Aeluropus littoralis seeds were collected from the Isfahan province in Iran and plants were cultivated at IPK Gatersleben (Germany) and Sari Agricultural Sciences and Natural Resources University (Iran). A specimen of the analyzed plants was deposited at the herbarium GAT under voucher number 70486. Sterilized seeds were plated on full strength MS medium (Murashige and Skoog, 1962) with vitamins, 3% sucrose and 0.7% agar (pH 5.8). The cultures were incubated in a germinator at 25 ± 2°C with 16 h light/8 h dark photoperiod at 100 μmol m^–2 s^–1 photon flux density using cool-white fluorescent light. Two weeks after germination, the seedlings were transferred to hydroponic culture containing Hoagland’s solution (Arnon and Hoagland, 1939). Hoagland’s nutrient solution comprised 3 mM KNO₃, 2 mM Ca(NO₃), 2.1 mM NH₄H₂PO₄, 0.5 mM MgSO₄, 1 μM KCl, 25 μM H₃BO₃, 2 μM MnSO₄, 2 μM ZnSO₄, 0.1 μM CuSO₄, 0.1 μM (NH₄)₆Mo₇O₂₄ and 20 μM Fe(Na) EDTA, in demineralized H₂O buffered with 2 mM 2-(N-morpholino) ethanesulphonic acid, pH 5.5, adjusted using KOH. Transferred plants were grown in a phytochamber at approximately 240 μmol m^–2s^–1 under longday photoperiodic conditions (16 h light, 22°C/8 h dark, 18°C) at 70% relative humidity.

For salt stress treatments soil grown plants (pots 12 cm diameter) were watered with 1 m NaCl once per week.

Aeluropus littoralis leaves of three biological replicates of plants grown under controlled conditions or exposed to salt stress were used for comparative histological and ultrastructural analyses. Cuttings of a size of 1 mm × 2 mm from the central part of fully developed leaves were used for combined conventional and microwave assisted chemical fixation, substitution and resin embedding as defined in the given protocol (Supplementary Table 1). Sectioning, histological staining, light and transmission electron microscopy analysis was performed as described (Daghma et al., 2011).

Mitotic chromosomes were prepared from root tips, which were pretreated in ice water for 24 h to accumulate synchronized cells at metaphase, fixed in Carnoy’s fixative [ethanol and glacial acetic acid, 3:1 (v/v)] at room temperature for 20 h, and kept in 70% ethanol at 20°C for later use. Fixed roots were digested in an enzyme mixture (2% cellulose, 2% pectinase, 2% pectolyase in citrate buffer containing 0.01 M sodium citrate dihydrate and 0.01 M citric acid) at 37°C for 30–40 min. Cell suspension from root meristems in Carnoy’s fixative was dropped onto slides on a hot plate at 50°C, slides were further fixed in the fixative for 1 min, air-dried, and kept at 4°C.

The clones pTa71 and pAt T4 were labeled with fluorophore ATTO550 and ATTO488, respectively, using nick translation labeling kit (Jena Bioscience) to detect 45S rDNA loci and Arabidopsis-type telomeres. Fluorescence in situ hybridization was performed as described in Aliyeva-Schnorr et al. (2015) with pretreatment for 10 min in 45% acetic acid at room temperature, followed by 0.1% pepsine in 0.01N HCl at 37°C. Slides were supplied with the hybridization mix (50% (v/v) formamide, 10% (w/v) dextran sulfate, 2 × SSC, and 3 ng/μl of telomere probe) and denatured at 75°C for 2 min. After stringency wash in 2 × SSC at 57°C for 20 min, chromosomes were counterstained with 4′,6-diamidino-2-phenylindoline (DAPI). Images were captured using an epifluorescence microscope BX61 (Olympus) equipped with a cooled CCD camera (Orca ER, Hamamatsu) and pseudocolored with Adobe Photoshop.

For estimation of nuclear genome size by flow cytometry, approximately 10 mm² of leaf tissue from individuals of Aeluropus littoralis plants was chopped with a sharp razor blade together with roughly 5 mm² of leaf material of Raphanus sativus cv. “Voran” (Genebank Gatersleben accession number: RA 34; 2C = 1.11 pg) as internal reference standard (Schmidt-Lebuhn et al., 2008) in a Petri dish containing 1 ml Galbraith nuclei isolation buffer (Galbraith et al., 1983) supplemented with 1% PVP- 25, 0.1% Triton X-100, DNase-free RNase (50 μg/ml). The nuclei suspension was filtered through a 35-μm mesh cell strainer cap to remove large fragments and stored on ice until measurement. The relative fluorescence intensities of 7,000–10,000 events (nuclei) per sample were measured using a CyFLow Space flow cytometer (Sysmex-Partec, Germany) quipped with a 30 mW green solid state laser (532 nm). The absolute DNA amounts of samples were calculated based on the values of the G1 peak means.

DNA of A. littoralis was extracted according to Dellaporta procedures (Dellaporta et al., 1983). The quality and quantity of the extracted DNA were controlled by measuring absorbance at 260/280 nm using a NanoDrop spectrophotometer (Biochrom WPA Biowave II, United Kingdom). Further, the purity and integrity of DNA were tested by running on 0.7% agarose gel electrophoresis.

Library preparation (Illumina TruSeq DNA Sample Prep Kit) and sequencing by synthesis using the Illumina HiSeq2500 device involved standard protocols from the manufacturer (Illumina, Inc., San Diego, CA, United States). The library was quantified by qPCR (Mascher et al., 2013) and sequenced using the rapid run mode (on-board cluster generation, paired-end, 2 × 101 cycles. In total, 125,600,517 Illumina paired end reads were produced having a total output of residues of 42.5 Gb. Prior to the assembly process reads were quality trimmed using clc_quality_trim with a minimal cut-off threshold of Q30 and default settings on remaining parameters. 85.6% of reads and 83.77% of residues passed this initial pre-processing. Subsequently, the quality of the sequence data was checked using fastQC¹. After this quality enrichment a genome coverage of 62-fold was reached.

Our A. littoralis de novo sequence was constructed using CLC assembly cell version 4.3 and the quality trimmed WGS data. The de novo assembly pipeline was applied with automatic detection of best parameters by CLC assembly cell. In accordance with good practice, all contigs below a length threshold of 200 bp were removed. For purification of the constructed assembly we checked our constructed contigs for contamination by E. coli using BLAST + (Camacho et al., 2009). As parameter settings we used a sequence identity of 60% and a word size of 28. Critical contigs were fully removed if the BLASTN analysis resulted in a hit with length > 500 bp. For smaller contigs we reduced the minimal length of a hit to 200 bp, while at the same time at least 10% of the length of the contig is identified as E. coli contamination. From the remaining sequences we removed contigs in case a bacterial origin was detected within the BLASTN analysis against the NCBI non-redundant nucleotide database nt. In addition, we filtered for contigs having a length of 500 bp. The descriptive statistics of both datasets (200 bp and 500 bp) are given in Table 1. The list of all contigs is available at https://doi.ipk-gatersleben.de/DOI/ca99c593-ffdd-4d49-8eab-f1c891953776/d5b041b5-b2c1-4696-bc7c-bc5a32a0c7ec/2/1847940088.

We used the purified WGS assembly without a threshold on contig sizes to predict gene models. Gene prediction was done with GeMoMa (Keilwagen et al., 2016) using gene models of Brachypodium distachyon (Brachypodium_distachyon_v3.0, INSDC Assembly GCA_000005505.4, Feb 2018), Oryza sativa (IRGSP-1.0, INSDC Assembly GCA_001433935.1, Oct 2015) and Sorghum bicolor (Sorghum_bicolor_NCBIv3, INSDC Assembly GCA_000003195.3, Apr 2017) downloaded from Ensembl Plants (Bolser et al., 2017). In total, 15,916 gene models were predicted in 12,130 different contigs. For all detected gene models CDS (FASTA), protein sequence (FASTA) and genomic positions (GFF) are provided. We further investigated these datasets performing a gene annotation with AHRD version 2.0² using UniProt, trembl and TAIR10 (downloaded January 4th 2016). For 13,921 genes (87.5%, Supplementary Table 2) a functional annotation could be assigned (Supplementary Table 3). The complete dataset of gene models and functional annotation is available for download. The coding sequences of all annotated genes are available at https://doi.ipk-gatersleben.de/DOI/ac423f10-971e-481e-bcab-6ac261e27f5c/15d455e1-da91-4e78-82d8-7c7607cb05b9/2/1847940088 (provisional DOI). DOIs of datasets released in this manuscript were constructed using the e!DAL system (Arend et al., 2014).

The repetitive fraction analysis was performed with 89 Mbp of reads of the total genomic DNA (0.26x genome coverage). Quality trimmed reads were grouped with the graph-based clustering algorithm based on sequence similarity, implemented in the RepeatExplorer pipeline (Novak et al., 2013). The paired-end reads clustering was performed with a minimum overlap of 55% and a similarity of 90%. Three independent analyses were performed, using a different dataset of reads of the same sequencing, to confirm the proportions of each cluster within the total genome. Repeat annotation and classification was performed for those clusters with an abundance of at least > 0.01%. For basic repeat classification, protein domains were identified using the tool “Find RT Domains” within RepeatExplorer (Novak et al., 2013). Searches for sequence similarity, using different databases (RepeatMasker and GenBank) were performed and graph layouts of individual clusters were examined using the SeqGrapheR program (Novak et al., 2013). Satellite DNAs were identified based on the TAREAN tool implemented in the pipeline, graph layouts and further examined using DOTTER (Sonnhammer and Durbin, 1995).

A total of 57 DREB proteins from rice (Oryza sativa) and Arabidopsis (Arabidopsis thaliana) were retrieved from the Michigan State University (MSU) rice genome annotation database³, and the Arabidopsis information resource (TAIR) database⁴, respectively.

These data were utilized to find DREB genes (AlDREB) in A. littoralis genomes as the query. AlDREB protein sequences were searched using two approaches. The first technique employed a tBlastn to search against A. littoralis genome sequences, while the second employed BLASTP (E-value 1e5) against A. littoralis protein sequence. Genomic, protein, and CDS (coding DNA sequence) sequences of AlDREB gene family have been identified. To verify the search results, we inspected and analyzed all candidate sequences using Pfam⁵ (El-Gebali et al., 2019), InterProScan⁶ (Jones et al., 2014), and SMART⁷ (Letunic and Bork, 2018) tools. The ProtParam Tool⁸ was used to determine the theoretical isoelectric point (pI) and molecular weight (MW) of the discovered proteins. The MEME website⁹ was used to find conserved motifs in AlDREB protein sequences (Bailey et al., 2009).

Protein motifs and gene structure were visualized by using TBtools software (Chen et al., 2018). Inferring phylogenetic relationships was done with MEGA7.0 software (Kumar et al., 2016) and the Maximum Likelihood (ML) approach based on the LG model.

In order to support our whole genome sequencing data, we were addressing the question of the A. littoralis genome structure. A chromosome number between 2n = 2x = 14 and 2n = 2x = 20 was previously reported for A. littoralis and deposited in different (Zouari et al., 2007; Modarresi et al., 2012) chromosome databases (Rice et al., 2015). Likewise, descriptions of the genome size vary from 342 Mb (Zouari et al., 2007; Modarresi et al., 2012) to 8,215 Mbp (Zonneveld et al., 2005). Therefore the nuclear genome size was estimated by flow cytometry (Figure 3A) using Raphanus sativus cv. “Voran” (Genebank Gatersleben accession number: RA 34; 2C = 1.11 pg) as reference (Schmidt-Lebuhn et al., 2008). Relative fluorescence intensities of around 7,000–10,000 events (nuclei) per sample were measured and the absolute DNA amounts of samples were calculated based on the values of the G1 peak means. The DNA content of the diploid A. littoralis was estimated to be 0.724 ± 0.01 pg/2C (354 Mbp/1C) and is therewith only slightly bigger than reported previously (Wang, 2004). To validate the chromosome number, karyotyping was performed on mitotic chromosome spreads. The chromosome number was determined to be 20 (2n = 2x = 20, Figure 3B). However, occasionally also metaphases with 21 or 22 chromatin units were found. To analyze if these additional chromatin units resulted from satellites being located distally from the corresponding chromosomes or were indeed B-chromosomes, as it was sporadically reported (Liu et al., 2005), fluorescence in situ hybridization (FISH) with the Arabibidopsis-type telomere repeat and 45S rDNA was performed (Figure 3B). The resulting hybridization pattern clearly indicates that the increased number of chromatin units are a consequence of extended nucleolus organizing regions (NORs).

The genome of A. littoralis was sequenced using a whole-genome sequencing approach (WGS) on Illumina’s HiSeq 2500 system. In total, 125 million paired end (PE) reads were produced, reaching genome coverage after quality trimming of approximately 62-fold. This data was sufficient to perform a de novo assembly to construct the first available genomic reference for the A. littoralis species. The constructed genome sequence reached a total size of ∼300 Mbp, which corresponds to 84.7% of 354 Mbp estimated here or 87.7% of the previously estimated genome size of 342 Mbp (Zouari et al., 2007). The assembly consists of 182,747 contigs with a N50 contig length of 3.6 kb. The constructed genomic resource was used for gene prediction and was complemented by a functional annotation of genes. In total, 15,916 gene models were predicted for A. littoralis and 87.5% of them could be assigned a function based on sequence similarity to known genes (Table 1 and Supplementary Figure 1).

In order to characterize the repetitive DNA fraction of A. littoralis the reads from the paired end WGS were used. Reads, comprising in total 0.26-fold genome coverage, were grouped based on sequence similarity into 33,385 clusters containing from 2 to 21,265 reads. Clusters included 32% of all reads, with the major 282 clusters representing at least 0.01% of the genome each. The cluster analysis revealed that 21.69% of the A. littoralis genome is composed of repetitive elements with nine satellite DNA families (satDNAs), nine transposable elements families (LTR-retrotransposons and LINE), two DNA transposons families (CACTA-like and Mutator-like), ribosomal DNA (35S and 5S) and microsatellites (Table 2 and Figure 4A). The most abundant repetitive families were satDNAs, ∼11% of the genome, with the five largest clusters being part of the superfamily AlSat140. Within this superfamily five variants could be identified: AlSat140a, AlSat140b, AlSat140c, AlSat70a and AlSat70b, with 85–96% similarity of the monomer sequence (Figure 4B). Beside these satDNAs, four other satDNAs families were identified in the genome: AlSat256, AlSat897, AlSat372 and AlSat80, with 0.62%, 0.42%, 0.03% and 0.02% of the genome, respectively. The LTR-like retrotransposons constituted 2.22% of the genome, with the Ty3/Gypsy superfamily exceeding 2.4 fold the genome proportion of the Ty1/Copia superfamily. Within the former, Tat/Retand, Tat/Ogre and Chromovirus were the only highly abundant lineages. Within Ty1/Copia retrotransposons the AleI, Ikeros and TAR lineages were identified, with the last being most abundant. Microsatellites were identified in several different clusters comprising 3.23% of the genome (Table 2 and Figure 4A).

As proof of concept for a possible utilization of our genome sequence data the dehydration responsive element-binding (DREB) transcription factor family in Aeluropus was investigated. Members of this family share a specific variant of the AP2 domain including a valine (Val14) and glutamine (Glu19) residue, as well as the YRG motif and the RAYD motif (Sakuma et al., 2002). The name giving AP2-domain is indicated in Figure 5 by the motif combination 3-2-1-4. The YRG domain is depicted as motif 3 (pink) and the RAYD domain as motif 1 (green). The genome wide analysis of the DREB-subfamily in Aeluropus was performed as described (Huang et al., 2019). In total, 16 non-redundant genes (Supplementary Table 4) encoding proteins containing DREB-related motifs (Supplementary Table 5) were identified from the genomic sequences of A. littoralis (Figure 5). All of these proteins contain the family determining motifs 1–4. Figure 6 shows the phylogenetic comparison of the A. littoralis sequences in comparison to the available information for rice (Oryza sativa) derived DREB sequences (Dubouzet et al., 2003). The classification is based on the encoded domain structure and the nomenclature defined by Sakuma et al. (2002). In comparison with the rice DREB gene family sequence (Chai et al., 2020), the identified Aeluropus sequences can be grouped into the subfamilies as depicted in Figure 6. In order to extend this analysis to the dicotyledonous model plants Arabidopsis thaliana, the available sequence information was retrieved and a combined phylogenetic tree generated (Supplementary Figure 2). The identified sequences represent at least one member of each subfamily. The subfamily A2 (II), usually is the subfamily with the highest variation and largest number of representatives. In our dataset the subfamily A2 (II) is represented by four members. An unusual high number of five genes in the Aeluropus genome can be assigned to subfamily A6 (V). Members of this subfamily (like RAP2.4) are involved in stress-specific changes of leaf morphology (Yang et al., 2020).