SUMOylation controls stem cell proliferation and regional cell death through Hedgehog signaling in planarians

Abstract

Mechanisms underlying anteroposterior body axis differences during adult tissue maintenance and regeneration are poorly understood. Here, we identify that post-translational modifications through the SUMO (Small Ubiquitin-like Modifier) machinery are evolutionarily conserved in the Lophotrocozoan Schmidtea mediterranea. Disruption of SUMOylation in adult animals by RNA-interference of the only SUMO E2 conjugating enzyme Ubc9 leads to a systemic increase in DNA damage and a remarkable regional defect characterized by increased cell death and loss of the posterior half of the body. We identified that Ubc9 is mainly expressed in planarian stem cells (neoblasts) but it is also transcribed in differentiated cells including neurons. Regeneration in Ubc9(RNAi) animals is impaired and associated with low neoblast proliferation. We present evidence indicating that Ubc9-induced regional cell death is preceded by alterations in transcription and spatial expression of repressors and activators of the Hedgehog signaling pathway. Our results demonstrate that SUMOylation acts as a regional-specific cue to regulate cell fate during tissue renewal and regeneration.

Electronic supplementary material

The online version of this article (10.1007/s00018-017-2697-4) contains supplementary material, which is available to authorized users.

Introduction

The planarian Schmidtea mediterranea provides unique opportunities to dissect mechanisms controlling cellular decisions in the adult body. Signals influencing cellular behavior during adult tissue maintenance, repair, and cancer are evolutionarily conserved between planaria and mammals [2–8]. Systemic disruption of homologous recombination in S. mediterranea leads to dramatic increases in DNA double-strand breaks (DSBs) throughout the body [9]. Intriguingly, cells in the anterior region of the body survive and continue dividing with DSBs, while most cells in the posterior region of the planarian body undergo cell death. Brain signals and the retinoblastoma pathway contribute to the survival of cells with DNA damage in the anterior region [9]. However, it remains elusive which mechanisms operate in the posterior region, where cells with DSB are collectively eliminated by apoptosis.

To better understand mechanism(s) leading to collective cellular death, we searched for phenotypes involving regional defects across different animals. Disruption of the SUMOylation pathway during embryonic development in Drosophila and zebrafish lead to AP axis abnormalities and regional defects associated with cell death [10–12]. SUMOylation is an evolutionarily conserved post-translational modification that regulates a myriad of proteins and cellular processes including chromatin organization, transcription, DNA repair, apoptosis, cell cycle control, protein trafficking, and signal transduction [13–15]. SUMOylation is a highly dynamic and reversible process whereby the small ubiquitin-like modifier (SUMO) is attached to target proteins, acting as a molecular switch that quickly controls protein activity [13]. The SUMOylation cascade requires Ubc9, the sole E2 conjugating enzyme necessary for properly attaching SUMO to target proteins [16–20]. Inactivation of Ubc9 is embryonically lethal, limiting our understanding about the role of SUMOylation in the maintenance and repair of adult tissues [10–12, 21].

Our present findings show that the SUMO pathway is evolutionarily and functionally conserved in the Lophotrocozoan S. mediterranea. Disruption of the SUMOylation pathway in planarians with RNA-interference of Ubc9 lead to regional defects characterized by loss of the posterior half of the body (tail), systemic loss of the adult stem cells and increase in DNA damage, specifically DSBs. In addition, Ubc9 is required for posterior specific tissue maintenance and regeneration through the regulation of the Hedgehog (Hh) pathway. Altogether, we propose that SUMOylation controls regional fate decisions by balancing activators and repressors of the Hh signaling.

Results

The SUMOylation pathway is conserved in the planarian Schmidtea mediterranea

The SUMOylation pathway is cyclic in nature and can be broadly summarized in five steps: SUMO maturation, activation, conjugation, attachment and its removal from the target protein (Fig. 1a). Components of the SUMOylation pathway are evolutionarily conserved from yeast through mammals [13]. The S. mediterranea genome includes homologs of all components of the general SUMOylation pathway (Fig. 1b). Interestingly, the two SUMO proteins we identified in planarians are the homologs of SUMO-2/3 and not SUMO-1 (Figure S1A, B). Consistent with other organisms, there is only one homolog for the SUMO E2 conjugating enzyme UBC9 (Smed-Ubc9, henceforth Ubc9). In most cases, Ubc9 is known to be the linchpin controlling the SUMOylation pathway and acts as a rate-limiting step for SUMO conjugation [13]. Phylogenetic analysis of UBC9 across different phyla confirms its evolutionary conservation, which in the case of planarians is 73% identical to its human counterpart (Fig. 1c, S1C). These results suggest that the SUMO pathway is evolutionarily conserved in S. mediterranea, a Plathyhelmith member of the Lophotrocozoan clade.

Fig. 1.

The SUMOylation pathway is conserved in the planarian Schmidtea mediterranea. a A general outline of the SUMOylation pathway with the names of enzymes involved and a brief description of major steps in the process. b A list of SUMO pathway components found in humans and their homologs in yeast and S. mediterranea. c Protein alignment of planarian UBC9 and homologs found in humans (NP_003336.1), mice (EDL06246.1), Drosophila (NP_476978.1), C. elegans (NP_001023158.1) and Xenopus (NP_001080758.1). Human and planarian UBC9 protein sequences are 73% identical

SUMOylation is required for regional tissue maintenance

To functionally characterize the role of Ubc9 in S. mediterranea, we performed RNA interference by feeding with dsRNA expressing bacteria mixed with liver six times every three days, and animals were fixed 25 days after the first dsRNA feeding (25 dpf) (Fig. 2a). Animals subjected to Ubc9(RNAi) exhibited macroscopic signs of tissue degradation, characterized by darkening of the epithelia in the post-pharyngeal region, followed by formation of lesions and complete loss of their tail by day 25 and onward (Fig. 2b). After 25 dpf, the experimental group undergoes rapid tissue deterioration and begins to lyse (Fig. 2c). We noticed that Ubc9(RNAi) animals have an apparent reduction in size before losing their tail, which was confirmed by the reduction in surface area that started 15 dpf (Fig. 2d). Despite the tissue loss being restricted to the posterior region, qPCR experiments revealed that Ubc9 gene expression was effectively downregulated throughout the whole organism at 25 dpf (Fig. 2e).

Fig. 2.

Ubc9 is required for regional tissue maintenance. a RNAi schedule based on feeding with bacterially expressed dsRNA to knockdown Ubc9 expression. All controls were fed either Unc22 or gfp. Black bars represent feeding days and red represents fixation. b Representative images of control and Ubc9(RNAi) animals 25 dpf. The arrows indicate different abnormalities including dark tail, lesions, and tail loss. n = 75 total animals used in three biological replicates. Scale bar = 200 µm. Note the scale bar on the right applies for the two pictures on the right portion of the figure. c Histogram illustrates the progression of the Ubc9 phenotype based on macroscopic abnormalities observed in “b”. d Change in surface area over the 25-day RNAi time course in controls and Ubc9(RNAi) animals. The experiment involves 15 animals per time point and three biological replicates. e Levels of Ubc9 expression measured with qPCR. The results show that the RNAi knockdown protocol is effective at reducing transcript levels of Ubc9 in both anterior and posterior regions. Data obtained from triplicates per experiment of at least two biological replicates. ****p < 0.0001; two-way ANOVA

Ubc9 is expressed in neoblasts and post-mitotic cells

To further understand Ubc9 function in planarians, we characterized its spatial distribution with in situ hybridization and found it is ubiquitously expressed throughout the organism (Fig. 3a). Interestingly, Ubc9 expression is dramatically reduced 24 h after lethal irradiation (6 k rads), which is known to eliminate planarian stem cells or neoblasts [22], and remains at that level 7 days post-irradiation (dpi) (Fig. 3a). In addition, we detected residual Ubc9 expression surrounding the pharynx and the periphery of the brain at 7 dpi (Fig. 3a). Distribution of Ubc9 expression among neoblasts and post-mitotic cells was further confirmed using the single-cell sequencing database [1] (Fig. 3b, c). Furthermore, the single-cell analysis also reveals that Ubc9 expression is present across all neoblast subclasses and in a subset of differentiated cells including neural, epidermal, and gut (Fig. 3c, S2). Together, these findings suggest that Ubc9 is mainly expressed in neoblasts but it is also transcribed in post-mitotic cells.

Fig. 3.

Ubc9 is necessary for stem cell function and proper cell cycle transition. a Whole mount in situ hybridization using antisense probe against Ubc9. Gene expression is found ubiquitously distributed along the AP axis (left) and is dramatically reduced 24 h after lethal exposure to gamma irradiation (6 K rads) and remained at that level for 7 days post-irradiation (red arrows). Experiments involved three biological replicates with 10 animals per experiment. b Ubc9 gene expression levels in different cell populations (X1: proliferative cells, X2: early post-mitotic progeny, and Xins: late post-mitotic progeny). c t-SNE plot of single cells displaying clusters of neoblasts and differentiated cells (left), along with the overlaid Ubc9 expression (right). The respective reference for the level of expression based on the colored gradient scale blue to red (low–high, respectively). The gene expression result for “b and c” were obtained from the planaria single-cell database hosted by the Reddien Lab at the Whitehead Institute for Biomedical Research (https://radiant.wi.mit.edu/app/) [1]. d Time course of mitotic activity along the AP axis, expressed as fold change in reference to the control at each time point. e Spatial distribution of mitotic activity in whole mount immunostaining against phospho-histone H3 (Ser10) (H3P) at 25 dpf Ubc9(RNAi). Dashed line divides anterior and posterior regions of the animal. All scale bars = 200 µm. f Fold change mitotic counts obtained independently from the anterior or posterior regions. Mitotic levels involved three biological replicates and more than 40 animals. g FACS analysis using DRAQ5, a nuclear dye and Calcein, a live cell marker, in either anterior or posterior regions of control and Ubc9(RNAi) animals 25 dpf. Blue, green and red squares represent X1, X2 and Xins populations, respectively. h Cell cycle analysis using DRAQ5 DNA dye in AP regions of control and Ubc9(RNAi) animals 25 dpf. Red bars represent percentage of cells at different phases of cell cycle. All FACS analysis performed with more than 10,000 cells and results are representative of three experiments with about 40 animals total. i Gene expression levels of various cell cycle markers necessary for proper G1/S and G2/M transition in AP regions of control and Ubc9(RNAi) animals 25 dpf. Gene expression portrayed as fold change normalized to control. **p < 0.01; ***p < 0.001; ****p < 0.0001; two-way-ANOVA

Ubc9 is required for the proper function of stem cells

Neoblasts are the only proliferative cells in planarians, and thus we evaluated mitotic activity by whole-mount immunostaining against phosphorylated histone-3 (H3P) at different time points after Ubc9(RNAi) (Fig. 3d). These experiments revealed a three-fold reduction in H3P⁺ cells in the experimental group but residual mitotic activity remained (Fig. 3d, e). The macroscopic regional effects are visible by day 15 after Ubc9(RNAi) (Fig. 2c), but the mitotic decrease is noted by day 20–25 post-RNAi. This suggests that tissue loss precedes effects in neoblast proliferation. To evaluate whether there was a regional reduction in mitotic activity, we separately quantified H3P⁺ cells in the anterior and posterior regions. Each animal was split into two parts along the AP axis, considering the prepharyngeal as the anterior region and the remaining tissue including the pharynx as the posterior region (Fig. 2e). Such delimitation was made to account for the dramatic reduction in the tail region in the phenotype. This analysis demonstrated that loss of mitotic activity was halved along both the anterior and the posterior regions (Fig. 3f).

Next, we analyzed the effects of Ubc9 downregulation on cell populations and cell cycle dynamics using flow cytometry (FACS) [23]. Pooled cells from anterior and posterior regions were processed separately in both control and Ubc9(RNAi) animals 25 dpf. First, we examined cellular populations associated with neoblasts (X1) and early and late post-mitotic progeny (X2 and Xins, respectively) [23, 24]. There was an overall reduction in the proliferative X1 population across the AP axis, which was more noticeable (~ 4 fold) in the posterior region of the experimental group (Fig. 3g). The decline in the X1 cells is consistent with the reduction in mitotic activity (Fig. 3d–f). However, Ubc9(RNAi) has an unexpected increase on post-mitotic X2 cells in the anterior region, whereas the posterior region shows the expected decline (Fig. 3g). The mechanism behind this observation remains to be investigated.

Taken together, whole mount in situ hybridization and gene expression levels demonstrate that Ubc9(RNAi) lead to severe downregulation in markers of X1, X2 and Xins cells (Smedwi-1, Smed-prog-1 and Smed-AGAT-1, respectively) (Figure S3A, B). Interestingly, the dramatic reduction in the expression of early progeny marker (Smed-prog-1) in the tail region and the persistence of some positive Smed-prog-1 cells in the anterior region may imply that the enrichment of the X2 fraction is specific to the anterior region of Ubc9(RNAi) animals. Nonetheless, this finding demonstrates that Ubc9 is required for the proper function of neoblasts and differentiated cells. Previous studies have shown that blockage in G1 phase of the cell cycle is observed in other organisms after Ubc9 depletion [25–28]. Consistently, we found in planarians that Ubc9(RNAi) leads to an increase in G1 phase (74.1 vs 81.1) and a decrease in S and G2/M (13.9 vs 6.02) phases of the cell cycle along the AP axis (Fig. 3h). In addition, the expression levels of genes that are evolutionarily conserved and commonly associated with G1, S and G2/M transitions were significantly decreased (p < 0.0001 two-way ANOVA) in Ubc9(RNAi) animals (Fig. 3i). This is consistent with both a reduction in neoblast numbers and a concomitant reduction in cell cycle as observed in Figs. 3d–h. Together, these results suggest that Ubc9 is required for maintenance of neoblasts and proper cell cycle transition.

Downregulation of SUMO proteins results in regional defects

We also tested whether RNAi of S. mediterranea SUMO proteins lead to gross morphological defects observed with Ubc9(RNAi). We identified two SUMO homologs in the S. mediterranea genome, which we termed Smed-SUMO2 and Smed-SUMO3 (Figure S1A, B). Individual RNAi of each component did not lead to behavioral or gross morphological defects, probably due to compensatory roles. However, double RNAi between the small ubiquitin-related modifiers Smed-SUMO2 and Smed-SUMO3 did produce regional defects like those noted after Ubc9(RNAi). Performing Smed-SUMO2 + SUMO3(RNAi) led to accelerated tail loss in a fraction of the animals (i.e. 15% animals, 20 dpf) (Figure S4A). The remaining animals gradually underwent regional tissue loss and eventually died. Further evaluation of mitotic activity revealed that simultaneous downregulation of Smed-SUMO2 + SUMO3 dramatically reduced the number of proliferative cells across the animal (Figure S4B, C). Analysis of gene expression with markers for neoblasts and post-mitotic cells also demonstrated a dramatic decrease in expression that is consistent with the effects of Ubc9(RNAi) (Figure S4D). Collectively, these results suggest that regional defects and systemic neoblast dysfunction are specific to the disruption of SUMOylation through RNAi of Ubc9 and/or Smed-SUMO2 + SUMO3. For simplicity and consistency, we focused on analysis of the SUMO pathway through the downregulation of Ubc9.

Ubc9-systemic loss of function leads to regional cell death

To investigate the cause of regional tissue loss after Ubc9(RNAi), we analyzed cell death through whole-mount immunostaining and FACS [9, 23]. First, immunostaining assay with cell death marker caspase-3; which has been previously characterized in planarians [29], revealed an increase in cell death throughout the animal, with a prominent expression in the posterior region 25 dpf (Fig. 4a). In some instances, cell death signal was associated with cellular clusters, limiting our capacity to manually count dying cells. Instead, we chose to quantify the signal by generating an intensity profile of the fluorescence from anterior to posterior body axis. We found the intensity to be higher in Ubc9(RNAi), with a four time increase in cell death in the posterior region (Fig. 4b). Ubc9(RNAi)-induced cell death distribution was further analyzed by FACS using Annexin V, a marker of apoptosis that we have previously characterized in planarians [23]. Anterior and posterior fragments were processed separately and the results show that Annexin V⁺cells are markedly increased in the posterior region of the experimental group (Fig. 4c). These results are consistent with regional cell death found in immunostaining using caspase-3 antibody. Furthermore, gene expression of BCL2, a survival signal marker [30], is importantly reduced in the posterior region when compared to the anterior region (Fig. 4d). Taken together, our findings demonstrate that Ubc9 is essential to cellular survival and regional tissue maintenance.

Fig. 4.

Ubc9-loss of function leads to regional cell death. a Whole mount immunostaining against caspase-3, a marker for cell death, in control and Ubc9(RNAi) animals. About 65% of experimental animals showed similar caspase-3 signal distribution at 25 dpf. Immunostainings involved three biological replicates and more than 40 animals. Scale bars = 200 µm. b Intensity of capase-3 signal from anterior to posterior (white line) of control and Ubc9(RNAi) animals. Intensity signal quantification involved three biological replicates and more than 20 animals. c FACS analysis staining against Annexin V, a marker for apoptosis, and 7AAD, a cell viability marker, in AP regions of control and Ubc9(RNAi) animals 25 dpf. Annexin V-/7 AAD- quadrant includes viable cells (outlined green). Annexin V +/7 AAD– and Annexin V +/7 AAD + indicate cells that are in early and late (necrotic) stages of cell death, respectively (outlined red). Blue numbers in each quadrant indicate the percentage of cells with that staining profile. Data is representative of two experiments with n > 40 each. d Fold change of BCL2 gene expression in anterior and posterior regions normalized to control group

Ubc9 is required to maintain genomic integrity along the anteroposterior axis

Ubc9 downregulation in planarians resembles regional features of the Rad51 phenotype, which is characterized by an increase in DNA double stranded breaks (DSBs) throughout the body [9]. To assess DNA integrity in Ubc9(RNAi) animals, we collected cells at different dpf and performed (1) COMET assay, (2) karyotyping and (3) immunostaining against markers required for early DNA damage response and repair through homologous recombination (HR). COMET assay was adjusted to detect DSBs [9, 31] and our results confirmed an increase in DNA damage equally distributed along the AP axis and is present as early as 10 dpf (Fig. 5a). The presence of DSBs was accompanied by a two fold increase in chromosomal abnormalities detected by karyotyping (Fig. 5b). Deletions, fusions and dicentric chromosomes characterized the abnormalities that affected all four pairs of chromosomes of the asexual S. mediterranea CIW4 strain. The spatial distribution of DNA damage in planarians can be detected with whole mount immunostaining using antibodies against ϒH2Ax (Figure S5) [31] and RAD51 [9, 32]. Our results with these antibodies revealed an increased signal for both in the experimental group (Fig. 5c, d). Interestingly, increase in signal intensities for both antibodies were consistently displayed in clusters in the pre-pharyngeal and post-pharyngeal area for 70% of the animals (Fig. 5c, d). Western blot also confirmed the increase in RAD51 signal after Ubc9(RNAi) (Fig. 5e, f). Increase in ϒH2Ax and RAD51 levels suggest that the mechanism for sensing DNA damage is active after Ubc9(RNAi).

Fig. 5.

Ubc9 is required to maintain genomic integrity along the anteroposterior axis. a COMET assay, single gel electrophoresis under alkaline conditions, was performed in AP regions of control and Ubc9(RNAi) animals at10, 15, and 25 dpf. Visual scoring was used to quantify DNA damage. Color-coded key at bottom represents undamaged (green), moderate damage (orange) and extremely damaged DNA (red) shown in either anterior or posterior regions. DNA damage scale reference is similar to the one used in Peiris et al. 2016b. Downregulation of Ubc9 leads to systemic accumulation of DNA damage as early as day 10 when compared to controls. Approximately 40 cells were counted for each DNA condition in two biological replicates. b Chromosome quality (not damage and damaged) normalized to control in Ubc9(RNAi) 25 dpf animals. Experiment was repeated twice. c Whole mount immunostaining against gamma H2AX antibody in control and Ubc9(RNAi) animals 25 dpf. Total number of animals was 20 and n = 2 biological replicates. d Whole mount immunostaining against human RAD51 antibody in control and Ubc9(RNAi) animals 25 dpf. Total number of animals was 15 and experiment was repeated twice. Scale bars = 200 µm and images are representative of approximately 70% of the animals in each condition. e, f Western-blot and subsequent quantification for RAD51 in control and Ubc9(RNAi) animals. Alpha tubulin was used as an internal control. Protein was extracted from n > 30 animals. g Spatial distribution of RAD51 immunostaining (green) in reference to the cell nucleus (stained with DAPI, blue) in control and Ubc9(RNAi) 18 dpf and 5 days after sub-lethal irradiation (1.25 k rads). RAD51 subcellular localization to the nucleus is at a maximum at this point in time. Scale bar = 10 µm. h Quantification of cells with RAD51 + foci in the nucleus and cytoplasm in control and Ubc9(RNAi) 18 dpf and 5 days after sub-lethal irradiation (1.25 k rads). Approximately 50 cells were counted for each condition in two biological replicates and all RAD51 stainings were performed with human RAD51 antibody

To repair DSBs, RAD51 must be translocated from the cytoplasm to the nucleus. Thus, we tested whether RAD51 translocation was compromised in Ubc9(RNAi) animals. Exposure to gamma irradiation induces DSBs, and RAD51 protein expression and nuclear translocation is at a maximum after 5 days post sub-lethal irradiation (1.25 K rads) [9]. Therefore, control and Ubc9(RNAi) animals were exposed to sub-lethal irradiation and the cells were harvested after 5 days to evaluate RAD51 sub-cellular location (Fig. 5g). This experiment showed that RAD51 nuclear localization was observed in most of the control cells (~ 80%) whereas smaller fraction of cells (31%) in the experimental group exhibited the same behavior (Fig. 5h). In summary, these data indicate that Ubc9 functions in maintaining DNA integrity and regulating RAD51 localization are conserved also in S. mediterranea.

Regional defects in Ubc9(RNAi) animals are mediated through modulation of the Hedgehog pathway

Posterior identity in planarians is associated with proper regulation of the evolutionarily conserved Hedgehog (Hh) pathway [16, 33, 34]. Specifically, regional determination of the posterior tissue is established through Patched (Ptc)-mediated signaling of Hh [33, 34]. Hh signaling in planarians is required for ciliogenesis, regeneration [16, 33, 34] and more recently, it has been linked to neural and glial function [17, 35]. However, it is still unknown whether this pathway is necessary for regional information in uninjured animals. Hh signaling activity can be assayed by gene expression analysis of anterior and posterior polarity markers, secreted frizzled protein 1 (sfrp1) and frizzled-like protein 4 (fz4), respectively [34]. Strikingly, the expression of both fz4 and sfrp1 was dramatically altered in uninjured Ubc9(RNAi) animals, while it remained without change in the control group. The expression of sfrp1 was elevated and appeared ectopically in the lateral and tail regions at day 10 when compared to control animals (Fig. 6a, red arrows). The pattern and level of expression of srfp1 decreased to wild-type at day 15. Inversely, fz4 appeared ectopically in the head and lateral cells at day 10 but was strongly elevated at day 15. Both mRNAs were downregulated when the tail phenotype became obvious (Fig. 6a). This suggests that the downregulation of Ubc9 may lead to transient ectopic activation of Hh signaling.

Fig. 6.

Regional defects after Ubc9(RNAi) are mediated through repression of Hedgehog pathway. a Whole mount in situ hybridization expression of sfrp-1 (anterior polarity marker) and fz4 (posterior polarity marker) in control and 10, 15 and 25 dpf Ubc9(RNAi) animals. Red arrows indicate abnormal gene expression. Scale bars = 200 µm. Note that controls for each time point were executed but no apparent change in gene expression was observed over time –not shown. b Gene expression levels of Patched, Hedgehog and Smoothened after Ubc9(RNAi) animals at 10, 15 and 25 dpf. Gene expression is given in fold change normalized to control. c Whole mount in situ hybridization expression of Patched and Hedgehog in control and 10, 15 and 25 dpf Ubc9(RNAi) animals. Scale bars = 200 µm. d. RNAi schedule based on feeding with bacterially expressed dsRNA to perform double knockdown of Hedgehog or Patched and Ubc9. All controls were fed either Unc22 or gfp. Black bars represent feeding days and red represents fixation. e Histogram depicts occurrence of tail abnormalities in Hedgehog + Ubc9 and Patched + Ubc9 animals. Data is representative of n = 30 animals consisting of three biological replicates

We next analyzed the expression of the inhibitor Ptc and activators Hh and Smoothened (Smo) at early and late phases of the Ubc9 phenotype (i.e. 10, 15 and 25 dpf). Ptc expression shows a steady increase in days 15–25 that is more than double at late phases of the phenotype (Fig. 6b, c). Hh expression decreased at day 10 but increases at day 15 dpf and returned to basal levels at the late stage, while Smo expression appears stable. This suggests that depletion of Ubc9 drives the pathway off equilibrium, leading to abnormal fluctuations in gene expression (Fig. 6b, c). To further examine the importance of the Hh pathway in mediating the loss of posterior homeostasis in Ubc9(RNAi) animals, we performed double RNAi experiments involving Ptc, Hh, and Ubc9. Different RNAi strategies were assayed and the most consistent results were obtained by first knocking-down either Ptc or Hh followed by Ubc9 in a 40 days schedule that included a total of 14 feedings with dsRNA (Fig. 6d). Animals subjected to double RNAi Ptc + Ubc9 displayed similar dynamics as in Ubc9 alone (Figs. 2c, 6e). However, simultaneous downregulation of Hh + Ubc9 led to more aggressive defects in the posterior region that appeared at day 10 and were lethal for 20% of the animals by day 20 of the Ubc9(RNAi) (Fig. 6e). The increased regional defects after Hh + Ubc9(RNAi), suggest synergistic/parallel contribution between these factors.

Sumoylation is required for planarian regeneration

We reasoned that if Ubc9(RNAi) leads to deregulation of the Hh pathway, then amputation would result in deficient tail and brain regeneration [33–35]. Control and Ubc9(RNAi) animals were amputated pre- and post-pharyngeally at 18 dpf, which is an early stage of the phenotype where most animals still retain posterior tissue but fluctuations in both Ptc expression and the downstream Hh components (fz4 and sfrp1) are detected (Fig. 6b, c). Animals were then allowed to regenerate for 7 days (7 dpa) and blastema formation was evaluated (Fig. 7a). Ubc9(RNAi) animals formed small anterior blastemas and remarkably, lacked posterior blastema formation (Fig. 7b, c). To further examine the mechanism limiting regeneration, we assayed the mitotic response after amputation and revealed that Ubc9(RNAi) animals can sense injury, but respond with limited mitotic capability (Fig. 7d). Next, we considered whether regeneration of the nervous system was compromised after Ubc9(RNAi). Two main observations were made: (1) the size of the regenerated brain was reduced by half, consistent with compromised neurogenesis [35] (Fig. 7e, f) and (2) ventral nerve cords were truncated, likely due to the lack of posterior regeneration (Fig. 7e). Furthermore, we performed similar amputation experiments on double RNAi animals (Fig. 7g) and the regenerative outcome of these double knockdowns between Ptc, Hh, and Ubc9 revealed that 100% of animals subjected to double RNAi Ptc + Ubc9 and Hh + Ubc9 presented lack of posterior blastema formation (Fig. 7h). This observation is in accordance with the posterior homeostasis synergistic/parallel effects (Fig. 6e). In addition, during regeneration, the anterior defects were more prevalent in the Ptc + Ubc9(RNAi) (over 90% of the animals), whereas 35% of the Hh + Ubc9 showed regenerative defects in the anterior end as in single Ubc9(RNAi) (Fig. 7b). These results demonstrate that Ubc9 is required for anterior and posterior blastema formation and organogenesis following injury.

Fig. 7.

Ubc9 is required for proper regeneration. a Schematic representation of regeneration experiments on control and Ubc9(RNAi) animals 18 dpf. b Live pictures of control and Ubc9(RNAi) 18 dpf animals 7 days post amputation (7dpa). c Blastema size in control and Ubc9(RNAi) animals expressed as fold change relative to control. In all experiments used more than 30 animals in three biological replicates. d Trunk fragments generated from control and Ubc9(RNAi) 18dpf animals were fixed at 0, 6, 10, 20, 30 and 48 h post amputation. Mitotic cell numbers were determined via H3P staining and expressed as H3P + foci divided by surface area in mm². Three biological replicates with more than 15 animals per time point were used. e Whole mount immunostaining against SYNORF-1 (synapsin), a marker for the nervous system, in control and Ubc9(RNAi) 18 dpf 7DPA. Green signal denotes the planarian central nervous system and insets provide detailed amplification of anterior and posterior ends including brain and tip of the tail, respectively. White dotted lines underscore differences between control and experimental group regarding brain morphology (top) and ventral nerve cords (bottom). More than 20 animals were stained in three biological replicates. f Regenerated brain size obtained from determination of surface area in both control and experimental group. Differences in brain size were calculated with Two way-ANOVA. g RNAi schedule based on feeding with bacterially expressed dsRNA to perform double knockdown of Hedgehog or Patched and Ubc9. All controls were fed either Unc22 or gfp. Black bars represent feeding days and red represents amputation. h Histogram depicts occurrence of regenerative defects in Patched + Ubc9 and Hedgehog + Ubc9 animals 7 days after amputation. Representative images of defects is shown on the right side. i Schematic summary representing Smed-Ubc9 acts as an upstream regulator of different functions in the adult body. At the organismal level, downregulation of Ubc9 lead to a systemic increase in DNA damage and decrease in in both stem cell renewal and cell cycle progression. However, it remains unclear whether inhibition of Hedgehog (Hh) signaling is associated with any of the systemic effects. Regionally, dysfunctional Ubc9 triggers collective cell death and posterior specific abrogation of tissue regeneration, which are mediated through Hh signaling. At the organ level, Ubc9 also regulates proper brain regeneration likely through disruption of Hh signaling that is required for neurogenesis. It is possible that lack of specific neurons in Ubc9(RNAi) animals also affect neural regulation of cell cycle progression but additional experiments are required. ****p < 0.0001 and all other statistical comparisons were made with Sidak’s multiple comparisons test NS (no significance) p > 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Scale bars = 200 µm

Discussion

Regional signals contribute to cell fate decisions in the adult body. There is a tendency for cells in the anterior part of the body to have a superior proliferative response in health and disease [36]. However, the mechanism underlying this regional difference remains poorly understood. Our results provide critical insight on the regulation of regional cell fate during tissue renewal and regeneration. Specifically, we demonstrate that the sole E2 conjugating enzyme Ubc9, an essential component of the SUMOylation pathway, controls cell cycle, stem cell renewal, adult tissue homeostasis and regeneration. In addition, Ubc9 function is critical for subcellular translocation of RAD51 into the nucleus to repair DNA damage through homologous recombination. Finally, we establish Ubc9 is an upstream regulator that exerts control over Hedgehog signaling to mediate regional cell fate decisions during tissue homeostasis and regeneration (Fig. 7i).

Post-translational modification (PTMs) with small ubiquitin-like modifier (SUMO) is a key regulator of various cellular processes in metazoans [13, 15]. We establish that SUMOylation is evolutionarily conserved in S. mediterranea, a member of the under studied group of Lophotrochozoans. This finding is critical because non-Ecdysozoan protostomes include the largest number of animal phyla across metazoans and the study of members of this group will complement, at the very least, information obtained in more traditional experimental organisms [37–39]. The lack of redundancy of Ubc9 in the planarian genome recapitulates its evolutionarily conservation in the regulation of the SUMO pathway across eukaryotes. This is further evidenced by the close relationship in the protein sequence between human and planarian UBC9. A major challenge facing the study of Ubc9 is the fact that its complete abrogation is embryonically lethal, which hampers our capacity to understand the systemic contribution of SUMO pathway at post-embryonic stages [40, 41]. Studies in S. mediterranea will provide unique opportunities to analyze PTMs through SUMOylation during simultaneous renewal of many tissues and regeneration in the context of the adult body.

The role of Ubc9 in regulating stem cells depends on the context and the experimental models in which the studies are performed [42–45]. For example, Ubc9 function is required for embryonic stem cell survival in vitro and reprogramming of induced pluripotent cells (iPS) but Ubc9 function is not pertinent to the maintenance of mouse embryonic fibroblasts (MEFs) [46]. Our analysis looks at the function of Ubc9 in planarians, which contain large pools of adult pluripotent stem cells (neoblasts). Ubc9(RNAi) led to rapid depletion of neoblasts across the AP body axis. This may be due to its effect on the cell cycle dynamics as knockdown of Ubc9 causes an accumulation of cells in G1 phase and a reduction in G2/M phase. Previous studies have suggested knockdown of Ubc9 is a key regulator of G1/S phase transition through various mechanisms of action [25, 28, 42, 44]. Additional findings suggest Ubc9 knockdown induces DNA damage and chromosomal abnormalities, which would block cells from passing the critical cell cycle checkpoints [21, 47–49]. Therefore, this blockage in G1 phase could also impair the progression of post-mitotic progeny that will negatively impact the maintenance of differentiated tissues. Our findings in planarians are consistent with these observations and support a conserved role for SUMOylation in cell cycle control and adult stem cell maintenance.

Collective regulation of cellular decisions along the AP axis is subject to mechanisms that remain poorly understood. Earlier studies have demonstrated that downregulation of Ubc9 induces regional cell death that precede lethality during embryonic development in Drosophila and zebrafish [10, 11]. In accordance, our results demonstrate consistent regional cell death followed by lethality, supporting functional conservation across metazoans and developmental stages. We propose that Ubc9 functions as a regulator of regional cell death that prevails during embryonic development and adult stages.

Recently, we reported that loss of function of Rad51, a protein required for DNA repair through HR, induces regional defects in planarians [9]. Ubc9-mediated regional cell death follows a similar pattern as in Rad51(RNAi). Double RNAi experiments involving Rad51 and Ubc9 demonstrate a synergistic effect in reducing mitotic activity throughout the animal (Figure S6A, B). Furthermore, this complementary effect is also supported by the observation that each gene’s expression is reduced but not completely abrogated after Rad51 or Ubc9(RNAi) (Figure S6C, D). Taken together, these results suggest that the SUMO pathway overlaps with Rad51 in some functions as RAD51 is probably also a SUMO target also in planarian, but still exert mitotic control through different mechanisms. For example, signals from the nervous system contribute to maintain mitotic activity in the anterior region of Rad51(RNAi) animals [9], whereas these neural inputs are limited after Ubc9(RNAi). It is possible that inhibition of the Hh pathway through Ubc9 loss of function affects the neuronal subtype(s) [ [35] and see discussion below] that send survival signals after abrogation of Rad51. However, the specific molecular mechanisms regulating anterior cellular fate in Ubc9 animals remain to be investigated.

Downregulation of Ubc9 leads to an increase in DSBs that are similarly distributed along the AP axis. Chromosomal abnormalities were also evident after Ubc9(RNAi), which is consistent with both defective DNA repair response and chromosomal segregation [21, 47–49]. We also noted that Ubc9 downregulation leads to DNA damage response that occurs in clusters in both anterior and posterior regions, suggesting there may be focus of cells that are more susceptible to accumulate DSB. Future experiments will address the cause of DNA damage response in clusters after Ubc9(RNAi). Mechanistically, our results demonstrate that the increase in DSBs after Ubc9(RNAi) are due to a disruption in DNA repair through HR. Ubc9(RNAi) restricts nuclear translocation of RAD51, which is consistent with previous studies showing UBC9 is involved in trafficking of nuclear proteins [21, 50–54]. These data together reveal Ubc9 is an essential component to maintain genomic integrity during renewal of adult tissues.

The Ubc9 phenotype displays some resemblances in intact and regenerating animals. Specifically, during homeostasis, Ubc9(RNAi) animals lose their tail and during regeneration, worms are unable to form posterior blastema. In the anterior region, we show (Fig. 2c) that intact animals develop anterior abnormalities over time, which leads to loss of tissue in the anterior region but only at very late stages of the phenotype when they start dying (> 25 days post-RNAi). The regenerating worms also show defective anterior regeneration that is associated with a small blastema and defective neurogenesis and brain formation. These examples demonstrate that anterior and posterior defects are present in both the intact and the regenerating animals. However, differences in the manifestation of these defects are visualized depending on the timeline in which the animals are analyzed (Fig. 2c). Our work suggests that SUMO targets in the posterior regions are more sensitive to changes in their SUMOylation when compared to the anterior counterpart and this is likely the reason for the early posterior defect.

Finally, we provide mechanistic insights behind the cell death-driven regional loss of tissue after Ubc9(RNAi). Deregulation of Hh pathway by Ubc9(RNAi) is evident through the steady increase of Ptc expression, which is detected early in the phenotype. Previous work has demonstrated that Ptc ultimately controls the hedgehog pathway, as high levels of Ptc serve to sequester any free Hh and limit the activity of the pathway [55]. Our findings about a role for SUMOylation in regulating Hh signaling are also supported by recent observations in Drosophila [56, 57]. Consistently, many members of the Hh pathway are prone to regulation by post-translation modification, such as SUMOylation. For example, Hh stimulates Smo sumoylation by dissociating it from a desumoylation enzyme Ulp1. Sumoylated Smo prevents its ubiquitination and degradation, leading to Smo accumulation on the membrane and elevated Hh pathway activity [56, 57]. Uniquely, our data demonstrate that Ubc9 is a key regulator of collective cellular decisions such as proliferation and death that are critical for tissue maintenance in the adult body (Fig. 7i). The striking lack of tail regeneration highlights SUMOylation as a region-specific requirement for tissue repair by affecting transcription of Hh genes by regulation of yet to be identified substrate(s). This result is consistent with previous observations in planarians where repression of Hh signaling results in lack of tail regeneration [33, 34]. Nonetheless, it will be interesting to determine whether (1) other mechanisms besides Ptc signaling also affect SUMOylation-mediated cell death or (2) if increased Ptc expression is involved in regional non-canonical Hh signaling related to cell death. Our findings also highlight the importance of Ubc9 as an early requirement for neoblasts-driven injury response and suggest that depending on the circumstances SUMOylation may play synergistic/parallel roles or even act as an upstream component of the Hh pathway (Fig. 7i). The Ubc9 downregulation also extend to Hh-mediated impairment of neurogenesis [35], revealing Ubc9 also controls behavior of neural progenitors and brain function. Hh signaling is also associated with planarian glial cells and future studies will aim to determine whether Ubc9 loss-of-function affects the function of glial cells through non-canonical Hh signaling. Altogether, our results demonstrate that SUMOylation, in the context of the whole adult body, regulates important aspects of cellular proliferation and cell death at regional and systemic levels (Fig. 7i).

Materials and methods

Planarian culture

Planarian species used in these experiments was Schmidtea mediterranea CIW4. The culture was maintained as previously described by [2].

Identification of homologs and phylogenetic analysis

Components of the SUMOylation pathway were identified by BLASTing human SUMO protein sequences into available genomic resources for S. mediterranea [58, 59]. Identified sequences underwent a six-frame translation using PFAM and conservation was further confirmed with UNIPROT. The sequences were further verified by Blastn and Blastp in NCBI. The identified sequences were aligned by CLUSTALW with sequences obtained using HomoloGene (http://www.ncbi.nlm.nih.gov/homologene).

Phylogenetic tree for Ubc9 was constructed using 30 sequences from 30 metazoans and for the SUMO tree we used 31 sequences from 22 metazoans. The sequences were aligned using MAFFT L-INS-I http://mafft.cbrc.jp/alignment/server/ [ 14 ]. The alignment was manually adjusted to remove large gaps. Phylogenetic analysis was conducted first by determination of the optimal substitution model for each alignment with ProtTest 2.4 [15]. We constructed maximum likelihood trees in RAxML version 7.3.2 [18], with the appropriate model of protein substitution using the RaxML GUI front end. The best tree and bootsrap proportions were determined by 1000 iterations using Maximum likelihood + Slow bootstrap (100 runs), and PROTGAMMA + LG options using empirical frequencies. Tree images were generated using FigTree (http://tree.bio.ed.ac.uk/software/figtree/), which were then processed using Adobe Photoshop (http://www.adobe.com/products/photoshopfamily.html).

RNAi experiments

Primers with attb flanking sites were generated for the gene of interest [60]. PCR products were generated using cDNA synthesized from RNA of asexual S. mediterranea animals. Thermo Scientific VERSO cDNA synthesis kit was used to make cDNA. The gene was cloned into the pPR244 vector using BP Clonase II kit. The newly generated vector was transformed into NEB5a bacteria and then into HT115 bacteria. HT115 bacteria with the vector was grown in 2XYT media until it reached an OD595 of 0.400. 50 μL of 0.1 M IPTG was added and bacteria were incubated for an additional 4 h. The media was pelleted at 5000 rpm for 5 min and mixed with 25 μL of calf liver and fed to planarians. RNAi feeding times were adjusted for each experiment time.

For Ubc9(RNAi), planarians were fed UBC9 dsRNA expressing bacteria every 3 days for a total of six feedings. All experiments were performed after 25 days of RNAi starting from the first day of feeding. For double knockdown experiments involving hedgehog and patched RNAi alone, animals were fed with the appropriate dsRNA 8 times every 3 days and then with gfp 6 times every 3 days. For Ubc9 alone, animals were first fed gfp 8 times and then Ubc9 six times. For Hh + Ubc9 or Ptc + Ubc9(RNAi), animals were fed Hh or Ptc first and then Ubc9 after. All animals were fixed at the same time. Same steps were taken to generate animals for the regeneration experiments. For double knockdown experiments involving Sumo2 and Sumo3, 5 ml of bacteria for each gene was grown as before and mixed together before centrifugation. This pellet was once again mixed with 25 μL of calf liver and fed to the planarians. For Rad51 and Ubc9 double knockdown, a total of three RAD51 microinjections were administered over three consecutive days and then the animals were fed with Ubc9 dsRNA expressing bacteria 4 times every 3 days.

Whole mount immunofluorescence

Planarians were killed in 5.7% 12 N HCl solution and fixed in Carnoys solution for 2 h on ice. Carnoy’s solution was replaced with cold 100% MeOH and animals were placed in −20 °C for 1 h. Worms were bleached overnight in 6% H₂O₂ solution. Worms were then rehydrated in series from 100% MeOH to 100% PBSTx and stained as previously described [9]. Primary antibodies: α-H3p 1:250 (Millipore Cat# 05-817R), α-synapsin, 1:100 (Developmental Studies Hybridoma Bank); and activated caspase-3, 1:500 (Abcam ab13847) [29], RAD51, 1:500 (Abcam ab109107) and Phospho-Histone H2AX 1:1000 (ThemoFisher LF-PA0025) [9]. Similar results were obtained after testing side-by-side the SMED-RAD51 antibody [32] kindly provided by the Sánchez Alvarado/Hawley labs and the commercial RAD51 described above. The results in the manuscript involving RAD51 were obtained with the commercial RAD51 antibody. Secondary antibodies include: Goat anti-mouse Alexa488, 1:400 (Invitrogen Cat# 673781) for α-synapsin, α-arrestin, acetylated tubulin, Goat anti-rabbit Alexa568, 1:800 (Invitrogen Cat# 11036) for H3P, HRP-conjugated Goat anti-rabbit antibody (Millipore Cat# 12-348) with TSA-Alexa568 anti-HRP for caspase-3 (1:2000).

In situ hybridization

Riboprobes for in situ hybridization (ISH) were synthesized using T3 or T7 polymerase (Promega) and digoxigenin labeled ribonucleotide mix (Roche) with specific PCR templates as previously described [9]. Whole-mount ISH (WISH) was performed as previously described [61].

Quantitative real-time PCR

Quantitative real time PCR (qPCR) was performed as previously described [9, 62]. The ubiquitously expressed gene H.55.12e was used as the control. Each individual experiment consisted of triplicates per condition and experiments were independently repeated at least twice. RNA was extracted from anterior and posterior region of animals (> 20 per condition) and converted to cDNA using the Verso cDNA synthesis kit. Gene expression is expressed of fold change in comparison to the control.

Protein extraction and Western blot

Protein studies were performed as previously described [9].

Planarian dissociation

Planarians were amputated, whole or A/P, into small pieces using a razor, suspended in cold calcium, magnesium-free (CMF) media and placed on a rocker for 30 min at 4 °C in the dark. Cells were forced through a 20 micron filter and centrifuged at 1500 rpm followed by resuspension in 1 ml of CMF media before cell counting [23].

Fixation and immunostaining of dissociated cells

Dissociated cells were pipetted onto glass slips and left to anneal for 30 min in the dark. After, the slips were transferred to a glass beaker and fixed in Carnoys solution for 2 h on ice. Carnoys solution was replaced with 100% methanol and transferred to − 20 °C for 1 h. Samples were then rehydrated using mixtures of 75, 50 and 25% Methanol in PBSTx. [2]. After final rehydration with 100% PBSTx (PBS + 0.3% triton), the cells were blocked in PBST containing bovine albumin serum (BSA) for 4 h RT and primary antibody added overnight: α-RAD51 1:500 (Abcam ab109107). Wash 6 × 10 min in PBSTB and replace with secondary antibody overnight: 1:500 HRP-conjugated Goat anti-rabbit antibody (Millipore Cat# 12-348). Wash 6 × 10 min in PBSTB. Add TSA-Alexa488 in PBSTI 1:1000 for 20 min. Wash 6 × 10 min in PBSTB. Add DAPI (0.1 μg/1 mL) for 15 min and mounted using VECTASHIELD^® Mounting Medium.

FACS analysis

FACS analyses were performed as previously described [9, 23].

COMET assay and karyotyping

COMET assay was performed as previously described [9] and karyotyping was performed as previously described [63].

Single-cell sequencing (SCS) data

The Ubc9 SCS expression analysis was obtained from the planaria single-cell database hosted by the Reddien Lab at the Whitehead Institute for Biomedical Research (https://radiant.wi.mit.edu/app/) [1].

Imaging and data processing

Animal behavior was recorded using a Nikon AZ-100 multizoom microscope and NIS Elements AR 3.2 software. Area measurements were calculated with ImageJ and the difference in animal size were determined as fold change in reference to control group at each time point. Digital pictures were collected using a Nikon AZ-100 multizoom microscope and NIS Elements AR 3.2 software. Brightness and contrast were adjusted with Adobe Photoshop. Neoblasts were counted and normalized to the area (mm²) using ImageJ. Caspase-3 signal was quantified by measuring levels of fluorescence using NIS element software (Nikon).

Statistical analysis

Data are expressed as mean ± standard error of the mean (SEM) or fold change ± SEM. Statistical analyses were performed in Prism, GraphPad software Inc. (http://www.graphpad.com).

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contribution statement

MT, PGB, AT, LB, and NJO performed research and analyzed data. MT, LB, and NJO wrote the manuscript. All authors read the manuscript, provided comments and approved the final version.

Compliance with ethical standards

Conflict of interest

The authors declare no competing or financial interest.

Funding

We acknowledge support from University of California Merced and the Israel Science Foundation (ISF 1878/15) and the Israel Cancer Research Fund 14-101-PG to LB. This research was funded by the National Cancer Institute and National Institute of General Medical Sciences of the National Institute of Health, awards CA176114 and GM109372 to NJO.