Even though the CMS flower phenotypes resemble homeotic mutant phenotypes, homeotic genes in the CMS lines are not mutated. Rather, the aberrant phenotype is the result of the alloplasmic combination of a nucleus from one species with mitochondria from another species (Budar et al., 2003; Hanson and Bentolila, 2004). In those CMS systems that have been characterized molecularly, the different CMS-associated mitochondrial ORFs show very few similarities but still the products of the novel mtDNA regions result in similar effects (Hanson and Bentolila, 2004; Schnable and Wise, 1998). Furthermore, in most CMS systems, the new polypeptides encoded by CMS-induced mitochondrial genes are found in the mitochondrial membrane fractions or inside the mitochondria (Abad et al., 1995; Budar and Pelletier, 2001; Grelon et al., 1994; Song and Hedgcoth, 1994). Ultrastructural studies of the CMS lines revealed that the inner-membrane integrity of most mitochondria was disrupted in the flower tissues, while mitochondria in the fertile B. napus cultivar exhibited ordinary morphology. Farbos et al. (2001) reported similar results from studies of alloplasmic male-sterile lines of N. tabacum. Moreover, disrupted mitochondria were found in male-sterile tobacco plants obtained after transformation with an unedited copy of the mitochondrial atp9 gene (Hernould et al., 1998). Thus, once the structure of the mitochondria is disrupted in tissues that require a high energy level, male sterility can occur. Reduced ATP levels were found in the alloplasmic CMS lines included in this study (Teixeira et al., 2005) and in flower tissues of the alloplasmic CMS lines of tobacco (Bergman et al., 2000).
The homeotic conversion of anthers is correlated with decreased levels of AP3 and PI at later floral stages
A reduction in BnPI mRNA levels for the initial six stages was noted in CMS line 41:17 which has a more delayed flower development when compared with the CMS line 4:19. BnAG activation was not affected in the CMS plants during early flower development. At later stages, however, AG is only expressed in specific types of the ovule and stamen cells (Bowman et al., 1991b; Deyholos and Sieburth, 2000); this might account for the reduced BnAG levels we observed in the CMS lines after stage 6 as the CMS lines lack stamens which, thus, results in reduced numbers of cells accumulating BnAG mRNA.
Plants mutated in the SCFUFO complex genes resemble CMS plants in many aspects of vegetative and floral development by producing shorter plants and displaying conversion of anthers into carpelloid structures (Leino et al., 2003). Indeed, in such mutants, decreased levels of AP3 and PI expression were observed after stage 6 (Ni et al., 2004; Wang et al., 2003; Zhao et al., 1999), which was also found in our CMS system. ASK1 and UFO are components of the SCF complex which interacts genetically with LFY to facilitate the degradation of a negative regulator of B-class gene expression (Ni et al., 2004; Wang et al., 2003; Zhao et al., 2001). It is known from animal and yeast systems that association of F-box proteins with SCF complexes targets specific proteins for degradation (Patton et al., 1998). The levels of BnLFY and the two SCFUFO genes, BnUFO and BnASK1, were higher in the CMS lines but the accumulation of BnAP3 and BnPI mRNA was reduced. Reduction in the expression of these genes suggests that activation of the B-genes by the SCFUFO complex and LFY did not occur.
The apparent contradictory upregulated upstream genes and reduced expression levels of downstream genes might suggest that the phenotypes result from a blocked activation of BnAP3 and BnPI. Because anther development requires efficient mitochondrial activity (Mackenzie and McIntosh, 1999; Smart et al., 1994) during a short period of time, the reduced ATP levels formed in the flower tissues (Teixeira et al., 2005) might slow down the protein degradation necessary for the correct activation of AP3 and PI by the SCFUFO complex.
In turn, the BnAP3/PI autoregulation occurring after stage 6 of flower development (Goto and Meyerowitz, 1994; Jack et al., 1994; Samach et al., 1997) is not accomplished in CMS plants. At these later floral stages, BnLFY and BnUFO levels are higher in the CMS lines than in B. napus, suggesting that the role of these genes in activating BnAP3 and BnPI is affected. The reduced levels of BnPI, especially in the CMS line 41:17 at earlier floral stages, also reflect the hypothesized deficiency in the autoregulatory pathway, as the PI and AP3 genes are differentially activated (Honma and Goto, 2000; Tilly et al., 1998).
CycB1 proteins contain a motif that targets it for degradation through proteolysis (Capron et al., 2003). Tobacco plants expressing a non-degradable form of CycB1 showed mis-shaped cells, especially in mitotically active tissues like meristems. Accumulation of CycB1 mRNA was also associated with the mutant phenotype (Weingartner et al., 2004). Although the expression patterns revealed by in situ hybridizations of the two mitosis-related genes, BnCycB1 and BnTON1, showed no differences between the three lines, differences in expression levels were noticed for one of the CMS lines with real-time RT-PCR. BnCycB1 and BnTON1 mRNA concentration was higher in CMS line 4:19 compared with B. napus during the first floral stage.
The hypothesis that reduced energy levels produced by the mitochondria in CMS systems affect cell divisions in flower meristems and/or the general metabolism rate, leading to impairment of pollen production due to high energy demands during this process, was advanced by Farbos et al. (2001), Sabar et al. (2003) and Hanson and Bentolila (2004). We extend the hypothesis that the CMS phenotype in our Brassica lines is caused by reduced ATP levels in flower tissues (Teixeira et al., 2005) affecting a key mechanism responsible for proper cell cycling like the process of protein degradation driven by phosphorylation steps (Koepp et al., 1999).
Cytoplasmic male-sterile plants displaying a phenotype of abnormal anther formation found in the Brassica system reported here, in the alloplasmic lines of tobacco (Farbos et al., 2001) and in carrot (Linke et al., 2003) suggest that a common mechanism is involved. In all these CMS systems, the initial expression pattern of AP3 and PI homologous genes resembles their expression in the fertile parental lines whereas expression levels were reduced later in development. Among the homeotic genes, AP3 and PI genes have the most complex regulation in flower development involving numerous other genes and co-factors (Bowman et al., 1993; Honma and Goto, 2000; Lamb et al., 2002; Ng and Yanofsky, 2001). This complex regulation which must occur between stage 3 until third whorl organs are completely differentiated may not be achieved correctly in CMS systems, if they are hampered by reduced ATP levels. In this work we showed that the downregulation of BnAP3 and BnPI after floral stage 6 does not result from a downregulation of upstream genes. Rather, the regulation by BnLFY and the SCFUFO complex seems to be impaired.