| Interspecific hybridization and allopolyploidization contribute to the origin of many important crops. The cultivated Brassica species comprise three diploids(B. rapa, B. nigra and B. oleracea) and three allotetraploids (B.juncea, B. napus and B. carinata) evolving from pair crosses of three diploid, which are important edible oil, vegetable and feed crops. Synthetic Brassica allotetraploids have become one of the most widely used models to study the genetic recombination and "fixed heterosis" in allopolyploids, and are important germplasm resources. In this study, the digenomic diploid and triploid hybrids (A.B and BB.A, A.C and CC.A, B.C and CC.B) and trigenomic triploid hybrids (A.C.B and C. A.B) were produced from the reciprocal crosses of three Brassica diploids. The chromosomes in the resultant dihaploids were doubled to obtain three allotetraploids (B. juncea, AABB; B. napus, AACC; B. carinata, BBCC). The intra-and intergenomic chromosome pairing, genetic, epigenetic, gene expression changes and rRNA gene expression in these hybrids and allotetraploids were assessed by genomic in situ hybridization and BAC-FISH, amplified fragment length polymorphism (AFLP), DNA methylation amplified fragment length polymorphism (mAFLP), cDNA-amplified fragment length polymorphism (cDNA-AFLP) and cDNA single-strand configuration polymorphism (cDNA-SSCP).The main results were as follows:1. These hybrids generally showed an intermediate morphology, while the reciprocal hybrids inclined more to the maternal parents, especially BB.A, CC.A and CC.B, probably due to the one more copy of the maternal genome. The three digenomic hybrids (A.B, A.C, B.C) had very low pollen fertility, and the two trigenomic hybrids (A.C.B, C.A.B) were male sterile. Among the three digenomic triploids (BB.A, CC.A, CC.B), BB.A and CC.B had much higher pollen fertility than CC.A, and produced some seeds after open-pollination. The pairing frequency in A.C was obviously higher than those in A.B and B.C. The average of bivalents in C.A.B was significantly lower than that in A.C.B. The pairing frequency in BB.A and CC.B was similar, but much lower than that in CC.A. 2. FISH analysis indicated the maximum autosyndetic bivalents were3,2and3pairs for A, B and C genome, respectively. The mean autosyndetic bivalents ranged from0.32to1.73pairs for A genome, from0.23to0.70pair for B genome, from0.57to1.14pairs for C genome. The maximum homologous bivalents for B genome were7pairs,with mean of5.35in BB.A. The maximum homologous bivalents for C genome were9and8pairs in CC.A and CC.B, respectively, with mean of6.29and6.70pairs. The maximum allosyndetic bivalents were8pairs for A-C,5for A-B, and5for B-C. The total chromosomes per cell involved in auto-and allosyndesis and no pairing (univalents) within A genome in these hybrids were2.07,3.72and4.21, respectively. The total chromosomes per cell involved in auto-and allosyndesis and no pairing within B genome in these hybrids (excluding BB.A) were1.19,2.47and4.34, respectively. The total chromosomes per cell involved in auto-and allosyndesis and no pairing within C genome in these hybrids (excluding CC.A and CC.B) were2.34,2.87and3.79, respectively. Among16chromosomes of B genome in BB.A,10.70formed homologous bivalents,1.06and2.68gave auto-and allosyndesis, and1.56unpaired. The18chromosomes of C genome in CC.A exhibited12.58for homologous pairing,0.60and4.09for auto-and allosyndesis,0.74for no pairing. The association of18chromosomes of C genome in CC.B occurred as13.40for homologous pairing,1.20for autosyndesis,1.80for allosyndesis, and1.60for no pairing. The total chromosomes per cell involved in auto-and allosyndesis within B genome in these hybrids were generally lower than those in A and C genomes and varied in narrower ranges, but the rates to the total chromosome number of the three genomes were similar in most hybrids, and only in A.C.B, the rates for auto-and allosyndesis within B genome were significantly lower than those within A and C genomes. The averages and rates for autosyndesis in each hybrid were significantly lower than those for allosyndesis, but not for those of A and C genomes in A.C, of C genome in B.C, and of B genome in CC.B. The averages and rates for autosyndesis within each genome across all hybrids were also significantly lower than those of allosyndesis. The autosyndetic bivalents and chromosomes within A genome were insignificantly different between A.B and BB.A, but significantly different between A.C and CC.A. This further showed that the haploid or diploid of B or C genome had different effects on the pairing of A genome. The means and rates for autosyndetic chromosomes involved in B genome were insignificantly different between B.C and CC.B, this showed that the haploidy or diploidy of C genome had no or limited effects on the pairing of B genome.3. The chromosome pairings in the synthetic allotetraploids (AA.BB, AA.CC/CC.AA, and BB.CC) were not fully diploidized, the univalents and multivalents appeared frequently. The mean homologous bivalents for A genome were significantly higher in AA.BB than in AA.CC, there were similar between AA.CC and CC.AA. The mean homologous bivalents for B genome were significant higher in AA.BB than in BB.CC. The mean homologous bivalents for C genome were similar among AA.CC, CC.AA and BB.CC. The homoeologous bivalents and trivalents were only observed in AA.CC and CC.AA, respectively. The A-A-A-A quadrivalents occurred with the similar rates in AA.BB and CC.AA. The A-A-B-B quadrivalents appeared in AA.BB. The A-A-C-C quadrivalents occurred in AA.CC and CC.AA at relatively high rates, and two such pairing per cells were observed. The B-B-C-C quadrivalents also occurred in BB.CC at low rate.4. AFLP, mAFLP and cDNA-AFLP analyses of hybrids and allotetraploids revealed that21.61%the parental genomic bands,17.70%parental methylation bands and9.34%parental transcriptional bands were altered. The loss of parental bands was much more than the gain novel bands. About31.28%A-genome-specific bands,31.97%B-genome-specific bands and13.40%C-genome-specific bands were lost for genomic fragments. The proportions of A and B-genome-specific fragment losses were similar, but much higher than that of C-genome-specific fragment losses. About31.72%A-genome-specific bands,19.09%B-genome-specific bands and13.91%C-genome-specific bands were lost for methylation fragments. The proportions of A-genome-specific fragment losses were much higher than that of B-genome-specific fragment losses, then much higher than that of C-genome-specific fragment losses. About12.92%A-genome-specific transcriptional bands,16.63%B-genome-specific transcriptional bands and8.40%C-genome-specific transcriptional bands were lost. The proportions of B-genome-specific transcriptional fragment losses were much higher than that of A-genome-specific transcriptional fragment losses, then much higher than that of C-genome-specific transcriptional fragment losses. These results indicated the C genome was largely maintained in the hybrids and allotetraploids. The difference of genome-specific losses between hybrids and allotetraploids was related to the genome combination, cytoplasm and ploidy. By comparing the genomes in hybrids and allotetraploids with the parental species; cytoplasm, ploidy levels and the relationships between parental genomes were factors in determining the frequency of genome biased sequence elimination. The comparison of genomic, methylation and gene expression changes between hybrids and resultant allotetraploids should reveal the effect of genome duplication. In the vast majority of cases, the allotetraploids retained the genomic, methylation and gene expression state observed in the diploid hybrids. For genomic changes, on average,77.66%of these cases involved additive, whilst18.87%of bands involved retention of a nonadditive profile. An average of1.38%of bands displayed a shift relative to the diploid hybrids in the additive favoured, whilst2.07%of bands displayed a shift relative to the diploid hybrids in the nonadditive state. For methylation changes,80.68%of these cases involved additive methylation, whilst9.75%of bands involved retention of a nonadditive methylation profile.4.54%of bands displayed a shift relative to the diploid hybrids in the additive methylation state favoured, whilst4.22%of bands displayed a shift relative to the diploid hybrids in the nonadditive methylation state. Finally,0.48%of bands showed novel nonadditive methylation not observed in the diploid hybrids. For gene expression changes,80.68%of these cases involved additive, whilst9.75%of bands involved retention of a nonadditive profile.4.54%of bands displayed a shift relative to the diploid hybrids in the additive favoured, whilst4.22%of bands displayed a shift relative to the diploid hybrids in the nonadditive.5. Combined data of DNA, methylated DNA and cDNA-AFLP exhibited that DNA fragment losses were significant correlated with cDNA-AFLP fragments losses; and mDNA fragment losses were insignificantly correlated with DNA and cDNA-AFLP fragments losses. Significant linear relationships were also detected among A-DNA, A-mDNA and A-cDNA fragment losses, among B-DNA, B-mDNA and B-cDNA fragment losses, and among C-DNA, C-mDNA and C-cDNA fragment losses. The losses of A genome specific genomic fragments losses were significantly correlated with A genome specific methylation fragments losses. And the losses of B genome specific cDNA-AFLP fragments losses were significantly correlated with B genome specific genomic and methylation fragments losses. They were insignificantly correlated with each other for C genome. Those suggested that genetic and epigenetic changes were largely responsible for alteration of gene expression, and genetic changes were also largely responsible for epigenetic changes.6. cDNA-SSCP analyses of hybrids and allotetraploids indicated that:In BB.A, A.B, AA.BB, CC.B, B.C, BB.CC, C.A.B, A.C.B hybrids, only rRNA genes from B genome were transcribed, while those from A or C genome were silent. In CC.A, A.C, AA.CC, CC.AA hybrids, transcripts from A genome were detected, while those from C genome were silent. This suggested that rRNA genes hierarchy was B>A>C, and the hierarchy could convert from the low degree to the high degree and dominance was independent of maternal effect, ploidy or rRNA gene dosage, and was reversible. |
PDF Full Text Request