Cytogenetic relationships among Citrullus species in comparison with some genera of the tribe Benincaseae (Cucurbitaceae) as inferred from rDNA distribution patterns

Comparative mapping of 5S and 45S rDNA by fluorescent in situ hybridization (FISH) technique is an excellent tool to determine cytogenetic relationships among closely related species. In this study, the number and position of 5S and 45S rDNA loci in all Citrullus species and subspecies were determined. The cultivated watermelon (C. lanatus subsp. vulgaris), C. lanatus subsp. mucosospermus, C. colocynthis and C. naudinianus (or Acanthosicyos naudinianus) had two 45S rDNA loci and one 5S rDNA locus which was located syntenic to one of the 45S rDNA loci. C. ecirrhosus and C. lanatus subsp. lanatus had one 45S rDNA locus and two 5S rDNA loci, each located on a different chromosome. C. rehmii had one 5S and one 45S rDNA locus positioned on different chromosomes. The distribution of 5S and 45S rDNA in several species belonging to other genera in Benincaseae tribe was also investigated. The distribution pattern of rDNAs showed a great difference among these species. The present study confirmed evolutionary closeness among cultivated watermelon (C. lanatus subsp. vulgaris), C. lanatus subsp. mucosospermus and C. colocynthis. Our result also supported that C. lanatus subsp. lanatus was not a wild form of the cultivated watermelon instead was a separate crop species. In addition, present cytogenetic analysis suggested that A. naudinianus was more closely related to Cucumis than to Citrullus or Acanthosicyos, but with a unique position and may be a link bridge between the Citrullus and the Cucumis.

that three subspecies of 'Citrullus lanatus' are unrelated species and there are seven species in Citrullus including C. naudinianus.
Although various analyses have been carried out aiming to establish phylogenetic relationships among different Citrullus species, the conclusions are inconsistent and the progenitor of cultivated watermelon has not yet been determined [5][6][7][8][9][10][11]. Cytological [12] and crosscompatibility [13] observations favored C. colocynthis as the ancestor of cultivated watermelon. Several studies established that C. lanatus var. citroides was phylogenetically sister to cultivated watermelon [3,[14][15][16]. However, C. ecirrhosus has been found to have closer relative relationship with cultivated watermelon based on sequencing analysis of cpDNA regions [17,18]. Jarret and Newman [9] proposed that C. rehmii might be the ancestral to cultivated watermelon by the analysis of internal transcribed spacer sequence heterogeneity.
Mapping of 5S and 45S rDNA by fluorescent in situ hybridization (FISH) technique is an excellent tool to determine phylogenetic relationships because their map positions can reveal similarities and differences between chromosomes of related species [19][20][21]. To date, the position and number of rDNA loci have been determined in more than 1000 plant species with FISH [22]. These studies showed that the number and position of the 5S and 45S rDNA were usually characteristic of a given species, genus or group [23] and 5S and 45S rDNA tended to occupy similar chromosomes and positions in closely related species [24][25][26][27][28]. The chromosomal localization of 5S and 45S rDNA loci has been reported using the FISH technique in three C. lanatus subspecies [29]. The study [29] showed there were one 5S locus and two 45S rDNA loci on chromosomes 4 and 8 and the 5S rDNA locus was located syntenic to one of the 45S rDNA loci on chromosome 8 in C. lanatus subsp. vulgaris and C. lanatus subsp. mucosospermus. Whereas C. lanatus subsp. lanatus contained one 45S locus on chromosomes 4 and two 5S rDNA loci, one 5S rDNA locus was on chromosome 8 and the other 5S rDNA locus was on chromosome 11. The rDNA distribution has also been investigated in C. colocynthis and C. rehmii besides C. lanatus var. lanatus and C. lanatus var. citroides [30]. However, the chromosomes with rDNA sites have not been identified due to the lack of chromosome identification markers in Reddy et al.' study [30]. Recently, we successfully labeled specific chromosomes using oligonucleotides (oligos) libraries as probes in cucumber [31]. In this study, the distribution of the 5S and 45S rDNA in all Citrullus species including three C. lanatus subspecies were investigated and the chromosomes with rDNA sites were identified using synthesized watermelon oligos probes to gain cyto-evolution at the chromosome level and resolve relationships among Citrullus species.

rDNA probes and oligo probes
The plasmids of 5S and 45S rDNA which were cloned in the vector pUC18 were kindly provided by K. Arumuganthan, University of Nebraska. The 5S and 45S rDNA were labeled with digoxigenin-dUTP and biotin-dUTP via nick translation, respectively. Two oligos libraries were developed to identify cultivated watermelon chromosomes 4, 8 and 11 with rDNA sites reported by Guo et al. [29]. Library 1 contained 10000 oligos from watermelon chromosomes 4 and 8 at the densities of 2-3 oligos per kilobases, respectively. Library 2 contained 10000 oligos from watermelon chromosomes 8 and 11 at the densities of 2-3 oligos per kilobases, respectively. The region spanned by library 2 was located adjacently the region spanned by library 1 on chromosome 8. Thus, the chromosome with overlapping signals must be chromosome 8 and the others are either 4 or 11. Two oligos libraries were also developed to identify melon chromosomes 8, 10 and 12. Library 1 contained 10000 oligos from melon chromosomes 8 and 10 at the densities of 2-3 oligos per kilobases, respectively. Library 2 contained 10000 oligos from melon chromosomes 10 and 12 at the densities of 2-3 oligos per kilobases, respectively. The region spanned by library 2 was located adjacently the region spanned by library 1 on chromosome 10. Thus, the chromosome with overlapping signals must be chromosome 10 and the others are either 8 or 12. The oligo libraries were synthesized by MYcroarray (Ann Arbor, MI). Each synthesized oligo contained 48 bp of genomic sequence, a 5′ F primer, which included the T7 RNA polymerase promoter sequence, and a 3′ R primer. Amplification and labeling of oligos libraries was the same as the protocol developed by us [31].

Fluorescence in situ hybridization (FISH)
FISH was performed according to published protocols [32]. Firstly, slides were hybridized with oligo librariy 1 (biotinlabeled) and oligo librariy 2 (digoxigenin-labeled). After the first round of probing and image capture, coverslips were taken off carefully and wash the slides three times in 1X PBS (phosphate-buffered saline) (5 min each). The slides were then dehydrated in an ethanol series (70, 90, and 100 %, 5 min each), denatured again in 70 % formamide at 80°C for 2 min, dehydrated in a second ethanol series, and reprobed with the 45S rDNA (biotin-labeled) and 5S rDNA (digoxigenin-labeled) sequences simultaneously. Biotin-labeled and digoxigenin-labeled probes were detected using a fluorescein isothiocyanate (FITC)-conjugated antibiotin antibody (Vector Laboratories) and a rhodamine-conjugated antidigoxigenin antibody (Roche Diagnostics), respectively. Chromosomes were counterstained with 4,6-diamidino-2-phenylindole (DAPI) in a VectaShield antifade solution (Vector Laboratories). Images were captured digitally using a CCD camera (QIMA-GING, RETIGA-SRV, FAST 1394) attached to an Olympus BX63 epifluorescence microscope. Gray-scale images were captured for each color channel and then merged. Final image adjustments were done with Adobe Photoshop (Adobe Systems).

Chromosome identification using watermelon oligos probes in Citrullus species and A. naudinianus
It is difficult to distinguish each chromosome in watermelon because the chromosomes are relatively small and are morphologically similar. Two oligos probes were developed to identify watermelon chromosomes 4, 8 and 11. As expected, two oligo probes produced bright FISH signals on cultivated watermelon chromosomes 4, 8 and 11 (Fig. 1a1, a3). We also performed FISH on metaphase chromosomes in all Citrullus species including three C. lanatus subspecies and A. naudinianus in order to investigate the potential of oligo probes for cross-species hybridization. Two probes generated FISH signals with similar intensity on three pairs of chromosomes in C. lanatus subsp. mucosospermus (Fig. 1b1, b3), C. colocynthis (Fig. 1c1, c3), C. lanatus subsp. lanatus (Fig. 1d1, d3), C. ecirrhosus (Fig. 1e1, e3), C. rehmii (Fig. 1f1, f3), and on two pairs of chromosomes in A. naudinianus (Fig. 1g1, g3). The counterparts of cultivated watermelon chromosomes 4, 8 and 11 in other Citrullus species or subspecies and A. naudinianus were also designated as chromosomes 4, 8 and 11 based on the homeologous relationship with the corresponding chromosome of watermelon.

Chromosomal distribution of 5S and 45S rDNA in Citrullus species and A. naudinianus
To determine the distribution of 5S and 45S rDNA in Citrullus species and A. naudinianus, following the FISH analysis of using oligos probes, the slides were washed and reprobed with 5S and 45S rDNA sequences simultaneously. The results were summarized Table 1. Consist with previous results [29], the genome of cultivated watermelon contained one 5S locus (red) and two 45S rDNA loci (green) on the chromosomes 4 and 8 (Fig. 1a2, a4). The 5S rDNA locus (red) was colocalized with the 45S rDNA loci (green) on the chromosome 8 (Fig. 1a2, a4). The number and location of 5S (red) and 45S (green) rDNA in the genomes of C. lanatus subsp. mucosospermus (Fig. 1b2, b4) and C. colocynthis (Fig. 1c2, c4) were identical to those in the genome of cultivated watermelon (Fig. 1a2, a4).

Discussion
The genus Citrullus belongs to the Benincaseae tribe of the Cucurbitaceae family [1]. The Benincaseae is a relatively large tribe with 204-214 species in 24 genera [33]. The phylogenetic relationships among species in this tribe were assessed by morphological [34,35], geographical [4,36], biochemical [37] or nucleotide data [4,36,38,39]. In this study, we investigated the distribution of 5S and 45S rDNA in several genera located on the different branches of the phylogenetic tree of the tribe Benincaseae [36]. We found L. siceraria and cultivated watermelon had exactly same rDNA distribution. Similarly, C. sessilifolia had identical rDNA distribution pattern as D. palmatus. This was in accordance with the results based on the molecular phylogenetic studies, which revealed a close relationship between the genera Citrullus and Lagenaria, and between the genera Coccinia and Diplocyclos [4,36].
Consist with previous studies [29,30], the present FISH analysis demonstrated that the rDNA gene loci showed wide differences among the Citrullus species. C. lanatus subsp. vulgaris (cultivated watermelon) had same number and location of 5S and 45S rDNA loci to those in C. lanatus subsp. mucosospermus and C. colocynthis, but differed from those in C. lanatus subsp. lanatus. Our results are concordant with assumptions that the progenitor of the cultivated watermelon might be C. colocynthis [12,13] and C. lanatus subsp. mucosospermus is the recent ancestor of C. lanatus subsp. vulgaris [29]. Several reports based on nuclear and plastid data of the genus Citrullus also confirmed evolutionary closeness among cultivated watermelon, C. lanatus subsp. mucosospermus and C. colocynthis. For example, the analyses of the sequence variation at cpDNA regions also confirmed the relationship because some of the C. lanatus var. lanatus accessions shared a unique substitution at the trnE-trnT region with all C. colocynthis accessions [40]. Schaefer et al. [36] also confirmed that the watermelon and C. colocynthis evolved from a common ancestor. Our results also supported Chomicki and Renner' findings [4], which revealed that C. lanatus subsp. lanatus was not a wild form or progenitor of the cultivated watermelon but instead was a separate crop species, domesticated independently. In addition, present cytogenetic analysis showed that C. lanatus subsp. lanatus and C. ecirrhosus were more closely related to each other than they are to other Citrullus species. Based on the current FISH analysis and previous molecular phylogenetics [4] and cytogenetic studies [29,30], a scheme about phylogenetic relationships among the Citrullus species was given (Fig. 4).
One of the most striking observations from the present results was that the number, size and location of 5S and 45S rDNA signals were exactly identical between A. naudinianus and C. metuliferus in genus Cucumis. Moreover, two melon oligos libraries generated distinct FISH signals in A. naudinianus. The number and location of rDNA and oligos probes signals were completely identical between A. naudinianus and C. metuliferus. However, unambiguous signals weren't detected when melon oligos probes hybridized to chromosomes of Citrullus species and watermelon oligos probes hybridized to chromosomes of Cucumis species. In addition, A. naudinianus had 24 chromosomes while other Citrullus species had 22 chromosomes. In the genus Cucumis, C. sativus is the only species with chromosome number 14, whereas the rest of the Cucumis species have a basic number 24 [41]. Our previous study also confirmed that the A. naudinianus had closer genetic affinity with the genus Cucumis than with Citrullus by comparative genomic in situ hybridization [42]. All above results suggest that A. naudinianus is more closely related to Cucumis than to Citrullus or Acanthosicyos. Interestingly, both watermelon and melon oligos probes generated distinct FISH signals on A. naudinianus chromosomes although cross-genus hybridization was not feasible using oligo probes in other species. Therefore, A. naudinianus could be a link bridge between the Citrullus and the Cucumis.
Consistent identification of individual chromosomes in a species is the foundation for successful cytogenetic research. FISH signals are reliable markers for chromosome identification. Most common FISH probes used in chromosome identification have been repetitive DNA elements or large genomic DNA clones [43][44][45][46][47]. Large genomic clones often contain dispersed repetitive sequences that will cause high background signal in FISH and cannot be used as chromosome-specific FISH probes [48]. The watermelon chromosomes are relatively small in size and are morphologically similar. Repetitive DNA elements were unavailable for differentiating all watermelon chromosomes. Recently, eleven BAC clones were used to differentiate eleven watermelon chromosomes and assign linkage groups to their corresponding chromosomes [49]. Furthermore, using the eleven BAC clones, 5S and 45S rDNA loci were assigned to chromosomes 4 and 8 in the genomes of cultivated watermelon and C. lanatus subsp. mucosospermus, and chromosomes 4, 8 and 11 in C. lanatus subsp. lanatus [29]. However, in present study we found that one of 5S rDNA sites was on an unidentified chromosome rather than on chromosome 11 by using oligos probes to identify chromosome 11 in C. lanatus subsp. lanatus.

Conclusions
The present study confirmed evolutionary closeness among cultivated watermelon, C. lanatus subsp. mucosospermus and C. colocynthis. Our result also supported that C. lanatus subsp. lanatus was not a wild form of the watermelon instead was a separate crop species. In addition, present cytogenetic analysis suggested that A. naudinianus was more closely related to Cucumis than to Citrullus or Acanthosicyos, but with a unique position and may be a link bridge between the Citrullus and the Cucumis. In future, we will develope chromosome-specific oligos probes for identifying each watermelon chromosome. Furthermore, bulked oligo probes will be designed to cover an entire watermelon chromosome. Thus, interchromosomal rearrangements and karyotype evolution among related Citrullus species can be revealed in more detail through cross-species chromosome painting. The information will be useful for collection and utilization of genetic resources for cultivated watermelon improvement.

Ethics
Not applicable.

Consent to publish
Not applicable.

Availability of data and materials
All data are contained within the manuscript.
Abbreviations FISH: fluorescent in situ hybridization.

Competing interests
The authors declare that they have no competing interests.
Authors' contributions YHH designed the study, performed the analysis and wrote the paper. YX, JMW and ZYL participated in the design of the study, analyzed the data and critically revised the manuscript. KPL, YXW, HZ, YW and XML carried out the experiments and acquisition of original data. KPL analyzed the data and participated in writing. All authors read and approved the final manuscript.