An RNAi screen in S. mediterranea
RNAi has been demonstrated to result in specific and near complete elimination of detectable S. mediterranea
mRNA and proteins (Newmark et al., 2003
; Sánchez Alvarado and Newmark, 1999
; data not shown). Nonetheless, since it is impractical to monitor protein levels for the products of each gene in an RNAi screen, some phenotypes resulting from dsRNA treatment could reflect incomplete gene loss of function. The RNAi by feeding methodology used for our screen involves expressing dsRNA from a planarian gene in bacteria and suspending those bacteria with blended liver and agarose (Newmark et al., 2003
). We optimized the feeding method and protocol used in this manuscript (, methods) through extensive experimentation (data not shown). We generated an RNAi vector (pDONRdT7) that contains two T7 RNA polymerase promoters flanked by two class I T7 transcriptional terminators () that results in more effective RNAi than does the conventional vector (data not shown). pDONRdT7 utilizes a modified Gateway cloning strategy (Invitrogen) to facilitate cDNA transfer.
Figure 1 RNAi screening strategy. (A) S. mediterranea cDNAs were transferred into pDONRdT7, which contains two T7 promoters and terminators, using a single-step Gateway (Invitrogen) reaction (see methods). (B) Screening procedure (see methods for details). (C) (more ...) S. mediterranea
cDNAs randomly selected from two cDNA libraries were inserted into pDONRdT7 and introduced into the RNaseIII-deficient bacterial strain HT115 (Timmons et al., 2001
). The two cDNA libraries were derived from a neoblast-enriched cell population and animal heads; all cDNAs used define unique genes (see methods). 1065 genes were inhibited using RNAi by feeding. Given the S. mediterranea
genome sequence is currently incomplete, it is unknown what percentage of the total number of S. mediterranea
genes these 1065 represent. The screen protocol consisted of three feedings, two rounds of regeneration following amputation of heads and tails, and three scorings (). Animals were scored for size of head blastemas on trunks and tails, size of tail blastemas, ability of tails to regenerate a pharynx in pre-existing tissues, shape of blastemas, presence and pattern of photoreceptors, light response, vibration response, touch response, flipping, locomotion, turning, and head lifting. Many animals, both with and without a detectable defect, were fixed and analyzed by antibody labeling to detect additional phenotypes at the cellular level (). Multiple RNAi feedings and two rounds of regeneration likely minimized protein perdurance. The multiple scoring timepoints served to determine degrees of phenotype expressivity, since aspects of a particular phenotype might be observed in the initial scoring and precluded by a more severe aspect of the phenotype in the latter scoring.
Development of a nomenclature system for comparisons of phenotypes
This manuscript contains a large amount of new phenotype terminology and involves the presentation of large datasets. Comparisons of phenotypes can allow the clustering of genes into candidate functional categories. There is, however, a practical problem involving the comparisons of phenotypes that are descriptive. Because essentially all of the phenotypes reported in this manuscript are new, we have devised a nomenclature system that allows common usage of descriptive terms for different phenotypes and comparison of different regeneration defects. To assist the reader, major terms used to define phenotypes are listed in ; planarian body regions related to locations where defects were observed, and the terms used in phenotype descriptions to identify these regions, are presented in .
240 genes for which RNAi generates a phenotype have been identified
Of the 1065 genes perturbed by RNAi, 240 (22.5%) conferred specific phenotypes when perturbed (, , S1). A sampling of the spectrum of phenotypes observed can be found in and . Genes are identified in Tables and Figures with an RNAi clone identification name in which the letter(s) “H” or “NB” refer to genes from the head or neoblast-enriched cDNA libraries, respectively, “E” indicates that clones are “entry” clones in the Gateway naming system, and the alphanumerical code refers to 96-well plate coordinates. Many phenotypic categories were uncovered, including the inability to regenerate (). Blastema-size abnormalities have been categorized on a scale from 0 to 3, with “BLST(0)” referring to no regeneration and “BLST(3)” referring to normal regeneration (, ). Other major phenotypic categories include ventral curling (), blastema shape and morphology abnormalities (), a variety of photoreceptor abnormalities (), behavioral defects (, S1), tissue regression (), lesions (), and lysis (). We also uncovered a large number of unexpected and surprising phenotypic categories that occurred with lower frequency. Examples include defects unique to caudal blastemas (TLBLST, , ), animals that glide sideways (), animals with signs of asymmetry (, S1, , ), animals with abnormal posture (), animals with pigment “freckles” in the normally unpigmented blastemas or darkened body spots (), and animals with ectopic growths and photoreceptors (, ). These defined phenotype categories can be used as tools for the study of cellular events and genetic regulatory mechanisms that control the poorly understood biology of planarians.
Summary of S. mediterranea RNAi screen results
240 genes confer phenotypes in S. mediterranea
Figure 2 Representative phenotypes from the RNAi screen. Phenotype nomenclature and homologies are in . White arrowheads, defects. Anterior, left. v, ventral surface. Bar, 0.2 mm. (A) Control, unc-22 (a C. elegans gene) RNAi animal. Irradiation at 6000rad (more ...)
Figure 5 Analyses of blastema differentiation and pattern formation in animals with screen phenotypes. (A–N) Anterior, left. (A) Photoreceptor system defect terminology. EXTNT, photoreceptor regeneration extent abnormal; descriptors: nopr, no photoreceptors; (more ...)
Figure 3 Identification of genes with candidate functions in planarian regeneration. (A) Planarian regeneration is divided into seven stages, “I” through “VII”. Anterior region of a decapitated planarian is shown, facing up. Wound (more ...)
85% of S. mediterranea genes associated with RNAi phenotypes are conserved
Of the 240 genes associated with RNAi phenotypes, 205 (85%) are predicted to encode proteins with significant homology (BLAST, E≤ 10−06
) to those encoded in the genomes of other organisms (, S1). This high frequency, coupled with the diverse set of predicted functions for these genes (), demonstrates the utility of studies of S. mediterranea
for broadly informing general metazoan biology. For example, 38 of the genes associated with RNAi phenotypes are related to human disease genes (, S2). These genes cause an array of phenotypes, ranging from aberrant regeneration following RNAi of a spastic paraplegia homolog (Casari et al., 1998
) to aberrant photoreceptor regeneration and function following RNAi of an RGS9-like encoding gene, which is associated with bradyopsia (vision defects) in humans (Nishiguchi et al., 2004
). Given that only eight of these 38 genes have a corresponding mouse knockout model, the S. mediterranea
phenotypes provide new functional information and demonstrate the utility of S. mediterranea
for the study of orthologs of human genes involved in genetic disorders. The remaining 35 genes associated with RNAi phenotypes, for which no obvious homologues were found in other phyla, may also be of medical relevance. These genes may be specific to the Platyhelminthes and required for the survival of related pathogenic platyhelminthes, the cestodes and trematodes (, S1). Considering such pathogens are estimated to cause disease in nearly 300 million people throughout the world (www.who.int
), these genes might make attractive drug targets.
Genes associated with similar and specific RNAi phenotypes may act together
We obtained a wide gamut of regeneration phenotypes from the RNAi screen. We categorized genes associated with these dsRNA-induced phenotypes into 11 groups as shown in . also includes the homology of these genes (if any), and a code that describes the RNAi phenotype in detail (see as guides for phenotype terminology). RNAi-induced phenotypes and sequence homologies for some genes that are listed as “Other” in can be found in Table S1.
We observed that RNAi of a number of different genes resulted in similar phenotypes, suggesting that such genes may be acting together. Sequence data supports this hypothesis. Specifically, genes predicted to encode proteins homologous to those known to act together in other organisms conferred similar phenotypes when perturbed independently. For instance, RNAi of two genes that encode different subunits of the ARP2/3 complex (HE.2.11E, HE.2.12A, ), which is known to mediate actin filament nucleation (Weaver et al., 2003
), caused early lysis; RNAi of two genes encoding components of TGF-β signalling (HE.2.07D BMP1, , HE.3.03B SMAD4) caused indented blastemas; and RNAi of α and β-tubulin-encoding genes (HE.1.01H, , HE.1.03G) caused uncoordinated behavior, blisters, and bloating (, S1). These examples indicate that RNAi screening in S. mediterranea
can readily generate specific phenotypes that allow identification of previously unknown functional associations. For instance, RNAi of NBE.3.07F or NBE.5.04A () caused spots, blisters, and bloating. The first gene is similar to Drosophila
hunchback, a gene known to regulate embryonic patterning (Irish et al., 1989
), and the other encodes a POU domain protein (). Given the rarity of this phenotype, these two transcription factors may act together. Additionally, RNAi of only two genes, HE.1.08G and NBE.8.03C (), caused the “freckles” phenotype. HE.1.08G encodes an a-spectrin-like protein, a membrane cytoskeletal protein (Bennett and Baines, 2001
), and NBE.8.03C encodes a protein with no known predicted function that may act with a-spectrin (). These examples are presented simply to illustrate that for the many phenotypic categories, including regeneration and neoblast abnormalities (see below), our data in , S1, S3, S4, and – identify shared properties that point to many candidate functional associations (see below for further examples).
Figure 6 Homeostasis defects in dsRNA-fed animals. (A–D), Arrowheads, defects. v, ventral. Bar, 0.4 mm. Anatomy and nomenclature, . Additional terms: all, entire animal; ant, anterior half or the anterior end of a region; int, gastrovascular (more ...)
Strategy for identifying genes that control distinct steps in regeneration
A series of stereotypical events following wounding produce a fully functional regenerated planarian (). I. Regeneration begins with the spreading of an epidermis over a wound surface followed by signaling that triggers the initiation of regeneration. II. Neoblasts are maintained in the parenchyma in appropriate numbers where they respond to wounds by proliferating. III. Neoblast progeny migrate and generate a blastema. IV. The cells within the regeneration blastema differentiate and organize to produce properly patterned structures. V. Changes occur in the pre-existing tissue of an amputated animal to generate new structures and a new animal with the proper proportions (morphallaxis). VI. Pre-existing tissue and the new tissue are maintained, involving the functioning of differentiated cells and the replacement of aged cells by neoblast progeny. VII. Finally, the regenerated animal restores the capacity to respond to its environment with appropriate behaviors. How can we identify the step of regeneration for which a given gene is needed? We performed additional experiments outlined below that, together with the regeneration data from the screen, allowed us to cluster genes into categories that correspond to the different phases of regeneration shown in .
I. Wound healing and regeneration initiation
In the absence of the ability to heal wounds, animals should lose tissue through the wound site and lyse. We observed that RNAi of at least nine genes caused lysis after wounding (, S1). We reasoned that if a gene were needed specifically for wound healing, inhibition of that gene with dsRNA would not cause lysis in intact, non-amputated animals. We identified one gene, HE.3.04D, that fits these criteria (, , S4, see homeostasis in Experimental procedures
for details). HE.3.04D is predicted to encode a novel protein.
We reasoned that genes specifically involved in the initiation of regeneration following wound healing might be required for normal blastema formation, but not for the extensive cell turnover that occurs during normal adult planarian life (Newmark and Sánchez Alvarado, 2000
). We inhibited 143 genes that were associated with dsRNA-induced regeneration defects in the screen and examined intact, non-amputated animals (, see Experimental procedures
). We found that genes needed for regeneration also tend to be needed for homeostasis (P<0.005) (see below). However, RNAi of 35 out of 143 genes conferred no or only minor defects in intact animals (Table S4). 25 of these 35 genes were associated with smaller than normal blastemas in two separate RNAi experiments (Table S4, ). One gene was important for the formation of caudal blastemas (HE.4.06F) and is predicted to encode a novel protein (Table S4). Four genes within this dataset, such as an FKBP-like immunophilin (NBE.3.05F), caused tissue regression following RNAi and regeneration (, S4). Genes needed for complete regeneration but apparently not necessary for homeostasis include those predicted to encode proteins similar to chondrosarcoma-associated protein 2 (NBE.3.11F), nucleostemin (NBE.7.07H), a DEAD box RNA-binding protein (HE.1.06D), SMAD4 (HE.3.03B), Baf53a (HE.3.10F), and a WW-domain protein (HE.3.02A) (Table S4). Some of these genes could identify signaling mechanisms that specifically activate neoblasts following wounding or control other processes needed for blastema generation. One of these genes, SMAD4, stands apart as a gene necessary for any blastema formation but dispensable for neoblast function in homeostasis (). Since SMAD proteins mediate TGF-β signals (ten Dijke and Hill, 2004
), this observation indicates that TGF-β signaling may control regeneration initiation in planarians.
II. Neoblast function: Comparison of phenotypes to defects in irradiated animals identifies candidate neoblast regulators
Irradiation of planarians is known to specifically kill the neoblasts, block regeneration, and result in lethality (Bardeen and Baetjer, 1904
). We observed that amputated, irradiated (e.g., 6000rad) animals were incapable of regenerating (), curled their bodies around their ventral surface within 15 days (), and subsequently died by lysis. Genes for which RNAi causes defects similar to those of irradiated animals may be needed for neoblast function in regeneration. In total, 140 gene perturbations blocked, limited, or reduced regeneration (, S1). RNAi of 48 of these genes caused curling (CRL), similar to that seen in irradiated animals (, S1, ). Lysis was the typical fate of these curled animals (, S1). Although many of these genes may not have functions specific to neoblast regulation, we suggest that most if not all of these genes are required for neoblast function. These genes include basal cell machinery factors, RNA binding proteins (HB.14.6D, NBE.4.06D, NBE.7.07D, NBE.8.12D), signal transduction factors (NBE.4.08C, phosphatidyl inositol transfer protein; NBE.2.09G, WD40 repeat protein), chromatin regulators (e.g., HE.2.01H, histone deacetylase), and disease genes (e.g., NBE.3.08C, human spastic paraplegia protein) (, S1).
II and III. Neoblast maintenance, proliferation, and progeny function
The genes for which RNAi caused defects similar to that caused by irradiation could be needed for neoblast maintenance and/or proliferation or the functions of neoblast progeny. We reasoned that direct observation of neoblast presence and proliferation in dsRNA-fed animals could help distinguish between these possibilities. We labeled dsRNA-fed animals with an antibody (αH3P, anti-phosphorylated histone H3 (Hendzel et al., 1997
) that recognizes mitotic neoblasts (Newmark and Sánchez Alvarado, 2000
). Numbers of mitotic nuclei were quantified and categorized as described in the legend. As a control, we observed no labeling of irradiated animals with αH3P, confirming αH3P specifically labeled mitotic neoblasts (Newmark and Sánchez Alvarado, 2000
Figure 4 Representative mitotic defects in amputated animals. Anterior, left. (A) αH3P-labeling examples from the RNAi of 140 genes (see text). 14d, 14 days. Bar, 1 mm. Irradiated animals, 6000 rads. Control unc-22 RNAi animals had an average of 212±37 (more ...)
Two sets of animals were used: animals with visible phenotypes in the screen from the RNAi of 140 genes, fixed 14 days following amputation (“14dH3P;”; ), and animals from the RNAi of 139 genes, which were fixed shortly following wounding to assess proliferation at the time of regeneration initiation (“24hH3P”; ). The 139 genes in the 24hH3P dataset were selected because they were associated with a range of blastema-size phenotypes following RNAi and amputation (, see Experimental procedures
). The genes associated with these two datasets (14dH3P and 24hH3P) only partially overlap, and the conclusions drawn from the data are similar. 14dH3P data are presented in Table S3. Data associated with overlapping genes in the 24hH3P and 14dH3P datasets are presented in Table S4 and grouped into four categories that incorporate homeostasis data (see below). Because the data can be grouped in multiple different and informative ways, we also present 24hH3P data in the form of scatter plots in , that allow visualization of how different aspects of phenotypes (regeneration, mitoses after wounding, and homeostasis) associate with one another following RNAi of individual genes.
We identified three main groups of mitotic numbers following the RNAi of genes in both datasets: too few mitoses, normal mitoses, and too many mitoses (). RNAi of 48 of 140 genes in the 14dH3P dataset and RNAi of 50 out of the 139 genes in the 24hH3P dataset led to low mitotic cell numbers. A large majority of animals with lower than normal numbers of mitotic cells also had defects in the production of normal sized blastemas (Tables S3, S4, ). These genes might be important for neoblast maintenance or deployment. Such genes include those predicted to encode multiple components of the ribosome, cell cycle and chromatin regulators, and a phosphatidyl inositol transfer protein (Table S4). RNAi of eight of 140 genes in the 14dH3P dataset and of four genes in the 24hH3P dataset led to abnormally high numbers of mitotic neoblasts as compared to the control, indicating animals to be abnormal due to mitotic defects or misregulation of the neoblast population (Tables S3, S4, ). Among these genes are two predicted to encode components of the proteasome, one predicted to encode gamma tubulin, and two predicted to encode anaphase promoting complex subunits (Tables S3, S4). Given the role of the anaphase-promoting complex and proteolysis in the progression of mitosis (Peters, 2002
), the numerous candidate metaphase nuclei observed in these animals indicate possible defects in chromosome separation at mitosis. Genes for which RNAi caused curling after amputation were very likely to be required for regeneration (P<0.0001) and were often, but not always, associated with reduced mitoses following RNAi and amputation (). Therefore, genes for which RNAi caused curling, blocked regeneration, and caused low mitotic numbers may be needed for neoblast maintenance or mitoses.
RNAi of 84 of 140 genes in the 14dH3P dataset and RNAi of 85 of 139 in the 24hH3P dataset led to relatively normal numbers of mitotic cells (Tables S3, S4, ). Many of these genes are needed for regeneration; for example RNAi of 38 of the 85 genes in the 24hH3P dataset allowed normal numbers of mitotic neoblasts after wounding but caused regeneration of very small blastemas and curling (BLST≤1.5, ). Among these 38 genes are five predicted to encode RNA-binding proteins and five predicted to encode signal transduction proteins (Table S4). These genes may control regeneration initiation, or the ability of neoblast progeny to form differentiated cells or to organize into a blastema. RNAi of 13 genes did not reduce mitotic numbers, but, nonetheless, blocked regeneration and caused curling--suggesting they were involved in neoblast functions (Table S4, ). These genes might be needed for the functioning of neoblast progeny rather than neoblasts per se. Examples of such genes include those encoding a striatin-like protein and an RNA-binding protein (Table S4).
IV. Differentiation and patterning of the regeneration blastema: morphological analyses
Following initial blastema formation, blastemal cells differentiate to produce missing structures. We identified a large number of genes needed for normal blastema morphology and patterning (). Defects observed include indented, pointed, and flat blastemas, as well as wide, faint, and no photoreceptors (, ). The molecular identities of these blastema-patterning genes can be found in and S1. These phenotypes reveal unexpected aspects of planarian biology and identify at least some genes that govern how a collection of undifferentiated cells within a cephalic blastema becomes organized to produce a new, functional head. Below, we highlight several examples of the diversity and implications of the phenotypes within this category.
Wild-type planarians are bilaterally symmetric with no known asymmetry (Hyman, 1951
); however, RNAi of five genes caused asymmetric regeneration of photoreceptors (, S1, , ). Given that we have not seen asymmetric effects of dsRNA treatment on gene expression and that asymmetric phenotypes are rare, these observations indicate active mechanisms may exist for maintaining symmetry in animal species that lack asymmetry. These genes include a Zn transporter (NBE.2.08E) and a Mak16-like protein (NBE.7.09G). 18 genes were associated with regression (RGRS) of blastemas following RNAi, possibly the result of defects in blastema maintenance (, , ). Examples include HE.2.11C myosin II light chain and NBE.3.05F FKBP. RNAi of the candidate axon guidance regulator H.68.4A Slit resulted in the regeneration of ectopic midline neuronal tissue and ectopic axis formation (, S3, ). The formation of photoreceptors may thus be regulated by the spatial location of the brain. Finally, indented blastemas in HE.2.07D BMP1(RNAi)
animals indicate BMP signalling, which regulates dorsal-ventral patterning and other morphogenic events (De Robertis and Kuroda, 2004
), may control regeneration of midline tissues (, , ). Although it was anticipated that rare and unusual phenotypes might be uncovered, what these unusual phenotypes would be and what they would indicate about planarian biology were entirely unknown. Defects such as these not only illuminate the genetic control of specific aspects of planarian biology, but also illustrate that undiscovered roles for known genes in understudied biological processes can be identified in planarians.
IV. Differentiation and patterning of the regeneration blastema: cellular analyses
Next, we utilized animals that regenerated abnormally in the screen from the RNAi of 140 genes, and assessed defects in patterning and differentiation using immunohistology (). Animals were fixed after 14 days of regeneration and labeled with an antiarrestin antibody (VC-1) that recognizes planarian photoreceptors (Sakai et al., 2000
), a kind gift of K. Agata). These same animals were also scored for mitoses with αH3P (, see above). We chose the photoreceptor neurons for study because they exist in two well-defined clusters of ~24 cells and extend easily visualized posterior and ventral processes to the cephalic ganglia (Carpenter et al., 1974
) (). The photoreceptors, therefore, serve as simple landmarks for pattern formation in cephalic blastemas.
A large variety of photoreceptor abnormalities were uncovered (Table S3, , ). Phenotypes include limited regeneration of the photoreceptor system (), photoreceptor cell bodies dispersed posteriorly from the main neuron cluster (“tears” phenotype) and/or ectopic photoreceptors (), diffuse clusters of photoreceptor neurons (), asymmetric photoreceptor cell body clusters (), optic chiasmata defects (), axon abnormalities (), and general disorganization (). These defects revealed not only the various degrees of differentiation that are possible in abnormal blastemas, but also the cellular and patterning abnormalities associated with specific gene perturbations (Table S3, ). For example, RNAi of NBE.3.03D serum response factor resulted in diffuse pigment cups in the visible screen; dispersed photoreceptor cell bodies were uncovered in the VC-1 screen (tears, Table S3). In another example, RNAi of NBE.6.04A HMGB2 caused faint photoreceptors in the visible screen; severely disorganized axons were uncovered in the VC-1 screen (ectoax, ). Homologies of other genes associated with RNAi-induced patterning defects can be found in Table S3.
A scatter plot depicting the correlation between blastema size and the degree of photoreceptor system formation in dsRNA-treated animals identifies several trends (). First, the vast majority of genes for which RNAi severely compromised cephalic regeneration were needed for detectable photoreceptor development (BLST(0–0.5), ). By contrast, many medium-sized blastemas, BLST(1–2.5), can differentiate and organize reasonably well (). Defects within slightly small blastemas can thus be the result of specific defects in pattern formation rather than non-specific results of the blastema being smaller than normal (). The few genes for which RNAi allowed regeneration of a medium-sized blastema with severely disrupted differentiation include candidate specific factors (e.g., tubedown-100 transcription factor, , and a TIMM50 phosphatase). Our data suggests it should be possible to readily identify specific blastema patterning and differentiation defects such as those illustrated in .
Animals lacking a visible phenotype from the RNAi of 677 out of the total 1065 genes in the original RNAi screen were also labeled with antibodies. 13 genes associated with cellular phenotypes following RNAi were identified in this manner (Tables S1, S3). New phenotypes were therefore rare; i.e., about 1–2 % of genes that did not confer a visible defect had a cellular defect. These findings indicate that it should be possible to design a multitude of future screens, coupling our RNAi screening methodology to high-throughput whole-mount immunohistology, to identify genes controlling the specific cellular events of regeneration in planarians.
V. Morphallaxis: genes needed for changes in pre-existing tissue
In addition to the production of new tissues within blastemas, a major element of planarian regeneration involves changes in pre-existing tissues. For example, some organs such as the pharynx form within old tissues, and old tissues can change in length and width. These changes in the proportion and distribution of organs in the differentiated tissues of regenerating planarians was first recognized by T. H. Morgan in 1898 (Morgan, 1898
), and he termed the process of such change morphallaxis. The cellular and molecular mechanisms underlying morphallaxis are nearly completely unknown. One of our screen assays allowed for an assessment of the production of new tissues in old tissues: tail fragments lack a pharynx () and produce a new pharynx in pre-existing tissues. Most genes that were needed for blastema formation were needed for pharynx formation indicating that similar cellular events are involved in both processes (, S1). However, RNAi of 11 genes, such as a nuclear migration nudC-like gene (NBE.1.11B), resulted in weak defects in blastema formation but perturbed pharynx regeneration (, ).
VI. Homeostasis: The function and replacement of differentiated cells
Regenerated animals maintain differentiated tissues with the constant replacement of aged cells by neoblast progeny (). The functions in homeostasis of 143 genes associated with defects in the RNAi screen were assessed (see Experimental procedures
). RNAi of 108 of 143 genes conferred robust defects that define the major planarian homeostasis phenotypes (Table S4, ). By observing irradiated animals, i.e., animals lacking neoblasts, we defined a neoblast-defective homeostasis phenotype. Irradiated animals displayed tissue regression within eight days (), curling within 15 days (), and lysis thereafter. The tissue anterior to the photoreceptors, where regression is typically observed, is normally incapable of regeneration (Morgan, 1898
) and is constantly replaced by neoblast progeny (Newmark and Sánchez Alvarado, 2000
RNAi of many genes caused defects in intact animals similar to those observed in irradiated animals; these genes may be needed for neoblast function in homeostasis (Table S4, ). Tissue regression and curling of intact animals are attributes that tend to appear together in RNAi experiments (48 out of 63 cases) as well as with lysis (40/48), suggesting a common underlying defect (Table S4, ). Genes that cause regression and curling following RNAi in intact animals tend to be needed for regeneration (40 of 48 genes, BLST≤0.5, P<0.0005) indicating these genes may be required for all neoblast functions (). Decrease of αH3P-labeled cells following amputation correlates with curling and regression defects in intact animals (). Of the 66 genes in this homeostasis study that were needed for regeneration (BLST(0/0.5), RNAi of 46 caused intact animals to display tissue regression and RNAi of 42 caused intact animals to curl, indicating about 2/3 of the genes that are needed for regeneration may be needed for neoblast function in homeostasis. Among the 63 genes that caused curling and/or regression in intact animals following RNAi are 33 genes predicted to encode proteins involved in translation or metabolism, 2 in vesicle trafficking, 3 in cell cycle, 4 chromatin factors, 1 cytoskeletal protein, 4 RNA-binding factors, 1 similar to a disease protein, 3 in protein folding, 1 in protein transport, 2 in RNA splicing, 3 signal transduction proteins, and 6 with unknown function (Table S4). This gene set provides a profile of gene functions likely required for the homeostatic functions of neoblasts.
RNAi of some genes caused robust, inviable homeostasis defects but did not block blastema formation or affect neoblast mitoses following amputation (). Therefore, cellular events required for homeostasis need not be required for regeneration or always involve neoblast proliferation. A major category of homeostasis phenotypes involved the formation of a variety of types of lesions (). Genes for which RNAi caused lesions in intact animals did not have strong tendencies to be required for regeneration or neoblast proliferation (). Since irradiation of planarians does not result in lesions (), lesions may arise due to defects in differentiated cells. By contrast, RNAi of some genes caused lesions and did block regeneration. Of the 18 genes for which RNAi blocked regeneration but did not cause regression or curling in intact animals, RNAi of 16 caused lesions to develop in the intact animals (Table S4). There may therefore be two main categories of genes, albeit not mutually exclusive, needed for regeneration and viability in adult animals: one that regulates the functions of neoblasts and another that is needed for differentiated cells.
VII. Behavior of regenerated animals
After completing regeneration, new animals acquire the ability to respond to their environment with normal behaviors (). Planarians locomote via the beating of ventral cilia, can move their body to turn and respond to objects by use of their muscular system, and control their behavior with bicephalic ganglia, two ventral nerve tracts, a variety of sensory systems, and a submuscular nervous plexus (Hyman, 1951
). RNAi of 44 genes conferred uncoordinated locomotion (36 robustly), with RNAi of two additional genes giving uncoordinated flipping (flp) (, S1, ). Following the RNAi of some genes, such as a proprotein convertase-encoding gene (HE.2.02B), which is known to regulate neuropeptides (Bergeron et al., 2000
), animals became completely paralyzed (). Five genes conferred blistering (BLI) and bloating (BLT) as well as lack of coordination following RNAi, including those predicted to encode cytoskeletal proteins such as tubulins (HE.1.01H, HE.1.03G) and rootletin (HE.1.02E), a component of cilia (Yang et al., 2002
) (, ). Since ciliated cells are needed for both locomotion (ventral epidermis) and the excretory system (flame cells of the protonephridia), these genes may play a role in controlling the function of cilia (Hyman, 1951
). RNAi of four genes caused animals to become uncoordinated and to adopt abnormal body postures, such as becoming flattened (flattened) following RNAi of a secretory granule neuroendocrine protein-encoding gene (HE.4.05F), or becoming narrower in the middle than at the ends (hourglass) following RNAi of a tropomyosin-encoding gene (NBE.1.12G) (, ). RNAi of one gene, predicted to encode a protein similar to a hepatocellular-associated antigen (NBE.8.11C), caused animals to stick to a surface and stretch their bodies out to a very thin morphology (stick&stretch) (, ). RNAi of one gene, predicted to encode an outer dense fiber of sperm tails-like protein (NBE.8.03E), caused animals to move sideways to the right (sidewinder) (). Other genes associated with abnormal behavior are predicted to encode proteins including G-protein factors, transcription factors, and 12 novel proteins (, S1). These results assign behavioral functions to an assortment of genes and identify functions for previously uncharacterized genes. Model organism studies of animal behavior have increasingly become a powerful strategy for understanding how specific genes control neural functioning and circuitry development (Hobert, 2003
). Studies of the genetic control of behavior in varied organisms can begin to address how proteins with similar biochemical functions are utilized to control the development of diverse nervous systems and behaviors.