Litchi (Litchi chinensis Sonn.), known as “the queen of fruit” (Menzel and Waite, 2005), is an important fruit tree in the tropical and subtropical regions of the world (Menzel, 1983). Given its long reproductive cycle and highly heterozygous genetic background (Litz, 1988; Raharjo and  Litz, 2005), new litchi cultivars are difficult to create via conventional breeding methods. Modern breeding techniques such as gene manipulation have the advantages of high efficiency and directional improvement of specific traits, providing a new way for the improvement of litchi cultivars (Das and Rahman, 2012). The  establishment  of  a  regeneration   system  invitro for litchi bioengineering breeding is vital. Many efforts have been made regarding the subject, but few successful results have been published. Fu and Tang (1983) reported that they obtained 24 plantlets via organogenesis from pollen callus. Subsequent reports were made on somatic embryogenesis and plantlet regeneration in litchi via the culture of anthers (Deng, 2005; Xie et al., 2006; Wang et al., 2013), immature embryos (Zhou et al., 1993, 1996; Kuang et al., 1997), or protoplasts (Yu et al., 1996). Das et al. (1999) reported a high proliferation rate from mature seeds using two kinds of methods. Yu et al. (2000) achieved somatic embryogenesis and plantlets from ‘Xiafanzhi’ litchi protoplasts isolated from embryogenic suspensions. Puchooa (2004) used young unvaccinated ‘Tai So’ leaves as explants to study various factors on the regeneration of the leaf blade and eventually obtained regenerated plants. By ‘Heli’ anther culture, Guo et al. (2014) obtained somatic embryos with root and no sprout. However, Huang and You (1990) and Yu (1991) and Guo et al. (2014) reported that the stem sprout buds, failed to produce (Huang and You, 1990; Yu, 1991). All the above results showed that the regeneration of different litchi varieties is inconsistent with their medium. The same culture medium plays different roles on the regeneration of different litchi varieties. Therefore, different genetic backgrounds of litchi varieties have a remarkable effect on in vitro regeneration ability. To date, only a few cultivars, such as ‘Nuomici’ (Kuang et al., 1997), ‘Xiafanzhi’ (Lai and Sang, 2003) and ‘Hushanjiaohe’ (Fu and Tang, 1983), have been successfully regenerated in vitro. ‘Feizixiao’ is an early maturity variety with a tender, juicy, sweet aril and high and stable yield. In vitro regeneration of ‘Feizixiao’ is potentially a valuable method for conservation, mass propagation, and genetic transformation of this species. Although ‘Feizixiao’ micropropagation has been described (Deng, 2005), no report exists regarding its in vitro plant regeneration by a somatic embryogenesis system.

Plant materials and callus induction

Branches with floral buds of 10-year-old ‘Feizixiao’ were collected from the base of Hainan Leihu Fruit Ltd. on March 12, 2011. Immature flowers were washed in running tap water to remove dust. They were then dipped into 70% (w/v) ethanol, and 0.1% (w/v) HgCl₂, followed by three rinses in sterile distilled water under aseptic conditions. The anthers were separated from the flowers for callus induction.

The anthers were cultured on MS (Murashige and Skoog, 1962) solid (containing 0.7% (w/v) agar) medium with 0.4 g/L LH and 30  g/L   sucrose.   The  medium  was  supplemented  with  different concentrations and combinations of plant growth regulators [PGRs: BA (0, 0.89, 2.22, and 4.44 µM), KT (0, 2.32, 4.65, and 9.29 µM), NAA (0, 1.07, 2.69, and 5.37 µM), 2,4-D (0, 4.52, 9.05, and 13.57 µM)] for callus induction. An orthogonal experimental design was used for these experiments with three repeats for each treatment, seven bottles per repeat, and 9 to 11 anthers per bottle. The callus was record after 8 weeks of culture. The callus obtained from anthers was subcultured in MS medium containing 4.52 µM 2,4-D to obtain synchronized friable embryogenic callus.

Somatic embryogenesis

The 18-day-old friable embryogenic calli (light, yellow, vigorous, fine-grained) were transferred to MS medium supplemented with 6% (w/v) sucrose, 0.4 g/L LH, 0.56 µM inositol, 10% (w/v) coconut water, and a combination of auxin NAA (0–0.54 µM) and cytokines KT (13.94, 23.23, and 32.53 µM), ZT (13.68, 22.81, and 31.93 µM), and TDZ (13.62, 22.71, and 31.79 µM). A completely randomized design with 10 replications (0.1 g of fresh callus per replicate) for each treatment was used for these experiments. Different developmental stages of SEs (>0.3 cm) were recorded after 7 weeks of culture.

Plant regeneration

Different developed stages of SEs (>0.8 cm) were transferred to different MS basal media (MS, 1/2 MS) supplemented with various concentrations of GA₃ (0, 1.44, 2.89 and 5.77 µM) and 3% (w/v) sucrose for regeneration. A completely randomized design with 15 replications (4 to 6 explants per replicate) was used for these experiments. Regeneration was recorded after 9 weeks of culture. These explants were incubated under 16 h (lightness)/8 h (darkness) photoperiod (provided by cool-white fluorescent lamps at a photon flux of 27 µmol/m²s¹) at 26±2°C.

All media were adjusted to pH 5.8 prior to autoclaving at 121°C for 20 min. All explants were incubated in the dark, unless stated otherwise.

Flow cytometry and chromosome counting

The ploidy level of regenerated plantlets was analyzed using flow cytometry. Root tips were chopped and measured using a Cell Lab Quanta SC (Beckman Coulter, CA, USA). Nuclei obtained from the root tip of air layering seedlings were used as a diploid control. Histograms of DNA content were evaluated using Cell Lab Quanta SC MPL Analysis software (Beckman Coulter, CA, USA).

Chromosome counts were carried out on root tip cells of the plantlet pretreated for 3 h with a saturated solution of 1,4-dichlorobenzene. Roots were fixed in 3:1 ethanol-glacial acetic acid, hydrolyzed in 1 M hydrochloric acid, and rinsed three times with distilled water. Each rinsed root tip was placed on a glass slide. The root cap was removed, and the remaining root was stained with a carbol fuchsin solution. Each root tip was moderately squashed as the cover slip was applied to help spread the chromosomes. Micrographs were taken with an FSX-100 microscope camera system (Olympus, Tokyo, Japan).

Specimen preparation for light microscopy

According to the methods described by He (1989) and Lee and Mu (1966), sections of friable embryogenic callus, white opaque SEs, non-friable embryogenic callus, different abnormal SEs, and vitrified somatic embryos were fixed in FAA for 20 h. The specimens were then rinsed  three  times  in  70%  (w/v)  ethanol.  For  staining,  the specimens were stained in Ehrilich’s dyeing liquid. The specimens were dehydrated in a range of different concentrations of dimethylbenzene and embedded in different concentrations of paraffin. The specimen block was roughly trimmed to exclude excess plastic matrix and expose the surface of the specimen. The roughly trimmed block was mounted into a Reichert Ultracut microtome, and 5 to 6 µM sections were cut using glass knives.

Induction of embryogenic callus

Embryogenic calli were achieved from anther explants on MS medium, 3% (w/v) sucrose, and various PGRs in combination. One callus formed at the yellowish anthers after approximately 3 weeks of culture, other calli developed in all treatments, except the control, after 6 weeks (Figures 1a to c). The first callus was more light yellow in color, heavily watery, and non-granular (Figure 1d). The other was non-friable callus. Other calli emerged after 6 weeks. These calli were light yellow, compact, friable, tiny and/or nodular, and pre-embryogenic (Figures 1e to g). Amin and Razzaque (1995), Yu and Chen (1997), Puchooa (2004), Rahario and Litz (2007), Ma et al. (2009), Xu and Lai (2013), and Guo et al. (2014) also reported a similar type of callus development in other litchi cultivars. Similar phenomena were also found in longan (Lai et al., 1997), ‘Premier’ honey peach (Lai et al., 2008), Heavea (Tan et al., 2009), and Pinus koraiensis (Wang et al., 2015).

Different combinations of PGRs were used to test their effect on the induction of embryogenic callus. The experiment revealed significant differences in the number of callus produced among treatments, indicating that the combination and concentration of PGRs exerted a significant effect on callus induction (Table 1). LSD test revealed that a significantly higher number of callus was obtained when anthers were cultured in medium L9 (2.22 µM BA, 2.69 µM NAA, and 13.57 µM 2,4-D), followed by medium L8 (0.89 µM BA, 9.29 µM KT, 2.69 µM NAA, and 4.52 µM 2,4-D) and medium L3 (4.65 µM KT, 2.69 µM NAA, and 9.05 µM 2,4-D). No significant difference was observed among L4, L12, L14, and L15, which produced fewer FECs compared with the other media. Medium L1, which contained no added PGRs, produced no callus. Interestingly, medium L12, in which callus induction was low, produced callus similar to those in medium L9. These observations suggested that certain auxin-cytokinin ratios in medium may be necessary for callus induction from anther culture of litchi. In this study, R value revealed that 2,4-D was most effective in FEC induction, followed by NAA, BA, and KT. The effect of 2,4-D on FEC induction was also observed in other litchi species (Fu and Tang, 1983; Zhou et al., 1993). Yu and Chen (1997) reported that induction of embryogenic callus from immature embryos after NAA, instead of 2,4-D, was a failure. Su et al. (2004) showed that the callus in medium without 2,4-D did not survive the following experiment. Many species confirmed that FEC induction is influenced by 2,4-D, such as in pine (Yu et al., 2011), in longan (Lai et al., 1997), and Hevea brasiliensis Mull. Arg (Huang et al., 2014).

In the same medium, continuous EC subculture will lead to embryonic loss, thus, the addition of KT and AgNO₃ in the medium synchronized embryogenesis (Lai et al., 1997). The regulated EC differentiated into somatic embryogenesis after 18 days of culture. In the present study, three  different combinations of PGRs were used in SE formation.

Somatic embryo formation

Somatic embryogenesis was achieved from EC on MS medium, 6% (w/v) sucrose, and various PGRs alone or in combination. The response of SEs to three different media was extremely variable, especially at five weeks after culture initiation (Figures 1 and i). Simon et al. (2007) also reported a similar type of embryo development in ‘Brewster’. Duncan’s multiple range test indicated that the medium supplemented with KT and NAA produced a significantly higher number of SEs compared with the other two media; the average SE induction numbers were 489, 344 and 101 FW/g in medium supplemented with KT and NAA, ZT and NAA, and TDZ and NAA, respectively. In this study, SE shapes varied after 7 weeks in treatments (Table 2). The embryo development stages observed were double cotyledon, no cotyledon embryo, single cotyledon, multi-cotyledon embryo, flake  embryo, conjoined embryo, goblet embryo, trumpet embryo, and mass embryo (Figure 2). Similar patterns of embryo development stages were reported for other plant species (Kuang et al., 1996; Tan et al., 2011).

These experiments resulted in significant differences in the   number    of    SEs    produced   among   treatments, indicating that the combination and concentration of PGRs had a significant effect on SE yield (Table 2). For double cotyledonary embryos, Duncan’s multiple range test revealed a significantly higher number of SEs when FECs were cultured in medium T3 (0.54 µM NAA and 23.23 µM KT), followed by medium T7 (0.54 µM NAA and 22.81 µM ZT) and medium T2 (0.54 µM NAA and 13.94 µM KT). No significant differences were observed among T8, T9, T10, T11 and T12 media which produced fewer SEs compared with the other media. Interestingly T1 and T4, with only cytokinin (KT or ZT) and no auxin (NAA) added, produced smaller and fewer SEs but medium T9 displayed contrasting results. These observations suggested that certain auxin-cytokinin ratios in medium may be necessary for SE formation.

Plant regeneration

Successful plant regeneration on the plant induction medium was achieved from post-somatic embryogenetic developmental stages: Globular, heart, torpedo, cotyledonary, and plumule embryos. Most double cotyledonary and conjoined mature embryos germinated to produce complete plantlets after 4 weeks in culture. Most abnormal embryos failed to germinate on the entire medium.

Half-strength solid MS medium produced a higher frequency of rooting (23.67%) but a lower frequency of germinated plantlets, whereas MS medium produced a lower frequency of rooting (13.78%) and well-germinated plantlets (1.11%) (Table 3). The frequency of rooting in half-strength solid MS medium was higher than that in full-strength solid MS medium in Pyrus communis L. (Predieri et al., 1999). Different salt concentrations would change the osmotic pressure, thereby affecting nutrient absorption and release substances into the medium. Perhaps the decrease in the concentration of nitrogen in half-strength solid MS medium stimulates rooting.

Meanwhile  2.89 µM  GA₃ produced higher frequency of rooting (68.33%) but lower germination (1.33%) in half-strength solid MS medium, whereas full-strength solid MS medium produced lower frequency of rooting (19.67%) and well germinated plantlets (6.67%). The beneficial effect of gibberellins on somatic embryo germination has been reported in other plant species. For example, in Clematis, enhanced frequency of shoot development and greater internode elongation resulted from the influence of GA₃ (Wang et al., 2014). Compared with other studies on shoot bud differentiation such as Quercus (Vengadesan and Pijut, 2009) and Ammopiltanthus mongolicus (Yang et al., 2014), in this paper, the tap root of litchi produced black spots (1.5 cm), which ultimately led to plant death. Thus, GA₃ promoted rooting, inhibited root growth, and suppressed lateral root growth.

Successful germination and plant regeneration via SEs have been reported with combinations of KT and NAA in ginger and buffel grass (Lincy et al., 2009; Carloni et al., 2014). After germination, GA₃ is required for plant development in buffel grass. Well-developed plantlets were separated (Figure 1l) and then transferred to plastic pots containing a 1:1:1 (v/v/v) vermiculite: sand: coconut chaff mixture. The plantlets were successfully acclimatized in a greenhouse (Figure 1m) with a 100% survival rate.

The present work is the first report on plant regeneration via indirect somatic embryogenesis from anther explants of ‘Feizixiao’ litchi. Promising plant regeneration from SEs via indirect somatic embryogenesis was remarkably influenced by PGRs. The system will be helpful for conservation, mass clone propagation, production of bioactive compounds, and genetic transformation studies.

Ploidy level of regenerated plant

All the plantlets were diploid. Ploidy levels of regenerated plants were determined by flow cytometry, and chromosome numbers were determined by chromosome counting  (Figures  1n  and   o).   ‘Feizixiao’   litchi  has  a chromosome numbers of 2n = 2x = 30. The same chromosome numbers were confirmed in the diploid plants.