A novel micropropagation of Lycium ruthenicum and epigenetic fidelity assessment of three types of micropropagated plants in vitro and ex vitro

Lycium ruthenicum is an excellent eco-economic shrub. Numerous researches have been conducted for the function of its fruits but scarcely focused on the somaclonal variation and DNA methylation. An efficient micropropagation protocol from leaves and stems of L. ruthenicum was developed in this study, in which not only the leaf explants but also the stem explants of L. ruthenicum were dedifferentiated and produced adventitious buds/multiple shoots on one type of medium. Notably, the efficient indirect organogenesis of stem explants was independent of exogenous auxin, which is contrary to the common conclusion that induction and proliferation of calli is dependent on exogenous auxin. We proposed that sucrose supply might be the crucial regulator of stem callus induction and proliferation of L. ruthenicum. Furthermore, results of methylation-sensitive amplified polymorphism (MSAP) showed that DNA methylation somaclonal variation (MSV) of CNG decreased but that of CG increased after acclimatization. Three types of micropropagated plants (from leaf calli, stem calli and axillary buds) were epigenetically diverged more from each other after acclimatization and the ex vitro micropropagated plants should be selected to determine the fidelity. In summary, plants micropropagated from axillary buds and leaves of L. ruthenicum was more fidelity and might be suitable for preservation and propagation of elite germplasm. Also, leaf explants should be used in transformation. Meanwhile, plants from stem calli showed the highest MSV and might be used in somaclonal variation breeding. Moreover, one MSV hotspot was found based on biological replicates. The study not only provided foundations for molecular breeding, somaclonal variation breeding, preservation and propagation of elite germplasm, but also offered clues for further revealing novel mechanisms of both stem-explant dedifferentiation and MSV of L. ruthenicum.


Introduction
Lycium ruthenicum, belonging to the Solanaceae family, which inhabits northwestern China [1], is a perennial desert pioneer shrub with saline-alkali tolerance, drought resistance, wear (w/v) agar (Jinan Zhongtian Plant Tissue culture Center), adjusted to pH 5.8 with KOH prior to autoclaving at 121˚C for 15 min. The inoculated seeds were cultured in dark until they germinated, thereafter the germinated seeds were cultured under 48 μmol m -2 s -1 light provided by LED fluorescent lamps at a photoperiod of 12 h. The temperature is 25±2˚C throughout the course [31]. At 45 days after inoculation, two healthy in vitro seedlings of PlantD and G were selected as explant donors. For convenience, the two donors in vitro were renamed as inDdonor (a seedling from PlantD) and inGdonor (a seedling from PlantG). After transplanting, the two donors were renamed as exDdonor and exGdonor, respectively. Notably, genetic background of the two donors is not identical because they were seedlings from different L. ruthenicum which shows self-incompatibility.

Callus, adventitious bud and multiple shoot induction
Expanded leaves of inDdonor and inGdonor were cut perpendicularly to their main vein into explants about 0.5-0.8 cm, and then inoculated with the abaxial side upwards in flasks with the leaf medium (Fig 1). The leaf medium was Murashige and Skoog (MS) medium [30] supplemented with 4% (w/v) sucrose, 0.50% (w/v) agar, 0.89 μM 6-benzyladenine (6-BA) and 0.54 μM a-naphthaleneacetic acid (NAA). The shoot tips of inDdonor and inGdonor were removed, the middle stems without leaves were cut into explants with two leaf axils, the remaining base parts of stems with roots and leaves were sub-cultured on the forgoing 1/2 MS medium (Fig 1). The new shoots of the remaining base parts could also be used as leaf and stem explants. The stem explants were inoculated in flasks with the stem medium. It is worth noting that the lower leaf axils of stem explants should touch the stem medium (Fig 1). The stem medium was MS medium supplemented with 4% (w/v) sucrose, 0.50% (w/v) agar and 0.44 μM 6-BA. The concentrations of 6-BA and NAA were chosen in our stem or leaf medium because they result in better micropropagation than other PGR concentrations in our previous experiences. All the inoculated stem and leaf explants were cultured at a photoperiod of 12 h under 48 μmol m -2 s -1 light provided by LED fluorescent lamps. Prior to autoclaving at 121˚C for 15 min, the PGRs of the leaf and stem medium were added and then the pH of the leaf and stem medium was adjusted to 5.8 with KOH. The temperature (25±2˚C) was maintained throughout the course of in vitro culture. Moreover, all the cultures were sub-cultured on the same media every 45 days.

Rooting and acclimatization of plantlets
When the regenerated shoots were about 2 cm high, they were cut from the leaf-derived calli and transplanted onto the forgoing 1/2 MS medium without any PGR for root induction. Meanwhile, when the shoots from both axillary buds and calli of stem explants were at least 2 cm high, they were cut and inoculated on the forgoing 1/2 MS medium for rooting [31]. The conditions of light intensity, photoperiod and temperature for rooting were identical to those of callus induction. In vitro rooted plantlets were acclimated under natural sunlight until both leaves were dark green and stems were no longer tender. Thereafter, the in vitro stronger plantlets were transferred into sterilized substrate with the mixture of humus and sphagnum moss (1:1) and the pot was covered with plastic wrap with holes for at least 10 days in a growth room at 25 ± 5˚C under indirect light. Moreover, the matrix were sterilized at 121˚C for 60 min, and the plantlets without medium were immersed in 0.33% (m/v) carbendazim turbid liquid for 5 min before transplanting. After removing the plastic wrap film, the transplanted ex vitro plantlets were originally acclimated under indirect sunlight for 10-20 days and then were exposed to direct sunlight.

DNA isolation and quantification
Genomic DNA was extracted from the expanded leaves of the two donor plants in vitro and ex vitro, from expanded leaves of both in vitro and ex vitro micropropagated plants derived from leaf calli, stem calli and axillary buds using a small-scale DNA isolation method (NuClean Plant Genomic DNA Kit-CVVBIO) according to the manufacturers' instructions. Notably, the expanded leaves from the new shoots of the transplanted plants were used for DNA extraction in order to exclude the non-heritable DNA MSV in response to the in vitro environment. Two replicate extractions from all the samples above were performed. DNA purity, integrity and concentration were assessed by the method mentioned in our previous report [22]. Fig 2  shows the processes used to obtain the samples used in MSAP analysis and the sample numbers. Meanwhile, for convenience, the micropropagated plant samples were renamed as follows (Fig 2): inDaxil-plant [1][2]

MSAP analysis
For the purpose of (a) detecting the differences of MSV among in vitro plants from leaf calli, axillary buds and stem calli and among ex vitro plants from leaf calli, axillary buds and stem calli, (b) revealing the differences of MSV between the micropropagated plants in vitro and ex vitro, and (c) finding the micropropagated plant-specific MSAP markers, the MSAP method was employed to assess cytosine methylation differences in D and G groups (Fig 2). All the

Data analysis
All the data of micropropagation were subjected to statistical analysis using paired sample ttest (2-tailed, P<0.05). The scored MSAP bands were transformed into a binary character for the absence (0) or presence (1). All the binary data of MSAP were generated by software Gene-Marker V2.2.0 (SoftGenetics, USA). The levels of cytosine (CCGG sites) methylation and locus-specific methylation differences among samples within D or G group were subjected to statistical analysis using one sample t-test (2-tailed, P<0.01 and 0.05) and one-way ANOVA (2-tailed, P<0.05) by software SPSS ver. 20.0 (IBM Co., Armonk, NY, USA) [29]. Meanwhile, independent-sample t-test of SPSS ver. 20.0 was used to compare the in vitro and ex vitro locus-specific MSV of the same micro-propagated plants. Moreover, the principal coordinate analysis (PCA) and UPGMA cluster analysis of MSAP profiles were carried out by software MVSP ver. 3.2 (Kovach Computing Services, Wales, U.K.) [32]. The specific MSAP markers were determined as follows: (1) all MSAP sites that showed a monomorphic pattern or a 'Suspected' by GeneMarker V2.2.0 in only one sample were excluded from the binary data matrices [33]; (2) the remaining binary data matrices were transformed into quaternion matrices (00!0, 01!1, 10!2, 11!3) by excel data processing; (3) the micro-propagated-or ex vitro plant-specific MSAP markers were found and determined [34].

Leaf explant produced both calli and adventitious buds on the same medium
In the leaf medium, not only the leaf explants from the two donors (inGdonor and inDdonor) produced calli but also adventitious buds were regenerated from the calli ( Fig 3A). After 30 days of culture on the leaf medium, the leaf explants of two donors showed similar efficiency for callus induction; however, the percentage of callus producing adventitious buds was significantly lower for materials from inGdonor than that from inDdonor (Table 1). This indicated that adventitious buds were regenerated in a genetic background-dependent pattern. Moreover, after 60 days of culture on the leaf medium, shoots produced by the calli could reach the height of 2-3 cm ( Fig 3B) and be used for in vitro root induction.

Stem explants produced two types of shoots on auxin-free medium
On the auxin-free MS medium supplemented with 0.44 μM 6-BA (stem medium), the stem explants of inGdonor and inDdonor produced not only multiple shoots from axillary buds but also nodular calli from cross section enwrapped by the medium (Fig 3D and 3E). On the same stem medium, adventitious buds were regenerated from the nodular calli soon (Fig 3F and  3G) and grew rapidly ( Fig 3H). After 30 days of culture on the stem medium, both the percentage of D stem explants producing multiple shoots and producing calli with adventitious buds were significantly higher than that of G stem explants (Table 1). This suggested that both the

PLOS ONE
axillary bud germination and regeneration of L. ruthenicum stem explants were affected by the genetic background.

Rooting and acclimatization of plantlets
A total of 59.47-93.87% shoots rooted in vitro on the 40th day after being transferred onto the 1/2 MS medium without any PGR (Table 1, Fig 3C and 3I). Rooting rate of shoots from stem explants was higher than that from leaf explants of the same donor (Table 1). Meanwhile, rooting rate of G clones was significantly higher than that of D clones (Table 1). Totally, both genetic background and explant type affected the rooting capability of L. ruthenicum shoots.
The transplanting survival rate of L. ruthenicum plantlets was up to 95.65% by using the acclimatization protocol above ( Fig 3J).

Selection of suitable primer pairs for MSAP analysis
Fourteen primer combinations were selected (S1 Table) based on the criteria of our previous report [22]. Using the 14 primer pairs, we scored 1,751 and 1,743 reproducible bands from D and G group, respectively (Tables 2 and 3). Within D group, 1,458 (83.27%) of the 1,751 bands

Changes in cytosine methylation level occurred between donors and micropropagated plants of in vitro and ex vitro
Of the CCGG sites assessed in plants in vitro and ex vitro, 53.64-56.81% and 53.24-58.55% are methylated, respectively (Tables 2 and 3). Moreover, compared with ex vitro plants, the average internal cytosine (CG) methylation levels in in vitro plants of the two groups are all lower (Tables 2 and 3). Compared with the corresponding in vitro donors of the two groups, all the in vitro plantlets regenerated from stem callus showed two types of significant alterations in three types of detectable cytosine methylation levels (CG, CNG and CG & CNG, Tables 2 and  3). However, there was only one type of significant difference between the ex vitro donors and the plants from stem calli within each group after the transplant (Tables 2 and 3). Interestingly, for plants regenerated from stem calli, the types of significant alterations ex vitro were different from those of in vitro within each of the two groups (Tables 2 and 3). This revealed that after acclimatization, the significant alterations in vitro might be erased but the in vitro non-significant alterations became significant. Compared with inDdonor, the alterations of methylated CG and CG & CNG in plants from D leaves are statistically significant; however, after acclimatization the alterations were never statistically significant (Table 2). Nevertheless, that was not the same as plants from leaf calli of G group, whose significant different levels in vitro (CG and CNG) still significant after transplanting (Table 3). Compared with donors, only in vitro D plants from axillary buds showed significant alterations in methylated CNG and CG & CNG ( Table 2). These results indicated that methylation level SV of L. ruthenicum micropropagated plants was mainly reduced after acclimatization but some level SV in plants from stem calli increased after acclimatization.

Locus-specific methylation alterations occurred both in vitro and ex vitro
Compared with the in vitro donor plants, all the patterns of locus-specific methylation alterations in three types of micro-propagated plants in vitro were statistically significant at 0.01  (Table 4). All the patterns of locus-specific methylation alterations in three types of micropropagated plants ex vitro versus the ex vitro donor plants were statistically significant with Both Hypo and Both Hyper of plants from axillary buds at 0.05 levels and the others at 0.01 levels ( Table 5). After transplanting, the level of CG Hyper in plants from stem calli was significantly higher than that in plants from axillary buds. However, the level of CG Hypo was contrary to that of CG Hyper (Table 5). Unlike in vitro plants, the total level of locus-specific methylation alterations in the ex vitro plants from stem calli was the highest but there was no statistical significance. Just like that in vitro, for each type of the micropropagated plants ex vitro, the level of hypermethylation was higher than that of hypomethylation (CG Hyper>CG Hypo, CNG Hyper<CNG Hypo, Both Hyper>Both Hypo, Total Hyper>Total Hypo).
From the data of Tables 4 and 5 we concluded that internal cytosine MSV of CCGG sites (CG Hyper and CG Hypo) in three types of micropropagated plants increased and that of CNG (external cytosine) decreased after acclimatization (Table 6). Also, the levels of CG Hyper in plants from calli of both leaf and stem were significantly increased (P<0.01) after acclimatization (Table 6). However, after acclimatization the levels of CNG Hyper in all the three types of micropropagated plants were significantly decreased (P<0.01) ( Table 6). There were one and two patterns of significant methylation alterations in plants from axillary buds and calli, respectively ( Table 6), suggesting that the MSV of plants from two types of calli showed the sharpest changes in response to acclimatization and the smallest changes were in Table 4

PLOS ONE
plants from axillary buds. In total, neither methylation level SV or locus-specific MSV of L. ruthenicum showed a simple reduction after acclimatization because there was not only decrease but also increase, however, the decrease was predominant.
All the patterns of locus-specific methylation alterations in leaves of various micropropagated plants ex vitro versus the corresponding in vitro leaves of the same plants were found to be statistically significantly. However, the total locus-specific methylation alterations in the two donors (44.24%) were the sharpest (S2 Table), which indicated that donors and micropropagated plants did not show the identical response to acclimatization. For all the three types of micropropagated plants, the rates of CNG Hypo and Total in plants from stem calli were the highest (S2 Table). This might account for why the significant methylation level alterations in plants from stem calli were erased but the novel significant alterations appeared after acclimatization (Tables 2 and 3).

Epigenetic divergence among all the plants within each group
Cluster analysis of D group based on the MSAP profiles revealed that (a) the in vitro and ex vitro plants from stem explants were clustered into two separate groups, respectively; (b) all the in vitro plants regenerated from leaves were clustered into another group; (c) all the in Table 5.

PLOS ONE
vitro micropropagated plants were diverged more from the in vitro donor than from each other but (d) after the acclimatization, the plants from leaf calli were clustered with ex vitro donor preferentially (Fig 4A). The PCA further validated the results of cluster analysis (Fig 4B). The above (a)-(c) implied that MSV in vitro are not random and are somewhat consistent between three types of in vitro micropropagated plants, consistent with the finding that consistent alterations of DNA methylation are induced by tissue culture in maize [28]. Moreover, the above (c)-(d) suggested that it was easier to select the most fidelity micropropagated plants after transplanting. Both cluster and PCA results of group G and D were similar. However, the special sample exGstem-plant 4 was clustered into one group with plants in vitro; the special sample inGaxilplant 3 was clustered into one group with plants ex vitro (Fig 5A). It was difficult to conclude which type(s) of in vitro plants was more similar to the in vitro donor. However, after acclimatization, the plants from stem calli exhibited significantly more epigenetic divergence from donor than those from both axillary buds and leaf calli with two plants from axillary buds most fidelity (Fig 5B). It was difficult to select the most fidelity micropropagated plants in vitro but easier ex vitro.

Micropropagated plant-specific MSAP sites
Compared with the in vitro donors, 14 and two in vitro micropropagated plant-specific MSAP sites were found in D and G groups, respectively (S3 and S4 Tables), but none of these were transmitted to the ex vitro leaves. This indicated that some main MSV in L. ruthenicum cannot be transmitted through mitosis. The sites of H4-153 are '11' in all the in vitro micropropagated plants of D group and '01' in inDdonor (S3 Table). Meanwhile, the site of H4-153 in G group except inGaxil-plant 3 (a special sample in cluster) was the same as that in D group, which indicated that methylation modification in site of H4-153 was usually removed during in vitro culture, and the site can be regarded as in vitro MSV hotspot but the demethylation of H4-153 site is not a necessory event for micropropagation of L. ruthenicum, because not all the micropropagated plants show the same alteration. Nevertheless, all the in vitro plants from calli show the same alteration. Moreover, the majority of the micropropagated plant-specific sites are only observed in one of the two groups, indicating that MSV of L. ruthenicum is depend on genetic background. Although the significantly locus-specific methylation alterations between the in vitro and ex vitro leaves of the same plants existed, we did not find any in vitro plant-specific MSAP site.

Micropropagation of L. ruthenicum
Dedifferentiation and redifferentiation rates of stem explants (Table 1), redifferentiation rates of leaf callus ( Table 1), percentage of stem explant producing multiple shoots (Table 1) and rooting rates (Table 1) were all significantly different between the two groups, indicating that stem explant dedifferentiation and redifferentiation, multiple shoots formation from axillary bud, redifferentiation of leaf callus and rooting of L. ruthenicum were all affected by genetic background.
Calli of L. ruthenicum stem and leaf could regenerate adventitious buds on the callus induction medium soon. That is to say, shoot regeneration of L. ruthenicum did not need change the proportion of exogenous auxin and cytokinin, which is contrary to the common accepted view that calli can be induced from explants on the medium supplemented with an optimal concentration of exogenous auxin and cytokinin, and subsequent culture of the calli with high cytokinin/auxin ratio leads to shoot regeneration [35,36]. Moreover, the roots of L. ruthenicum were

PLOS ONE
induced without exogenous auxin, which is similar to that of the Clivia miniata [31] and Brassica juncea var. Tsatsai [37] but in contrast to the majority of previous studies which found proper exogenous auxin promotes rooting of in vitro shoots [35]. It is commonly accepted that exogenous auxin plays a key role for plant callus induction and proliferation, however, exogenous cytokinin, may play a coordinating role for callus induction and proliferation [35,38]. In addition, proliferation of habituated Arabidopsis callus is dependent on exogenous auxin but not on cytokinin [20]. Furthermore, exogenous auxin-picloram or indolo acetic acid was necessary for callus induction from undifferentiated cambial meristematic cells [39]. Surprisingly, the stem explants of L. ruthenicum could produce calli without exogenous auxin. Above all, L. ruthenicum showed abnormal auxin demand during its micropropagation process. Not only did stem explants of L. ruthenicum produce calli but also in vitro shoots of L. ruthenicum rooted without exogenous auxin. It was proposed that the leaf-to-callus process is not a dedifferentiation process but a transdifferentiation process [36,40] because many studies in Arabidopsis had demonstrated that callus is a group of root meristem tip cells and that callus induction resembles lateral root formation [41][42][43][44]. Therefore, we proposed that callus induction from stem cross section enwrapped by medium exceedingly resembles root induction from shoot of L. ruthenicum. Also, not only the roots but also the forgoing stem callus should be originated from vascular cambium of L. ruthenicum stem [45][46][47]. Therefore, the stem callus induction without exogenous auxin in fact is identical to root induction without exogenous auxin. Notably, exogenous auxin is an essential regulator of callus initiation and proliferation in other plants [36] but not in L. ruthenicum stem of this study. Thus, a novel mechanism of callus initiation and proliferation might exist in L. ruthenicum and even in other plants. Recent researches have shown that sugar demand, not auxin, is the initial regulator of apical dominance [48,49]. Moreover, developmental transitions in plants require adequate sugar which acts as sugar signaling and carbon energy-supply [49,50]. We concluded that sucrose supply maybe the crucial regulator of stem callus initiation and proliferation in L. ruthenicum because only the lower stem cross section enwrapped by medium with 4% (w/v) sucrose and without auxin could produce callus (Fig 3D) but the upper stem cross section did not. However, further study should be carried out to clarify the hypothesis.

DNA methylation variation and specific MSAP sites
In this survey, the total cytosine methylation levels in expanded leaves of L. ruthenicum are 53.64-56.81% in vitro and 53.24-58.55% ex vitro. The methylation levels are certainly higher than those surveyed in almost all other plants [37,[51][52][53][54][55] but comparable to those in Clivia miniata [22]. However, the CG methylation levels in L. ruthenicum are significantly lower than those in diploid C. miniata [22] and tetraploid cotton [56]. To the best of our knowledge, these are the first data on DNA methylation of L. ruthenicum.
The PCA revealed total clear separation between leaves of plants in vitro and leaves of the same plants ex vitro within D group. The result was similar to a previous report which found clear separation between in vitro propagated plants and their field counterparts from cuttings for five Manihot esculenta cultivars [57]. Notably, plants compared in Kitimu's study are different plants cloned by two methods but leaves compared in this study are from the same plants before and after acclimatization. Thus, we can deduce more rigorous conclusion of plasticity in genomes of L. ruthenicum growing under two different environments.
Initially, we proposed that after acclimatization the MSV should decrease because only the heritable SV existed in the new leaves of transplanted plants. It was supported by the MSV hotspot which was not transmitted to the ex vitro leaves in the study. Nevertheless, neither methylation level SV or locus-specific MSV of L. ruthenicum showed a simple reduction after acclimatization because there was not only decrease but also increase; however, the decrease was predominant. On one hand, after acclimatization not only micropropagated plants but also donor plants of L. ruthenicum show significant DNA methylation changes. On the other hand, micropropagated plants and donors of L. ruthenicum show different response to acclimatization (S2 Table). Thus, some patterns of MSV, such as CG Hyper, increased significantly after acclimatization.
The SV is undesirable for long-term germplasm preservation [27] but can be used in strain improvement during plant breeding [25,26]. The three types of transplanted micropropagated plants were diverged more from each other than the in vitro counterparts. Thus, it was difficult to select the most fidelity micropropagated plants in vitro but easier ex vitro (Figs 4 and 5). In conclusion, after acclimatization plants from stem calli were diverged more from the ex vitro donors than plants from axillary bud or leaf calli. Thus, we proposed that direct organogenesis from axillary buds might be suitable for preservation or propagation of elite L. ruthenicum germplasm. Meanwhile, the SV in plants from stem calli might be used to SV breeding. Also, the stem explants can be used in researching the novel sugar mechanism of dedifferentiation. However, propagation using L. ruthenicum leaf as explants can be used in germplasm preservation or propagation and is suitable for transformation. Furthermore, one MSV hotspot of L. ruthenicum was found based on 41 replicates of two groups. The hotspot indicated that certain regions of the L. ruthenicum genome are consistently exhibiting DNA demethylation in tissue culture. This is similar to recent studies in rice and maize which have shown that losses of DNA methylation following tissue culture are more common than gains of DNA methylation [28,58]. The hotspot of L. ruthenicum could be used for revealing the epigenetic mechanism of SV.

Conclusions
We developed a novel efficient micropropagation protocol from leaves and stems of L. ruthenicum and found that stem explant dedifferentiation and redifferentiation, multiple shoots formation from axillary bud, redifferentiation of leaf callus and rooting of L. ruthenicum were all affected by genetic background. Notably, the optimal medium for indirect organogenesis of stem explants was auxin-free medium with 4% sucrose. This indicated that sucrose supply might be the crucial regulator of stem callus induction and proliferation in L. ruthenicum. One MSV hotspot was found based on MSAP analysis, which offer an important clue for revealing the epigenetic mechanism of SV. Furthermore, MSAP analysis indicated that DNA methylation SV of CNG decreased but that of CG increased after acclimatization; the three types of micropropagated plants (from leaf calli, from stem calli and from axillary buds) were epigenetically diverged more from each other after acclimatization. Thus, we proposed that the ex vitro micropropagated plants should be selected to determine the fidelity. In summary, micropropagation from axillary buds and leaves of L. ruthenicum was more fidelity and might be suitable for preservation or propagation of elite germplasm. Propagation using L. ruthenicum leaf as explants is suitable for transformation. Meanwhile, the micropropagation from stem calli showed the highest MSV and could be used in both SV breeding and researching the novel sugar mechanism of dedifferentiation. The findings above not only provided foundations for molecular breeding, somaclonal variation breeding, preservation and propagation of germplasm, but also offer clues for further theoretical researches.
Supporting information S1