Genomic Characterization of Somatic Hybrids

Selection of somatic hybrids

After fusion, protoplast suspensions con- tain parental, homofused, heterofused and multifused protoplasts. Regeneration is done without any selection pressure. It is therefore necessary to select the somatic hybrids and alloplasmic plants among the regenerated plants. Morphological charac- ters can be useful for somatic hybrid identi- fication (Grosser and Gmitter, 1990). However, a much more effi cient selection is generally done by the combination of molecular or isozyme marker analysis and ploidy evaluation.

Restriction fragment length polymor- phism (RFLP) analysis with rDNA probes was one of the fi rst molecular techniques for somatic hybrid identifi cation (Ohgawara et al., 1985; Takayanagi et al., 1992; Miranda et al., 1997). However, isozyme analysis is still a very easy and powerful tool for such applications. Several isozyme systems are routinely used, e.g. peroxidase (Tusa et al., 1990; X.X. Deng et al., 1992; Grosser et al., 1992a; Ye et al., 1992), phos- phoglucomutase (Grosser et al., 1992a; Tusa

et al., 1992; Ye et al., 1992; Ollitrault et al., 1996b) and phosphoglucose isomerases (PGIs; Tusa et al., 1990, 1992; Ollitrault et al., 1996b). Relative band intensity is gener- ally effi cient for identifi cation of somatic hybrids when a homozygous genotype is combined with a heterozygous one sharing a common allele with diploid plants. It is particularly true with dimeric enzymes such as PGI or isocitrate dehydrogenase (IDH). Moreover, these enzymes generally allow distinction between triploid and tetraploid hybrids arising from diploid + haploid protoplast fusion (P. Ollitrault, 2002, France, unpublished data). Random amplified polymorphic DNA (RAPD; Grosser et al., 1996b; Kobayashi et al., 1997; Shi et al., 1998b; Guo et al., 2000) has also been used and should be particularly useful when co-dominant markers such as isozymes or RFLP do not exhibit any poly- morphism between parents. STMS (sequence-tagged microsatellites) display- ing high polymorphism between mandarins (Luro et al., 2001) should also be interesting for combinations inside this cultivar group. Chromosome counting can now be favourably by-passed by fl ow cytometry for quicker ploidy analysis (Ollitrault et al., 1996b; Grosser et al., 2000). By using fl ow cytometry and isozymes or polymerase chain reaction (PCR) markers, it is possible to select somatic hybrids at the in vitro stage of plant regeneration.

Accurate nuclear genome characterization and somatic hybrid meiotic analysis

Nuclear genome instability has been observed for several crops, particularly for wide somatic hybridization (Té oulé, 1992; Kisaka et al., 1997; Oberwalder et al., 1998). Correlations between genetic distance and chromosome elimination have been described in Brassica (Sundberg and Glimelius, 1991). The multiple subcultures generally necessary to regenerate wide hybrids should be favourable to chromo- some elimination or recombination (Oberwalder et al., 1998; Guo and Deng,

1999). Moreover, the systematic use of embryogenic callus lines as one of the par- ents in citrus somatic hybridization could be another source of nuclear genome insta- bility. In the case of asymmetric hybridiza- tion, it is clearly necessary to manage tools to evaluate the contribution of the donor parent in the hybrid genome. It is therefore important to be able to analyse precisely the nuclear genome constitution for both sym- metric and asymmetric hybridization. Tetraploid somatic hybrids are used as par- ents for sexual crosses for triploid scion or tetraploid rootstock breeding (Grosser et al., 2000). Thus, the knowledge of meiotic behaviour of allotetraploid somatic hybrids appears essential to establish efficient breeding schemes.

Molecular tools currently used for somatic hybrids nuclear genome characterization

Several tools are currently used for somatic hybrid nuclear genome analysis. Schematically, two main classes of markers can be distinguished. Markers allowing a large random coverage of the genome such as RAPD (Forsberg et al., 1998b), single sequence repeats (SSRs; Harding and Millam, 2000; Cheng et al. 2002), inter- simple sequence repeats (ISSRs; Scarano et al., 2002) and amplifi ed fragment length polymorphism (AFLP; Tian and Rose, 1999; Guo et al., 2002) are very useful. Specifi c locus analysis allowing marking of each individual chromosome will be the better approach when addition or deletion lines are studied. For this purpose, RFLP (Rutgers et al., 1997) or STMS analysis with mapped markers will be preferred.

None of these methods will allow the display of evidence of interspecifi c chromo- some recombination. For such structural studies and global visualization of complex genomes, genomic in situ hybridization (GISH) has proved to be effi cient for several crops including woody persimmon (Choi et al., 2002). For example, this approach has been used to demonstrate that hexaploid hybrids arising from somatic hybridization

between Lycopersicon esculentum (2 x) and Solanum lycopersicoï des (2 x) contain two sets of L. esculentum with several chromo- some rearrangements between the two genera (Escalante et al., 1998). Specifi c elimination of Allium cepa in somatic hybrids between A. cepa and Allium ampeloprasum has also been demonstrated (Buiteveld et al., 1998). GISH is also very powerful for studying somatic hybrid meio- sis (Garriga-Calderé et al., 1999; Gavrilenko et al., 2001).

Potential of GISH for analysis of citrus somatic hybrid genomes

To determine the potential of GISH for citrus, Ollitraut et al. (2000d) have analysed three diploid species of Citrus (C. medica cv ‘Poncire’, C. reticulata cv ‘Willow leaf’ and

C. maxima cv ‘Pink’), Poncirus trifoliata cv ‘Pomeroy’ and Fortunella japonica cv ‘ Marumi’. GISH has also been applied on diploid sexual and tetraploid somatic hybrids between these genotypes. GISH did not allow colouring of all the set of chromo- somes. In most cases, only fi ve pairs from nine were coloured. Moreover, the staining appeared limited to the extremity of chro- mosomes and some central rDNA sites. The double staining of the C. reticulata + P. tri- foliata intergeneric somatic hybrid, with genomic DNA of each parent marked with different colours, showed that genomic dif- ferentiation between the two parental species was suffi cient to identify the stained chromosomes of each species (Fig. 10.1). In the case of C. reticulata + F. japonica as well as C. reticulata ´ C. maxima and C. maxima

´ C. medica, the differentiation of chromo-

somes appeared much more diffi cult and many spots displayed similar affi nity with the genomic DNA probes of both parents. A better specifi city of in situ hybridization at the interspecifi c level was found for the C. reticulata ´ C. medica hybrid.

In situ hybridization with 18S–25S rDNA has also been tested by Ollitrault et al. (2000d). They found that the number of chromosomes marked with the rDNA probe

Fig. 10.1. Double in situ hybridization of an allotetraploid hybrid C. deliciosa + P. trifoliata with total genomic DNA of C. reticulata (green) and P. trifoliata (red), from Ollitrault et al. (2000d).

(pTA71) varied between two for C. medica and C. limon to six for P. trifoliata (Fig. 10.2). Three sites were found for C. sinensis and C. aurantium, and four for F. japonica. These results are similar to those of Roose et al. (1998), who found six major sites and one occasional minor site for P. trifoliata, and three strong and two minor sites for C. sinensis.

From these results, it appears that the incomplete staining of the chromosome set and the very partial coloration of each chro- mosome will limit the application of GISH for citrus somatic hybrid genome analysis. It could be used to identify the relative con- tribution of each parental species to multi- ploid intergeneric genomes. Such an

application should also be found at the inter- and intrageneric levels, with rDNA probes if parental species present a differ- ent number of rDNA sites. The use of bacte- rial artifi cial chromosome (BAC) probes for in situ hybridization could perhaps be an interesting complementation of rDNA for chromosome distinction and then for the analysis of specifi c introgression or elimi- nation in breeding schemes.

Nuclear genomic structure of citrus somatic hybrids

The majority of publications report on the symmetric addition of nuclear parental

Fig. 10.2. In situ rDNA hybridization on chromosomes of six species of citrus (1, C. medica; 2, C. lemon; 3, C. sinensis; 4, C. aurantium; 5, F. japonica; 6, P. trifoliata); from Ollitrault et al. (2000d).

genomes in citrus somatic hybrids arising from polyethylene glycol (PEG) or electri- cally mediated protoplast fusions at the intrasubtribal level (Citrinae) (Grosser et al., 1996b; Guo and Deng, 2001). However, unexpected ploidy levels are observed among somatic hybrids for some combina- tions. Triploid plants were obtained from Severinia buxifolia (2 x) + C. sinensis (2 x) (Grosser et al., 1992b) as well as pentaploid hybrids from tetraploid Fortunella hindsii + diploid P. trifoliate (Miranda et al., 1997). A somatic hybrid between Fortunella crassi- folia cv ‘Meiwa’ and C. sinensis cv ‘Valencia’ with abnormal growth was proven to be a chimera containing non- tetraploid cells along with amphidiploids (Shi et al., 1998a; Guo and Deng, 2001). Aneuploid cells and chromosomal varia- tions were also observed in embryoids aris- ing from C. sinensis + C. reticulata Blanco hybridization (Ye et al. 1992). These results suggest that chromosome elimination could occur in citrus somatic hybrids as has been observed for other crops (Sundberg and Glimelius, 1991; Kisaka et al., 1997; Oberwalder et al., 1998). At the intertribal

level, Guo and Deng (1998) obtained sym- metric tetraploid hybrids between Citrus and Murraya, while all the plants regener- ated from C. sinensis (2 x) and Clausena lansium (2 x) were hexaploid (Guo and Deng, 1999). Froelicher (1999) also obtained unexpected polyploid hybrid embryoids from intersubtribal and inter- tribal combinations: hexaploids from Triphasia trifolia (2 x) + C. aurantifolia (2 x), and 10 x and 11 x from the 2 x C. aurantifolia

+ 6 x Clausena excavata androgenetic line. As discussed by Guo and Deng (2001), these high ploidy levels observed in wide somatic hybrids could be due to: (i) multi- fusion leading to more favourable genomic structures; (ii) ploidy variation at the level of parental embryogenic callus lines; and

(iii) chromosome doubling of either parent or global doubling of the hybrid genome fol- lowed by specifi c chromosome elimina- tions.

Ollitrault et al. (1998) observed a greater ploidy level diversity on regener- ated material from diploid + haploid somatic hybridization than that generally described for diploid + diploid combina-

tions. Indeed, triploid, tetraploid and pen- taploid hybrids were obtained from most of the 2 x + x combinations. This result could be due to the ploidy instability in the ‘hap- loid callus line’. Indeed, the presence of diploid and triploid cells in this callus has been demonstrated by flow cytometry analysis (Ollitrault et al., 1998). Furthermore, the fact that Kobayashi et al. (1997) have obtained only diploids (with one of the nuclear parental genomes) or triploid somatic hybrids by combining pro- toplasts from diploid calli and haploid leaves confi rms that triploid cells arising from haploid + diploid combinations do not present a specifi c nuclear genome insta- bility during mitosis.

Cytoplasmic genome analysis

Somatic hybridization allows combining nuclear, chloroplastic and mitochondrial genomes in new patterns with no pre- defi ned rule, unlike sexual hybridization. Moreover, organelle genome recombination (mainly mitochondria) has been observed in several crops following somatic hybridization (Belliard et al., 1979; Rothenberg et al., 1985; Galun et al., 1987; Kanno et al., 1997). Thus, the organelle genomic constitution must be characterized for each regenerated plant. For citrus plants, cytoplasmic genome analysis is classically done by RFLP (Kobayashi et al., 1991; Saito et al., 1993, 1994; Yamamoto and Kobayashi, 1995; Grosser et al., 1996a; Ollitrault et al., 1996b; Moriguchi et al., 1997; Moreira et al., 2000a, b; Cabasson et al., 2001). This method is powerful but not applicable to plants in vitro because it requires too much fresh leaf material. Thus PCR methods for cytoplasmic analysis are required for easier and earlier organelle analysis. Direct sequencing or restriction analysis of fragments amplifi ed with uni- versal primers defi ned by Demesure et al. (1995) for organelle DNA has been used for different plants (Gielly and Taberlet, 1994), and was applied to somatic hybrids of Brassica by Bastia et al. (2001).

CAPS analysis of citrus organelle genomes

Following Luro and Ollitrault (1996), Ollitrault et al. (2000c) have analysed the potential of cleaved amplifi ed polymorphic sequences (CAPS) for differentiation of organelle genomes at the intrageneric level (Citrus) and intergeneric levels within the Aurantioideae subfamily. Fourty-four geno- types from 31 species were studied combin- ing tagged PCR amplification with two mitochondrial and fi ve chloroplast univer- sal primers (Demesure et al., 1995) revised for citrus by Lotfy et al. (2002) and restricted with 12 enzymes.

From this study, the CAPS technique appears more powerful for displaying poly- morphisms for chloroplasts than for mito- chondria. Polymorphisms have only been found at the intergeneric level for mitochon- dria, while intrageneric diversity was revealed for chloroplasts. The chloroplast genomes of all cultivated Citrus species can be distinguished, except C. maxima, C. sinensis and C. paradisi for one group, and

C. aurantium and C. limon for a second group. This is in agreement with the gener- ally accepted hypothesis that sweet orange and grapefruit originated from female pum- melo interspecifi c hybridization. One combi- nation of primer pair/enzyme (trnT3/trnD2/ Dra I) allows the distinction of

C. lemon/C. aurantium plastones from those of C. maxima/C. sinensis/C. paradisi. Such differentiation was not observed in the study of plastone restriction carried out by Green and Vardi (1986). Nicolosi et al. (2000) hypothesized, from chloroplastic CAPS and nuclear genome analysis, that C. limon was derived from hybridization between C. aurantium (female) and C. medica. In the same study, including a broad range of citrus species, a very close relationship was found between C. micrantha and C. aurantifolia chloroplasts, and they are very different from the other cultivated species.

CAPS analysis has been successfully applied on interspecifi c and intergeneric somatic hybrids for mitochondrial and chloropast genomes (Fig. 10.3, Lofty et al. (2002) and Fig.10.4). This method is much

more simple, rapid and less expensive than traditional methods, and can be applied to small in vitro plants. Recently, new PCR markers for cytoplasmic genomes have been successfully developed: microsatellite markers for chloroplast (Lofty et al., 2003; Cheng et al., 2005) and SSCP markers for mitochondria (Olivares et al., in press).

Cytoplamic genomes of citrus somatic hybrids

Genetic studies on the regenerated citrus somatic hybrids and cybrids after fusion of callus-derived protoplasts with leaf-derived protoplasts demonstrate the non-segregation of mitochondrial genomes (the mitochondr- ial genome from the embryogenic parent always prevails in cybrids, as well as com- plete somatic hybrids) and segregation of chloroplastic genomes (random segregation of the chloroplast genome in both cybrids and hybrids) (Kobayashi et al., 1991; Saito et al., 1993, 1994; Yamamoto and Kobayashi, 1995; Grosser et al., 1996a; Ollitrault et al., 1996b; Moriguchi et al., 1997; Moreira et al., 2000a, b; Cabasson et al., 2001; Guo et al., 2002). In the case of fusion between callus-

derived protoplasts of the two parents, random segregation of the chloroplast and mitochondria is observed, allowing four dif- ferent cytoplasmic constitutions with the same nuclear genome (Grosser et al., 2000). Moreover, rearrangements of the cytoplas- mic genomes that have often been observed in somatic hybrids and cybrids of many plant species were also reported in some citrus studies following asymmetric (Vardi et al., 1987, 1989) and standard (Motomura and Hidaka, 1995; Moriguchi et al., 1997; Cheng et al., 2002) somatic hybridizations. Moreira et al. (2000a, b) observed non- parental mitochondrial fragments in several somatic hybrids and cybrids. Moreira et al. (2000b) also made a unique observation of the addition of choroplastic parental genomes on all the 14 ‘Succari’ + Citropsis gilletiana somatic hybrid plants analysed.

The sole presence of the mitochondrial genome from the embryogenic parent in all regenerated cybrids and somatic hybrids suggests a critical role for these organelles in plant regeneration via somatic embryogene- sis (Kobayashi et al., 1991; Saito et al., 1993; Grosser et al., 1996a; Moriguchi et al., 1996; Ollitrault et al., 1996b; Moreira et al., 2000a, b; Cabasson et al., 2001). Moreira et al.

pb

1 2 3 4 5 6 7 8 9 10 11 12 13

Fig. 10.3. CAPS analysis of chloroplast segregation in a population of somatic hybrids between ‘Star Ruby’ grapefruit and ‘Willow Leaf’ mandarin (from Ollitrault et al., 2000a). Products generated after amplifi cation and restriction using trnT3/trnD2 primers and Dra I restriction enzyme are from somatic hybrids (lanes 2–11), ‘Willow Leaf’ mandarin (lane 12) and ‘Star Ruby’ grapefruit (lane 13). Lane 1 = molecular size marker.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Fig. 10.4. Molecular analysis of G1 satsuma + HB pummelo diploid cybrid plant. Genotype identifi cation per lane: 1, 8 and 12, 100 bp DNA ladder; 2, 7, 9 and 13: satsuma mandarin cv. Guoqing No 1 (G1); 3, 6,

10 and 14: G1 + HB pummelo diploid regenerant; 4, 5, 11 and 15: HB pummelo. Lanes 1–4: SSR analysis by primer pair TAA15 confi rmed that the nuclear genome was derived from HB pummelo. Lanes 5–11: CAPS analysis of the mitochondria genome by primer pairs 18S rRNA/5S rRNA and Nad4exon1/Nad4exon2, respectively, and cut by the enzyme Taq I, which showed that the mtDNA was derived from G1. Lanes 12–15: CAPS analysis of the chloroplast genome by primer pair TrnD/TrnT and cut by the enzyme Taq I, which showed that the chloroplast DNA was derived from HB pummelo. Note: a 3% Metaphore agarose gel was used to separate the small size DNA fragments.

(2000a, b) hypothesized that this may be a quantitative effect and that only cultured cells have adequate quantities of mitochon- dria to provide the necessary energy for somatic embryogenesis. They also verifi ed that the number of mitochondria per embryogenic culture cell was signifi cantly higher than per leaf cells.