New chromosome number and cyto-molecular characterization of the African Baobab (Adansonia digitata L.) - “The Tree of Life”

The African baobab (Adansonia digitata L.), also referred to as the “Tree of Life”, is a majestic, long-lived and multipurpose tree of sub-Saharan Africa. Internationally, a growing demand for baobab products in the food, pharmaceutical and cosmetics industries has been observed. Considering this, there is a need for scientific information on the genetics and breeding of A. digitata, including cytogenetics, genetic diversity and reproductive biology. The objectives of our cytogenetic research were to determine the genome size, chromosome number, and organization of ribosomal DNA (45S and 5SrDNA) of A. digitata. Flow cytometry analysis revealed a 2C-DNA value of 3.8 ± 0.6 pg (1Cx monoploid genome size 919.1 ± 62.9 Mbp). Using our improved chromosome preparation technique, we were able to unequivocally count the chromosomes resulting in 2n = 4x = 168, a revised chromosome number for A. digitata. Fluorescent in situ hybridization (FISH) analysis revealed two massively large variants of 45S rDNA and their corresponding nucleolus organizer regions (NOR). The NOR variants were about two to four times larger than the main body of their respective chromosomes. To our knowledge, this is the first report of this phenomenon in a plant species. Furthermore, we found that FISH analysis using the Arabidopsis-type telomere repeat sequence probe clarified and confirmed the new chromosome number and characterized the 45S rDNA structural organization.


Results
Flow cytometry. For each sample, flow cytometry analysis provided counts of nuclei to fluorescence intensity values (displayed as histograms), fluorescence intensity means, and coefficients of variation (Fig. S1). The CV values ranged from 1.98 to 4.12 with a mean of 2.53. An illustrative histogram of the relative DNA content with two peaks corresponding to the G1 nuclei of A. digitata and Solanum lycopersicum L. cv. 'Stupicke' is shown in Fig. S1. The mean 2C nuclear DNA content and 1Cx monoploid genome size of A. digitata seedlings were 3.8 ± 0.6 pg and 919.1 ± 62.9 Mbp, respectively (Table 1).
Chromosome count. We counted chromosomes numbers from 32 intact cells with well separated chromosomes. Three independent counts were made on the same 32 individual cells, providing strong support for 168 chromosomes (Table 1; Fig. 1; Fig. S2). Furthermore, following FISH with Arabidopsis-type telomere repeat sequence (ATRS) probe [(TTT AGG G)n], each chromosome arm showed a pair of telomere signals, making it easier to count the total chromosome number in each cell (Fig. 7). These data provide strong evidence that the A. digitata chromosome number is 2n = 4x = 168 (Table 1), and structurally they can be designated as metacentric, near-metacentric, sub-metacentric, near sub-metacentric and acrocentric (Fig. 1).
45S rDNA FISH. Four 45S rDNA signals were observed across the cell spreads for both seedlings studied; however, as revealed by the intensity of the FISH signals, the 45S rDNA copy number is massively large for each of the four chromosomes for Seedling-1 (Fig. 2) and for three of the four chromosomes for Seedling-2 (Fig. 3). For Seedling-2, the 45S rDNA signal observed was much reduced (approximately 90 to 95%) for the fourth chromosome compared to any of the other three showing FISH signals (Fig. 3). This indicates that the 45S rDNA sites Table 1. Adansonia digitata L. 2C (pg), 1Cx (pg) nuclear DNA content and 1Cx monoploid genome size (Mbp) determined by flow cytometry, and chromosome count determined by cytogenetic analysis. * a The mean is the average of 20 runs from 10 plants (2 runs/plant). * b The mean is the average of 94 counts (3 counters, 31 or 32 cells per counter).   7a (arrowheads)] were consistently observed in both seedlings across the cell spreads (interphase to metaphase chromosomes). The 5S rDNA sites are located interstitially and proximally (near the centromere positions) in two pairs of homologous chromosomes along with bright DAPI bands, which are co-localized with the 5S rDNA FISH signals on the long arms of both homologues (Fig. 3d). One of the homologous pairs is a sub-metacentric chromosome and the other is near-metacentric. A diagrammatic representation of each of the 5S rDNA bearing chromosomes is shown in Fig. 3d (also see enlarged images in inserts, DAPI stained and FISH signals chromosomes). These enlarged chromosomes clearly show the physical location of the 5S rDNA sites. The signal intensities observed varied for the sub-metacentric pair and could be an indication of 5S rDNA copy number variation 32,43,44 . Interphase and prophase 45S and 5S rDNA FISH. We observed numerous dispersed 45S rDNA FISH signals in interphase and prophase cells. DNA in interphase nuclei are highly decondensed, so as expected numerous (as many as 50 minor to major) 45S FISH signals are sporadically distributed in the nucleus (green, Fig. 5a,b; Fig. S4a,c, S7a). Depending on the orientation of interphase nuclei, there is a maximum of four hollow areas that can be observed, and these are the potential bodies of nucleoli, which cannot be stained with DAPI because they are mainly composed of RNA. A group of four hollow areas can be clearly observed in Fig. 5a Fig. S3). Sometimes one larger signal was observed instead of two smaller ones, which is due to the orientation of a chromosome, i.e., if one chromatid is aligned on top of the other one, polar view, yielding one larger signal (Fig. 6a, arrowheads). Furthermore, due to the decondensed nature of DNA numerous ATRS signals were seen throughout the interphase nuclei (Figs. S7a). Due to the extended exposure time necessary for capturing ATRS signals (red), some of the signals appeared to be very large (Fig. 6). In terms of chromosome counting, sometimes a large, metacentric chromosome can be counted as two due to its sharp primary constriction (see insert, Fig. 7a), but this can be clarified by ATRS FISH as shown in Fig. 7b (see inserts) where each chromosome end showed a pair of signals.  www.nature.com/scientificreports/ clumps or clusters causing difficulties in correct estimation 18,49 . In this study, we overcame this common problem with addition of PVP-40 (≥ 10%) to the staining buffer which resulted in improved resolutions of histograms and lower CV values (mean 2.53) for G1 peaks. Our results suggest that the choice and optimization of the staining buffer is very critical for genome size determination of baobab species.

Discussion
Chromosome count. The commonly accepted chromosome number for A. digitata, the only tetraploid species in Adansonia, is 2n = 4x = 160 26,30,35 . Numerous earlier reports (1960 to 1974) on chromosome numbers of A. digitata ranged from 96 to 144 26 . In our study, only chromosome spreads that appeared to be intact, i.e., well-separated and containing all chromosomes of a complete cell, were photomicrographed and then processed (see "Materials and methods") for chromosome counting. The chromosomes typically spread out well in midprophase to early-metaphase where the terminal ends of some chromosomes appeared to be stained lightly with DAPI ( Fig. 1). We provided 32 high quality chromosome spreads, independently, to three individuals trained in cytology for chromosome counting. The mean and standard deviation of the number of chromosomes per cell are 167.87 ± 1.12, with 168 being the median and the mode (Table 1). These data provide strong support for the 2n = 4x = 168 chromosome number for A. digitata. The quality of these spreads and the numbers of replications and independent counts of the chromosomes provided a high degree of confidence, notwithstanding the high chromosome number.
In an earlier study, it was reported that the primary constrictions (the centromere positions) of A. digitata chromosomes were not clearly visible 26 . A more recent karyotypic analysis in Ceiba species, which, like A. digitata, belong to Bombacoideae (Malvaceae), reported that the chromosomes were mostly metacentric 50 . In contrast to Ceiba species, we observed different chromosome types such as from metacentrics to acrocentrics ( It is often difficult to obtain well-separated chromosome spreads with full chromosome complements, to accurately determine a species' chromosome number (and/or ploidy), especially from plants with a high number of small chromosomes (~ 50 and higher). Our modified enzymatic digestion of protoplast technique 40 works well to prepare chromosome spreads from plant species with a higher number of small chromosomes like A. digitata (this report) or Hibiscus hamabo 41 . A key feature of this method is allowing the chromosomes to spread naturally on the glass slides without squashing them with cover slips (Figs. 1, 2, 3, 4, 6, 7; Figs. S2, S3).

45S rDNA FISH.
Recently, 21 different species of the Bombacoideae, including A. digitata, were analyzed for CMA3 banding, and nine and ten species excluding A. digitata for 45S rDNA and 5S rDNA, respectively, with FISH 35 . It is commonly accepted that the 45S rDNA loci in different species are associated with CMA3 banding [51][52][53] . Adansonia digitata of South American origin was reported to have four-terminal CMA3 bands and occupied the entire half of each chromosome as shown in Costa et al. 35 , but 45S rDNA was not specifically addressed in their report. Our results in A. digitata indicate that the structural organization of the 45S rDNA bearing chromosomes are quite different than that of the A. digitata studied by Costa et al. 35 . In our study, the main chromosome body of A. digitata was about half or less of the size of the NOR when condensed at its maximum, or in other words, the NORs are two to four times larger (depending on the degree of condensation,) than the main body of each 45S rDNA bearing chromosome (Figs. 1, 2, 3, 4, 6, 7; Fig. S3). To our knowledge these are the largest NORs relative to chromosome size observed to date in any plant species.
The organization of the ribosomal DNA is often straightforward to determine; however, it can be more complicated, for example, when the copy number varies between homologues. In the case of A. digitata we observed two variants of the 45S rDNA. For Seedling-1, both homologous pairs were found to be homozygous with a slight variation in FISH signal intensity (Figs. 2, 6), while for Seedling-2, one homologous pair is homozygous, and the other pair is heterozygous (Figs. 3, 4, 6, 7; Fig. S3). For the heterozygous locus, most of the secondary constriction (i.e., the NOR) of one of the homologues is missing. Loss and gain of rDNA loci and variation of repeat unit number reported in other plant species is a common feature of genome evolution and speciation 44,53-58 . Our results show structural polymorphism in the 45S rDNA loci of A. digitata, consistent with ongoing genome evolution, but with only two seedlings analyzed we cannot suggest what the cause may be. The results of Costa et al. 35 showed that the NORs in their samples (i.e., a tree of South American origin) were similar to those of Seedling-1. This is consistent with the hypothesis that continental Africa was the origin of A. digitata 59,60 , with 45S rDNA representative of Seedling-1 being the ancestral type. In any case, the heterozygous variant may eventually become fixed through the evolutionary process, i.e., one pair with the massive 45S rDNA sites and the other pair with a much reduced 45S rDNA locus (see Fig. 8, a diagrammatic representation). Further studies are needed to verify this hypothesis.

5S rDNA FISH.
For each of the Bombacoideae species reported to date, one 5S rDNA locus has been observed, and with no chromosomal locations specified for A. digitata 35 . The diagrammatic sketch provided by Costa et al. 35 shows metacentric chromosomes containing 45S and 5S rDNA in the 10 Bombacoideae species, with the 5S rDNA locus being placed interstitially towards the middle of the short arm for each species. In contrast, we observed two 5S rDNA loci in A. digitata, each located proximally (near the centromeric positions and co-localized with bright DAPI bands) on the long arms of two different chromosomes (Fig. 3d, see inserts, enlarged image of each). One of the 5S rDNA sites was on a near-metacentric chromosome pair and the other on a sub-metacentric pair. This result indicates that A. digitata could be an allotetraploid species since two different homologous chromosome pairs carry the 5S rDNA loci, although previous reports suggest it is an Interphase and prophase 45S and 5S rDNA FISH. In our study, numerous 45S rDNA FISH signals were observed throughout interphase nuclei and prophase chromosome spreads. Sometimes more than 35 and 50 FISH signals with various intensities (minor to major) were counted in prophase (Fig. S6) and interphase (Fig. S7a), respectively. The interphase nuclei DNA is highly decondensed compared to prophase and metaphase stages indicating the variations of signal numbers are due to the highly decondensed nature of DNA in interphase and, to a lesser degree, in prophase. No prophase, even in late stage, showed four 45S rDNA FISH signals, and this was true for pro-metaphase and early-metaphase stages [ Fig. 3, (~ nine 45S FISH signals), 7 (13 45S FISH signals, arrows)]. Data from interphase nuclei provide an upwardly biased count of the number of tandemly repetitive DNA loci such as 45S rDNA. It has been reported that the interphase nuclei cannot be used to determine the 45S rDNA loci number due to their decondensed nature 41,43,48,61 and this is supported by genome sequencing where rDNA loci are often missing from genome assemblies as they cannot be correctly assembled 47 .
We suggest this is also the case for prophase and pro-metaphase when the 45S rDNA is massively large like in A. digitata. As the cell progresses from interphase to metaphase, the 45S rDNA FISH signals get markedly denser, so the number of signals gets smaller, and as the chromatin condensation process reaches its maximum level at metaphase, it yields a single strong signal [Figs. 2, 4, 6; Figs. S3, S4 (see model)]. In contrast, the number of 5S rDNA FISH signals remains consistent throughout the cell cycle, from interphase to metaphase. In addition, the 5S rDNA signal intensity is significantly reduced compared to the 45S signals, because the 5S locus contains many fewer copies of rRNA genes compared to the 45S loci.

Arabidopsis-type telomere repeat sequence (ATRS) FISH. Previous work in angiosperms, including
forest trees, has shown that the ATRS are confined to the terminal ends (telomeres) of chromosome arms [46][47][48] . They are composed of five to eight nucleotides and function to protect the chromosome structures from degradation, i.e., maintain the structural integrity of chromosomes 46,62,63 . The 45S rDNA loci (NORs) appeared to be located at the end of four chromosomes (two homologous pairs of chromosomes) in A. digitata, without a distal DAPI-stained satellite region, thus hindering the investigation of the chromosome regions distal of the NORs. To explore these regions, we used FISH with the ATRS probe to localize the telomeres relative to the NORs 48,64-67 . The ATRS FISH signals (red) were observed at the distal end of each 45S rDNA signal (green) confirming the presence of telomere and the lack of a typical DAPI-stained satellite (Figs. 4c-e, 6, 7b; Fig. S3). Furthermore, since every chromosome end is protected by telomeres, the ATRS-FISH signals should provide an accurate chromosome count. However, the FISH signals tended to be overexposed when captured using the automated (default) camera mode for TexRd-X (far red fluorochrome) labeled ATRS probe, thus making it difficult to count the chromosomes using the ATRS-signals (red, see Fig. 6). Therefore, to confirm our chromosome counts, we took a different approach, which was to re-FISH the same slides with 45S rDNA as a control (labeled with Alexa-Flour 488) and ATRS probe (labeled with Cy3) after washing off the first FISH probes (see "Materials and Methods"). The lack of Cy3 signals (red) from the 5S rDNA sites [ (Fig. 7b (arrowheads); S3d (yellow arrowheads)] clearly demonstrates that the probes from the first FISH were completely washed off. The chromosome spread in Fig. 7 Fig. S2). The count perfectly matched the first FISH and the second FISH, providing further support of our chromosome number of 2n = 168 (2n = 4x = 168) for A. digitata.

Conclusion
In this study, we determined the nuclear DNA content, chromosome number, and the distribution and organization of ribosomal DNA (45S and 5S rDNA) of the African baobab tree (Adansonia digitata L.). This is the first report of rDNA characterization for the species while 2n = 2x = 168 constitutes a newly revised chromosome number. Furthermore, our FISH analysis with Arabidopsis-type telomere repeat sequence probe clearly demonstrated that the NOR (major 45S rDNA loci) termini are protected by telomere repeat DNAs, and the telomere FISH signals conclusively demonstrated that the chromosome number of A. digitata is 2n = 4x = 168.  www.nature.com/scientificreports/ The cytogenetic methods and cyto-molecular data presented here can serve as the basis for future in-depth investigations in other Adansonia species to elucidate their genome structure and evolution as well as for other plant species with a high number of chromosomes. In addition, these findings can be beneficial in developing genetic improvement and conservation strategies for A. digitata. Determining the chromosome pairing behavior and utilizing total genomic DNA from diploid Adansonia species as labeled probes in genomic in situ hybridization may provide additional information as to the nature of the ploidy of A. digitata, revealing whether the species is an autotetraploid or allotetraploid or a diploidized tetraploid.

Materials and Methods
Plant materials. Seeds of A. digitata, collected in northeastern Senegal, were acid-scarified by soaking them in 98% sulfuric acid (H 2 SO 4 ) for 24 h, rinsed thoroughly with tap water in a sink under a fume hood, and sown in soil-containing pots (for flow cytometry) or potting media (for cytology) in a greenhouse. Leaves and root tips were collected from fully developed seedlings and used for flow cytometry analysis and cytology investigation.
Flow cytometry. Flow cytometry was performed as described by Sakhanokho et al. 41 with minor modifications to the nuclei staining solution and procedure. In this study, we stained A. digitata nuclei with propidium iodide (PI) for analysis with a BD Accuri C6 flow cytometer and a BD Accuri C6 software version 1.0.264.21 (BD BioSciences, Ann Arbor, MI). The staining kit was used following the manufacturer's instructions. The staining recipe, which was prepared on the day of the flow cytometry procedure, consisted of 20 mL of staining buffer per sample mixed with 120 µL of PI solution, 60 µL of RNAse solution (05-5022; Sysmex Partec GmbH, Görlitz, Germany) and 10% polyvinylpyrrolidone-40,000 (PVP-40, Sigma-Aldrich, St. Louis, Missouri, USA). Fresh leaves of A. digitata seedlings and the internal standard Solanum lycopersicum L. cv. 'Stupicke' (2C 1.96 pg) 70 were co-chopped for 30-60 s using 102 mm razor blades (Electron Microscopy Sciences, Hatfield, PA, USA) to equal size (~ 0.5 cm 2 ) and placed in a Petri dish before addition of 0.5 mL nuclei extraction buffer. The extraction mixture was filtered through 50 μm nylon-mesh filters (Sysmex America, Inc., Lincolnshire, IL, USA) before addition of 2 mL of staining solution (see above).
The mixture was covered with aluminium foil to protect against light and incubated in a refrigerator at 4 °C for 15 min before being submitted to flow cytometry analysis. We used 10 A. digitata seedlings and two leaf samples  rDNA probes (1st hybridization, "a") and re-hybridized with 45S rDNA (as a control) and ATRS probes (2nd hybridization or re-hybridized, "b") a) As many as thirteen 45S rDNA (green signals, arrows) from four NOR bearing chromosomes (dotted lines over green signals numbered as 1, 2, 3 and 4) and four 5S rDNA (red, arrowheads) signals, the spread is divided into eight sections to facilitate accurate counting of chromosomes and the total number is 168 (19 + 19 + 24 + 18 + 25 + 19 + 16 + 28 = 168). One large metacentric chromosome (see insert in the bottom left-hand side section) may appear as two chromosomes (due to its sharp constriction in the center of the chromosome). In such scenario, (a) ATRS hybridization (see insert in "b") would be the option to confirm a single chromosome; (b) The same cell was re-hybridized (i.e., 2nd hybridization, see Method for details) with 45S rDNA (as control, green signal) ATRS probes (red signals) after washing off probes from the 1st hybridization. The 5S rDNA signals are missing (arrowheads), which confirmed the 1st hybridization probes were completely washed off. Though the terminal end of each 45S rDNA-bearing chromosomes showed no visible sign of satellites, the telomere signals confirmed (arrows) that these chromosomes are protected by telomere repeat DNA sequences. The same large metacentric chromosome (insert in "a") showed a pair of ATRS signals (red) at each end (inserts in "b") and no signals in the middle of the chromosome that further confirmed as one chromosome, and the total count was 168. Scale bars are 5 µm. Chromosome spread preparation. Two seedlings (Fig. S8, Seedling-1 and Seedling-2) were used for root tip chromosome spreads and processed separately. Chromosome spreads were prepared following procedures previously described 40,41 with some modifications. Actively growing root tips about 1.0 cm long were harvested and immediately pre-treated with 2 mM (0.036% (w/v)) 8-hydroxyquilonine for 4.0 h in the dark at room temperature (RT, 22-24 °C), rinsed with ddH 2 O and then fixed in 4:1 (95% ethanol (EtOH): glacial acetic acid (GAA)) and stored at RT over-night before processing for enzyme digestion for chromosome spread. The root tips were processed for enzyme treatment within a week after harvest. Fixed root tips were rinsed with deionized water (di-H 2 O ) to remove the fixative for 30 min at RT, mildly hydrolyzed (0.2 N HCl at 60 °C for 10 min and then 10 min at RT), and rinsed with di-H 2 O (20 min at RT) followed by cold (4 °C Freshly prepared slides (3-10 days stored at RT) were treated with RNase-A (30 µg/mL) in 2 × SSC in a coupling jar in a water bath at 37 °C for 60 min followed by two washes in 2 × SSC at 37 °C, 5 min each. The slides were then dehydrated through an ethanol series, 5 min each (70%, 85%, 95% and 100%) at RT and airdried overnight. The hybridization mixture (25 µL/slide) consisted of 50% deionized formamide (12.5 µL of diformamide), 10% dextran sulfate (5 µL of 50% ds), 5.0 µg of E. coli DNA (used as blocking DNA), 25 ng of each of 45S rDNA (total 100 ng/slide) and 5S rDNA (total 100 ng/slide) oligo probes in 2 × SSC (2.5 µL of 20 × SSC) and adjusted the volume to 25 µL with TE buffer. For telomere site detection, we used 45S rDNA (total 100 ng/ slide) as a control along with the ATRS oligo probes (total 100 ng/slide). The hybridization mixture was placed on the spread and covered with a glass-cover slip (22 × 30 mm) without sealing. The slides were then placed in a pre-heated (at 80 °C convection oven) humidity chamber, and then the chromosomal DNA was denatured at 80 °C for 4 min. After denaturation, the slides were cooled down for 2 to 3 min at RT and then placed at 37 °C for incubation for 2 h. After incubation, the glass-coverslip was slid off using a 2 × SSC squeeze bottle, and the slides immediately washed in 2 × SSC at RT for 5 min, two washes in 0.1 × SSC at 40 °C 5 min each, and then washed in 2 × SSC at RT for 5 min followed by quick rinsed in di-H 2 O 41 .  Telomere TTT AGG GTT TAG GGT TTA GGG TTT AGGGT 29 5′TexRd-X, 5′Cy3