Development of the first axillary in vitro shoot multiplication protocol for coconut palms

The coconut palm or “tree of life” is one of nature’s most useful plants and the demand for its fruit is increasing. However, coconut production is threatened by ageing plantations, pests and diseases. Currently, the palm is exclusively propagated via seeds, limiting the amount of planting material. A novel micropropagation method is presented, based on axillary shoot formation. Apical meristems of in vitro coconut seedlings are cultured onto Y3 medium containing 1 µM TDZ. This induces the apical meristem to proliferate through axillary shoots in ~ 27% of the initiated explants. These axillary shoots are seen as white clumps of proliferating tissue and can be multiplied at a large scale or regenerated into rooted in vitro plantlets. This innovative micropropagation method will enable the production of disease-free, high quality in vitro plantlets, which will solve the worldwide scarcity of coconut planting material.


Results
The effect of cutting in the presence or absence of TDZ on inducing proliferating meristems. Forty-five days after initiation, 36 of the 141 (25.5%) explants that were subjected to the "cut protocol" combined with culture on 1TDZ, proliferated by exhibiting a white enlarged meristematic zone with many multiplying meristems arising from the former center of the plantlet (Fig. 1). These meristematic zones always originated from only one of the two halves that were initiated; the other part blackened and died, irrespective of the culture medium. The plantlets of the two other treatments ("control treatment" + 1TDZ and "cut protocol" + Y40) did not show any signs of proliferation. The Scanning Electron Microscope (SEM) picture shows the formation of new meristems at the base of older meristems in a structured (spiral-like) pattern (Fig. 2), pointing at the axillary origin of the side meristems. www.nature.com/scientificreports/ The effect of different cytokinins or cytokinin-like PGRs on inducing proliferating meristems. All 3 tested compounds were able to induce meristematic proliferation after 45 days (Table 1), however only 1 µM TDZ and 10 µM CPPU resulted in a proliferation rate significantly higher than treatments resulting in no proliferation. It was also observed that 8.3% of plantlets that were transferred to 100 µM BA produced proliferating meristems but also that 41.7% died, which is significantly higher than the mortality rate of the other treatments (6.3%).
The effect of the "meristem protocol" vs "cut protocol". After    www.nature.com/scientificreports/ Proliferation of six cultivars. No significant differences were observed between the responses of the six cultivars (Table 2). All showed comparable proliferation (average 27.1%) and mortality (average 13.5%) rates after 45 days. Also the type of the coconut variety (tall versus dwarf) did not influence their reaction. We observed that the proliferating material of one of the cultivars, MVT, showed a pink coloration ( Fig. 3) but no other disfigurations or colorations were observed.
Subculture of proliferating material. The proliferating meristem clumps could be maintained, in their proliferating state, on the 1TDZ medium over the course of 3 years (Fig. 3) provided they were subcultured monthly, as the tissue tended to become brown if left on the same culture medium for longer periods. During their maintenance, we observed that the size of each meristem clump almost doubled every month.
Regeneration. Over the course of 2 years, samples were regularly isolated from the continuous proliferating meristems for regeneration. Each regeneration attempt, resulted in rooted in vitro plantlets, regardless if the samples were taken directly after initiation or after two years of subculturing on the 1TDZ proliferation medium. Eight to 70% of the meristems present per ~ 0.8cm 2 , measured at the base, sized proliferating clumps, depending on the clone, regenerated into shoots that could be rooted (Fig. 4). As such, 5 to 18 cloned plantlets were produced on Y40 AC medium per clump after six to eight months (Fig. 4). The remaining 30 to 92% meristems per clump that did not develop into normal shoots showed flowerlike structures (Fig. 5) or became necrotic and died.

Discussion
The development of a reliable and fast clonal multiplication technique for coconut palms is essential to meet the increasing industry demand for quality planting material. This would allow breeders and nurseries to bring large numbers of cloned plantlets onto the market, that are free of diseases and have moreover superior characteristics. In addition, this would also greatly help conservation efforts of coconut genetic resources through in vitro conservation and cryopreservation 32 . For more than 50 years, the search for an efficient clonal propagation method for coconut has been the subject of many studies. This has initially led to the development of various SE protocols 33 . However, somatic embryogenesis still bears major challenges that need to be resolved 17,19,34 before it can be used in a commercial setting. This study thus focuses on the development of an alternative micropropagation method,  www.nature.com/scientificreports/ that would mitigate some of the problems associated with SE such as somaclonal variation, low induction rates and cultivar dependence 19,32,35 . In our first experiment, we tested the hypothesis that latent axillary meristems could develop into shoots provided the apical dominance of the apical shoot was eliminated. A first method consisted of cutting the explant, that contains one apical meristem, in two equal pieces. This cut would either damage the apical meristem irreversibly, provided it went exactly through the apical meristematic dome or remove the meristem from one of the two pieces. The second method involved the addition of a cytokinin-like PGR (TDZ) to the culture medium. In banana, another monocot species, TDZ already proved to be very powerful in reducing the apical dominance resulting in highly proliferating meristem clumps 36 . While these two procedures, combined or alone, have been successfully applied in many plant species, both monocotyledons and dicotyledons, e.g. banana 36 , ensete 37 & pitaya 38 , these techniques proved to be ineffective in coconut. Sisunandar and coworkers developed an embryo incision method that when combined with the addition of cytokinins in the medium, resulted in only two cloned plantlets per embryo 39 . Our results significantly differed from this report in two ways; firstly the plantlets that were cut in two, both in the presence and absence of TDZ, always resulted in the death of one part. Secondly, in 25% of the plantlets that were treated with the "cut + TDZ" treatment, a new structure appeared, i.e. a cluster of meristems, somewhat comparable to a cauliflower structure.
As this meristem cluster appears only in one of the two halved shoots, the apical meristem is likely the meristem cluster's source. This statement is also supported by the location of the meristem cluster, which is in the former center of the explant (Fig. 1). In case the meristem cluster would have originated from a somatic cell (de novo organogenesis) or from a "sleeping" axillary meristem, in the absence of the apical dominance, both halves should have formed new meristems.
While these meristem clusters share some general visual similarities with the structures that have previously been generated through coconut SE, such as their white-opaque-yellow lobbed exterior 40,41 , there are key differences between the two tissues. The meristem clusters have a more uniform white color and are not as transparent. Also they are structurally more orderly arranged (Fig. 2), compared to the somatic embryo clusters. This difference can be explained by their different origins; somatic embryos do appear more randomly, while the organized pattern observed in the meristem cluster is caused by the formation of new axillary meristems at the basis of the older apical meristems. Such structures are also reported in two other monocotyledonous species; date palm bud cultures obtained by Sidkey 42 and banana shoot clusters by Strosse and coworkers 36 .
We observed that 100 µM BA was also able to induce the same structures as with much lower concentrations of TDZ. This observation strengthens the hypothesis that TDZ, along with its direct cytokinin-like activity, acts upon multiple pathways in the plant 29 as well as confirming earlier reports claiming that TDZ has a stronger cytokinin-like activity compared to adenine based cytokinins such as BA, kinetin and zeatin 36,43 . Moreover, the high BA concentrations needed to induce meristem clump formation in coconut proved to be more toxic compared to TDZ. We also observed (results not shown) that prolonged culture of the meristem clumps on 100 µM BA leads to severe tissue necrosis in line with other reports 36,44 .
The cytokinin like activity of TDZ 45 is explained through the direct interaction of TDZ with cytokinin receptors 46 and the inhibition of endogenous cytokinin degradation 47 . CPPU, another phenylurea derivative, which was found to be a more potent inhibitor of cytokinin degradation compared to TDZ in maize 47 , resulted in our study in a lower amount of meristem clumps being formed, when using similar concentrations. This suggests that a direct interaction might be TDZ's main course of action in coconut, as otherwise the explants treated with CPPU would result in a similar or even higher proliferation rate. However, earlier comparisons between CPPU and TDZ show that their relative potency might be plant and tissue dependent 48,49 .
Since no significant difference in frequency of axillary shoot formation was observed between the "meristem" and "cut" method, shoot proliferation originates from the intact apical meristem. The facilitated uptake or transport of PGRs to the apical meristem are thus the main factors in successfully inducing proliferating meristems. The full sized in vitro plants, containing many "buffering" tissue layers, were not successfully induced, which was probably due these layers diluting or restricting the PGR flow to the meristem.
The meristematic induction rate of the six different cultivars under investigation, did not significantly differ and was on average 27%. As these cultivars represent the genetically distinct tall and dwarf types, it is expected that the protocol is applicable to more if not most coconut palms. It would therefore also be worthwhile to investigate if this method would be applicable other members of the palm family (Arecaceae) as in vitro shoot multiplication protocols are also needed in date and oil palm industries.
This successful regeneration of proliferating meristems into rooted in vitro plantlets was obtained even after 2 years of subculture and multiplication. This proves that the morphology of the proliferating meristems is maintained for multiple years resulting in a virtually unlimited source of clonal plant material.
While each regenerated meristem clump produced 5 to 18 normal in vitro plantlets, after 7 to 8 months, on average 52% of the meristems grew into flowerlike structures. The latter resemble the coconut flower initiation stages described by Krisanapook and coworkers 50 and are probably initiated by TDZ as this PGR is known to induce flowering in some plants 51 . While these structures can be regarded as undesirable, they could be of potential use for speed breeding practices. If for example pollen or ovaries could be produced from these immature flowers, breeding cycles could be shortened significantly. Currently it takes 10 years from seed to assessment 52 . Other researchers have already attempted to induce flowering in oil and date palm with varying results 53,54 . Masmoudi-Allouche and coworkers were able to induce in vitro date palms to develop ovaries, however, to date these findings in the palm family are not yet applied in speed breeding 55 .
While in this report in vitro multiplication was initiated from zygotic embryos, this technique has the potential to be applied on fully characterized mature palms. In other palm species it has been shown that apical shoots isolated from field grown palms (off-shoots) could be used 56 as starting material for clonal propagation, and for bananas (monocot) it has been proven that even immature male inflorescence could be used 30  www.nature.com/scientificreports/ technique would have the advantage of being non-destructive. However, this still needs to be experimentally confirmed.
During the next few years, regenerants derived from our multiplication protocol will be verified under field conditions. It is known that clonal propagation of oil palm through somatic embryogenesis often results in a high frequency of the unwanted mantled phenotype 19 , only visible during fruiting. However, we believe that by using a multiplication protocol based on axillary shoot multiplication, the chance to obtain such off types is considerably reduced.

Conclusions
The "cut" treatment, where the plantlets are vertically cut in two at the middle, on in vitro coconut plantlets combined with exposure to 1 µM TDZ in the culture medium, resulted in the meristematic proliferation of the apical meristem in up to 33.3% of the initiated tissues. This high rate of initiation combined with the potential of multiplying material indefinitely to regenerate thousands of plants, provides the coconut industry with a solution for their current need of quality planting material. This work provides a welcome alternative to the current coconut palm micropropagation practices that rely solely on somatic embryogenesis. Due to its high efficiency, ease of application and putative lower chance of off types this is of interest to the industry as well.

Materials and methods
Plant material. Zygotic embryos from six cultivars were provided by the Philippine Coconut Authority  31 . The zygotic embryos were extracted at PCA from 10-11 month old coconuts after which they were locally sterilized by washing the coconut plugs (a disk, 2-3 cm diameter, of solid endosperm containing the embryo) in commercial bleach and rinsing with water 57 . After sterilization, the embryos were excised from the plugs and individually transferred to 2 mL cryotubes containing 1 ml sugarless semi-solid Y3 medium 58 , then shipped to the Laboratory of Tropical Crop Improvement Leuven, Belgium by air courier.
Previous research has shown that upon arrival an extra sterilization step needs to be included to eradicate remaining contaminations 59,60 . Therefore, the embryos were exposed to a 95% EtOH solution for 3 min, followed by 20 min washing in a 0.5% NaOCl solution containing two drops of liquid soap followed by a triple rinse with sterile demineralized water 59 .
In vitro culture and germination of coconut embryos. Sterile embryos were transferred onto Y60 AC medium and incubated for 1 month in the dark at 24 °C. Then they were subcultured twice monthly on the same medium and incubated in the light (18 h/6 h light cycle; the light was provided by 36 W (cool white)/ 840 Lumilux fluorescent lights). After the 4 th month, the plantlets were transferred to Y40 AC medium and subcultured every month. At each subculture, the plants were trimmed to 3-4 cm in length and transferred to fresh medium.
Induction of proliferation. The plantlets used for induction of proliferation were at least 4 months old.
To induce meristematic proliferation, several methods and media were tested. In the control treatment plantlets were trimmed to 3-4 cm in length and transferred to the induction medium. For the "cut protocol" the outer layers of leaves were cut at the base. Then, horizontal cuts were made ± 0.5 cm above and underneath the estimated location of the meristem. The remaining 1 cm long cylinder was vertically halved and both pieces were transferred upright to the induction medium, making sure that the explant is 50% submerged in the medium. For the "meristem protocol", leaf tissues were removed from the base of the coconut plantlet under a binocular microscope until its diameter was around 3 mm, leaving a cylindrical/conic shaped tissue with 3-4 leaf primordia. This cone was then transferred to the induction medium in an upright position, fully submerging the base in the medium, but leaving the meristem and leaf primordia just out of it.
Tissues were cultured at 24 °C with an 18 h/6 h light cycle provided by 36 W (cool white)/ 840 Lumilux fluorescent lights. The induction medium used was varied according the experimental set up in the following 4 paragraphs.
The effect of cutting in the presence or absence of TDZ on induction of proliferating meristems. Plantlets  The effect of the "meristem protocol" vs "cut protocol". Plantlets of the MYD cultivar were subjected to the "meristem protocol" or "cut protocol", and transferred to 1TDZ medium. Each treatment contained 48 plantlets. After 45 days the explants showing proliferating meristems were counted.
Proliferation of six different cultivars. The "cut protocol" in combination with 1TDZ was executed on six cultivars, 3 dwarf (CATD, EGD, MYD) and 3 tall ( LAGT, MVT, WAT) cultivars. Three repetitions of each 24 plants were used. The explants that showed proliferating clumps or died were counted after 45 days.
Subculture of proliferating material. The proliferating tissues that were obtained in the previous experiments were isolated from their leaf base and transferred to 1TDZ medium and grown in the dark. These proliferating tissues were subcultured monthly and transferred onto fresh 1TDZ medium. For this, the clumps of meristems were divided in smaller pieces (~ 0.8 cm 2 measured at the base) by removing non-meristematic tissues and parts that showed browning. These were then transferred one by one to new culture tubes and further maintained in the dark at 24 °C.

Regeneration.
Clumps of proliferating meristems of around 0.8 cm 2 in size, measured at the base, were transferred to Y40 AC medium and subsequently stored in darkness at 24 °C for three months. Each month, tissues were divided in 2-5 pieces depending on the size of the regenerating meristem clump and subcultured on Y40 AC medium until only 1-6 shoots were left per tube. After three months these regenerating plantlets were transferred to light (18 h/6 h light cycle provided by 36 W (cool white)/ 840 Lumilux fluorescent lights) and further subcultured every month on Y40 AC medium until fully regenerated rooted plantlets were obtained. During regeneration, the development of the meristems was investigated and the shoots showing browning or abnormal growth were counted. Statistics. The data were analysed using a contingency analysis with an alpha level of 0.05 in the statistical program JMP. If the different treatments differed significantly, subsequent analysis was done using a new contingency analysis excluding the significantly different treatments, until no significant differences were found.
The percentages were transformed using the following formula: y = arcsin x 100 to create a normal distribution. After the transformation the averages of the different cultivars were compared with each other, using an ANOVA with an alpha level of 0.05 in the statistical program JMP.

Data availability
The datasets generated in current study, after executing the experiments described in this manuscript are available as a supplementary Data S2.