Preexisting fixed rearrangements in Brassica parental genotypes

Brassica genotypes used to produce the carirapa and NCJ allohexaploids were analyzed for fixed rearrangements in the genome. The parental genotypes used to produce the carirapa hexaploids, B. carinata accession “03,949,” and the B. rapa genotypes “Ankangzhong” and “WulitianYC” did not have any detectable fixed rearrangement events in their corresponding genomes. In the genotype “03949,” we did observe a segregating event at the top of B01, where a single copy was missing between 0 and 13.6 Mb. In the case of the genotypes used in the crossing of NCJ hexaploids, we only observed fixed events in the B. napus genotypes. In the case of the B. napus genotype “Surpass400_024DH” (N1), we detected deletions in chromosomes A01 and A04, located at 2.2–4.7 Mb and 20.7–23.3 Mb, respectively. In the C genome of “Surpass400_024DH,” we identified a deletion at the top of chromosome C01, located between 1.4 and 3.3 Mb. We also found two other deletion events located in chromosome C09, at positions 0–2 and 52.9–60.1 Mb, respectively. For the “Ag-Spectrum” genotype (N6), we observed a deletion on chromosome C02, located between 8–10.7 Mb.

Chromosome inheritance and fixed karyotype changes in NCJ and carirapa allohexaploid types

From the available NCJ and carirapa collection, the most advanced and fertile genotype combinations were selected. From the NCJ type, the genotypes selected were N1C2.J1 and N6C2.J2, and from the carirapa type, C13, and C21. The carirapa plants were not fully homozygous, suggesting that pollen cross-contamination with other genotypes may have occurred during propagation in the field. In the NCJ-type hybrids, residual heterozygosity was present based on the method of producing these hybrids, but no cross-contamination was indicated by the presence of alleles other than those from the parent genotypes. In total, six plants per allohexaploid type were grown and later on crossed to produce F1 hybrids between the lines (Fig. 1c).

First, the relative chromosome number for the 12 allohexaploid parental plants was determined using SNP data and Log2R ratio values (Supplementary Table 1). For the “carirapa” type, the chromosome numbers were 51 (1 plant), 52 (2 plants), 53 (1 plant), and 54 (2 plants). In the case of the allohexaploid NCJ, the chromosome numbers observed were 50 (1 plant), 51 (1 plant), 52 (1 plant), 53 (1 plant), 54 (1 plant), and 55 (1 plant). The two carirapa plants with 54 chromosomes were aneuploids in terms of genome composition. In contrast, the NCJ plant with 54 chromosomes was an euploid individual (genotype combination N1C2.J1). Overall, in both hexaploid types, the chromosomes A05, A06, A08, A10, C03, C05, C07, B01, B02, B04, B06, and B08 showed no changes (Supplementary Fig. 1). Specifically, in the carirapa plants, no B chromosome changes were observed. Changes in B chromosomes were, however observed in the NCJ plants for chromosomes B03 (two plants), B05 (two plants), and B07 (one plant). For both allohexaploid types, most of the chromosome number changes were observed in the A01–A03, C01, and C02 chromosomes. Also, the chromosomes A01, B05, and B07 were both lost in one carirapa and two NCJ plants, respectively.

As the NCJ and carirapa allohexaploids have been independently selected by fertility (total seed number) through several generations, we analyzed the presence of fixed rearrangements (rearrangement present in both homologous chromosomes) in all the three genomes. To be able to have a better overview and frequency of all the fixed events, we combined them into one figure (Fig. 3). A single fixed event involving the B genome was detected putatively between B01/A04, where the B01 segment was lost (deletion) and replaced by the end of A04 segment (duplication) for this homologous region. The end of A04 (~ 3 Mb) was frequently lost, as was the case in seven out of 12 allohexaploid parents (four carirapa and three NCJ plants). However, this deletion event was already present in the B. napus cv. “Surpass400_024DH” used in the cross and was inherited and fixed in the three NCJ plants analyzed. In the case of the carirapa plant C21-1 that had this deletion at the end of A04, the region was most likely replaced by the end of chromosome C04, but evidence of this was inconclusive in the NCJ plants analyzed.

Fig. 3

Karyotype of fixed events in Brassica allohexaploids NCJ (B. napus × B. carinata × B. juncea) and carirapa (B. carinata × B. rapa) type. The fixed partial or whole chromosome events are colored according to the legend for each of the three Brassica genomes

For chromosomes B05 and B07, both copies were missing in the plants belonging to the two independent N6C2.J2 lineages. Another region that was frequently lost was located at the top of chromosome C03, ranging from approximately 0–1.2 Mb in size: this region was missing in all four carirapa C13 parents (representing four plants of one lineage). The three plants from the genotype combination N1C2.J1 (one lineage) had in common a duplication/deletion event involving an extra copy of the top of chromosome A01 (~ 2 Mb) being putatively translocated into chromosome C01. The deletion of C01 was already present in the Brassica napus cv. “Surpass400_024DH” (N1) used to produce the hexaploid, unlike the duplication of A01, which was a new rearrangement event. In some regions, we found an overlap of events (duplications and deletions) between the genotypes. The region at the end of chromosome A01 was duplicated in two carirapa C13 parents (one lineage), deleted in one carirapa parent (C21), and deleted in one NCJ parent (N6C2.J2). In the homologous region, there was no detectable duplication for any of the plants, hence it was more likely to be only a deletion event. Another region of overlap was found for putative homologous translocations between chromosomes A10 and C09. The end of A10 had three duplication events (N1C2.J1) and one deletion event (C13), while C09 had one duplication (C13) and four deletions located at the end of C09 at ~ 53–60 Mb in three N1C2.J1 and at ~ 52–55 Mb for one C13 parent. For the NCJ plants from the genotype combination N1C2.J1, the putative translocation between C09/A10 was already present in the B. napus cv. “Surpass400_024DH” and was inherited and fixed in the hexaploids. The rest of the events observed in the allohexaploids NCJ and carirapa were more parent-independent (occurred in only one plant). In total, we observed 50 fixed deletion events (32 carirapa, 18 NCJ) and 33 fixed duplication events (23 carirapa and 10 NCJ) involving the A, B, and C genomes of Brassica hexaploids. The fixed events observed in the carirapa hexaploids were not identified in either the B. carinata or the B. rapa accessions used as parents.

Crossing NCJ and carirapa allohexaploids: F1 hybrid and allele segregation

Seven out of eight F1 hybrids were the result of a cross between carirapa and NCJ hexaploids, as expected (Table 1); one hybrid appeared to result from an outcross to an unknown paternal plant. The second step was to analyze the cross between the F1 hybrid and the test-cross parent. To do this, the same approach as in the F1 hybrid was used. In this case, populations 1 and 3 had only 7 and 5 individuals that corresponded to true test-cross progeny, and the remaining individuals resulted from unintended self-pollination to produce F2 progeny seeds. The other five populations were used to analyze allele segregation from the F1 hybrid.

Table 1 Brassica allohexaploid populations. Genotypes crossed between carirapa (C13 and C21, generations H5–H7) and NCJ (N1C2.J1 and N6C2.J2, generations H3–H5) allohexaploid types. Plant number is also specified after the genotype code. Test-cross progeny correspond to the total number of progeny that were determined to comprise true test-cross progeny between the F1 and the test-cross parent. *Populations removed from further analysesAllele segregation per populationPopulation 2

The F1 hybrid was the result of a combination between the genotypes N1C2.J1 × C13. This hybrid had 52 chromosomes distributed between the A genome (19), B genome (16), and C genome (17) (Fig. 4a). In the A genome, the chromosomes A01 and A05 were present in only a single copy, with just the chromosome from the carirapa parent present. Chromosome A01 was already a single chromosome in the NCJ parent, unlike A05, where both copies were present in the parent. The other chromosomes (A02–A09) were present in two copies, while chromosome A10 had at least one extra copy from the carirapa parent. In the A genome, putative nonreciprocal translocations between A05/C04, A05/C05, A09/C09, and A10/C09 were present in the chromosomes inherited from the carirapa parent. No translocations were observed in A-genome chromosomes from the NCJ parent. The bottom of chromosome A04 from NCJ was missing 2.5 Mb, corresponding to a known deletion present in the B. napus cv. “Surpass400_024DH” used as a parent in the crossing. The top of chromosome A09 from the NCJ parent was missing ~ 3 Mb that corresponded partially to an extended deletion that was initially inherited from a deletion also present in the B. napus cv. “Surpass400_024DH”. For the B genome, the F1 hybrid contained the expected 16 chromosomes with no translocations or deleted regions detected. In the C genome, the chromosomes C01–C08 were correctly inherited from the carirapa parent. In the case of chromosome C09, just a portion (~ 5 Mb) from the end of the chromosome was present as a translocated region into chromosome A10. Part of the top of chromosome C03 (~ 7 Mb) was lost from the carirapa parent. The chromosomes C01–C09 were correctly inherited from the NCJ parent, despite the presence of only one copy of chromosome C02 in the parent. Putative translocations were observed in chromosomes coming from both parents. In the case of the carirapa parent chromosomes, the translocations present in the F1 were C01/A01 and C02/A02, and in the NCJ parent chromosomes there were C01/A01, C05/A05, C06/A07, and C09/A10. From the above translocations, only two corresponded to putative de novo events identified in the F1 hybrid.

Fig. 4

a Molecular karyotype and b allele segregation for F1 hybrid population 2. Chromosomes are colored based on hexaploid parent: carirapa in purple, NCJ in orange. Rearrangements are colored in blue. De novo translocations (present in the F1 hybrid but not in the parents) are marked with a star in a different color depending on the type (see legend). Chromosome sizes are represented in megabases (Mb). Expected segregation ratio of the alleles (50%) is marked with a red dotted line for each chromosome. Significant allele distortion (X2 test, p < 0.05) is indicated with dark orange (NCJ) or dark purple (carirapa). Seg., segregation

Allele inheritance from the F1 hybrid into the test-cross progeny was also analyzed (Fig. 4b). Segregation distortion was established if the observed allele segregation ratio was significantly different from the expected (X2 test, p < 0.05). This test was performed independently for each of the parents of the F1 hybrid. In population 2, 18 test-cross progeny individuals were used to assess allele segregation (for more details, see materials and methods). The allele segregation from the A01 and C02 chromosomes could not be established due to the lack of enough polymorphic alleles. The segregation distortion observed in the chromosomes A05 and C09 corresponded directly to the presence of only a single chromosome in the carirapa or NCJ parent, respectively. The allele segregation distortion observed in A07, A09, A10, and C05 corresponded to rearrangements present in the F1 hybrid. Allele segregation distortion not directly explained by the karyotype of the F1 parent was observed in chromosomes A02 (extension of a known duplication present in the F1 hybrid, carirapa), A03 (although few markers were represented, carirapa), C01 (NCJ), C04 (NCJ), and B02 (NCJ). Potentially, these events correspond to de novo rearrangements produced during meiosis in the F1.

Population 4

The F1 hybrid resulting from the combination between NCJ N6C2.J2 × carirapa C13 had 51 chromosomes distributed between the A (21), B (15), and C (15) genomes (Supplementary Fig. 2a). In the F1 hybrid, chromosomes A02 and A07 had at least one extra copy from the carirapa and from the NCJ parent, respectively. These chromosomes were also doubled in the corresponding allohexaploid parent. Chromosome A04 was present as a single copy in the F1 hybrid, inherited only from the NCJ parent. In the carirapa parent, chromosome A04 was present as a single copy, and it was not inherited from the F1. In F1, the chromosome A05 inherited from the carirapa parent had a small region missing (1.6 Mb) located at chromosome position ~ 24 Mb, and no potential candidate translocation corresponding to this region was identified. In the B genome, the F1 hybrid had only one B07 chromosome (carirapa origin). Chromosome B07 was completely absent in the NCJ parent. In the B03 chromosome of the F1 hybrid (NCJ origin), a duplicated region was identified, but it was not possible to determine the position in the genome of this extra copy (colored gray, see Supplementary Fig. 2a).

In the C genome, chromosomes C01, C02, and C06 were present in single copies (NCJ). The chromosomes C01 and C02 were also present as a single copy in the carirapa parent, unlike C06, which was present as two copies in the carirapa parent and as a single copy in the NCJ parent. In the A genome, putative translocations between homoeologs A01/C01, A02/C02, A07/C07, and A09/C08 were observed. In the B genome just one putative translocation between B01/A04 (NCJ) was observed. In the C genome, we had putative translocations involving C02/A02, C04/A04, C05/B01, and C07/B02.

In the analysis of the allele segregation, most significant distortions were explained by the translocations described above (Supplementary Fig. 2b). The exceptions were present in chromosomes A02 (carirapa), A07 (carirapa), A10 (carirapa), C06 (NCJ), B01 (carirapa), and B02 (NCJ). In addition, the single B07 chromosome in the F1 hybrid was expected to be present in 50% of the test-cross population, but was only present in 18% of the individuals (3 out of 17).

Population 5

The F1 hybrid (N6C2.J2 × C13) had 54 chromosomes distributed between the A (21), B (15), and C (18) genomes (Supplementary Fig. 3a). Chromosomes with at least one extra copy were present for A04 (NCJ) and C03 (NCJ). Chromosome A04 was already doubled in the NCJ parent, unlike chromosome C03, which resulted from a new chromosome duplication event. Single copies were observed for chromosomes B05 (carirapa) and C09 (NCJ). In the case of B05, both copies were missing in the NCJ parent. Chromosome C09 was present as a single copy in the carirapa parent, and did not get inherited into the F1 hybrid. In the chromosomes from NCJ, putative translocations between A01/C01, A03/C03, A04/C04, B01/A05, C03/A03, and C08/A09 were found. In the case of chromosomes coming from the carirapa parent, we observed translocations involving the chromosomes A02/C02, A05/C04, A09/C09, C01/A01, C02/A10, and C03/A03 (Supplementary Fig. 3a).

In the population segregating for alleles from the F1 hybrid, most of the allelic distortion was explained by CNV events in the parent F1 (Supplementary Fig. 3b). The exceptions to this (putatively novel CNV events) were located in A02 (carirapa), A08 (NCJ and carirapa), B03 (NCJ), B04 (NCJ), B07 (NCJ), C02 (NCJ), C04 (NCJ and carirapa), and C08 (NCJ). Chromosome C09, as mentioned before, had just one copy in the F1 hybrid, and it was present in fewer test-cross individuals than expected (25% presence vs. 50% expected). In the case of B05 as a single chromosome, only a few polymorphic alleles were present with which to make a proper comparison, but it also seemed to be present less often than expected (25% presence vs. 50% expected).

Population 6

The F1 hybrid in this population was a cross between N6C2.J2 × C13 and had 51 chromosomes distributed between the A (19), B (15), and C (17) genomes (Supplementary Fig. 4a). Single-copy chromosomes were A04, B05, and C08. Chromosome B05 was already missing both copies in the NCJ parent. In the case of A04 and C08, both copies were present in the corresponding carirapa and NCJ parents, respectively. No extra chromosomes were observed. Putative translocations between A01/C01 and A09/C08 were detected in NCJ chromosomes. In the case of the carirapa chromosomes, observed possible translocations were located between A09/C09-C08, C01/A01, and C02/A02.

When analyzing the allele segregation for alleles from the F1 hybrid (Supplementary Fig. 4b), most of the distortion was explained by rearrangement events, with the exception of the following: end of A02 (carirapa), A03 (carirapa and NCJ), end of chromosome A04 (NCJ), A09 (carirapa), C01 (NCJ), C04 (NCJ), and C09 (NCJ). The A09 chromosome was a special case, as it was present in two copies in the F1 hybrid, but in the test-cross population, the A09 of carirapa origin was present at a higher frequency than expected (expected 50%, observed 86.6%). Interestingly, this chromosome had two translocations involving chromosomes C9 and C8. The single chromosomes A04 and C08 segregated as expected in the population, unlike B05, which was present in only two out of the 15 test-cross population individuals and not in half of them, as expected. No other changes were observed in the B genome.

Population 8

The cross between N1C2.J1 (euploid, 2n = 54) × C21 gave rise to an F1 hybrid with 52 chromosomes distributed between the A (19), B (16), and C (17) genomes (Supplementary Fig. 5a). Single chromosomes were observed for A03 and C01. Both chromosomes were present as two copies in the corresponding NCJ and carirapa parents. No chromosomes were doubled. In the NCJ chromosomes, we observed potential translocations between A02/C02, C01/A01, C06/A07, and C09/A10. Chromosomes A04 and A09 inherited from the NCJ parent had a deletion at the bottom and at the top of the chromosome, respectively. These deletions were already present in the parents, and we did not observe a duplicated homologous region that could have replaced this fragment in the F1 hybrid. In the chromosomes from the carirapa parent, we observed putative translocations between A01/C01, C02/A02, C03/A03, and C09/A09.

In the allele segregation from the F1 hybrid (Supplementary Fig. 5b), most of the events were explained by the rearrangements, with the exception of areas on chromosomes A03 (NCJ), A06 (carirapa), B02 (NCJ), B06 (NCJ), B08 (NCJ and carirapa), C08 (NCJ), and C09 (NCJ).

De novo rearrangements and inheritance in the F1 hybrids

Overall, 50 new rearrangement events (22 duplications and 28 deletions) were observed in the F1 hybrids. Out of these events, 28% (6 duplications and 8 deletions) were triggered by a previous event nearby or overlapping the chromosomic location of the new event already present in the hexaploid parent. On average, there were 10 new events per population, affecting mostly the A (52%) and C genome (44%). Both parents contributed almost equally to the de novo rearrangements observed in the F1 hybrids, with the exception of population 5, where the NCJ parent contributed to 10 rearrangements compared to 4 coming from the carirapa parent. As mentioned before, the least affected genome was the B, where just two events in two populations were observed. In the A genome, the chromosome most affected by rearrangements was A09, with a total of 8 events (2 duplications and 6 deletions). In the case of the C genome, the chromosome with more de novo rearrangements was C02, with six events (2 duplications and 4 deletions). We also observed nine de novo events involving whole chromosomes, where six chromosomes were lost, and three were present in an extra copy in the F1 hybrids.

In the F1 hybrids, we also observed that some of the rearrangement events present in the parental hexaploid plants were either inherited in the same size or reduced in size due to cross-overs. In total, 72 rearrangements were inherited from the hexaploid parental plants, with an average of 14.4 rearrangement events per F1 hybrid. Out of the 72 rearrangements, 8.3% had a reduction in size due to a cross-over. Most of the events inherited from the parents corresponded to deletions (66.7%), with seven events involving the loss of a chromosome, affecting mostly the C genome (27 events).

When analyzing the putatively de novo events produced by the F1 hybrids, we observed a total of 43 events (8.6 events on average per population). Out of these events in the F1 hybrids, 25 events were potential duplications and 18 deletions. Interestingly, many of these events also affected the B genome (9 events total, 8 from NCJ and 1 from carirapa origin, respectively). Overall, more events were produced by one meiosis in the F1 hybrid than by meioses coming from the grandparents, although the difference was only significant between the carirapa parent and the F1s (Fig. 5, one way-ANOVA, p < 0.05).

Fig. 5

De novo rearrangements produced by meiosis in Brassica hexaploids carirapa (derived from B. carinata crossed with B. rapa), NCJ (derived from crosses between B. napus, B. carinata, and B. juncea) and their F1 hybrids. Per group, there are 5 different meiosis events representing individual gametes produced from the parents and the F1 (red: carirapa, green: NCJ and the F1 hybrid between them: blue). The number of de novo rearrangements (occurring during this specific meiosis and not inherited from a previous meiotic event in the lineage) for each of these five meiotic events for the two parents and their hybrid is indicated as a dot. The average per group is drawn with a black triangle. Significant differences are indicated with letters (Tukey’s HSD, p < 0.05): groups represented with a different letter (a, b) are significantly different, while groups that have a letter in common are not significantly different (a, ab and b, ab)