Plant-a-Seed

Welcome | Summary | History | Why self-publish? | Disclaimer | Citing | Research Paper | Submit Comments

To my colleagues,

I know that some of my peers will be critical of my decision to launch this site. Perhaps some will think this action demonstrates my complete disregard for the scientific process and the peer review system. Let me state clearly and emphatically that this is not the case. Rather, it is out of respect for and belief in the power of science and the scientific process that I take this unconventional step.

I do not believe that the opinions of a few should hold unilateral sway such that they effectively prevent the larger community of scientists from scrutinizing our research. I argue instead to let any scientists thus inclined consider our results and judge for themselves. If alternative explanations for our findings exist, even if they fall somewhere between the trivial or the extraordinary, let us explore them.

It is the very possibility that these data reveal the existence of something truly astonishing and novel that motivates me to deviate from accepted protocol despite the dissention we have already encountered. I am inviting my scientific colleagues to draw their own conclusions by openly planting a seed that hopefully promotes scientific dialogue. For the more daring, perhaps this will also inspire them to do the experiments that will either confirm or refute our results.

The experimental results detailed in the accompanying paper suggest that the model plant Arabidopsis thaliana keeps a back-up copy of genetic information that can be used to overwrite different parts of its genome. Although this back-up copy is normally not active or detectable, our experiments suggest that under certain conditions this hidden genome will show itself and leave a telltale footprint. The first hint that this genomic overwriting system existed came from studies we did on a mutant plant that had abnormal flowers. Oddly, when this mutant plant set seed, about 10% of those seeds grew into plants that had normal flowers. According to Mendel's law of stable inheritance, this should not happen. In trying to understand how this might work, evidence pointed to the astonishing possibility that the mutant plants were able to fix their flawed gene by overwriting the defective copy with a working copy inherited from the grandparent.

On March 24th 2005, the first experimental evidence hinting at the existence of a secret cache of genetic information was published (Lolle et al., Nature 434, 505). This finding captured the imagination of the public and starting with front-page stories in The New York Times and The Washington Post spurred an international media flurry. Naturally, these findings were highly controversial and sparked considerable debate among scientists. Subsequently, two different labs found that they were unable to reproduce these results (Peng et al., Nature 443, E8; Mercier et al., Genetics 180: 2295) and claimed that there was a much simpler explanation, namely, experimental error. Surprisingly, even after these two reports were published, interest among scientists has remained high. In 2008, The Scientist wrote a featured cover story, "Mendel Upended" (The Scientist, 22 (2): 31) and in December 2010, the 2005 paper was ranked the #1 paper in the all time top ten by the Faculty of 1000.

The research paper posted on this site is the first study from my laboratory that follows up on the research published in 2005. We believe that the data presented not only support our original findings but also offer reasons why this genetic phenomenon may not have been found previously. We invite you to make up your own mind.

This manuscript has been considered for publication by four different scientific journals. Two of the journals did not send the manuscript out for review but instead made editorial decisions to decline publication. The two other journals, however, did have the manuscript peer reviewed. In total six scientists with the appropriate expertise rendered their opinions on the technical quality of the research and our interpretation of the results.

In the opinion of several editors and these six reviewers the primary reason that our paper was unacceptable for publication was that all of our results could be explained by experimental error. In other words, reviewers felt that contamination was the most likely explanation for our data. Specifically, they felt it was more likely that our seed stocks were not pure or that the tissue samples used for our DNA fingerprinting experiments were contaminated with other plant material.

Because we make "extraordinary claims" about a phenomenon that has never been described before we have been told that we must meet a very high burden of proof. Others have stated it even more strongly saying that our findings have no validity without independent replication by other research laboratories. In my opinion, this creates an untenable situation that is impossible to resolve. In light of these comments I believe that we are unlikely to satisfy the impossibly high standards that our colleagues have set even if we were to include additional data showing more evidence of this phenomenon. If we are not given an opportunity to share our findings, how can the scientific community either confirm or refute these experimental results?

As a scientist I consider it my duty to make these findings public and by extension, to invite my scientific colleagues to fully engage in the scientific process. The scientific community has the right and arguably, the duty, to arrive at their own conclusions. Given the potential benefits of this discovery, I feel that it is essential that our research findings enter the global forum.

The objective of this site is to provide information about our research and our experimental findings, as described here. The views and opinions expressed herein are solely those of "Plant-a-Seed" and the senior author, S.J. Lolle and not those of any funding agencies, academic institutions or affiliated organizations. All rights reserved (© S.J. Lolle).

Hopkins MT, Author AMK, Chang P-C, Author KCV, Lai D, Doerr MD & Lolle SJ. 2011. De Novo Genetic Variation Revealed in Somatic Sectors of Single Arabidopsis Plants. [Internet]. In Plant-a-Seed. Available at http://www.plant-a-seed.com/#research-paper

De Novo Genetic Variation Revealed in Somatic Sectors of Single Arabidopsis Plants

M.T. Hopkins, Author A.M.K., P-C. Chang, Author K.C.V., D. Lai, M.D. Doerr and S.J. Lolle*.

Abstract

Concern over the tremendous loss of genetic diversity among many of our most important crops has prompted major efforts to preserve seed stocks derived from cultivated species and their wild relatives. Like many crop plants, Arabidopsis thaliana propagates mainly by self-fertilizing and theoretically has a limited potential for producing genetically diverse offspring. Despite this, inbreeding has persisted in Arabidopsis for over a million years suggesting that some underlying mechanism buffers the deleterious consequences of this reproductive strategy. Using presence-absence molecular markers we demonstrate that single Arabidopsis plants are not genetically uniform but produce somatic sectors with distinct molecular profiles. Sequence analyses revealed changes in single nucleotides, loss of insertions and, surprisingly, acquisition of insertions that share identity with sequences that were absent in the experimental plants themselves and their immediate parent. Estimates based on quantitative analyses showed that sectors tended to be very small and that their genetic makeup was complex but also showed that sectoring events were more extensive than revealed by standard methods. After ruling out more trivial explanations for these data, we suggest that Arabidopsis plants harbor an intrinsic cache of templates, possibly of ancestral origin, that can drive genome sequence changes. Given the evolutionary advantage afforded to populations with greater genetic diversity, we hypothesize that organisms that primarily self-fertilize or propagate clonally counteract the genetic cost of such reproductive strategies by leveraging a sequestered reserve of cryptic genomic information. We propose the use of the term 'restoration' to describe this novel form of genome over-writing.

Running title: Arabidopsis plants are genetic mosaics

1. Department of Biology, University of Waterloo, 200 University Ave. West, Waterloo, Ontario, N2L3G1 Canada.

*To whom correspondence should be addressed. Email: plantaseedfeedback@gmail.com

Introduction

One of the biological features that distinguish plants from animals is their ability to spontaneously form genetically distinct but stable sectors. Individuals that are genetic mosaics remain viable in large part because plant growth and development is modular in nature, allowing for individuals to be composed of genetically distinct modules without compromising their viability. Nonfunctional modules can simply be discarded. In addition, the majority of plant cells remain developmentally labile through much of their life allowing differentiated cells to reprogram and proliferate should tissue replacement be required. As a consequence of this profound developmental versatility, individuals composed of cell populations derived from different plant species are not only viable but can coordinate the growth and development of chimeric organs that can be stably maintained by grafting [1].

For organisms such as plants that are constrained to sessile life styles, potentially spanning decades, the formation of sectors that vary in genotype may provide a robust strategy allowing individuals to adapt to extrinsic factors encountered during their life span or to challenges that persist over multiple generations. In an elegant paper published in 1981, Whitham and Slobodchikoff [2] propose that the formation of genetic mosaics offers plants a unique adaptive advantage by allowing introduction of genetic variants into the gene pool either through vegetative propagation or through sexual reproduction. They further propose that mutations arising somatically have a greater probability of being incorporated into the gene pool than mutations that arise in the gametes [2] because germ line cells are derived from somatic tissues that arise late in the developmental history of the plant [3,4].

The relatively frequent occurrence of mosaics among various plant species has been extensively utilized in the development of novel ornamentals and for the selection and maintenance of desirable traits in many cultivated crops. Any desirable cultivars that have arisen in this manner have been maintained through vegetative propagation and, to date, are responsible for a significant fraction of agriculturally important perennial plants. On the other hand, desirable traits in many important annual crops, such as rice, soybean, maize and wheat, have been introduced through classical genetic manipulations using directed breeding strategies. Once generated, annuals with good agronomic performance are generally maintained by inbreeding.

In recent years, concern has grown over the presumed loss of genetic diversity resulting from the application of modern horticultural and breeding practices. Therefore, the benefit of excellent performance may come with a significant risk [5,6]. However, recent and surprising results suggest that even highly inbred species can harbor unanticipated sources of intrinsic genetic variation. For example, highly inbred soybean cultivars have been shown to manifest significant phenotypic variation in the absence of sexual manipulation [7,8]. Such high intrinsic genetic variation has also been demonstrated for a number of other crop plants [9].

In the natural world, inbreeding occurs in numerous highly successful flowering plant species including wild relatives of Arabidopsis thaliana [10]. In nature, therefore, highly inbred species have persisted despite the predicted reduction in genetic diversity. Why would such inbreeding strategies be successful and what are the implications from an adaptive perspective? One possibility put forward by Barrett [11] is that such populations are very successful in their particular niches and benefit from producing large numbers of genetically identical offspring. Nevertheless, selection should favor plant species that can co-evolve on time scales reflecting particular environmental challenges such as fluctuations and variations in pathogen populations. In keeping with this view, it has been shown that sequence variation in 20 diverse strains of Arabidopsis is highly non-random. In gene families mediating biotic interactions, such as those implicated in pathogen defense, variation far exceeds that seen in families involved in basic biological processes [12].

The underlying mechanisms driving phenotypic variation in highly inbred lines, whether domesticated or wild, have often been inferred and have had limited experimental verification. Nevertheless, relatively simple molecular approaches have provided insight into some of the genomic events coinciding with visible changes in phenotype. In flax, for example, molecular assays have demonstrated that heritable phenotypic changes induced by environmental shifts are accompanied by reproducible changes in genomic DNA including changes in total DNA content, non-random changes in DNA sequences or sequence rearrangements [13-16]. In soybean, reproducible non-random DNA sequence changes induced by in vitro culturing of root explants have also been demonstrated using restriction fragment length polymorphic markers [17]. Genomic changes manifesting similar hallmarks of biased sequence alterations have also been described for banana [18] and in rice hybrids [19].

In the work described by Roth et al. [17], soybean root explants were shown to repeatedly give rise to particular alleles that were absent in the donor plants but had previously been found and characterized in other varieties of cultivated soybean. To account for the appearance of these particular allelic variants, the authors proposed that these organisms had evolved "internal generators of genetic variation" that mediated genome changes through some type of recombination process. In 2005, Lolle and colleagues [20] described a genome-wide phenomenon in Arabidopsis hothead (hth) mutants that was very reminiscent of that described by Roth et al. [17]. Based on the nature and genome-wide locations of the sequence changes detected, it was proposed that a template-directed process mediated these changes and that these cryptic but stable extra-genomic templates themselves had persisted since at least the grandparental generation. Not surprisingly, this proposal met with considerable skepticism and numerous alternative explanations for these data have since been published [21-26].

In this study, we have employed presence-absence molecular markers to test for non-Mendelian inheritance and found that Arabidopsis plants can inherit sequence tracts that were absent in their immediate parents. Furthermore, we show that tissue samples taken from multiple parts of an individual plant can have distinct DNA-based marker profiles. This includes profiles where novel insertions can be detected that are exclusive to only a small subset of samples. These experiments demonstrate that individual plants produce somatic sectors and are genetic mosaics. Strikingly, sequence analyses show that the newly acquired sequence tracts share identity with sequences found in the grandparent. Since genetic variation can occur in the same plant during vegetative development and in the absence of sexual reproduction, these novel sequences must originate from cryptic reserves within the host plant itself. The data presented support the original contention that a previously unknown template-directed mechanism exists that drives the de novo acquisition of insertions [20]. Our findings raise the encouraging possibility that other inbreeding species, including crop plants, may also harbor a cryptic reserve of genetic variation.

Results

Molecular markers deviate in the absence of out-crossing

Homozygous hth mutant Arabidopsis plants were previously shown to give rise to wild type (wt) progeny at relatively high frequencies [20,27]. Although an intrinsic mechanism was proposed [20], cross-pollination with neighboring plants was subsequently put forward as the more likely explanation for the appearance of these wt revertant offspring [24,25]. To test the susceptibility of hth plants to out-crossing under our growth conditions, experiments were conducted using a pollen donor harboring a dominant gene conferring resistance to the herbicide glufosinate. Herbicide-resistant transgenic lines were grown together with hth and eceriferum-10(cer-10) [28] floral fusion mutants and wt Landsberg plants. These analyses showed that mutants with floral fusion phenotypes experienced consistently higher levels of out-crossing than wild type plants (0.02-0.43% for hth-4, 8 and 10 mutants, 0.89% for cer10 mutants, 0.01% for wt plants). Out-crossing rates varied considerably between individual hth mutant plants, but results also confirmed that the majority of plants did not cross-pollinate. However, factors such as donor-recipient proximity, the severity of the floral fusion phenotype, growth chamber airflow patterns and plant handling influenced the propensity to cross-pollinate. Nevertheless, growing hth mutant F2 plants in isolation did not eliminate wt progeny from F3 progeny pools and, on average, 1.53% of F3 progeny were phenotypically wt for HOTHEAD despite being derived from self-fertilized F2 hth mutant parent plants (2/133 hth-4 , 2/131 hth-8 and 2/127 hth-10 gave rise to wt F3 progeny). Under our laboratory conditions, cross-pollination could be reduced to 0.1% but could not be eliminated, even if hth mutant plants were reared in transparent plastic cones. For all experiments described here, seeds were collected exclusively from hth mutant plants grown in cones.

While conducting segregation analyses and scoring offspring for herbicide resistance, a single hth mutant plant with a large phenotypically wt floral sector was identified (Figure 1). Sampling of shoot tissues confirmed that phenotype corresponded to genotype and that both mutant hth-4 and wt HTH alleles could be detected in tissue derived from this large wt sector (Figure 1B).

The identification of this sectored individual provided the first unequivocal phenotypic evidence that hth plants were capable of producing somatic sectors. This finding suggested that perhaps some of the wt revertants originally found among hth mutant progeny might have arisen from genetically heterozygous sectors on the parent plant [20]. Since well over 300,000 mutant plants were screened in the course of our out-crossing experiments and only one plant found such as that shown in Figure 1B, we reasoned that if sectoring does occur, the vast majority of sectors would be very small, possibly only alter the genotype of a few cells and therefore would have no obvious phenotype. This possibility prompted us to take a more comprehensive molecular approach and test whether novel genotypes could be detected in tissue samples obtained from single hth plants.

Insertion-deletion markers are unstable in mutant and wild type hybrids

We chose to focus exclusively on molecular markers consisting of genomic DNA sequence tracts between 45-94 nucleotides in length that are either present or absent in the Columbia and Landsberg Arabidopsis accessions (insertion-deletion polymorphic markers or indels; Figure S1, Table S1). Hybrid lines were constructed between Columbia and Landsberg accessions and descendants used as experimental material. For all of the indel markers used, Columbia is homozygous for the insertion. In choosing to use indel markers we reasoned that deletions would be recalcitrant to enzyme repair or modification and therefore would help differentiate between enzyme-based mechanisms such as the one put forth by Comai and Cartwright [22] and a template-directed mechanism like the one previously proposed [20].

Initially, F3 progeny derived from hybrid F2 parent lines with known indel marker profiles were screened to test whether or not these markers were stable. When marker profiles were compared between hth-4 parent plants and their F3 adult offspring, 2.16% [6/277] deviated from the expected profile . When F3 progeny were assayed as seedlings, similar frequencies were seen, with 2.5% [15/600] of the F3 seedlings showing distinct marker profiles. Altogether six hundred seedlings were tested using a total of thirty seedlings per F2 plant (eleven hth-4, five hth-7, two hth-8 and two hth-10 F2 plants). Of the 15 F3 seedlings that tested positive for at least one non-parental marker, 7 had acquired insertions. In all cases, F2 parent plants used for these experiments were reared in plastic cones.

To test whether the observed genetic discordance between parent and offspring was symptomatic of sectoring, multiple tissue samples were collected from individual adult plants and indel marker profiles compared between these different samples. Molecular analyses confirmed that individual hth mutant plants were not always genetically uniform and that some tissue samples had novel marker profiles. For the plant shown in Figure 2A, seven out of eight samples scored as expected and were homozygous for the Landsberg deletion marker, however, one sample produced two amplicons, one of which co-migrated with the larger Columbia insertion band.

In general, when plants were genotyped and novel amplicons detected, the corresponding DNA band tended to show weaker fluorescence. This could be due to poor in vitro replication of target loci resulting in unequal DNA amplification or it could reflect the fact that tissue samples were heterogeneous and contained a mix of cells with different molecular genotypes. To distinguish between these possibilities, seedlings were cut at the shoot-root junction and the molecular genotype of the two halves determined separately. On the assumption that wild type plants would not produce sectors, identical tests were also conducted on wt hybrid lines as negative controls. In the majority of cases, as expected, there was a perfect correspondence between the molecular profiles of the root and shoot. However, in some cases, individual seedlings were found to have molecular signatures that differed between the two organ systems (10/44 hth-3 ; 1/50 hth-4; 9/76 hth-7 ; Figure 2B). Surprisingly, wt hybrid seedlings also showed novel genotypes when roots and shoots from the same seedling were compared (10/184 wt hybrids; Figure 2B).

Non-parental markers share sequence identity with the grandparent

A subset of amplicon samples were subjected to DNA sequence analyses in order to determine their molecular features. Sequence analyses of DNA clones derived from individuals where the non-parental amplicon co-migrated with the smaller deletion allele showed identity with the Landsberg deletion marker (Figure 3). In two instances, polymorphisms immediately upstream of the deletion were also detected (Figure 3A). As indicated, the Landsberg accession differs from Columbia at these exact three nucleotides. DNA sequence analysis of novel amplicons that co-migrated with the larger insertion allele showed that this seedling shoot had acquired a 54-nucleotide insertion that shares identity with the Columbia reference genome (Figure 3B). This same insertion was absent in the F2 parent plant. These particular seedlings descended from the same wt hybrid parent plant as the F3 progeny whose profiles are shown in Figure 2B.

Individual plants produce multiple, genetically distinct somatic sectors

To obtain an estimate of sector size, tissue samples were subjected to quantitative assays where the copy number of a genomic reference sequence immediately flanking the marker of interest was compared to the copy number of a sequence internal to that particular insertion marker (Figures 4 and 5, Table S2). Hybrid plants verified to be homozygous for a deletion at specific indel markers were subjected to the quantitative assays. The quantitative polymerase chain reaction (qPCR) data reveal two surprising findings. First, the majority of tissue samples collected from individual hth mutant plants tested positive for the presence of at least one insertion marker (Figure 4). In addition, multiple insertion sequences could be detected in many of the tissue samples tested (Figures 4B). In most instances, the copy number of any given insertion sequence, relative to the reference, was very low (less than one copy per 1000). Second, wt hybrid plants also showed evidence of sectors with novel genotypes (Figure 5). Only two out of four plants tested, however, showed evidence of unexpected insertions.

Discussion

Inbreeding is an integral part of the reproductive strategy used by numerous highly successful flowering plant species [29] including wild relatives of Arabidopsis where self-fertilization is thought to have evolved approximately one million years ago [10]. Despite have limited genetic variation, however, highly inbred species have continued to thrive. Why do organisms that exercise these reproductive strategies remain evolutionarily robust? In wild relatives of Arabidopsis this may reflect the fact that the genome is highly dynamic and can respond to changes in environmental conditions or other extrinsic factors. Genome responses include elevated rates of homologous recombination that persist for multiple generations [30], changes in copy number [31] and modulation of epigenetic gene regulation [32,33]. Pervasive genetic buffering [34,35] ensures that phenotypes with potentially deleterious consequences are attenuated. In addition to the genome responses listed above, our findings suggest that an intrinsic source of genetic variation can be leveraged to enhance the diversity in genetic output achieved by Arabidopsis plants. DNA sequence analyses suggest that these newly acquired sequences are of ancestral origin, sharing identity with sequences that most recently existed in the grandparental generation.

In showing that single individuals are genetic mosaics, cross-pollination and seed contamination can be completely and unequivocally discounted as sources of experimental error. Although it is possible that spurious contamination may have occurred, especially when tissues were collected from plants reared in growth chambers, this explanation becomes less tenable if consideration is given to reproducibility. By employing classical genetic approaches in conjunction with low and high resolution molecular methods, we have repeatedly detected novel genotypes in multiple individuals with distinct genetic backgrounds, reared in different growth chambers, at different times, as well as in seedlings grown under sterile conditions. Furthermore, a higher incidence of genetic discordance was found in hth mutants as compared to wt plants whether shoot and root systems were compared between aseptically grown seedlings or if tissue samples were taken from adult plants and subjected to qPCR. The risk of sample contamination would be minimal for samples grown aseptically and error due to contamination should not show a genotype-dependent bias.

In considering alternate template-dependent mechanisms, such as gene conversion or homologous recombination, none can account for the de novo appearance of unique sequence insertions. Nevertheless, it is possible that the insertion or deletion of small DNA sequence tracts, as described here, could reflect the activity of transposable elements [36,37]. However, numerous lines of evidence argue against this possibility. For instance, when novel amplicons were detected, they co-migrated with their corresponding insertion or deletion allele and did not show size heterogeneity, as would have been expected for transposon-driven excision or insertion events. Sequence data confirm that deletion events reproducibly eliminate a fixed length of sequence while insertion events reproducibly introduce a fixed sequence tract and both events repeatedly target specific genomic sites. Insertion and deletion events do not appear to produce obvious junction sites with altered nucleotides. Similarly, insertion events introduce sequences that share identity with the Columbia reference genome and do not appear to be chimeric gene or genome fragments. Furthermore, transposable element-mediated events cannot account for the fact that these insertion sequences appear to be generated de novo since no comparable conserved region of homology exists elsewhere in the host genome, as demonstrated by our qPCR data. Lastly, as determined by DNA database searches, none of the indel markers used in this study share significant sequence homology with annotated Arabidopsis transposable elements.

If the genome of an intensely studied model organism such as Arabidopsis is subject to modification by the mechanism we propose, why has this phenomenon not been described previously? Our research shows that target choice and methodological approach are critical in differentiating genome restoration events from other processes that also modify DNA sequences. Based on our findings, the only genomic targets that are truly diagnostic of genome restoration are deletions. To the best of our knowledge, deletions alleles have been used in genetic studies precisely because they are known to be stable and not to revert but have not been used to study phenomena related to epigenetic inheritance. There is no generalized precedent for genetic instability of deletions and assuming otherwise would go against an established biological paradigm. Polymorphic molecular markers such as single nucleotides, simple sequence repeats, or insertions that are subject to alterations by other processes will not provide sufficient resolution to differentiate mechanism, even though they are also likely targets for restoration. In particular, our findings may explain why genomic sequencing efforts have failed to register these sequence deviations or, if detected, why they may have been attributed to sequencing error and eliminated during curation. One possibility that immediately emerges from this prediction is that raw sequence data contained in existing genome database archives may actually already contain evidence of genome restoration, revealed by features such as highly biased loci-specific "errors".

Collectively, our genetic and molecular data show that many, and perhaps most, genome restoration events occur somatically in both seedlings and adult plants. Sectoring may therefore take place throughout development but be limited such that, at any given time, only a few cells host these genetic changes. Importantly, this may explain why sequence changes seen in revertant hth progeny have rarely been found to affect both alleles. Although sexual transmission of non-parental markers clearly does occur [20], the fact that we have not found wt HTH progeny among seed-derived offspring suggests that sectors populating the gamete forming lineages are tiny. The qPCR data are consistent with this supposition. However, it is also possible that mechanistic differences exist between somatic and reproductive tissues or that genome restoration events remain dynamic, limiting sexually transmitted changes to those that stabilize. It is also possible that certain genetic backgrounds condition this process as suggested by the greater number of events in detected in hth mutants.

In addition to validating our genetic and molecular data, the qPCR results extend those findings and suggest that individuals can produce multiple discreet sectors each with unique marker profiles and at many different growing points. This finding implies that sectoring may be a relatively common occurrence, even in wt genetic backgrounds. Since the adult plants used for these experiments were left largely intact and only a small proportion of the plant sampled, many more sectors may actually have been present than quantified. As such, it is possible that our current census underestimates the frequency with which these smaller islands of genetic variation arise. Although sectors are more readily detected using qPCR, this method cannot distinguish, for example, between single cells hosting multiple copies of an insertion sequence versus multiple clonally related cells that remain strictly diploid. Similarly, it is not possible to distinguish whether a single sector hosts the full complement of detected insertions, whether independent insertion events occur in multiple discreet sectors, or if those sectors overlap. Visualization of sectors in living tissue or tissue sections should help distinguish between these possibilities.

Recent work using computational modeling provides robust support for an ancestrally based "error-correcting" mechanism such as the one we propose to exist in Arabidopsis plants [38]. In these constrained-optimization simulations, the evolutionary robustness of "genetic repair" strategies was compared between populations that access repair templates derived either from parents, grandparents or great-grandparents. Interestingly, the data show strong support for a grandparent- or great grandparent-based genetic repair strategy over parental repair strategies. Furthermore, simulation results show that using a randomly selected template consistently gave superior results to those achieved using templates from the fittest parent or grandparent. From a biological perspective, such a strategy has considerable merit. Retaining a cache of templates derived from grandparental lineages would guarantee greater allele diversity precisely because the reservoir of allele variants would be deeper and allele redundancy would be less likely to occur. Half of the parental alleles are already represented in the offspring. Random selection of templates would be the most parsimonious strategy to affect genome repair, again because it would promote allele diversity. Since only those individuals that survived in previous generations would contribute to these cached templates, represented alleles would be biased to those that have proven robust under a spectrum of selective pressures.

If intrinsic drivers of genetic variation exist in inbreeding plant species, have incidents of this type of cryptic variation been documented in other systems? We believe that such events have indeed manifested as reports of enigmatic cases of phenotypic variation in soybean and cauliflower [7,8,39]. In other studies, molecular data have also been featured. In flax, for example, molecular assays have demonstrated that heritable phenotypic changes induced by environmental shifts are accompanied by reproducible changes in genomic DNA [14-16]. In soybean, reproducible non-random DNA sequence changes induced by in vitro culturing of root explants have also been demonstrated using restriction length polymorphic markers [17]. Genomic changes manifesting similar hallmarks of biased sequence alterations have also been described in rice [19,40] and corn [41] hybrids, as well as, in Arabidopsis [42,43].

The research presented here brings to light five striking findings. First, individual Arabidopsis plants are capable of producing somatic sectors during the course of normal vegetative development. Second, those sectors can have distinct and unique marker profiles and can differ in single nucleotide composition, can acquire small DNA insertions or can experience DNA sequence loss. Third, the de novo appearance of genomic insertions supports our original contention that cryptic sequence templates, possibly of ancestral origin, drive some of these changes [20]. Fourth, this phenomenon can be detected in wt genetic backgrounds raising the possibility that many Arabidopsis lab strains may be genetic mosaics. Finally, these genomic events can take place across all 5 chromosomes, whether or not the target loci reside within genes or between genes.

As posited by Whitham and Slobodchikoff [2], it may be that the formation of genetic mosaics is another strategy that allows plants to circumvent the detrimental effects of inbreeding [2]. Somatic sector formation permits the introduction of genetic variants into the gene pool either through vegetative propagation or through sexual reproduction. As these authors point out, germ line cells are derived from somatic tissues that arise late in the developmental history of the plant and therefore somatic mutations are more likely to introduce genetic variation than mutations that arise in the gametes [2,3,35]. By expanding the window of tolerance for genetic variation, plants may be afforded a better adaptive strategy given lifestyle constraints. The versatility of modular development combined with tolerance for genetic variation may allow plants to adapt at rates tailored to pathogen life cycles [44] or to relatively expanded time scales, such as those affecting climate change.

Our data expand on the ideas put forth by Whitham and Slobodchikoff [2] and suggest that sector formation, even in a short-lived organism like Arabidopsis, may be a normal part of development and, furthermore, that the formation of sectors serves to capture novel genetic variation, irrespective of the source of that variation. As our data show, this phenomenon may not necessarily affect every individual in the population but, for those individuals affected, the frequency of events varies greatly. Our findings raise the possibility that inbreeding plants and, perhaps other organisms that predominantly propagate asexually, may sequester cryptic sources of genetic variation that can be harnessed to promote greater genetic diversity. How these templates replicate, where they reside within the cell, how long they persist and whether equivalent mechanisms exist outside of the plant Kingdom are among the many questions that await further investigation.

Materials and Methods

Plant growth conditions

Seeds were sown onto moistened potting mix (1:1 mixture of LC1:LG3 Sungro Sunshine potting mixes, Sungro Horticulture, Seba Beach, AB) and vernalized at 4ºC for 2-5 days. Plants were maintained in growth chambers (Econoair AC60, Ecological Chambers Inc., Winnipeg, MB; GC8-VH/GCB-B, Environmental Growth Chambers, Chagrin Falls, Ohio; Conviron PGW36/E15, Controlled Environments Ltd., Winnipeg, MB) and illuminated with a mixture of incandescent and fluorescent lights (140 - 170 µmol m^-2 sec^-1 at pot level) with a 24 h photoperiod. Growth chambers were maintained at 20 ± 4ºC at 40 - 60% relative humidity. Plants were grown in flats or in 3- or 6-inch pots and watered as needed. Seeds used for seedling root-shoot comparison were surface sterilized using bleach and plated on agar medium containing half strength MS basal salts (Sigma, St. Louis, USA). Seedlings were harvested approximately 5 days post-germination. Hybrid lines were generated between Landsberg and Columbia accessions by manual pollination and all crosses were done reciprocally. F2 seed was obtained from self-fertilized F1 plants. Individual F2 plants were reared in plastic cones and F3 seed collected from each F2 plant individually. Tissue samples were collected from individual F2 and F3 plants, and genotypic profiles were determined ±using insertion-deletion polymorphic molecular markers (see Figure S1).

Out-crossing experiments

Experimental set ups were replicated twice and the net out-crossing frequencies determined. A 1:1 ratio of herbicide resistant transgenic pollen donors and mutant test plants were arranged in randomized positions (www.random.org). Out-crossing frequencies were also compared to plants under the same conditions but reared within plastic tubes. Progeny were sprayed with glufosinate (40 micrograms ml^-1 active ingredient: WipeOut, Nu-Gro IP Inc., Ontario) to test for herbicide resistance and resistant plants tested for segregation of hth mutant progeny plants.

DNA extraction andmolecular genotyping

For DNA extraction, rosette or cauline leaf tissue was collected and DNA extracted according to the method of Edwards et al. [45]. Samples not processed immediately were stored at -20ºC. Sixteen sets of PCR primers were designed to amplify approximately 150-300bp genomic regions, each containing one 45-94bp marker which is present in the Columbia but absent in the Landsberg accession (Table S1). PCR products were size separated by agarose gel electrophoresis.

Isolation, cloning and sequencing of PCR products

Portions of genomic DNA were PCR amplified and sequenced directly or products cloned into standard pGEM TA vectors (Promega). Amplified or cloned PCR products were sequenced at the Centre for Applied Genomics (http://www.tcag.ca/, Toronto, Ontario). Sequence alignments were generated using CLC Sequence Viewer 6.4 software (CLC bioA/S;www.clcbio.com).

qPCR Methods

Quantitative PCR was performed on a Bio-Rad Real-Time thermal cycler CFX96 attached to a computer running CFX Manager. SsoFast EvaGreen Supermix (Bio-Rad) was used according to manufacturer's instructions. The positive control was a PCR product amplified from the Columbia accession, flanking the indel sequence of interest by ~700-900bp. The positive control was gel purified and used to generate a standard curve for conversion of C(t) value to copy number of the insertion sequence and the external reference sequence.

Acknowledgements

We would like to express our gratitude and sincere appreciation to the many individuals who provided technical assistance and critical evaluation of experiments or editorial input on this manuscript. In particular, SJL would like to thank members of the Department of Biology, the Faculty of Science and University of Waterloo administration for their continued support and encouragement.

References

1. Szymkowiak EJ, Sussex IM (1996) What chimeras can tell us about plant development. Annual Review of Plant Physiology and Plant Molecular Biology 47: 351-376.

2. Whitham TG, Slobodchikoff CN (1981) Evolution by individuals, plant-herbivore interactions, and mosaics of genetic-variability - the adaptive significance of somatic mutations in plants. Oecologia 49: 287-292.

3. Satina S, Blakeslee AF (1941) Periclinal chimeras in Datura stramonium in relation to development of leaf and flower. American Journal of Botany 28: 862-871.

4. Youngson NA, Whitelaw E (2008) Transgenerational epigenetic effects. Annual Review of Genomics and Human Genetics 9: 233-257.

5. Hopkin M (2008) Biodiversity: Frozen futures. Nature 452: 404-405.

6. Walck J, Dixon K (2009) Time to future-proof plants in storage. Nature 462: 721-721.

7. Fasoula VA, Boerma HR (2005) Divergent selection at ultra-low plant density for seed protein and oil content within soybean cultivars. Field Crops Research 91: 217-229.

8. Fasoula VA, Boerma HR (2007) Intra-cultivar variation for seed weight and other agronomic traits within three elite soybean cultivars. Crop Science 47: 367-373.

9. Rasmusson DC, Phillips RL (1997) Plant breeding progress and genetic diversity from de novo variation and elevated epistasis. Crop Science 37: 303-310.

10. Tang CL, Toomajian C, Sherman-Broyles S, Plagnol V, Guo YL, et al. (2007) The evolution of selfing in Arabidopsis thaliana. Science 317: 1070-1072.

11. Barrett SCH (2002) The evolution of plant sexual diversity. Nature Reviews Genetics 3: 274-284.

12. Clark RM, Schweikert G, Toomajian C, Ossowski S, Zeller G, et al. (2007) Common sequence polymorphisms shaping genetic diversity in Arabidopsis thaliana. Science 317: 338-342.

13. Cullis C, Swami S, Song Y (2004) RAPD polymorphisms detected among the flax genotrophs. Plant Molecular Biology 41: 795-800.

14. Chen YM, Schneeberger RG, Cullis CA (2005) A site-specific insertion sequence in flax genotrophs induced by environment. New Phytologist 167: 171-180.

15. Chen Y, Lowenfeld R, Cullis C (2009) An environmentally induced adaptive (?) insertion in flax. International Journal of Genetics and Molecular Biology 1: 038-047.

16. Schneeberger RG, Cullis CA (1991) Specific DNA alterations associated with the environmental induction of heritable changes in flax. Genetics 128: 619-630.

17. Roth EJ, Frazier BL, Apuya NR, Lark KG (1989) Genetic-variation in an inbred plant - variation in tissue-cultures of soybean [Glycine-max (L) Merill]. Genetics 121: 359-368.

18. Oh TJ, Cullis MA, Kunert K, Engelborghs I, Swennen R, et al. (2007) Genomic changes associated with somaclonal variation in banana (Musa spp.). Physiologia Plantarum 129: 766-774.

19. Xu PZ, Yuan S, Li Y, Zhang HY, Wang XD, et al. (2007) Genome-wide high-frequency non-Mendelian loss of heterozygosity in rice. Genome 50: 297-302.

20. Lolle SJ, Victor JL, Young JM, Pruitt RE (2005) Genome-wide non-mendelian inheritance of extra-genomic information in Arabidopsis. Nature 434: 505-509.

21. Chaudhury A (2005) Plant genetics - Hothead healer and extragenomic information. Nature 437: E1-E1.

22. Comai L, Cartwright RA (2005) A toxic mutator and selection alternative to the non-Mendelian RNA cache hypothesis for hothead reversion. Plant Cell 17: 2856-2858.

23. Krishnaswamy L, Peterson T (2007) An alternate hypothesis to explain the high frequency of "revertants" in hothead mutants in Arabidopsis. Plant Biology 9: 30-31.

24. Mercier R, Jolivet S, Vignard J, Durand S, Drouaud J, et al. (2008) Outcrossing as an explanation of the apparent unconventional genetic behavior of Arabidopsis thaliana hth mutants. Genetics 180: 2295-2297.

25. Peng P, Chan SWL, Shah GA, Jacobsen SE (2006) Plant genetics - Increased outcrossing in hothead mutants. Nature 443: E8-E8.

26. Ray A (2005) Plant genetics - RNA cache genome trash? Nature 437: E1-E2.

27. Lolle SJ, Hsu W, Pruitt RE (1998) Genetic analysis of organ fusion in Arabidopsis thaliana. Genetics 149: 607-619.

28. Koornneef M, Hanhart C, Thiel F (1989) A genetic and phenotypic description of Eceriferum (cer) mutants in Arabidopsis thaliana . Journal of Heredity 80: 118-122.

29. Wright SI, Ness RW, Foxe JP, Barrett SCH (2008) Genomic consequences of outcrossing and selfing in plants. International Journal of Plant Sciences 169: 105-118.

30. Molinier J, Ries G, Zipfel C, Hohn B (2006) Transgeneration memory of stress in plants. Nature 442: 1046-1049.

31. DeBolt S (2010) Copy Number Variation Shapes Genome Diversity in Arabidopsis Over Immediate Family Generational Scales. Genome Biology and Evolution 2: 441-453.

32. Lang-Mladek C, Popova O, Kiok K, Berlinger M, Rakic B, et al. (2010) Transgenerational Inheritance and Resetting of Stress-Induced Loss of Epigenetic Gene Silencing in Arabidopsis. Molecular Plant 3: 594-602.

33. Boyko A, Kovalchuk I (2011) Genome instability and epigenetic modification - heritable responses to environmental stress? Current Opinion in Plant Biology 14: 260-266.

34. Queitsch C, Sangster TA, Lindquist S (2002) Hsp90 as a capacitor of phenotypic variation. Nature 417: 618-624.

35. Sangster TA, Salathia N, Lee HN, Watanabe E, Schellenberg K, et al. (2008) HSP90-buffered genetic variation is common in Arabidopsis thaliana. Proceedings of the National Academy of Sciences of the United States of America 105: 2969-2974.

36. Lisch D (2009) Epigenetic Regulation of Transposable Elements in Plants. Annual Review of Plant Biology 60: 43-66.

37. Tenaillon MI, Hollister JD, Gaut BS (2010) A triptych of the evolution of plant transposable elements. Trends in Plant Science 15: 471-478.

38. FitzGerald A, O'Donoghue DP, Liu XY (2010) Genetic Repair Strategies Inspired by Arabidopsis thaliana. In: Coyle L, Freyne J, editors. Artificial Intelligence and Cognitive Science. pp. 61-71.

39. Chable V, Rival A, Cadot V, Boulineau F, Salmon A, et al. (2008) "Aberrant" plants in cauliflower: 1. Phenotype and heredity. Euphytica 164: 325-337.

40. Gao DY, He B, Zhou YH, Sun LH (2011) Genetic and molecular analysis of a purple sheath somaclonal mutant in japonica rice. Plant Cell Reports 30: 901-911.

41. Tracy WF, Talbert LE, Gerdes JT (2000) Molecular variation and F-1 performance among strains of the sweet corn inbred P39. Crop Science 40: 1763-1768.

42. Yi H, Richards E (2008) Phenotypic instability of Arabidopsis alleles affecting a disease Resistance gene cluster. Bmc Plant Biology 8.

43. Yi H, Richards E (2009) Gene duplication and hypermutation of the pathogen Resistance gene SNC1 in the Arabidopsis bal variant. Genetics 183: 1227-1234.

44. Todesco M, Balasubramanian S, Hu TT, Traw MB, Horton M, et al. (2010) Natural allelic variation underlying a major fitness trade-off in Arabidopsis thaliana. Nature 465: 632-U129.

45. Edwards K, Johnstone C, Thompson C (1991) A simple and rapid method for the preparation of plant genomic DNA for PCR analysis. Nucleic Acids Research 19: 1349-1349.

Financial Disclosure

We gratefully acknowledge funding from the Natural Sciences and Engineering Research Council of Canada (NSERC) and the University of Waterloo (UW). MTH, PCC, and DL each were supported by NSERC Fellowships. AMK was supported by an Ontario Graduate Scholarship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Supplemental Information

View Supplemental Figures and Tables - Figure S1, Table S1, Table S2 - (PDF)