Agromorphologic, genetic and methylation profiling of Dioscorea and Musa species multiplied under three micropropagation systems

Plant in vitro vegetative propagation using classical semi-solid culture medium is limited due to the low degree of automation, suboptimal nutrient availability and induced physiological stress which often reduce its efficiency. Temporary Immersion System (TIS) emerged as an innovative approach to optimize and eliminate the drawbacks associated with the conventional system of micropropagation. In this study, both Dioscorea and Musa spp. were subjected to conventional semi-solid culture media, complete immersion in shaking liquid culture media and TIS using RITA bioreactor. In vitro grown plantlets were screened for possible vegetative changes using agro-morphological descriptors while genetic and methylation differences were assessed using amplified fragment length polymorphism (AFLP) and methylation-sensitive amplification polymorphism (MSAP). In vitro results showed that the number of shoots produced in Musa spp. varied significantly (P≤0.001) with the type of culture system. The highest mean shoot produced was observed with TIS (28.40) and the least using semi-solid culture medium (1.13). For Dioscorea spp., there was no significant interaction between the hormone combination and the culture system. However, the lowest mean shoot value (1.55) was observed in the semi-solid culture medium. Genetic analysis via AFLP using 15 primer pair combinations revealed that the 3 culture systems maintained genetic variation for Musa and Dioscorea spp. under in vitro and field conditions. Results showed 99% and 91% of the total bands were polymorphic under in vitro and field conditions respectively for Musa and 100% polymorphism for Dioscorea under in vitro and field conditions. Methylation investigation via MSAP using 12 primer pair combinations showed 25% and 46% polymorphic methylated-sensitive loci, 100% and 78% of non-methylated loci of the total bands generated under in vitro and field conditions respectively. Unmethylated (HPA+/MSP+) levels were highest in TIS (0.0842) as compared to CI (0.0227) and SS (0.0161) while full methylation or absence of target (HPA-/MSP-) was lowest in TIS (0.5890) and highest in SS (0.7138). For Dioscorea, 52% and 53% methylated sensitive loci and 100% non-methylated loci were polymorphic under in vitro and field conditions respectively. Although in vitro plant tissue culture techniques led to methylation at some loci of both species, there were no observable changes in the phenotype of both crops under field conditions. This also confirmed that not all methylation events lead to phenotypic changes.

Although in vitro plant tissue culture techniques led to methylation at some loci of both species, there were no observable changes in the phenotype of both crops under field conditions. This also confirmed that not all methylation events lead to phenotypic changes.

Introduction
In vitro plant tissue culture is recognized as one of the most valuable biotechnology tools for rapid multiplication of disease-free and true-to-type genotypes. The technology is used extensively in clonally propagated horticultural, food and tree crops. There are, however some challenges related to formation of aberrant plantlets and low survival during acclimatization stages in the field [1][2][3]. In both Dioscorea and Musa spp., in vitro clonal propagation can be used either for large scale propagation or conservation. However, somaclonal variation in plant material under in vitro plant tissue culture is significantly influenced by DNA methylation changes, although the occurrence of such events is unclear [4][5][6].
Culture media is a crucial aspect of in vitro plant propagation. It determines its effectiveness and can be targeted for improvement. The response of plant tissue to in vitro culture medium depends on several factors including the genotype itself, the nutrient content of the culture medium, the source and physiological state of the explant and the physical culture conditions such as temperature, pH, photoperiod and aeration [7][8][9] of the culture systems. Semi-solid and liquid culture are some of the common culture media systems used for in vitro plant propagation [10]. Although these methods have certain advantages but there are limitations too. The major disadvantages are asphyxiation, hyperhydricity, induced stress on agitated cultures, explant blackening (oxidation), poor diffusion rate and sub-optimal nutrient uptake which may lead to severe physiological disorder [10]. The temporary immersion system (TIS) that involves automated system provides an optimal environment for plant tissue and organ in vitro cultures. The method emerged as an approach to scale up the conventional method of propagation [11]. Over the years, several TIS have been developed and successfully used in various plant in vitro systems [12][13][14]. In 1993, Center de Cooperation Internationale en Recherche Agronomique pour le Development (CIRAD) developed a new temporary immersion system named RITA (recipient á immersion temporaire automatique) which eliminated the limitations of previously developed bioreactors, thus promoting massive in vitro plant production.
The objective of this study was therefore to explore the efficiency of TIS (RITA bioreactor) on Musa and Dioscorea spp. with respect to its comparative advantage over other conventional culture systems and to assess probable genetic and methylation modifications in the regenerated plants using amplified fragment length polymorphism (AFLP) and methylation-sensitive amplified polymorphism (MSAP) markers.

Materials and methods
The accessions of Musa and Dioscorea spp. (Table 1) were collected from the Genetic Resources Center (GRC) of the International Institute of Tropical Agriculture (IITA) Ibadan, Nigeria.
Initiation of multiple shoots for Musa spp. was enhanced by wounding the apical meristem during subculture, a technique described by Jarret et al. [17] and Gupta [18] which involves vertical cuts through the meristematic dome while keeping the base of the explant intact.
The culture systems used for both species were: semisolid culture medium in test tube (SS), complete immersion in liquid culture medium in baby food jar (CI) with shaking on a rotary shaker at 100 rpm, and temporary immersion system (TIS) in RITA bioreactor with immersion time of 15min every 2 hours for both crops.
Five replicates of each of the seven accessions of Musa and one accession of Dioscorea were cultured under in vitro conditions in the three culture systems (SS, CI, TIS). The cultures were kept in growth chamber at 26˚C ± 2.0, 38 μmol/m 2 /s and 12-hour photoperiod for three and six weeks for Musa spp. and Dioscorea rotundata respectively. The number of shoots were determined by counting the number of shoots (for Musa) and number of nodes (for yam) per single plant for each of the three culture systems. For better rooting system development, MS mineral-based culture medium was supplemented with either 0.18 mg/l IAA or no auxin for 3 weeks.

Agro-morphological characterization
The in vitro grown planting materials were acclimatized (using sterilized top soil and chicken manure in ratio 2:1 for Musa and Jiffy peat pellet for Dioscorea) in the screen-house. After 12 weeks, the established plants were transplanted to the field at IITA (Latitude 7.50338˚Longitude 3.90427˚, Altitude 248.00m) following usual agronomic management practices such as regular watering, weeding and mulching. Data of agro-morphological parameters were captured every month following the available descriptor list for both Musa and Dioscorea [19][20] using 33 and 83 traits respectively until the crop senesced completely. However, data on majority of the traits were recorded at harvest for Musa spp. while for Dioscorea rotundata was carried out through the entire growth period. Mini tubers obtained from yam, were planted the second year and characterization data were collected on selected traits (S2 and S3 Tables).

Molecular characterization a) Sample collection and DNA isolation
About 100 mg of young leaf samples of Musa (cigar-like leaf) and Dioscorea (first fully expanded) genotypes were collected both from field and in vitro grown plants and labeled  [21] with the addition of DIECA and ascorbic acid to inhibit phenoloxidase and other impurities. Ground yam samples were first washed with 1000 μl HEPES buffer (10ml of 0.1 M HEPES + 90mg L-ascorbic acid + 102 mg PVP + 200μl β-mercaptoethanol) to remove secondary metabolites prior to extraction procedure. The quality of the DNA was determined by 1% agarose gel electrophoresis while the quantity and purity were measured through absorbance ratio (240/280) using NanoDrop spectrophotometer (Thermo scientific Nanodrop 2000 spectrophotometer).

b) AFLP and MSAP analysis
The AFLP and MSAP analysis followed a modified version of Vos et al., Vroh-Bi et al. and Reyna-Lopez et al. [22][23][24]. The isoschizomer restriction enzyme pair HpaII and MspI was used for MSAP, which recognizes CCGG site with differential sensitivity to methylation at cytosine, while MseI was used for AFLP. When the internal cytosine is fully (methylation on both DNA strand) or hemi (methylation on one DNA strand) methylated, MspI recognizes and cleaves the motif but it cannot cleave an outer cytosine. However, HpaII has the capacity to recognize and cleave outer cytosine. Profiles generated from MspI and HpaII isochizomeric pair not only provide events associated with inner and outer methylation but give a comprehensive picture of genetic and epigenetic variations linked to methylation. Accordingly, we generated MSAP profiles of in vitro and field grown plants of both Musa and Dioscorea spp. to understand the methylation pattern associated with different culture system used in the study.

c) Scoring
The fragment peaks and intensity from the AFLP and MSAP analysis were evaluated using GeneScan software (Applied Biosystems) after scanning the signals from all samples for each crop separately. Following fragment analysis on ABI3730, AFLP and MSAP profiles were visualized using GeneMapper software v4 (Applied Biosystems, Foster City, CA, USA). Raw data generated were scored following a band-based strategy described by Bonin et al. [25] using the GeneMapper v4.0 software (applied Biosystems). Allelic profile was transformed into binary matrixes which were scored as '1' for presence of allele and '0' for absence of allele. In order to reduce eventual impact of size homoplasy [26], binning of allelic sizes was followed with a size range between 100-500 bp with peak height �100. All reactions were repeated twice and only distinct, polymorphic and informative bands across all samples were considered for analysis. Fragments that could not be visually distinguished with low intensity were regarded as ambiguous and were not scored.

Statistical analysis
Phenotypic data generated from field grown plants were subjected to generalized linear model (PROC GLM and PROC GENMOD) in Statistical Analysis System (SAS-V9.2 and V9.3) [27] to obtain the variance components. The least significant mean (LSMEANS) was used to compare the means of different traits across both species. Principal coordinate analysis (PCoA) was carried out on standardized morphological data, eigen values and eigen vectors were calculated to generate two-dimensional plots under different growth conditions (in vitro and field). Similarly, genotypic (presence/absence) data was analyzed with msap software using R script (msap_score.r) [28][29] and GenAlEx version 6.5 [30]. Different statistical parameters such as PCoA, population differentiation test using analysis of molecular variance were also estimated using the Shannon diversity index.

In vitro performance of genotypes in different culture systems
The efficiency of the three culture systems used in the study were examined on two tropical crop species (Musa spp. and Dioscorea rotundata) after three and six weeks respectively. The number of shoots per explant were counted for each plant cultured on these culture systems. For Musa spp. the multiplication rate varied significantly both with the type of culture system and the genotype tested. However, TIS (RITA bioreactor) gave significantly higher shoot mean across the genotypes, followed by complete immersion in liquid media system with shaking while the least was observed with semi-solid medium (Fig 1). There was a significant interaction between the performance of the accessions and culture system ( Table 2). On average, a higher multiplication rate was observed with genotypes having AA genome as compared to those with AAB genome. In terms of number of shoots produced under TIS (RITA bioreactor), the highest mean shoot of 28.40 was observed in TMb 28, while least mean shoot value of 3.90 was observed for TMp 100 in RITA bioreactor.
For yam (Dioscorea rotundata), the result indicated that there was no significant difference in performance between TIS and CI within 6 weeks of plant culture while they differed significantly to SS across all the treatments used in the study (Table 2). Similarly, TIS and CI favored better multiplication rate than SS culture system (Fig 2). Culture media supplemented with low sucrose without hormone or with low level of Kinetin (T1, T2 & T4) promoted better shoot growth in TIS and CI. However, a previous study indicated that an increase in the culture duration in T1 and T4 favored shoot vigour in TIS (Jekayinoluwa et al. Unpublished, S4 Table).

Agro morphological characterization a) Screen house acclimatization
The in vitro plantlets were transferred to the screen-house for acclimatization and no significant difference was observed for the parameters (NL: Number of leaves, PH: plant height, LW: leaf width, LL: leaf length) measured (S5 Table) between the different culture systems for Musa spp.
There was significant difference in growth parameters of Dioscorea spp. when transferred to screen-house across the plants generated from different culture systems. (S5 Table). TIS  Agro-morphologic, genetic and methylation profiling of clonal crops under three micropropagation systems showed better acclimatization success rate which differed significantly in terms of leaf width and plant height in comparison to other culture systems under treatment 4 (T4). However, T4 in CI produced higher number of leaves compared to plants cultured in other systems (S5 Table).

b) Field morphological characterization
The acclimatized plantlets of both Musa and Dioscorea accessions were transplanted under field conditions for morphological characterization. The Musa plants were screened based on thirty-three agro-morphological parameters. Only 1 (Weight of Bunch (WB)) variable out of 33 agro-morphological descriptors showed significant differences between the plants derived using the three culture systems. TIS recorded the highest value for WB in comparison to SS and CI (S6A & S6B Table) for TMb 19, TMp 59 and TMp 100. While for qualitative variables, the p-value of the Chi-square distribution showed no significant difference across the three culture systems.
For Dioscorea, about 8% of the total (85) agro-morphological variables showed a varying level of differences across the culture systems used. However, six qualitative traits such as leaf density, plant vigour, spine length, number of inflorescence, tuber shape and place of root on tuber were observed on plants cultured on TIS (S7A-S7E Table).

c) Molecular characterization
Amplified fragment length polymorphism (AFLP) profiles were generated for in vitro and field-grown plants of Musa and Dioscorea spp. with respect to different culture systems and treatments ( Table 3). The profiles recorded 99-100% polymorphism regardless of the type of  culture systems in both crops indicating the polymorphic nature of the markers used in the study. Principal Coordinate Analysis (PCoA) revealed the relatedness of the culture systems across the genotype used (Fig 3).  (Fig 4). The variance among the three culture systems (SS, CI & TIS) for Musa spp. under field condition was up to 3% (S8 Table) while there was no variation under in vitro conditions (S8 Table) indicating similar growth patterns among plantlets with negligible variances across different culture systems under study. Conversely, Dioscorea plants cultured under in vitro conditions recorded 17% variation among the different culture systems (S8 Table) while no variation was observed under field conditions (S8 Table). Methylation event was revealed by the PCoA analysis developed from binary matrix of combined MSAP profiles using EcoRI/MspI and EcoRI/HpaII. For Musa spp., the profile ( Table 4) revealed similarity in both methylation susceptible loci (MSL) and non-methylated  loci (NML) for plants grown using all the three culture systems, and in the field (Fig 3B and  3C). There were 564 (46% of total MSL) polymorphic methylated susceptible loci (MSL) and 797 (78% of total NML) non-methylated loci (NML) for field grown Musa plants. On the other hand, in vitro plants recorded 137 (25% of total MSL) polymorphic methylated susceptible loci and 15 polymorphic NML. The profiles of isoschizomer pair HpaII, ( m CCGG) for field grown plants showed a variation of 1% among the culture systems (S8 Table) and CI culture system recorded the highest number of private bands (143) while SS had the least (100 bands). For MspI (C m CGG), there was no observable variation among the three culture systems. However, samples collected from SS culture system recorded the highest private bands (141) while those from CI showed the least with 100 private bands. On the other hand, SS-derived plants recorded the least private bands in both profiles of the HpaII and MspI isoschizomer for in vitro grown Musa plants. The PCoA of HpaII ( m CCGG) isoschizomer showed clustering of SS and CI grown plantlets together regardless of the genotype whereas, TIS grown plantlets are separated (Fig 5).
Twenty-five field-grown yam plants of TDr 1228 accession generated 757 polymorphic MSL bands and 563 polymorphic NML bands. The PCoA generated from HpaII profile, showed clustering based on different treatments used in culture system (Fig 6). T1 and T4 were closely related while T2 and T3 clustered together. There were 627 polymorphic MSL and 367 polymorphic NML for in vitro grown plants. Further clustering based on culture system in relation to the treatment used revealed TIS with minimum MSL across all treatments. Of all the treatments, T1 and T4 recorded 334 and 338 MSL while 311 and 270 NML, respectively ( Table 5). The measure of genetic diversity was estimated by the Shannon diversity index and comparison with Wilcoxon test revealed a significant difference (P<0.0001) between MSL and NML for both crops.

Discussion
In this study, three in vitro culture systems were used, including semisolid medium in test tube, complete immersion in liquid culture medium with shaking in baby food jar and temporary immersion system in RITA bioreactor for Musa and Dioscorea spp. It was noted that the type of culture system had a significant effect on the number of polyshoot produced in vitro. In general, the temporary immersion system was more efficient for rapid propagation of Musa spp., owing to the higher number of shoots produced within a relatively short period of time compared to other conventional (semi-solid) methods. An optimal relatively short period of 21 days generated a multiplication rate of 28.40 and 12.93 for Musa AA and AAB respectively in TIS (RITA bioreactor). This can enable a faster rate of availability of plantlets for use and conservation. Similar results were observed by Roels et al. [31] at 28 days with multiplication https://doi.org/10.1371/journal.pone.0216717.g006 Table 5. Summary of band pattern of in vitro Dioscorea rotundata. Treatment  Culture system  MSL  NML   1  TDr 1228  T1  CI  518  231   2  TDr 1228  T1  TIS  334  311   3  TDr 1228  T1  SS  542  271   4  TDr 1228  T2  CI  468  129   5  TDr 1228  T2  TIS  366  130   6  TDr 1228  T2  SS  573  258   7  TDr 1228  T3  CI  397  176   8  TDr 1228  T3  TIS  380  242   9  TDr 1228  T3  SS  577  243   10  TDr 1228  T4  CI  375  197   11  TDr 1228  T4  TIS  338  270   12  TDr 1228  T4  SS  572  rate of 13 and Hui et al. [32] at 5 weeks with an average rate of 4.9. The studies of Businge et al. [33], Georgieva et al., [34] and Steinmacher et al., [35] also confirmed the positive effect of TIS over conventional in vitro propagation in other crops. This may be because TIS has the advantage of allowing better contact between the explant and the culture medium, thus easy diffusion and uptake of nutrient is achieved. The TIS nullifies the drawbacks of liquid culture medium by making the explant immersed only for a while and aerating them inducing relatively less stress on the tissues [36]. Conversely, there was no significant effect of culture system on the multiplication rate of Dioscorea spp. TDr 1228 within 6 weeks of culture. However, the lowest multiplication rate was observed in semi-solid culture medium followed by TIS and CI. This suggested that an increased nutrient-to-plant contact period was needed for increased multiplication rate of TDr 1228. Also, culture media composition influenced the multiplication rate of TDr 1228. The multiplication rate was highest in MS medium supplemented with 0.5mg/l Kinetin (T1) in CI, TIS and SS. While the absence of hormone supplement (T4) also supported growth in CI and TIS but minimal in SS. Polizin et al. [37], observed a similar trend, wherein there was no significant difference between TIS and SS at 8 weeks and suggested increasing immersion frequency as a possible way of optimizing the potential of TIS for Dioscorea spp. However, Yan et al. [38], observed significant difference in the growth rate of Dioscorea fordii and Dioscorea alata in TIS indicating that the multiplication rate in Dioscorea spp. may be genotype/cultivar dependent. TIS is also known to improve plant quality and increase shoot vigor as well as quantity of morphologically normal somatic embryos [39]. Hyperhydricity that seriously affects cultures in liquid medium is eliminated in TIS [10] as the explants are not permanently immersed. Hvoself-Eide et al., [40] confirmed that in TIS, there is increased multiplication rate when shoots are appropriately exposed to culture media at correct intervals. TIS provides an excellent way of using liquid medium at the same time controlling the gaseous environment thereby increasing the growth and multiplication rate of cultures. Also due to lack of agitation, the mechanical stress on plant tissues are low compared to other micropropagation methods.

S/N Accession
For in vitro plant tissue culture, type of growth regulator plays an important role for the physiological response of explants. A higher level of BAP supplement in MS medium promoted the production of polyshoot for Musa spp. As reported by several authors [41][42][43][44], polyshoot production in Musa spp. was improved by wounding the apical meristem to break the apical dominance thereby stimulating the axillary buds to produce multiple shoots in Musa. For Dioscorea rotundata, low sucrose concentration has been identified to reduce exudation of phenolic compounds, which may hamper regeneration and growth [45]. This in combination with no or low levels of plant hormone have been identified to promote its multiplication rate [46][47][48].
Molecular characterization of plants obtained from the three culture systems was essential to compare their genetic differences in relation to phenotypic characteristics. Amplified Fragment Length Polymorphism (AFLP) and Methylation Sensitive Amplification Polymorphism (MSAP) are useful molecular markers that help to understand effect of methylation on genetic diversity within and among a population. This is possible because AFLP marker is highly polymorphic and evenly distributed in the genome, giving a broad understanding of genomic variation. MSAP is a modification of AFLP marker that reveals methylation pattern in a population.
AFLP profiles for Dioscorea and Musa spp. under in vitro and subsequent field conditions revealed a level of conserved genetic variability across the genotypes. A 3% variation among the different culture systems was explained on the basis of presence of private alleles peculiar to each type of culture system. In addition, the pairwise genetic distances were calculated to investigate the allelic differences among the 3 culture systems. A low genetic distance was observed between SS and TIS for both Dioscorea and Musa spp. either under in vitro or field conditions, indicating the relatedness between the two culture systems.
In this study, MSAP using EcoRI/HpaII and EcoRI/MSpI as restriction enzymes to identify and cleave methylation regions, thus generating methylation profiles which can help in understanding genetic diversity between and among a population. In plant genome, DNA methylation is a common phenomenon that does not alter the main genetic code but may show somatic or phenotypic variations. Schulz et al [49] and Herrera and Bazaga [50] described different forms of cytosine methylation that explained the basic principle of methylation scoring and profiling. Cleavage of methylated cytosine by MspI and HpaII could either result in full methylation (when the internal cytosine of the double stranded DNA is methylated) or hemimethylation (methylation of internal cytosine on one DNA strand) with the exception that HpaII cleaves external cytosine. CG methylation is said to be an important factor for promoter function [51]. This is evident in the work of Hafiz et al [52] who made it clear that DNA methylation plays a significant role in the transition from vegetative to reproductive growth and ploidy level of plants. Polymorphism in DNA methylation is an important form of genetic variation which plays a significant role in cell division, higher growth rate of plants, rooting ability and a potential capacity of silencing plant viruses [53][54].
Methylation profiles generated from Musa and Dioscorea spp. under field and in vitro conditions revealed a significant level of full and partial methylation pattern. The value of Shannon diversity accounts for the richness and evenness of MSL and NML for Musa and Dioscorea spp. under field and in vitro conditions. The level of diversity of MSL under in vitro (1.78) condition-was higher than field (1.59) condition for Musa spp. A similar trend was observed for Dioscorea spp. The relative lower diversity value under field condition may be a reflection of environmental influence on the crops. It has been reported that factors such as plant growth hormone, increased level of salt, biotic or abiotic stress may contribute to methylation/genetic variation in crops. This was confirmed with the number of MSL for Dioscorea spp. across four hormone treatments and the 3 culture systems used under in vitro conditions. Plants grown in TIS recorded lowest number of MSL across all hormone treatment used while SS system had the highest number of MSL. A closer look at the hormone combination revealed that T1 and T4 had a lower number of MSL compared to other hormone combinations. LoSchiavo et al [55] and Arnholdt-Schmitt [56] reported hypo-methylation with increasing level of cytokinin (kinetin) in carrot root while higher concentration of auxin (2, 4-D) increased methylation level from 15 to 70%. This explained how plants react or adjust to stressful conditions when developing different cell types [57][58]. Rico et al [59] also confirmed increase in hemi-methylation level and decrease in full methylation of drought effect on forest trees. The DNA methylation highlights the capacity of plants to acclimatize and adapt to changing environmental conditions. An exponential of the Shannon diversity index provided information on the effective number of species, which is the actual measure of diversity as it shows the richness and evenness of a population [60][61].

Conclusion
The present study confirmed the advantage of temporary immersion system (TIS, RITA bioreactor) in improving the multiplication rate of polyshoot production in both Musa and Dioscorea spp. The adoption of TIS over other propagation system is to assist in overcoming the challenges of mass production of good quality planting materials within a relatively short period of time and at a lower cost. The suitability of tissue culture-based system depends on their effect on genetic uniformity. Methylation-sensitive amplification polymorphism (MSAP) is a valuable tool for detecting methylation, which could be a potential indicative signal of possible somaclonal variation in clonal crops with respect to the culture duration and/or systems. The culture systems used in this study did not show significant alteration on the genetic integrity of Musa and Dioscorea spp. The high level of genetic polymorphism showed the ability of the culture system to conserve crop genetic variability, which can make the crop adaptable and promote their use and conservation in genebanks, breeding and biotechnology programs. However, factors such as plant growth hormone, culture system type, mode of propagation and induced stress revealed the cause of variation in plants. It was also observed that certain type of plant growth hormone could either trigger increase or decrease in methylation, which could lead to activation or deactivation of certain genes in the plant genome. The variation observed is marked by increase or decrease in methylation events and could be further explored to understand and assess epigenetic changes in these two plants.
Supporting information S1  Table. ANOVA summary