Apomixis or asexual seed formation in Hieracium praealtum (Asteraceae) is controlled by two independent dominant loci. One of these, the LOSS OF APOMEIOSIS (LOA) locus, controls apomixis initiation, mitotic embryo sac formation (apospory) and suppression of the sexual pathway. The LOA locus is found near the end of a hemizygous chromosome surrounded by extensive repeats extending along the chromosome arm. Similar apomixis-carrying chromosome structures have been found in some apomictic grasses, suggesting that the extensive repetitive sequences may be functionally relevant to apomixis.
Fluorescence in situ hybridization (FISH) was used to examine chromosomes of apomeiosis deletion mutants and rare recombinants in the critical LOA region arising from a cross between sexual Hieracium pilosella and apomictic H. praealtum.
The combined analyses of aposporous and nonaposporous recombinant progeny and chromosomal karyotypes were used to determine that the functional LOA locus can be genetically separated from the very extensive repeat regions found on the LOA-carrying chromosome.
The large-scale repetitive sequences associated with the LOA locus in H. praealtum are not essential for apospory or suppression of sexual megasporogenesis (female meiosis).
Apomixis refers to the suite of reproductive processes that facilitate asexual seed production in plants, which form clonal progeny genetically identical to the maternal plant. Introducing apomixis into sexual crop species has the potential to radically alter, hasten and simplify plant breeding and seed production practices through indefinite propagation of desirable genetic traits and fixation of hybrid vigour. To date, efforts at stable introgression of apomixis into crop species from close apomictic relatives have been unsuccessful as the apomictic lines generated have additional chromosomes and increased seed abortion. Genetic engineering has the potential to offer a more flexible approach to introducing apomixis into crops but is likely to require a thorough mechanistic understanding of both sexual and apomictic developmental processes (Spillane et al., 2004; Pupilli & Barcaccia, 2012).
The apomictic members of eudicot Hieracium subgenus Pilosella undergo a form of gametophytic apomixis known as apospory in which somatic ovule cells, termed aposporous initial (AI) cells, differentiate adjacent to cells undergoing meiosis. AI cells avoid meiosis and develop chromosomally unreduced embryo sacs via mitosis, a process commonly termed apomeiosis. In the case of apomictic Hieracium subgenus Pilosella species, the developing unreduced aposporous embryo sacs supplant the sexual cells and the sexual pathway terminates. Fertilization-independent embryogenesis and the atypical component of fertilization-independent endosperm development then occur in these aposporous embryo sacs, giving rise to maternally derived seed independent of paternal genomic input (Tucker & Koltunow, 2009).
Genetic and γ-deletion mutagenesis studies in apomict Hieracium praealtum (R35) identified the dominant LOSS OF APOMEIOSIS (LOA) locus that regulates apospory and displacement of the sexual pathway, and the independent dominant LOSS OF PARTHENOGENESIS (LOP) locus which controls the processes of autonomous embryogenesis and endosperm formation (Catanach et al., 2006). Deletion of either the LOA or LOP locus results in partial reversion to sexual reproduction, whereas deletion of both loci results in complete reversion to sexual development indicating that the sexual pathway is the underlying reproductive mode in examined apomictic Hieracium species (Koltunow et al., 2011). A recent cytological and genetic study of progeny in crosses of sexual and apomictic Hieracium species has shown that the component of fertilization-independent endosperm formation is genetically separable from autonomous embryogenesis and from the apospory component controlled by LOA (Ogawa et al., 2013).
Molecular markers linked to the LOA and LOP loci were identified by comparing sequences lost between the apomictic R35 parent and derived deletion mutants (Catanach et al., 2006). Sequence-characterized amplified region (SCAR) markers obtained through conversion of the original amplified fragment length polymorphism (AFLP) markers (Catanach et al., 2006) were used to identify genomic sequences in a bacterial artificial chromosome (BAC) library made from R35 genomic DNA (Okada et al., 2011). This enabled the isolation of c. 1.2 Mb of R35 genomic sequence in three separate contigs associated with the central region of the LOA locus (Fig. 1a–c). Analyses of the partial LOA locus sequence indicated that the region is gene poor and enriched for simple and complex repeats as well as transposons (Okada et al., 2011). Cytogenetic fluorescence in situ hybridization (FISH) analyses have demonstrated that the LOA locus in R35 is sub-telomerically located on a single elongated hemizygous chromosome, and is associated with extensive repeats that cover almost half of the long chromosome arm (Fig. 1a). Despite the hemizygous nature of the LOA locus and associated repeat region, recombination does occur at the locus, albeit at suppressed levels, when crosses are made between sexual and apomictic Hieracium plants. Progeny analyses from these crosses had delineated the genomic region essential for LOA function between two SCAR markers (14-T7 and 9-HR; Fig. 1c) present on two contigs encompassing 400 kb of known sequence and an intervening region of sequence which remains to be identified (Okada et al., 2011).
The structural characteristics of the LOA locus, associated SCAR markers and repeats found on the LOA-carrying chromosome in R35 are conserved in some other apomictic Hieracium species. They include tetraploid Hieracium caespitosum (C36), and tetraploid (D36) and diploid (D18) forms of apomictic Hieracium piloselloides. In D36 and D18, the extensive hybridizing repetitive sequence structure is smaller relative to R35 and C36. Conversely, the LOA locus and associated SCAR markers are absent in sexual tetraploid Hieracium pilosella (P36). Apomictic tetraploid (A36) and aneuploid (A35) accessions of Hieracium aurantiacum lack LOA-linked markers and associated hybridizing repeat sequences, and it is unclear if a different locus regulates apospory in these species (Okada et al., 2011).
Structural characteristics of the repeat-rich LOA-carrying chromosome resemble those of the hemizygous repeat-rich chromosome carrying the apospory-specific genome region (ASGR) in the monocot grasses Pennisetum squamulatum and Cenchrus ciliaris. In these monocot apomicts, the components of apospory and fertilization-independent embryo formation are linked on the ASGR-carrying chromosome (Sherwood et al., 1994; Ozias-Akins et al., 1998). Comparison of partial genomic sequences in the LOA and ASGR loci show enrichment in Ty3-gypsy and Ty1-copia transposons, respectively, although no obvious DNA sequence conservation between the two loci has been identified as yet (Okada et al., 2011).
Conservation of repeat-rich features of hemizygous chromosomes carrying apospory loci in monocot and eudicot apomicts raises questions concerning the function of the extensive repeats in the initiation and progression of apomixis. Here, we identified recombinants in the LOA region amongst hybrid progeny from a sexual and apomictic cross. Karyotype and FISH analyses of these recombinants and deletion mutants were used to determine whether or not the structural integrity of the LOA-carrying chromosome and association with repeats is required for apospory and/or autonomous seed formation.
Materials and Methods
Plant materials, growth conditions and phenotyping
Three Hieracium subgenus Pilosella accessions were used in this study and consisted of the apomictic aneuploid (4x – 1 = 35) Hieracium praealtum (R35), apomictic tetraploid (4x = 2n = 36) Hieracium caespitosum (C36), and the sexual tetraploid (4x = 2n = 36) Hieracium pilosella (P36) (Koltunow et al., 2011). In addition, four F1 progeny from a H. pilosella × H. praealtum (P36×R35) mapping population (n = 833) with different phenotypes consisting of the presence or absence of apospory (± LOA) and the presence or absence of autonomous seed formation (±LOP) were used as listed: plant 2-16 (+LOA; −LOP), 16-8 (+LOA; LOP not determined), 20-13 (−LOA; −LOP) and 28-36 (+LOA; +LOP) (Okada et al., 2011). The loa124 (−LOA; +LOP) and loa135 (−LOA; +LOP) γ-deletion mutants derived from H. praealtum were also used (Catanach et al., 2006). Characterized plants were vegetatively propagated to maintain clonal integrity. The plant growth conditions and cytological phenotyping methods used were as described previously (Koltunow et al., 1998, 2011). Phenotypes were based on observations of AI cell formation at stage 4, and embryo and endosperm development at stage 10. Development is stochastic and the frequency does not represent total frequency of viable seed formed. A minimum of 80 ovules were examined from unpollinated florets which were collected from a minimum of three individual plants.
Development of LOA-linked molecular markers
SCAR markers associated with the LOA locus were developed from BAC sequences isolated from H. praealtum and diploid H. piloselloides (D18) BAC libraries whilst chromosome walking at the LOA locus as previously described (Okada et al., 2011). Oligonucleotides for SCAR markers and associated PCR conditions are listed in Supporting Information Table S1. DNA extraction from leaf samples and analysis of LOA-linked SCAR markers was performed as described previously (Okada et al., 2011).
Vigorously growing root tips were collected from Hieracium plants aseptically grown on 0.5× Murashige and Skoog medium agar in sterile pots (Koltunow et al., 1998). To accumulate metaphase cells, the root tips were pre-treated in ice water for 17–19 h. The root tips were then fixed in 3 : 1 ethanol : acetic acid. Chromosome preparations for FISH were performed as described previously (Mukai et al., 1990). Metaphase chromosome spreads of Hieracium plants used in this study are shown in Fig. S1.
LOA267.14 BAC DNA and 18S-5.8S-26S rDNA were purified by the QIAGEN Plasmid Midi kit (Qiagen) and labelled with biotin-16-dUTP using a Biotin-Nick Translation Mix kit (Roche Diagnostics) and digoxigenin-11-dUTP using a Dig-Nick Translation Mix kit (Roche Diagnostics), respectively.
Chromosomal DNA was denatured in 70% formamide-2× SSC for 2 min at 69°C and dehydrated in an ethanol series at −20°C. The hybridization mixture consisted of 50% formamide, 10% dextran sulfate, 50 μg of salmon sperm DNA, 2× SSC, and the labelled BAC DNA or rDNA. The mixture was denatured for 10 min at 100°C, immediately quenched in ice for at least 10 min, and applied to each slide. Hybridization took place overnight in a moist chamber at 37°C. After hybridization, the slides were washed in 2× SSC at room temperature for 5 min, 50% formamide-2× SSC at 37°C for 15 min, 2× SSC at room temperature for 15 min, 1× SSC at room temperature for 15 min, and 4× SSC at room temperature for 5 min. For the simultaneous detection of digoxigenin and biotin, slides were incubated in 2 μg ml−1 rhodamine-conjugated anti-digoxigenin (Roche Diagnostics) and 2 μg ml−1 fluorescein isothiocyanate (FITC)-conjugated avidin (Roche Diagnostics) in the detection buffer containing 4× SSC-1% BSA for 1 h at 37°C. After incubation, the slides were washed in 4× SSC for 10 min, 0.1% Triton X-100 in 4× SSC for 10 min, 4× SSC for 10 min, and 2× SSC for 5 min, all at room temperature. The slides were mounted in a fluorescence anti-fade solution (1.25% DABCO, 90% glycerol). DAPI was used as chromosome DNA counterstaining in the anti-fade solution. Each fluorescent signal on the slide was captured with an Axioskop fluorescence microscope (Zeiss) coupled to a cooled CCD camera (Hamamatsu Photonics, model 4880, Hamamatsu, Japan). Images were pseudo-coloured and merged using Photoshop software (Adobe, San Jose, CA, USA).
Further delineation of the genomic region essential for LOA function
We previously reported the use of a chromosome walking strategy with four SCAR markers linked to the central region of the LOA locus to generate three independent DNA contigs totalling c. 1.2 Mb of genomic sequence, known as Contigs A, B and C (Okada et al., 2011). In this study, we used the same strategy to isolate an additional seven BACs for Contig A and eight BACs for Contig B which has extended the contigs by c. 700 kb to a combined total length of c. 1.9 Mb (Fig. 1c). A 26-kb overlap between Contig C and Contig A was identified and enabled the three contigs at the LOA locus to be consolidated into two contigs (Fig. 1c). This resulted in a change in the SCAR marker order reported previously (Okada et al., 2011), with marker LOA219 now located within Contig A proximal to a gap of unknown sequence between Contigs A and B (Fig. 1c). The identification of genomic sequences in the region between Contigs A and B is continuing. BAC end sequences were used to develop additional SCAR markers distributed across both contigs for physical mapping of the LOA locus in Hieracium mapping populations (Fig. 1c). The suite of new SCAR markers obtained from the expanded LOA contig sequences were used to examine the rare recombinant progeny at the LOA locus arising from a cross between sexual P36 and apomictic R35. These progeny had been phenotyped for their ability to undergo apospory and also autonomous seed formation (Fig. 1c,d).
In this study, we have further characterized two recombinant F1 progeny (16-8 and 28-36) from a P36 × R35 cross with the complete set of LOA-linked SCAR markers and sequenced selected marker amplicons to confirm SCAR marker specificity. The 16-8 and 28-36 recombinants contained none of the Contig A markers and all the Contig B markers indicating that recombination had occurred within the gap region of unidentified sequence between the two contigs in both plants (Fig. 1c). Cytological ovule examination indicated that 16-8 and 28-36 were aposporous with AI cell frequencies of 12% and 36%, respectively (Fig. 1d). Sexual demise was also observed, indicating that these plants contained a functional LOA locus.
Our previous mapping study had shown that the sequences required for LOA function were located between SCAR markers 14-T7 on Contig A and 9-HR on Contig B (Okada et al., 2011) (Fig. 1c). As recombinant plants 16-8 and 28-36 were aposporous and had none of the Contig A SCAR markers, we have been able to exclude the Contig A sequence as being essential for LOA function. Thus, we have further delineated the genomic sequences required for the initiation of apomixis in H. praealtum to be between the 25-T7 SCAR marker on Contig A and the 9-HR marker on Contig B (Fig. 1c). This delineated region contains c. 400 kb of isolated genomic sequences and unidentified sequence present within the gap between Contigs A and B.
The presence of large-scale repetitive sequences in an LOA deletion mutant does not suppress the early stages of sexual reproduction
We sought to investigate if the structural integrity of the LOA-carrying chromosome and association with large-scale hybridizing repetitive sequence observed in H. praealtum was necessary for apospory and autonomous seed formation. To address this question, FISH analysis was performed with plants that had partial or complete loss of LOA locus-associated sequences as determined by phenotypic and molecular analyses with markers linked to the LOA locus (Fig. 1c,d). FISH was initially used to determine whether the highly repetitive chromosomal structure surrounding the LOA locus was present or absent within two LOA deletion mutants, loa124 and loa135, which were selected from a suite of LOA deletion mutants unable to form unreduced female gametophytes but were able to undergo autonomous seed formation (Fig. 1d). Although the physical sizes of the deletions encompassing the LOA locus in these mutants were unknown, previous characterization of AFLP markers enabled the relative deletion sizes to be determined. Mutant loa124 was shown to have the largest LOA-associated deletion with all AFLP markers being absent, whereas both flanks of the smaller deletion in loa135 had been delineated by AFLP markers (Catanach et al., 2006; Koltunow et al., 2011) (Fig. 1b).
Cytological characterization of metaphase chromosome spreads indicated that the loa124 and loa135 deletion mutants had 2n = 34 and 36 chromosomes, respectively, whereas the apomictic R35 progenitor plant had 2n = 35 chromosomes (Fig. 1d). FISH analysis with an 18S-5.8S-26S ribosomal DNA (rDNA) probe resulted in consistent hybridization to six chromosomes in loa124, loa135 and R35 (Fig. 1d). The differences in chromosome number can most likely be attributed to the γ-irradiation of H. praealtum seed used to generate the deletion mutants (Catanach et al., 2006) as ionising radiation is known to induce multiple deletions and other genetic aberrations such as the gain or loss of whole chromosomes (Vizir & Mulligan, 1999; Touil et al., 2000). The leaf and shoot morphology of the loa124 plant grown in tissue culture is distinct from the R35 progenitor, with the mutant having multiple shoots and short, slender leaves. When grown in soil, the loa124 leaf, shoot and flower morphology appear similar to R35, although the mature mutant plant is shorter than R35. The loa135 mutant is morphologically indistinguishable from R35 when grown in tissue culture and in soil.
FISH characterization of the extensive repetitive sequence structure within the loa124 and loa135 deletion mutants was undertaken using BAC LOA267.14 as a probe which had previously been shown to hybridize to the repeat-rich region surrounding the LOA locus (Okada et al., 2011). In this report, reference to the extensive repetitive sequence in Hieracium refers specifically to the LOA-associated repeat region which hybridizes to the LOA267.14 BAC probe. The specificity of this probe was confirmed with strong hybridization signals observed in the lower arm of the elongated chromosomes in the R35 and C36 apomict accessions and the absence of signal in sexual P36 (Fig. 1d). No hybridization signal was detected in the FISH of the loa124 deletion mutant, indicating that this mutant has a very large deletion that appeared to encompass all of the large-scale sequence structure associated with the LOA locus on the elongated chromosome of R35 (Fig. 1d). Alternatively, it is possible that the entire chromosome carrying LOA had been lost as loa124 only had 34 chromosomes and did not have any of the AFLP markers linked to the LOA locus. The absence of repetitive sequence structure in the loa124 mutant did not prevent this plant from generating fertilization-independent embryos and endosperm, suggesting that the extensive repeat region is not essential for autonomous seed formation in apomictic R35. Both mutants were capable of autonomous seed production although the embryo and endosperm formation frequencies were lower in loa124 (8%) and loa135 (33%) relative to R35 (50%) (Fig. 1d). Although the possibility that the penetrance of the autonomous seed trait may be influenced by the presence or absence of extensive repetitive sequence structure cannot be excluded, it is likely that the difference in autonomous seed formation in the two mutants reflects genetic variation caused by γ-irradiated mutagenesis of the R35 progenitor plant.
In contrast to loa124, strong hybridization of the LOA267.14 BAC probe to one of the loa135 chromosomes demonstrated that loa135 retains an extensive region of repetitive sequence. The presence of the repeat-rich region in loa135 and its absence in loa124 confirmed previous AFLP analysis indicating that the deletion in loa135 was substantially smaller than the deletion in loa124 (Koltunow et al., 2011). The length of the repetitive region in loa135 was estimated to be c. 1.09 μm ± 0.3, which was significantly (P < 0.05, t-test) smaller than the 1.58 μm ± 0.25 repeat region in apomictic R35 (Fig. 2) that had been determined previously (Okada et al., 2011), and presumably reflects the deletion of the LOA locus. The loa135 mutant generated meiotically derived embryo sacs, as can be inferred from the cytological observation that this mutant did not produce AI cells but had an embryo and endosperm frequency of 33% (Fig. 1d), suggesting that the presence of large-scale repetitive sequence structure alone does not facilitate the avoidance of meiosis. FISH characterization of the loa124 and loa135 mutants indicates that in the absence of linked LOA sequences, the extensive repeat sequences do not appear to influence early sexual development or autonomous seed formation.
Large-scale repetitive sequences surrounding the LOA locus are not required for apospory
We performed FISH characterization of four F1 progeny from a P36×R35 cross, including two plants recombinant at the LOA locus (Fig. 1c). By this means, we assessed how the structural integrity of the LOA-carrying chromosome may have been affected by the processes involved in sexual reproduction. The karyotypes of the P36×R35 F1 progeny were assessed cytologically and by FISH using a probe for the tandemly arrayed 18S-5.8S-26S rDNA genes. All the progeny with the exception of 20-13 were aposporous and must have inherited LOA from the apomictic R35 parent. All F1 progeny contained 2n = 36 chromosomes and the rDNA probe hybridized with six chromosomes in 28-36 and 20-13, and with seven chromosomes in 2-16 and 16-8, indicating that the progeny appeared to have hybrid P36/R35 chromosomal constitutions expected from sexual reproduction (Fig. 1d).
The structural integrity of the LOA-carrying chromosome in the P36×R35 F1 progeny was assessed by FISH with the LOA267.14 BAC probe. FISH characterization of the aposporous F1 plant (2-16) containing all of the LOA-linked SCAR markers indicated that this plant had a single chromosome with a large region of highly repetitive sequence (Fig. 1d). There were no significant differences (P > 0.05, t-test) in the sizes of the repetitive regions between 2-16, C36 and R35 (Fig. 2) indicating that the LOA locus and surrounding hemizygous large-scale repetitive sequence structure can be inherited in progeny from a cross between sexual and apomictic Hieracium species. The sexual F1 plant (20–13) did not have any of the known LOA-associated SCAR markers and did not produce AI cells; this corresponded with the lack of detectable signal by FISH, indicating that the sexual hybrid did not contain the large-scale repetitive sequence (Fig. 1c,d).
A hybridization signal associated with repetitive sequence was not detected in FISH analysis of the 16-8 and 28-36 recombinants (Fig. 1d). Given that both 16-8 and 28-36 are aposporous, the FISH results indicate that the large-scale repetitive sequence structure is not required for the initiation or maintenance of apospory in these hybrid plants. Furthermore, the frequency of AI cell formation in 28-36 (36%) was comparable to that observed in the apomictic R35 parent (36%) (Fig. 1d), indicating that the absence of large-scale repetitive sequence surrounding the LOA locus does not affect the penetrance of apospory. However, the lower frequencies of AI cell formation in ovules of the aposporous 2-16 (21%) and 16-8 (12%) progeny suggests that other unknown factors segregate in the genetic backgrounds of the hybrids which influence the frequency of AI cell formation (Fig. 1d).
Recombination at the LOA locus
Unlike most other apomixis loci where recombination is heavily suppressed (Ozias-Akins & Van Dijk, 2007), map-based cloning is a feasible strategy for isolating the apomeiosis gene(s) in Hieracium praealtum because recombination does occur in the vicinity of the LOA locus, albeit at partially suppressed levels. Chromosome walking enabled us to consolidate the three LOA-associated contigs we had generated in a previous study (Okada et al., 2011) into two contigs, and to correctly order the LOA219 marker with respect to other SCAR markers associated with the LOA locus. Overall consideration of SCAR markers analysed in phenotyped recombinant progeny from crosses between sexual and apomictic Hieracium species in this study and in our previous study (Okada et al., 2011) have enabled us to further delineate the LOA locus to be between markers 25-T7 and 9-HR (Fig. 1c). We are continuing our efforts to isolate the DNA sequences between these two markers and to identify the gene(s) responsible for apomeiosis in H. praealtum.
Recombination clearly occurs in the vicinity of the LOA locus, which is interesting given that the majority of the lower chromosome arm that contains the LOA locus is hemizygous with no apparent homologous counterpart. Since the analysed P36×R35 progeny were of the F1 generation and meiosis occurs in each parent of a cross before double fertilization, meiotic recombination in the LOA-carrying chromosome must have occurred within the R35 genome only. The SCAR marker patterns in the 16-8 and 28-36 recombinants demonstrate that recombination had occurred in the intervening region between Contigs A and B. Although the extensive repetitive sequence and the LOA locus-associated contigs in R35 are hemizygous, it is unknown whether the intervening sequence between Contigs A and B is also hemizygous, or whether there is an homologous region within another R35 chromosome. There are at least two possible ways in which recombination may have occurred within the LOA associated region in the P36×R35 F1 recombinant progeny. First, it is possible that allelic recombination may have occurred between the intervening sequence and homologous loci at distinct genomic locations. Second, ectopic (nonallelic) recombination may have occurred between repetitive sequences in different R35 chromosomes or possibly within the LOA-carrying R35 chromosome itself. Recombination in natural populations would be expected to reduce hemizygosity unless there were selective advantages in maintaining hemizygosity, for example, if apomictic recombinants had reduced fitness. We were unable to evaluate the adaptive fitness of the aposporous recombinants, as these plants were obtained from an experimental mapping population that was maintained through vegetative propagation and had not been subjected to selective pressures. We are currently analysing molecular markers from both contigs linked to the LOA locus in naturally sourced populations of sexual and apomictic Hieracium species, and also experimental hybrid populations recovered in soil without prior tissue culture. This may enable us to better assess the conservation of our LOA locus-linked markers in natural apomictic populations, and possibly the prevalence of recombinants in wild populations and under more stringent experimental conditions.
The extensive repetitive structure is not required for the avoidance of meiosis, formation of unreduced embryo sacs or autonomous seed formation in H. praealtum
The convergent evolution of repetitive sequence elements surrounding the apospory LOA locus in eudicot H. praealtum and the ASGR in monocot P. squamulatum and C. ciliaris suggests that the hemizygous, repeat-rich regions may be functionally important for apomixis in these species (Okada et al., 2011). We had previously postulated that apomixis may be controlled by genetic and/or epigenetic regulation. The extensive repetitive sequence elements associated with the LOA locus may directly or indirectly influence functional attributes of apospory through various mechanisms including alterations in chromatin structure, epigenetic marks, trans or cis induction of gene silencing and/or modification of gene expression patterns (Koltunow & Grossniklaus, 2003; Okada et al., 2011).
The LOA locus in apomictic R35 facilitates the differentiation of somatic ovule cells into AI cells and the subsequent suppression of the sexual pathway, which enables the formation of unreduced female gametophytes. The loa124 and loa135 deletion mutants do not have LOA so are unable to form AI cells or avoid meiosis. Thus, the sexual pathway proceeds in these deletion mutants to form meiotically reduced embryo sacs, while the presence of LOP stimulates autonomous seed formation which contain embryos with reduced ploidy. FISH characterization of loa135 shows that this mutant contains the majority of the large-scale repetitive sequence. Thus, the repetitive region found in loa135 is not sufficient to facilitate the suppression of sexual reproduction in the absence of the LOA locus. In addition, we have shown that the extensive repetitive sequence structure associated with the LOA locus in apomictic R35 is absent in two P36×R35 F1 progeny (16-8 and 28-36). These plants are recombinant at the LOA locus but are still able to avoid meiosis and generate somatically derived female gametophytes, and therefore must contain the essential LOA sequences required for apospory. This result clearly demonstrates that the extensive repeat-enriched structure is not essential for the regulation of the factors encoded by the LOA locus. Furthermore, the frequency of AI cell formation in 28-36 where the extensive repetitive region is absent is comparable to that of the R35 apomict parent, indicating that the extensive repetitive region does not appear to substantially influence the penetrance of the apospory trait. This observation also suggests that elements at the LOA locus required for qualitative apomixis initiation and sexual suppression are tolerant of substantial structural changes in the chromosome carrying the LOA locus. The lower frequencies of AI cell formation in the 16-8 and 2-13 F1 progeny which lack the extensive repetitive region indicate that there are likely to be other apomixis-modifying elements segregating in the hybrid background, as has been observed previously (Koltunow et al., 2000; Tucker et al., 2012).
The chromosomal location of the LOP locus in Hieracium has not yet been determined. LOP is known to be unlinked to LOA, unlike the situation in the ASGR locus of Pennisetum where the components of apomeiosis and autonomous embryo formation are tightly linked (Ozias-Akins et al., 1998; Conner et al., 2013). The possibility that the extensive repetitive sequence structure may function in trans as a source or sink for LOP-requiring/activating factors is unlikely because mutant loa124 had no detectable repetitive region, yet was able to form autonomous seed. Together, our results indicate that the extensive repetitive sequence structure is unlikely to have a role in the qualitative initiation or progression of apospory, sexual suppression or autonomous seed formation in H. praealtum. However, we cannot exclude the possibility that apomixis is influenced by relatively small regions of repetitive sequence at or near the genic sequences in the currently defined region of the LOA locus.
What is the function of the extensive repeats on the LOA-carrying chromosome?
Although numerous apomixis loci have been defined, no apomixis genes have yet been identified. Hence, it is unknown if the components of apomixis are regulated by single or multiple genes co-located at these loci. Partial or complete suppression of recombination between apomixis traits and/or multiple linked molecular markers has often been observed in the mapping of apomixis loci in various apomictic species of Pennisetum, Cenchrus, Paspalum, Tripsacum and Hieracium (Grimanelli et al., 1998; Ozias-Akins et al., 1998; Roche et al., 1999; Pupilli et al., 2001; Okada et al., 2011). Inhibition of recombination may serve to preserve linkage between putative genes conferring apospory and sexual suppression to ensure inheritance of the LOA locus as a functional unit. Although the precise mechanisms for recombination suppression at apomixis loci are unknown, the hemizygosity and repetitive chromosome structure associated with the LOA locus in H. praealtum and the ASGR in monocot P. squamulatum and C. ciliaris could play a role in maintaining the structural integrity of these loci. An alternative possibility is that the extensive repeat structure associated with the LOA and ASGR-carrying chromosomes may somehow aid in promoting the transmission or maintenance of the chromosome. Repeat regions have been postulated to be involved in the regulation of chromosome segregation to ensure maintenance of supernumerary chromosomes in natural plant and animal populations (Banaei-Moghaddam et al., 2012).
Evolution of apospory loci in Hieracium subgenus Pilosella species
The extensive repeat sequence associated with the LOA locus may be a consequence of apomixis rather than functioning as a regulatory element of LOA. The repeat-rich region could have evolved as a result of continual repetitive insertion of retrotransposons that have become fixed due to the asexual mode of reproduction. In maize, some LTR-retrotransposons have been shown to preferentially insert into each other in intergenic regions to form clusters of nested elements (SanMiguel et al., 1996; SanMiguel & Bennetzen, 1998). These types of nesting LTR-retrotransposons include the Ty3-gypsy and Ty1-copia elements that are highly represented within the LOA locus in H. praealtum (Okada et al., 2011).
Hieracium subgenus Pilosella species have highly variable morphology and fit into two divergent chloroplast haplotype network groups known as Pilosella I and Pilosella II (Fehrer et al., 2007a,b). The aposporous R35, C36 and D36 accessions are placed in Pilosella II whereas two other aposporous accessions of Hieracium aurantiacum (A35 and A36) are located in Pilosella I (Koltunow et al., 2011). In contrast to R35, C36 and D36, the A35 and A36 accessions do not have any of the LOA-linked SCAR markers or the extensive repetitive sequence structure surrounding LOA, and exhibit a slightly varied mode of aposporous embryo sac formation. In the absence of a gene-specific marker for apospory, these differences had suggested Hieracium subgenus Pilosella members may employ independent mechanisms for apospory (Okada et al., 2011). However, we have shown here that the extensive repeat-rich structure is not essential for apospory expression in H. praealtum, which leaves open the possibility that other Hieracium subgenus Pilosella species such as H. aurantiacum may potentially utilise the same apospory mechanism as H. praealtum, but may not have formed – or have lost – the flanking sequences of the core region linked to LOA and the associated extensive repetitive sequence structure.
Characterization of additional apomictic subgenus Pilosella species with the suite of LOA-associated SCAR markers and the repetitive sequence probe may help determine the degree of conservation of the delineated core region at the LOA locus and the surrounding extensive repeat-rich sequence structure. This would provide insight into the distribution of the LOA-carrying chromosome, and evolution of apospory within Hieracium subgenus Pilosella.
FISH characterization of the repetitive region in F1 progeny from a cross between sexual and apomictic species, and in LOA deletion mutants has shown that the extensive repetitive region is not essential for the initiation and progression of apomixis in H. praealtum. This finding has positive implications for the potential engineering of apomixis in agriculturally important plant species as it suggests that a relatively small genic region needs to be transferred and a large repetitive chromosomal segment is not required.
We thank Mandy Walker, Paul Boss and Melanie Hand for critical comments on the manuscript and helpful discussions. We also thank Karsten Oelkers for assistance with marker development. This research was supported by a Science and Industry Endowment Foundation grant to A.M.K.