Main

The infrequency of planarian fission behaviour has largely precluded its mechanistic dissection. However, recently optimized worm husbandry techniques augmented fission activity11,12, and enabled us to study the integration of worm size with fission behaviour. Large planaria (Schmidtea mediterranea) from recirculation culture systems exhibited robust and reproducible increases in fission activity when transitioned to static culture systems and starved (Fig. 1a, Supplementary Video 1). Live imaging provided detailed characterization of the fission process. Planarians first elongate and adhere their posterior tissue to a substrate. Next, periodic body contractions concentrate body mass towards the head region while thinning out tissues immediately anterior to the adherent tail. After 20–40 minutes, progressive stretching ruptures connecting tissue with rapid recoil, which separates the anterior parent from the posterior fission progeny (Extended Data Fig. 1a, Supplementary Video 1).

Fig. 1: Planarian fission is a size-dependent behaviour.
figure 1

a, Optimized fission protocol. b, c, Representative images of 5–12-mm worms and fission fragments less than 24 h (b) and 14 days (c) after the first fission event (n = 46 worms from 1 experiment). Scale bars, 5 mm. d, e, Length of the first fission fragments (d) and progeny number (e) over 2 weeks relative to parent length (n = 46 (d) or 30 (e) worms from one experiment). f, Webcam live-imaging schematic (left) and example timeline depicting successful (middle) and unsuccessful (right) fission attempts. g, Representative fission behaviour timelines from a range of parent lengths. h, i, Total fission attempts (h) and successful attempts per total attempts (i) relative to parent length (n = 39 (h) and 21 (i) worms). Data are from a single experiment. Pearson correlation co-efficient (PCC), linear regression (red line), and R2 values are provided.

Full size image

Observation of fission behaviour in worms of increasing size showed that the length of first posterior fission fragments did not correlate with parent length (Fig. 1b, d). Instead, larger worms produced additional progeny, each approximately 1 mm in length, such that the number of progeny after 2 weeks linearly correlated with parent size (Fig. 1c, e, Extended Data Fig. 1b–d). Thus, the size of fission fragments is fixed independently of anterior–posterior position or parent length. The frequency of the production of fission fragments—that is, the fission rate—did correlate with worm length (Extended Data Fig. 1e, f), and both the time to the first fission event and the time between sequential fission events was inversely related to parent size (Extended Data Fig. 1g–1). Automated webcam imaging of individual worms allowed us to generate timelines chronicling successful (upward displacement) and unsuccessful (downward displacement) fission attempts (Fig. 1f, Supplementary Video 2). Fission attempts occurred only in worms above 4–5 mm in length, which indicates a minimal size required for fission (Fig. 1g, h, Extended Data Fig. 2a, b). Furthermore, larger worms produced fission progeny more frequently owing to more fission attempts (Fig. 1h, Extended Data Fig. 2c, d), rather than higher rates of success (Fig. 1i). Together, these results confirm that planarian fission is a size-dependent behaviour, with both progeny number and fission rate coupled to parent size.

We tested the hypothesis that patterning cues are required to coordinate worm size and planarian fission. Genes from the Wnt13,14,15,16, TGFβ17,18,19 and Hh20 signalling pathways that regulate anterior–posterior identity were screened using RNA-dependent genetic interference (RNAi) techniques21 (Fig. 2a, b, Extended Data Fig. 3a, b). Rescreening confirmed six presumptive activators of fission (actR-1, smad2/3, β-catenin, dsh-B, tsh and wnt11-6) and a presumptive inhibitor (apc) (Fig. 2c). The morphology of parent worms was observed at days 0 and 14 of the fission assay and in regenerating tissue fragments. RNAi knockdown reproduced published anterior–posterior patterning defects in regenerating tissue fragments (Extended Data Fig. 4a), but few morphological defects were observed in parent worms (Fig. 2d). On day 0, β-catenin RNAi worms exhibited morphological abnormalities, whereas other RNAi conditions were indistinguishable from controls. By day 14, RNAi of actR-1 and smad2/3 elicited motility defects, but RNAi of dsh-B, wnt11-6, tsh and apc significantly altered fission rates without changes in morphology. In situ staining of the central nervous system (CNS), intestine and muscle confirmed published anterior–posterior polarity regeneration phenotypes, but no gross morphological defects in parent RNAi worms (Extended Data Fig. 4b–d). Therefore, we conclude that Wnt and TGFβ signalling components modulate fission behaviour independently of overt body plan repolarization.

Fig. 2: Wnt signalling and TGFβ signalling components modulate fission activity.
figure 2

a, RNAi screen workflow (see also Extended Data Fig. 3). b, c, Heat maps depicting fission activity after RNAi treatment for both the two-phase primary (b) and secondary (c) RNAi screens. Normalized cumulative fissions over time are displayed for individual worms from each RNAi condition (n = 10 worms for phase I, n = 12 worms for phase II and secondary screen). Targets in secondary screening (independently repeated three times) depicted in green (activators) and red (inhibitors). P values determined by two-way analysis of variance (ANOVA) interaction factor. Ctrl, control. d, Representative parent images on days 0 and 14 of the fission assay (n = 10–12, independently repeated 3 times). Scale bars, 1 mm.

Full size image

Serendipitously, we discovered that compression of planaria reveals cryptic mechanically vulnerable planes that divide the worm at regularly spaced intervals along the anterior–posterior axis (Fig. 3a, b, Supplementary Video 3). The number of these ‘compression planes’ scaled with worm size (Fig. 3b, c) and their position along the anterior–posterior axis overlapped with the position of fission planes (Fig. 3d). Furthermore, incomplete fission formed tears similar to those observed with compression (Extended Data Fig. 5a). Therefore, we conclude that compression planes are fission planes revealed by mechanical compression. Fission plane number and distribution correlated with worm length during tissue rescaling and regeneration. After starvation, worms reduced body length and lost fission planes to restore number and distribution (Extended Data Fig. 5b–d). To assay regeneration of the fission plane, we amputated worms around the pharynx such that 90% of fragments contained a single plane (Extended Data Fig. 5e–g). One week after amputation, worms remodelled, doubled in length and increased fission plane number (Extended Data Fig. 5f–j). Subsequent feeding increased worm length and fission plane number (Extended Data Fig. 5f–j). After starvation, worms exhibited little to no elongation or plane addition despite rescaling and regenerating their other tissues (Extended Data Fig. 5f–j). In summary, fission planes are pre-established in planarians and correlate dynamically with worm size and form.

Fig. 3: Pre-established fission planes determine progeny size independently of Wnt and TGFβ signalling.
figure 3

a, Schematic of compression assay (Supplementary Video 3). b, Pre- and post-compression worm (inset) and compression planes revealed in 3–6-mm worms (independently repeated 5 times). c, Compression plane number relative to worm length (n = 117 worms). PCC, linear regression (red lines), R2 values and 95% confidence interval (black lines) are shown. d, Fission (n = 196 fission progeny from 50 worms) and compression plane (n = 173 planes from 30 worms) overlap along the anterior–posterior axis of the worm. e, f, Representative images of post-compression worms after knockdown of Wnt and TGFβ signalling components using the specified number of RNAi feedings and rounds of regeneration (n depicted by dot plot quantification; experiment performed three times (e) or once (f)). Scale bars, 1 mm. g, Plot of the number of fission planes per worms length after RNAi treatment of 2 experiments (n = 20 and 10 worms to the left and right of the dotted line, respectively). P values determined by two-sided t-test. NS, not significant. Data are mean ± s.d. (d) or mean ± s.e.m. (g).

Full size image

Given the role of Wnt and TGFβ signalling in body patterning, we tested whether genes of these signalling pathways regulate fission planes. Worms treated with RNAi were mechanically compressed and the quantity and relative distribution of fission planes was measured (Fig. 3e–g). Notably, whereas RNAi of actR-1 and smad2/3 moderately reduced the number of fission planes, RNAi of Wnt signalling components had no effect on fission plane number or position (Fig. 3e, g, Extended Data Fig. 6a). Even knockdown of wnt11-6 by three rounds of amputation and regeneration did not alter fission-plane patterning (Fig. 3f, g, Extended Data Fig. 6b). Hypomorphic RNAi knockdown of β-catenin, actR-1 or smad2/3 revealed little or no effect on the size of fission fragments (Extended Data Fig. 6c–e), which further supports the conclusion that neither Wnt nor TGFβ signalling regulate fission behaviour through the anterior–posterior patterning of fission planes.

We tested whether Wnt and TGFβ signalling instead regulated the frequency of fission attempts. Using the automated webcam image-capture system (Fig. 1f), we recorded fission behaviour in RNAi-treated worms (Fig. 4a). RNAi of β-catenin, actR-1, smad2/3 and wnt11-6 reduced fission attempts, whereas RNAi of apc increased fission attempts (Fig. 4b–d, Extended Data Fig. 7a–l, Supplementary Videos 4–6). RNAi of β-catenin and smad2/3, which resulted in observable morphological abnormalities, also significantly reduced the fission-success ratio (Figs. 2d, 4e, Extended Data Fig. 7k–n). dsh-B RNAi reduced the fission success ratio without altering the number or frequency of fission attempts (Fig. 4d, e, Extended Data Fig. 7k–n). Finally, apc RNAi reduced the time between fission attempts by approximately 50%, and worms initiated fission attempts independently of remaining tissue, markedly reducing their success ratio (Fig. 4e, Extended Data Fig. 7i–n, Supplementary Video 6). These findings demonstrate that Wnt and TGFβ signalling regulate the frequency of fission behaviour.

Fig. 4: Wnt and TGFβ signalling regulates fission frequency by size-dependent patterning of mechanosensory neurons in the CNS.
figure 4

a, Schematic depicting RNAi treatment, live imaging and data analysis. b, c, Representative activity timelines. d, e, Total fission attempts (d) and successful attempts per total attempts (e) for RNAi-treated worms (n = 16 (d) and 10 (e) worms, from 1 (d) or 2 (e) independent experiments). f, Schematic depicting in situ staining strategy. g, h, Representative images of pkd1L-2+ and gabrg3L-2+ neurons in worms of increasing size (g), or after RNAi treatment (h). Scale bars, 0.5 mm. i, k, Diagrams depicting quantification of the angle of pkd1L-2+ cells (i) or the range of gabrg3L-2+ cells (k). j, l, Staining quantification of pkd1L-2 (j) and gabrg3L-2 (l) in worms of increasing size or after RNAi treatment (n = 3–6, exact n depicted in dot plot quantification). m, Fission activity heat maps after treatment with pkd1L-2 and gabrg3L-2 RNAi (n = 12; Fig. 2). n, Representative parent images on days 0 and 14 of fission assay (n = 12, 2 independent experiments). Scale bars, 1 mm. o, Representative fission activity timelines of worms treated with gabrg3L-2 RNAi. p, q, Total fission attempts (p) and successful attempts per total attempts (q) for worms treated with gabrg3L-2 RNAi (n = 10 worms). P values determined by two-sided t-test (j, l, p, q) or two-way ANOVA (d, e). Data are mean ± s.e.m. (d, e, j, l, p, q).

Full size image

We proposed that components of the Wnt and TGFβ signalling pathways might regulate fission behaviour through the planarian CNS. Double fluorescent in situ hybridization (FISH) with the CNS marker pc2 confirmed that Wnt and TGFβ fission regulators were detected in pc2-positive cells in the anterior CNS (Extended Data Fig. 8a, b). Removal of anterior tissue that contains the cephalic ganglia delayed the onset of fission behaviour (Extended Data Fig. 8c–f). Restoration of fission activity coincided with regeneration and re-establishment of anterior, pc2 co-localized, tsh expression (Extended Data Fig. 8g). Notably, removal of anterior tissue that contained just one cephalic ganglion did not alter the total number of fission progeny produced (Extended Data Fig. 8c–f), which indicates that half of the CNS is sufficient to initiate fission. Finally, RNAi against coe, a transcription factor essential for the patterning of the CNS22,23, markedly reduced planarian fission (Extended Data Fig. 8h, i). Together, these data support a model in which an anterior CNS expressing Wnt and TGFβ signalling components regulates fission initiation.

We tested whether polarity genes could modulate size-dependent behaviour via size-dependent patterning of the CNS. To identify neuronal subpopulations that regulate fission downstream of Wnt and TGFβ, we analysed 17 neuronal markers24,25,26,27,28,29 in small, medium and large planaria and 10 markers in worms treated with smad2/3 RNAi (Fig. 4f, Extended Data Fig. 9a, b). Patterning of pkd1L-2+, gabrg3L-2+ and sargasso-1+ mechanosensory neurons exhibited the clearest changes in worms of increasing size and after smad2/3 RNAi treatment (Extended Data Fig. 9a, b). In large worms, mechanosensory neurons are tightly restricted to the anterior and knockdown of either smad2/3 or wnt11-6 broadened their distribution akin to that of smaller worms (Fig. 4g–l). RNAi against pkd1L-2 and gabrg3L-2 (homologous to cation and chloride channel genes, respectively) increased planarian fission activity (Fig. 4m, n), and live imaging of gabrg3L-2 RNAi worms confirmed an increase in fission attempts without a reduction in fission success (Fig. 4o, p, Extended Data Fig. 10, Supplementary Video 7). These results indicate that mechanosensory neurons are differentially patterned during growth, inhibit fission behaviour and require Wnt and TGFβ for their appropriate patterning in accordance with worm size. Therefore, we conclude that Wnt and TGFβ signalling coordinates worm size and behaviour via size-dependent patterning in the adult CNS.

In conclusion, we used planaria as a model for the integration of size, patterning and function and established fission as a robust, reproducible and quantifiable size-dependent behaviour (Fig. 1, Supplementary Video 1). Although previous studies have generated physical models for the process of transverse fission9, mechanisms that couple worm size and fission frequency have remained unknown. We discovered two independent mechanisms by which fission is coordinated with worm size in S. mediterranea. First, previously undescribed iterative structures patterned in accordance with anterior–posterior axis length couple worm size with the number of fission progeny produced (Fig. 3, Supplementary Video 3). Second, the Wnt and TGFβ signalling pathways mediate size-dependent patterning of mechanosensory neurons, which regulate fission frequency (Fig. 4, Extended Data Figs. 9, 10). Thus, we demonstrate that differential patterning of key cell populations in accordance with tissue size provides a mechanistic link between worm growth and the acquisition or modulation of tissue function. Together, our results identify a role for Wnt and TGFβ patterning genes in the regulation of size-dependent behaviour and show that developmental patterning cues coordinate tissue growth with size-dependent functions.

Methods

Worm husbandry

Clonal CIW4 S. mediterranea were maintained in 1× Montjuic salts as previously described. CIW4 worms were sourced from a large recirculation culture as previously reported11. In brief, worms are housed in three culture trays (244 cm length × 61 cm width × 30.5 cm height) stacked vertically. Water is recirculated through the system by a sump pump, which moves water through a chiller, a canister filter, a UV sterilizer and the three housing trays. Water is then passed through two vertically stacked sieves and a set of filter/floss pads before being returned to the sump pump. Worms were pulled from this system and placed directly into fission assays, starved for at least seven days before tissue fixation for imaging, or transferred to a unidirectional flow system culture for controlled feeding or RNAi feeding experiments.

Gene cloning and RNAi feeding protocol

Candidate genes analysed in this study were cloned from a CIW4 cDNA library into a pPR-T4P vector as previously described20 (Supplementary Table 1). These served as template for in vitro synthesis of dsRNA for RNAi feedings. Unc22 dsRNA was used for control RNA treatment. RNAi food was prepared by mixing 1 volume of dsRNA at 1,600 ng ml−1 with 1.5 volumes of beef liver paste. For RNAi experiments that target neuronal genes, 1 volume of dsRNA at 1,400 ng μl−1 was mixed with 1 volume beef liver paste. The amount of food administered was 10 μl of food per 1 mm of worm length present in the worm flow container. Worms were allowed to feed for 6–10 h with 2 rounds of light stimulation to facilitate additional consumption. Worms were fed every three days for a total of three RNAi feedings, unless otherwise specified. After RNAi feedings, worms were transferred to the relevant biological assay.

Fission assay

A detailed protocol for fission induction has been made available through Protocol Exchange10. To induce fission, worms were removed from recirculation culture or unidirectional flow system culture and washed 5–10 times with fresh 1× Montjuic salts. Individual worms were placed in 15-cm tissue culture dishes with 50 ml 1× Montjuic salts and their body length was measured. Representative images of day-0 parents were captured using a Leica M205 microscope. Plates were stacked 6–12 dishes high and placed in a dark incubator at 20 °C. Daily, plates were removed from the incubator and fission fragments for each worm were counted and removed from the 15-cm dish. For some experiments, images of fission fragments were taken on the day they were collected to allow for quantification of fission fragment length. The 1× Montjuic salts in each individual dish was replaced weekly.

For data analysis, the number of daily cumulative fissions was divided by initial body length and then normalized to the average of the control RNAi fissions. This normalized fission score for each day was converted to a heat colour code. Daily scores for each individual worm were aligned in descending order along the y axis and the average score of each column was calculated and used to sort individual worms in ascending order along the x axis. The average fission score of each RNAi condition was then sorted in ascending order from left to right. This resulted in a heat map visualization ranking the effects of RNAi treatments on fission activity.

Fission plane compression assay

Fission planes were revealed by compression between a plastic tissue culture dish and a glass coverslip (Supplementary Video 3). Worms were inverted with their ventral side up, compressed using four fingertips, then imaged. To ensure that all compression/fission planes were revealed for every worm, images were acquired sequentially using a Leica M205 microscope as each fission plane was revealed by mechanical compression. Position of fission planes and distance between fission planes was quantified using Fiji (https://fiji.sc/). Video depicting compression assay was captured with an iPhone 6 (Apple).

Whole-mount FISH

For RNA expression analyses, FISH was performed as previously described30,31. Antibodies were used in MABT containing 5% horse serum for FISH (Roche anti-DIG-POD 1:1,000 and Roche anti-FLCN-POD 1:1,000) or NBT/BCIP in situ hybridization (Roche anti-DIG-AP 1:1,000). For double FISH, peroxidase activity was quenched between tyramide reactions using 100 mM sodium azide for at least 1 h at room temperature with agitation. Nuclear staining was performed using 1:1,000 Hoescht 33342 (Invitrogen) in PBST (1× PBS with 0.5% Triton-X-100).

Microscopy

Images of live worms and regenerating fission fragments were acquired using a Leica M205 microscope. Confocal images were acquired on an LSM-700-Vis and stitching was performed in Fiji using built-in grid collection plugins.

Live imaging of fission behaviour

Videos of worms from two orthogonal views were acquired using two webcams (Logitech C910/920). Webcams were mounted using a variety of ring stands and test tube clamps. The imaging chamber was a clear plastic square lid obtained from a box of coverslips. Lighting of the chamber was achieved using a Volpi illuminator (NCL-150). Each camera was connected to its own computer running micro-manager (https://micro-manager.org/). The cameras were set up in micro-manager using OpenCVgrabber to set the pixel density (1,920 × 1,080) and to acquire the images. The camera gain, exposure and all other settings were set using the Logitech Webcam Controller software (https://download01.logi.com/web/ftp/pub/video/lws/lws280.exe). Data were acquired using the Multi-Dimensional Acquisition mode of micro-manager. The two computers were synchronized for acquisition manually at the beginning of the experiment.

For the high-throughput screening of fission behaviour, worms were placed in six-well dishes with cameras mounted above the plates using optics components (Thor Labs). Illumination was obtained using four LED ring lights (AmScope) mounted upside down and above the cameras to provide diffuse light. Image acquisition was performed using two different camera configurations: four cameras connected to one computer via a USB hub or one 4K camera connected to a USB port. In the four-camera configuration, images where captured sequentially from the cameras every ten minutes. A script written in Python 3.6 (https://www.python.org/) was used as a wrapper for FFMPEG (https://www.ffmpeg.org/) to acquire images. The size of the images (1,920 × 1,080) and the pixel format (yuv420p) were set in the python script. The camera gain, exposure and other settings were controlled with the Logitech Webcam Controller software (https://download01.logi.com/web/ftp/pub/video/lws/lws280.exe). The DirectShow framework was used to interface between the cameras and FFMPEG. In the single 4k camera setup, a 4096 × 2160-pixel image was captured every ten minutes from a Logitech BRIO webcam. The same Python script was used as a wrapper for FFMPEG in this configuration.

Quantification of live imaging

Videos of individual worms were manually annotated. For each fission attempt, the start time and completion time were recorded and the success or failure of the attempt was recorded. To depict fission behaviour, a timeline was constructed and a numerical value was given to each frame of a video. A value of 0 was assigned to any frame in which no fission behaviour was observed; a positive value was given to any frame during a successful fission attempt; and a negative value was given to any frame during a failed fission attempt (see Fig. 1f). A prolonged diagonal line in a timeline indicates a period in which frames were not acquired owing to failed communication between the image acquisition software and the webcam.

Statistical tests

For all pairwise comparisons, significance was tested using an unpaired Student’s t-test. GraphPad Prism was used to calculate PCC values with a two-tailed 95% confidence interval and to perform linear regression analyses. Two-way ANOVA analysis was performed in GraphPad Prism to determine the significance of RNAi treatment over time. No statistical methods were used to predetermine sample size. The experiments were not randomized, and investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.