Introduction

Cytoplasmic male sterility (CMS), an important agronomic trait in crops, is controlled by the mitochondrial genome (Kumar and Cocking 1987). Protoplast fusion plays an impressive role in the transfer of CMS in higher plants, which otherwise requires at least several years to realize by sexual repetitive backcrossing. To date, numerous asymmetric somatic hybrids or cybrids have been produced as a means to transfer CMS into crop plants such as rice, rapeseed, potato, tobacco, etc. (Zelcer et al. 1978; Sakai and Imamura 1990; Creemers-Molenaar et al. 1992; Melchers et al. 1992; Akagi et al. 1995; Kirti et al. 1995; Liu et al. 1996; Cardi and Earle 1997; Sigareva and Earle 1997; Atanassov et al. 1998; Bhattacharjee et al. 1999; Akagi 2001; Motegi et al. 2003). There are two principal fusion methods to achieve CMS transfer. The first is by asymmetric protoplast fusion using irradiated protoplasts as a donor parent and chemically inactivated protoplasts as the recipient parent (Melchers et al. 1992; Sigareva and Earle 1997), and the second is by cytoplast-protoplast or microprotoplast fusion (Sakai and Imamura 1990; Spangenberg et al. 1991; Louzada et al. 2002). It is generally difficult to generate highly asymmetric hybrids or cybrids by asymmetric fusion, and the cytoplast-protoplast or microprotoplast fusion method is even more technically difficult (see above-mentioned references).

During the past two decades, great progress has been made in citrus somatic hybridization, and somatic hybrid plants from more than 200 different intergeneric and interspecific parental combinations have been produced (Grosser et al. 2000; Guo and Deng 2001). Somatic hybridization in citrus has circumvented some of the problems such as nucellar polyembryony, pollen/ovule sterility, and long juvenility encountered in sexual conventional breeding, and new hybrids are showing potential in citrus cultivar improvement (Grosser et al. 2000). For citrus somatic hybridization, the fusion model of “diploid embryogenic protoplasts + diploid mesophyll-derived protoplasts” has been used extensively. Normally, unfused mesophyll protoplasts do not divide and regenerate into plants. However, diploid plants morphologically resembling the leaf parent have been recovered unexpectedly from more than 40 symmetric fusion combinations (Grosser et al. 2000; Guo and Deng 2000; Deng et al. 2000a, 2000b). In all such cases examined, RFLP analysis showed that the regenerated diploid plants were not directly regenerated from unfused mesophyll protoplasts, but rather were cybrids with the nuclear DNA from the leaf parent and the mitochondria (mt)DNA from the corresponding embryogenic parent (Saito et al. 1993, 1994; Yamamoto and Kobayashi 1995; Grosser et al. 1996; Moriguchi et al. 1996, 1997; Moreira et al. 2000a; Cabasson et al. 2001). This indicates that the mitochondrial genome of the embryogenic parent plays an important role in mesophyll parent-type cybrid plant regeneration. Chloroplast (cp)DNA was randomly inherited in these plants. The same organelle inheritance pattern also occurs in all citrus allotetraploid somatic hybrids (Kobayashi et al. 1991; Moreira et al. 2000b; Guo and Deng 2001).

The above-mentioned phenomenon provides a novel opportunity to transfer mitochondrion-controlled specific agronomic/horticultural traits simply by conducting symmetric fusion experiments. Seedlessness is a prerequisite for new fresh-market citrus cultivars, and seedlessness in diploid citrus generally relates to male/female sterility and/or self-incompatibility (Yamamoto et al. 1995). Since most citrus species exhibit some level of parthenocarpy, seedless fruits can develop normally. The seedless Satsuma mandarin is typically male-sterile, and its male sterility has been identified to be of a CMS type (Yamamoto et al. 1997). In the investigation reported here, we designed fusions between embryogenic callus protoplasts of Satsuma mandarin cv. Guoqing No. 1 (Citrus unshiu Marc.) and mesophyll protoplasts from four seedy cultivars in an effort to produce diploid seedless cybrids. Hirado Buntan pummelo [Citrus grandis (L.) Osbeck] produces a high-quality pink-fleshed fruit that is becoming increasingly popular in Florida. However, it is excessively seedy, averaging 91 seeds per fruit (Melinda Grosser, personal communication), a trait that inhibits its consumer acceptance. Murcott tangor [C. reticulata × Citrus sinensis (L.) Osbeck] and Sunburst mandarin hybrid (Clementine × Orlando tangelo) are the two most popular tangerine cultivars in Florida. However, their high seed content is allowing seedless Clementine fruit from Spain, Morocco, and California to displace them in the marketplace. The Orie Lee hybrid (Clementine × Murcott) is an unreleased selection with exceptional quality, but it is also seedy and susceptible to citrus scab disease. Seedless fruits of any of these cultivars should provide excellent marketing opportunities to Florida and other growers.

Materials and methods

Protoplast fusion and plant regeneration

A friable embryogenic callus line of Satsuma mandarin, Citrus unshiu Marc. cv. Guoqing No 1 (abbreviated as G1) was induced from unfertilized ovules by Huo et al. (1999) and maintained on MT (Murashige and Tucker 1969) basal medium. Suspension cultures of this line were established and maintained using a 2-week subculture cycle on agar-free H+H medium (Grosser and Gmitter 1990). A recently budded plant of Hirado Buntan pummelo [Citrus grandis (L.) Osbeck] (abbreviated as HBP) and nucellar seedlings of Sunburst [C. reticulata (L.) Osbeck], Orie Lee (C. reticulata cv. Clementine tangerine × Murcott tangor), and Murcott [C. reticulata × C. sinensis (L.) Osbeck] were maintained in a heavily shaded greenhouse (using double shadecloth), and fully expanded leaves were used for protoplast isolation. The procedures for protoplast isolation, PEG-mediated fusion, and culture were according to Grosser and Gmitter (1990). Regenerated embryos were also cultured over 0.22 μm acetate plus membrane filters, which were put on fresh EME solid medium (Grosser and Gmitter 1990) containing 0.15 M maltose instead of sucrose to normalize and enlarge the embryos (Niedz et al. 2002). The regenerated shoots were grafted in vitro or onto greenhouse rootstock seedlings to expedite growth.

Ploidy analysis

Ploidy analysis was carried out using a Partec flow cytometer (Partec, Münster, Germany) according to Miranda et al. (1997) with minor modifications. Approximately 1 cm2 of young leaf was chopped in a plastic petri dish containing 0.4 ml Partec HR-A buffer. After being filtered, the samples were stained with 0.8 ml of HR-B buffer and the relative fluorescence of total DNA measured. Each histogram was generated by the analysis of at least 3,000–5,000 nuclei.

Genomic DNA extraction

For SSR and CAPS analysis, we used a Sigma GenEluteTM plant genomic DNA miniprep kit (Sigma, St. Louis, Mo.). For RFLP analysis, DNA extraction was performed according to Cheng et al. (2003a): 2–5 g of fresh healthy leaves were harvested and ground in liquid nitrogen, mixed with SDS extraction buffer and incubated at 65°C. After removal of the RNA, the crude DNA solution was then extracted with water-saturated ether and NaCl and precipitated with isopropanol; the DNA was then dissolved in TE buffer.

SSR and cp-SSR analysis

SSR analysis of nuclear genomes using three primer pairs (Table 1) and cp-SSR analysis by one pre-screened primer pair (NTCP9, primer 1: 5′-CTT CCA AGC TAA CGA TGC-3′; primer 2: 5′-CTG TCC TAT CCA TTA GAC AAT G-3′) (Cheng et al. 2003b) from the cp-SSR primer pairs list described by Bryan et al. (1999) were conducted according to the procedure of Kijas et al. (1997) with minor modifications. Approximately 20 ng of genomic DNA, 0.2 mM dNTPs, 1.5 mM MgCl2, 0.5 U Taq DNA polymerase (Promega, Madison, Wis.) and corresponding 1× reaction buffer and 0.2 μM of each primer pair were mixed in a 25-μl reaction volume. PCR cycles were programmed as follows: one initial denaturing cycle at 94ºC for 3 min; 32 cycles of 1 min at 94ºC (denaturing), 40 s at 55ºC (annealing), 2 min at 72ºC (elongation); a final cycle of 10 min at 72ºC. The products were analyzed on 3.0% (w/v) Metaphore agarose (BMA, Rockland, Me.) gels and visualized under UV light.

Table 1 SSR primer pairs used to verify the nuclear genome origin
Full size table

CAPS (or PCR-RFLP) analysis

Organelle DNA was amplified using three chloroplast and three mitochondrial universal primer pairs (Table 2). The PCR reactions were conducted in a PTC-100 thermocycler. The reaction mixture (50 μl) consisted of 0.2 mM dNTPs, 1.5 mM MgCl2, 2 U Taq DNA polymerase (Promega) and corresponding 1× reaction buffer, 0.2 μM of each primer, and 100 ng of sample DNA. PCR conditions of the CAPS analysis were the same as that of the SSR analysis. The PCR products were digested with 5 U of restriction enzymes (NEB, Beverly, Mass.), including DraI, EcoRI, FspI, HaeIII, HindIII, HinfI, PstI, SacII, or TaqI, and electrophoresed on 3.0% Metaphore agarose gels.

Table 2 Mt-DNA and cp-DNA universal primer sequences and references
Full size table

RFLP analysis

Approximately 10 μg of genomic DNA was digested with EcoRI, HindIII, BamHI, DraI, and PstI. The digested DNA was electrophoresed on a 0.8% agarose gel and blotted onto a Hybond-N+ membrane (Amersham, UK) using an alkali-downward capillary blotting procedure according to the manufacturer’s recommendation. Probe labeling, hybridization, and stringency washing were conducted using the procedure described by Feinberg and Vogelstein (1983). The mtDNA probes—aptA, cob, cox I, cox II, and cox III—were labeled with d(CTP)-[32P]. Hybridization was performed in tubes at 65ºC overnight, and membranes were washed in high-stringency solution (0.1× SSC, 0.1% SDS) at 65°C for 4 h. The membranes were exposed to X-ray film at –80ºC overnight to 1 week before the film was developed.

Results and discussion

Fusions were conducted combining embryogenic suspension protoplasts of G1 with mesophyll protoplasts from HBP, Sunburst, Orie Lee, and Murcott, respectively. The G1 callus line had been in culture for several years and had lost its capacity for direct protoplast regeneration. Embryos were recovered from all four fusion combinations and transferred to 0.22 μm acetate plus membrane filters placed directly on fresh EME solid medium containing 0.15 M maltose instead of sucrose (Perez et al. 1998) to expedite growth and development (Niedz et al. 2002). Great differences in embryo and plantlet recovery were found from the four fusion combinations. Only one embryo was germinated from the G1 + HBP fusion combination, and approximately 20, 5, and 2 shoots were produced from the G1 + Sunburst, G1 + Orie Lee, and G1 + Murcott combinations, respectively (Table 3). The regenerated shoots were then grafted in vitro or grafted directly onto greenhouse rootstocks of a sour orange + Carrizo citrange somatic hybrid to expedite their growth.

Table 3 Parental combinations and number of plants regenerated
Full size table

Flow cytometry (Fig. 1) analysis confirmed that the regenerated plants of the G1 + HBP, G1 + Sunburst, and G1 + Orie Lee fusion combinations were all diploids, whereas the two plants from the G1 + Murcott combination were tetraploid. In addition, the regenerated plant from G1 + HBP was morphologically identical to HBP, which suggested its putative cybrid nature. The diploid plants from the other two fusion combinations also exhibited a leaf morphology similar to that of their respective leaf parents.

Fig. 1
figure 1

Ploidy determination by flow cytometry analysis. Peak 1 Diploid (G1 + HBP, G1 + Sunburst, G1 + Orie Lee leaf mixture, peak 2 tetraploid (G1 + Murcott)

Full size image

To determine the origin of the nuclear genotype, we conducted SSR analysis. The results showed that the regenerated diploids from the three different fusion combinations had a banding pattern identical to that of their corresponding leaf parent, indicating that the regenerated diploid plants were putative cybrids with the nuclear genome from their corresponding leaf parent (Figs. 2, 3). The regenerated tetraploid from G1 + Murcott had complementary specific bands from both parents, thus confirming its somatic hybrid nature (Fig. 4A). This hybrid should have excellent potential to serve as a breeding parent for use in interploid crosses to produce seedless triploids (Grosser et al. 2000).

Fig. 2
figure 2

SSR analysis of G1 + HBP diploid cybrid plant. Lanes: 1 100-bp ladder; 2, 5 G1; 3, 6 G1 + HBP diploid cybrid plant; 4, 7 HBP. Primer pairs: lanes 2–4 TAA15, lanes 5–7 TAA41. Arrows indicate specific bands

Full size image
Fig. 3
figure 3

SSR analysis of G1 + Sunburst and G1 + Orie Lee hybrid diploid regenerated plants. Lanes: 1, 4, 8, 11 G1; 2, 5 G1 + Sunburst diploid plant; 3, 6 Sunburst; 7100-bp ladder; 9, 12 G1 + Orie Lee hybrid diploid plant; 10, 13 Orie Lee hybrid. Primer pairs: lanes 1–3, 8–10 TAA15, lanes 4–6, 11–13 TAA41. Arrows indicate specific bands

Full size image
Fig. 4
figure 4

SSR (A) and mt-DNA (B) analysis of G1 + Murcott tetraploid somatic hybrid. Lanes: 1, 8 100-bp ladder; 2, 5, 9, 12 G1, 3, 6, 10, 13 G1 + Murcott tetraploid plant; 4, 7, 11, 14 Murcott. SSR primer pairs: lanes 2–4 TAA15, lanes 5–7 TAA41. Mt-DNA primer pair/enzyme combinations: lanes 9–11 nad4Exon1-nad4Exon2/TaqI, lane 12–14 18S rRNA-5S rRNA/TaqI. Arrows indicate specific bands

Full size image

To further confirm the cybrid nature of the diploid regenerates, we carried out CAPS, RFLP, and cp-SSR analysis. CAPS analysis results showed that mt-DNA primer pair/enzyme combinations of nad4exon1-nad4exon2/TaqI, nad4exon1-nad4exon2/HaeIII, nad4exon1-nad4exon2/HinfI, and 18S rRNA-5S rRNA/TaqI were able to distinguish both parents of G1 and HBP and that the mtDNA of the G1 + HBP diploid plant was always from G1 (Fig. 5, lanes 1–6). RFLP analysis of the mtDNA further confirmed the results of CAPS analysis (Fig. 6A). For cpDNA, the primer pair/enzyme combinations of rbcL-psaI/TaqI and trnD-trnT/TaqI were effective in distinguishing both parents, and the cpDNA of G1 + HBP diploid plant was verified to be from HBP (Fig. 5, lanes 8–13); cp-SSR analysis further confirmed this inheritance pattern (Fig. 6B). With respect to the inheritance of the mitochondrial genome of the G1 + Murcott tetraploid somatic hybrid, the mtDNA primer pair/enzyme combinations of nad4exon1-nad4exon2/TaqI and 18S rRNA-5S rRNA/TaqI were able to distinguish both parents; the mtDNA of the hybrid was always from G1 (Fig. 4B). We could not distinguish the cpDNA of this fusion combination by either CAPS or cp-SSR analysis. In addition, CAPS analysis with the primer pair/enzyme combinations used could not identify the origin of the cytoplasmic genomes in the G1 + Sunburst and G1 + Orie Lee diploid regenerates due to the close phylogenetic relatedness of both fusion parents, although they were thought to be cybrid plants. CAPS analysis has only recently been applied to identify the cytoplasmic inheritance of somatic hybrids in higher plants and has proven to be very efficient (Bastia et al. 2001; Guo et al. 2002; Liu et al. 2002; Cheng et al. 2003a; Lotfy et al. 2003). Chloroplast microsatellite (cp-SSR) analysis is a new tool for plant ecology and evolution studies (Provan et al. 2001) and has only recently been introduced into citrus phylogenetic study (Cheng et al. 2003b). Compared with CAPS analysis, it is even more convenient and efficient since enzyme cutting following PCR reaction is not needed.

Fig. 5
figure 5

Mt- and cp-DNA analysis of G1 + HBP diploid cybrid plant by CAPS markers. Lanes: 1, 4, 8, 11 G1; 2, 5, 9, 12 G1 + HBP diploid cybrid plant; 3, 6, 10, 13 HBP; 7, 14100-bp ladder. Primer pair/enzyme combinations: lanes 1–3 nad4Exon1-nad4Exon2/HaeIII, lanes 4–6 nad4Exon1-nad4Exon2/HinfI, lanes 8–10 rbcL-psaI/TaqI, lanes 11–13 trnD-trnT/TaqI. Arrows indicate specific bands

Full size image
Fig. 6
figure 6

RFLP analysis of mt-DNA (A), and cp-SSR analysis of cp-DNA (B) of the G1 + HBP diploid cybrid plant. Lanes: 1 λ/HindIII marker; 2, 6 G1; 3, 7 G1 + HBP cybrid plant; 4, 8 HBP; 5 100-bp ladder. Enzyme/probe: HindIII/atpA. Cp-SSR primer pair: NTCP9. Arrows indicate specific bands

Full size image

The molecular analysis confirmed that we had produced not only a cybrid plant between G1 and HBP with mtDNA from G1 but also many putative cybrid plants between G1 and Sunburst tangerine and G1 and the Orie Lee hybrid. Additional plants from G1 + Sunburst and G1 + Orie Lee were tested by flow cytometry and SSR analysis. The results also showed that they were all diploids and that their nuclear genome was always from their corresponding leaf parent. There is also strong circumstantial evidence that the plants regenerated from these latter two combinations are also cybrids. In the more than 25 cases of citrus callus/suspension + leaf protoplast fusion experiments examined where diploid plants of leaf-parent morphology were recovered, all were confirmed as cybrids containing the nuclear DNA of the leaf parent but only the mtDNA of the callus/suspension parent (Saito et al. 1993; Yamamoto and Kobayashi 1995; Grosser et al. 1996; Moriguchi et al. 1996; Guo and Deng 2000; Moreira et al. 2000a; Cabasson et al. 2001; Y.J. Cheng, unpublished data; M.T. Scarano, personal communication). The same phenomenon has been shown for allotetraploid somatic hybrids, as all of the regenerated plants examined always contain only the mtDNA of the callus/suspension parent (Kobayashi et al. 1991; Ohgawara et al. 1994; Moreira et al. 2000b). The current research demonstrates that it is possible to take advantage of this phenomenon to transfer CMS from Satsuma mandarin to other seedy cultivars simply by symmetric fusion. This method is easier and more efficient than cytoplast-protoplast fusion or asymmetric fusion, which are technically complicated with difficult cybrid plant recovery.

Several possible mechanisms leading to the regeneration of cybrid plants by symmetric fusion have been proposed by various researchers (Gleba and Sytnik 1984; Grosser et al. 1996; Moriguchi et al. 1996; Guo and Deng 2000; Moreira et al. 2000a), but none are conclusive. Moreira et al. (2000a) proposed that quantitative differences in the mitochondria content of callus/suspension cells versus leaf cells could be responsible. Leaf cells and corresponding protoplasts contain much fewer mitochondria per cell than callus/suspension cells and may not be capable of producing the energy required for direct regeneration. Cytoplasmic fusion with callus/suspension cells may provide the necessary mitochondria and energy to achieve regeneration. In such cases, it is not known what happens to the callus/suspension donor cell nuclei. Our experiences with Citrus indicate that the phenomenon of cybrid production by symmetric fusion is dependent on the genotype of the embryogenic parent and the parental combination and that the possibility of obtaining cybrid plants is higher when higher ratios of mesophyll protoplasts to callus/suspension cells are used in the fusion.

Several seedless sexual hybrid cultivars have been obtained and released with Satsuma mandarin as the seed parent in China and Japan. One cybrid plant between Juman Satsuma and Washington navel orange has also been produced (Yamamoto and Kobayashi 1995; Yamamoto et al. 2001). Since Washington navel orange is a typical seedless cultivar just as Satsuma, it is difficult to determine if the seedlessness of the cybrid plant is due to sterile cytoplasm from Satsuma. In our investigation reported here, we were able to produce cybrid plants between G1 Satsuma and other seedy cultivars within a year; a task that would require years by conventional repetitive backcrossing—if it is possible at all. In this sense, the transfer of CMS by cell fusion in citrus may be even more efficient than in other annual crops. These cybrid plants have been top-worked to mature trees to expedite flowering and fruiting, which is necessary to determine the practical value of this approach. It must still be determined if the substitution of Satsuma mtDNA in the present combinations can result in new functional mitochondria-nuclear interactions that could result in making these cultivars seedless without otherwise altering their cultivar integrity. Any resulting seedless fruit with expected cultivar integrity would have instant marketability. If successful, this strategy could be applied to eliminate seed production from many superior diploid citrus cultivars. In addition, even if the cybrid HBP containing the Satsuma CMS cytoplasm turns out to be seedy, it will still have value as a mono-embryonic female parent that could be used in crosses with mandarins of the appropriate nuclear background to produce seedless diploid tangelos.