Stratified seeds of three ginseng cultivars ‘Sunun’, ‘Chunpoong’, ‘Gopoong’, and a local landrace ‘Hwangsook’ were provided by the Korea Ginseng Corporation (KGC) Natural Resources Research Institute (Daejeon, Korea). Stratified seeds were allowed to germinate in petri dishes with wet filter papers at 10–15°C. The root meristems were then excised (about 2 cm from the root tips), pretreated with 0.002M 8-hydroxyquinoline for 5 hours at 18°C, fixed in 90% acetic acid for 15 min at room temperature (RT, ~24°C), and then stored in 70% ethanol until use.

Somatic chromosome spreads were obtained using a modified version of the technique described by Kato et al. (2004). After thorough washing with distilled water, the meristematic regions of the fixed root tips (~2 mm) were excised and digested in a pectolytic enzyme mix [2% cellulase (MB Cell, Korea), 1.5% macerozyme (Maxim Bio, USA) and 1% pectolyase (Sigma, Japan) in 150 mM Citrate Buffer, pH 4.5] for 75 min at 37°C. The digested meristems were then pipetted into a petri dish with chilled distilled water and incubated on ice for 15 min to wash out the enzymes. Using a stereomicroscope, the root epidermis was removed, and the protoplasts were gently pipetted into a 1.5 ml tube containing 40 μl chilled Carnoy’s fixative. The protoplasts were then suspended by gently vortexing the tube for 30 sec at room temperature, after which the sample was centrifuged at 4, 000 ×g for 3 min and the pellet was resuspended in acetic acid-ethanol (9:1) solution. Finally, the protoplast suspension was pipette-mounted onto ethanol cleaned glass slides, which were placed in a humid chamber to facilitate spreading of the chromosomes and allowed to dry.

A 9-kb fragment of 45S rDNA (18S-5.8S-25S) (Gerlach and Bedbrook 1979) was labeled with biotin-16-dUTP (Roche, Germany) by nick translation. The 5S rDNA was obtained according to the procedure described by Hwang et al. (2009) and then labeled with digoxigenin-11-dUTP (Roche, Germany) by nick translation. Labeled DNA fragments within the range of 200–500 bp were used as probes.

Slide pretreatment. To remove contaminating RNA, the slides were treated with RNase A buffer (RNase A final conc. 100 μg ml^-1 in 2× SSC) for 1 hr at 37°C. The slides were then incubated in 0.01 M HCl for two minutes, followed by subsequent treatment in pepsin buffer [stock: 10% (w/v) pepsin in dH₂O, working: 1:100 dilution in 0.01 M HCl] for 10 min at 37°C to lyse endogenous proteins that could cause background signals. Next, the chromosomes were fixed by treating the slides with 4% paraformaldehyde in 2× SSC. Finally, the slides were dehydrated in ethanol series (70%, 90%, 100%, 3 min each) and air-dried. The slides were washed in 2× SSC for 5 min (3×) between each step. All incubation steps at 37°C were conducted in a humidified chamber.

Probe hybridization.The hybridization mixture contained 50% formamide, 10% dextran sulfate, 2× SSC, 5 ng μl^-1 salmon sperm DNA and 500 ng μl^-1 of each probe DNA adjusted with DNase- and RNase-free water (Sigma, USA, #W4502) to a total volume of 40 μl/slide. The mixture was denatured at 90°C for 10 min and immediately kept on ice for at least 5 min prior to mounting on slides. After covering with a glass cover slip, the chromosomes were denatured at 80°C for 3–5 min on a hot plate. The slides were then immediately transferred into a humid chamber preset at 37°C and incubated overnight (~16 hr). The following day, the slides were washed in 2× SSC (15 min at RT), 0.1× SSC (35 min at 42°C), and finally 2×SSC (30 min at RT).

Signal detection. The slides were treated with TNB [0.1 M Tris-HCl, 0.15 M NaCl, 1% (w/v) blocking reagent] at RT for 5 min, after which they were subjected to antibody detection. Briefly, biotinylated 45S rDNA probe was detected with streptavidin-Cy3 conjugate (Zymed, USA), while digoxigenin-labeled 5S rDNA probe was detected using anti-digoxigenin-FITC conjugate (Sigma, USA). Both antibodies were diluted in TNB to a ratio of 1:100, and the slides were then incubated at 37°C for one hour. Excess reagents were subsequently washed off in TNT [0.1 M Tris-HCl, 0.15 M NaCl, 0.2% (v/v) Tween-20] at 37°C for 5 min (3×), after which they were subjected to dehydration in ethanol series (70%, 90%, 100%, 3 min each) and air-dried. Chromosomes were then counterstained with a premixed DAPI solution [1 μg ml^-1 DAPI in Vectashield (Vector Laboratories, USA)].

The three cultivars and one landrace of Panax ginseng evaluated in this study were all confirmed to have a chromosome complement of 2n=48 (Fig. 1). With reference to the centromere position (i.e. arm ratio), the complement comprised 12 metacentric (1–7, 11–13, 15, and 18), 9 submetacentric (8–10, 16–17, 19, and 22–24), and 3 subtelocentric (14 and 20–21) homologous pairs with a karyotype formula of 24m+18sm+6st. The chromosome lengths ranged from 3.27 to 6.30 μm (Table 1).

Only one pair of satellited chromosomes (pair 14) was observed, and the only locus of 45S rDNA in the genome was localized at the secondary constriction of this subtelocentric chromosome (Figs 2, 3 and Table 2). Moreover, one locus of 5S rDNA signal was detected at the intercalary region of the short arm of chromosome 11. This locus was flanked by two DAPI bands. There was no variation in the number of rDNA loci among the three cultivars and one landrace of Panax ginseng investigated in this study (Fig. 1).

Numerous DAPI-binding heterochromatic regions were dispersed along all chromosomes and were visible as DAPI dots. These dots, similar to those in chromosomes 5 and 8, did not form distinct DAPI bands. Both the DAPI dots and bands were made more easily visible by inverting the images (Fig. 2c and d).

In addition to the rDNA loci, the presence of several observable DAPI bands along the chromosome complement made identification of homologous pairs possible. The number of the observed bands further increased as the resolution increased after subsequent enhancement of the image sharpness in Genus™. A total of 32 DAPI bands were initially observed in the sharpness-enhanced DAPI images, but six additional DAPI bands were observed after using the inverse tool of Genus™ with adjustment to the brightness and contrast, resulting in a total of 38 bands (Fig. 2b and d).

Twelve of the observed DAPI bands were localized on the short arms, while 26 were on the long arms (Table 2). Among the 24 chromosomes, four had no band (5, 8, 15, and 23), six had one band (2, 3, 9, and 12–14), 11 had two bands (1, 4, 7, 16–22, 24), two had three bands (6 and 10), and one had four bands (11). Furthermore, chromosomes were grouped according to the presence or absence of DAPI bands on each arm (Fig. 3). Group A had no band on either arm (pairs 5, 8, 15, and 23), group B had no band on the short arm, but one band on the long arm (2, 3, 9, and 12), group C had no band on the short arm, but two bands on the long arm (7, 17, and 21), group D had no band on the short arm, but three bands on the long arm (6 and 10), group E had one band on the short arm, but none on the long arm (13 and 14), group F had two bands on both arms (11), and group G had one band on each arm (1, 4, 16, 18–20, 22, and 24).

In addition to the chromosome length, centromeric position, and rDNA loci distribution, we utilized the observed DAPI bands to characterize the chromosomes. Collectively, these DAPI bands could be very useful in identifying homologues for further cytogenetic analyses, especially of the Panax ginseng genome, which comprises a large number of chromosomes with mostly similar sizes. The distinguishing features of each chromosome are presented in Table 2.

We detected only one locus each for 5S and 45S rDNA, which is in agreement with the results reported by Choi et al. (2009). However, the 45S rDNA signal was more intense than the 5S rDNA signal. Owing to the semi-quantitative nature of FISH (Maluszynska and Heslop-Harrison 1991), this could indicate that the 45S rDNA has more repeating units than the 5S rDNA in the ginseng genome.

Localization of the rDNA resulted in our only being able to easily pair two out of the 24 homologues. However, the existence of several DAPI bands distributed along most of the chromosomes greatly facilitated the identification of the other homologous pairs, which otherwise would have been challenging owing to the very low size diffe-rence among most ginseng chromosomes. As a result, DAPI banding, which has been utilized in several previously conducted studies (e.g. Schweizer 1976, Heng and Tsui 1993, Costa Silva et al. 2011), was found to also be an excellent cytogenetic marker in ginseng. Further analysis of the chromosomes based on the DAPI banding pattern on each arm enabled us to categorize them into seven groups (Fig. 3). This technique, which utilizes the presence or absence of DAPI-binding heterochromatin, has the potential for use in future karyotype analyses of ginseng varieties and other Panax species.

Chromosomal DAPI bands are caused by the preferential binding of DAPI to AT-rich heterochromatic DNA segments (Schweizer 1976, Eriksson et al. 1993, Heng and Tsui 1993, Kubota et al. 2000) that are long enough to be seen using a fluorescence microscope, suggesting that these DAPI-intense heterochromatic regions in ginseng are AT-rich DNA segments. This information should be useful in the ongoing ginseng genome sequencing because it enables identification of possible characteristics of heterochromatin types present in its genome. Nevertheless, further molecular and cytogenetic analyses are necessary to quantify the AT content of these regions and isolate DNA sequences specific to these heterochromatic bands, like the DAPI-intense signal of the 180-bp knob-specific satellite repeat in maize (Lamb et al. 2007), which is about 56% AT (Peacock et al. 1981, Ananiev et al. 1998).

The use of the rDNA loci number and distribution pattern of other Panax species can be useful in deducing the phylogenetic relationship among these species. Choi et al. (2009) showed that wild ginseng and American ginseng (Panax quinquefolius), although geographically isolated, have equal numbers of 5S and 45S rDNA loci (2 and 1, respectively), while the cultivated ginseng, although found in the same geographic area with the wild ginseng has only one locus of each rDNA. Although further research is needed to confirm the possible phylogenetic significance of this report, we found only one locus for each type of rDNA in all three cultivars and one local landrace of Panax ginseng.

Karyotype data are essential to understanding the phylogenetic relationships among species belonging to the same family (Heslop-Harrison and Schwarzacher 2011, Mendes et al. 2011), making them useful to cyto-taxonomic studies (Pinto et al. 2012). Additionally, comparative cytogenetics provide knowledge regarding the cytogenetic relationships between diploid species and their polyploid cytotypes, as well as between allopolyploids and their ancestral genomes (Kovarik et al. 2005, Leflon et al. 2006, Snowdon 2007, Wang et al. 2007, Kolano et al. 2008, Xiong and Pires 2011).

Most species belonging to the family Araliaceae are 2n=24 or 2n=48, except for a few genera that have little chromosomal number variation (Yi et al. 2004). In a review of the chromosomal evolution of the family Araliaceae, Yi et al. (2004) discussed that, although the actual basic chromosome number of the family was thought to be x=12, some species were 2n=36. These species would be triploids if the basic chromosome number 12 is considered, but triploids are genetically unstable. This caused a challenge in establishing the basic chromosome number of the family. The x=12 hypothesis was further challenged after the genus Hydrocotyle which has several taxa with 2n=18, 36, and 60 were moved into Araliaceae form Apiaceae, giving an alternative basic chromosome number x=9 and x=6. Nevertheless, x=12 is generally accepted as the basic chromosome number in the family, but this does not eliminate the possible ancestral x=6 (Yi et al. 2004). One hypothesis cannot easily rule out the other but further phylogenetic and karyotype analyses in the family are necessary to resolve these competing hypotheses.

Considering a basic chromosome number of 12 or 6, ginseng would be considered a tetraploid or octoploid, respectively; the latter having a more ancient nature. Recently, Choi et al. (2011) showed the high replication of homologous genes in ginseng using SSR markers and suggested that the polyploidy could range from tetra- to octoploidy. Nevertheless, in practice, Panax ginseng is regarded as a tetraploid species with a basic chromosome number of 12 (Wen and Zimmer 1996, Court 2000, Yi et al. 2004, Choi et al. 2009).

Our data showed a somatic cell chromosome complement of 2n=48, supporting previously reported chromosome numbers (Ko et al. 1993, Choi et al. 2009) and polyploidy (Wen and Zimmer 1996, Court 2000, Yi et al. 2004, Choi et al. 2009). However, evaluation of the rDNA loci number revealed only one locus for each 5S and 45S rDNA, despite its polyploid nature. This reduction of rDNA loci may be explained by the non-additive nature of rDNA loci and other genomic DNA segments after polyploidization (Snowdon et al. 1997, Ozkan et al. 2003, Yoshikazu et al. 2006). More over, loss of the duplicate loci may be brought about by single-generation or rapid genome/chromosomal reorganization (Wendel 2000, Heslop-Harrison and Schwarzacher 2011), or from the gradual action of concerted evolution after genome duplication or alloploidization (Kovarik et al. 2005). In the former case, it would be difficult to tell whether ginseng is an ancient polyploid, while in the latter, the loss of these loci would provide obvious evidence of an ancient polyploidization event. However, some species do not really reflect a correlation between the rDNA loci number and the level of ploidy; in fact, polyploids can even have half the number of rDNA signals than their diploid counterparts (Yoshikazu et al. 2006). This rDNA reduction phenomenon has been well-documented in the Artemisia species (Pellicer et al. 2010).

Additionally, based on localization of the 45S rDNA near the centromere area and the intercalary position of the 5S rDNA, it is just as likely that these loci were favored to survive locus loss from non-additive recombination over their duplicated counterparts, which probably would have been in more distal positions, or epigenetically silenced (Kovarik et al. 2008).