSAT14 was not detected in the species of subgenus Limnochloa (Supplementary Fig. 6). Only E. geniculata and E. maculosa (Subgenus Eleocharis, Section Eleogenus, Series Maculosae) accumulated EmaSAT14 sequences in comparison to the other nine genomes (Supplementary Table 6).
Given the scattered pattern of EmaSAT14 across the chromosomes of E. geniculata and E. maculosa (Supplementary Fig. 7), we searched for any association between this satDNA and TEs that could explain its dispersion. Results of the search of EmaSAT14 monomers against E.D.T.A. output files for retrotransposons, transposons and helitrons (> 70% identity) showed 16 non-autonomous TEs invaded by EmaSAT14 sequences. To illustrate this association, we annotated a transposon from E. geniculata where EmaSAT14 appeared as a very small array inside a fragmented transposase (Supplementary Fig. 7). When searching the E. maculosa genome, insertions of EmaSAT14 in transposons of the type hAT and retrotransposons of the LINE and LTR-RT groups were also detected. EmaSAT14 arrays were commonly found on the flanks of these transposition elements, as shown in Supplementary Fig. 7. Based on these data, it can be inferred that EmaSAT14 invaded various TEs in these two genomes in different ways.
EmaSAT14 is unevenly distributed along the chromosomes
To better understand the distribution of the EmaSAT14 satellite along Eleocharis holocentric chromosomes, we mapped EmaSAT14 monomers in four randomly selected scaffolds from E. geniculata. The occurrence of multiple arrays interspersed along the chromosomes was revealed by this approach. They appeared in clusters with a small number of copies to large arrays distributed irregularly throughout the scaffolds (Fig. 4). The second analysis consisted of screening EmaSAT14 monomers in the five largest scaffolds (16 to 25 Mb in length) generated by PacBio sequencing in E. geniculata. EmaSAT14 showed an irregular accumulation in these scaffolds. When distances between blocks using EmaSAT14 were estimated, it was observed that the average distance between the arrays was very variable and had a very high standard deviation. This happened both for sequences with > 70% similarity and for those with > 80% similarity (Supplementary Table 7 and Supplementary Figs. 8, 9 and 10).
FISH using EmaSAT14 probe in species of different clades showed that only three species from Section Eleogenus, Series Maculosae (E. maculosa, E. geniculata and E. filiculmis) showed cross hybridization. When this probe was used in E. maculosa with 2n = 6 and 10, it produced stronger line-like FISH signals in some chromosomal regions, while very weak signals were produced in other regions (Figs. 5a-e and Supplementary Fig. 11). On the other hand, the oligomer FISH probe with similarity to the conserved stretch of EmaSAT14 (see also the Figs. 3 and 4) produced FISH signals of various sizes in the interphase nucleus. This probe was more effective to show the irregularly distributed FISH signals throughout the chromatids in prometaphases and metaphases of E. maculosa, E. geniculata and E. filiculmis (Figs. 5d, g, h and i). Remarkably, in all these three species line-like signals were observed, reinforcing the putative association of EmaSAT14 with the holocentromeres in different Eleocharis species. Thus, our results suggest a possible common repeat-based holocentromere organization for several Eleocharis species.
Other satDNAs
Based on the dotplot and alignments, we observed that EleSAT99 monomers have a small coincidence, organized as three subrepeats (see Fig. 2). As the identity between them is < 30%, we consider it one monomer with internal subrepeats. EleSAT99 is an array of two sets of simple sequence repeats (SSR) rich in ATG and ATC trimers (Supplementary Fig. 12). This satellite was found in all high and low coverage genomes we evaluated, but it was only found in small arrays, except for E. elegans, where it formed a block observed after FISH (Fig. 6). Four scaffolds obtained from E. geniculata genome assembly were used to perform a comparative alignment using EleSAT99, EmaSAT14, and LTR retrotransposons. Data shows that all groups of elements are irregularly distributed along scaffolds, with peaks of accumulation at various distances (Supplementary Fig. 13). EleSAT99 appears to be present in different regions of the genome, including within TEs. To exemplify, we have annotated a non-autonomous Helitron with 14,544 nucleotides, which contains five EleSAT99 insertion sites (Supplementary Fig. 12e).
FISH experiments carried out using the other satDNA probes also revealed hybridization signals in distal and interstitial positions, in the form of small blocks or dots (Fig. 6). Intraspecific diversity and polymorphisms were observed in two diploid samples of E. geniculata (2n = 10), as well as in the polyploid E. montana (2n = 40). In E. geniculata, for example, in addition to a very small chromosome without FISH signals, a large chromosome exhibited 35S rDNA signals in both distal regions (Fig. 6c). Interstitial FISH sites were clearly identified by the EgeSAT112 probe in this same species (Fig. 6c). The EpaSAT248 probe obtained from the E. parodii produced two distal FISH signals in this species, adjacent to 35S rDNA, not to the 5S, on a pair of chromosomes (Fig. 6a and b). FISH signals were also observed for two other phylogenetically very close species (E. elegans and E. montana). In the first, this probe EpaSAT248 showed distal-interstitial signals in two chromosomes (Fig. 6d). In E. montana, six distal signals were detected by probe EpaSAT248 (Fig. 6i). The EmoSAT273 probe derived from E. montana resulted in small interstitial FISH signals in E. elegans (Fig. 6e). Despite several rounds of FISH, we were unable to reproduce these signals on E. parodii chromosomes.
Two populations of E. montana exhibited differences in their number of FISH signals detected by probe EmoSAT234. Six distal signals in one sample and two distal-interstitial signals in the other were observed (Figs. 6j-k). Probe EmoSAT273 also showed some polymorphisms. Four chromosomes of the first population exhibited distal signals, whereas in the other, there were two stronger distal signals in one pair. In addition, the second population showed another pair with smaller FISH signals in both distal regions (Fig. 6j-k). We also compared all these selected satDNA sequences with 5S and 35S rDNA sequences obtained from the E. maculosa genome. The results of Dotter tool and double-FISH confirmed that these satDNA sequences are not related to any repetitive fraction originating from rDNA (Fig. 6 and Supplementary Fig. 14).