Abstract
The complex polyploid genomes of three Saccharum species have been aligned with the compact diploid genome of Sorghum (2n = 2x = 20). A set of 428 DNA probes from different Poaceae (grasses) detected 2460 loci in F1 progeny of the crosses Saccharum officinarum Green German × S. spontaneum IND 81-146, and S. spontaneum PIN 84-1 × S. officinarum Muntok Java. Thirty-one DNA probes detected 226 loci in S. officinarum LA Purple × S. robustum Molokai 5829. Genetic maps of the six Saccharum genotypes, including up to 72 linkage groups, were assembled into “homologous groups” based on parallel arrangements of duplicated loci. About 84% of the loci mapped by 242 common probes were homologous between Saccharum and Sorghum. Only one interchromosomal and two intrachromosomal rearrangements differentiated both S. officinarum and S. spontaneum from Sorghum, but 11 additional cases of chromosome structural polymorphism were found within Saccharum. Diploidization was advanced in S. robustum, incipient in S. officinarum, and absent in S. spontaneum, consistent with biogeographic data suggesting that S. robustum is the ancestor of S. officinarum, but raising new questions about the antiquity of S. spontaneum. The densely mapped Sorghum genome will be a valuable tool in ongoing molecular analysis of the complex Saccharum genome.
THE consequences of polyploid formation in plants are exemplified by comparison of the closely related genera Saccharum (sugarcane) and Sorghum. In as little as 5 million years since Saccharum and Sorghum diverged from a common ancestor (Al-Janabiet al. 1994b), Saccharum species have reached gametic chromosome numbers ranging from 18 to 85 or more. Two wild Saccharum species are recognized: Saccharum spontaneum (2n = 36-128), with a putative center of origin in India, and S. robustum (2n = 60-170), putatively from New Guinea. The cultivated species S. officinarum (2n = 70-140) is thought to be derived from S. robustum, and also has a center of origin and diversity in New Guinea. While sugarcane breeders regard S. officinarum as a separate species, plant taxonomists generally view it as a cultivated derivative of wild S. robustum. Cultivated taxa referred to as S. barberi in India and S. sinense in China are believed to be natural hybrids of S. spontaneum and S. officinarum (Sreenivasanet al. 1987; Roach 1995; J. E. Irvine, unpublished data). Molecular analyses of chloroplasts (Sobralet al. 1994), mitochondria (D’Hontet al. 1993), and ribosomal DNA (Glaszmannet al. 1990) support the concept of S. spontaneum as a separate species. The aborted inflorescence species S. edule (2n = 60, 70, 80, and some aneuploids) may be a hybrid of S. officinarum or S. robustum with Miscanthus sp. (Daniels and Roach 1987).
Modern sugarcane varieties derived from hybrids between S. officinarum and S. spontaneum are among the world’s most valuable crops, at about $143 billion annually (Christouet al. 1992). High sugar content is thought to be contributed by S. officinarum while vegetative vigor and resistance to biotic/abiotic stresses are attributed to S. spontaneum.
A high level of chromosome duplication and autogamous chromosome pairing in sugarcane preclude genetic mapping based on codominant “alleles.” DNA markers showing simplex or “single-dose” segregation (Wuet al. 1992) have been used to construct genetic linkage maps of S. spontaneum SES208 (2n = 64) and its doubled haploid ADP068 (Al-Janabiet al. 1993; da Silvaet al. 1993). This map consisted of 64 linkage groups and 8 homologous groups based on 276 restriction fragment length polymorphism (RFLP) and 208 single-dose (SD) arbitrarily primed polymerase chain reaction (PCR) loci (da Silvaet al. 1995). Mapping of 408 RFLP loci in a cultivated S. officinarum × S. spontaneum hybrid R570 (2n = 107-115) revealed 96 linkage groups and 10 putative homologous groups (Grivetet al. 1996). Fifty-one linkage groups of S. officinarum LA Purple were mapped on the basis of 160 randomly amplified polymorphic DNA markers and 1 morphological marker, in a cross with S. robustum (Mudgeet al. 1996).
Molecular analysis of the complex autopolyploid genome of Saccharum might be simplified by exploiting its close relationship with the compact genome of Sorghum. The chromosome number of Saccharum ranges from 2n = 36 to 170 (Daniels and Roach 1987; J. E. Irvine, unpublished data), while chromosome number of Sorghum ranges from 2n = 10 to 40 (Doggett and Prasada Rao 1995). A high-density sorghum map has been made in a cross between Sorghum bicolor and S. propinquum (Chittendenet al. 1994; A. H. Paterson et al., unpublished data), both 2n = 20. The range 2547-4183 Mbp/1C in DNA content of several Saccharum spp. with unspecified chromosome number (Arumuganathan and Earle 1991) is approximately consistent with a three- to sixfold duplication of the genome of S. bicolor (2n = 20; 760 Mbp/1C). This suggests that the difference in DNA content of the Saccharum and Sorghum genomes may largely be attributable to the larger number of chromosomes in the Saccharum nucleus, rather than a higher percentage of repetitive DNA.
Two prior studies have suggested parallels in gene order between Saccharum and Sorghum, but had available only sparsely distributed DNA markers. Dufour et al. (1997) evaluated colinearity on the basis of 84 anchor probes distributed at intervals of ∼20 cM across the Sorghum genome of >1500 cM (Chittendenet al. 1994). Guimarães et al. (1997) compared Saccharum and Sorghum on the basis of only 68 probes, and 63 (69%) of their Saccharum linkage groups contained two or fewer loci, making it impossible to test whether or not intrachromosomal rearrangements are present.
A long-term goal of our work is to study the basis, and the consequences, of the rapid increase in chromosome number of Saccharum. The specific objectives of this study were as follows: (1) to construct detailed linkage maps for the three Saccharum species using a high-density sorghum linkage map as a template, (2) to assess in detail the levels and patterns of chromosome structural rearrangement among Saccharum species and between Saccharum and Sorghum, (3) to examine genome coverage of the Saccharum maps through comparative genome analysis, (4) to compare the extent of preferential chromosome pairing in different Saccharum species, and (5) to reexamine the elusive question of the basal chromosome number(s) for Saccharum.
MATERIALS AND METHODS
Plant materials: Three interspecific F1 populations were used, each derived from crosses between heterozygous parents: (1) 85 F1 plants from S. officinarum Green German (GG, 2n = 97-117) × S. spontaneum IND 81-146 (IND, 2n = 52-56); (2) 85 F1 plants from S. officinarum Muntok Java (MJ, 2n = 140) × S. spontaneum PIN 84-1 (PIN, 2n = 96); (3) 80 F1 plants from S. officinarum LA Purple (LP, 2n = 80) × S. robustum Molokai 5829 (MOL, 2n = 80). The SES 208 population was previously described (da Silvaet al. 1995). Populations Green German × IND 81-146 (G × I) and PIN 84-1 × Muntok Java (P × M) were genotyped at Texas A&M, LA Purple × Molokai 5829 (L × MOL) was genotyped at the Hawaiian Sugar Planters’ Association, and SES 208 was genotyped at Cornell. New DNA probes were added to a previously described sorghum population (Chittendenet al. 1994) at Texas A&M.
Genotyping: Genomic DNA extraction was slightly revised from Li et al. (1995). About 7.5 μg sugarcane DNA per lane was digested with EcoRI, HindIII, or XbaI, according to the manufacturer’s instructions. Southern blotting, radioactive labeling, and autoradiography were as described, except that 3 μl 32P per reaction was used (Chittendenet al. 1994). A total of 829 DNA probes were surveyed on G × I and P × M, including 132 cDNA and 18 genomic clones from published sugarcane maps (da Silva et al. 1993, 1995): 201 hypomethylated (PstI) Sorghum genomic clones (pSB, SHO); 199 random maize cDNA probes (CSU); 40 maize cDNA (CSU) and 45 genomic (BNL and UMC) clones previously mapped in sorghum; 89 rice cDNA (RZ) and 41 rice genomic (RG) clones; 19 barley cDNAs (BCD) and 61 oat cDNAs (CDO). Probe-enzyme combinations that generated the largest numbers of polymorphic bands were chosen for mapping. For L × MOL, 132 sugarcane cDNA probes, 25 sugarcane genomic probes, and some additional probes from rice, maize, and oats were surveyed. Data for the SES 208 population are as previously described (da Silvaet al. 1995).
Data analysis: Polymorphic restriction fragments from each parent were scored present vs. absent in the mapping progeny. Each marker was tested against the expected ratios for SD or simplex (1:1), double-dose (DD) or duplex (11:3 for x = 8, or 7:2 for x = 10), and “triple-dose” (TD) or triplex segregation (13:1 for x = 8, or 11:1 for x = 10; da Silva 1993; da Silva and Sorrells 1996). For determining significance thresholds, we assumed that S. spontaneum is x = 8 but S. officinarum and S. robustum are x = 10, on the basis of the periodicity observed among known accessions (J. E. Irvine, unpublished data). A chi-square threshold of 2% was used to keep both type I and type II errors below 5% (Wuet al. 1992). There is variation in the chromosome counts available for GG (97-117) and IND (52-56). When analysis involved chromosome number, the middle values of 107 for GG and 54 for IND were used.
SD markers were mapped using Mapmaker v2.0 for the Macintosh (Landeret al. 1987). Linkage groups were built at threshold LOD = 5.0 and θ = 0.25. Genetic distances (in centimorgans) were calculated using the Kosambi mapping function. DD and TD markers were superimposed on the SD map by the method of da Silva (1993). Homologous groups (HGs) were assembled on the basis of at least two probes common to two linkage groups. The nomenclature of HGs 1-8 was consistent with the previously published SES 208 map (da Silvaet al. 1995). New HGs that shared no homologous loci with SES 208 HGs 1-8 were denoted HGs 9-17 and are consistent across the six maps from the three new crosses described herein. Inverted segregation data of each genotype were used to investigate repulsion phase linkages (Al-Janabiet al 1994a; Grivetet al. 1996).
Summary of sugarcane genome mapping data
RESULTS
DNA polymorphism, segregation, and recombination: Of the 829 probes surveyed, 496 were polymorphic between S. spontaneum and S. officinarum (G × I and P × M crosses) with at least one restriction enzyme. Forty-six probes were polymorphic for G × I only while one probe was polymorphic for P × M only. The overall level of polymorphism detected was 65.4% for G × I and 60.0% for P × M. A total of 428 probes could be genotyped on G × I and/or P × M yielding 615, 575, 536, and 418 RFLP markers for GG, MJ, IND, and PIN, respectively (Table 1). Increasing ploidy level was closely correlated with increasing percentages of DD (r = 0.99) and TD (r = 0.99) markers. The number of DD and TD markers that could be mapped were 14 and 3 in GG, 10 and 1 in MJ, 13 and 0 in IND, and 4 and 1 in PIN. A total of 47, 40, 99, and 28 loci for GG, MJ, IND, and PIN, respectively, showed segregation ratios that could not be explained by SD, DD, or TD, and might be due to distorted segregation. Among them, 27 (57%), 19 (48%), 68 (69%), and 11 (39%) have significantly less than 50% “present” bands, as clear evidence that some chromosomes are not being reliably transmitted, particularly from IND. Only the markers that fit the expected segregation ratio were used for mapping. On L × MOL, 31 probes generated 109 markers for LA Purple and 85 for Mol 5829.
Seventy-two linkage groups were formed for both S. officinarum Green German and Muntok Java, and 69 for both S. spontaneum IND 81-146 and PIN 84-1 (Table 1). Consistent with its higher chromosome numbers, the total number of polymorphic loci detected was much higher for S. officinarum (737 for GG and 739 for MJ) than for S. spontaneum (492 for both IND and PIN). The overall length of each of the four maps (2304, 1443, 2063, and 1303 cM) is approximately proportional to the number of SD markers mapped (270, 206, 248, 182). The high significance threshold for inferring linkage reduced false positives but excluded distant linkages (>25 cM), contributing to the large number of unlinked markers: 148 in GG, 149 in MJ, 137 in IND, and 115 in PIN. By assuming that unlinked markers are at least 20 cM from the nearest marker(s) linked to the map, we estimate the minimum overall lengths of these maps as 5246, 4222, 4803, and 3603 cM. The average distance between linked markers was 8.5, 7.0, 8.3, and 7.2 cM for GG, MJ, IND, and PIN, respectively, and was more closely correlated with the cross than with the parental species. Twenty-three linkage groups were assembled in S. officinarum LA Purple and 19 in S. robustum Mol 5829; however, a large proportion of unlinked markers impels caution in interpreting the LA and MOL maps.
Assembly of Saccharum homologous groups, and alignment with Sorghum chromosomes: Multiple loci detected by single probes permitted the identification of 10, 9, 10, and 7 HGs for GG, MJ, IND, and PIN, respectively. Alignment was based on two or more common markers among homologs as illustrated in Green German HGs 3 and 8 (Figure 1). For example, Green German linkage groups 63 and 69 share related loci detected by four common DNA probes, pSB173, SG212, SG202, and pSB167. Other common probes contributing to assembly of this HG 3 include pSB173 and SG305 on linkage groups (LGs) 48 and 69; CSU527 and CDSR157 on LGs 48, 49, and 60; pSB173, CSU536, and SG302 on LGs 60 and 63; CSU537, SG212, SG202, UMC147, and pSB167 on LGs 61 and 63; and CSU536 and SG305 on LGs 48, 60, and 62. HG 8 was assembled by LGs 1, 2, 3, 4, 7, 8, 9, 24, 26, and 35.
Virtually all markers on sugarcane HGs 3 and 8 were located on sorghum LG C (Table 2). If a sugarcane LG did not clearly correspond to other sugarcane homologs, but the majority of its markers did correspond to a particular sorghum LG, the sugarcane LG was aligned with the sorghum linkage group as well. Only four (5.5%, 13 loci), five (6.9%, 13 loci), three (4.3%, 9 loci), and four (5.8%, 8 loci) sugarcane LGs for GG, MJ, IND, and PIN, respectively, could not be aligned with any of the 10 sorghum LGs.
—Conserved genomic regions between sorghum linkage group (LG) C and sugarcane Green German homologous groups (HGs) 3 and 8. Loci connected by a line are detected by the same probe in both genomes and demonstrate how HGs were assembled (See results). ∩ indicates that the relevant probe may correspond to a locus at approximately the indicated location, but no DNA polymorphism was found at the (inferred) locus. Genetic distance in Kosambi centimorgans is shown at left. The underlined markers on HG 8 (BCD1424, CDSC25, and CDSB73) show linkage in repulsion phase. Approximate map positions of double-dose (*) and triple-dose (**) loci are inferred by the method of da Silva (1993). Because these locations are approximate, conflicts in marker order that are solely due to DD or TD loci cannot be inferred to represent chromosomal rearrangements.
Intrachromosomal rearrangements between Saccharum and Sorghum chromosomes: To test possible intrachromosomal rearrangements between Saccharum and Sorghum, the markers involved in the possible rearrangement, plus two flanking markers when available, were used to calculate LOD scores for the alternative orders. Only if the LOD difference between alternative orders exceeded 2.0 in both Saccharum and Sorghum was a rearrangement inferred. Because apparent intrachromosomal translocations might actually result from multiple inversions, we referred to both simply as “rearrangements.” For example, in GG HG 3 (Figure 1), LG 49 showed pSB071 mapping between CSU448 to CSU527, a major rearrangement relative to sorghum. On LGs 61 and 63, pSB167, SG212, and SG202 mapped to the region between CSU536 and CSU537. These two differences in gene order between Saccharum and Sorghum may reflect a single chromosomal rearrangement; however, we do not yet have DNA markers necessary to prove this. An inversion on LG 7 of HG 8 was detected by CDSR97, CDSB15, and UMC76. This inversion was supported by the positions of CDSR97 and CDSB15 on LG 2, although the position of the DD locus UMC76 was less certain.
Homologous loci between Saccharum and Sorghum
On the basis of present data, we conclude that at least 13 chromosomal rearrangements appear to differentiate sorghum from sugarcane. These rearrangements involved parts of sugarcane HGs 2, 3, 8, and 13 that corresponded to sorghum LGs A, B, C, and H. Two of 13 rearrangements were supported by evidence from two sugarcane species, S. officinarum and S. spontaneum. Among the 11 additional rearrangements found, 5 (45%) were detected on more than one LG within S. officinarum or S. spontaneum, while the remaining 6 were detected on one LG only (Figure 2, Table 3). No evidence that contradicted these rearrangements was found within or among Saccharum species; however, in many cases there were insufficient loci to test the rearrangements on other homologs.
Interchromosomal rearrangements between Saccharum and Sorghum chromosomes: Among the total of 10, 9, 10, and 7 putative HGs from GG, MJ, IND, and PIN, respectively, some were found to be segments of a single HG when aligned with the sorghum linkage map (Figure 2, Table 4). For example, sugarcane HGs 3 and 8 were likely a single HG corresponding to sorghum LG C. One sorghum LG, F, corresponded to only one sugarcane HG. Seven sorghum LGs, C, D, E, G, H, I, and J, each corresponded to 2 sugarcane HGs. The remaining two sorghum LGs, A and B, each corresponded to 3 sugarcane HGs. The 17 sugarcane HGs could be reduced to 10 nonoverlapping sets that corresponded to different regions of the sorghum genome—9 of these sets were found in S. officinarum and 9 in S. spontaneum. The missing sets were different in each species.
—RFLP linkage maps of sorghum generated from S. bicolor × S. propinquum F2 population and corresponding genomic regions from five sugarcane maps (see materials and methods). The filled region in the sorghum map indicates approximate location of the centromere based on maize (LGs A, B, C, and F) and wheat (LG D; Patersonet al. 1995) genetic maps. These five sugarcane maps each correspond to the genome of individual parental varieties: Green German (G); Muntok Java (M); IND 81-146 (I); PIN 84-1 (P); SES 208 (S). Within each cell, the first number in parentheses is the number of loci detected by the corresponding DNA probe. The second number is the number of these loci that could be assigned to LGs; zero indicates that all loci were unlinked to the map. After the letter indicating the parent, the next number is the corresponding HG(s). HGs 1-8 are as previously described (da Silvaet al. 1995), and HGs 9-17 are newly identified. The letter A-J followed by a number in the data column indicates a subset of incongruous (not colinear between Saccharum and Sorghum) loci detected by an otherwise congruent DNA probe. If it is at the end of a cell, it indicates duplication in the Saccharum genome; if it is at the beginning of a cell, then it is not mapped in this part of the Saccharum genome. For example, the probe CDSR125 on sorghum LG D was found in two IND LGs corresponding to sorghum LGs A and D, respectively, hence it is designated [(2/2)I 5: A2]; the same probe was found in a GG LG corresponding to sorghum LG A, but not LG D, and designated [A1: (6/1)G 13]. These incongruous loci are also listed on the bottom of each linkage map with the probe name and the LG where the probe maps in sorghum. The approximate position of each incongruous locus in sugarcane is shown to the right of the individual map (data column) that each locus belongs to. Differences in the order of loci between sorghum and sugarcane (possible chromosomal rearrangement) are indicated by a line with arrows at both ends for a possible inversion, or a line with an arrow on one end for a possible translocation. For those DNA probes that could not be mapped in sugarcane (far right column), gb indicates background hybridization was prohibitively high; np, no polymorphism; ns, no signal; ls, light signal; d, segregation distortion—markers showed <50% present band.
Cases where 1 sugarcane HG corresponds to parts of two or more sorghum LGs are likely to reflect interchromosomal rearrangements. Because the sorghum map is complete, with the number of LGs corresponding to the number of chromosomes, we can rule out the possibility that such overlap means that the sorghum LGs are part of a single chromosome. Four of the 10 sugarcane HGs (4, 5, 6, and 10) show correspondence to two sorghum LGs each (B and E, D and J, B and F, and E and J). These rearrangements could be accounted for by as few as three chromosomal breaks (one each in LGs B, E, and J).
Only one of the possible interchromosomal rearrangements (HG 4 with LGs B, E) is found in both S. officinarum and S. spontaneum.
Details of correspondence between sugarcane and sorghum maps: A total of 439 loci (84%) detected by 242 probes (Figure 2, Table 2) could be compared between Saccharum and Sorghum. Their corresponding arrangements are as follows:
Sorghum LG A: This group corresponded to HGs 2, 12, and 13 of GG, HGs 2 and 12 of MJ, HG 2 of IND, HGs 2 and 12 of PIN, HG 2 of SES, and HGs II and III of R570 (Grivetet al. 1996; Figure 2, Table 2). The numbers (and percentages) of loci in the respective sugarcane maps that support this correspondence are 18 (90%) on 8 LGs in GG, 15 (88%) on 13 LGs in MJ, 20 (95%) on 11 LGs in IND, 12 (92%) on 10 LGs in PIN, 3 (50%) in SES, and 4 (80%) in R570 (Table 2). New HGs 12 and 13 may be terminal segments of SES HG 2. A cluster of 4 loci—CSU382 and SG442 of IND HG 2, CDSR94 of PIN HG 2, and CDSR153 of SES HG 2—were located on sorghum LG G. A second cluster of 4 loci, CDSR125 of GG HG 12 and IND HG 2, and CDSC5 and SHO74 of MJ HG 12 were located on sorghum LG D. CDSB29 of GG and MJ HG 2, and CDSC24 of SES HG 2 were located on sorghum LG J and LG H, respectively. Seven (19%) of 36 LG A probes surveyed were not polymorphic in sugarcane. Based on data from five sugarcane maps, about 90% of sorghum LG A was mapped in sugarcane (Table 5).
An intrachromosomal rearrangement differentiates sugarcane HG 2 from sorghum LG A, marked by relocation of the pSB079-CDSC056 interval to a region between CSU456 and CDSR067 in sugarcane. This rearrangement was supported by two LGs in the IND map (S. spontaneum) and one LG in the GG map (S. officinarum), and not contradicted by any LGs, indicating that it likely happened before the divergence of the Saccharum species from each other. MJ and PIN did not segregate at appropriate loci to detect this rearrangement.
Two other rearrangements were suggested by individual GG and IND linkage groups (respectively), in regions for which other LGs did not segregate at the affected loci. In GG, pSB581 was relocated to the segment between pSB1368 and CDSB62, while in IND, an inversion occurred in the segment between SG293 and BNL9.11 (Figure 2). The corresponding regions of MJ and PIN were too sparsely mapped to test these rearrangements.
Sorghum LG B: This group corresponded to HG 11 of GG, part of HG 4 of MJ, HG 4 of PIN, parts of HGs 4 and 6 of SES, and HG X of R570. Although about 65% of this sorghum LG was collectively covered by the five sugarcane maps, only small HGs could be formed in individual GG, PIN, and SES maps. The numbers (and percentages) of loci supporting correspondence to sorghum LG B are 11 (85%) from nine LGs in GG, 8 (89%) from six LGs in MJ, 5 (71%) from five LGs in IND, 7 (70%) from seven LGs in PIN, 5 (36%) on SES HGs 4 and 6, and 5 (63%) on R570 HG X. Four contiguous loci, CDO686, RZ567, RZ166, CDO1380, of the seven in SES HG 6 were located on a clustered region of sorghum LG F. Since only half of the HG 4 loci of SES and MJ, and half of the HG 6 loci of SES were located on sorghum LG B, this LG may not be well conserved in sugarcane. Two incongruous loci, CDSR91 in PIN and CDSR49 in IND, were located on sorghum LG E. Four more exceptions, CDSR155 in GG, pSB604 in MJ, CDSC57 in IND, and CSU536 in PIN mapped to sorghum LG C. CDSR126 in PIN and pSB204 in GG were located on sorghum LGs A and J, respectively. Seven (26%) of 27 probes surveyed on LG B were not polymorphic in sugarcane.
An inversion involving CSU81 and CSU422 was detected in one LG of GG. In other GG linkage groups, as well as in MJ, IND, and PIN, the loci needed to further test this rearrangement did not segregate for polymorphisms.
Sorghum LG C: This group corresponded to HGs 3 and 8 of GG, HG 8 of MJ, HGs 3 and 8 of IND and SES, and HGs I, VI, and IX of R570. The numbers (and percentages) of loci supporting correspondence to sorghum LG C are 30 (86%) on 16 LGs in GG, 21 (88%) on 12 LGs in MJ, 25 (100%) on 12 LGs in IND, 16 (94%) on 10 LGs in PIN, 16 (73%) in SES, 7 (78%) in R570. Among 15 exceptions (12%) of the total 122 loci mapped on HGs 3 and 8 in these five sugarcane maps, 6 loci detected by 4 probes were located on sorghum LG D, 4 loci on sorghum LG I, 2 on LG B, and 1 each on LGs A, E, F, and G (Figure 2; Table 2), respectively. Six (11%) of 53 probes on LG C surveyed were not polymorphic in sugarcane, with no marked clustering. About 90% of sorghum LG C was collectively covered by the sugarcane maps.
An important feature is that two HGs in GG, IND, and SES corresponded to sorghum LG C, with “breakpoints” at similar positions. However, HG 8 of MJ extended over these break points. HGs 3 and 8 in IND overlapped at locus RZ421. The pooled data from these five sugarcane maps indicates that sorghum LG C is conserved as a single HG in sugarcane, which is comprised of both HGs 3 and 8. These HGs were previously thought to be independent (da Silvaet al. 1995).
Extensive intrachromosomal rearrangement appears to have occurred between sorghum LG C and sugarcane HGs 3 and 8. An inversion involving loci CSU450, CSU28, and UMC76 occurred in two LGs of MJ and two LGs of IND. Two other inversions involving UMC76, CDSB15, and CDSR97, and SG370 and CDSC57 of GG (two and one LGs, respectively), were also observed. A rearrangement in three LGs of GG involving loci CDSB6, CSU450, and CSU28 was detected in a nearby (but different) region to a rearrangement in IND involving CDSB15, pSB600, and CDSR97. No LG in MJ and PIN covered these rearranged regions. Two additional rearrangements were detected in GG (one and two LGs, respectively), but could not be tested with available loci in MJ, IND, and PIN. A rearrangement was also observed in SES from SG212 to the region between RZ421 and CDO66. This is the same region as one of the GG rearrangements, but in the opposite direction. While extensive intrachromosomal rearrangements are apparent in this part of the sugarcane genome, only the inversions in MJ and IND can be asserted to be consistent across taxa.
Sorghum LG D: This group corresponded to HG 5 of GG, HGs 5 and 14 of MJ, HG 5 of IND and PIN, part of HG 5 of SES, and HG VIII of R570. The numbers (and percentages) of loci supporting correspondence to sorghum LG D are 13 (93%) loci on 10 LGs in GG, 11 (85%) loci on 9 LGs in MJ, 14 (100%) loci on 8 LGs of IND, 15 (88%) loci on 14 LGs in PIN, and 3 (50%) loci in R570. HGs 5 and 14 of MJ overlapped on the same region of sorghum LG D, suggesting that they are a single HG. HG 5 of SES covered part of sorghum LGs D and J with only two markers on each LG. Among exceptions, CDSB29 in GG and MJ were on sorghum LG J; CDSR35 in MJ and PIN were on LG C; CSU539 in IND, CDSR95 in PIN, and RZ143 in SES were on LG E. CDO365 of SES was on LG H. Six out of 32 (19%) probes on LG D surveyed were not polymorphic in sugarcane. About 75% of sorghum LG D was mapped in sugarcane, and no rearrangement could be detected.
Sorghum LG E: This group corresponded to part of HG 10 of GG, part of HG 4 of MJ and SES, and HG X of R570. A putative recently discovered translocation between S. bicolor and S. propinquum (A. H. Paterson, unpublished data) complicated the analysis of this region. The numbers (and percentages) of loci supporting correspondence to sorghum LG E are four (100%) on four LGs in GG, four (80%) on four LGs in MJ, two (67%) on three LGs in IND, six (86%) on seven LGs in PIN, and four (80%) in SES. In R570, the only two probes corresponding to LG E were located on HG X that also partly corresponded to sorghum LG B. HG 4 of MJ and SES also spanned parts of LGs B and E, and HG 10 of GG covered segments of LG E and J. Among four incongruous loci, one was on LG B, two on LG C, and one on LG J. Ten (53%) of the 19 LG E probes surveyed were not polymorphic in sugarcane. About 40% of sorghum LG E was mapped in sugarcane. The short LGs in sugarcane maps corresponding to LG E made it impossible to detect rearrangements in this part of the sugarcane genome.
Sorghum LG F: This group corresponded largely to HG 6 of MJ, IND, and SES. The numbers (and percentages) of loci supporting correspondence to sorghum LG F are four (80%) on four LGs in GG, seven (88%) on six LGs in MJ, eight (80%) on seven LGs in IND, three (100%) on four LGs in PIN, and four (57%) in SES. On R570 HG VII, only 2 of the 5 probes were mapped in sorghum; one was on LG D and the other was on LG F. A small HG 6 was assembled in MJ and IND, but there were not enough common markers in GG and PIN to form a HG. Among the exceptions, three loci on SES HG6 were located on sorghum LG B as noted previously. Others are CDSC30 of MJ on LG A, SG305 of GG on LG C, CSU549 of IND on LG G, and CDO87 of MJ on LG I. Seven (25%) of the 28 probes surveyed on LG F were not polymorphic in sugarcane. About 90% of sorghum LG F was mapped in sugarcane, and no rearrangement could be detected.
Sorghum LG G: This group corresponded to HG 1 of GG and MJ, HGs 1 and 15 of IND, HG 1 of SES, and HGs IV and V of R570. The numbers (and percentages) of loci supporting correspondence to sorghum LG G are nine (82%) on 10 LGs in GG, six (86%) on 6 LGs in MJ, five (63%) on 5 LGs in IND, five (100%) on 3 LGs in PIN, four (67%) in SES, and four (100%) in R570. HG 1 of SES spanned two segmental HGs 1 and 15 of IND, suggesting that they are different parts of the same HG. Incongruous loci CDSR67 in IND and SG293 in SES were located on LG A, UMC4 and pSB521 in IND on LG B, and CDSR49 in GG on LG E. Mapped loci on sugarcane linkage groups varied from two to eight. Six (29%) of 21 LG G probes surveyed were not polymorphic. About 60% of sorghum LG D was mapped in sugarcane, and no rearrangement could be detected.
Sorghum LG H: This group corresponded to HG 9 of GG and MJ, and HGs 9 and 16 of IND. No HG in SES corresponded clearly to sorghum LG H—only CDSC24 on HG 2 and CDO365 on HG 5 were located on sorghum LG H. No probe mapped in R570 corresponded to LG H. The numbers (and percentages) of loci supporting correspondence to sorghum LG H are four (100%) on three LGs in GG, five (83%) on six LGs in MJ, five (71%) on five LGs in IND, and two (100%) on three LGs in PIN. With the limited number of homologous loci, HG 9 of GG and MJ covered HGs 9 and 16 of IND, suggesting that they are parts of the same HG. Within these HGs, pSB302 in MJ and CDSR95 in IND were located on sorghum LG E; pSB1108 in IND was on LG J.
A rearrangement was observed in GG where pSB1248 mapped between pSB419 and pSB300, supported by two LGs. Eleven (55%) of the 20 LG H probes surveyed showed no polymorphism in G × I or P × M. Seven of the 11 nonpolymorphic loci were clustered in a small genomic region and homologous loci were detected within a very short genetic distance. About 60% of sorghum LG H was mapped in sugarcane.
Sorghum LG I: This group corresponded to HGs 7 and 17 of PIN, and HG 7 of SES. Although nine LG I loci were genotyped in S. officinarum MJ, none were linked. Only three LGs were formed in GG. In the other species S. spontaneum, seven and five LGs were formed in IND and PIN, respectively. The numbers (and percentages) of loci supporting correspondence to sorghum LG I were four (100%) on three LGs in GG, six (75%) on seven LGs in IND, five (100%) on five LGs in PIN, and five (83%) in SES. Incongruous locus CDSB15 in IND was on LG C, CDSR49 in SES on LG E, and CDSC42 in IND on LG F. Fourteen (54%) of the 26 LG I probes were not polymorphic. About 55% of sorghum LG I was mapped in sugarcane, and no rearrangement could be detected.
Sorghum LG J: This group corresponded to part of HG 10 of GG, HG 10 of IND and PIN, and HG 5 of SES. The numbers (and percentages) of loci supporting correspondence to sorghum LG J were three (100%) on two LGs in GG, four (80%) on four LGs in MJ, six (75%) on seven LGs in IND, and seven (87%) on five LGs in PIN. Incongruous loci pSB289 and pSB95 of MJ were on LGs A and D, respectively, pSB302 of IND on LG E, and pSB240 of PIN on LG H. There were more loci homologous to LG J in S. spontaneum than S. officinarum, a property shared only by LG I. However, homologous groups were established in both species. Fifteen (58%) of the 26 LG J probes showed no polymorphism in populations G × I and P × M. About 75% of sorghum LG J was mapped in sugarcane, and no rearrangement could be detected.
Recombination rate: Disregarding regions where chromosomal rearrangements were found, 25 sets of syntenic chromosomal intervals could be compared for recombination rate in sorghum and sugarcane (Table 6). Among these, 22 pairs showed more recombination in sugarcane. Over all 25 intervals, recombination was 311 cM in sugarcane and 105 cM in sorghum, a highly significant difference (P < 0.0001 by paired t-test). Eleven of these intervals could be compared between S. officinarum (GG and MJ) and S. spontaneum (IND and PIN) with total map units of 92.7 and 106.9 cM, respectively (data not shown), a nonsignificant difference.
Preferential pairing: Chromosome pairing behavior was investigated by reanalysis of each genetic locus, with segregation data recoded in the opposite linkage phase (Al-Janabiet al. 1994a; da Silvaet al. 1995; Grivetet al. 1996). Pairwise comparisons were conducted at each locus with all linked or unlinked SD markers that are putatively borne by corresponding homologous chromosomes using a LOD threshold of 3.0.
In GG, 8 (12.5%) of the 72 LG had markers showing evidence of preferential pairing, including one pair of loci in HG 5, two pairs in HG 8, and five pairs not yet linked to HGs (Table 7, Figure 1). Among eight pairs of markers involved in preferential pairing in GG, three pairs each were detected by the same probes (CSU393, CDSB6, and CDSB73).
Prior reports of preferential pairing in S. robustum (Al-Janabiet al. 1994a) were strongly supported by present data. In MOL5829, 7 (50%) of 14 LG and two pairs of unlinked markers showed evidence of preferential pairing. S. officinarum loci segregating in the same population also showed a high degree of preferential pairing. In LA Purple, 8 (35%) of 23 LGs and one pair of unlinked markers showed preferential pairing.
No preferential pairing was found in S. spontaneum.
Proximal duplication: A total of 47 proximal duplications of individual DNA clones to two or more loci on the same linkage group (or homologous group) were observed. Eight were shared by different Saccharum species, detected by CDSB28 and BNL9.11 for MJ and IND; CDSR133 for GG and IND; CDSR78 for GG and PIN; and CDSB4, CDSB35, CDSB67, and CDSC45 for LP and MOL. There were another 6, 8, 3, 3, 2, and 9 proximal duplications in GG, MJ, IND, PIN, MOL, and LP, respectively.
DISCUSSION
Levels and patterns of chromosome structural rearrangement in Saccharum and Sorghum: Their close relationship, high degree of colinearity, and crosshybridization of DNA probes, all impel use of the small genome of Sorghum to guide molecular mapping and positional cloning in Saccharum (sugarcane), one of the world’s most valuable crops. At least five Saccharum HGs appear to correspond largely if not completely to single Sorghum chromosomes (LGs A, C, G, H, and I). In no case does any HG correspond clearly to more than two LGs, suggesting that no more than four or five major interchromosomal translocations have occurred since the Saccharum-Sorghum divergence (involving homologs of sorghum LGs B, E, and J).
We found only 1 interchromosomal and 2 intrachromosomal rearrangements that distinguished both S. spontaneum and S. officinarum from Sorghum—suggesting that many of the rearrangements found may have occurred after the Saccharum-Sorghum divergence. Chromosome structural polymorphism within the Saccharum genus was suggested by the discovery of 11 rearrangements that distinguished S. spontaneum and S. officinarum. Five of the 11 were supported by evidence from multiple LGs within a HG (Table 3), and thus can be asserted with a high degree of confidence. Finally, 10 and 5 rearrangements were found that distinguished among individual S. officinarum and S. spontaneum linkage groups, respectively. While the greater divergence within S. officinarum is consistent with its higher chromosome number, this last category of possible rearrangement needs to be interpreted with special caution. Independent verification from other LGs is usually not possible, and various artifacts such as non-Mendelian segregation (see above) or simply bands that are difficult to score could account for it.
Summary of chromosomal rearrangements
Some deviations from colinearity in Saccharum and Sorghum may be due to ancient chromosomal duplications (Chittendenet al. 1994; Pereiraet al. 1994) or proximal duplications of individual DNA elements that predate the Saccharum-Sorghum divergence. For example, 4 loci on the Saccharum region homeologous to sorghum LG B were located on LG E (Table 2). Likewise, 2 loci on the region homeologous to LG E were mapped on LG B in sorghum. Recent results (Y.-R. Lin and A. H. Paterson, unpublished results) suggest extensive duplication between sorghum LGs B and E. At least 25 (40%) of the 63 incongruous loci found mapped to chromosomal regions that showed proximal duplication in the sugarcane genome.
Correspondence of Saccharum HGs to Sorghum LGs
A fascinating topic being further investigated is the extent to which genes/quantitative trait loci (QTL) that account for phenotypic variation in sorghum also account for the same traits in sugarcane. Already, there is evidence suggesting that the major determinant of short-day vs. day-neutral flowering of sorghum may also regulate the trait in sugarcane (Patersonet al. 1995; Liuet al. 1997). Correspondence of QTL may add another dimension to the ways in which prior information from sorghum and other grasses can be used to simplify the genetics of sugarcane.
Diploidization of Saccharum chromosomes: Selection for preferential pairing (“diploidization”) is thought to be important in stabilizing the transmission genetics and seed fertility of recently formed polyploids. The extent of diploidization of chromosomes, as reflected by bivalent inheritance (Sreenivasanet al. 1987), is one indicator of the antiquity of autopolyploid formation. In this study, moderately strong preferential pairing was found in S. robustum, adding further evidence in support of the beginnings of disomic inheritance in this species (Al-Janabiet al. 1994a; Mudgeet al. 1996). S. officinarum appears variable in the extent of preferential pairing—showing much preferential pairing when crossed with S. robustum, but little when crossed with S. spontaneum. Preferential pairing was not observed in S. spontaneum SES 208 (da Silvaet al. 1995), IND 81-146, or PIN 84-1; however, some markers of S. spontaneum origin in a hybrid (S. officinarum × S. spontaneum) sugarcane cultivar R570 do show evidence of preferential pairing (Grivetet al. 1996). The stronger preferential pairing of S. robustum is consistent with its proposed antiquity. The lack of preferential pairing in S. spontaneum is consistent with the possibility of a recent origin, also suggested by incongruity in its putative basal chromosome number relative to other Saccharum and Sorghum taxa (see below).
Estimated percentage of Sorghum genome covered by Saccharum LGs
Comparison of recombination rates in Saccharum and Sorghum: With so many chromosomes, Saccharum can generate new multilocus genotypes simply from segregation. The finding that it still showed threefold higher recombination than Sorghum across comparable genomic regions suggests a fitness advantage to creation of new gene combinations in Saccharum. Such a fitness advantage could be one factor that contributed to the evolution of high chromosome number in Saccharum.
Genome coverage of Saccharum maps: Only ∼70% of the sorghum genome is covered collectively by the four newly described sugarcane maps (GG, MJ, IND, PIN), plus the previously published SES208 map (da Silvaet al. 1995), suggesting that the sugarcane map(s) remain incomplete. The R570 map (Grivetet al. 1996) shares only 30 common RFLP probes with our sorghum map; however, even from these limited data it is clear that this map is incomplete as well (Figure 2). The large number of LGs in our maps cannot be explained by “female restitution,” the transmission of the somatic chromosome number in crosses using S. officinarum as female and S. spontaneum as male (cf. Burner 1997). In the cross involving S. officinarum as female in this experiment, the female parent Green German is an intraspecific hybrid with 2n = 97-107. A sampling of F1 progeny from GG × IND has 2n = 73-85, indicating n + n transmission (Burner 1997).
In addition to random factors associated with DNA probe distribution, several other factors may reduce the genome coverage of individual Saccharum maps, such as:
All Saccharum maps made to date are based on “preselection” of DNA polymorphisms that fit simplex segregation ratios (“single-dose restriction fragments” or SDRFs). However, it is well established from a host of genetic mapping data that deviations from Mendelian transmission ratios are common. Such deviations, together with cotransmission of neighboring markers due to genetic linkage, may introduce gaps into SDRF-based maps. For example, in IND, 12 LGs (8 of them with only 2 markers) were formed from 68 markers that showed <50% present bands; 2 short LGs were also detected from 27 such markers in GG.
Levels and patterns of genetic variation across the Saccharum genome vary widely. In regions corresponding to sorghum LGs A, B, C, D, F, and G, Saccharum showed polymorphism for 71-89% of DNA probes surveyed. By contrast, in regions corresponding to sorghum LGs E, H, I, and J, Saccharum showed DNA polymorphism with only 42-47% of probes surveyed. The basis for these marked differences is unknown.
Comparison of map distance in conserved intervals of Saccharum and Sorghum
Gaps in sugarcane HGs often correspond to locations of centromeric DNA in other taxa: Although the sorghum centromeres have not been mapped directly, their possible locations on five LGs could be inferred from centromeres of maize (LGs A, B, C, and F) and wheat (LG D) (Patersonet al. 1995). Curiously, many sugarcane HGs show prominent gaps in the inferred centromeric regions, in marked contrast to the clustering of markers near centromeric regions in most plants. Such gaps separated sugarcane HGs 2 and 12 (MJ), 4 and 11 (GG, MJ, and PIN), and 3 and 8 (GG and IND), corresponding to sorghum LGs A, B, and C, respectively (Figures 1 and 2). We have found DNA markers that showed overlap between sugarcane HGs 2 and 12 (MJ), and 3 and 8 (GG and IND). It remains unclear why sugarcane maps would be marker deficient near the centromere, in contrast to most other organisms that have been mapped.
Basic chromosome number of Saccharum: The basic chromosome number (x) for Saccharum remains uncertain. In the Andropogoneae tribe, x = 10 is common (Whalen 1991) but not without exceptions, such as 2n = 6 and 8 for Iseilema (Clayton and Renvoize 1986). An ancestral chromosome number of x = 5 has been suggested for the Andropogoneae (Celarier 1956; Mehra and Sharma 1975; Clayton and Renvoize 1986) and is strongly supported by evidence from both isozyme (Wendelet al. 1985) and DNA markers (Helentjariset al. 1988) in maize (n = 10). A growing body of evidence also supports ancient duplication of much of the (n = 10) sorghum genome (Chittendenet al. 1994; Pereiraet al. 1994; Linet al. 1995; Paterson et al. 1995, 1996a,b). The basic chromosome number of Erianthus, a close relative of sugarcane, was suggested as x = 10 (D’Hontet al. 1995).
Based on data from 10 sources (J. E. Irvine, unpublished results), chromosome numbers ranging from 2n = 36 to 170 are known in Saccharum. Among 1086 S. spontaneum samples for which chromosome counts were available, a series involving multiples of 8 (2n = 40, 48, 56, 64, 72, 80, 88, 96, 112, 120, and 128) accounts for 77% of the data reported. Among 96 S. robustum samples, 72% are multiples of 10 (2n = 60, 70, 80, 90, 100, 110, 140, and 170). Of the 497 S. officinarum samples 92% are 2n = 80, providing no clue for a basic number of x = 8 or x = 10. Among 95 samples from S. edule, S. barberi, and S. sinense, 28% can be accounted for by a series involving multiples of 10 (2n = 60, 70, 80, 90, 110), but none by a series involving multiples of 8 (excluding 6 samples of 2n = 80).
Markers linked in repulsion phase in Green German
Due to the surprisingly incomplete genome coverage of the Saccharum maps available to date, firm conclusions regarding basal chromosome numbers still remain premature. da Silva et al. (1995) reported a linkage map with 8 homologous groups in S. spontaneum, apparently supporting the series of eight (above); however, we show that HGs 3 and 8 reported by da Silva et al. (1995) appear to join up into a single group. Similarly, Grivet et al. (1996) reported a map with 10 HGs in hybrid R570 (90% S. officinarum), but our comparative data reveal that this map covered no more than 6 of the previously established sugarcane HGs. Our maps appear to suggest a total of 10 HGs for Saccharum. The maps constructed from G × I and P × M detect all the previously known HGs plus 3 new HGs (9, 10, and 11). HG 7 was not detected in S. officinarum, while HG 11 was not found in S. spontaneum (Table 4).
If x = 10 were the ancestral chromosome number for Saccharum, as it appears to be for Sorghum, the most parsimonious explanation of the possible x = 8 set of S. spontaneum would be a recent origin. Saccharum HGs 5 and 6 in S. spontaneum each corresponded to parts of two sorghum LGs (D and J, and B and F, respectively) while the corresponding regions of S. officinarum corresponded only to a single sorghum LG (D and F, respectively). Fusion of ancestral chromosomes would be one possible explanation of both this observation and the n = 8 series of S. spontaneum. A recent origin of S. spontaneum might also explain the fewer rearrangements that distinguished its chromosomes from one another and the absence of preferential pairing of its chromosomes. However, the recent origin of S. spontaneum is contraindicated by its geographic distribution (much wider than S. robustum) and its generally lower levels of polyploidy. Further characterization of the complex genome of Saccharum, with the simple genome of Sorghum as a guide, will help to clarify these and other questions.
Acknowledgments
We thank Ryan Dalrymple, Melanie Marine, and Linghua Zhu for technical assistance, and Zhikang Li, Peter Morrell, and Mark Burow for valuable comments. We appreciate funding from the International Consortium for Sugarcane Biotechnology (A.H.P., J.E.I., M.E.S.), U.S. Department of Agriculture-Agricultural Research Service and Hawaiian Sugar Planters’ Association (P.H.M., K.K.W., T.F.W.), Copersucar Technologies Corporation (D.B., J.D.S., W.B.), U.S. Department of Agriculture National Research Initiative (M.E.S.), and Texas (A.H.P., J.E.I.) and Cornell (M.E.S.) Agricultural Experiment Stations.
Footnotes
-
Communicating editor: M. A. Asmussen
- Received February 5, 1998.
- Accepted September 10, 1998.
- Copyright © 1998 by the Genetics Society of America
