smedinx-11 is a planarian stem cell gap junction gene required for regeneration and homeostasis

The planarian flatworm exhibits a remarkable ability to regulate complex pattern formation, providing an exciting opportunity to understand well-conserved and widely relevant molecular controls of morphogenesis. Planarian stem cells (neoblasts) have the potential to differentiate into all of the different tissue types found in the adult flatworm, and worms show no evidence of senescence as neoblasts continually replace damaged or aging cells(Reddien and Sánchez Alvarado,2004). Thus, neoblasts must effectively integrate the signals required for self-renewal, controlled proliferation, migration,differentiation and cell incorporation into target tissues during physiological turnover and maintenance of complex pattern (morphostasis). Furthermore, after injury, neoblasts proliferate and regenerate all missing tissues in a short period of time, restoring the original form and proportion of the organism (Reddien and Sánchez Alvarado, 2004). All of these properties highlight the interactions between adult stem cells and the context of their existing differentiated neighbors (Joseph and Morrison, 2005; Li and Xie,2005; Scadden,2006; Yamashita et al.,2005). Understanding the endogenous controls of stem cell behavior is of high priority for therapeutic applications in regenerative medicine and cancer biology, as well as for the fundamental understanding of highly regulative morphogenesis. However, the molecular mechanisms of the crosstalk between stem cells and their surrounding differentiated tissues are largely unknown.

Gap junctional (GJ) communication is increasingly revealed as an important regulatory modality functioning alongside traditional secreted signaling molecules. In metazoans, gap junction proteins are conserved membrane channels involved in direct cell-cell communication via small molecule signals during numerous physiological events (Levin,2002; Nicholson,2003; Phelan,2005). Gap junction-permeable signals have been implicated in embryonic morphogenesis and neoplasm(Bruzzone et al., 1996; Levin, 2002; Mesnil et al., 2005). Importantly, GJ communication is not just a housekeeping event necessary for cell survival; several studies have revealed subtle alterations of embryonic patterning, but not generalized inhibition of cell proliferation or toxicity,following abrogation of GJ communication (GJC)-dependent signaling (reviewed in Levin, 2007). Invertebrate gap junctions are composed of proteins from the innexin family, whereas vertebrate gap junctions consist of the homologous pannexins as well as connexins. Innexins have been shown to be a crucial component for tissue polarity, morphogenesis, locomotion, electrical cell coupling, and germ-line cell maintenance and differentiation (Bauer et al., 2004; Gilboa et al.,2003; Nogi and Levin,2005; Phelan,2005; Tazuke et al.,2002); however, no specific gap junction protein has yet been functionally shown to regulate adult somatic stem cells in vivo. Moreover, no specific gap junction gene product has been implicated in the interplay of signals required during the regeneration of complex structures.

In recent work, we characterized the expression of a family of innexin genes in planaria (Nogi and Levin,2005). Crucially, however, the specific innexins involved in patterning, and the molecular mechanisms by which innexin function could regulate cell growth, fate specification and regeneration, remain completely unknown. The molecular implication of a specific innexin gene in regulative morphogenesis and stem cell behavior, as well as the characterization of the consequences of its abrogation, are the key steps that must now be addressed. Thus, we investigated the potential roles of GJ-mediated signals in regeneration and homeostasis, using the planarian species Schmidtea mediterranea as a tractable model system in which to probe novel cellular mechanisms of communication between stem cells and their niche.

We identified 12 transcripts encoding innexin proteins(smedinx-1-12). Additional qualitative and quantitative expression studies suggested that neoblast-associated innexins are present and enriched in specific adult stem cell subpopulations. Functional analysis using RNA interference (RNAi) revealed a neoblast-associated gene, smedinx-11,that is specifically required for tissue regeneration, homeostasis and neoblast maintenance. Reduction of smedinx-11 expression resulted in a phenotype that is different in several key properties from those obtained with known genetic and radiation-based treatments. Our data show that smedinx-11 is required for the normal expression of well-conserved key stem cell regulators such as PIWI and other proteins. Taken together, our data reveal a novel role for GJC proteins and characterize a new element functioning in the regulation of adult stem cells in their natural environment.

In all experiments, the clonal line CIW4 of S. mediterraneaprovided by A. Sánchez Alvarado (University of Utah, UT, USA) was used and maintained as previously described(Cebrià and Newmark,2005).

Full-length innexin gene sequences were obtained by RACE and subcloned into the pCS2+ expression vector. Full-length smedinx-11 was entered into Genbank under accession number DQ851133 (formerly AY067505). Ventral or dorsal blastomeres of four-cell Xenopus embryos were injected with in vitro-synthesized, full-length, capped mRNA encoding smedinx-11, Dugesia japonica innexin-11 (DjInx-11; the smedinx-11 homolog)or a truncated mutant of smedinx-11 (for additional details, see Fig. S2 in the supplementary material). Effects were assayed as described previously (Levin and Mercola,1998). Injection of mRNA encoding smedinx-11 or DjInx-11 into ventral blastomeres, in which cells are normally gap-junctionally isolated (Levin and Mercola, 1998), specifically randomized embryonic laterality (35%and 26% heterotaxia, respectively, n=119, P≪0.01 byχ ² test with Pearson correction); by contrast, injections into dorsal blastomeres (expressing the innexin in a zone that is already well-coupled by gap junctions) had no effect on embryonic asymmetry. Similarly, a non-functional smedinx-11 mutant did not randomize in any injection locale (P=0.3). Thus, smedinx-11 behaved in this assay identically to all of the connexin gap junction constructs tested previously (Levin and Mercola,1998).

The RNA isolation procedures and internal control used were as previously described (Reddien et al.,2005b). We used the iScript cDNA synthesis kit and iQ SYBR green supermix (BioRad); PCR reactions and analyses were performed using iCycler(BioRad).

In vitro double-stranded (ds)RNA preparation (including the 65-1469 bp fragment) and injection procedure was as previously described(Reddien et al., 2005b). The injection schedule was: a total of three injections distributed over two consecutive days, one day of rest, followed by one more day of injections (one injection per day with three pulses of 32 nl of respective dsRNA). To rule out the possibility of local neoblast depletion, microinjections in some experiments targeted either pre- or post-pharyngeal areas. In most cases,dsRNA injections were targeted to the gastro-vascular system, to assure effective dsRNA incorporation as well as homogeneous distribution throughout the body. Some animals were amputated at 8 days after first injection and,when possible (for all S. mediterranea innexin genes other than smedinx-11), a second round of injections plus amputation was performed. The control group in RNAi experiments was injected with water. Some animals remained intact for more than 3 weeks, and then were fixed and processed for immunostaining or in situ hybridization (ISH).

ISHs were performed as previously described(Reddien et al., 2005b). Once the ISH signal for positive controls was reproducible, experiments were carried out with different S. mediterranea innexin riboprobe dilutions to determine the optimal ratio of positive-signal:background. The specificity of the ISH pattern was also confirmed by three sets of experiments performed in parallel (Fig. 2A): (i) γ-irradiation (6000 rad), which resulted in the selective removal of stem cells (neoblasts), significantly reduced the signal of two S. mediterranea innexins compared to the un-irradiated worms(see below; results were additionally confirmed by qRT-PCR); (ii) sense probe,which resulted in no detectable signal; and (iii) specific RNAi exposure,which reduced the signal of the target genes.

Planarian dissociation and cellular labeling was performed as described(Reddien et al., 2005b) with only one filtration through a 35 μm Nitex filter. Propidium iodide was added (5 μg/ml) to eliminate dead cells. Labeled cell fractions were sorted using a MoFlo high-speed cell sorter and analyzed with Summit software (Dako). Combinations of Hoechst and calcein stains were used in standard protocols to separate cells on the basis of dye efflux and DNA content(Goodell et al., 1996).

Digital images were collected using a Nikon SMZ1500 microscope connected to a Cool Snap CF camera (Roper Scientific Photometrics), and with Openlab software (Improvision), a Leica TCS SP2 Spectral Confocal Imaging System and an Olympus BX61 motion-controlled compound microscope attached to a Hamamatsu Orca AG high-resolution, monochrome, cooled CCD camera with IP Labs analysis software.

Among planarian species, S. mediterranea is the most convenient model in which to perform molecular and genetic analyses(Newmark and Sánchez Alvarado,2002; Saló,2006; Sánchez Alvarado,2006; Sánchez Alvarado and Kang, 2005). Twelve cDNAs coding for innexins were identified in two public S. mediterranea databases(Sánchez Alvarado et al.,2002; Zayas et al.,2005) and numbered smedinx-1 to smedinx-12(Fig. 1A, and see Fig. S1 in the supplementary material). Phylogenetic analysis and in situ hybridization(ISH) revealed considerable similarity of sequence and expression pattern to Dugesia japonica innexins (inter-species comparison in Fig. S1 in the supplementary material). These transcripts were segregated into four categories of expression: excretory, digestive, mesenchymal and nervous tissue(Fig. 1A). Although innexin genes are now widely known as encoding gap junction proteins, we indirectly confirmed this function in a Xenopus assay (see Materials and methods and Fig. S2 in the supplementary material).

The expression of mesenchymal S. mediterranea innexin genes is widely distributed. To confirm that these patterns are correct and specific,and to conclusively establish expression of these genes, several rounds of optimization of the ISH protocol used were performed with negative (sense probe) and positive (other genes with expression patterns known to involve different tissues) (Newmark and Sánchez Alvarado, 2002; Sánchez Alvarado et al.,2002) controls (data not shown). Moreover, wild-type intact animals were sectioned (transversally and longitudinally along the anteroposterior axis) and ISH was performed on sections(Reddien et al., 2005b) to ensure that probe penetration differences did not bias the results(Fig. 1B and see Fig. S3 in the supplementary material). The results of whole-mount and section ISH under a variety of conditions consistently revealed expression in nervous, excretory and digestive systems, as well as in the mesenchyme - a tissue that contains a number of different cell types (Fig. 1 and see Fig. S3 in the supplementary material). Please see the legend to Fig. S1 in the supplementary material for a discussion of the comparison of innexin expression patterns among S. mediterranea and D. japonica.

Because neoblasts are the only proliferative cells in adult planaria,γ-irradiation provides a convenient way to selectively kill them and thus identify neoblast-associated gene expression(Guo et al., 2006; Hayashi et al., 2006; Ogawa et al., 2002; Orii et al., 2005; Reddien et al., 2005b; Rossi et al., 2006; Salvetti et al., 2000; Salvetti et al., 2005; Shibata et al., 1999). ISH and quantitative real-time PCR (qRT-PCR) revealed that the expression of only two S. mediterranea innexin genes (smedinx-11 and smedinx-2) was reduced by γ-irradiation(Fig. 2A,B and data not shown),suggesting a relationship between these S. mediterranea innexins and neoblasts.

Interestingly, the expression patterns of these two neoblast-associated S. mediterranea innexin genes revealed an additional domain not present in some neoblast markers (Guo et al., 2006; Ogawa et al.,2002; Orii et al.,2005; Reddien et al.,2005b; Rossi et al.,2006; Salvetti et al.,2000; Salvetti et al.,2005). Proliferative neoblast-associated gene expression is commonly observed throughout the mesenchyme, but not in tissue in front of the photoreceptors or in the pharynx (Guo et al., 2006; Ogawa et al.,2002; Orii et al.,2005; Reddien et al.,2005b; Salvetti et al.,2000; Salvetti et al.,2005). However, in addition to the smedinx-11 signal in mesenchymal cells, irradiation-sensitive expression of smedinx-11 was also present in areas known to be postmitotic (i.e. cells anterior to the photoreceptors, surrounding cephalic ganglia, and in the pharynx) (Figs 1, 2 and see Fig. S3A,B in the supplementary material) as well as in regenerating blastemas (see Fig. S3C in the supplementary material). Such irradiation-sensitive gene expression in areas in front of the photoreceptors and pharynx (see Fig. S3A,B in the supplementary material) has also been described for another stem cell-associated transcript, vasa-related(Shibata et al., 1999). Moreover, although the expression of both smedinx-11 and smedinx-2 was reduced after irradiation, a small irradiation-insensitive component remained(Fig. 2B).

Recent work using fluorescent activated cell sorting (FACS) identified two irradiation-sensitive subpopulations (X1 and X2) with neoblast-like morphology(Hayashi et al., 2006; Reddien et al., 2005b). The X1 subpopulation in S. mediterranea consists mainly of dividing neoblasts expressing the piwi genes smedwi-1 and smedwi-2,and smedcyclinB; however, no marker or functional characterization are yet available for the X2 population (which consists of both irradiation-sensitive and -insensitive cells). Gene expression experiments using FACS-isolated cells demonstrated that smedinx-11 is mainly expressed in the X1 and X2 compartments, and its expression levels are specifically enriched in X2 cells (Fig. 2C).

Together, smedinx-11 quantitative and qualitative expression analyses in FACS-isolated cells also revealed that this gene is expressed at low levels in most (∼60%) proliferative neoblasts (X1), whereas, in X2 cells, in which the total smedinx-11 levels are higher, the number of cells expressing this gene only reach approximately 24%(Fig. 2C,D). Because there is more total expression of smedinx-11 in X2 cells than in X1 cells, but fewer X2 cells express detectable smedinx-11 mRNA, we suggest that smedinx-11 is expressed at high levels in a subset of X2 cells.

The expression of S. mediterranea piwi genes is strongly concentrated in proliferative neoblasts (∼90%)(Reddien et al., 2005b),suggesting that the expression in many X1 cells must overlap for S. mediterranea piwi genes and smedinx-11. To investigate whether smedinx-11 expression in X2 cells might be related to other neoblast markers, we stained FACS-isolated cells with a recently characterized SMEDWI-1 antibody that recognizes postmitotic irradiation-sensitive cells(Guo et al., 2006). In both X1 and X2 cell groups, over 90% were positive for SMEDWI-1 protein (see Fig. S4 in the supplementary material), suggesting that smedinx-11 expression in X2 neoblasts overlaps with SMEDWI-1-positive cells. Together, these quantitative and spatial expression data suggest that smedinx-11represents a novel marker for irradiation-sensitive cells that provides a molecular entry-point into understanding the differences among stem cell subpopulations (see also the legend for Fig. S1 in the supplementary material).

To investigate whether S. mediterranea innexin proteins regulate neoblast biology, RNAi experiments were performed(Sánchez Alvarado and Newmark,1999) for each S. mediterranea innexin clone, and the effects on regenerating fragments and intact animals were evaluated(Reddien et al., 2005a). Although five sequences encoding S. mediterranea innexin(smedinx-5 to smedinx-9) were previously included in an RNAi screen, no significant phenotypes associated with regeneration or homeostasis were described (Reddien et al.,2005a). We used RNAi to screen all S. mediterraneainnexin genes and found that only two out of 12 (smedinx-10 and smedinx-11) produced a detectable phenotype (100% penetrance) in both regenerating and intact worms. smedinx-10 is expressed in the excretory system and its abrogation produced the disruption of body homeostasis, revealed by a conspicuous swelling followed by subsequent lysis of the animals within 10 days (data not shown). The swelling phenotype,together with the spatial expression of smedinx-10, might suggest that loss of innexin function results in a failure in the excretory system to maintain efficient osmoregulation.

By contrast, smedinx-11 RNAi treatment [referred to as smedinx-11(RNAi)] in regenerating worms strongly inhibited blastema formation (Fig. 3A, the `early'phenotype - inability to regenerate) with no toxicity or other visible abnormalities and normal wound healing. In intact planarians, smedinx-11(RNAi) did not produce any external abnormality until approximately 14 days after RNAi exposure (the `late' phenotype, Fig. 3B and Fig. 7A-D). In both regenerating and intact worms, visible signs of the late phenotype began with tissue bending, restricted to the pre-pharyngeal area, which progressed and extended through the borders of the body with the exception of the anterior end (Fig. 3B, white arrows). Curling around the ventral surface and the inability to regenerate have been associated with impaired function of neoblasts following γ-irradiation or RNAi for several genes, including smedwi-2(Guo et al., 2006; Reddien et al., 2005a; Reddien et al., 2005b). Although regression of the tissue in front of the photoreceptors (head regression) was not observed even after 1 month following the first injection of smedinx-11 dsRNA, all other phenotypes of irradiated worms(Reddien et al., 2005b) were recapitulated. These results suggest that smedinx-11 is required for proper neoblast function.

To link innexin gene function with known molecular pathways and to further characterize the smedinx-11(RNAi) phenotype, we studied whether smedinx-11 is required for the proper expression of neoblast markers(Reddien et al., 2005b). qRT-PCR analyses showed that, at 7 days after smedinx-11(RNAi),smedwi-1 expression was reduced, whereas smedwi-2 expression was slightly increased; but, by 1 week later, the expression of both S. mediterranea piwi genes was severely reduced(Fig. 4A). We characterized the spatial expression of the S. mediterranea piwi genes, smedcyclinB and smedbruno-like (smedbruli) in smedinx-11(RNAi) worms. Expression of these neoblast markers disappeared gradually (Fig. 4B,C), consistent with the quantitative assay data(Fig. 4A). By contrast, markers of tissues other than neoblasts (S. mediterranea innexins expressed in excretory, nervous and digestive tissue) did not disappear in smedinx-11(RNAi) worms at 14 days after RNAi(Fig. 4C). Interestingly, smedbruli expression, which is observed in neoblasts and differentiated tissues (Guo et al.,2006), was strongly reduced in the neoblast compartment but remained in differentiated CNS cells (Fig. 4C). Although it is not known whether the regulation of other neoblast genes by smedinx-11 is cell-autonomous, taken together,these data firmly establish smedinx-11 as an endogenous regulator of neoblast biology (it is required for the proper expression of key stem cell genes in the piwi, cyclin B and bruno families).

The inability to regenerate within the first 7 days after smedinx-11(RNAi) demonstrates that the smedinx-11 phenotype actually begins well before external signs are visible (the early phenotype). Within the first week after smedinx-11(RNAi), animals failed to form a regeneration blastema despite a slight increase in smedwi-2expression and the presence of mitotic activity. Thus, the early obligate role of smedinx-11 in regeneration does not function via smedwi-2but is required to form a blastema upon amputation. By 14 days after smedinx-11(RNAi), there was a strong reduction in the expression of S. mediterranea piwi genes. By contrast, 8 days after smedwi-2(RNAi) exposure [when smedwi-2(RNAi) animals are unable to regenerate], smedinx-11 expression was not affected (see Fig. S5 in the supplementary material). This epistatic analysis strongly suggests that smedinx-11 function is required for the expression of the neoblast genes smedwi-1 and smedwi-2 during the late phenotype. It is possible that a significant component of the reduction in the expression of neoblast markers might result from a physical loss of neoblasts after smedinx-11(RNAi). However, the simple possibility that the loss of marker expression is due entirely to the disappearance of expressing cells is refuted by the upregulation of smedwi-2 following smedinx-11(RNAi); instead, this indicates the control of gene expression by SMEDINX-11 function. Taken together, our results suggest that SMEDINX-11 activity is required for the normal expression of S. mediterranea piwi genes in the animal and has additional consequences that are not dependent upon smedwi-2.

FACS analysis of smedinx-11(RNAi) worms demonstrated that the decline in neoblast markers was accompanied by a reduction of neoblast subpopulations (≥10 days after RNAi)(Fig. 4D). At 10 days after smedinx-11(RNAi), the number of dividing neoblasts was reduced,whereas the X2 subpopulation was not affected in this way(Fig. 4E). However, as the phenotype progressed, the X2 irradiation-sensitive subpopulation became strongly reduced (comparable to the irradiated group), a phenomenon observed only after γ-irradiation and in the late phases (≥18 days) of smedwi-2 RNAi treatment (Reddien et al., 2005b) (Fig. 5). The time-dependent way in which neoblast subpopulations disappear in smedinx-11(RNAi) animals suggests that smedinx-11 is crucial in order to maintain a pool of proliferative stem cells and to support continuity of its postmitotic progeny. Thus, these data reveal functional, molecular differences between X1 and X2 neoblast subpopulations. Based on these expression and functional studies, we propose that smedinx-11 is required for stem cell maintenance and that both neoblast subpopulations require smedinx-11 in order to participate in body homeostasis and regeneration.

Strikingly, after smedinx-11(RNAi), the mitotic activity of neoblasts disappeared in anterior-posterior (AP) progression(Fig. 6A). We quantified three regions (pre-pharyngeal, pharyngeal and post-pharyngeal, Fig. 6B)(Reddien and Sánchez Alvarado,2004) and counted mitotic activity in each region(Reddien et al., 2005a; Reddien et al., 2005b). In wild-type animals, we detected an intrinsic AP gradient of mitotic activity(more dividing cells in anterior than in posterior areas, Fig. 6C), consistent with previous reports on the distribution of neoblasts and mitotic activity(Baguñà, 1976; Brondsted, 1969). However, we noted a sharp reduction in mitotic activity in pre-pharyngeal and pharyngeal,but not in post-pharyngeal, regions within the first week after smedinx-11(RNAi). Eventually, as the phenotype progressed, the reduction in mitotic activity also reached the post-pharyngeal region, ending with the depletion of mitotic cells in the whole animal after 14 days following smedinx-11(RNAi). A similar microinjection schedule targeting only post-pharyngeal areas was performed to test whether the depletion of mitotic activity in the anterior areas was due to a local neoblast response to repair the wounds caused by the three successive rounds of microinjections. In both cases (microinjections in the pre- or post-pharyngeal area, data not shown), mitotic activity disappeared in the same AP spatiotemporal fashion, suggesting that this is a specific response to the functional abrogation of SMEDINX-11 and not because of localized effects at the injection site. These data also revealed that, even though the anterior region normally possesses more mitotic cells than the posterior region in the worm, this is also the area with the highest propensity to lose mitotic activity after the loss of SMEDINX-11 function. Remarkably, this pattern of mitotic-activity disappearance has not been reported for any type ofγ-irradiation treatment or RNAi of genes required for stem cell maintenance, and is unique for the reduction of SMEDINX-11.

Our functional data indicate that smedinx-11 is a neoblast-associated gene that regulates somatic adult stem cells during self-renewal, regeneration and homeostasis(Fig. 7A-C). The smedinx-11(RNAi) phenotype differs in important ways from the loss of regenerative ability following knockdown of other previously characterized transcripts (Guo et al., 2006; Reddien et al., 2005b; Salvetti et al., 2005). First,in known genes required for homeostasis, animals usually die <1 month after first exposure to RNAi. By contrast, smedinx-11 inhibition did not lead to head regression; fragments of pre-existing tissue survived without brains or other head organs for over 1 month (verified by synaptotagmin expression, data not shown). This long survival will enable the characterization of morphallaxis and the transitions among stem cell states. The way in which irradiation-sensitive cells disappear after smedinx-11(RNAi) is unique to this phenotype (i.e. first proliferative cells, followed by the X2 population). Moreover, although smedwi-2 is already known to be required for regeneration, abrogation of this gene seems not to affect stem cell maintenance - a stark contrast to smedinx-11. Lastly, the reversion of the AP gradient of proliferative cells that we observed after smedinx-11(RNAi) has not been described for any previous RNAi phenotype or γ-irradiation treatments.

One major consequence of the smedinx-11(RNAi) phenotype is the inability to regenerate (despite the presence of cell proliferation and the lack of early toxicity), unlike in bipolar heads previously described in D. japonica using heptanol (Nogi and Levin, 2005). This probably reflects the different dynamics of GJ closure caused by the two techniques: external heptanol exposure targets different gap junctions non-selectively, changing the open-state probability in cells accessible to the drug; by contrast, RNAi abrogates smedinx-11-dependent GJC. The RNAi is specific for one innexin transcript and thus allows us to separately characterize GJ signals involved in AP polarity versus stem cell regulation. Combinatorial analysis of individual innexins will be necessary to understand these distinct roles and to determine whether the lack of phenotype from RNAi of the other innexin family members might indicate that the roles of individual innexins can be compensated for by other innexin family members (as is evident in the connexin literature on vertebrate gap junctions in patterning).

Because smedinx-11 and smedinx-2 expression was irradiation-sensitive (Fig. 2A,B), these postmitotic S. mediterranea innexin expression domains are likely to reveal committed progeny of neoblasts, in addition to cells expressing the known smedwi-1 and smedwi-2markers (dividing neoblasts) (Reddien et al., 2005b). The smedinx-11 gene expression pattern in X1 and X2 cells opens an important opportunity to dissect the transition between proliferative and postmitotic neoblasts. Moreover, expression studies for smedinx-11 revealed its irradiation-sensitive expression in postmitotic areas (Fig. 2 and see Fig. S3 in the supplementary material), its presence in X1 with enrichment in X2 cells (Fig. 2C,D) and its overlapping expression with other neoblast markers (see Fig. S4 in the supplementary material). Together with the way in which the X1 fraction disappears before the X2 cells after smedinx-11(RNAi) (Figs 4, 5), these data are suggestive of a progression between X1 and X2 cells. Therefore, one likely possibility is that X2 represent the postmitotic progeny of the proliferative X1 compartment[consistent with our observation that both smedinx-11 and SMEDWI-1 are observed in irradiation-sensitive postmitotic areas(Guo et al., 2006), and that,whereas SMEDWI-1 protein is expressed in both X1 and X2, its mRNA is expressed mostly in X1 cells]. Future work will further elucidate the complexity and composition of the X2 cells and the possible roles of SMEDINX-11-mediated signaling as a component of the transition of stem cells between sub-population states (Fig. 7D).

Analysis of smedinx-11(RNAi) also revealed unique effects on the endogenous neoblast AP gradient (Fig. 6). Because we detected no obvious AP gradient in smedinx-11 expression, the higher number of neoblasts in the pre-pharyngeal area might be due to the anterior regions of planarians possessing a higher density of complex organs and systems that are subject to greater cellular turnover compared with posterior regions. The neoblasts anterior to the pharynx support more-varied tissues and/or tissues with higher cellular demand. Thus, after smedinx-11(RNAi) (which limits the available neoblast pool), the anterior regions might deplete the neoblasts more rapidly than posterior tissues, consistent with the requirement of neoblast maintenance for smedinx-11. Another possibility is that the propensity of neoblasts to respond to gap junction-mediated signal differs in a graded manner along the AP axis. This AP progression confirms the existence of AP gradients that were hypothesized to exist by classical workers in the regeneration field (Brondsted,1969), provides a molecular entry-point into the known but still not understood relationship between position along the AP axis and proliferation rates of neoplastic and normal cells in mammals(Auerbach and Auerbach, 1982),and suggests that a primary component of physiological gradients are the neoblasts themselves. These data additionally demonstrate that the control of neoblast proliferation by gap junction signals functionally interacts with the morphogenetic polarity of the animal; future work must address the mechanisms integrating neoblast behavior with large-scale axial patterning.

Innexins are widely conserved membrane channels mediating regulatory signals; we propose that smedinx-11-containing gap junctions provide a conduit for small-molecule messengers that control the maintenance (X1 subpopulation) and migration/differentiation of the progeny of proliferative neoblasts (the X2 subpopulation). Although the primary cells expressing and dependent on smedinx-11 are the neoblasts, this effect is probably not cell-autonomous; we are currently pursuing the plausible hypothesis that heterotypic gap junctions allow communication between neoblasts and their surrounding cells.

Abrogating smedinx-11 function partially isolates neoblasts from their environment; their subsequent disappearance suggests that, within the X1 population, gap junction-mediated signals might inform the neoblasts that they are in the midst of differentiated tissue. Future work will focus on identifying the small-molecule or electric-current signals that traverse smedinx-11 gap junctions, but this mechanism is of general significance to stem cell biology. Innexin homologs (pannexins) have recently been identified in mammals (Baranova et al., 2004; Bruzzone et al.,2003); human pannexin 2 (PANX2) has been associated with bone marrow stem cells (Baranova et al.,2004), and gap junctions have been described in human embryonic stem cells and neural progenitors (Cheng et al., 2004; Leung et al.,2002; Pearson et al.,2005; Rozental et al.,1998; Wong et al.,2004). In vertebrates, connexin43 functions in the proper targeting of neural crest cells (Xu et al., 2006) - a cell population heavily involved in embryonic morphogenesis of diverse body regions.

Our results in a regenerating system confirm in vitro studies in mammalian cells that suggest that gap junction expression is required to maintain proliferative state in neural progenitor cells and that the expression of gap junction-specific genes (e.g. connexin43) decreases as neuronal differentiation takes place (Cheng et al.,2004; Leung et al.,2002; Rozental et al.,1998). Although our data implicate gap junctions in the control of somatic, adult stem cells in a morphogenetic context for the first time, a number of recent studies have indicated that GJC is crucial for germ cell function in Drosophila (Gilboa et al., 2003; Tazuke et al.,2002), suggesting that GJ signaling is likely to be broadly important in stem cell regulation.

Taken together, these findings suggest that similar GJ-dependent mechanisms might be widely relevant for controlling the behavior of migratory, plastic cells. Moreover, tantalizing mammalian data(Gersdorff Korsgaard et al.,2001; Goodell et al.,1996) have suggested that ion transport is a hallmark of stem cell state, but functional roles have not yet been probed. Likewise, exciting recent data have linked ion and small-molecule transport to regenerative ability in vertebrates (Adams et al.,2007; Nuccitelli,2003; Zhao et al.,2006). Thus, GJC might provide an ancient and powerful mechanism for mediating the complex interplay of regulatory signals among differentiated and stem cells in multicellular systems(Sánchez Alvarado and Kang,2005). Further molecular analyses of GJ roles in planaria and vertebrate systems will provide crucial opportunities to understand the evolution of morphogenetic controls as well as identifying novel therapeutic targets for manipulating stem cell behaviors in biomedical settings.

We are grateful to: A. Sánchez Alvarado for providing the S. mediterranea CIW4 strain and innexin clones used in this work; P. Reddien for providing S. mediterranea piwi and smedcyclinB clones;P. Newmark and T. Guo for anti-SMEDWI-1 antibody and bruli clone; F. Dewhirst, D. Adams, P. Yelick, S. Rittlingand M. Montgomery for comments on the manuscript; D. Qiu for molecular biology; P. Koustubhan, D. Sorocco, A. Gandhi, and K. Gallant for laboratory assistance; G. Eisenhoffer, J. Jenkin and B. Pearson for FACS advice; J. LaVecchio, B. Girijesh, the Joslin Diabetes Center and the Harvard Stem Cell Institute Flow Cytometry Cores for FACS analysis; P. Walentek for planarian care; Y. Hosogi for qRT-PCR advice; and J. Morokuma for assistance with phylogenetic analysis. N.J.O. thanks Oliver and E. Oviedo-Pfister for their encouragement and the American Physiological Society for a fellowship (Professional Skills Course). N.J.O. was supported by NIH under a Ruth L. Kirschstein National Research Service Award: F32 GM078774. This work was also supported by NSF grant IBN#0347295, NHTSA grant DTNH22-06-G-00001 and NIH grant R21 HD055850 to M.L. This research was conducted in a Forsyth Institute facility renovated with support from Research Facilities Improvement Grant Number CO6RR11244 from the National Center for Research Resources, NIH.