Perspective Modeling Regeneration: A Primer for Reverse- Engineering the Worm

Daniel Lobo, Wendy S. Beane, Michael Levin* Center for Regenerative and Developmental Biology, and Department of Biology, Tufts University, Medford, Massachusetts, United States of America

chines and fault-tolerant, robust systems. Abstract: A mechanistic under- the cellular mechanisms that are A medical treatment that would enable a standing of robust self-assembly and thought to underlie them. By es- person to regenerate a completely new repair capabilities of complex systems tablishing an engineering-like style head, or a robotic system that could would have enormous implications for reviews of the molecular devel- automatically recover its proper structure for basic evolutionary developmental opmental biology of biomedically and function after losing more than 99% biology as well as for transformative important model systems, signifi- of its constitutive parts, is still only a applications in regenerative biomed- cant fresh insights and quantitative dream. However, there does exist a icine and the engineering of highly computational models will be de- natural system capable of performing these fault-tolerant cybernetic systems. Mo- veloped by new collaborations amazing feats: the planaria. lecular biologists are working to between biology and the informa- are nonparasitic identify the pathways underlying the tion sciences. remarkable regenerative abilities of that have bilateral symmetry, a true brain model that perfectly regener- driving a complex behavioral repertoire ate limbs, brains, and other complex [2], and an extraordinary capacity to body parts. However, a profound regenerate due to the presence of a large disconnect remains between the adult stem cell population [3]. Individual deluge of high-resolution genetic Possibly the people who are trying to planarians are practically immortal—able and protein data on pathways re- discover how to set up a computer to to regenerate aging, as well as severely quired for regeneration, and the learn to play good chess, or bridge, damaged or lost, tissues [4]. A trunk desired spatial, algorithmic models are among those most likely to make fragment cut from the middle of an adult that show how self-monitoring and a major contribution to the funda- planarian will regenerate into a whole growth control arise from the synthe- worm, always growing a new head and sis of cellular activities. This barrier to mental theory of evolution. — C. H. Waddington [1] new tail in the same orientation as the progress in the understanding of original worm. As little as 1/279th of a morphogenetic controls may be planarian [5], or a fragment with as few as breached by powerful techniques 10,000 cells [6], can regenerate into a new from the computational sciences— Introduction using non-traditional modeling ap- worm within 1–2 weeks. Planaria are a proaches to reverse-engineer systems The ability to control the pattern popular model for molecular-genetic and such as planaria: flatworms with a formation of organs and appendages is a biophysical dissection of pathways that complex bodyplan and nervous sys- key aim of regenerative medicine. Trans- underlie regenerative patterning [4,7,8], tem that are able to regenerate any formative impact in areas such as birth having more genes in common with body part after traumatic injury. defects, traumatic injury, cancer, and humans than with the fruit fly Drosophila. Currently, the involvement of experts degenerative disease requires that we A mechanistic understanding of the com- from outside of molecular genetics is understand the molecular mechanisms munication and control networks that hampered by the specialist literature that allow living beings to detect and maintain complex shape against radical of molecular developmental biology: repair damage to complex biological perturbations will revolutionize our ability impactful collaborations across such structures. A similar goal is pursued by to regulate stem cell behavior in the different fields require that review engineers seeking to build resilient ma- context of the host organism. Thus, literature be available that presents the key functional capabilities of important biological model systems Citation: Lobo D, Beane WS, Levin M (2012) Modeling Planarian Regeneration: A Primer for Reverse- while abstracting away from the Engineering the Worm. PLoS Comput Biol 8(4): e1002481. doi:10.1371/journal.pcbi.1002481 often irrelevant and confusing details Editor: Takashi Gojobori, National Institute of Genetics, Japan of specific genes and proteins. To Published April 26, 2012 facilitate modeling efforts by comput- Copyright: ß 2012 Lobo et al. This is an open-access article distributed under the terms of the Creative er scientists, physicists, engineers, and Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, mathematicians, we present a differ- provided the original author and source are credited. ent kind of review of planarian Funding: This work was supported by the Telemedicine and Advanced Technology Research Center (TATRC) regeneration. Focusing on the main at the U.S. Army Medical Research and Materiel Command (USAMRMC) through award W81XWH-10-2-0058. ML patterning properties of this system, is also grateful for the support of the G. Harold and Leila Y. Mathers Charitable Foundation, and NSF CDI grant - EF-1124651. WSB gratefully acknowledges the support of the NIH Kirschstein-NRSA grant F32 GM08354. The we review what is known about the funders had no role in the preparation of the manuscript. signal exchanges that occur during regenerative repair in planaria and Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

PLoS Computational Biology | www.ploscompbiol.org 1 April 2012 | Volume 8 | Issue 4 | e1002481 reverse-engineering the remarkable system processes [14,15] (for excellent introduc- In this review, we hope to close the gap that is planarian regeneration would have tions to biological pattern formation mod- between regenerative biology and the profound impacts on regenerative medi- eling see [16–19]). Unlike bare gene or fields of mathematics, computer science, cine, bioengineering, synthetic biology, protein networks, such models are con- and engineering and lower the barrier for and robotics. structive in the sense that they make experts from the information and systems Regeneration in planarians involves a explicit the events that need to occur to engineering sciences to apply their knowl- truly complex interaction of several sys- create a specific shape. Only a handful of edge to unraveling the mechanisms of tems at the organismal level. After an algorithmic models have been proposed large-scale regeneration. Here we provide injury, the stem cells in the worm over the years to explain regeneration in an overview of the planarian regeneration proliferate and migrate to form a protec- planarians [20–23] (see ‘‘Existing Models’’ system, explain what is known about the tive mass of new cells (blastema) at the section below and Supplemental Text S1), signaling mechanisms, summarize the wound site. This cell proliferation is tightly and none of them successfully integrate proposed partial models in the literature, coordinated with the selective destruction more than one or two key features of and frame the specific issues that must be of some old cells (apoptosis), effectively regeneration. addressed to bring the power of interdis- remodeling both the new and old tissues to There is a gap between the success of ciplinary investigation to fruition. Our recreate exactly those regions and organs high-resolution genetic analysis and the goal is to present the basic features of this the worm is missing, adjust the propor- needed level of insight into systems-level system from an engineering perspective to tions of the remaining regions and organs mechanisms that enable adaptive control facilitate modeling approaches [35–38]. If to the new smaller worm size, and of pattern formation. A fresh set of ideas the modeling and engineering communi- maintain the original patterning orienta- may be helpful, from areas of science that ties can be engaged to produce algorith- tion of the worm with the new tissues. have developed techniques for reverse- mic models that can accurately explain These complex interactions are controlled engineering complex systems, utilizing the regeneration process, the application by a diverse set of signals, including analytical methods and types of models of biologically inspired computational molecular pathways, gap junctional com- that are distinct from those familiar to ideas will feed back into biology and aid munication, ion fluxes, and nervous sys- most cell biologists today. Construction of our understanding of complex biological tem signals. Although essential for regen- in silico implementations is especially systems [39]. Conversely, the insights eration, the mechanisms by which these crucial; for any but the most trivial set of gained from the construction and appli- signals integrate to maintain and restore relationships among subunits, running a cation of these regenerative models will the correct geometry of the are still simulation on a computer is the only way equally benefit computer science, artificial not well understood. to determine the predictions of a given life, robotics, and many areas of engi- After more than 100 years of research, system of rules, ascertain the model’s neering. Moreover, we hope this review no single model has been proposed that quality of fit to the known data, derive will have the broader impact of establish- explains comprehensively the mechanisms testable predictions for driving real exper- ing a precedent for much-needed different of all the known components of planarian iments, and determine which manipula- kinds of reviews that lower the barrier for regeneration; the majority of current tions can give rise to desired patterning true interdisciplinary cross-fertilization. models are descriptive in nature and outcomes. Planaria constitute an excellent test limited to only one or two observed To facilitate the application of engi- case with which to explore this type of properties [9–12]. Current research efforts neering and information sciences to this approach. capitalize on molecular and cell biology fascinating problem [24–26], experts out- techniques to produce an ever-increasing side of molecular and developmental The Building Blocks for set of detailed data on genetic components biology need to become aware of the basic Modeling Planaria that are necessary for normal regeneration capabilities of the planarian model system Basic Anatomy and Physiology [13]. However, making use of such and the current state of knowledge about Several species are used for research; information for biomedical or engineering the control mechanisms involved. The first Figure 1 summarizes the basic anatomy of purposes requires the integration of pro- reviews to highlight the remarkable regen- Schmidtea mediterranea planaria and outlines tein or gene networks into constructive erative capacity of planaria were mainly their major anatomical axes. Planaria models that are sufficient to predict and descriptive collections reporting on various possess an intestine (gastrovascular tract), explain geometry of tissues and organ cutting experiments [27,28]. Later, func- a body-wall musculature, a well-differenti- systems, and reveal what changes must be tional experiments were also described, ated nervous system (including brain) with made in specific signals to drive necessary including starvation, transplantations, ir- most of the same neurotransmitters as alterations of tissue topology. If we hope to radiation, and pharmacological exposures humans, three tissue layers (endoderm, understand and tame powerful regenera- [29,30]. Given the revolution in available ectoderm, and mesoderm), and bilateral tive mechanisms, we will need to develop molecular methods, the most recent re- symmetry [3]. The gastrovascular tract algorithmic models that are consistent views have superbly summarized the consists of a highly branched gut spread with the existing experimental datasets genetics of regeneration [4,10,31–34], throughout the entire body, with a single but also bridge the gap between functional detailing the growing number of gene ventral opening from which a long mus- genetic data and self-assembly of three- products whose experimental inhibition cular tube (the pharynx) both takes in food dimensional shape and dynamic morphos- results in various kinds of regenerative and expels wastes [40]. The central tasis. Algorithmic (also called mechanistic failures. Unfortunately, these reviews are nervous system is comprised of a bi-lobed or computational) models, in contrast to largely unusable by computer scientists or cephalic ganglia (the brain) connected to descriptive ones, explain precisely at every engineers, as the molecular details of two ventral nerve cords that run longitu- step what information a system needs and pathways and protein–protein interactions dinally throughout the animal and fuse in what logical steps should be performed, obscure the main features and control the tail [41]. Planarians possess a diverse i.e., what algorithm governs the observed functions to be modeled. set of sensory receptors that can detect

PLoS Computational Biology | www.ploscompbiol.org 2 April 2012 | Volume 8 | Issue 4 | e1002481 gy is restored by a tightly regulated combination of new tissue growth and selective loss of old tissues, producing a new worm that has all its parts in the correct proportion for its now smaller size [59]. Planarians also use this extraordinary regenerative ability to reproduce asexual- ly. Through the process of transverse fissioning, planarians anchor their tails and essentially pull themselves apart, resulting in two fragments (one head and one tail) that will regenerate into two genetically identical worms [3].

Signaling Mechanisms Regeneration in planarians has been shown to be the result of carefully orchestrated communication both within the blastema and between the newly regenerating tissues and the old existing ones [8,60,61]. However, the exact nature of this communication is not well under- stood and is the focus of a majority of planarian experiments, which together support four main types of signaling mechanisms: cell signaling pathways, gap junctional communication, ion fluxes, and nervous system signals. In planarians, as in other multi-cellular organisms, classical cell signaling pathways Figure 1. Planarian anatomy and body axes. (A) Dorsal side of the planarian Schmidtea exist that allow cells to communicate with mediterranea.(B) Planarian diagram showing the brain lobes, nerve cords, and secondary nerves (green); the two eyes (black and white); the gastrovascular tract (gray); and the pharynx (light each other [10]. Diffusing chemical factors brown). (C) The three main axes of the planarian anatomy: anterior-posterior (AP), dorsal-ventral act as messenger molecules, triggering a (DV), and medial-lateral (ML). specific cascade of events either in the doi:10.1371/journal.pcbi.1002481.g001 same cell or, more often, in neighboring cells. Typically, developmental signaling pathways involve the secretion of a light [42,43], chemical gradients [44,45], processes together restore the normal molecule (morphogen) from one cell that vibration [46], electric fields [47], mag- morphology of the worm. binds to its specific receptor on a second netic fields [48,49], and even weak c– Wound closure is facilitated by muscle cell (which can be located quite distant radiation [50]. At least 20–30 different contraction [55], but the molecular trigger from the originating cell). Sometimes, the types of planarian cells have been de- for this reaction is unknown. However, same morphogen is dispersed throughout scribed [51,52], including broadly distrib- migration of planarian epithelial cells to the animal in a concentration gradient, uted pluripotent adult stem cells called the wound site is known to be an essential where different concentration ranges of neoblasts that constitute 20%–30% of the component for wound closure [56], as is the same morphogen produce different total planarian cell population [53]. Neo- the juxtaposition of dorsal and ventral outcomes. This type of signaling is wide- blasts are the only proliferative cells in the tissues [57]. Wound closure is followed by spread and used in both short-range and body, with the ability to differentiate into an initial body-wide peak of cell division long-range cell communication for basic any other planarian cell type, and are a (mitosis) from the neoblasts within 6 h cellular activities such as rearranging the key component of the planarian’s ability to after injury; if tissue was lost, this is cytoskeleton, changing gene expression, regenerate [54]. accompanied by the migration of neo- and global tissue patterning during em- blasts towards the wound by 18 h and a bryogenesis. Regeneration Primer second, local mitotic peak at the injury site Alternatively, cells can communicate Planarians have the remarkable ability to around 48–72 h after injury to promote with their immediate neighbors through regenerate an entire worm from a fragment the formation of the blastema [53]. the direct exchange of cytoplasm in a that may lack any brain, central nerve Conversely, these proliferative spikes are process known as gap junctional commu- cords, or pharynx. Regeneration is com- counterbalanced by an initial local in- nication (GJC) [62]. Gap junctions are pleted through (1) closure of the wound crease of cell death (apoptosis) at the membrane channels that dynamically within the first 30–45 min, (2) formation of wound site within 1–4 h after injury, allow for the direct transmission of small a mass of new cells (called the blastema) at followed by a second, systemic apoptotic molecules and ions between cells; gap the injury site, which is visible by 48–72 h, increase throughout the body that peaks at junctions are passive channels that can and (3) re-patterning of both the old and the about 72 h after injury as old tissues are control the amount and type of small new tissues over the next 1–2 weeks. These remodeled [58]. Thus, normal morpholo- molecules that pass through. Hence, GJC

PLoS Computational Biology | www.ploscompbiol.org 3 April 2012 | Volume 8 | Issue 4 | e1002481 enables regulated quick bursts of commu- anterior tissues in a fragment following tissue loss results in neoblast migration nication, permitting synchronization amputation [8]. It is clear that a compre- to the wound site and a second mitotic among nearby cells, while inhibition of hensive picture of regenerative regulation peak at the wound resulting in blastema GJC can create regions of isolation often will include the integration and coordina- formation [9]. The signal that causes needed for morphogenesis [62]. The tion of all of these overlapping signaling neoblast cells to stop proliferating and movement of charged substances through mechanisms. start differentiating within the blastema GJC-connected cells can be driven by requires the activation of the extracellular electrophoretic forces [63,64]. Further- Planarian Experiments: The signal-related kinase (ERK), since phar- more, voltage gradients can be transmitted Current Dataset macological experiments blocking ERK and altered through gap junctions, such result in blastema formation without that cells are able to sense the membrane A large amount of functional data has neoblast differentiation [78]. Finally, c- potential of neighboring cells. In planaria, been generated over the last century that Jun N-terminal kinase (JNK) signaling is about a dozen gap junction genes (the remains to be integrated into models that required for blastema formation [79], innexin family) have been found [65,66], explain the generation and repair of while epidermal growth factor (EGF) and various combinations create junctions complex structures during planarian re- receptors have been shown to regulate that selectively allow for different degrees generation. Before the emergence of the differentiation of several cell types of communication among cells, such as modern molecular tools, the most basic during regeneration [80]. that between neoblasts and their differen- of regeneration experiments involved cut- tiated neighbors [66]. ting assays, in which a section of the How Polarity Is Established While ion flux is more commonly planarian was amputated. Almost all Polarity refers to an asymmetric distri- associated with nerve conduction, all cells conceivable types of amputations are bution of a particular property [81]. generate and receive bioelectrical signals documented in the classical literature Planarians have several polarities along through channels and pumps within their [30], with most of them resulting in the their body axes, with anterior-posterior membranes, producing ion currents regeneration of a complete worm. The (AP, or head versus tail polarity) and which, in turn, create voltage gradients majority of current planarian experiments dorsal-ventral (DV, or top versus bottom [67]. Recent experiments have begun to consist of inhibiting or silencing the polarity) being the most prominent in the identify these long-term, steady-state expression of a protein encoded by a literature. Remarkably, this polarity is membrane voltage gradients and ion specific gene through pharmacology or somehow maintained even when anatom- flows, the genetic networks that regulate RNA interference (RNAi) and using the ical cues, such as the brain and pharynx, them, and the mechanisms that transduce results to determine the regeneration are removed. For instance, a worm trunk electrical signals into changes in patterning mechanism in which that gene is involved fragment generated by removing both the and morphogenesis [68]. Thus, bioelectric [76,77]. head and tail will always re-grow its head signals (often transmitted by GJC) and Any useful model of planarian regen- in the same orientation as the original signaling pathways interact, and together eration must exhibit the behavior ob- worm (never producing a head in the they regulate regenerative outgrowth, cell served in these functional experiments. direction of the original tail), thus main- proliferation, differentiation, and migra- To facilitate efforts at constructing such taining its AP polarity (Figure 3). tion [69]. Recent work has begun to models, in this section we highlight a Although regeneration usually results in elucidate the bioelectric signaling that selection of the main types of experiments the formation of a whole worm regardless carries patterning information during pla- found in the literature (Figure 2). For of the injury that initiated it (Figure 2A), narian regeneration, including membrane convenience, we organized the experi- there are exceptions. Very thin cut voltage gradients, and fluxes of hydrogen, ments into several broad categories of the fragments sometimes regenerate with sym- potassium, and calcium [70–72]. Investi- kind of regenerative questions these metrical double heads, although never gations in multiple model species have experiments attempt to answer. This is double tails [5]. Also, really long worms shown that several cellular mechanisms not to suggest that the mechanisms in sometime spontaneously produce double can transduce such electrical signals into planarians are independent or modular- heads after fissioning [82]. Such failures intracellular responses, including: control ized in this way. Indeed, computational represent excellent opportunities for dis- of Ca2+ flux and calcium-mediated genetic models that do not have any pre-deter- secting the molecular mechanisms that regulation [73], phosphorylation of key mined organizational bias are more likely drive planarian regeneration as well as regulatory enzymes induced by electric to uncover significant, novel regenerative guiding the choice of signaling modes used fields [74], subcellular translocation of mechanisms. for modeling this system. transcription factors due to depolarization Several signaling pathways are involved [75], modulation of voltage-sensitive trans- How Regeneration Is Initiated in the regulation of planarian AP polarity. porters, redistribution of membrane re- Following injury, a cascade of signaling The Wnt/b-catenin pathway, comprising ceptors, electrophoresis of signaling mole- events is triggered that initiates the a large number of regulatory proteins and cules such as morphogens, and activation regeneration process. Experiments using signaling molecules, is essential for the of signals by voltage-induced conforma- markers for both neoblasts and their formation of the primary body axis in most tional changes in membrane proteins [67]. mitotic activity have provided evidence [83]. This pathway is known to be Finally, the nervous system itself may for the existence of two different signaling necessary for posterior polarity (tail for- mediate instructive signaling during pla- events, which distinguish between simple mation) during regeneration in planarians, narian regeneration [30]. Recently, it has wounding (resulting in wound healing and its perturbation causes head regener- been shown that the planarian’s ventral alone) and the loss of tissue (requiring ation at every wound regardless of the nerve cords have the capacity to transmit regeneration) [9]. After wounding, an original polarity [84,85]. The Wnt/b- long-range information to a wound site increase in neoblast mitoses occurs catenin pathway in planarians is in turn regarding the presence or absence of throughout the animal; however, only regulated by the hedgehog (Hh) pathway,

PLoS Computational Biology | www.ploscompbiol.org 4 April 2012 | Volume 8 | Issue 4 | e1002481 PLoS Computational Biology | www.ploscompbiol.org 5 April 2012 | Volume 8 | Issue 4 | e1002481 Figure 2. Diagrams of the main planarian regeneration experiments found in the literature. (A) Cutting experiments amputate part of the planarian body (shadowed); normally, a complete regenerated worm results within 1–2 weeks. (B) Transplantation of diverse parts also regenerates into a complete worm. (C) Planarians degrowth when starved; they restore their original size upon feeding. (D) Octanol blocks gap junction communication between the worm cells; a trunk fragment treated with octanol regenerates into a double-headed worm. (E) A post- pharyngeal fragment treated with octanol and with the nerve cords partially amputated regenerates into a quadruple-headed worm. (F) An external electric field applied to a trunk fragment disturbs AP polarity during regeneration when the anode is located in the head wound; low-intensity currents cause double-headed worms, whereas high-intensity currents cause reversed-polarity worms. (G) The drugs ivermectin (IVM) and SCH-28080 (SCH) disturb the ion pumps in the worm cells, altering their membrane voltage. IVM causes cell depolarization (more positive) and trunk fragments to regenerate into double-headed worms; SCH causes cell hyperpolarization (more negative) and trunk fragments to regenerate into worms with no heads. doi:10.1371/journal.pcbi.1002481.g002 which also is required for posterior Bioelectric signals also regulate planar- macological experiments targeting endog- polarity and when blocked similarly results ian AP polarity. A series of classical enous ion channels and pumps in worm in only head regeneration [86,87]. Con- experiments showed that applying exter- cells revealed a membrane voltage signal- versely, both the b-catenin destruction nal electric fields to regenerating trunk ing pathway that is required early for the complex member adenomatous polyposis fragments can result in AP polarity regeneration of heads [70]. Membrane coli (APC) and the Wnt/b-catenin path- reversals [92,93] (Figure 2F). Regenera- depolarization of the blastema (endoge- way inhibitor notum are required for tion proceeded normally when the ante- nously regulated by hydrogen and potas- anterior polarity (head formation), and rior cut faced the cathode (negative), sium flux through membrane ion translo- their loss leads to the regeneration of tails while double-headed worms were pro- cator proteins) is required for head only [88,89]. Loss of axin genes (other duced when the anterior cut faced the regeneration; the data suggest that a negative regulators of the Wnt/b-catenin anode (positive). With the application of voltage-mediated influx of calcium into pathway), or the exogenous administration higher current densities, there occurred the blastema triggers anterior gene ex- of retinoic acid (a small molecule that is the regeneration of morphologically nor- pression [70,72]. In fact, forced depolar- also important for vertebrate AP pattern- mal worms but whose AP polarity was ization of a blastema is sufficient to cause ing), are also required for planarian completely reversed compared to the head regeneration even at posterior anterior regeneration [90,91]. original fragment. More recently, phar- wounds (Figure 2G) [70].

Figure 3. Planaria restore their AP polarity similarly to bar magnets. (A) A bar magnet restores the original polarity after being cut transversally. (B) Similarly, after bisecting a worm, polarity is restored correctly in each fragment. Note that even though cells on either side of the amputation plane were direct neighbors before the cut, the ones facing posterior make a completely different structure (a tail) than the ones facing anterior (a head). The drastically different fates of cells with essentially the same positional information suggest models based on non-local control of orientation [8]. doi:10.1371/journal.pcbi.1002481.g003

PLoS Computational Biology | www.ploscompbiol.org 6 April 2012 | Volume 8 | Issue 4 | e1002481 There is evidence that the organization cells and restricts the expression of the slit neoblasts and the appearance of abnormal of the cellular cytoskeleton may also play a gene to the midline; loss of Wnt5 results in outgrowths reminiscent of tumors [111]. role in AP polarity, as inhibition of the expression of slit beyond the midline, In mammals, loss of PTEN activates the microtubule organization (using the inhib- disrupting ML polarity and leading to target of rapamycin (TOR) pathway, itor colchicine) induces bipolar head regeneration phenotypes such as multiple, resulting in tumor formation [112]. Phar- phenotypes in regenerating planarian ectopic pharynxes [103,104]. Interestingly, macological inhibition of TOR (with trunk fragments [94,95]. Gap junctional as in vertebrates, in which the dorsal- rapamycin) in planarians rescues loss of communication has also been demonstrat- ventral and left-right axes are linked, in PTEN, preventing abnormal outgrowths ed to be necessary for AP polarity in planaria DV polarity also appears to affect and restoring regenerative ability [111]. planaria. Both pharmacologically (through ML polarity. For instance, as with Wnt5, Understanding the factors that link cell- exposure to long-chain alcohols such as loss of ADMP (which regulates BMP level regulation of neoblast proliferation octanol, which inhibit GJC) and geneti- signaling) also results in the lateral expan- and differentiation to system-level needs cally (using RNAi to abrogate expression sion of slit expression [11]. Finally, it should such as polarity and patterning is a crucial of planarian innexin gap junction proteins), be noted that planaria are not quite next step for the field. it has been shown that isolating cells from symmetric about the left-right axis: al- The planarian nervous system contains gap junction-based communication with though the extent of consistent asymmetry a full set of vertebrate neurotransmitter other regions of the worm leads to and its underlying mechanisms are com- homologs, possessing among others dopa- inappropriate generation of secondary pletely unknown, the left eye has a minergic, serotonergic, cholinergic, and heads [8,65,66] (Figure 2D). The ventral significantly better capacity to regenerate GABAergic neurons [113–116]. Neural nerve cords seem to similarly transmit under pharmacological perturbation of connectivity is mediated in part by the information along the planarian AP axis eye-relevant ion currents following head disheveled family of proteins, a Wnt signaling during regeneration [8] (Figure 2E). Thus, amputation [71]. pathway member, most likely Wnt5 sig- like Wnt/b-catenin and Hh signaling, naling [117]. Interestingly, several of the GJC- and neural-mediated signaling ap- How Tissue Identity Is Determined ML polarity regulators are also involved in pear to be equally necessary for blastema A central component of most planarian the regulation of nervous tissue organiza- cells to determine the needed identity of regeneration studies is the question of how tion and when inhibited result in neural the structures they assemble. It is tempting cell type and tissue identity are specified in patterning defects in which nerves collapse to hypothesize that both of these systems each location [105]. Although the ultimate towards the midline (slit inhibition [102]) underlie the long-range information ex- goal is to understand how every planarian or, conversely, displace laterally (Wnt5 change between existing tissues and sites of tissue and cell type is regenerated and inhibition [117]). The fibroblast growth active morphogenesis that is needed for maintained, here we concentrate on the factor (FGF) pathway, important in verte- the regeneration of the needed structures, most widely studied tissue types into which brate neural formation and patterning, and only those structures. neoblasts differentiate [106]. Such studies also participates in planarian brain regen- The establishment and maintenance of are aided by the fact that the stem cell eration. Loss of function of the gene nou- the DV axis during regeneration in planar- population can be selectively killed by darake, a component of the FGF pathway ians is regulated by the secreted bone irradiation, which prevents planarians specifically expressed in the head region, morphogenetic protein (BMP) pathway. from regenerating [30]. This method has leads to the expansion of the brain through While in vertebrates BMP is expressed on been used to show that a single trans- the body [118]. Finally, the netrin family the ventral side, in invertebrates BMP is planted neoblast can rescue the regenera- of axon guidance proteins is also required expressed dorsally [96,97]. In planaria, tive capacity of irradiated planarians, as in planarians for the regeneration and BMP signaling drives dorsal fates, while well as induce the production of gonads in maintenance of neural patterning; when BMP inhibition results in ventralized pla- asexual hosts [54]. Irradiation experiments the netrin receptor is inhibited the overall narians [98–100]. Similarly, silencing of have also been used to elucidate molecular organization of the nervous system is lost, noggin genes, which are inhibitors of BMP identifiers of neoblasts. The piwi family of disrupting the relationship between the signaling, results in dorsalized planarians regulatory genes, known to be essential for brain and VNCs [119]. [101]. Also important for DV polarity is the maintaining stem cell populations by Graft transplantation experiments have antidorsalizing morphogenetic protein preventing cell differentiation, are ex- historically been used to probe the func- (ADMP), which promotes, but also is pressed solely in neoblasts and are re- tional identity of different regions and inhibited by, BMP; they are thought to quired for regeneration [107–109]. Bio- tissues in planaria and have provided create a BMP/ADMP regulatory circuit to logical markers that distinguish both early important information about the types of control DV polarity [11]. Thus ADMP and late neoblast prodigy (descendants signaling that occur between different ensures that BMP continues to be expressed, that will differentiate into tissue-specific tissues (Figure 2B). For example, when a while BMP expression subsequently turns off cells) have also been identified [110]. small piece is cut out, flipped along the ADMP expression, producing two non- GJC is also important for neoblast DV axis, and grafted back into its original overlapping domains of ADMP (ventral) survival, as inhibition of the gap junction location in the worm, a cup-shaped and BMP (dorsal) that establish the DV axis. protein innexin-11 results in a loss of both projection is formed on the boundary Animals with bilateral symmetry, includ- piwi-positive cells and of the ability to between the host and the graft; for ing planarians, also have medial-lateral regenerate [66]. Additionally, the phos- anterior grafts this projection will develop (ML) polarity. A gene from the slit family phatase and tensin homolog (PTEN) protein a head-like morphology, while posterior is expressed along the planarian midline; (a known tumor suppressor) and its projections will appear tail-like [21]. This when knocked down, ML polarity collapses respective pathway have been shown to suggests that the juxtaposition of dorsal and and gives rise to phenotypes such as the regulate neoblast activity in planarians. ventral tissues is a cue that signals the regeneration of a single, cyclopic eye [102]. Disruption of planarian PTEN signaling regenerative process. In contrast, if two The protein Wnt5 is secreted from lateral results in both the hyperproliferation of dorsal halves are grafted together, no

PLoS Computational Biology | www.ploscompbiol.org 7 April 2012 | Volume 8 | Issue 4 | e1002481 regeneration occurs [22]. Such results have (e.g., AP polarity) the field lacks a truly Many algorithmic models propose the driven many models in vertebrate systems cohesive model that explains the system- existence of a coordinate system that in which growth is dictated by juxtaposition level (patterning) data in terms of the facilitates the patterning of the organism. of regions with distinct ‘‘positional infor- behavior of lower-level (cell pathway) The positional information model [125] is mation values’’ [120]. Similarly, if the components. based on diffusible substances that create pharyngeal region is removed and the concentration gradients; depending on the remaining head and tail fragments are then Existing Models of Planarian specific concentration of such substances joined together, a new pharynx regenerates Regeneration at different locations, different cell pro- between them [121]; however, when two grams are triggered [126]. The serial fragments from the same region of the Most modern planarian regeneration threshold model [20] combines positional worm are grafted together, no regeneration studies result in a gene regulatory network information and cell migration to explain is observed [21]. Finally, if a small fragment or protein interaction pathway. However, how planarians can restore AP polarity. of the head is grafted into the tail region, a these networks largely do not constrain Reaction-diffusion models [127] are based new head and pharynx will appear with shape or shape regulation, and such on diffusible substances that react with reversed polarity from the original AP axis; pathways are compatible with a very each other; they can generate most of the conversely, if a tail piece is grafted anterior broad range of morphologies. A few patterns found in biology [81] and explain to the pharynx, a second pharynx will form researchers have proposed algorithmic regeneration of polarity in planaria [128]. between the head and the transplanted models, which precisely define the steps Other algorithmic models proposed for tissue, again with reversed AP polarity that cells (or tissues) must take to assemble planarian regeneration are based on [60,61]. These results clearly indicate that, or repair a given morphology. In specify- bioelectrical signals [23], dorsal-ventral instead of adopting the identity of the host ing the information needed to make interactions [21], and the intercalation of tissue, a grafted piece can signal the host to decisions, such models could be fleshed missing regions [129]. re-pattern and change tissue identities. out in molecular terms to provide testable However, none of the proposed models During everyday life, planarian cells are hypotheses about the mechanisms under- in the literature are able to explain more continuously replaced by the differentiating lying information exchange, computation, than one or two facets of planarian progeny of the neoblast population in a and links to the ultimate execution of regeneration. There is a tremendous process known as morphostasis [4]. The morphogenesis via cellular effectors. Due opportunity to integrate existing knowl- mechanism is not well understood, al- to space constraints, the basic features edge of pathways with the anatomical and though it is known that mitogen-activated (assumptions and outputs) are given in physiological data to produce novel mod- protein kinase (MAPK) signaling is re- Supplemental Text S1. els of the worm as a system that senses quired for neoblasts to stop proliferation and undergo differentiation [78]. This homeostatic process of cell turnover, which Box 1. Key Functional Questions Yet to Be Solved constantly renews all cells without changing tissue size or proportion [55,122], is Detection of missing tissues: What is the signaling mechanism that triggers common among organisms. However, the regeneration of only the exact missing regions, structures, and organs? planaria exhibit a unique remodeling Growing boundaries delimitation: What signals trigger growth and cause ability that enables them to dynamically remodeling to cease when regeneration is completed? scale their body in relation to the available cell number. Upon starvation, planaria use Plasticity of organism size: How are planarians able to remodel existing their own cells for energy, reducing their tissues to maintain the correct internal organ proportions as size changes due to body size in a phenomenon known as available cell number? degrowth [3]; this is reversible, and growth resumes upon feeding (Figure 2C). The Neoblast migration: What are the driving forces that permit neoblasts to worm’s total cell number changes linearly migrate in the direction of the wound to initiate blastema formation and how are with respect to body length and time, while positional values encoded (and to what spatial resolution)? keeping the proportion of all cell types Specification of cell types: How are different cell types generated from the constant. This requires active remodeling same neoblast population, and how are anatomical regions of different size (head of existing tissues to scale the structure in a vs. eye vs. photoreceptor cell) specified spatially? process known as morphallaxis and is quantitatively described by an allometric Organization of cells into organs: Once cells are committed to a given cell equation [52,123]. The mechanisms regu- type, what is the process that orchestrates their development into specific lating this exquisite rearrangement are not organs? well understood, but it is accomplished in part through coordinated cell death and Small vs. large fragments paradox: Experimentally suppressing essential neoblast proliferation [58,124]. mechanisms (such as anterior-posterior positional gradients) in large fragments This brief overview of some of the (for instance middle-third trunk fragments), disturbs normal regenerative questions that the planarian regeneration patterning; so why are small pieces cut from a worm often still able to regenerate field has investigated, both historically and normally, despite such loss-of-function treatments, when they have suffered currently, highlights the increasing wealth relatively more damage? of information that continues to be Specification of target morphology: What is the mechanism (whether generated and must be integrated into directly encoded or an emergent property of the remaining tissue) that specifies any comprehensive model of planarian target morphology during regeneration? regeneration. Even within a narrow area

PLoS Computational Biology | www.ploscompbiol.org 8 April 2012 | Volume 8 | Issue 4 | e1002481 damage [130], carries out the distributed that at least basic AP polarity in worms may tions. The concept is highlighted in the processing needed to restore a target state, be directly stored. A transient modulation following hypothetical experiment, illus- and then ceases growth. The major of physiological events in a worm results in trated in Figure 4. Take one planarian outstanding questions that a complete a bipolar 2-headed outcome; remarkably, from each of two species with clearly algorithmic model should be able to although this change did not affect the different morphologies: S. mediterranea with explain are summarized in Box 1, as a DNA sequence, the patterning change a rounded head, and P. felina with a challenge to modelers from the computer persists upon multiple subsequent amputa- hammerhead. In this experiment, the programming, engineering, physics, math- tions in the absence of any other perturba- neoblasts are killed off (by irradiation) in ematics, complex systems science, and tion. The shape to which the animal half of the S. mediterranea worm. Subse- artificial life communities. regenerates upon damage (the target mor- quently, live neoblasts from the P. felina A key question concerns the possible phology) has been stably altered, suggesting worm are transplanted into the irradiated existence of a direct encoding of ‘‘target the possibility that the large-scale axial worm. If this chimeric worm is now cut, morphology’’. It is commonly held that anatomic plan may be encoded in physio- forcing it to regenerate its head, which head shape is an emergent property of cell logical networks and thus directly modifi- shape will be regenerated (round or ham- interactions. However, recent data suggest able by non-genetic experimental interven- mer, or perhaps a hybrid, or perhaps

Figure 4. Hypothetical experiment illustrating the concept of target morphology. (A) S. mediterranea (left, rounded head) and P. felina (right, hammerhead) are planarian species with different head morphology. (B) Half the neoblasts of the rounded-head worm are killed by using irradiation and a lead shield. (C) Half the neoblasts of the hammerhead worm are transplanted to the rounded-head worm. (D) After the neoblasts have diffused, the head of the rounded-head worm is amputated. Without a model of how target morphology is determined, it is impossible to predict what shape will regenerate. doi:10.1371/journal.pcbi.1002481.g004

PLoS Computational Biology | www.ploscompbiol.org 9 April 2012 | Volume 8 | Issue 4 | e1002481 regeneration will never cease as each set of (3) A standardized formal modeling en- tion, and computational methods for auto- neoblasts works to remodel the head)? vironment and representation format. mated model discovery. This approach can Without a theory of how the organism’s Models arising from functional studies be also extended with little effort to other final morphology determines stem cell- could be reported in the primary regenerative model organisms, such as driven processes during planarian regener- literature in a formal way that facili- hydra, axolotl, Xenopus,newt,zebrafish, ation, it is impossible to predict the tated both their comprehension and and mammalian organs (e.g., liver, deer outcome of this experiment. Truly success- implementation, to determine wheth- antlers, etc.). The incorporation of strategies ful algorithmic models, that can be used er they indeed match the patterning from other fields, and the integration of truly predicatively to forward the regeneration behavior that was supposed to be interdisciplinary approaches with the now- field, should be able to answer these types of explained by the genetic pathway accepted molecular genetics-bioinformatics- questions. being elucidated. computational biology efforts will greatly (4) Artificial intelligence tools to augment facilitate fundamental insight into the gen- Summary and Conclusion human ingenuity in proposing testable eral questions of how complex systems (organized on many scales of size) ascertain In addition to testable in silico models models. We are currently inundated [131–134] (or indeed, hardware imple- with functional data to the point that it their shape, alter it to match functional mentations based on swarm intelligence is becoming increasingly difficult to needs, and recover from injury. With the models [35]), there are several other areas come up with models that are consis- ability to truly understand the generation in which information sciences can make a tent with available results. As this and adaptive self-regulation of complex transformative impact on regeneration problem grows, human reasoning morphology will come exciting advances in research. We suggest the urgent need for needs to be assisted by computational evolutionary developmental biology, regen- the development of a bioinformatics be- tools that can infer testable, algorithmic erative medicine, and synthetic biology. yond sequence and regulatory networks— models from databases in which mo- Moreover, the payoffs will extend far a bioinformatics of shape, including: lecular-genetic perturbations are linked beyond biology, contributing significantly to their morphological outcomes. to cybernetics, computer science, dynamical (1) A formalization of patterning out- control theory, robotics, and many areas of comes in model systems. We currently Using the planarian regeneration data engineering that can benefit from under- have no standard formal language in as a proof-of-principle test case, our lab is standing how living systems actually perform which outcomes such as 1-headed vs. pursuing these directions. We hope the the remarkable tricks developed by millions 2-headed worm can be encoded for above primer on the system will of years of evolution. informatics approaches, and precise motivate others with complementary ex- quantitative morphometrics need to pertise to join the molecular regeneration Supporting Information be augmented by symbolic represen- community in attempting to solve the tations that focus on large-scale pat- puzzle of autonomous large-scale pattern Text S1 Previously proposed models of terning changes. formation and repair. Paradigms such as patterning in planarian regeneration. (2) Creation of databases where patterning cellular automata [135–137], formal (DOCX) phenotypes (and the associated manip- grammars [138–142], formal rules ulations that produced them) can be [143,144], neural networks [145], and Acknowledgments Boolean circuits [146,147] are just some stored, queried, and mined. The field is We thank Junji Morokuma and other members currently limited to searching abstracts of the tools that remain to be brought to of the planarian regeneration community for for keywords, and a new investigator in bear on this crucial problem at the center many helpful discussions and the three (anon- this field cannot easily do a search to of developmental biology. ymous) reviewers for very helpful comments and find out if any existing experiment The planarian model organism represents suggestions. produced, for example, a worm with an excellent opportunity for the application three eyes in a row. of the proposed model-building, formaliza-

References 1. Waddington CH (1968) Towards a theoretical 8. Oviedo N, Morokuma J, Walentek P, Kema I, 14. Priami C (2009) Algorithmic systems biology. biology. Nature 218: 525–527. Gu MB, et al. (2010) Long-range neural and gap Commun ACM 52: 80–88. 2. Sarnat HB, Netsky MG (1985) The brain of the junction protein-mediated cues control polarity 15. Fisher J, Henzinger TA (2007) Executable cell planarian as the ancestor of the human brain. during planarian regeneration. Dev Biol 339: biology. Nat Biotechnol 25: 1239–1249. Can J Neurol Sci 12: 296–302. 188–199. 16. Ball P (1999) The self-made tapestry: pattern 3. Reddien P, Sa´nchez Alvarado A (2004) Funda- 9. Wenemoser D, Reddien P (2010) Planarian formation in nature. Oxford; New York: Oxford mentals of planarian regeneration. Annu Rev regeneration involves distinct stem cell responses University Press. 287 p. Cell Dev Biol 20: 725–757. to wounds and tissue absence. Dev Biol 344: 17. Murray JD (2002) Mathematical biology. New 4. Aboobaker AA (2011) Planarian stem cells: a 979–991. York: Springer. simple paradigm for regeneration. Trends Cell 10. Adell T, Cebria` F, Salo´ E (2010) Gradients in 18. Harrison LG Kinetic theory of living pattern: Biol 21: 304–311. planarian regeneration and homeostasis.. Cold Cambridge University Press. 354 p. 5. Morgan T (1898) Experimental studies of the Spring Harb Perspect Biol 2: a000505. 19. Schnell S, Maini P, Newman SA, Newman T, regeneration of Planaria maculata. Dev Genes 11. Gavin˜o MA, Reddien PW (2011) A Bmp/Admp eds (2007) Multiscale modeling of developmen- Evol 7: 364–397. regulatory circuit controls maintenance and tal systems. Elsevier. 604 p. 6. Montgomery JR, Coward SJ (1974) On the regeneration of dorsal-ventral polarity in planar- 20. Slack J (1980) A serial threshold theory of minimal size of a planarian capable of ians. Curr Biol 21: 294–299. regeneration. J Theor Biol 82: 105–140. regeneration. Trans Am Mic Sci 93: 386– 12. Molina MD, Salo E, Cebria F (2011) Organiz- 21. Kato K, Orii H, Watanabe K, Agata K (1999) 391. ing the DV axis during planarian regeneration. The role of dorsoventral interaction in the onset 7. Gentile L, Cebria` F, Bartscherer K (2011) The Commun Integr Biol 4: 498–500. of planarian regeneration. Development 126: planarian flatworm: an in vivo model for stem 13. Newmark PA (2005) Opening a new can of 1031–1040. cell biology and nervous system regeneration. worms: a large-scale RNAi screen in planarians. 22. Meinhardt H (2009) Beta-catenin and axis Dis Mod Mech 4: 12–19. Dev Cell 8: 623–624. formation in planarians. Bioessays 31: 5–9.

PLoS Computational Biology | www.ploscompbiol.org 10 April 2012 | Volume 8 | Issue 4 | e1002481 23. Lange C, Steele V (1978) The Mechanism of 48. Brown FA (1966) Effects and after-effects on H,K-ATPase-mediated membrane voltage is anterior-posterior polarity control in planarians. planarians of reversals of the horizontal mag- required for planarian head regeneration. Chem Differentiation 11: 1–12. netic vector. Nature 209: 533–535. Biol 18: 77–89. 24. Chaplain MAJ, McLachlan JC, Singh GD 49. Brown FA, Chow CS (1975) Differentiation 71. Nogi T, Yuan YE, Sorocco D, Perez-Tomas R, (1999) On growth and form: spatio-temporal between clockwise and counterclockwise mag- Levin M (2005) Eye regeneration assay reveals pattern formation in biology. New York: Wiley. netic rotation by planarian, Dugesia dorotace- an invariant functional left-right asymmetry in 413 p. phala. Physiol Zool 48: 168–176. the early bilaterian, Dugesia japonica. Laterality 25. Ellner SP, Guckenheimer J (2006) Dynamic 50. Brown FA, Park YH (1964) Seasonal variations 10: 193–205. models in biology. Princeton, NJ: Princeton in sign and strength of gamma-taxis in planar- 72. Nogi T, Zhang D, Chan JD, Marchant JS University Press. 329 p. ians. Nature 202: 469–471. (2009) A novel biological activity of praziquantel 26. Szallasi Z, Stelling J, Periwal V (2006) System 51. Bagun˜a´ J, Romero R (1981) Quantitative requiring voltage-operated Ca2+ channel beta modeling in cell biology: from concepts to nuts analysis of cell types during growth, degrowth subunits: subversion of flatworm regenerative and bolts. Cambridge, MA: MIT Press. 448 p. and regeneration in the planarians Dugesia polarity. PLoS Negl Trop Dis 3: e464. 27. Newmark P, Sa´nchez Alvarado A (2001) mediterranea and Dugesia tigrina. Hydrobiolo- doi:10.1371/journal.pntd.0000464. Regeneration in Planaria. Encyclopedia of life gia 84: 181–194. 73. Sasaki M, Gonzalez-Zulueta M, Huang H, sciences. John Wiley & Sons, Ltd. 52. Takeda H, Nishimura K, Agata K (2009) Herring WJ, Ahn SY, et al. (2000) Dynamic 28. Morgan T (1901) Growth and regeneration in Planarians maintain a constant ratio of different regulation of neuronal NO synthase transcrip- cell types during changes in body size by using tion by calcium influx through a CREB family Planaria lugubris. Dev Genes Evol 13: 179–212. the stem cell system. Zool Sci 26: 805–813. transcription factor-dependent mechanism. Proc 29. Brøndsted HV (1955) Planarian regeneration. 53. Bagun˜a´ J, Salo´ E, Auladell C (1989) Regener- Natl Acad Sci U S A 97: 8617–8622. Biological Reviews 30: 65–126. ation and pattern-formation in planarians .3. 74. Zhao M, Song B, Pu J, Wada T, Reid B, et al. 30. Brøndsted HV (1969) Planarian regeneration. Evidence that neoblasts are totipotent stem-cells (2006) Electrical signals control wound healing New York: Pergamon Press. and the source of blastema cells. Development through phosphatidylinositol-3-OH kinase-gam- 31. Salo´ E, Bagun˜a´ J (1984) Regeneration and 107: 77–86. ma and PTEN. Nature 442: 457–460. pattern formation in planarians. I. The pattern 54. Wagner DE, Wang IE, Reddien PW (2011) 75. Yang SJ, Liang HL, Ning G, Wong-Riley MTT of mitosis in anterior and posterior regeneration Clonogenic neoblasts are pluripotent adult stem (2004) Ultrastructural study of depolarization- in Dugesia (G) tigrina, and a new proposal for cells that underlie planarian regeneration. Sci- induced translocation of NRF-2 transcription blastema formation. J Emb Exp Mor 83: 63–80. ence 332: 811–816. factor in cultured rat visual cortical neurons. 32. Newmark P, Alvarado A (2002) Not your 55. Handberg-Thorsager M, Fernandez E, Salo E Eur J Neurosci 19: 1153–1162. father’s planarian: a classic model enters the (2008) Stem cells and regeneration in planari- 76. Petersen CP, Reddien PW (2008) Smed-betaca- era of functional genomics. Nat Rev Genet 3: ans. Front Biosci 13: 6374–6394. tenin-1 is required for anteroposterior blastema 210–219. 56. Pascolini R, Tei S, Vagnetti D, Bondi C (1984) polarity in planarian regeneration. Science 319: 33. Agata K (2003) Regeneration and gene regula- Epidermal-cell migration during wound-healing 327–330. tion in planarians. Curr Opin Genet Dev 13: in Dugesia-Lugubris - observations based on 77. Cebria` F, Newmark PA (2007) Morphogenesis 492–496. scanning electron-microscopy and treatment defects are associated with abnormal nervous 34. Salo´ E, Abril JF, Adell T, Cebria` F, Eckelt K, with cytochalasin. Cell Tissue Res 236: system regeneration following roboA RNAi in et al. (2009) Planarian regeneration: achieve- 345–349. planarians. Development 134: 833–837. ments and future directions after 20 years of 57. Chandebois R (1979) The dynamics of wound 78. Tasaki J, Shibata N, Nishimura O, Itomi K, research. Int J Dev Biol 53: 1317–1327. closure and its role in the programming of Tabata Y, et al. (2011) ERK signaling controls 35. Rubenstein M, Sai Y, Chuong CM, Shen WM planarian regeneration I—Blastema emergence. blastema cell differentiation during planarian (2009) Regenerative patterning in Swarm Ro- Dev Growth Differ 21: 195–204. regeneration. Development 138: 2417–2427. bots: mutual benefits of research in robotics and 58. Pellettieri J, Fitzgerald P, Watanabe S, 79. Tasaki J, Shibata N, Sakurai T, Agata K, stem cell biology. Int J Dev Biol 53: 869–881. Mancuso J, Green DR, et al. (2010) Cell death Umesono Y (2011) Role of c-Jun N-terminal 36. Gordon R, Melvin CA (2003) Reverse engi- and tissue remodeling in planarian regeneration. kinase activation in blastema formation during neering the embryo: a graduate course in Dev Biol 338: 76–85. planarian regeneration. Dev Growth Differ 53: developmental biology for engineering students 59. Salo´ E (2006) The power of regeneration and 389–400. at the University of Manitoba, Canada. Int J Dev the stem-cell kingdom: freshwater planarians 80. Fraguas S, Barbera´n S, Cebria` F (2011) EGFR Biol 47: 183–187. (Platyhelminthes). Bioessays 28: 546–559. signaling regulates cell proliferation, differentia- 37. Perkins TJ, Jaeger J, Reinitz J, Glass L (2006) 60. Santos FV (1929) Studies on transplantation in tion and morphogenesis during planarian re- Reverse engineering the gap gene network of Planaria. Biol Bull 57: 188–197. generation and homeostasis. Dev Biol 354: Drosophila melanogaster. PLoS Comp Biol 2: 61. Kobayashi C, Nogi T, Watanabe K, Agata K 87–101. e51. doi:10.1371/journal.pcbi.0020051. (1999) Ectopic pharynxes arise by regional 81. Meinhardt H (1982) Models of biological 38. Jaeger J (2009) Modelling the Drosophila reorganization after anterior/posterior chimera pattern formation. Academic Press. embryo. Mol Biosys 5: 1549–1568. in planarians. Mech Dev 89: 25–34. 82. Jenkins MM (1963) Bipolar planarians in a stock 39. Hogeweg P (2011) The roots of bioinformatics 62. Levin M (2007) Gap junctional communication culture. Science (New York, NY) 142. in theoretical biology. PLoS Comp Biol 7: in morphogenesis. Prog Biophys Mol Biol 94: 83. Petersen C, Reddien P (2009) Wnt Signaling e1002021. doi:10.1371/journal.pcbi.1002021. 186–206. and the polarity of the primary body axis. Cell 40. Forsthoefel DJ, Park AE, Newmark PA (2011) 63. Esser AT, Smith KC, Weaver JC, Levin M 139: 1056–1068. Stem cell-based growth, regeneration, and (2006) Mathematical model of morphogen 84. Petersen C, Reddien P (2008) Smed-bcatenin-1 electrophoresis through gap junctions. Dev remodeling of the planarian intestine. Dev Biol is required for anteroposterior blastema polarity Dyn 235: 2144–2159. in planarian regeneration. Science 319: 356: 445–459. 64. Levin M, Buznikov GA, Lauder JM (2006) Of 327–330. 41. Cebria` F (2007) Regenerating the central minds and embryos: left-right asymmetry and 85. Iglesias M, Gomez-Skarmeta JL, Salo´ E, Adell T nervous system: how easy for planarians! Dev the serotonergic controls of pre-neural morpho- (2008) Silencing of Smed-bcatenin1 generates Genes Evol 217: 733–748. genesis. Dev Neurosci 28: 171–185. radial-like hypercephalized planarians. Devel- 42. Brown HM, Ito H, Ogden TE (1968) Spectral 65. Nogi T, Levin M (2005) Characterization of opment 135: 1215–1221. sensitivity of the planarian ocellus. J Gen Phys innexin gene expression and functional roles of 86. Rink JC, Gurley KA, Elliott SA, Sa´nchez 51: 255–260. gap-junctional communication in planarian Alvarado A (2009) Planarian Hh signaling 43. Brown FA, Park YH (1975) A persistent regeneration. Dev Biol 287: 314–335. regulates regeneration polarity and links Hh monthly variation in responses of planarians to 66. Oviedo N, Levin M (2007) smedinx-11 is a pathway evolution to cilia. Science 326: light, and its annual modulation. Int J Chronobio planarian stem cell gap junction gene required 1406–1410. 3: 57–62. for regeneration and homeostasis. Development 87. Yazawa S, Umesono Y, Hayashi T, Tarui H, 44. Mason PR (1975) Chemo-klino-kinesis in pla- 134: 3121–3131. Agata K (2009) Planarian Hedgehog/Patched narian food location. Anim Behav 23: 460–469. 67. Levin M (2007) Large-scale biophysics: ion flows establishes anterior-posterior polarity by regu- 45. Miyamoto S, Shimozawa A (1985) Chemotaxis and regeneration. Trends Cell Biol 17: 261–270. lating Wnt signaling. Proc Natl Acad Sci USA in the freshwater planarian Dugesia japonica. 68. Levin M (2009) Bioelectric mechanisms in 106: 22329–22334. Zool Sci 2: 389–396. regeneration: unique aspects and future per- 88. Gurley KA, Rink JC, Alvarado AS (2008) b- 46. Aoki R, Wake H, Sasaki H, Agata K (2009) spectives. Semin Cell Dev Biol 20: 543–556. catenin defines head versus tail identity during Recording and spectrum analysis of the planar- 69. Adams DS (2008) A new tool for tissue planarian regeneration and homeostasis. Sci- ian electroencephalogram. Neuroscience 159: engineers: ions as regulators of morphogenesis ence 319: 323–327. 908–914. during development and regeneration. Tissue 89. Petersen C, Reddien P (2011) Polarized notum 47. Brown HM, Ogden TE (1968) The electrical Eng Part A 14: 1461–1468. activation at wounds inhibits Wnt function to response of the planarian ocellus. J Gen Phys 51: 70. Beane WS, Morokuma J, Adams DS, Levin M promote planarian head regeneration. Science 237–253. (2011) A chemical genetics approach reveals 332: 852–855.

PLoS Computational Biology | www.ploscompbiol.org 11 April 2012 | Volume 8 | Issue 4 | e1002481 90. Iglesias M, Almuedo-Castillo M, Aboobaker AA, planarian stem cells. Dev Genes Evol 216: 128. Meinhardt H (2009) Models for the generation Salo E (2011) Early planarian brain regenera- 335–346. and interpretation of gradients. Cold Spring tion is independent of blastema polarity medi- 109. Reddien P, Oviedo N, Jennings J, Jenkin J, Harb Perspect Biol 1. ated by the Wnt/beta-catenin pathway. Dev Alvarado A (2005) SMEDWI-2 is a PIWI-like 129. Agata K, Tanaka T, Kobayashi C, Kato K, Biol 358: 68–78. protein that regulates planarian stem cells. Saitoh Y (2003) Intercalary regeneration in 91. Romero R, Bueno D (2001) Disto-proximal Science 310: 1327–1330. planarians. Dev Dyn 226: 308–316. regional determination and intercalary regener- 110. Eisenhoffer G, Kang H, Alvarado A (2008) 130. Bongard J, Zykov V, Lipson H (2006) Resilient ation in planarians, revealed by retinoic acid Molecular analysis of stem cells and their machines through continuous self-modeling. induced disruption of regeneration. Int J Dev descendants during cell turnover and regenera- Science 314: 1118–1121. Biol 45: 669–673. tion in the planarian Schmidtea mediterranea. 131. Basanta D, Miodownik M, Baum B (2008) The 92. Marsh G, Beams HW (1952) Electrical control Cell Stem Cell 3: 327–339. evolution of robust development and homeosta- of morphogenesis in regenerating Dugesia 111. Oviedo N, Pearson B, Levin M, Alvarado A sis in artificial organisms. PLoS Comp Biol 4: tigrina. I. Relation of axial polarity to field (2008) Planarian PTEN homologs regulate stem e1000030. doi:10.1371/journal.pcbi.1000030. strength. J Cell Comp Physiol 39: 191–213. cells and regeneration through TOR signaling. 132. Andersen T, Newman R, Otter T (2009) Shape 93. Dimmitt J, Marsh G (1952) Electrical control of Dis Mod Mech 1. homeostasis in virtual embryos. Artif Life 15: morphogenesis in regenerating Dugesia tigrina. 112. Hollander MC, Blumenthal GM, Dennis PA 161–183. II. Potential gradient vs. current density as (2011) PTEN loss in the continuum of common 133. Larson D, Johnson R, Swat M, Cordero J, control factors. J Cell Comp Physiol 40: 11–23. cancers, rare syndromes and mouse models. Nat Glazier J, et al. (2010) Computer simulation of 94. Mcwhinnie MA (1955) The effect of colchicine Rev Cancer 11: 289–301. cellular patterning within the Drosophila pupal on reconstitutional development in Dugesia- 113. Nishimura K, Kitamura Y, Taniguchi T, eye. PLoS Comp Biol 6: e1000841. Dorotocephala. Biol Bull 108: 54–65. Agata K (2010) Analysis of motor function doi:10.1371/journal.pcbi.1000841. 95. Kanatani H (1958) Formation of bipolar heads modulated by cholinergic neurons in planarian 134. Kennaway R, Coen E, Green A, Bangham A induced by demecolcine in the planarian Dugesia japonica. Neuroscience 168: 18–30. Dugesia gonocephala. J Fac Sci Tokyo Univ 8: (2011) Generation of diverse biological forms 114. Nishimura K, Kitamura Y, Umesono Y, through combinatorial interactions between 253–270. Takeuchi K, Takata K, et al. (2008) Identifica- 96. Lowe CJ, Terasaki M, Wu M, Freeman RM, Jr., tissue polarity and growth. PLoS Comp Biol 7: tion of glutamic acid decarboxylase gene and e1002071. doi:10.1371/journal.pcbi.1002071. Runft L, et al. (2006) Dorsoventral patterning in distribution of GABAergic nervous system in the hemichordates: insights into early chordate 135. de Garis H (1992) Artificial embryology: the planarian Dugesia japonica. Neuroscience 153: genetic programming of an artificial embryo. In: evolution. PLoS Biol 4: e291. doi:10.1371/ 1103–1114. journal.pbio.0040291. Soucek B, ed. Dynamic, genetic, and chaotic 115. Nishimura K, Kitamura Y, Inoue T, programming: New York Wiley. 97. Brown FD, Prendergast A, Swalla BJ (2008) Umesono Y, Sano S, et al. (2007) Reconstruc- Man is but a worm: chordate origins. Genesis 136. Glazier JA, Graner F (1993) Simulation of the tion of dopaminergic neural network and differential adhesion driven rearrangement of 46: 605–613. locomotion function in planarian regenerates. biological cells. Phys Rev E 47: 2128–2154. 98. Molina MD, Salo´ E, Cebria` F (2007) The BMP Developmental neurobiology 67: 1059–1078. 137. Savill NJ, Hogeweg P (1997) Modelling mor- pathway is essential for re-specification and 116. Cebria` F (2008) Organization of the nervous phogenesis: From single cells to crawling slugs. maintenance of the dorsoventral axis in regen- system in the model planarian Schmidtea erating and intact planarians. Dev Biol 311: J Theor Biol 184: 229–235. mediterranea: an immunocytochemical study. 138. Lindenmayer A (1968) Mathematical models for 79–94. Neurosci Res 61: 375–384. 99. Orii H, Watanabe K (2007) Bone morphoge- cellular interaction in development: Parts I and 117. Almuedo-Castillo M, Salo´ E, Adell T (2011) netic protein is required for dorso-ventral II. J Theor Biol 18: 280–315. Dishevelled is essential for neural connectivity patterning in the planarian Dugesia japonica. 139. Gautier H, Mech R, Prusinkiewicz P, Varlet- and planar cell polarity in planarians. Proc Natl Dev Growth Differ 49: 345–349. Acad Sci USA 108: 2813–2818. Grancher C (2000) 3D architectural modelling 100. Reddien PW, Bermange AL, Kicza AM, San- of aerial photomorphogenesis in white clover 118. Cebria` F, Kobayashi C, Umesono Y, chez Alvarado A (2007) BMP signaling regulates (Trifolium repens L.) using L-systems. Ann Bot Nakazawa M, Mineta K, et al. (2002) FGFR- the dorsal planarian midline and is needed for related gene nou-darake restricts brain tissues to 85: 359–370. asymmetric regeneration. Development 134: the head region of planarians. Nature 419: 140. Allen MT, Prusinkiewicz P, DeJong TM (2005) 4043–4051. 620–624. Using L-systems for modeling source-sink inter- 101. Molina MD, Neto A, Maeso I, Gomez- 119. Cebria` F, Newmark PA (2005) Planarian actions, architecture and physiology of growing Skarmeta JL, Salo´ E, et al. (2011) Noggin and trees: the L-PEACH model. New Phytol 166: noggin-like genes control dorsoventral axis homologs of netrin and netrin receptor are required for proper regeneration of the central 869–880. regeneration in planarians. Curr Biol 21: 141. Prusinkiewicz P, Erasmus Y, Lane B, 300–305. nervous system and the maintenance of nervous system architecture. Development 132: Harder LD, Coen E (2007) Evolution and 102. Cebria` F, Guo TX, Jopek J, Newmark PA development of inflorescence architectures. Sci- (2007) Regeneration and maintenance of the 3691–3703. 120. Mittenthal JE, Nuelle JR (1988) Discontinuities ence 316: 1452–1456. planarian midline is regulated by a slit ortholo- 142. Lobo D, Vico FJ, Dassow J (2011) Graph gue. Dev Biol 307: 394–406. of pattern and rules for regeneration in limbs of crayfish. Dev Biol 126: 315–326. grammars with string-regulated rewriting. 103. Gurley KA, Elliott SA, Simakov O, Theor Comp Sci 412: 6101–6111. Schmidt HA, Holstein TW, et al. (2010) 121. Salo´ E, Bagun˜a´ J (1985) Proximal and distal 143. Ferna´ndez-Blanco E, Dorado J, Rabun˜al J, Expression of secreted Wnt pathway compo- transformation during intercalary regeneration Gestal M, Pedreira N (2007) A computational nents reveals unexpected complexity of the in the planarian Dugesia (S) mediterranea - morphogenesis approach to simple structure planarian amputation response. Dev Biol 347: evidence using a chromosomal marker. Wilhelm development. Lecture Notes in Artificial Intel- 24–39. Rouxs Arch Dev Biol 194: 364–368. ligence 4648: 825–834. 104. Adell T, Salo´ E, Boutros M, Bartscherer K 122. Pellettieri J, Sa´nchez Alvarado A (2007) Cell (2009) Smed-Evi/Wntless is required for beta- turnover and adult tissue homeostasis: From 144. Haddow P, Hoye J (2007) Achieving a simple catenin-dependent and -independent processes humans to planarians. Annu Rev Genet 41: development model for 3D shapes: are chemi- during planarian regeneration. Development 83–105. cals necessary? GECCO ’07: Proceedings of the 136: 905–910. 123. Oviedo N, Newmark P, Sa´nchez Alvarado A 9th annual conference on Genetic and evolu- 105. Reddien PW (2011) Constitutive gene expres- (2003) Allometric scaling and proportion regu- tionary computation. London, England: ACM. sion and the specification of tissue identity in lation in the freshwater planarian Schmidtea pp 1013–1020. adult planarian biology. Trends Genet 27: mediterranea. Dev Dyn 226: 326–333. 145. Devert A, Bredeche N, Schoenauer M (2007) 277–285. 124. Gonza´lez-Este´vez C, Felix DA, Aboobaker AA, Robust multi-cellular developmental design. 106. Newmark PA, Sa´nchez Alvarado A (1999) Salo´ E (2007) Gtdap-1 and the role of GECCO ’07: Proceedings of the 9th annual Planarian regeneration. Embryonic encyclope- autophagy during planarian regeneration and conference on Genetic and evolutionary com- dia of life sciences. London: Nature Publishing starvation. Autophagy 3: 640–642. putation. London, England: ACM. pp 982–989. Group. 125. Wolpert L (1969) Positional information and the 146. Miller J (2004) Evolving a Self-Repairing, Self- 107. Palakodeti D, Smielewska M, Lu YC, Yeo GW, spatial pattern of cellular differentiation. J Theor Regulating, French Flag Organism. GECCO Graveley BR (2008) The PIWI proteins Biol 25: 1–47. ’04: Proceedings of the 6th annual conference SMEDWI-2 and SMEDWI-3 are required for 126. Meinhardt H (1978) Space-dependent cell on Genetic and Evolutionary Computation. pp stem cell function and piRNA expression in determination under the control of morphogen 129–139. planarians. RNA 14: 1174–1186. gradient. J Theor Biol 74: 307–321. 147. Lobo D, Vico FJ (2010) Evolution of form and 108. Rossi L, Salvetti A, Lena A, Batistoni R, Deri P, 127. Turing AM (1952) The chemical basis of function in a model of differentiated multicellu- et al. (2006) DjPiwi-1, a member of the PAZ- morphogenesis. Philos Trans R Soc Lond, Ser lar organisms with gene regulatory networks. Piwi gene family, defines a subpopulation of B: Biol Sci 237: 37–72. BioSyst 102: 112–123.

PLoS Computational Biology | www.ploscompbiol.org 12 April 2012 | Volume 8 | Issue 4 | e1002481