• methylation-sensitive amplified fragment length polymorphisms;
  • microsatellites;
  • methylation-sensitive amplified polymorphism;
  • somaclonal variation;
  • somatic embryogenesis;
  • single sequence repeat;
  • Theobroma cacao


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Relatively little is known about the timing of genetic and epigenetic forms of somaclonal variation arising from callus growth. We surveyed for both types of change in cocoa (Theobroma cacao) plants regenerated from calli of various ages, and also between tissues from the source trees.
  • For genetic change, we used 15 single sequence repeat (SSR) markers from four source trees and from 233 regenerated plants. For epigenetic change, we used 386 methylation-sensitive amplified polymorphism (MSAP) markers on leaf and explant (staminode) DNA from two source trees and on leaf DNA from 114 regenerants.
  • Genetic variation within source trees was limited to one slippage mutation in one leaf. Regenerants were far more variable, with 35% exhibiting at least one mutation. Genetic variation initially accumulated with culture age but subsequently declined. MSAP (epigenetic) profiles diverged between leaf and staminode samples from source trees. Multivariate analysis revealed that leaves from regenerants occupied intermediate eigenspace between leaves and staminodes of source plants but became progressively more similar to source tree leaves with culture age.
  • Statistical analysis confirmed this rather counterintuitive finding that leaves of ‘late regenerants’ exhibited significantly less genetic and epigenetic divergence from source leaves than those exposed to short periods of callus growth.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The regeneration of plant material from in vitro culture has long been associated with the occasional to frequent appearance of individuals exhibiting some form of phenotypic abnormality (Lin et al., 2006; Zhao et al., 2006). This phenomenon is known as ‘somaclonal variation’ and has two main causes: de novo genetic mutation and epigenetic change. The absence of phenotypic variation does not necessarily mean a concomitant lack of genetic or epigenetic variation, with some seemingly normal plants nevertheless containing major (biochemical) flaws of commercial significance (e.g. Tsuro et al., 2001). It is therefore important to measure both causes of variation for any in vitro regeneration system intended for commercial or large-scale use.

The passage of plant material through in vitro culture often results in a marked increase in the abundance of genetic variants (Phillips et al., 1994) to rates more normally associated with the use of mutagens (e.g. Kharabian & Darabi, 2005; Burg et al., 2007). Types of genetic change that arise during the regeneration process range in scale from genome duplication (autopolyploidy) down to point mutations. The numerous reports of spontaneous genome doubling include petal-derived azalea regenerants (De Schepper et al., 2001, 2003) and polyploid callus cell lines of Arabidopsis (Fras & Maluszynska, 2004). Minor changes to chromosome number (aneuploidy) appear also to be common, featuring in megagametophyte-derived larch (Larix decidua) embryos (Von Aderkas et al., 2003) and somatic embryo-derived asparagus (Asparagus officinalis) (Raimondi et al., 2001). There are also several reports of more subtle changes, varying from large deletions (63–76 kb) of the chloroplast genome in rice (Oryza sativa) calli (Abe et al., 2002) through to point mutations leading to the creation of a giant embryo mutant in rice (Park et al., 2009) or to a nonfunctioning Adh1 (alcohol dehydrogenase 1) mutant in maize (Zea mays) (Dennis et al., 1987).

In addition to genetic change, there can be profound epigenetic perturbations associated with the regeneration of plants from culture. Most studies have used the genome-wide but anonymous marker system methylation-sensitive amplified polymorphism (MSAP) to investigate global changes to the distribution of cytosine methylation. Indeed, several authors have used this technique to uncover tissue culture-induced epigenetic variation in a wide taxonomic spread of plant species (Nicotiana tabacum, Schmitt et al., 1997; rice (Oryza sativa), Xiong et al., 1999; strawberry (Fragaria Xananassa), Hao et al., 2002; potato (Solanum tuberosum), Joyce & Cassells, 2002; Arabidopsis, Bardini et al., 2003; oilpalm (Elaeis guineensis), Jaligot et al., 2004). The attachment of a methyl group to cytosine residues is an important feature of epigenetic regulation, and a series of locus-specific studies have implicated C-methylation in the targeted silencing of genes involved in developmental progression or tissue differentiation (e.g. Messeguer et al., 1991), in the response of plants to abiotic or biotic stress (e.g. Bossdorf et al., 2008), and in the repression of viruses and retroelements (Wassenegger et al., 1994; Ingelbrecht et al., 1999). Widespread alteration in the DNA methylation profile of a genome has the potential to drive physiological or developmental variation among plants recovered from tissue culture. Characterization of such change has utility for the diagnosis and quantification of epigenetic change amongst regenerants, and may even provide information on the nature of the explant material used. For example, Joyce et al. (2003) proposed that cultured cells may be partially ‘developmentally imprinted’ so that, while capable of continued organized growth in vitro, the organs formed may retain some characteristics of the phase of growth at which the explants were excised. This characteristic could be partly attributed to incomplete cell de-differentiation processes that occur during callus growth, as in Arabidopsis (Bardini et al., 2003). In a study by Henry (1998), in which epigenetic changes associated with callus-phase de-differentiation were recorded, DNA methylation features characteristic of tomato (Solanum lycopersicum) callus were detected in regenerant plants and even a proportion of their progeny. Tissue culture medium components such as antibiotics (hygromycin, kanamycin and cefotaxime) commonly used as selective agents in transgenic plant production cause hypermethylation in Nicotiana tabacum (Schmitt et al., 1997) and hypomethylation in calluses of Arabidopsis (Bardini et al., 2003). However, changes of this type are not limited to passage through a callus stage, as illustrated by the epigenetic variation reported among micropropagated potato plants, multiplied via callus-free regeneration (Joyce & Cassells, 2002). There have also been correlative links made between physiological changes arising from culture and associated changes to the methylome. Kaeppler & Phillips (1993) noted links between variability in flowering time among maize adventitious regenerants and an increased frequency of modification of cytosine methylation patterns in their genomic DNA. Thus, global assessments of methylation status provide a useful means by which to measure epigenetic ‘somaclonal’ changes associated with passage through in vitro culture.

Cocoa (Theobroma cacao) is a tropical tree crop with a typical juvenile period of 3 yr when grown from seed. The crop is relatively poorly developed from its wild relatives (Cuatrecasas, 1964) and is susceptible to many pests and diseases (Kenedy & Mooleedhar, 1992). Difficulty in generating commercial quantities of cocoa ramets via cuttings has led to an increasing interest in the application of somatic embryogenesis for clonal multiplication (Lopez-Baez et al., 1993; Li et al., 1998; Traore et al., 2003). However, cocoa plants recovered via somatic embryogenesis carry elevated numbers of genetic mutations (Rodríguez López et al., 2004), some of which may contribute to the phenotypic variation reported in these plants (e.g. Lanaud, 1998; Rodríguez López et al., 2004). The aim of the current study was therefore to compare the genetic and epigenetic changes during normal somatic growth of cocoa source trees with that found in plants regenerated from primary and secondary cocoa somatic embryos.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Plant material

Cocoa trees (Theobroma cacao L.) held in the intermediate quarantine facility of the University of Reading, Reading, UK were used to supply explant material. Closed immature flower buds (4–5 mm) were collected from genotypes LCT EEN 37/I, LCT EEN 162/S-1010, SC 3, and SIAL 93 between 08:00 h and 12:00 h. Staminodes were isolated from these buds and used to induce and regenerate somatic embryos according to Li et al. (1998).

DNA isolation

DNA was extracted from recently expanded leaves of ‘maternal’ cocoa trees used as a source of explant material, and from the cotyledons and leaves of regenerated plantlets using the DNeasy 96 Plant Kit (Qiagen, UK) and the Mixer Mill MM 300 (Retsch, Haan, Germany) according to the manufacturers’ instructions. Isolated DNA was diluted in nanopure water to produce working stocks of 10 ng μl−1. Twenty separate DNA leaf extractions from each maternal tree and two replicate extractions from leaves of each regenerant plantlet were performed.

Single sequence repeat (SSR) analysis

To evaluate variation within maternal trees, SSR-PCRs were performed on DNA samples from 20 phenologically dispersed leaves from ortet trees (LCTEEN 37/I, LCTEEN 162/S-1010, SC3, and SIAL93) using primers mTcCIR15, mTcCIR3, mTcCIR17, mTcCIR10, mTcCIR6, mTcCIR7, mTcCIR1, mTcCIR8, and mTcCIR61 (Supporting Information Table S1) according to the protocol of Lanaud et al. (1999). In all cases, one primer was 5′-end-labelled using 6-carboxyfluorescein (FAM), 6-carboxy-2′,4,4′,5′,7,7′-hexachlorofluorescein (HEX) or NED (Applied Biosystems, Inc, Carlsbad, California, USA) to allow product detection during capillary gel electrophoresis on an ABI 373XL (Applied Biosystems, Inc, Carlsbad, California, USA). Variation among regenerated plants was assessed using the same primers and conditions applied to template DNA extracted from a random selection of in vitro regenerants (LCTEEN 37/I (13 samples), LCTEEN162/1010 (180 samples), SC3 (20 samples), and SIAL93 (20 samples)). Products were visualized on the resultant electropherograms and compared to identify cases of de novo slippage mutation. Profiles were only considered to provide evidence of novel genetic change among the regenerants when both replicate DNA extractions exhibited the same deviation from the maternal SSR profile. Regenerant ‘off types’ were then re-examined along with 51 randomly selected samples from the ‘wild-type’ regenerants in which no deviations were noted from the parental profile using the following additional set of SSR primers: mTcCIR19, mTcCIR18, mTcCIR2, mTcCIR26, mTcCIR25, and mTcCIR24.

DNA sequencing

Amplification products obtained using primer pair mTcCIR15 from the wild-type and off-type maternal samples were subjected to direct cycle sequencing using the ABI PRISMTM Big DyeTM terminator cycle sequencing ready reaction kit (Applied Biosystems). Twenty single-stranded sequencing reactions were performed per sample (10 for each forward and reverse primer). Each of these included 3 μl of PCR product, 0.16 μl of the primer, 4 μl of the Terminator Ready Reaction Mix and 2.8 μl of sterile nanopure water. The thermal profile used for all PCRs was: 35 cycles of 30 s at 96°C, 15 s at 50°C, and 4 min at 72°C. The extension products were purified and then fractionated by capillary electrophoresis on an ABI 373XL automated sequencer according to the manufacturer’s protocols (Applied Biosystems).

MSAP procedure

A modification of the MSAP methods of Reyna-Lopez et al., (1997) and Xiong et al. (1999) was used on the 20 separate DNA extractions from phenologically dispersed leaves collected at the same time from each explant source tree and on DNA samples from leaves of a random selection of somatic embryogenesis regenerants of LCT EEN 162/S-1010 (96 samples) and SIAL 93 (17 samples). The method involves digestion of genomic DNA with methylation-sensitive or -insensitive restriction enzymes (isoschizomer pairs), ligation of adaptors and selective PCR amplification with primers complementary to the adaptors but with unique 3′ overhangs.

Enzyme choice and design of adapters and primers  In the MSAP method, the original amplified fragment length polymorphism (AFLP) technique is modified to incorporate methylation-sensitive isoschizomer restriction enzymes. Central to the technique is the differential behaviour of the two restriction enzymes in the presence of cytosine methylation. We therefore verified this property using DNA fragments (synthesized by Sigma, UK) that were sequence-identical but differed in methylation content. These were ligated into plasmid pCR2.1-TOPO (Invitrogen), diluted to approximately the same template concentration, restricted with the isoschizomers and subjected to real-time quantitative PCR (RT-qPCR) amplification with plasmid-specific primers (see Methods S1). In this way, we assayed for differential sensitivity of the isoschizomers to methylation. The design of new adapters and primers (Transgenomics, Glasgow, UK) was necessary (see Table S2). The designations primer + X, primer + XY, or primer + XYZ are used henceforth to indicate primers modified by addition of extra nucleotide(s) at the 3′-end of the primers. In order to ensure that only fragments generated by methylation-sensitive digestion (i.e. those delimited by one or two TCCGGA sites) would be visible, only BseAI/MroI primers were end-labelled using 6-FAM. Thus, all products of selective amplification using BseAI contain one (or two) TCCGGA sites that could contain methylated cytosine residues. Conversely, amplicons produced using MroI lack methylated cytosine residues.

Enzymatic digestion of DNA  The restriction enzymes used in this study were Tru9I, BseAI (Thanos et al., 1989) and MroI (Kato et al., 1988). BseAI and MroI recognize the sequence T/CCGGA and generate fragments with the same 5′-cohesive ends. Conversely, MroI is inhibited by 5-methylcytosine, while its isoschizomer BseAI is largely insensitive to methylation of the cytosine residues on the recognition site. As MroI and Tru9I do not perform effectively using a common incubation temperature, a two-step reaction was implemented: genomic DNA (55 ng) was initially restricted using one unit of Tru9I and 5 units of BseAI or MroI for 2 h at 37°C and then for a further 1 h at 60°C. The enzymes were then inactivated by heating to 75°C for 15 min. In all cases, 5 μl of restriction products were size-fractionated by electrophoresis through a 2.5% w/v metaphor agarose gel for 2–3 h to confirm complete digestion.

Ligation of adapters to restricted DNA  We annealed and ligated 1 μl of rare and frequent cutter double-stranded adapters to 10 μl of genomic DNA that had previously been digested into fragments using 1 unit of T4 DNA ligase (Roche) in a final volume of 15 μl of ligase buffer (Roche) for 2 h at room temperature. Ligation products were diluted 1 : 10 in nanopure water and stored at −20°C.

Preselective PCR  For preselective amplification, 3 μl of the diluted ligation products described above was incubated with 0.5 μl of preamplification primer + X mix (both primers at 10 μM) (see Table S2 for preselective primers sequences), 1 unit of Bioline Taq (0.2 μl), 2.5 μl ×10 reaction buffer, 0.75 μl of MgCl2 50 mM buffer and 0.3 μl of dNTP mixture (Bioline, London, UK) in a final volume of 25 μl. PCR conditions were 2 min at 72°C followed by 30 cycles of 94°C for 30 s, 56°C for 30 s and 72°C for 2 min with a final extension step of 10 min at 60°C. The preselective amplification PCR products (10 μl) were separated on a 1.5% w/v agarose gel by electrophoresis to confirm that PCR amplification was successful. The remaining 15 μl of preselective product was diluted in 100 μl of nanopure water and stored at −20°C.

Selective PCR  We mixed 2.5 μl of diluted preamplification products with 1.25 μl of ×10 reaction buffer (Bioline), 0.25 μl of +2Tru9I primer, 0.15 μl of +3BseAI/MroI primer (both primers at 10 μM), 1 unit of Taq (0.2 μl), 0.5 μl of MgCl2 50 mM buffer, 0.15 μl of dNTP mixture and nanopure water to a final volume of 12.5 μl. All 24 +2Tru9I and +3BseAI/MroI primer combinations were evaluated using the preamplification products obtained from the cocoa maternal tree DNA samples restricted both with Tru9I and BseAI, and with Tru9I and MroI. This was done to assess the level of intra-varietal variation for each primer combination and the ability of each primer combination to generate informative and consistent MSAP profiles. Primer combinations T+2(1)/Brc+3(1), T+2(1)/Brc+3(6), and T+2(2)/Brc+3(4) (see Table S2 for selective primer sequences) were chosen for the comparative selective amplification of the regenerant leaf DNA and those from the maternal staminode and leaf tissues. Aliquots (5 μl) of the selective amplification products were tested on agarose and the remaining 15 μl of selective amplification PCR product was stored at −20°C.

Sample fractioning  Labelled SSR-PCR and MSAP products were diluted 1 : 10 in nanopure sterile water and 1 μl was combined with 1 μl of ROX/HiDi mix (50 μl ROX plus 1 ml of HiDi formamide). Samples were heat-denatured at 95°C for 3–5 min and snap-cooled on ice for 2 min. Samples were fractionated on an ABI PRISM 3100 at 3 kV for 22 s and at 15 kV for 45 min at 60°C. Resultant profiles were analysed using Genotyper 3.7NT software (Applera Corporation, Norwalk, Connecticut, USA).

Data analysis

For the analysis of the effect of tissue type and of in vitro culture on the production of polymorphisms detected using MSAP, reproducible products were scored as present (1) or absent (0) to form a raw data matrix. For this analysis, the small number of markers exhibiting polymorphisms between MSAP profiles obtained from replicate DNA samples extracted from the same tissue types in ortet plants were discarded. However, profile polymorphisms between DNA samples from the same plant but extracted from different tissues (floral and leaf tissue) were retained as inter-tissue methylation differences. Finally, markers present in all replicates obtained from each ortet plant were selected as true-to-type products for each genotype.

Estimation of tissue-dependent epigenetic distance  The epigenetic distance between cocoa floral and leaf tissues from genotypes LCT EEN 162/S-1010 and SIAL 93 was estimated using Jaccard’s similarity coefficient (Patwary et al., 1993). In this way the genetic distance between compared samples is assigned a value between 0 and 1, where 1 indicates no genetic distance detected between samples. Epigenetic distances between tissues obtained from MSAP profiles using MroI (cytosine methylation sensitive) and BseAI (insensitive) were compared in order to establish which enzyme reveals greater differences between cocoa tissue types.

Estimation of culture-induced genetic and epigenetic distance  In contrast to the differences detected between tissues from the same plant, when tissues from ortet and regenerants are compared, differences are likely to be attributable both to genetic and to epigenetic forms of genome change. Genetic and epigenetic similarities between three groups of tested samples (maternal leaf and staminode tissues, and leaf tissue from somatic embryo-derived plants) were determined by principal coordinates Euclidean analysis (PCoA) (Gower, 1966) based on the MSAP profiles obtained from all three primer combinations using mvsp software (Kovach Computing Services, Anglesey, Wales, UK). Somatic embryo-derived samples were grouped for PCoA considering different in vitro culture variables (i.e. culture age, and primary or secondary embryogenesis). The effect of these variables on the induction of somaclonal variation was also estimated comparing the genetic/epigenetic distances (calculated using Jacard’s similarity coefficient). Comparisons were made between cocoa regenerants from genotypes LCTEEN 162/S-1010 (96 samples) and SIAL 93 (17 samples) and floral and leaf tissues from donor plants from both genotypes. LCT EEN 162/S-1010 (96 samples) and SIAL 93 (17 samples) applied to the MSAP profiles.

The effect of culture age on somaclonal variation was then tested by grouping screened regenerant samples according to their time of excision from the source callus. The mean genetic/epigenetic distance between each time group (i.e. 8, 10, 12, 14, 16, 18, 20 and 22 weeks on ED medium) (Li et al., 1998) was then compared with that observed between maternal leaf/staminode samples. As all studied embryos underwent this subculture at the same stage of development, the time of excision from the callus is related to the culture time taken to generate each somatic embryo. SIAL 93 samples were subcultured after 10, 12, 14, and 18 wk on ED medium (Li et al., 1998), while LCT EEN 162/S-1010 samples were subcultured after 8, 10, 12, 14, 16, 20, and 22 wk on ED medium.

LCT EEN 162/S-1010 embryo-derived plants divided into primary embryos (83 samples) and secondary embryos (13 samples). The influence of primary and secondary embryogenesis on somaclonal variation was tested by comparing the estimated mean genetic/epigenetic distance of regenerated embryos with the donor plant tissues via each type of embryogenesis.

We used an adaptation of Paetkau’s method specific for AFLP markers to test the significance of PCoA clusters identified using MSAP data. For this we used the software aflpop (Duchesne & Bernatchez, 2002) to assemble 500 simulated genotypes for each subpopulation (early vs late regenerant populations, and also primary vs secondary embryos) using the observed allele frequencies from each. The software then calculated the likelihood that either the leaf or the staminode profiles from the source tree could be allocated to either regenerant subpopulation. A log-likelihood difference was set at = 2 to establish a significant difference in the likelihood of allocation to the different subpopulations (this imposes the condition that the likelihood of allocation to one population has to be 100 times higher than that to the other for an allocation to be deemed significant).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Somatic variation in source material

Seven hundred and twenty SSR profiles were obtained for 20 independent DNA extractions from each source tree (LCTEEN 37/I, LCTEEN 162/S-1010, SC3, and SIAL93) using primers mTcCIR15, mTcCIR3, mTcCIR17, mTcCIR10, mTcCIR6, mTcCIR7, and mTcCIR1. Profiles were conserved between replicated leaves in all parents except for locus mTcCIR15 (linkage group 1) in SIAL93. This locus revealed an off-type regenerant (Fig. 1a) in which one parental allele had reduced in size. Sequencing of the three alleles revealed that the difference was attributable to loss of a dinucleiotide GA repeat motif (Fig. S1).


Figure 1.  Detection of genetic and epigenetic changes in cocoa plants using single sequence repeat (SSR) markers and methylation-sensitive amplified polymorphism (MSAP). (a) Slippage mutation in the 238-bp allele of mTcCIR15 in parental plant SIAL93. (b) Allele gain by slippage mutation of mTcCIR6 in a somatic embryo-derived cocoa plant from genotype LCT EEN 162/S-1010. (c) Allele gain for mTcCIR7 in a somatic embryo-derived plant from genotype LCT EEN 37/I. (d) Allele loss in mTcCIR9 in a somatic embryo-derived plant from genotype SIAL93. Allele size is indicated below each peak. (e) MSAP electropherograms of leaf and staminode tissues from the LCT EEN 162/S-1010 genotype generated using MroI and BseAI, and primers T+2(1)/Brc+3(1), T+2(1)/Brc+3(6), and T+2(2)/Brc+3(4). MSAP traces from donor plant leaves obtained using MroI; β trace for leaves using BseaI; δ trace for staminodes using MroI; ε trace for staminodes using BseaI. Polymorphic peaks (indicated by arrows in grey areas) within each trace are indicated as follows: 1, class II, present only in one of the tissues when both enzymes were used. 2–3, class I, products whose presence is tissue-specific when restricted with MroI or BseAI, respectively; 4, class III, products appear in both tissues only when restricted with BseAI.

Download figure to PowerPoint

Isoschizomer characterization

The use of MroI to cleave methylated DNA template before PCR resulted in amplification profiles that closely matched those of the untreated template, indicating that the template had failed to restrict. The finding thereby illustrates the sensitivity of this enzyme to cytosine methylation. Only weak and late amplification was observed after 35 cycles from the unmethylated template (as is used in MSAP preamplification), demonstrating a clear differential ability of MroI to cleave unmethylated template. In comparison, BseAI cleaved the same methylated and unmethylated templates with greater efficacy relative to the same untreated control (Fig. S2). Thus, we inferred that MroI and BseAI could be used in combination as isoschizomers for MSAP.

Variation in methylation patterns between cocoa tissues

Methylation patterns were deduced from fragment polymorphisms generated by MSAP (using MroI and BseAI) for each tissue and every sample from LCTEEN 162/S-1010 and SIAL93. Once the small number intra-tissue polymorphisms were discarded, comparison of the leaf and staminode profiles revealed three types of differences (Fig. 1e): class I products, whose presence was diagnostic of tissue type but which only appeared in profiles generated by one enzyme; class II products, which also appeared only in one of the tissues but did so when both enzymes were used; and class III products, which appeared in both tissues only when restricted with the C-methylated insensitive enzyme BseAI. In general, fewer fragments amplified using the C-methylated sensitive enzyme (MroI) than with the C-methylated insensitive enzyme (BseAI), and also fewer in staminodes than in leaves (Table S3). Overall, when the total numbers of products obtained from both tissues and using all three primers were compared in both genotypes, unmethylated TCCGGA sites (products common to both digests) were the most abundant. For the remaining loci, products present in BseAI but not in MroI and products present in MroI but not in BseAI lanes accounted for 4.2% and 10.6% of the leaf and staminode profiles, respectively, from LCTEEN 162/S-1010 and for 2.6% and 5.8%, respectively, of those from SIAL 93. Overall, staminode profiles contained approximately twice the number of methylated markers as were observed in leaf profiles.

Thus, although minimal genetic variation was detected between samples from the same source tree, there was significant epigenetic variation between tissues from the same plant.

Detection of somaclonal variants using SSR markers

Overall, we scored 4194 SSR profiles from 466 DNA extractions (two per in vitro sample) across nine loci, revealing 5676 amplicons. Eighty-one in vitro samples, 34.8% of the total somatic population (LCTEEN 162/S-1010, 62 out of 180; LCTEEN37I, 5 out of 13; SC3, 4 out of 20; and SIAL93, 10 out of 20), exhibited some form of deviation from the parental genotype. Observed polymorphisms divided into two categories (Fig. 1b–d): (1) allele loss and (2) DNA strand slippage mutation leading to the formation of new alleles. In all, we found 148 polymorphisms using the first set of nine SSR primers. These comprised 87 allele losses and 61 new alleles. The abundances of the two classes of polymorphism differed significantly (χ2 = 0.002, α ≤ 0.001).

Assuming that mutational events were independent, the probability that any single regenerant carries a mutant allele was estimated to be 0.0353 (148/4200, dividing the total number of mutations detected by the total number of alleles analysed). However, the number of samples detected carrying one, two, three, four, five or six mutations was 44, 19, 9, 7, 1, and 1, respectively. The frequency of samples carrying more than one mutation was therefore significantly higher than expected if mutational events were truly independent (χ2 = 2.45 × 10−6, α ≤ 0.001).

Inferred chromosome loss/major deletions using linked SSR markers  Somatic nondisjunction could lead to aneuploidy through the gain or loss of chromosomes. We estimated the maximal frequency of chromosome loss by surveying for the shared elimination of SSR markers mapped from the same chromosome (noting that major deletions would also exhibit similar features). When 81 polymorphic regenerants from all genotypes (containing 57 allele losses across six SSR loci) were screened for variation at a further six heterozygous SSR loci, an additional 24 allele losses were detected. Of these, just six of 233 regenerants (2.6% of plants screened) had lost an allele from two linked SSR markers and so could be considered as possible cases of chromosomal loss (Table 1).

Table 1.   Detection of chromosome loss induced during somatic embryogenesis detected using single sequence repeat (SSR) markers
 Cocoa genotypeTotal
LCT EEN 37/ILCT EEN 162/S-1010SC 3SIAL 93
  1. Presumption of chromosome loss was based on the coincidence of allele loss in two SSR markers linked to a single chromosome. Number of samples presenting allele loss and number of allele loss mutational events were scored for the first (A) and second (B) sets of primers. The ratio of chromosome loss to allele loss was obtained by dividing the number of coincidences by the total number of allele losses using both sets of primers (A + B).

No. plants tested10624581
No. mutant samples A–B1–127–132–29–439–20
% mutant samples2055757055.5
No. allele losses A–B1–145–152–29–657–24
No. allele loss coincidences03126
% chromosome loss/allele loss052513.37.4

Identification of factors involved in genetic and epigenetic instability

A random selection of 113 somatic embryo-derived clonal lines (96 samples from genotype LCT EEN 162/S-1010 and 17 from SIAL 93) were subjected to MSAP analysis. Minimal genetic variation but significant epigenetic variation was anticipated between tissue samples from the source trees, although SSR evidence had shown that there is significant genetic variation amongst the regenerants. In accordance with this expectation, the first principal component of the PCoA generated by the methylation-tolerant BseAI included comparatively little variation between tissue types; accounting for just 0.36 of that seen between regenerants. By contrast, variability between tissues predominated in the MroI MSAP profiles, and was 3.3 times greater than that seen between regenerants (i.e. a ninefold increase). This suggests that the vast majority of variation revealed by MroI is epigenetic in nature. Moreover, the PCoA using MroI data indicated wide separation between leaf and staminode tissues from the source tree, with the profiles of regenerant leaves being somewhat intermediate although closer to source leaf tissue (Fig. 2a).


Figure 2.  Effect of time in culture on genetic/epigenetic instability. (a, c) Principal coordinate diagrams based on the Euclidian analysis of methylation-sensitive amplified polymorphisms (MSAP) (a) and amplified fragment length polymorphisms (AFLP) distances (c) using primer combinations T+2(1)/Brc+3(1), T+2(1)/Brc+3(6), and T+2(2)/Brc+3(4), between tissues from the LCT EEN162/1010 ortet plant (leaf and staminode (a), leaf (c) and 96 somatic embryo derived samples grouped by callus age (8–20 or more (a) and 8–16 or more (c) weeks in ED medium indicated by the number associated with each symbol). Larger graphic, genetic/epigenetic distances between all the samples studied, smaller graphic, calculated mean distances for each of the groups. (a) The dashed line separates samples generated before and after 10 wk in ED medium. Distances between source tree tissues can be both genetic and epigenetic. (b) Detected ratio of variation using SSR markers. Vertical axis, number of variant regenerants/total number of regenerants; horizontal axis, culture age, measured in weeks, at which embryos were isolated. Grey columns, variation in the total population using the first set of markers; white columns, variation in off-type samples using the second set of markers; black columns, variation in wild-type regenerants usin the second set of markers. P-lv, parental leaves; P-st, parental staminodes.

Download figure to PowerPoint

Effect of culture age on somaclonal variation  The somaclonal variation detected by SSR analysis was found to gradually increase to a peak at 12 wk and thereafter to decrease. This pattern of novel mutation fitted a second-order polynomial regression influenced by different variables (time in culture, marker used, and genotype). The computational analysis revealed a statistically significant interaction between time in culture, mutation rate and the combination genotype/chromosome analysed. The simple regression model applied to the data followed the formula:

  • image(Eqn 1)

(g(x), the estimated mutation ratio; μ, the intercept, which varies according to which genotype/SSR marker was analysed; t, weeks in culture) (Fig. 2b).

Regenerants that diverged from the explant genotype on the basis of SSR markers used in the first experiment also showed a much increased tendency to contain nonexplant alleles when the regenerant population was re-examined using an independent second set of SSR markers (t-test, = 0.0026). Intriguingly, the detected variation in this set of samples sharply decreased after 10 wk in culture (Fig. 2b). Control samples (i.e. those showing the explant genotype in the first experiment) followed the same pattern as that seen in the population as a whole using the first set of SSR markers but at a markedly lower rate (Fig. 2b). Here, a peak in mutations was observed after 12 wk but thereafter mutants of both genotypes declined such that the lowest frequency of aberrant genotypes was noted in the final week of the experiment (wk 22). Interestingly, a similar pattern was also seen in the variation as revealed by AFLP using the methylation-tolerant BseAI. In this case, PCoA plots showed that leaf samples from plants regenerating after 12–14 wk were the most divergent from leaves of the source trees, and those appearing beyond 16 wk were the most similar (Fig. 2c).

The PCoA from the tissue-specific class I products (MroI, class Ia and BseAI, class Ib) emphasizes epigenetic differences between the donor tree tissues (staminodes and leaves) and leaves of the regenerants. This plot revealed a strong association between culture age and affinity of the regenerant leaves to those of the donor tree, with regenerants showing an increased resemblance to donor leaves with time in culture (Fig. 2a). In this instance, profiles of regenerants secured after 10 wk became markedly more similar to those of the source tree leaves.

We used an adaptation of Paetkau’s method specific for AFLP markers to test the significance of these differences. For this, we evaluated whether source leaf and staminode profiles were significantly more likely to be allocated to either late or early regenerant subpopulations. We found that the source leaves were significantly more likely to be allocated to the late regenerants than to the early regenerant group (= 0.33 for allocation to late regenerants compared with = 0.002 for allocation to early regenerants; = 2). The source staminodes were not allocated to either subpopulation (< 0.001; = 2).

Effect of primary vs secondary embryogenesis  We used SSR analysis to compare genetic mutation rates among 43 of the 48 secondary somatic embryo-derived plants regenerated from genotypes LCT EEN 162/S-1010 (30 samples), LCT EEN 37/I (three samples), and SC 3 (10 samples) with primary embryos regenerated during the same experiment (Table 2). Overall, regenerants obtained via secondary somatic embryogenesis exhibited a lower proportion of off-types (10 : 43; 23.3%) than were observed among primary embryo regenerants (71 : 189; 37.6%) (t-test, = 0.036). Of the 32 true-to-type secondary embryos, the majority (27) were produced by true-to-type primary embryos.

Table 2.   Effect of primary and secondary embryogenesis on the in vitro culture-induced somaclonal variation detected using single sequence repeat (SSR) markers
Cocoa genotype% primary off-types% secondary off-typesTotal (%)
  1. Percentage of primary and secondary embryos presenting somaclonal variation is shown for four genotypes. The total number of screened samples is shown in brackets for each genotype and embryogenic event. (–) Genotype SIAL 93 did not produce secondary embryos.

LCT EEN 162/S-1010135.6 (149)30 (30)34.4
LCT EEN 37/I40 (10)33 (3)38.5
SC 340 (10)0 (10)20
SIAL 9350 (20)50

Some 13 of the 96 somatic embryo-derived plants tested using MSAP from genotype LCT EEN 162/S-1010 were secondary embryos. These regenerants behaved similarly to primary embryo-derived plants. Here, once again profiles occupied intermediate locations between source tree tissues but slightly closer to donor tree leaves than primary embryos (Fig. 3). Class I products exhibited two major types of behaviour when primary and secondary embryos were compared. First, products present only in ortet tree staminodes tended to be less frequent in the secondary embryo-derived regenerants. Secondly, amplification fragments exclusive to source tree leaves tended to appear more frequently on samples regenerated through secondary embryos (Fig. S3). Overall, therefore, MSAP profiles of leaves from plants derived from secondary embryos were genetically and epigenetically more similar to the profiles associated with parental leaves than those generated from source staminodes (Fig. 3). However, when the same adaptation of Paetkau’s method analysis as used above was applied, this difference was found not to be significant (data not shown).


Figure 3.  Effect of primary and secondary embryogenesis on epigenetic instability. Principal coordinate diagrams based on the Euclidean analysis of methylation-sensitive amplified polymorphism (MSAP) distances obtained from class I cococa products (those whose presence is dictated by the tissue studied), using primer combinations T+2(1)/Brc+3(1), T+2(1)/Brc+3(6) and T+2(2)/Brc+3(4), between leaf and staminode tissues from the LCTEEN 162/1010 ortet plant (P-lv, parental leaves; P-st, parental staminodes) and 96 somatic embryo-derived samples grouped by primary (closed romboids) and secondary embryos (open romboids). Distances between source tree tissues should be epigenetic in origin, while distances between regenerants, and between regenerants and source tree tissues, can be both genetic and epigenetic. Larger graphic, genetic/epigenetic distances between all the samples studied; smaller graphic, calculated mean distances for each of the groups.

Download figure to PowerPoint


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Genetic variation

As expected, we found elevated frequencies of mutant genotypes among somatic embryogenesis regenerated clones when compared with those found in the parental plants (Rodríguez López et al., 2004). Indeed, the initial nine SSR markers revealed deviant profiles in 34.8% of regenerants compared with < 1.2% in leaves from source trees. Perhaps more surprising was the increased incidence of SSR allele loss (87 instances) compared with allele slippage (61 cases), which represents the mutation form most commonly observed in eukaryotes (Goldstein & Schlötterer, 1999; Chang et al., 2002; Vigouroux et al., 2002; Sibly et al., 2003). However, there are numerous possible causes of allele loss, including chromosome loss, chromosomal reorganization, and small indels or point substitutions that prevent primer binding. Changes to the SSR array can also cause apparent allele loss, but only on rare occasions when slippage gives rise to allele convergence such that a formerly heterozygous genotype becomes homozygous. Thus it may be the cumulative effect of all possible causes of allele loss that makes this category of mutation more frequent. However, the fact that we observed only six instances (2.6% of the tested samples) where two SSR linked markers were lost from the same regenerant implies that small-scale events such as indels and point mutations are most important. This inference accords with previous observations in other species (e.g. Gupta, 1998; Jin et al., 2008).

Among slippage mutations, cases of slippage leading to allele size change predominated (38 of 61 cases). In residual cases, an additional allele was observed. Extra alleles can arise from sample contamination, locus duplication followed by slippage mutation (Rolf et al., 2002) and slippage mutation followed by cell lineage chimerism (Rodríguez López et al., 2004). Contamination seems an unlikely explanation of the tri-allelic profiles, as (1) repeat DNA extractions invariably yielded identical abberant profiles; (2) all plants that exhibited a three-allele profile did so for one locus only and the identity of this locus varied between aberrant regenerants; and (3) the size of the additional allele sometimes differed between genotypes carrying three alleles for the same locus. The other possibilities arise from slippage mutation. Thus, the tri-allelic and simple profile changes can be combined and used to estimate slippage mutation rates among the regenerant populations. For plants regenerated after a single somatic embryogenic cycle, this gave an average mutation rate of 1.45 × 10−2. This is at the higher end of SSR mutation rates per generation seen in other species, which typically fall between 1 × 10−2 and 5 × 10−6 (Goldstein & Schlötterer, 1999; Vigouroux et al., 2002; Marum et al., 2009). However, it does not necessarily follow that other forms of mutation are similarly elevated for other component sections of the genome.

Time in culture

We found that SSR profile variation among regenerants follows a quadratic regression. Initially the relationship between time in culture and detected variation was found to be consistent with mutational events increasing with time in culture, supporting the widespread belief that genetic mutations accumulate with callus age (Tremblay et al., 1999; Bouman & De Klerk, 2001; Peredo et al., 2006). However, we found that older calli yielded fewer regenerants but that these plants also contained fewer SSR mutations, a finding seemingly supported by the BseAI AFLP profiles. One explanation is that the cell lineages accumulating heavy mutational loads lose totipotency. This leaves totipotent cells with few or no mutations and means that the callus becomes purged of totipotent mutant cell lines. This interpretation is supported by the skewed manner in which mutations appeared in regenerants; plants containing multiple mutations were more common than expected by chance alone. Furthermore, when mutant and true-to-type samples detected with the initial SSR markers were compared using six extra SSR markers, the former presented a statistically higher mutation rate than both the total population and the true-to-type subpopulation. This suggests a cascade effect, with initial mutation triggering the appearance of further mutations. The finding also echoes SSR instability patterns in human tumours, where increased mutation rates were found in cells already containing SSR mutations (Cotton, 1997; Li et al., 2002).

Primary vs secondary embryogenesis

Fewer genetic variants were detected amongst secondary embryos (23.3%) than primary embryos (37.6%). This contrasts with other in vitro multiplication systems, where variant rates apparently increase with the number of multiplication cycles (Brar & Jain, 1998; Cote et al., 2001), but supports previous reports where cocoa secondary somatic embryos were found to contain no mutations (Fang et al., 2009). A similar line of thinking used above to explain the decline in somaclonal variants with time can be applied here. Certainly, previous studies have revealed the appearance of somaclonal mutations in callus lines of Larix and Picea that do not appear in subsequent regenerants (DeVerno et al., 1994, 1999). Several authors have suggested that a correlation between DNA damage and apoptosis implies that plants can initiate programmed cell death upon sensing DNA damage (Whittle et al., 2001; Wang & Liu, 2006). Thus, cells carrying heavy mutational loads due to somaclonal variation may be triggered into initiating senescence and/or apoptosis, thereby negatively affecting their embryogenic capacity (Lam, 2004; Wang & Liu, 2006). It follows that cell lineages that are free (or relatively so) of genetic mutations may be expected to retain a greater embryogenic capacity.

Epigenetic variation

PCoA from MSAP profiles using the cytosine-methylation sensitive enzyme MroI from two tissues of the ortet plant (leaves and staminodes) and leaves of the regenerant plants revealed clear separation of the three tissue types. Greatest separation was seen between the two source plant tissues (leaves and staminodes). Significantly, the regenerant leaf profiles occupied intermediate positions between tissues from the source tree but were much closer to the leaves. It also appears that epigenetic profiles of leaves from the regenerants were more similar to those of leaves from the source plant than to those of the explant tissue from which they were derived. However, most regenerants retained some features of the staminode epigenetic profile, perhaps indicating incomplete de-differentiation.

Curiously, both genetic and epigenetic divergence between source and regenerant leaves sharply decreased after 10 wk in culture. It is tempting to speculate on a possible link between changes in DNA methylation profiles and the frequency of mutations. DNA methylation is believed to preserve genome integrity by controlling the appearance of aberrant recombinant events (Vigouroux et al., 2002), inactivating mobile elements such as transposons (Joyce et al., 2003) and detecting and correcting single base mismatches in prokaryotes (Cheng & Roberts, 2001), with several authors suggesting a homologous system in eukaryotes (Wu et al., 2003; Takeda et al., 2004; Kunkel & Erie, 2005). Several studies have highlighted associations between changes to global methylation profiles and increased mutation rates (Cheng & Roberts, 2001; Vigouroux et al., 2002; Joyce et al., 2003). However, the existence and nature of such a relationship require further study. Certainly, it should be remembered that the MSAP analysis used here only detected a subset of the epigenetic variation caused by C-methylation in the TCCGGA motif. We therefore cannot say whether our findings apply also to other sites of DNA methylation or indeed to other forms of epigenetic control (e.g. histone tail modification or siRNA-based gene silencing).

This study has shown that, for cocoa somatic embryogenesis, de novo mutations do not simply accumulate with callus age. Rather, we have found that, after an initial period of increased frequency of mutant regenerants, mutant cell lines progressively lose totipotency so that totipotent cell lineages free of mutations gradually predominate. As a result, late-forming regenerants (and, to a lesser extent, secondary embryos) tended to contain fewest genetic abnormalities. We have also found that late-forming and secondary embryos have an epigenetic profile that retains more features from the corresponding source tissue. We therefore hypothesize that there may be some form of link between stability of global methylation profile and repression of de novo mutation. Clearly, if our findings apply generally, they run contrary to the common practice of minimizing somaclonal variation by limiting time in callus culture (Potter & Jones, 1991; Sheidai & Hamta, 2008; Krizova et al., 2009). This could have a profound impact on in vitro protocols used for micropropagation, germplasm storage and transformation programmes. It follows that further work is required to examine other species and in vitro regeneration systems.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Sandro Leidi of the Statistical Services Centre at the University of Reading, Reading, UK for his help during the statistical analysis and Cocoa Research UK for funding this study. We also thank the anonymous referees for their constructive and helpful comments.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Abe T, Ii N, Togashi A, Sasahara T. 2002. Large deletions in chloroplast DNA of rice calli after long-term culture. Journal of Plant Physiology 159: 917923.
  • Bardini M, Labra M, Winfield M, Sala F. 2003. Antibiotic-induced DNA methylation changes in calluses of Arabidopsis thaliana. Plant Cell Tissue and Organ Culture 72: 157162.
  • Barret P, Brinkman M, Beckert M. 2006. A sequence related to rice Pong transposable element displays transcriptional activation by in vitro culture and reveals somaclonal variations in maize. Genome 49: 13991407.
  • Bossdorf O, Richards CL, Pigliucci M. 2008. Epigenetics for ecologists. Ecology Letters 11: 106115.
  • Bouman H, De Klerk GJ. 2001. Measurement of the extent of somaclonal variation in begonia plants regenerated under various conditions. Comparison of three assays. Theoretical and Applied Genetics 102: 111117.
  • Brar DS, Jain SM. 1998. Somaclonal variation: mechanisms and applications in crop improvement. In: Jain SM, Brar DS, Ahloowalia BS, eds. Somaclonal variation and induced mutations in crop improvement. Dordrecht, the Netherlands: Kluwer Academic Publisher, 1538.
  • Burg K, Helmersson A, Bozhkov P, von Arnold S. 2007. Developmental and genetic variation in nuclear microsatellite stability during somatic embryogenesis in pine. Journal of Experimental Botany 58: 687698.
  • Chang CL, Marra G, Chauhan DP, Ha HT, Chang DK, Ricciardiello L, Randolph A, Carethers JM, Boland CR. 2002. Oxidative stress inactivates the human mismatch repair system. American Journal of Physiology-Cell Physiology 283: C148C154.
  • Cheng X, Roberts RJ. 2001. AdoMet-dependent methylation, DNA methyltransferases and base flipping. Nucleic Acids Research 29: 37843795.
  • Cote FX, Teisson C, Perrier X. 2001. Somaclonal variation rate evolution in plant tissue culture: contribution to understanding through a statistical approach. In vitro Cellular and Developmental Biology-Plant 37: 539542.
  • Cotton RGH. 1997. Mutation Detection. Oxford, UK: Oxford University Press.
  • Cuatrecasas J. 1964. Cacao and its allies. A taxonomic review of the genus Theobroma. Contributions from the United States National Herbarium 35: 379614.
  • De Schepper S, Debergh P, Van Bockstaele E, De Loose M, Gerats A, Depicker A. 2003. Genetic and epigenetic aspects of somaclonal variation: flower colour bud sports in azalea, a case study. South African Journal of Botany 69: 117128.
  • De Schepper S, Leus L, Mertens M, Debergh P, Van Bockstaele E, De Loose M. 2001. Somatic polyploidy and its consequences for flower coloration and flower morphology in azalea. Plant Cell Reports 20: 583590.
  • DeVerno LL, Charest PJ, Bonen L. 1994. Mitochondrial DNAvariation in somatic embryogenic cultures of Larix. Theoretical and Applied Genetics 88: 727732.
  • DeVerno LL, Park YS, Bonga JM, Barrett JD. 1999. Somaclonal variation in cryopreserved embryogenic clones of white spruce (Picea glauca (Moench) Voss.). Plant Cell Reports 18: 948953.
  • Duchesne P, Bernatchez L. 2002. AFLPOP: a computer program for simulated and real population allocation based on AFLP data. Molecular Ecology Notes 3: 380383.
  • Fang J-Y, Wetten A, Adu-Gyamfi R, Wilkinson M, Rodriguez-Lopez C. 2009. Use of secondary somatic embryos promotes genetic fidelity in cryopreservation of cocoa (Theobroma cacao L.). Agricultural and Food Science 18: 152159.
  • Fras A, Maluszynska J. 2004. The correlation between the chromosome variation in callus and genotype of explants of Arabidopsis thaliana. Genetica 121: 145154.
  • Goldstein DB, Schlötterer C. 1999. Microsatellites: evolution and applications. Suffolk, UK: Oxford University Press.
  • Gower JC. 1966. Some distance properties of latent root and vectors methods used in multivariate analysis. Biometrika 53: 325338.
  • Gupta PK. 1998. Chromosomal basis of somaclonal variation in plants. In: Jain SM, Brar DS, Ahloowalia BS, eds. Somaclonal variation and induced mutations in crop improvement. Dordrecht, the Netherlands: Kluwer Academic Publisher, 149168.
  • Hao YJ, You CX, Deng XX. 2002. Analysis of ploidy and the patterns of amplified length polymorphism and methylation sensitive amplification polymorphism in strawberry plants recovered from cryopreservation. Cryoletters 23: 3746.
  • Henry RJ. 1998. Molecular and biochemical characterization of somaclonal variation. In: Jain SM, Brar DS, Ahloowalia BS, eds. Somaclonal variation and induced mutations in crop improvement. Dordrecht, the Netherlands: Kluwer Academic Publisher, 485499.
  • Ingelbrecht IL, Irvine JE, Mirkov TE. 1999. Posttranscriptional gene silencing in transgenic sugarcane. Dissection of homology-dependent virus resistance in a monocot that has a complex polyploid genome. Plant Physiology 119: 11871198.
  • Jaligot E, Beulé T, Baurens FC, Billote N, Rival A. 2004. Search for methylation-sensitive amplification polymorphisms associated with the “mantled” variant phenotype in oil palm (Elaeis guineensis Jacq.). Genome 47: 224228.
  • Jin SX, Mushke R, Zhu HG, Tu LL, Lin ZX, Zhang YX, Zhang XL. 2008. Detection of somaclonal variation of cotton (Gossypium hirsutum) using cytogenetics, flow cytometry and molecular markers. Plant Cell Reports 27: 13031316.
  • Joyce SM, Cassells AC. 2002. Variation in potato microplant morphology in vitro and DNA methylation. Plant Cell Tissue and Organ Culture 70: 125137.
  • Joyce SM, Cassells AC, Jain SM. 2003. Stress and aberrant phenotypes in in vitro culture. Plant Cell Tissue and Organ Culture 74: 103121.
  • Kaeppler SM, Phillips RL. 1993. Tissue culture induced DNA methylation in maize. Proceedings of the National Academy of Sciences, USA 90: 87738776.
  • Kato F, Suetake T, Murata A, Mukai T, Ueki H. 1988. Restriction endonuclease MroI useful for genetic engineering – obtained by culturing Micrococcus roseus S. Japanese Patent Office JP 63098383.
  • Kenedy AJ, Mooleedhar V. 1992. Conservation of cocoa in field genebanks – the international cocoa genebank, Trinidad. In: Proceedings of the international workshop on conservation, characterization and utilization of cocoa genetic resources in 21st century. 2126. 13–17 Sep. 1992, Port-of-Spain, Trinidad and Tobago.
  • Kharabian A, Darabi A. 2005. Characterization of some chromosomal aberrations in regenerated rice plants (Oryza sativa). Plant Cell Tissue and Organ Culture 83: 161168.
  • Krizova K, Fojtova M, Depicker A, Kovarik A. 2009. Cell culture-induced gradual and frequent epigenetic reprogramming of invertedly repeated tobacco transgene epialleles. Plant Physiology 149: 14931504.
  • Kunkel TA, Erie DA. 2005. DNA mismatch repair. Annual Review Biochemistry 74: 681710.
  • Lam E. 2004. Controlled cell death, plant survival and development. Nature Reviews Molecular Cell Biology 5: 305315.
  • Lanaud C. 1998. Origin of haploids and semigamy in Theobroma cacao L. Euphytica 38: 221228.
  • Lanaud C, Risterucci AM, Pieretti I, Falque M, Bouet A, Lagoda PJL. 1999. Isolation and characterization of microsatellites in Theobroma cacao L. Molecular Ecology 8: 21412152.
  • Li YC, Korol A, Fahima T, Beiles A, Nevo E. 2002. Microsatellites: genomic distribution, putative functions and mutational mechanisms: a review. Molecular Ecology 11: 24532465.
  • Li Z, Traore A, Maximova S, Guiltinan MJ. 1998. Somatic embryogenesis and plant regeneration from floral explants of cacao (Theobroma cacao L.) using thidiazuron. In Vitro Cellular and Developmental Biology-Plant 34: 293299.
  • Lin CS, Lin HL, Jane WN, Hsiao HW, Lin CC, Jheng FY, Chang WC. 2006. Changes in the morphology and cation content of a Bambusa edulis xylem mutant, vse, derived from somaclonal. Journal of the American Society for Horticultural Science 131: 445451.
  • Lopez-Baez O, Bollon H, Eskes A, Petiard V. 1993. Somatic embryogenesis and plant regeneration from flower parts of cocoa Theobroma cacao L. Comptes rendus de l’Académie des sciences 316: 579584.
  • Marum L, Rocheta M, Maroco J, Oliveira MM, Miguel C. 2009. Analysis of genetic stability at SSR loci during somatic embryogenesis in maritime pine (Pinus pinaster). Plant Cell Reports 28: 673682.
  • Messeguer R, Ganal MW, Steffens JC, Tanksley SD. 1991. Characterization of the level, target sites and inheritance off cytosine methylation in tomato nuclear DNA. Plant Molecular Biology 16: 753770.
  • Park DS, Park SK, Lee BC, Song SY, Jun NS, Manigbas NL, Cho JH, Nam MH, Jeon JS, Han CD et al. 2009. Molecular characterization and physico-chemical analysis of a new giant embryo mutant allele (ge(t)) in rice (Oryza sativa L.). Genes & Genomics 31: 277282.
  • Patwary MU, Mackay RM, Vander MJP. 1993. Revealing genetic markers in Geledium vagum (Rhodophyta) through the random amplified polymorphic DNA (RAPD) technique. Journal of Phycology 29: 216222.
  • Peredo EL, Revilla MA, Arroyo-García R. 2006. Assessment of genetic and epigenetic variation in hop plants regenerated from sequential subcultures of organogenic calli. Journal of Plant Physiology 163: 10711079.
  • Phillips RL, Kaeppler SM, Olhoft P. 1994. Genetic instability of plant tissue cultures: breakdown of normal controls. Proceedings of the National Academy of Sciences, USA 91: 52225226.
  • Potter R, Jones MGK. 1991. An assessment of genetic stability of potato in vitro by molecular and phenotypic analysis. Plant Science 76: 239248.
  • Raimondi JP, Masuelli RW, Camadro EL. 2001. Assessment of somaclonal variation in asparagus by RAPD fingerprinting and cytogenetic analyses. Scientia Horticulturae 90: 1929.
  • Reyna-López GE, Simpson J, Ruiz-Herrera J. 1997. Differences in DNA methylation patterns are detectable during the dimorphic transition of fungi by amplification of restriction polymorphisms. Molecular and General Genetics 253: 703710.
  • Rodríguez López CM, Wetten AC, Wilkinson MJ. 2004. Detection and quantification of in vitro-culture induced chimerism using simple sequence repeat (SSR) analysis in Theobroma cacao (L.). Theoretical and Applied Genetics 110: 157166.
  • Rolf B, Wiegand P, Brinkmann B. 2002. Somatic mutations at STR loci-a reason for three-allele pattern and mosaicism. Forensic Science International 126: 200202.
  • Schmitt F, Oakeley EJ, Jost JP. 1997. Antibiotics induce genome-wide hypermethylation in cultured Nicotiana Tabacum plants. Journal of Biological Chemistry 272: 15341540.
  • Sheidai M, Hamta A. 2008. Cytogenetic analysis of somaclonal variation in regenerated plants of Berseem clover (Trifolium alexandrium L.). Caryologia 61: 392396.
  • Sibly RM, Meade A, Boxall N, Wilkinson MJ, Corne DW, Whittaker JC. 2003. The structure of interrupted human AC microsatellites. Molecular Ecology and Evolution 20: 453459.
  • Takeda S, Tadele Z, Hofmann I, Probst AV, Angelis KJ, Kaya H, Araki T, Mengiste T, Scheid OM, Shibahara K et al. 2004. BRU1, a novel link between responses to DNA damage and epigenetic gene silencing in Arabidopsis. Genes & Development 18: 782793.
  • Thanos D, Scarpelis G, Papamatheakis J, Bouriotis V. 1989. Isolation and identification of restriction endonuclease BseAI. Nucleic Acids Research 17: 8881.
  • Traore A, Maximova S, Guiltinan MJ. 2003. Micropropagation of Theobroma cacao L. using somatic embryo-derived plants. In Vitro Cellular and Developmental Biology-Plant 39: 332337.
  • Tremblay L, Levasseur C, Tremblay FM. 1999. Frequency of somaclonal variation in plants of black spruce (Picea maritima, Pinaceae) and white spruce (P. glauca, Pinaceae) derived from somatic embryogenesis and identification of some factors involved in genetic instability. American Journal of Botany 86: 13731381.
  • Tsuro M, Inoue M, Kameoka H. 2001. Variation in essential oil components in regenerated lavender (Lavandula vera DC) plants. Scientia Horticulturae 88: 309317.
  • Vigouroux Y, Jaqueth JS, Matsuoka Y, Smith OS, Doebley J. 2002. Rate and pattern of mutation at microsatellite loci in maize. Molecular Biology and Evolution 19: 12511260.
  • Von Aderkas P, Pattanavibool R, Hristoforoglu K, Ma Y. 2003. Embryogenesis and genetic stability in long term megagametophyte-derived cultures of larch. Plant Cell Tissue and Organ Culture 75: 2734.
  • Wang C, Liu Z. 2006. Arabidopsis ribonucleotide reductases are critical for cell cycle progression, DNA damage repair, and plant development. Plant Cell 18: 350365.
  • Wassenegger M, Heimes S, Riedel L, Sanger HL. 1994. RNA-directed de novo methylation of genomic sequences in plants. Cell 76: 567576.
  • Whittle CA, Beardmore T, Johnston MO. 2001. Is G1 arrest in plant seeds induced by a p53-related pathway? Trends in Plant Sciences 6: 248251.
  • Wu P, Qiu C, Sohail A, Zhang X, Bhagwat AS, Cheng X. 2003. Mismatch Repair in methylated DNA structure and activity of the mismatch-specific thymine glycosylase domain of methyl-cpg-binding protein MBD4. The Journal of Biological Chemistry 278: 52855291.
  • Xiong LZ, Xu CG, Maroof MAS, Zhang QF. 1999. Patterns of cytosine methylation in an elite rice hybrid and its parental lines, detected by a methylation sensitive amplification polymorphism technique. Molecular General Genetics 261: 439446.
  • Zhao YP, Zhou Y, Grout BWW. 2006. Variation in leaf structures of micropropagated rhubarb (Rheum rhaponticum L.) PC49. Plant Cell Tissue and Organ Culture 85: 115121.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Fig. S1 Detected somatic genetic variation in linkage group 1.

Fig. S2 Visualization of the relative efficiency of enzymes MroI and BseAI in cutting methylated or unmethylated templates as revealed by real-time quantitative PCR (RT-qPCR).

Fig. S3 Effect of primary and secondary embryogenesis on the appearance of staminode I products in the LCT EEN 162/S-1010 regenerant population.

Table S1 Single sequence repeat (SSR) allele sizes for cocoa genotypes.

Table S2 Sequences (5′–3′) of primers used for microsatellite and methylation-sensitive amplified polymorphism (MSAP) analysis.

Table S3 Summary of the number of fragments displaying each class of epigenetic polymorphism between floral and leaf tissues in Theobroma cacao.

Methods S1 Experimental procedure to quantify differential sensitivity of MroI and BseAI to cytosine methylation.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

NPH_3242_sm_fS1.jpg75KSupporting info item
NPH_3242_sm_fS2.jpg62KSupporting info item
NPH_3242_sm_fS3.jpg30KSupporting info item
NPH_3242_sm_legends.doc23KSupporting info item
NPH_3242_sm_methods-S1.doc37KSupporting info item
NPH_3242_sm_tS1-S3.doc62KSupporting info item