A functional genomics screen identifies an Importin-? homolog as a regulator of stem cell function and tissue patterning during planarian regeneration

Changes in gene expression during head repair and regeneration

As a first step in our screen, we measured changes in gene expression on days two,
three, and four of regeneration following amputation of one side of the head. We chose
to focus on the head and these time points because this is a region and period where
many different cell and tissue types differentiate and pattern relative to one another;
we were also interested in studying the regeneration and patterning of the CNS. The
nascent CNS can be visualized with the neural marker anti-Synapsin by day three of
regeneration and has connected to the cephalic ganglion on the non-amputated side
of the head by day four (Fig. 1a). Removing only one side of the head allowed us to examine mechanisms of repair in
addition to those required for de novo regeneration of the entire head; we also aimed
to identify molecular interactions between the blastema and the non-regenerating tissue
that was left behind on the opposite side. Figure 1b outlines our sample collection procedure. We amputated half of the head, saved the
tissue that was removed as the non-regenerating control sample, and then collected
samples from the blastema and “opposite side” on days two through four of regeneration.
We extracted RNA from each of these samples, reverse transcribed the RNA to cDNA and
submitted these samples to the GeneChip Microarray Core at University of California
San Diego for labeling and hybridization to custom Nimblegen microarrays 33], 34]. Analysis of the resulting data revealed 637 genes with significant (p???0.05) changes
in gene expression in one or more of the blastema or opposite side samples compared
to control (Additional files 1 and 2). Of these, 420 genes were upregulated in the blastema and 179 were downregulated.
For the majority of the upregulated genes, expression peaked on day two and then declined
over the rest of the time course, but a cluster of 74 genes remained highly upregulated
throughout (Fig. 1c). On the opposite side of the head, we found 23 genes with increased expression and
33 with decreased expression; for 38 of these genes, the change in expression was
unique to the opposite side tissue and not mirrored in the blastema.

Fig. 1. Identification of genes differentially expressed during head repair and regeneration.
a Half-head regeneration. Anti-Synapsin antibody was used to label the central nervous
system of intact animals and animals that had regenerated for two to 4 days following
half-head amputation. Samples were imaged from the ventral side with anterior to the
left, and the amputated portion of the head in regenerates is toward the bottom. cg?=?cephalic
ganglia; vnc?=?ventral nerve cords. Asterisks mark the pharynx, and triangles point
to where the nervous system is regenerating in the blastema. Scale bar?=?0.5 mm. b Sample collection strategy. One side of the head was removed from intact animals
and saved as the non-regenerating control sample. After two to four days of regeneration,
samples were collected from the blastema and non-regenerating opposite side of the
head. c Heat map summary of changes in gene expression in the blastema and other side of
the head during a time course of half-head regeneration. Low expression is represented
in blue and high in red. Control?=?non-regenerating samples. Day 2 b, day 3 b, and
day 4 b?=?blastema samples collected on day two, three or four of regeneration. Day
2 os, day 3 os, and day 4 os?=?samples collected from the opposite side of the head
on day two, three, or four of regeneration. d Categorization of differentially expressed genes based on the function of their homologs
in other species using Clusters of Orthologous Groups. Red bars show the number of
upregulated genes in each functional group in either the blastema or opposite side
tissue relative to control. Blue bars represent downregulated genes

To begin characterizing the differentially expressed genes, we grouped them into functional
categories based on homology using Eukaryotic Clusters of Orthologous Groups (KOGs)
(Fig. 1d). Approximately half of the genes (332/637) could be definitively assigned to a KOG
group, and the others were unclassifiable due to low sequence conservation. The genes
upregulated in the blastema were divided across many functional groups. The most highly
represented category was “Posttranslational modification, protein turnover, chaperones”,
which included homologs of MEPRIN A metalloproteinase and other proteases, proteasome
subunits, and heat shock proteins. The category of “Replication, recombination and
repair” was the next most common and contained several DNA replication licensing factors
(MCM2, 3, 4, 5 and 7), exo- and endonucleases involved in DNA repair, and cell cycle
protein CDC45, among others; representation of genes in this category is likely due
to the inclusion of some postblastema tissue in our samples. Notably, the “Transcription”
category, which was of interest to us because it contains proteins such as transcription
factors and chromatin regulators likely to play important roles in controlling changes
in gene expression during regeneration, was also highly represented among the differentially
expressed genes. Nearly half (28 out of 62) of the annotated downregulated genes and
almost as many upregulated genes were identified as functioning in “Signal transduction
mechanisms”. Several genes potentially involved in neural function were among those
downregulated in the signal transduction class, including homologs of three human
GABA receptors (GBRB2, GBRB3, and GABRR2) and a sodium-dependent noradrenaline transporter,
SLC6A2.

In samples from the side of the head opposite to the blastema of regenerates, differentially-expressed
genes again fell into diverse functional categories. Genes upregulated in the opposite
side tissue but not in the blastema included a homolog of KCNN3, a small conductance
calcium-activated potassium channel with neuronal function in humans. “Signal transduction”
was again the most highly represented functional category among the downregulated
genes and included Wnt signaling protein Frizzled and homologs of an acetylcholine
receptor and a serine/threonine kinase. Given that most cell types in the animal are
contained within the samples we collected, we expected to find a broad range of functional
categories represented in our expression data.

Identification of genes expressed in the blastema, neoblasts, and CNS

Based on our expectation that genes involved in head regeneration and repair would
be expressed in the blastema, neoblasts, or CNS, we further analyzed the genes identified
from our microarray study by whole-mount in situ hybridization (WISH). We performed
these experiments to determine which cell and tissue types they are expressed in and
to validate our microarray results. Additional file 3 provides a list of the genes tested and a summary of the results. Among 260 genes
that were identified as upregulated in the blastema by our microarray analysis, we
found 243 that were expressed at a higher level in the blastema than in the non-regenerating
tissue by WISH. For some, the increased expression was tightly confined to the blastema
itself (e.g., F-box and leucine rich repeat 4), whereas others were strongly expressed both in the blastema and in the area beneath
it (e.g., monocarboxylate transporter) (Fig. 2a). Sixty genes appeared strongly upregulated in the blastema based on in situ hybridization,
but surprisingly only one of these fell within the group of genes showing the largest
fold-change by microarray (boxed region in Fig. 2b). This can be explained by noting that in non-regenerating animals, many of the genes
with the largest fold-change were detected almost exclusively in the secretory cells
surrounding the pharynx – any expression in the blastema represented a large change
compared to the control head region, even if the resulting expression was not particularly
strong. Taken together, the in situ staining validated the microarray results with
regard to blastema expression.

Fig. 2. Identification of genes expressed in tissues of interest and validation of microarray
results. Representative examples from the WISH screen. a Blastema expression. The lower half of the head of the animal on the right in each
pair was removed two to four days before fixation, on the day of greatest expression
change for each gene. Arrows indicate the blastema. b Volcano plot of p-value vs. fold change for blastema to control comparison colored
to indicate genes with no increased expression in the blastema (red), moderate blastema expression (green; e.g., F-box and leucine rich repeat 4) or strong blastema expression (blue; e.g., monocarboxylate transporter). c Neoblast expression. The animals on the right were irradiated (60 Gy) 3 days before
staining to destroy the neoblasts. The animals on the left are non-irradiated. d Volcano plot colored to indicate neoblast expression; green marks genes detected
in neoblasts and red marks non-neoblast genes. Genes were scored as expressed in neoblasts
if staining was reduced following irradiation, as for histone h2a and jerky homolog-like. e CNS expression. Arrows and arrowheads indicate expression in the cephalic ganglia
and ventral nerve cords, respectively. f Volcano plot colored to indicate genes with no CNS expression (red), weak CNS expression (green; e.g., Rb-like protein) or strong CNS expression (blue; e.g., prohormone-4, neurensin-1, and thioredoxin). g Venn diagram showing overlap of expression between tissues of interest. h Examples of genes not expressed in the blastema, neoblasts, or CNS, but in the epidermis
(fam166b protein), intestine (Contig3907_SE3), secretory cells (Contig7601_SE3), or other discrete cells (mannosyl phosphorylinositol ceramide synthase sur1). i Examples of WISH to genes with proposed opposite side of the head expression. The
lower half of the head of the animal on the right in each pair was amputated 2 to
4 days before fixation. The arrow indicates cephalic ganglia expression. Animals were
imaged ventrally, anterior to the left. Scale?=?0.5 mm

To identify genes expressed in the neoblasts, we performed WISH in control animals
and planarians treated with ?-irradiation to destroy the neoblasts 5], 47], 48]. We found 139 genes with reduced staining in the irradiated samples. Some of these
genes were expressed only in the neoblasts (e.g., histone H2A), whereas others were also strongly expressed in other tissues (e.g., jerky homolog-like), including the CNS and intestine (Fig. 2c). The vast majority of neoblast-expressed genes were also expressed in the blastema
(127 out of 139 genes; Fig. 2g), and many were found among the genes identified by microarray to be upregulated
in the blastema (Fig. 2d).

We identified 81 genes expressed in the CNS exclusively or in combination with other
tissues. Some genes were expressed in the CNS at levels similar to their expression
in other tissues and were classified as simply having CNS expression, whereas others
were more highly expressed in the CNS and were classified as “strong CNS” genes (Fig. 2e and f). Genes with strong CNS expression were enriched among those that appeared downregulated
in blastema samples compared to control on the microarrays (Fig. 2f). This is likely not due to actual downregulation but more representative of genes
that are highly expressed in the cephalic ganglia of intact animals and that are not
strongly upregulated in the blastema. Among the genes with strong CNS expression,
some appeared pan-neuronal (e.g., neurensin-1) whereas others were expressed in a subset of neuronal cells or tissues (e.g., prohormone-4 and thioredoxin) (Fig. 2e). Besides genes expressed in the blastema, neoblasts and/or the CNS, 52 genes were
found primarily in other tissues, including the epidermis, intestine, secretory cells,
or other discrete cells (Fig. 2h).

We also performed in situ hybridization to 19 genes that were identified by microarray
to have changed in expression only on the opposite side of the head (versus also changing
in the blastema) during regeneration. Of these, seven were identified as potentially
upregulated in the opposite side tissue. We did not observe an obvious increase in
expression in the opposite side of the head over that of the surrounding tissue, although
there may have been a more widespread increase in expression that included this region.
The staining pattern for several of these genes (5 of 7) was similar to that of the
Ca ²⁺ activated K ⁺ channel shown in Fig. 2i, with expression mainly in the mesenchyme. None of these seven genes were expressed
in the CNS. In contrast, the majority of genes (7 of 12) identified as downregulated
in the opposite side tissue showed strong CNS expression, as exemplified by Contig3034_SE3
(Fig. 2i).

In total, we characterized the expression patterns of 390 genes, most of which had
not been previously reported. We have created a publicly accessible database to house
our in situ data. Images of the expression patterns of all genes analyzed in our screen
are available at http://planaria.sdsu.edu.

A targeted RNAi screen to identify genes required for regeneration

We chose 156 genes from those differentially expressed during head regeneration to
knock down by RNAi (Additional file 4). We gave higher priority to genes that were expressed in tissues of interest (blastema,
neoblasts, CNS) from our WISH experiments or that we hypothesized would function in
head/CNS regeneration based on their homology. Planarians were fed double-stranded
RNA (dsRNA) twice per week for either three or six feedings and then amputated anterior
to the pharynx and observed through 14 days of regeneration (Fig. 3a). Knockdown of 25 genes resulted in phenotypes (Table 1), which included stem cell loss, impaired regeneration, effects on patterning or
differentiation, and defects affecting specific tissues such as the photoreceptors
and pharynx. The results for genes in each of the phenotype categories are described
below.

Fig. 3. Representative images of phenotypes observed following RNAi knockdown. a RNAi feeding schedules. Phenotypes in panels (b), (c), and (g) appeared after three feedings, panel (l) after 12 feedings, and the remainder after six feedings. b–c Loss of neoblasts in Smed-slbp(RNAi). Arrow indicates reduced-size blastema. Anti-PH3 labels the mitotic neoblasts. Live
animals were imaged on day seven of regeneration, and stained animals were fixed on
day 14. d–e Phenotypes of Smed-dkc1(RNAi). Arrows indicate missing blastemas. d?=?day seven following pre-pharyngeal amputation, e?=?ventral curling and lysis on day 14. f Impaired regeneration following six Smed-rbap46/48-2 dsRNA feedings. Animals imaged on day seven. g Abnormally elongated photoreceptor pigment (yellow triangles) in day seven regenerates following three feedings against Smed-rbap46/48-2. h Asymmetric photoreceptors of Smed-mcm2(RNAi). The red triangle indicates underdeveloped photoreceptor in day six regenerates.
i Reduced blastema (arrow) and underdeveloped photoreceptors (red triangles) in day six Smed-fen1(RNAi) regenerates. j Reduced blastema (arrow) and forked tail (blue triangle) in day seven Smed-ptbp1(RNAi) regenerates. kSmed-tph(RNAi) animals lacking photoreceptor pigment on day 14 of regeneration. Red triangles indicate
missing pigment. l Reduction of photoreceptor pigment in Smed-ddc(RNAi). Red triangles indicate photoreceptor pigment cups. m Elongated body shape following extended Smed-tph RNAi. n Edema following Smed-gas8 RNAi. The animal shown is affected in the posterior. o–qSmed-pgbd4(RNAi) phenotypes. White arrows indicate dorsal humps in uninjured animals (o) and day 14 regenerates (p). The blue triangle indicates a post-pharyngeal lesion, and the green arrowhead indicates
lateral bulging. q WISH to pharynx marker laminin in Smed-pgbd4(RNAi). Upper animals are shown ventrally, and lower animals were imaged from the side,
with dorsal toward the top. Anterior is to the left. Yellow arrows and arrowheads
indicate the base of the pharynx and its ventral opening, respectively. Brackets indicate
the pharynx in controls. Scale?=?0.5 mm for all panels

Table 1. Summary of 25 genes producing RNAi phenotypes

Differentially expressed genes required for neoblast survival/maintenance

Knockdown of 14 genes resulted in phenotypes associated with loss of the stem cells,
which include impaired blastema formation, head regression, lesions, ventral curling
and lysis (Table 1 and Fig. 3b, d and e). Staining with anti-phospho-Histone H3 (PH3) revealed a striking reduction in the
number of mitotic cells relative to controls following RNAi against four of the genes
that had the “loss of stem cells” phenotype (Fig. 3c, marked in Table 1); staining was not performed for the other genes in this group, in most cases due
to death of the animals by lysis prior to the end of the two week observation period.
All 14 of the genes in this phenotypic class were upregulated in the blastema based
on both microarray and WISH data, and 12 of them (all except Smed-smarcc-1 and PL08006B2E08) were expressed in the stem cells. These data suggest that these genes play important
roles in the survival or maintenance of planarian stem cells.

Genes required for blastema formation and regenerative capacity

Knockdown of five other genes resulted in a general reduction in regeneration ability
without the dramatic loss of stem cells observed in the previous group. For each of
these genes, blastemas developed but were smaller than those in controls, and photoreceptors
were delayed in formation or underdeveloped relative to controls at the same time
point (Fig. 3f-j). There was a three- to four-fold decrease in the number of mitotic cells following
knockdown of Smed-rbap46/48-2 (to 75?±?24 cells/mm
²
vs. 267?±?70 cells/mm
²
for gfp(RNAi), n?=?8 animals/gene) and Smed-mcm2 (to 69?±?28 cells/mm
²
vs. 292?±?70 cells/mm
²
for gfp(RNAi), n?=?8 animals/gene).

The impaired regeneration phenotype was accompanied by additional defects in Smed-rbap46/48-2(RNAi), Smed-mcm2(RNAi) and Smed-ptbp1(RNAi) animals. In contrast to the underdeveloped photoreceptors observed following six
dsRNA feedings targeting Smed-rbap46/48-2, shortening the treatment to three feedings uncovered a defect in photoreceptor morphology
in which the pigment cups appeared abnormally elongated at early stages of regeneration
(Fig. 3g). In some Smed-mcm2(RNAi) animals (n?=?9/30 tails after six RNAi feedings), the photoreceptors were asymmetric in size
or only one photoreceptor fully developed (Fig. 3h). Finally, following knockdown of Smed-ptbp1, head fragments regenerating a new tail sometimes developed a forked blastema with
two separate points rather than the typical tapered tail (14/30 heads after 6 RNAi
feedings; Fig. 3j), and both head and trunk fragments displayed abnormal inching movements. The knockdown
results for this group of genes as a whole suggest they may act in cell differentiation
and/or tissue patterning during regeneration.

Other defects and effects on specific tissues or cell types

Knockdown of five other genes resulted in phenotypes distinct from those described
above in that they did not affect general stem cell function or blastema formation
but instead had more specific effects. Two of these genes, dopa decarboxylase (Smed-ddc) and tryptophan 5-hydroxylase (Smed-tph), were required for production of photoreceptor pigment; in trunks regenerating their
heads, Smed-tph(RNAi) animals lacked photoreceptor pigment entirely (Fig. 3k), and the pigment cups in Smed-ddc(RNAi) animals were visible as faint brown spots rather than the usual dark black color
(Fig. 3l). Both Smed-tph and Smed-ddc are expressed in the pigment cup (Additional file 5) 20]. In addition to the photoreceptor phenotype described above, extended knockdown of
Smed-tph caused the worms to become longer and thinner (Fig. 3l). The length-to-width ratio of non-regenerating Smed-tph(RNAi) animals after 12 dsRNA feedings was 8.4?±?1.2 compared to 5.6?±?0.6 for gfp(RNAi) controls (students’ t-test p-value??0.0001). These longer Smed-tph(RNAi) animals also displayed abnormal inchworm-like movement rather than the usual gliding.

RNAi against growth arrest-specific protein 8 (Smed-gas8) led to edema (Fig. 3n), a phenotype typically associated with dysfunction of the protonephridia. Finally,
we observed movement defects (inching) following knockdown of a homolog of B9 domain-containing protein 2 (Smed-b9d2). This phenotype was most noticeable shortly after the final dsRNA feeding in both
uninjured and regenerating animals and seemed to wear off over time after the RNAi
feedings stopped.

Knockdown of a protein with weak homology to PiggyBac transposable element-derived protein 4 (Smed-pgbd4) led to the formation of a raised hump over the pharynx and a single lesion at the
posterior end of the pharynx on the dorsal side of the animal (Fig. 3o). In some (9/50 trunks after six RNAi feedings) regenerating animals, we also observed
lateral bulges near the pharynx (Fig. 3p). Staining with the pharynx marker laminin in Smed-pgbd4(RNAi) trunk fragments after 14 days regeneration or in uninjured animals following a period
of 10 days starvation after the final dsRNA feeding revealed that the pharynx was
lost after this treatment (Fig. 3q).

The observation that knockdown of genes in this group leads to specific defects rather
than a general loss of regenerative capacity suggests that their products may act
in pathways that direct regeneration or maintenance of particular tissues or cell
types, including the photoreceptors, ciliated cells and the pharynx.

Planarians have two homologs of importin-? with distinct expression patterns and function

The final gene that produced a phenotype upon knockdown in our screen was a homolog
of the nuclear transport factor Importin-? that we named Smed-ima-1. Knockdown of this gene led to defects in differentiation and patterning of regenerating
tissues along with a decrease in the number of mitotic cells (described in more detail
later). Importin-? homologs in other species aid in the transport of NLS-containing
proteins into the nucleus by acting as an adaptor between the target proteins and
Importin-?, which in turn interacts with the nuclear pore complex 50]. Intrigued by the possibility that regulated nuclear import of proteins could be
important for planarian regeneration, we decided to characterize the Importin-? gene
family in S. mediterranea. In addition to Smed-ima-1, we identified another member of this family in the S. mediterranea genome, which we named Smed-ima-2 (Fig. 4a). Each of these proteins contains two Armadillo repeats, which may mediate protein-protein
interactions, and an Importin-? binding domain (IBB), which is required for the interaction
between Importin-? and Importin-? (Fig. 4b). A handful of other transcripts (labeled in the phylogenetic tree by the mk4 names
given to them by the Maker gene prediction program in SmedGD 51]) shared some sequence similarity with this family but did not contain the Importin-?
binding domain (Fig. 4b) and could not be detected by in situ hybridization, suggesting that they are not
functional homologs. In contrast to Smed-ima-1, which was predominantly expressed in the stem cell population (Fig. 4c), Smed-ima-2 was ubiquitously expressed but still present in the stem cells based on its reduction
following ?-irradiation (Fig. 4d). Smed-ima-1 was strongly upregulated in the region beneath the blastema during regeneration.

Fig. 4. Analysis of importin-? homologs in S. mediterranea. a Phylogenetic analysis of Importin-? (???) homologs. Each protein is labeled with
Uniprot accession number, gene name, and species. Species abbreviations are as follows:
HUMAN?=?Homo sapiens, PONAB?=?Pongo abelii, BOVIN?=?Bos taurus, MOUSE?=?Mus musculus, DANRE?=?Danio rerio, RAT?=?Rattus norvegicus, CHICK?=?Gallus, YEAST?=?Saccharomyces cerevisiae, SCHPO?=?Schizosaccharomyces pombe, DICDI?=?Dictyostelium discoideum, ARATH?=?Arabidopsis thaliana, ORYSJ?=?Oryza sativa subspecies japonica, SOLLC?=?Solanum lycopersicum, SMED?=?Schmidtea mediterranea, CAEEL?=?Caenorhabditis elegans, DROME?=?Drosophila melanogaster, XENLA?=?Xenopus laevis. Arrows indicate Smed-ima-1 and Smed-ima-2. b Domain structure of Schmidtea mediterranea Importin-? proteins. IBB?=?Importin-? binding domain. ARM?=?Armadillo repeats. c–d Whole-mount in situ hybridization to Smed-ima-1 and Smed-ima-2. The irradiated worms were treated with 60 Gy ?-irradiation 3 days prior to fixation
to destroy the stem cells. The regenerating animals were amputated to remove the half
of the head oriented toward the bottom of the picture two days prior to fixation.
Arrows indicate the blastema. Animals were imaged from the ventral side with anterior
to the left. Scale?=?0.5 mm

The two Importin-? homologs had distinct functions based on RNAi knockdown experiments.
Knockdown of Smed-ima-2 caused a rapid loss of stem cells accompanied by ventral curing and lysis (data not
shown; also observed by Reddien et al. 48]), suggesting that Smed-ima-2 may serve an essential function in bulk nuclear import of NLS-containing proteins.
The Smed-ima-1 RNAi phenotype was more specific; when animals were fed Smed-ima-1 dsRNA four times over two weeks then amputated pre-pharangeally, the photoreceptors
often appeared closer together than in gfp(RNAi) controls (n?=?21/113) (Fig. 5a). In other animals, one photoreceptor was very small or absent (n?=?23/113), or a single cycloptic photoreceptor appeared near the center of the head
(n?=?32/113) (Fig. 5a). The abnormality of the photoreceptors extended to their neuronal connections (n?=?4/4), which we visualized by staining RNAi-treated animals with an antibody against Arrestin
52] (Fig. 5c). The cephalic ganglia in Smed-ima-1(RNAi) planarians were reduced in size, with an average area of 0.114?±?0.015 mm
²
/mm animal length compared to 0.145?±?0.019 mm
²
/mm animal length in controls (n?=?4-5/group, Students’ t-test p-value?=?0.031) (Additional file 6A). Smed-ima-1(RNAi) cephalic ganglia also appeared less developed than in controls and were also collapsed
toward the midline (n?=?5/5) (Fig. 5b). The smaller size of the cephalic ganglia may have been due to a defect in differentiation;
this defect is further illustrated by a reduction in the number of regenerated sensory
neurons, from 28.5?±?1.2 cintillo positive cells/mm length in control to 18.9?±?2.7 in Smed-ima-1(RNAi) (n?=?5/group, Students’ t-test p-value??0.0001) (Fig. 5d and Additional file 6B). The regenerating tails of trunk fragments also developed abnormally in many cases,
with the blastema growing asymmetrically slanted or forked (n?=?24/64) (Fig. 5a). Interestingly, we did not observe similar defects in the tail blastema morphology
of regenerating head fragments.

Fig. 5. Smed-ima-1 is required for normal stem cell function and regeneration. a–f Images of control and Smed-ima-1(RNAi) animals fed bacterially expressed dsRNA targeting each gene four times over 2 weeks
then amputated transversely both pre- and post-pharyngeally. Animals were imaged or
fixed on day 10 of regeneration. Anterior is toward the top. Live animals were imaged
from the dorsal side and all others were imaged ventrally. Scale bars?=?0.5 mm for
(a) and (g), 0.25 mm for (b) and (d–f), 0.1 mm for (c). a Live animals following RNAi treatment. Dashed lines indicate amputation sites. Red
triangles indicate photoreceptors forming abnormally close to the midline, and the
black arrow indicates a mis-positioned and underdeveloped photoreceptor. The white
arrow indicates forking of the tail blastema. b In situ hybridization to the neuronal marker Smed-pc2. White arrow indicates small cephalic ganglia collapsed toward the midline. c Staining with anti-Arrestin antibody (Arrestin) to mark photoreceptor neurons. Arrows
indicate aberrant neuronal projections. d In situ hybridization to Smed-cintillo, which labels sensory neurons. e In situ hybridization to Smed-inx to label the intestine. Arrows mark space between the two posterior intestinal branches.
f In situ hybridization to midline marker Smed-slit. g Uninjured animals stained with anti-phospho-Histone H3 antibody (PH3) to mark mitotic
cells following six dsRNA feedings. Anterior is to the left

The abnormal positioning of the photoreceptors and cephalic ganglia with respect to
the midline led us to investigate whether Smed-ima-1(RNAi) caused other defects in midline patterning. Knockdown of Smed-slit, a known regulator of midline patterning, leads to fusion of the two posterior branches
of the intestine during regeneration 53], however knockdown of Smed-ima-1 did not lead to the same phenotype (Fig. 5e). We also assayed Smed-slit mRNA expression by WISH in Smed-ima-1 knockdown animals and did not find any overt difference in its pattern or levels
(Fig. 5f). Therefore, the midline collapse phenotype of Smed-ima-1 RNAi does not appear to be caused by a defect in Smed-slit regulation.

We also observed a reduction in the number of mitotic cells in Smed-ima-1(RNAi) animals compared to controls. After six feedings of dsRNA over three weeks, the number
of PH3
⁺
cells in intact animals was reduced from 258?±?53 cells/mm
²
in gfp(RNAi) controls to 162?±?40 cells/mm
²
in Smed-ima-1(RNAi) (n?=?15–16 animals/gene, Students’ t-test p-value??0.0001) (Fig. 5g). Worms treated with dsRNA against Smed-ima-1 for extended periods of time (more than 4 weeks) began to show other signs of stem
cell dysfunction, including head regression and lysis. The RNAi phenotypes of the
two Smed-importin–? genes suggest that regulated nuclear import of proteins is a key factor in stem cell
function.

Our functional genomics screen identified several previously uncharacterized genes
required for neoblast maintenance and blastema formation and others affecting regeneration
or homeostatic maintenance of specific cell and tissue types. This, along with our
implication of a nuclear import factor in patterning of regenerating tissues provides
new insights into the molecular basis of planarian regeneration.