Comparative mapping in the Poaceae family reveals translocations in the complex polyploid genome of sugarcane

Assembly of Saccharum linkage groups into homo(eo)logous groups (HGs)

Alignment to sorghum

The sequence information from the DArT, SNP, RFLP and the EST derived SSR markers were used to align the sugarcane LGs to the sorghum genome sequence (Additional file 1, Table 2). This was achieved by comparing the DNA sequence of the 959 markers with a known sequence to the genome sequence of sorghum using BLASTN homology searching. At the significance threshold of P < 1 × 10⁻²⁰ there were 677 primary correspondences and a further 175 secondary and tertiary correspondences where the sequence aligned to a further one or two sites at P < 1 × 10⁻²⁰ but a lower significance level than the primary correspondence (Table 2). In combination with the sequence information from 677 markers, 282 alleles from SSR, SNP and RFLP loci gave a total of 959 links between the sugarcane genetic map and the sorghum genome. Using these 959 informative markers the majority of the sugarcane linkage groups could be assigned to a single sorghum chromosome with high confidence (P < 1 × 10⁻²⁰) at the macrosyntenic level. As S. spontaneum has a basic chromosome number of x = 8, the 8 HGs of Saccharum are aligned to the ten chromosome sequences of sorghum. A composite map was produced for each HG to increase the number of markers that could be used for alignment between the Saccharum HGs and the sorghum genome. The sugarcane LGs within a HG where possible were condensed into single consensus LGs. This was only possible where the LGs shared two or more markers in common. The consensus LG for each HG was compared with the sorghum genome (Figure 1). In some cases the LGs could not be included as there were not enough markers in common between the LGs within the HG. For example HG6 aligned to Sb9 and 3 using the consensus sequence but when all the LGs are considered parts of LGs also aligned to Sb8 and Sb5 (Table 3). In total over 92% of the sorghum genome was covered by these composite LGs.

Sorghum bicolorchromosomes	Primary correspondences	All correspondences	No. of incongruous loci	No. with no hit to sorghum
Sb 1	109	150	16	15
Sb 2	69	80	16	4
Sb 3	79	101	13	2
Sb 4	98	138	32	10
Sb 5	78	96	6	13
Sb 6	68	85	8	12
Sb 7	44	53	41	13
Sb 8	25	29	5	2
Sb 9	55	62	37	10
Sb 10	52	58	21	6
Total	677	852	195	87

Interchromosomal rearrangements between Saccharum and sorghum chromosomes

Rearrangements between Saccharum and Sorghum chromosomes

Four major rearrangements were identified from syntenic analysis between the sugarcane genetic map and the sorghum genome (Figure 2A-D). Two of these account for the reduction in basic chromosome number from x = 10 in sorghum and S. officinarum to x = 8 in S. spontaneum. The complicated nature of the hybrid sugarcane genome means that only a few of the LGs within a HG have the translocation, for example in HG2, 7 of the LGs display the translocation, the other LGs align only to one of the two chromosomes (Additional file 1). This recombination results in the two sorghum chromosomes combining into one sugarcane HG. This translocation event reduces the basic chromosome number of sugarcane (Figure 2A). The other translocation event is in HG8 which aligns to Sb8 and Sb2 and again reduces the chromosome number to its final basic chromosome number of x = 8 (Figure 2B). The other major rearrangements had no effect on basic chromosome number but involved rearrangements between chromosomes. HG6 aligns in the main to Sb9 but one LG contains a significant translocation from Sb3 covering approximately 50% of Sb3 (Table 3, Figure 2C) The last rearrangement is in HG5 which aligns to Sb7 and Sb5 (Figure 2D); this appears to be a simple translocation between Sb7 and half of Sb5 and is present in 5 LGs within this HG (Additional file 1).

Analysis of the DArT sequences

In total, 7846 DArT clones were sequenced and a proportion of these were also used to generate the genetic map. These sequences produced a total sequence length of 3,602,170 bp with an average length of 465 bp. These DArT sequences were assembled into a total of 5278 sequences, comprised of 1501 contigs and 3777 singletons. Of these 5278 unique sequences, 1750 hit against the Saccharum EST dataset, 978 to the S. bicolor EST dataset, 1226 to the Z. mays EST dataset and 825 to the O. sativa EST dataset. A Venn diagram representing the assembled DArT sequences that aligned to each of the relevant datasets is displayed in Figure 3. Table 4 contains the number of ESTs in each dataset from the different genomes and the total number of assembled DArT sequences that aligned to each dataset.

Genome sequence comparison

A results table of the DArT sequence comparison against the genomes of S. bicolor, Z. mays, and O. sativa is provided in the Additional file 2. A total of 1755 assembled DArT sequences aligned to an annotated gene in the S. bicolor genome, while a similar number (1783) did not align to the genome at all. A small proportion of the DArT sequences aligned to repeats (Figure 4). The categories of repeats against which the DArT sequences aligned on the S. bicolor genome are displayed in Figure 5.

In total, 2124 assembled DArT sequences which hit against the S. bicolor genome sequence also hit against the Z. mays genome sequence, of which 1888 (88.9%) aligned to regions syntenic between these two species [[31]]. In contrast, only 846 assembled DArT sequences hitting against the S. bicolor genome also hit against the O. sativa genome sequence, of which only 671 (79.3%) aligned in regions syntenic between these two species. The ratios of syntenic to non-syntenic region alignments for each category of annotation are displayed in Figures 6 and 7. In the case of both Z. mays and O. sativa, DArT sequences aligning to annotated genes most frequently aligned to syntenic regions (92.2% and 83.0% respectively).

Comparison of the sugarcane genetic map to the sorghum genome

To date only low resolution maps have been used to compare the sugarcane genetic map to the sorghum genome [[12],[32],[33]]. These maps were constructed with low-copy RFLP markers that were selected for their ability to provide a signal in cross-hybridisation thereby limiting the ability to assess orthologous and paralogous relationships in gene families, since comparative mapping by RFLP often identifies paralogous rather than orthologous sequences [[29]]. The large number of markers with known sequence in this study made it possible for the first time to align a significant part of the sugarcane genetic map to the sorghum genome sequence (Table 3, Additional file 1). Although with 954 sequences this analysis is at the macro level and only the major structural changes are identified, at the micro level there could be many more rearrangements but more markers and sequence information is needed to verify these. To use as many markers as possible in the comparison to sorghum a composite linkage map was generated where possible for each HG. This was then aligned to the sorghum genome sequence (Figure 1). Six HGs aligned to one sorghum chromosome and four HGs aligned in the main to two sorghum chromosomes. For all chromosomes there were regions that appeared to contain markers from other chromosomes (Additional file 1, Table 2) but to be considered for further analysis a number of linked markers from the same chromosomes showing homology to gene sequences had to be identified. This left four regions that showed strong likelihood of being due to rearrangement. A major translocation was identified in HG8, where LG3 aligned to Sb8 and Sb2. This result agrees with previous attempts to align sugarcane to sorghum [[32],[33]] although the greater number of markers in this map means a higher level of resolution of the alignment. Two sugarcane populations were studied by [[12]] and they determined that no more than four or five major interchromosomal translocations had occurred since the Saccharum-Sorghum divergence. They also identified a translocation between Sb2 and Sb8 in both S. spontaneum parents and the hybrid parents of their populations. They identified additional possible translocations between sorghum chromosomes 8 and 7 in the hybrid parent and 6 and 7 and 2 and 4 in the S. spontaneum parent, these were not verified in this study. They did not identify the Sb5/Sb6 translocation that was identified in this analysis and by [[32],[33]]. These two major translocations resulted in both cases in the condensation of two chromosomes into one. This process is known in many lineages and is demonstrated within the grass family where all grasses have evolved from an ancestral genome of 12 chromosomes [[29],[34]]. A further 2 HGs contained translocated chromosomes although these did not result in reduced chromosome number. HG5 aligned in the main to Sb7 but 5 of the 13 LGs also aligned to Sb5 indicating a translocation between Sb7 and Sb5. An interchromosomal rearrangement between LGs that aligned to Sb7 and Sb5 was also identified by [[12]] in the hybrid parent of the cross studied. The additional interchromosomal rearrangement in HG6 LG20 which aligns to Sb9 and Sb3 has not previously been identified (Table 3, Additional file 1).

Translocations

A more detailed comparison of the four major interchromosomal rearrangements (Figure 2A-D) revealed that in some cases the rearrangement of the grass ancestral genome that has occurred in maize is the same as in sugarcane [[29]]. The chromosomal rearrangement seen in HG6 on LG20, which aligns to Sb9 and part of Sb3, corresponds to maize chromosome 8 (Figure 2C). Maize diverged from sorghum approximately 12 million years ago (mya) followed by allopolyploidisation approximately 5 mya [[29]]. The present maize chromosomes are a result of numerous rearrangements which reduced the chromosome number from 20 to 10 [[29]]. Recent work by [[35]] has shown that the whole genome duplication (WGD) which occurred in the ancestor of both maize and sorghum 29 mya has resulted in large duplicated blocks in their genomes. Sb9 of sorghum shares a duplicated block with Sb3 these two chromosomes have recombined to form maize m8. The same rearrangement has occurred in sugarcane as seen in LG20 in HG6 (Figure 2C). The translocation identified in HG2 which aligns to Sb5 and Sb6 is also seen in the top of maize m2 (Figure 2A). Parts of HG2 also align to m10 and m4 which are the duplicated ancestral regions [[26]]. The rearrangement in HG5 which aligns to Sb7 and Sb5 again is present in maize and aligns to m4 (Figure 2D). HG8 aligns to sorghum Sb8 and Sb2 which also aligns to maize m1 and m7. Interestingly the ancestral fusion event in Sb2 is retained in both maize and sugarcane although a further rearrangement of this chromosome has occurred uniquely in sugarcane (Figure 2C).

Saccharum spontaneum

Sugarcane being a polyploid, is more tolerant of gene deletions and rearrangements than a diploid. It would appear that the rearrangements in HG2 and HG8 are major chromosome rearrangements to account for the smaller number of basic chromosomes in S. spontaneum. A sparse RFLP map was generated by [[10]] of a S. spontaneum clone SES208 using 234 single dose RFLP markers. From comparative mapping of the RFLP clones to sorghum and maize, HG4 of this map contains clones UMC130 and UMC102 which align to Sb8 and SG298 and UMC139 which align to Sb2 and are on the same LG. This is comparable to the Q165 HG8 which aligns to Sb8 and Sb2. Similarly RFLP markers in HG2 of SES208 align to Sb5 and Sb6 as does the Q165 HG2. The translocations in HG2 and HG8 appear to be inherited from the S. spontaneum part of the genome of Q165 but with the current data it is not possible to distinguish S. spontaneum chromosome from S. officinarum chromosomes. However, as S. officinarum is x = 10, the assumption is that at least the two rearrangements in HG2 and HG8 are inherited from the S. spontaneum part of the genome. The other rearrangements in HG5 and HG6 could be inherited from either or both species.

Comparison within the Poaceae

Recently a detailed genetic map was constructed for another member of the Andropogoneae: Poaceae, Miscanthus sinensis. This species has a diploid inheritance patterns (2 n = 38) and when aligned to sorghum contains two sub-genomes, each syntenic to the entire sorghum genome with one major structural rearrangement which resulted in the condensation of two chromosomes into one and a reduction in chromosome number to x = 19 [[36]]. In this case the translocated LG aligned to Sb7 and Sb4 a translocation not seen in Saccharum but again seen in maize on m4. Interestingly [[36]] found that the M. sinensis LGs that were syntenic to Sb5 had the lowest percentage of markers that mapped to sorghum, a similar finding to switchgrass. The data suggests that linkages syntenic to Sb5 have encountered more evolutionary changes than other chromosomes after the species diverged from each other. Similarly HG2 in Sacharrum which is syntenic to Sb5 has undergone more rearrangement than other HGs.

Recent work by [[37]] comparing sequenced genomes of grasses has demonstrated that the chromosome number variation/reduction from the n = 12 common paleo-ancestor to the n = 10 of sorghum could be due to the insertion of a chromosome in the centromeric region of another chromosome although these reductions could also come about through pericentric inversions followed by reciprocal translocations [[38]]. The comparison of the Q165 Saccharum genetic map to the sorghum and maize genomes demonstrates Saccharum contains the same two ancestral fusion events. Moreover, out of the 4 major translocations identified, 3 of them have also occurred in maize (Figures 2 and 8), one of which resulted in a reduction in chromosome number. S. officinarum is an octoploid and S. spontaneum ranges from 2n = 5x = 40 to 16x = 128 [[39],[40]] in chromosome number. It would seem probable that after the whole genome duplication that also occurred in maize the Saccharum ancestor underwent some of the same rearrangements as maize then diverged and under went further WGD events. More sequence analysis is needed to determine how these events occurred but in evolutionary terms they would have been recent (less than 5 mya).

Whole genome duplication and the rapid and massive structural changes which have occurred in the Poaceae and to a greater extent in the Saccharum might explain the ability of these polyploids to adapt and survive environmental conditions that are not tolerated by their diploid ancestors. Within the Saccharum there are no known diploids as all species are polyploid. S. spontaneum has the widest distribution from Japan and Indonesia through the Indian subcontinent to the Mediterranean and Africa [[39]] and the largest range of ploidy levels (2n = 40-128). It has been reported that polyploidy provides increased vigour [[41]] and provided an advantage for colonizing a wider range of environments [[42]].

Dart sequence comparison to EST and genome sequences

It is to be expected that the highest number of DArTs would align to ESTs from Saccharum as the genome of origin. Interestingly, a higher number of DArTs align to the maize EST dataset (1226) than to the sorghum (978) EST dataset, even though sorghum is generally thought to be a much closer relative to sugarcane [[25]] (Figure 3). This may be attributable to the higher volume of repetitive DNA in the maize genome [[25],[43]], some of which may be expressed and represented in the EST database but missing from the sorghum EST database which is much smaller than the maize database. As would be expected, the least number of DArT sequences aligned to the rice EST dataset, which is the most distant relative of the species surveyed.

A proportionally low number of DArT sequences aligned to annotated repeats (7.7%) in the sorghum genome compared to the number of DArTs aligned to annotated genes (35.1%), given that 55% of the sorghum genome is known to be repetitive and the repetitive content of sugarcane has been demonstrated to be similar to that of sorghum [[25],[44]]. This suggests that the enrichment for unique regions applied in the production of the DArT markers has largely succeeded. Interestingly, 1783 (33.8%) of the DArT marker sequences can be observed in Figure 2 to display no significant alignment to the sorghum genome, suggesting that these originate from divergent regions of sugarcane.

Plant material

The mapping population used to generate the Q165 genetic map consisted of 227 progeny derived from a cross between a S. officinarum clone IJ76-514 (2n = 80) as the female parent and Q165 (2n = 115) an Australian sugarcane cultivar and elite parent as the male parent.

Sequencing of the DArT clones

A total of 7846 DArT clones, the majority of which were present on the DArT array used to genotype the mapping populations were isolated and DNA purified using QIAGEN plasmid purification kit. Each of these clones were sequenced using the ABI PRISM BigDye Terminator Cycle Sequencing Ready reaction kit (Perkin-Elmer, Wellesley, MA) and analyzed on an ABI 3700 sequencer. Sequencing was done with both M13 Forward and Reverse primers. The sequences were analyzed using the Lasergene software program Seqman (DNAstar Inc., Madison, WI). The vector and adapter sequences were removed and the original restriction sites restored. The DArT sequences can be obtained on request from the corresponding author.

DArT sequence clustering and analysis

The 7846 raw DArT sequences were assembled using Sequencher with a minimum overlap length of 50 nucleotides. The EST sequences available in the NCBI EST database as at 27th October 2011 were downloaded to produce four species-specific EST datasets for Saccharum, Sorghum bicolor, Zea mays, and Oryza sativa. For each DArT contig, coding DNA sequences (CDS) were predicted by comparing the DArT sequence to these databases using TBLASTX [[45]] with an e-value cutoff of ≤10⁻²⁰, and the best hit for each DArT sequence identified by e-value.

Comparison with public genome sequences

Genome sequences for S. bicolor, Z. mays and O. sativa as well as the gene and RepeatMasker annotations for S. bicolor were downloaded from Phytozome version 7.0 [[46]]. The 5278 assembled DArT sequences were compared to each genome sequence using BLASTN [[40]] with an e-value cutoff of ≤10⁻²⁰, identifying the best hit for each DArT sequence along with the chromosomal location of these hits.

We compared the position of each top DArT marker hit on the S. bicolor genome sequence against the annotations for genes and repeats using a custom Python script to identify whether a DArT sequence could be ‘annotated’ as one of four categories; aligned to an annotated gene, aligned to an annotated repeat, aligned on the boundary between an annotated repeat and an annotated gene, or in a region without an annotation. Assembled DArT sequences without a BLAST hit against the S. bicolor genome were also recorded and DArT hits against repeats were categorised.

We then compared the position of hits against the S. bicolor genome with the position of hits against the genomes of Z. mays, and O. sativa to identify whether DArT sequences hit against syntenic regions in these genomes based on the syntenic regions identified by [[47]].

Marker analysis and genotyping

The DNA extraction and marker data were generated as in [[15]]. This consisted of the original data from [[15]] to which were added further AFLP, SSR, RFLP, SNP and DArT markers. Generation of further SSR, and Pst I/Mse I AFLP marker data were as reported by [[15]]. The SNP data was generated as in [[48]] and RFLP data as in [[49]]. The DArT data was generated as in [[23]]. The genetic map was produced using 2283 simplex markers that were present as a single copy in Q165, the map is reported in detail in [[24]]. Composite LGs were generated using JoinMap V4 [[50]].

Comparison to the sorghum genome

Sequence information from the DArT, SNP and EST - SSR markers were used to align the Q165 linkage map to the genome sequence of Sorghum bicolour. This was carried out using BLASTN homology search; sequences had to show at least 80% nucleotide identity for at least 50% of the marker sequence length with an e-value cutoff of ≤10⁻²⁰ to be used for comparative analysis. The results were visualised using Sigmaplot. This high stringency was used as it is difficult to infer orthologous (derived from a common ancestor by speciation) and paralogous (derived by duplication within one genome) relationships from sequence comparisons using a less stringent e-value.