Remembrances of an embryo: long-term effects on phenology traits in spruce
(*Author for correspondence:
tel +32 (0)9 272 2953; fax +32 (0)9 272 2901; email firstname.lastname@example.org)
Coping with their sessile life style, plants have become masters in adaptation. Through adaptation, they improve their chances of survival and reproduction in a potentially changing environment. Within the frame of optimal adaptation, phenology – the seasonal timing of growth processes – determines to a large extent the geographic distribution of many plants. By consequence, phenology traits often display some kind of latitudinal, altitudinal or drought clines that have been traditionally explored in breeding programs and provenance trials.
In Norway spruce (Picea abies), the translocation of northern ecotypes (64°–66°N) into a southern orchard (58°N), for better seed production, resulted in progeny that no longer resembled the phenology of their siblings produced in situ in the northern location. This effect persisted for years in the progeny (Skrøppa et al., 2007). In line with this, the inverse experiment (translocation of a southern ecotype to the north) also produced, within one generation, progeny that were adapted to the northern location (Ø. Johnsen & T. Skrøppa, pers. comm.). Importantly, this observation is not limited to ecotypes that are displaced into a different environment, but extends to phenology characteristics of seedlings that were produced in warm or cold years within the same stand (Kohmann & Johnsen, 1994). In addition to Norway spruce, similar effects have only been demonstrated in Picea glauca×Picea engelmannii and in Scots pine (Pinus sylvestris; Dormling & Johnsen, 1992; Webber et al., 2005). Quite surprising is the lack of substantial evidence on comparable phenomena in angiosperm trees, leaving the generality of the phenomenon an open issue.
‘... increased genome dynamics including epigenetic components might be a prerequisite for increasing the raw material for adaptive evolution.’
The molecular basis of these long-term effects in Norway spruce is currently unexplained. Johnsen and co-workers have established that the prevailing temperature during zygotic embryogenesis and seed maturation sets the phenology response and that the memory is probably of an epigenetic nature (Johnsen et al., 2005a,b). Now, Johnsen and colleagues have repeated the original experiments with somatic, instead of zygotic, embryos and present their findings in this issue of New Phytologist (Kvaalen & Johnsen, pp. 49–59). The strength of this approach is provided by the elimination of genetic effects through the use of clonal material and a focus on temperature effects during zygotic and somatic embryogenesis. Kvaalen & Johnsen demonstrate that somatic embryos derived from identical genetic material, but exposed to different temperatures during zygotic and somatic embryo development, show marked differences in the critical night length for bud set in the second year of seedling growth. The difference in critical night length for bud set amounts to a remarkable 2 h, corresponding to 4–6° latitude in Norway. Furthermore, genetic selection during prezygotic stages, or selection of a particular embryo, could differ with varying temperature environments, but were excluded through demonstrating the absence of any genetic marker distortion (Besnard et al., in press).
These temperature effects on the long-term phenology of the progeny do clearly differ from maternal provision or maternal effects sensu stricto. Maternal provision usually results in better-nurtured seeds that in turn have an obvious advantage during seedling establishment. The somatic embryos germinate at the same frequency, regardless of whether they have derived from a cold or a warm environment during embryogenesis and maturation, refuting an after-effect as a result of different maturity of the embryos or maternal provision. Moreover, the somatic embryos develop detached from the mother, identifying the embryo as the entity that senses and responds to temperature. In the provenance trials it is thus possible that zygotic embryos, rather than the mother tree, sense temperature. However, the independence of the memory effect from the mother plant or the female gametophyte remains to be established for zygotic embryos. Still, the maternal environment might influence the embryo sensitivity and response to temperature, or affect the magnitude of the epigenetic memory. Thus, the temperature-based memory effects in Norway spruce may not conform to all tenets of a typical maternal effect, still, the mechanisms underlying maternal effects can be useful for understanding the putative basis of the memory effects.
Maternal effects as a mechanism for adaptive transgenerational phenotypic plasticity
Maternal effects as a mechanism to pass an acquired adaptation to progeny are usually observed if the progeny are likely to live in the same environment as the mother plant (Galloway, 2005). In Norway spruce, the temperature during seed development depends on latitude, altitude and microclimate, and might thus correlate closely with day length and temperature that will govern the bud set of future progeny living in the same habitat. Maternal effects are particularly important if the spatial scale of environmental variation is greater than that of pollen movement or seed dispersal (Galloway, 2005), as is the case for Norway spruce. If habitat patches (micro-environments) display continuity over generations, then maternal effects will lead to phenotypic adaptation to the environment, thus, to habitat selection.
The effects of maternal environment have been documented in ecological studies, although the gain in fitness is often not described (Galloway, 2005). In yellow monkeyflower (Mimulus guttatus), for example, simulated insect damage on early leaves provokes increased trichome density on later leaves (within the generation) and in yet-unchallenged progeny originating from treated parents (across generation; Holeski, 2007). In American bellflower (Campanula americana), the offspring life style (annual vs biennial) is influenced by the maternal light environment in its forest habitat. This transgenerational plasticity is adaptive when offspring are grown in their maternal light environment (understory vs light gap), where seeds typically disperse. Offspring developing in light gaps that coincide with the environment of the mother show a 3.4-fold increase in fitness over those grown in understory (Galloway & Etterson, 2007).
Maternal effects can thus constitute a source of adaptive plasticity between generations, in which the offspring are predisposed with enhanced fitness to the environment they are likely to experience. The mechanism by which the maternal environment experienced during seed development imposes a particular disposition to the offspring is not yet understood. Moreover, it is unknown for how long these effects exert constraints upon the progeny.
Epigenetics as a basis for memory within and across generations
DNA methylation, chromatin and noncoding RNAs that, in turn, govern changes in DNA methylation and chromatin, are the molecular basis for epigenetic effects (Henderson & Jacobsen, 2007). The involvement of epigenetic components in the regulation of gene expression has clearly come to our attention. Environmental factors can play a substantial role in modulating the different epigenetic systems, as evident from the vernalization response in Arabidopsis. A prolonged period, but not short episodes, of cold lead to covalent histone modifications of particular loci that are involved in the stable repression of flowering time locus C (FLC), a repressor of flowering in Arabidopsis (Sung & Amasino, 2005). A similar mechanism may explain the memory effect in spruce, although vernalization is by comparison a short-term memory.
Perhaps less recognized is that various environmental stresses influence genome stability, as documented for the activation of transposons in maize, genomic rearrangements in flax and altered frequencies of homologous recombination (Bond & Finnegan, 2007). In one very instructive example from Arabidopsis, Molinier et al. (2006) applied ultraviolet light and a bacterial peptide mimicking pathogen infection to young plants. Both stresses provoked the parental generation and four (unchallenged) filial generations to have higher frequencies of homologous recombination. The adaptive value of higher recombination frequencies with respect to the applied stresses is unexplained. Molinier et al. (2006) propose an epigenetic mechanism that possibly acts through alterations in chromatin condensation to make loci more accessible for recombination. Another study of a methylation-deficient Arabidopsis mutant showed that methylated CpG are central to epigenetic memory across generations (Mathieu et al., 2007). Loss of methylated CpG triggered other epigenetic mechanisms, but these acted in an uncoordinated manner. Together, both studies established an epigenetic mechanism for transgenerational memory and underscore a central role for DNA methylation for its stability. Similar mechanisms, whether requiring a transgenerational component or not, might apply for the observed memory effects in Norway spruce.
The ‘problematic’ aspect of epigenetic mechanisms is their variable inheritance. They are frequently heritable through mitosis, sometimes for multiple generations (Richards, 2006). Thus, during sexual reproduction epigenetic information is partly reset during meiosis and partly transmitted through meiosis. The epigenetic marks at the FLC locus involved in the vernalization response are reset during meiosis, whereas transposon methylation is stably maintained (Bond & Finnegan, 2007). Nevertheless, increased genome dynamics, including epigenetic components, might be a prerequisite for increasing the raw material for adaptive evolution (Kalisz & Purugganan, 2004; Rapp & Wendel, 2005). Rapp & Wendel (2005) suggest that a population bottleneck, while reducing genetic diversity, might create at the same time epigenetic novelty. In contrast to genetic alleles, epialleles might react more quickly to environmental change, be reversible and persist for a number of generations only (Kalisz & Purugganan, 2004). If the epiallele were to cause a mild phenotype, through an alteration of the degree of gene expression, it might experience weaker selection than a loss-of-function sequence mutation (Kalisz & Purugganan, 2004). The significance of epialleles in wild populations will depend on their frequency and stability.
In conclusion, initial reports establish epigenetic effects as a basis for memory within and across generations. Future studies are needed to investigate whether and to what extent adaptation through phenotypic plasticity, such as reported by Kvaalen & Johnsen, relies on epigenetic mechanisms.
Implications of an epigenetic memory for bud set
Bud set is among the most differentiated genetic traits in many temperate trees (Howe et al., 2003). Along latitudinal and altitudinal clines strong genetic differentiation prevails for phenology traits, typically resulting in locally adapted ecotypes. The memory mechanism suggested by Johnsen and co-workers would counteract this strong differentiation through introducing new sources of phenotypic variation at the local community level, provided it were a widespread phenomenon in the wild and that successive seed years varied considerably in temperature.
The strong differentiation or genetic constraint, on the one hand, and counteracting mechanisms that introduce variation and/or phenotypic plasticity from various sources, on the other, might of course reflect flip sides of the same trait. Alternatively, the clinal gradient in bud set might mask a very large within-stand variation. Conventional provenance-level quantitative genetic variation (i.e. that among geographic origins) might reflect both the directional selection and the phenotypic plasticity. The ease of provenance transfer in Norway spruce is interpreted as a sign of phenotypic plasticity (Skrøppa et al., 2007). To achieve reliable estimates of phenotypic plasticity (magnitude of effect and its heritability), clonally replicated progeny of a cross would need to be assessed over a variety of environments (Pigliucci, 2005). Additionally, no quantitative methods currently exist to distinguish between genetic and epigenetic components affecting bud set. Whether contributing to trait variance (covered in the genetic variance) or to variation in developmental stability (contained in phenotypic variance), epigenetic components are currently undetected (Kalisz & Purugganan, 2004). Epigenetic effects might thus inflate the variation in bud set among the provenances, causing the clines to reflect more local adaptation than actual genetic differences can account for. Taken to the unlikely extreme, if provenances indeed adapt within one generation to the local conditions, much of the phenotypic plasticity might come through embryo memory and/or maternal effects.
There are uncertainties as to whether this long-term effect on phenology will lead to a comparable outcome and significance in wild populations that will experience warmer temperatures as a result of global warming. The discussion continues as to whether evolutionary adaptation could cope at all with the expected speed of climate change (Huntley, 2007; St Clair & Howe, 2007). The contribution of adaptive genetic evolution and phenotypic plasticity to the already observed phenological changes in the past decades is currently unknown. However, within a 100-yr timeframe for strong climate change, migration and evolution in place are rather unlikely scenarios for adequate genetic adaptation, particularly for forest trees. Eurasian Scots pine (P. sylvestris) was predicted to need as many as 12 generations to evolve to the new optima in future climates (Rehfeldt et al., 2002). Discussing this apparent dilemma for Douglas fir (Pseudotsuga menziesii), St Clair & Howe (2007) recommend human interference through movement of populations from south to north, from lower to higher altitude and the deployment of mixed-seed sources to ensure forests that can face future climates. However, perhaps the within-population variation in climatic responses is greater than often assumed, even in marginal areas (Johnsen & Østreng, 1994). Within this scenario, the observed memory effect might constitute a source of phenotypic plasticity, although the actual contribution of this memory effect and the putative maternal influences to the total phenotypic variation in natural populations remain to be uncovered.
Future experiments are needed to assess whether it is also the case for zygotic embryos that epigenetic memory is established within the embryo itself and whether or not it involves transgenerational inheritance from the parental generation. Reciprocal hybrids can unequivocally establish to what extent gene expression is affected by maternal effects (affecting both paternal and maternal alleles in trans) or by genomic imprinting (specific to only one parental allele). Furthermore, systematic analyses of the epigenome should ascertain the epigenetic nature of the phenomenon. A first estimate of the extent of methylation can be provided using methylation-sensitive amplified fragment length polymorphism (AFLP) markers. Other viable approaches include the study of specific genes involved in phenology responses, such as the three phytochrome genes whose expression levels in seedlings appeared to correlate with different temperatures during zygotic embryogenesis (Johnsen et al., 2005a). The methylation patterns of these and other genes should be determined in both parental gametes, in the zygote and in the seedlings to demonstrate whether methylation patterns are established de novo.
In addition to bud set, the warmer temperature during embryogenesis and seed maturation leads also to delayed bud burst and dehardening in spring, to later bud set, to delayed frost-hardiness development and to delayed lignification (Johnsen et al., 2005a,b). Although all traits could depend on one pleiotropic regulator, it is also possible that the higher temperature during seed development results in a nonspecific alteration of chromatin configuration that would affect many traits. Monitoring constructs, such as those used for quantifying homologous recombination in Arabidopsis (Molinier et al., 2006), could reveal such alterations. Similarly, general loss-of-function approaches to perturb or erase the memory should clarify whether DNA methylation or chromatin configuration were the molecular basis of the memory.
Much of the suggested experimentation is highly speculative and would be extremely challenging, if not impossible, in Norway spruce – a species in which intensive molecular studies are very difficult. This situation calls for a main research need in this area, namely whether comparable effects could be identified in model plants that are more amenable to molecular methods. Potentially, the lack of data in other (tree) species is the result of mere ignorance of these phenomena, as there appear to have been very few, if any, serious attempts at study. Conifers are, however, distinct in having very large genomes that possess a higher abundance of epigenetic mechanisms in place to control the repetitive parts of the genome. It is therefore feasible that in conifers such as Norway spruce, this epigenetic regulation might be more easily recruited for the regulation of other processes. Nevertheless, whatever the reasons underlying the observations of Kvaalen & Johnson, they reveal an important mechanism for adaptation to new and unstable environments. Knowledge on adaptation is so fundamental for forestry and forest genetic conservation that these memory effects deserve further investigation.