Vertically- and horizontally-transmitted memories - the fading boundaries between regeneration and inheritance in planaria

The Weismann barrier postulates that genetic information passes only from the germline to the soma and not in reverse, thus providing an obstacle to the inheritance of acquired traits. Certain organisms such as planaria – flatworms that can reproduce through asymmetric fission – avoid the limitations of this barrier, thus blurring the distinction between the processes of inheritance and development. In this paper, we re-evaluate canonical ideas about the interaction between developmental, genetic and evolutionary processes through the lens of planaria. Biased distribution of epigenetic effects in asymmetrically produced parts of a regenerating organism could increase variation and therefore affect the species' evolution. The maintenance and fixing of somatic experiences, encoded via stable biochemical or physiological states, may contribute to evolutionary processes in the absence of classically defined generations. We discuss different mechanisms that could induce asymmetry between the two organisms that eventually develop from the regenerating parts, including one particularly fascinating source – the potential capacity of the brain to produce long-lasting epigenetic changes.

KEY WORDS: Planaria, Regeneration, Memory, Inheritance, Epigenetics, Evolution, Generations, Transgenerational, Small RNAs, Chromatin

Most models of evolution, which are based on Mendelian genetics, depend conceptually on the existence of a distinct separation between generations across an ancestry. This distinction between parents and children is supposedly enforced by Weismann's barrier, which in theory precludes information transfer from the soma to the germline, and thus prevents inheritance of parentally-acquired traits (Poulton et al., 1889; Sabour and Schöler, 2012). The germline, according to this framework, is conceived as a ‘bottleneck’, which filters out epigenetic responses. In other words, all the changes that affect somatic cells, whether epigenetic or genetic (e.g. mutations, transpositions), are erased in the next generation. Lamarck's discarded theory of evolution, according to which somatic responses (and acquired traits) are carried over to the progeny, assumed a continuation between the generations, and until recently was considered to be entirely incorrect (Jablonka and Lamb, 2015). New discoveries in the field of epigenetics, some of which will be discussed here, suggest the need for reexamination of these original ideas in a new light.

As unicellular organisms have been shown to preserve cellular states over generations (Zacharioudakis et al., 2007), Weismann's barrier as originally suggested is relevant to organisms that have a well-defined and segregated germline (namely, only specific, designated cells will become germ cells). However, do similar restrictions on the process of evolution apply to plants (where the germline is not segregated), or to the many phyla of animals that can reproduce asexually without going through a germline bottleneck?

Even in metazoans, which segregate their germline and for which Weismann's barrier is supposedly relevant, different mechanisms are used to specify the primordial germ cells (Extavour, 2003). These different mechanisms allow different degrees of communication between the parent's environment and the germline. Recent evidence suggests that the variance between germline specification mechanisms could influence the process of evolution, and specifically, that a continuity with the previous generation could accelerate evolution. For example, it was shown that genes evolve faster in amphibians that define their germline by using maternally inherited determinants (‘preformation’), in comparison to the rates of gene evolution seen in related organisms that define their germline by inductive signals (‘epigenesis’), without inheriting ‘germplasm’ (which should be affected by the environment) from the mother (Evans et al., 2014).

One asexually reproducing animal, on which we will focus in this paper, which presents an interesting challenge to the Weismann barrier, is planaria. Planarians are an order of free-living flatworms which are complex bilaterians possessing a wide range of cell types, a true centralized brain, and a complex repertoire of behavioral responses (Saló et al., 2009). Planaria have advanced mechanisms of regeneration, and are able to coordinate their resident population of stem cells to recreate any portion of the animal that is surgically removed, including their brain, throughout adulthood (Roberts-Galbraith and Newmark, 2015). These attributes have made it a popular model system for studies of stem cell regulation, morphogenesis, behavioral plasticity, and physiological signaling (Gentile et al., 2011; Nicolas et al., 2008; Shomrat and Levin, 2013).

While many of the common planarian species (which are grown in the lab and are considered model organisms for regeneration) can reproduce sexually (Cardona et al., 2006), they most frequently reproduce asexually through fission followed by regeneration. Upon bisection (whether externally induced or self-initiated), a structure called the ‘blastema’ forms in each fragment (Birnbaum and Sánchez Alvarado, 2008). The blastema gives rise to new tissues, and a process of remodeling then scales both new and existing structures appropriately (Beane et al., 2013). When a head fragment regenerates its missing tail, or when a tail fragment regenerates a missing head, new cells differentiate from pluripotent stem cells known as ‘neoblasts’. These unique cells are required for regeneration, and also for the continuous remodeling and morphological rescaling observed in intact worms during growth and starvation (Oviedo et al., 2003). The neoblasts are instructed both by intrinsic state (cell-autonomous pathways) and information from surrounding cells (Oviedo and Levin, 2007; Oviedo et al., 2010; Wagner et al., 2011; Witchley et al., 2013).

Here, we explore a number of scenarios that could potentially defy classical models of evolution. Specifically, we ask whether in planaria and other organisms that reproduce by fission, different types of epigenetic information are asymmetrically passed across generations. Such stored information, which can be regarded as memory (see more below and in the glossary), could play many crucial roles in regulating behavioral and developmental patterns. In this manuscript we will discuss different types of memories that may persist upon regeneration/inheritance; memories of gene activity, memories which are encoded in the connectivity of neuronal circuits, and memories of non-neural physiological states.

In the broadest sense of the word, memory is what enables altering of future responses based on history. Biological memory is encoded at many levels: metabolic differences (Cameron et al., 2012; Ros et al., 2006), epigenetic factors (e.g. small RNAs, histone marks, DNA methylation and prions) (Bird, 2002; D'Urso and Brickner, 2014; Iwasaki and Paszkowski, 2014), stable bioelectrical circuit modes (Cervera et al., 2014; Law and Levin, 2015), or neuronally-encoded memories (Axmacher et al., 2006; Daoudal and Debanne, 2003; Herry and Johansen, 2014; Maren and Quirk, 2004; Zhang and Linden, 2003). A myriad of mechanisms exist to allow molecules, molecular pathways, cells, and cellular networks to transduce physiological or behavioral inputs (experiences) into stable state changes that guide future activities. In this sense, processes that ensure the persistence of different developmental fates or trajectories are also forms of memory.

The Weismann barrier is relevant to asexual organisms as well, because the issue is not only which cells will contribute to the next generation, but whether and how the life history of the body gets permanently encoded in cells so as to significantly alter the offspring in a stable manner. Indeed, the potential breaching of the Weismann barrier in planaria has previously been considered, in the context of tracking the source of the cellular contents of neoblasts that form a new organism (Solana, 2013). However, could parentally-produced alterations that encode biological memory breach Weismann's barrier and persist across generations? Even if information could travel from somatic tissues to the germline, several rounds of reprogramming events (in the germline and in the embryo) were previously thought to prevent the inheritance of epigenetic memory in animals (Mann and Bartolomei, 2002; Messerschmidt et al., 2014; Morgan et al., 2005; Vucetic et al., 2010). Nevertheless, in recent years it has become clear that complex and still poorly understood regulatory processes determine which epigenetic memories would persist, and which would be erased across generations. The removal of DNA cytosine methylation and histone marks during embryogenesis was thought to ‘clean’ the embryo of epigenetic modifications that were present on its parents' genome. The addition of de novo chromatin modifications in the next generation was similarly thought to depend solely on the current environmental conditions, and the dictation of the hard-wired, genomically-encoded developmental program. However, reprogramming is not complete and a few parental marks escape removal (Hackett et al., 2013).

How widespread are heritable memories and what types of memories avoid reprogramming? We will explore these questions through planaria, by focusing on the events that take place when animals reproduce by fission.

Memory: retention of information about a state of affairs for some time period; the ability of a system to specifically alter some aspect of a labile medium in response to stimuli, such that future responses to stimuli are altered. Memory requires latency between stimulus and salient response.

Epigenetic modifications: defined here as factors that alter the phenotype that are not stored in the genetic code, including but not limited to DNA methylation, histone modifications and small RNAs.

Bioelectric network/circuit: a group of cells, not restricted to neurons/muscle, often connected by gap junctions, which communicate via slow changes in resting potential and endogenous electric fields, which regulates cell state and large-scale morphogenesis.

Maternal effects: factors that alter the phenotype of the progeny that depend on the maternal environment, including genetic, epigenetic and physiological effects.

Epimutations: as opposed to DNA mutation, an epimutation is a molecular alteration to the DNA that does not alter the DNA sequence that can be stably transmitted across generations. Most commonly refers to differences in cytosine methylations between certain alleles. Epimutations can be segregated with the chromosomes in accordance with Mendel's roles.

Plant embryo: a phylogenetically conserved structure that develops from the zygote containing the shoot and root apical meristems, and the primordial tissues that will differentiate into tissues of the mature plant.

Meristem: in plant biology, meristems are self-maintaining structures of undifferentiated cells from which plant organs develop.

We hypothesize that the asymmetric fission of planaria, and similar organisms, and the resulting genetic and epigenetic differences in the individuals that regenerate from the different fragments, can create stable variation and therefore participate in the process of evolution.

A generation can be defined as ‘a single step in natural descent’ (http://www.dictionary.com/, accessed 2015). In planarian asexual reproduction, this definition does not necessarily apply, since after fission the relationship between the two resulting individuals does not display a clear hierarchy – which half is the ‘parent’ and which half is the ‘child’? Is one half ‘older’ than the other? Despite these ambiguities, we suggest that parentally-acquired information (the result of the parent's life experiences) could be transmitted from the worm that underwent splitting to the two organisms that form upon regeneration, and therefore the term ‘inheritance’ is relevant when discussing fission. The term ‘genetics’ could also be relevant in this regard, although, as will be elaborated below, the information that is inherited from the parent might not be restricted to changes in genes.

Fission and regeneration in planaria involve long-range instructive communication among cells (a signaling mode that can facilitate breaches of Weismann's barrier). When a worm is bisected, cells on the anterior- and posterior-facing sides of the cut must form a tail and head, respectively; the cut plane separates cells that were adjacent neighbors, and therefore had essentially the same positional information, yet these generate completely different anatomical structures. Thus, cell position (the local microenvironment) does not uniquely dictate the appropriate morphological outcome; instead, cells must communicate with other remaining tissues in order to determine which structures each blastema needs to build (Nogi and Levin, 2005; Oviedo et al., 2010; Reddien and Sánchez Alvarado, 2004). A similar long-range, highly integrated pattern control is seen in amphibians, where tails transplanted to the side of a salamander eventually remodel to limbs (including the transformation of the tail tip into fingers, which reveals that tissues can change their morphological structure in response to global patterning cues) (Farinella-Ferruzza, 1956).

The process of regeneration is essentially one of cell networks processing information about large-scale growth and form. A focus on information reveals an interesting analogy between generational descent and regeneration; that of space versus time. With classical generational inheritance, patterning information is passed on temporally from parent to offspring via the genome, conserved with high fidelity and yet susceptible to environmental influence. In regeneration, in addition to its temporal progression, instructive information is also propagated spatially, from the rest of the body to a wound region and thus to new tissues; planarian regeneration is a remarkable example of how these two distinct but highly parallel pattern control processes converge. It should be noted that while we focus on planaria as a uniquely tractable model for these studies, stable modifications to regenerative pattern occur also in mammals (trophic memory in deer antlers) and other invertebrate systems such as crab limbs (reviewed in Lobo et al., 2014).

The parallelism between development and regeneration is also seen at the cellular level, as manifested in the similarities between germ cells and the stem cells that enable regeneration in planaria (Solana, 2013). In asexual reproduction, both tail and head fragments regenerate their missing tissues through the proliferation and differentiation of neoblasts. Thus, when planarians reproduce asexually, the new generation does not originate from one cell, but from a ‘community of cells’ (generation/regeneration of a worm from a single neoblast without a surrounding mature body has never been shown). Because genomic changes arise during cell division, and as a result of DNA damage of different sorts, this ‘cell community’ is expected to be composed of a mixture of different neoblasts, and also from genomically-different surrounding cells, which are not totipotent. It was recently demonstrated that the different neoblasts are not completely genetically identical – even in the same individual, a large number of mutations and SNPs differentiate between neoblasts (Nishimura et al., 2015). Moreover, it is not clear that the information that is required for regeneration (where, when, and how much to make of the new cell types, how to arrange those new tissues in correct geometric patterns, and crucially, when to stop growing) is present in the neoblasts; thus, the genetic variance in the surrounding cells could also be crucial, and differentiate the organisms that grow from the two regenerating halves. The barriers to the interaction between the surrounding cells and the neoblasts are also analogical to the Weismann barrier, between somatic cells and germ cells.

Similarly to the germ cells of other animals, planarian neoblasts (unlike other cell types) express PIWI homologues (Friedländer et al., 2009). In other organisms, PIWI proteins, and PIWI-associated small RNAs, or piRNAs, are important for maintaining the immortality of the germline (Meister, 2013), and their role in somatic tissues in less clear (Rajasethupathy et al., 2012). In Caenorhabditis elegans, for example, animals without a germline are virtually devoid of piRNAs (Bagijn et al., 2012). PIWI proteins and piRNAs play a critical role in the silencing of transposons and enable distinction between ‘self’ and ‘foreign’ genes, and therefore preserve the progeny's genome (Rechavi, 2014). The heritable small RNA pool, which includes piRNAs and other types of small RNAs (e.g. endo-siRNAs in C. elegans) (Claycomb, 2014; Gent et al., 2010; Rechavi et al., 2014; Vasale et al., 2010), and tRNA-fragments in mice (Chen et al., 2016; Liao et al., 2014; Peng et al., 2012; Sharma and Rando, 2014), constitutes a germline ‘memory bank’ of sequences that were found in past generations to be ‘dangerous’ (mobile parasitic DNA elements) or ‘safe’ (genes that need to be expressed in the germline). Transmission of piRNAs to progeny ensures that transposons will not jump, thus preventing disruption of the germline's genome, and ensuring error-proof transgenerational information transfer (Malone and Hannon, 2009). Neoblasts, which grant planarians their powerful ability to regenerate endlessly, express PIWI proteins and piRNAs (Reddien et al., 2005), and were recently shown, like germ cells, to use piRNAs to preserve the integrity of their genomic heritage (Zhou et al., 2015).

Asymmetric retention or erasure of cellular memory, after cell division, is an important and well-studied mechanism in development, crucial both for renewal of pluripotency/proliferation, and for differentiation and establishment of cell fate (Armakolas et al., 2010; Di Laurenzio et al., 1996; Jan and Jan, 1998; Klar, 1987). Asymmetric cell division (in neurons and other cell types) is also used as a mechanism for preventing aggregated, damaged or misfolded proteins from being inherited to the cell progeny by confining them to only one daughter-cell (Ogrodnik et al., 2014). A similar phenomenon is familiar in budding yeast, where asymmetric division results in two daughter-cells; one of them contains large amounts of unfolded and aggregated proteins, usually associated with aging, while the other remains ‘young’ (Spokoini et al., 2012).

Similarly, asymmetric fission of an entire multicellular organism, such as planaria, could result in asymmetric inheritance of cells which, in theory, could have distinct expression patterns maintained by cell-specific epigenetic states. Could the uneven inheritance of epigenetic effects make the organisms that develop from the two separate fragments phenotypically unequal?

Indeed, planarian ‘clones’ that regenerate from fragments of a single animal and that live in the same container, can show variable responses to an external perturbation such as a pharmacological compound (Beane et al., 2011; Chan et al., 2014; Oviedo et al., 2010). At the molecular level, fission and the ensuing recreation of a new individual in planarians may not necessarily entail complete ‘resetting’ of modifications (such as histone marks, RNA content and synaptic connection strengths) that were acquired by the previous ‘generation’. Asymmetric fission could therefore be a mechanism that enables retention of life history memories; some epigenetic changes, specific to the tail or head sections, may persist, at least in the tissues that were not regenerated anew. As a result of these retained memories of the ancestor's gene activity, the resulting individuals might respond differentially to changes in the environment in the future. If indeed epigenetic marks are asymmetrically distributed, whether through a passive/random process, or via active mechanisms (similarly to the mechanisms that asymmetrically distribute aggregated proteins in dividing neurons or yeast, that were described above), then we suggest that the clonality of the resulting individuals should be questioned, and that the evolution of the species could be affected.

Therefore, are all clones created equal, or could epigenetic information survive splitting? The answer depends on the capacity of asymmetric fission to maintain long-term variability – the ability of each cloning product (each ‘individual’) to hold memories acquired by their ancestral body (or the relevant part thereof) in its lifetime. A few different mechanisms, which are not mutually exclusive, and could operate in tandem, could in theory establish asymmetry following planarian fission.

Since the new individual is regenerated from a ‘community of cells’ and not from one unique cell, asymmetric fission could non-randomly distribute genetically distinct neoblasts to the two fragments. The asymmetry in this regard may not be entirely random; genetic variability could be caused by differential mutation rates in different tissues of the body; it was suggested that neurons, for example, display more genetic variability (Muotri and Gage, 2006).

In theory, since the different genomes are packed into different cells, which do not fuse, the genetics of planaria that reproduce by fission could be dictated by the frequencies of multiple non-recombining alleles that are present within a single organism. This possible mosaicism also has practical considerations for planaria geneticists. Since each worm is created from multiple ‘germline-like’ neoblasts, genetic editing of an entire worm's genome (by CRISPR for instance) would require manipulation of all the neoblasts' genomes, or highly efficient selection of those neoblasts which were successfully edited; otherwise, only a mosaic animal would be achieved. Indeed, a recent study reveals that genetic mosaicism in planarian cells can create genetic diversity in a population of asexually reproducing animals (Nishimura et al., 2015).