Molecular characterization of a rice mutator-phenotype derived from an incompatible cross-pollination reveals transgenerational mobilization of multiple transposable elements and extensive epigenetic instability

Identification of a mutator-phenotype

We harvested more than 300 seeds from ca. 50 rice (Oryza sativa L. ssp. japonica, cv. Tong211) panicles that were artificially pollinated by fresh pollens taken from a single accession of a dicot plant, evening primrose (Oenothera biennis L.), followed by a second round pollination 48 hrs later with their own pollens collected from other individuals of Tong211 – a pollination method we termed "repeated pollination" [27]. We found that a substantial portion of the seeds showed abnormal development as a result of the manipulation, and which either did not germinate or died at early seedling stages. Among 84 rice plants (designated as S0 generation) growing to maturity, we identified a single individual plant (hereforth named Tong211-LP) that showed conspicuous phenotypic variation in multiple traits, including flowering time, seed-size, and particularly panicle-size (larger panicle, LP), relative to its rice parental cultivar. We then self-pollinated several panicles of this plant to produce seeds (also markedly enlarged), and the S1 progeny plants. Tong211-LP was partially sterile and produced a much smaller quantity of seeds than expected from a normal rice individual. Nonetheless, among the progenies produced, phenotypic variations that were observed in the individual mutator-phenotype (S0) were inherited at high frequencies to the S1 progenies; in addition, some new phenotypic variations that were apparently individual-specific appeared de novo in some of the S1 progeny plants. Typical phenotypic variations in the S1 progeny plants of the mutator-phenotype were shown in Figure 1. The transgenerational continuous occurrence of multiple phenotypic variations, i.e., sustained phenotypic instability, from a single individual is characteristic of a "mutator-phenotype", as originally reported in Drosophila [28].

Transpositional activation of multiple TEs in the mutator-phenotype and its selfed progenies

Genomic DNA was isolated from expanded young leaves of Tong211-LP (vegetatively propagated from root-stock cuttings), its eight selected S1 progeny individuals that exhibited the most extreme phenotypic variations, the parental rice line (cv. Tong211) and the pollen-donor, O. biennis. A series of DNA gel blot analysis was performed. First, to look for possible genetic introgression from the pollen donor O. bienni into rice, we did a genomic DNA probing assay [29], i.e., using labeled genomic DNA of O. biennis as a probe and autoclaved genomic DNA (150× excess in quantity) of rice as a blocker. We detected no evidence for genomic introgression (data not shown) based on this assay, as expected given the extreme phylogenetic remoteness of the two species. It should be pointed out however that this genomic blotting assay, though efficient [29], is by no means exclusive as only introgression of O. biennis species-specific genomic sequences would have been detected, and therefore cryptic introgression incidents may go undetected. Next, given the transgenerational mutability of multiple phenotypic traits, we asked the question if some normally quiescent transposable elements (TEs) might have been activated in the mutator-phenotype, as in the case of Drosophila [30, 31], and in bona fide wide hybrids of plants [32]. We thus checked stability of several previously characterized low-copy TEs endogenous to the rice genome, which were either documented to be active under certain stress conditions [33–36], or were suspected so based on bioinformatic predictions [37]. The studied TEs included a 430 bp MITE, mPing[34], its two transposase donors, Ping and Pong [34, 38], nine low-copy LTR retrotransposons, Osr2, Osr3, Osr7, Osr23, Osr35, Osr36, Osr42, Tos19 and Tos17 [33, 37]. We found that four (mPing, Osr7, Tos17 and Osr23) of the 12 studied TEs showed apparent transpositional activation in Tong211-LP, or more often, in its S1 progenies (Figures 2, 3 and 4). A notable feature of the gel blotting patterns of these four TEs was that some of the S1 progenies exhibited individual-specific patterns, i.e., the pattern of a given individual was not shared by other sib-individuals, thus suggesting stochastic, transgenerational inheritance of the destabilized state in these elements, which mirrors the observation of phenotypic novelty in these plants Among the eight studied S1 individuals, #5 showed the most transpositions for all four elements (Figures 2, 3 and 4). It remained to be determined if this "transpositional active state" would be continually inherited to future generations, or it would be converted to a stabilized, repressive state after a few generations.

Two (mPing and Tos17) of these four elements were shown previously as active under various stressful conditions, including tissue culture [34, 35] and irradiation [36], but the other two (Osr7 and Osr23) were only implicated as potentially active based on bioinformatic predictions but not empirically documented [37]. Thus, this is the first demonstration of transpositional activity of these two LTR retrotransposons in rice under this specific condition (Figures 4a and 4c).

Next, we studied the transposition of mPing in Tong211-LP and its S1 progenies in more detail, as this element is most active as well as amenable to characterization. First, the transpositional activity of mPing in these plants was further verified by transposon-display or TD [39], whereby > 60 putative mPing excision and de novo insertion events were isolated and sequenced. The sequence data indicated that at least 52 of the isolated events represent bona fide mPing activities (i.e., excisions or de novo insertions; see Additional files 1 and 2), as they each contained at their 5' terminus the expected portion of the mPing sequence encompassing the typical 15 bp terminal inverted repeats (TIRs) and the 3-bp target site duplication (TSD) of TAA or TTA, which is characteristic of the transpositional behavior of mPing [34–36]. Based on the sequence information, together with the complete genomic sequence of the standard laboratory rice genotype Nipponbare, locus-specific primers were designed for a set of 30 loci (see Additional files 1 and 2), which showed identical or high degree of sequence conservation between the studied rice cultivar (Tong211) and the genome-sequenced japonica rice cultivar Nipponbare. All these 30 loci were successfully amplified, cloned, sequenced and characterized (see Additional files 1 and 2). Sixteen of the 30 loci represent excisions in the S1 progenies of Tong211-LP, that is, compared with the parent Tong211 and the mutator-phenotype Tong211-LP (S0), they were excised in one or more of the S1 progenies (see Additional file 1). The predominant occurrence of mPing excisions in the S1 rather than the S0 generation (also evident in the gel blotting pattern, Figure 2) suggests that the timing of the excision events should be during late vegetative development and/or early gametogenesis in the Tong211-LP (S0) plant, and being manifested in the next generation via germline inheritance. Pairwise sequence comparison showed that none of 16 excisions had left behind any excision footprint (see Additional file 1), which is in agreement with some [38, 40] but not all (e.g., [26, 27]) previous reports on the excision properties of mPing. A BlastN analysis of these 16 excised mPing loci against the whole genome sequence of Nipponbare revealed an unexpected result in that eight of 16 loci were mapped to chromosome 3 and the rest to chromosomes 1, 2, 4, 11 and 12 (see Additional file 1), suggesting differential activity of the mPing copies with regard to their chromosomal locations.

Fourteen of the TD-identified loci were mPing de novo insertions in the S1 progenies of Tong211-LP, i.e., compared with the parent Tong211 and Tong211-LP (S0) they became larger-sized due to insertion of intact mPing copies in some of the S1 progenies (see Additional file 2). A BlastN analysis of these insertion-targeted sequences against the whole genome sequence of Nipponbare showed that all insertions mapped to unique- or low-copy regions (see Additional file 2), consistent with insertional propensity of mPing [41]. In contrast to the situation of excisions, these 14 mPing insertions did not show an obvious bias towards a particular chromosome (see Additional file 2).

Genome-wide genetic and epigenetic instability in the mutator-phenotype

To test whether the genetic variations in Tong211-LP and its progenies were confined to the activation of a few specific TEs, or a more general genomic instabilities have been elicited, we performed genome-wide analysis on the same set of plants used in the gel-blotting by using AFLP and MSAP markers, and assessed > 1,000 loci for each marker. In the AFLP analysis, both loss of the rice parental bands and gain of novel bands were detected in Tong211-LP and its S1 progenies relative to their rice parent Tong211, with variation frequencies of both types of genomic changes together ranged from 2.66% to 8.16% (Figure 5). In MSAP analysis, methylation alteration of both CG and CNG at the CCGG sites (a prominent site for methylation modification in eukaryotes) were also detected in Tong211-LP and its S1 progenies relative to their rice parent Tong211, with frequencies of both types of DNA methylation alterations together ranged from 21.44% to 27.30% (Figure 5), which are higher by more than three times than those of genetic changes. These data indicated that, apart from transpositional activation of a subset of TEs (Figures 2, 3 and 4), pollination by pollens of O. biennis had particularly induced genome-wide epigenetic instabilities in the form of altered cytosine methylation patterns in the mutator-phenotype and its S1 progenies, though lower frequencies of genetic changes also occurred. It is notable that, as in the case for the transposition of the four TEs, described above, many loss or gain of bands in the AFLP or MSAP profiles were singletons (a specific change occurred in only one individual), thus probably could not be attributed to genetic segregation from existing variations in the S0 generation alone. Instead, stochastic transgenerational inheritance of the induced metastable epigenetic chromatin state might be a major underlying cause for the new variations in the S1 progeny, which is apparently consistent with the de novo appearance of phenotypic variations in the S1 plants (e.g., Figure 1).

To gain some insights into the chromosomal distribution and possible functional relevance of the genomic loci that showed genetic and epigenetic instabilities in Tong211-LP and its progenies, a set of variable AFLP and MSAP bands were isolated, cloned and sequenced (see Additional file 3). It was found that all variable bands were chromosomal DNA sequences of the rice genome (thus again pointing to the lack of genetic introgression from the pollen donor O. biennis), and they mapped to all 12 rice chromosomes, with the numbers ranged from two to six for each chromosome. A notable observation from the inferred possible functionalities of these variable bands is that the majority of them (23 out of 41) appeared to be genic sequences, followed by TEs (11), and only seven showed no homology (see Additional file 3). This may explain the dramatic phenotypic variations in the mutator-phenotype and its selfed S1 progenies (e.g., Figure 1).

Correlations between two of the three kinds of instabilities, genetic variation, epigenetic variation, and TE activity

Heritable alteration in expression state of genes encoding for putative DNA methyltransferase, 5-methylcytosine DNA glycosylase and siRNA pathway-related protein in the mutator-phenotype and its progenies

Given that two known mechanisms, cytosine methylation and small interfering (si) RNA, often play critical roles in repressive control of TE activity [43–47], and the above documented correlation between alteration in cytosine methylation and TE (mPing) activity in the rice mutator-phenotype, it is reasonable to assume that the TE activity and epigenetic instability are probably related to perturbation of the homeostatic expression state of genes encoding for the enzymatic machinery responsible for maintaining the cytosine methylation states and/or other aspects of the chromatin epigenetic structure. We thus measured the mRNA steady-state abundance for a set of genes encoding for putative DNA methyltransferases, 5-methylcytosine DNA glycosylases, and the siRNA pathway-related proteins by real-time reverse transcriptase- PCR (q-RT-PCR) analysis with gene-specific primers. We found that of the 13 selected genes analyzed, 11 (the two exceptions are DDM1 and AGO4-2) showed significantly perturbed expressions in the mutator-phenotype (Tong211-LP) relative to its rice parental line (Tong211) (Figure 6). Furthermore, the perturbed expression states of these genes were transgenerationally heritable in the sense that none of the genes had reverted to the original expression states of their parental rice line (wild type) in all or most of the eight S1 plants analyzed (Figure 6). In fact, the extent of perturbation in the expression states of most of the genes was further augmented in some of the S1 progenies, and in one extreme case (i.e., the DDM1 gene), whose expression did not exhibit significant deviation from the parental line in Tong211-LP, nonetheless showed significant difference in six (except for S1-1 and -3) of the eight S1 individuals (Figure 6). This result of transgenerational perturbation in the expression states of the chromatin structure-maintenance genes is consistent with the heritable epigenetic instability in these plants, described above.

Recent studies have established that the intrinsic DNA methylation patterns in both plants and animals are faithfully maintained and perpetuated by coordinated function of at least two classes of DNA methyltransferases (maintenance and de novo), together with active demethylases, i.e., the 5-methylcytosine DNA glycosylases [48, 49]. On the other hand, small interference RNA (siRNA) was documented as playing pivotal roles in repressing activity of TEs in diverse organisms by specific targeting [44–47]. Furthermore, at least in plants de novo methylation is often related to the activity of certain species of siRNAs by a mechanism known as RNA-directed DNA methylation or RdDM [50]. It is therefore conceivable that the coordinated expression of these genes represent a default requirement for stable maintenance and perpetuation of intrinsic DNA methylation patterns and silent TE states. Thus, the transgenerational perturbation of these genes in the mutator-phenotype and its progenies (Figure 6) may conceivably disrupt the homeostatic expression state of these genes as a network in the rice cells. It is likely that at least one facet of the possible effect of alien pollination as a "biological stress" may lie in its perturbation of coordinated expression of these chromatin state maintenance and regulation genes in the maternal recipient somatic and germinal cells, and hence, result in transgenerational epigenetic instability and TE activation.

The mutator-phenotype was not caused by parental heterozygosity or contamination, but resultant from the alien pollination-imposed stress

There are three alternative possible causes that may be responsible for or contribute to the generation of the mutator-phenotype: one is segregation of pre-existing parental heterozygosity, the second is seed or pollen contamination from other rice cultivar(s), and the third is mutagenic effect from some unknown source. That we consider pollination by O. biennis as the only major underlying cause for the genetic and epigenetic instabilities in the mutator-phenotype (Tong211-LP) and its S1 progenies, are based on the following lines of evidence: First, the rice parental cultivar cv. Tong211 is a genetically pure line, as rice is a predominantly self-pollinating plant, and furthermore, the specific strain used for the present work had been maintained by strict selfing for > 10 successive generations in our hands, thus its inbred nature was ensured. In fact, the inbred nature of Tong211 has also been validated in this study by a parallel analysis on 30 random individuals, as in no case a variable pattern suggestive of heterozygosity was observed in either the gel-blotting patterns of the four active TEs or in PCR-based locus-specific mPing amplifications of all 30 loci (see Additional files 1 and 2; data not shown). Therefore, parental heterozygosity can be confidently ruled out as a causal factor for the markedly changing patterns of either the studied TEs, and by extension, the variable MSAP/AFLP profiles. Second, based on the following lines of evidence, contamination by pollens or seeds of other rice cultivars was considered as extremely unlikely. (1) Strict precautions were taken both in the cross manipulations (emasculation and pollination) and in later propagations by timely bagging of all panicles to endure 100% selfing. (2) It is notable that whereas most genetic and epigenetic changes that occurred in the mutator-phenotype were largely inheritable to its S1 progenies, many individual-specific new patterns appeared de novo in the S1 individuals (e.g., Figures 2, 3 and 4). Therefore, if contamination were a cause for the observed variable patterns, then the S1 plants need to have derived from S0 seeds that were contaminated independently by pollens of different rice cultivars, which obviously is extremely unlikely. (3) In the course of identifying the mPing-containing loci in cultivar Tong211, we uncovered 21 additional loci each contains a mPing copy in the standard cultivar Nipponbare but devoid the element in Tong211 (see Methods). We then amplified these 21 mPing-empty loci from the mutator-phenotype and its eight S1 progeny individuals, and we found that in all cases, only smaller-sized PCR products consistent with lacking of mPing in these plants were amplified (data not shown). Given the extremely high degree of presence vs. absence polymorphism of mPing among japonica rice cultivars [[34, 51]; our unpublished data], this result strongly suggests that the mutator-phenotype and its analyzed progenies were unequivocally originated from one cultivar (Tong211) only. Taken together, the possibility of pollen or seed contamination can be confidently ruled out. Finally, all plant lines used in this study were grown together under identical normal conditions, under which biotic (e.g., pathogen infestation) and abiotic stresses were not exerted. Therefore, it is also inconceivable that the mutator-phenotype and its progenies had been differentially stressed from their parental line by an unknown stress that elicited the genomic instabilities.

By ruling out each of the three alternative possibilities, we are confident to conclude that pollination by O. biennis is the major, if not the only, conceivable cause for the genetic and epigenetic variations in the mutator-phenotype and its sefled progenies. Nonetheless, the basis for the occurrence of such extensive genetic and epigenetic instabilities as a result of "abortive" alien pollination remains mysterious. The single mutator-phenotype (Tong211-LP) was identified out of 84 "pollinated" plants, based on its striking phenotypic variations, thus giving a mutation frequency of 1.2%. Although it is likely that genomic variations may also have occurred in some other treated plants, they did not reach the extent to cause apparent phenotypic variations. Because all these 84 plants were sequentially pollinated first by O. biennis and then by pollens from the same rice line (Tong211), it appeared likely that stochasticity have also played an important part in the genesis of the mutator-phenotype individual.

The phenomenon we reported here is reminiscent of what McClintock envisioned two decades ago that wide hybridization in plants might activate quiescent TEs and cause genomic restructuring [18]. Indeed, several lines of empirical evidence in both plants and animals have lend support to this prediction [8, 13, 30–32, 40, 52–55]. Although all these previous works involved documented genome merger and/or introgression, it can be envisioned that even in the absence of introgression a "shock" at multiple levels may be incurred if the pollination by alien species per se represents a kind of "biological stress". In principle, even between incompatible crosses, certain metabolites, particularly those that require only trace quantity to produce dramatic effects like signal molecules, phytohormones and siRNA species etc., may be released from the donor pollens and enter the recipient stigma cells, thus may conceivably produce various mismatches and elicit a stress response.

It can be imagined that the cellular machinery responsible for safeguarding the genetic and epigenetic stabilities is likely sensitive as well as responsive to perturbations by stress, and fine-tuning on a balance between genetic/epigenetic fidelity and instability is required for the sake of survival and adaptation. Thus, in this study the significantly perturbed expression of nearly all of the 13 studied genes involved in the cellular machinery responsible for maintenance of chromatin epigenetic state subsequent to alien pollination might represent a sensory and adaptive response by the plant genome. From this perspective, the findings of this study may have bearing to genome evolution, as similar incidents of alien pollination may occur frequently under natural conditions, and hence, implicate a novel role of hybridization in evolution. Thus, we propose the possibility that "accidental cross-pollination" by a certain unrelated species may be actually mutagenic and elicited dramatic genetic/epigenetic instabilities, which may be perceived by selection. Further judiciously designed experiments involving an array of cross manipulations between different plant species are needed to investigate generality of this phenomenon and its underlying mechanism.

Plant materials

Plants used in this study included a single rice individual named "Tong211-LP" that was identified from a set of plants derived from seeds of a "alien pollination experiment" between rice (Oryza sativa L.), ssp. japonica, cv. Tong211 and a dicot plant, evening primrose (Oenothera biennis L.), and followed by self-pollination with pollens of different individuals of the same rice cultivar, namely, using a procedure we termed "repeated pollination" [27, 40]. The choice for the particular crossing partners was based on two major considerations: (1) the two species (O. sativa L. and O. biennis L.) are hardly related, and hence, served well for the purpose of this study; (2) O. biennis L. produces a large amount of pollens that are viable for relatively long period after collection (our unpublished observation), and hence, convenient for the crossing manipulations. The identified individual plant (Tong211-LP) exhibited conspicuous phenotypic variation in multiple traits particularly enlarged panicles and seed-size, compared with its maternal parental cultivar Tong211. Seeds of individual panicles collected from Tong211-LP were selfed to produce the S1 progenies. In all cases, mutant plants were grown together with the parental line under identical, normal conditions, and strict bagging was practiced.

DNA gel blot analysis

Genomic DNA was isolated from expanded young leaves of individual plants by a modified CTAB method [56] and purified by phenol extractions. Genomic DNA (~3 μg per lane) was digested by XbaI (New England Biolabs Inc.), and run through 1% agarose gels. The choice of XbaI is because the studied TEs either do not have a restriction site or the site(s) being on one side of the probe region, such that copy number of the TEs can be estimated based on the blotting patterns. Fractionated DNA was transferred onto Hybond N+ nylon membranes (Amersham Pharmacia Biotech) by the alkaline transfer recommended by the supplier. For investigating stability of a set of 13 low-copy transposable elements (TEs), element-specific primers were designed (see Additional file 4), and the fragments were obtained by PCR amplifications by using genomic DNA of the parental line (Tong211) as the template. The fragments representing each of the TEs were then gel-purified, identities confirmed by sequencing, and labeled with fluorescein-11-dUTP by the Gene Images random prime-labeling module (Amersham Pharmacia Biotech). Hybridization signal was detected by the Gene Images CDP-Star detection module (Amersham Pharmacia Biotech) after washing at a stringency of 0.2 × SSC, 0.1% SDS for 2 × 50 min. The filters were exposed to X-ray film for 1–3 hrs depending on signal intensity.

Transposon-display (TD) and PCR-based locus assay on mPingexcision and insertion

The transposon-display (TD) technique [39], using nested mPing-specific primers together with a primer designed according to the restriction enzyme MseI-adapter sequence was as described [40]. To further verify mPing excisions and insertions, a subset of identified TD loci were sequenced, and by taking advantage of the complete genome sequence of the standard laboratory japonica rice cultivar Nipponbare http://rgp.dna.affrc.go.jp, a set of locus-specific primers (see Additional file 1) each bracketing an intact mPing in the parental rice cultivar Tong211 (for detecting excision) or in the mutant Tong211-LP and its S1 progeny individual(s) (for detecting insertions), was designed by the Primer 3 software http://biocore.unl.edu/cgi-bin/primer3/primer3_www.cgi. Likewisely, a set of 21 loci each of which does not encompass a mPing copy in the parental cultivar Tong-211 was also identified in the course of TD analysis. Primers specific to this set of loci were also designed, and used to validate single genotypic origin of the mutant (Tong-211-LP) and its progenies. PCR amplification with these primer pairs were then conducted on the corresponding plant materials. The amplicons were visualized by ethidium bromide staining after electrophoresis through 2% agarose gels. All identified sites for mPing excisions (along with the corresponding element-containing donor sites) and de novo insertions were isolated and sequenced, such that the excision prosperities (e.g., to leave footprint or not) and characteristics (e.g., chromosomal location and potential functionality) of insertion-targeted sequences could be determined or inferred.

AFLP and MSAP analysis

The protocols suitable for amplified fragment length polymorphism (AFLP) and methylation-sensitive amplified fragment (MSAP) in rice were exactly as reported [22, 57]. For each marker, > 1,000 loci were scored. Typical bands representing genetic changes (AFLP) and cytosine methylation alterations (MSAP) in the mutant or its S1 progenies, as compared with their parental cultivar Tong211, were isolated, cloned and sequenced. Homology analysis was performed by BlastX at the NCBI website http://www.ncbi.nlm.nih.gov/blast/Blast.cgi.

Real-time Reverse transcriptase (RT)-PCR analysis

Isolation of total RNA and cDNA synthesis was essentially as reported [24]. Specifically, total RNA was isolated from expanded young leaves at the same developmental stage as that used for DNA isolation by the Trizol Reagent (Invitrogen), following the manufacturer's protocol. The RNA was treated with DNaseI (Invitrogen), reverse-transcribed by the SuperScriptTM RNase H-Reverse Transcriptase (Invitrogen), and subjected to q-RT-PCR analysis using gene-specific primers. The q-PCR experiments were performed using a Roche LightCycler480 apparatus (Roche Inc.) according to the manufacturer's instruction and SYBR Premix Ex Taq (Takara) as a DNA-specific fluorescent dye. The primers for all 13 studied genes encoding for putative DNA methyltransferases (seven), 5-methylcytosine DNA glycosylases (two) and siRNA-related proteins (four) were designed by the Primer 5 software (see Additional file 5). Expression of a rice β-actin gene (Genbank accession X79378) was used as internal control with the primer pairs 5'-ATGCCATTCTCCGTCTT-3' and 5'-GCTCCTGCTCGTAGTC-3'. Thermal cycling conditions consisted of an initial denaturation step at 95°C for 30 s, followed by 45 cycles of 15 s at 95°C and 1 min at 60°C. Three batches of independently isolated RNAs were used as technical replications. The melting curve analysis with the LightCycler480 together with 1.5% agarose gel electrophoresis of the products were used to ensure that right size product without significant background was amplified in the reaction. The relative amounts of the gene transcripts were determined using the Ct (threshold cycle) method, as described by the manufacturer's protocol. Data were analyzed by using the software provided by Roche Company and calculated by the 2-ΔΔCt method. Quantitative results were given as mean expression (means ± SD).

Statistics

Statistical significance was determined using SPSS 11.5 for Windows http://www.spss.com/statistics/ and analyzed by Independent-Samples student's t-Test. Specifically, correlations were tested for the data of variation frequencies calculated based on the three markers, Transposon (mPing)-display, MSAP and AFLP, by using the Pearson correlation analysis. The same program was used to test for the statistical significance of differences in the relative expression of the set of 13 genes in the mutator-phenotype (Tong211-LP) and its sefled progenies relative to that of the rice parental line (Tong211).