The distribution of repetitive sequences along chromosomes in Cucumis species revealed by self-GISH
The distribution patterns of repetitive DNAs along chromosomes in Cucumis species were investigated based on the sGISH method in which gDNA probe was hybridized onto its own metaphase chromosomes. Five Cucumis species including C. sativus, C. hystrix, C. melo, C. metuliferus and C. anguria which are of Asia or African origin (Fig. 1) were investigated [11]. The sGISH signal patterns for each species were showed in Fig. 2.
Signal patterns from C. sativus sGISH
Self-GISH produced unique signal patterns in each chromosome of C. sativus (Fig. 2a-1). Except for the end of long arm of chromosome 6, obvious signals were found at the ends of all other chromosomes. Very strong signals appeared at the both ends of chromosomes 1 and 7, short arm ends of chromosomes 4 and 6, and long arm end of chromosome 5. In addition, the distinct signal patterns were also observed at the pericentromeric heterochromatin regions of each chromosome, and very strong signals appeared in chromosomes 1, 2, 4 and 7. The signals in chromosomes 2 and 4 covered a large region across centromeres. The signal patterns revealed sGISH at the pericentromeric heterochromatin regions colocalized with those from 45S rDNA and Type III (Additional file 1: Figure S1).
In order to confirm that the signals located on the ends of chromosomes are from other repetitive sequence rather than from telomeres, we probed simultaneously the Arabidopsis type telomere on C. sativus chromosomes. The telomere signals were detected at the ends of every chromosome of cucumber (green signals in Fig. 2a-2). The FISH from telomere and sGISH gave different signal patterns at the ends of each chromosome, which confirmed repeats-derived sGISH patterns (Fig. 2a-3). Interestingly, besides the terminal signals, some interstitial signals from telomere probe were also detected on chromosomes 1 and 2 (shown by arrows in Fig. 2a-2). Chromosome complement for C. sativus species was determined based on the signals of 45S rDNA (Fig. 3) and chromosome morphology according to previous reports [22, 25].
Signal patterns from C. hystrix sGISH
Due to no reference data about the karyotype of C. hystrix so far, the chromosome number was given here based on their sizes, in descending order (Fig. 2b). Self-GISH produced very strong signals at two ends or one end of chromosomes (Fig. 2b-1), which indicated a preferential location of repeats at the distal regions of chromosomes in this species. Chromosome 1 gave very strong signals at the both ends. Chromosomes 3, 4 and 8 display strong signals at the end of long arms, and weak signals at the end of short arms. Three weak signals produced in chromosomes 3, 4 and 8 were found to locate at the same positions with 45S rDNA loci as showed in Fig. 3g. All other chromosomes produced strong signal only at one end of chromosomes. We also compared the FISH mapping of telomere and sGISH. The FISH signals from telomere probe were detected at the ends of every chromosome and gave different patterns with sGISH, which indicted sGISH signals was produced by specific repeats rather than telomeres in C. hystrix (Fig. 2b-3).
Signal patterns from C. melo sGISH
For C. melo, signals from sGISH were detected at the primary constriction region of every chromosome (Fig. 2c-1). Further, we used CentM repeat located at the C. melo centromere regions [24], to compare the difference of sGISH and CentM signals. The FISH results showed that CentM produced exactly the same signal patterns as that from sGISH (Fig. 2c-2). In addition, we found that sGISH of this species did not produce any signals at the nucleolar organizing regions (NORs) (shown by arrow in Fig. 2c-3), which bear 45S rDNA loci (Fig. 3m). The FISH result from telomeric probe detected clear signals at the ends of each chromosome (Additional file 1: Figure S2). Unlike C. sativus and C. hystrix, sGISH of C. melo did not produce any signals at the ends of chromosomes.
Signal patterns from C. metuliferus sGISH
For C. metuliferus, except for two chromosomes without any sGISH signals, all other ten chromosomes possessed obvious sGISH signals. Among ten chromosomes having signals, nine chromosomes displayed signals at the both ends of chromosomes, while chromosome three produced signal only at one end (Fig. 2d-1). We also used the telomere probe to differentiate the signals of sGISH from telomeres. The different telomere signal pattern eliminated the possibility of sGISH signals from telomere (Fig. 2d-2). Like C. melo species, sGISH in C. metuliferus did not detected any signals at the nucleolar organizing regions (shown by arrow in Fig. 2d-3) which were detected by 45S rDNA probe (Fig. 3s). These results showed that the dominant repeats derived from GISH in C. metuliferus were located at the distal ends for majority of chromosomes.
Signal patterns from C. anguria sGISH
For C. anguria, sGISH detected the signals at the pericentromeric heterochromatin regions of all chromosomes (Fig. 2e-1). However, whether the signals are from the centromeric region or not could not be confirmed, because the centromere repeat sequence in this species is not available so far. Like in other species, the telomeric probe produced bright signals at the ends of each chromosome (Fig. 2e-2).
The homology of repeats among Cucumis species based on the comparative genomic in situ hybridization
To reveal the homology of repeats among five Cucumis species, comparative GISH (cGISH) was employed to probe the signals of gDNA on metaphase chromosomes of all other species (Fig. 3). We also compared the FISH mapping of 45S rDNA on metaphase chromosomes of these species (shown by arrows in Fig. 3). The FISH signal patterns were shown in Fig. 3. cGISH from different xspecies revealed the distinct signal patterns when different gDNA probes were used. It is found that on C. sativus metaphase chromosomes, the gDNA of C. hystrix probed obvious subtelomere signals on majority of chromosomes (Fig. 3b), which is similar to the signal patterns produced by C. sativus sGISH at the chromosome ends (Figs. 2a and 3a). However, cGISH on C. sativus chromosomes using gDNAs of C. melo, C. metuliferus and C. anguria showed obvious signal patterns at the pericentromeric heterochromatin regions of some chromosomes (Fig. 3c, d, and e), which have same position with 45S loci (data not shown).
In C. hystrix metaphase chromosomes, cGISH using C. sativus gDNA probe produced bright signal patterns at majority of chromosome ends (Fig. 3f), which is similar to that produced by C. hystrix sGISH (Fig. 3g). Interestingly, gDNA of C. melo gave clear signals at the pericentromeric heterochromatin regions of all chromosomes of C. hystrix, and NORs as well (Fig. 3h). However, gDNAs of C. metuliferus and C. anguria did not probe strong signals along chromosomes except for the six 45S rDNA loci (Fig. 3i and j).
In C. melo, both gDNA probes of C. sativus and C. hystrix only detected four bright signals on four chromosomes (Fig. 3k and l), which were found to be the 45S rDNA loci (as shown with arrows in Fig. 3m). The signal patterns detected by gDNA of C. metuliferus and C. anguria showed bright 45S rDNA loci signals, as well as some weak telomeric and scattered signals along every chromosome (Fig. 3n and o).
Similar to C. melo, in C. metuliferus metaphase chromosomes, gDNA probes from C. sativus and C. hystrix only produced signals at the NORs (Fig. 3p and q). Genomic DNA probe from C. melo detected the pericentromeric heterochromatin regions of every chromosome, and NORs as well (Fig. 3r). However, cGISH using C. anguria gDNA produced weak scattered signals on some chromosomes, and clear NORs signals.
In C. anguria chromosome spreads, gDNA probes of C. sativus and C. hystrix also only detected the 45S rDNA loci (Fig. 3u and v). cGISH using gDNA probe of C. melo produced clear signals around pericentromeric heterochromatin domains, though the signal intensity varied in different chromosomes, and 45S rDNA loci (Fig. 3w). Genomic DNA probe of C. metuliferus produced scattered signals along C. anguria chromosomes (Fig. 3x).
Different signal pattern of 45S rDNA were produced among five Cucumis species. In C. sativus, 10 chromosomes displayed 45S rDNA loci, including 6 very strong loci and 4 weak signals, which all were located adjacent to centromeric regions. Six chromosomes in C. hystrix bear 45S rDNA loci which were mapped at the distal of chromosomes. And only 4 chromosomes displayed 45S rDNA loci with interstitial or subtelomeric locations in C. melo, C. metuliferus and C. anguria species (shown by arrows in Fig. 3). However, all five species bore a pair of 5S loci (data not shown).
Comparative mapping of specific satellites revealed the significant divergence among Cucumis species
High similarity of signal patterns shown by cGISH from C. satvius and C. hystrix (shown by Fig. 3a, b, f and g) is likely to be explained by the high homology of satellites. To further reveal the homology of satellites between species, specific types of satellites, including Type I/II, Type III, Type IV and CentM were comparatively mapped on metaphase chromosomes of five Cucumis species. Among these satellites, Type I/II and Type IV have been mapped on the subtelomeric domains of C. sativus chromosomes, and Type III was located at the centromeric regions of C. sativus chromosomes [22, 23, 26], and CentM was identified as the centromeric satellite DNA of melon [27]. As expected, Type I/II (Fig. 4a) and Type IV (Fig. 4b) produced bright signals at the subtelomeric regions of almost all chromosomes, and Type I/II was mapped at the distal position compared with Type IV in C. sativus chromosome spreads (Fig. 4c). Interestingly, Type I/II and Type IV also produced clear signals at the end of chromosomes of C. hystrix (Fig. 4d, e and f), presenting the similar relative positions as them in C. sativus chromosomes, though their copy numbers varied significantly in individual chromosome based on the sizes and intensities of the FISH signals. However, Type I/II and Type IV did not probe clear signal in other three Cucumis species. Type III and CentM probes only detected the centromeric signals in C. sativus and C. melo, respectively, and no signal in any other species (Data not shown).
To characterize the molecular organization and the abundance of the specific tandem repeats observed by fluorescence in situ hybridization, southern blot was conducted to further analyze the genomic organization of them in Cucumis species. Genomic DNA of C. sativus, C. hystrix, C. melo, C. metuliferus and C. anguria was digested with the restriction enzymes EcoRI, blotted, and hybridized with Type I/II, Type IV, CentM and 45S rDNA probes. We found that C. sativus and C. hystrix shared almost the same patterns of Type I/II and Type IV, but the fragments from C. sativus are stronger than those from C. hystrix. The ladder of fragments indicates a tandem organization of these repeats. However, no bands appeared in other three species for Type I/II and Type IV probes. For CentM, obvious ladder of bands was observed only in C. melo species. For 45S rDNA, five species gave three types of patterns. C. sativus and C. hystrix produced the same two bands, and C. metuliferus and C. anguria shared the same one band. C. melo produced one band with shorter size compared with that from other species (Fig. 5).
The relationship between GISH patterns and gene distribution
The sGISH showed the distinct signal patterns in different Cucumis species. Some species such as C. sativus, C. hystrix and C. metuliferus produced bright telomeric signals, and some species, such as C. melo and C. anguria only detected signals in pericentromeric heterochromatin region. These results showed the obvious difference of chromosomal structure in Cucumis species. We further analyzed the gene density along chromosome to reveal the relationship between chromosomal structure and gene distribution. In Cucumis, C. sativus and C. melo are the only two species whose genome sequences are available [28, 29]. The chromosome 5 was selected randomly to conduct this analysis. According to the Cucumber Genome Database (http://cucumber.genomics.org.cn/page/cucumber/index.jsp) and Melon Genome Database (https://melonomics.net/), the numbers of annotated genes along chromosome 5 in two species are 3344 and 1840, respectively. The number of annotated genes per 300 kb was calculated to investigate the gene density along chromosome 5 of both species. The distributions of gene density along the two chromosomes were illustrated in Fig. 6a and b, respectively. The uneven distribution of gene density was observed along chromosomes 5 in both species. In general, lowest gene density was observed in the heterochromatin regions (showed by the gray color in Fig. 6) in two chromosomes, and relative high gene density in the euchromatin region of chromosome arms. However, obvious different distribution patterns were found between two chromosomes. In C. sativus chromosome 5, the highest gene density appeared at the central regions of chromosome arms, and decreased towards the ends of chromosome and pericentromeric heterochromatin region (Fig. 6a). However, in C. melo chromosome 5, the gene density increased towards the ends of chromosome, and the highest appeared at the ends (Fig. 6b). The distribution patterns of gene density of these two chromosomes were found to be relative with the sGISH patterns. The regions with strong sGISH signals, like the ends of chromosomes and pericentromeric heterochromatin regions in the chromosome 5 of C. sativus (Fig. 2a-1) have the lowest gene density (Fig. 6a). Similarly, in chromosome 5 of C. melo species, the lowest gene density (Fig. 6b) was detected at the pericentromeric heterochromatin region with strong sGISH signals. Meanwhile, a significant increase of gene density towards the ends of chromosome arms was observed where nearly no sGISH signals were detected (Fig. 2c-1).