A clonal strain of the planarian Dugesia japonica was used in all experiments (Wang et al., 2021). The colony was maintained in autoclaved stream water in the dark at 20°C. Planarians 6–8 mm in length were starved for at least one week. Two regenerating fragments were obtained via transverse amputation at the anterior of the pharynx.

DNAMAN software (7.0) was used to analyze the obtained cDNA and deduced the amino acid sequence of Djpi3k. The conserved domains of Djpi3k protein were predicted using the NCBI Conserved Domain Database (Lu et al., 2020) and the SMART database (Letunic et al., 2021). The ProtParam online tool¹ was used to calculate the molecular weight and isoelectric point (pI) of Djpi3k protein. Through the neighbor-joining (NJ) method, a phylogenetic tree was constructed via MEGA software (7.0), and further annotation was conducted by the Evolview tool using a Pfam database (He et al., 2016).

At 0 h, 4 h, 8 h, 12 h, 1 days, 3 days, 5 days, 7 days post-amputation, head- and tail-regenerating tissues were harvested and grinded using a homogenizer, respectively. For each replicate, 5–10 planarians were homogenized together, and total RNA was isolated using RNAiso Plus reagent (Takara, China) according to the standard protocol. Next, cDNA was synthesized from 1 μg of total RNA using Transcriptor First Strand cDNA Synthesis Kit (Roche, Switzerland). The qRT-PCR analysis was performed using SYBR Premix Ex Taq (TaKaRa, China) and a CFX96^TM Real-Time PCR Detection System (Bio-Rad, United States) with the following parameters: 95°C for 10 s; followed by 40 cycles at 95°C for 10 s, 60°C for 10 s, and 72°C for 10 s. The β-actin housekeeping gene was selected as an internal reference. The relative gene expression of each gene was calculated using the 2^–ΔΔCT method. PCR primers used in this study were shown in Supplementary Table 1.

The PI3K-specific inhibitor LY294002 (Abmole Bioscience, United States) or A66 (Abmole Bioscience, United States) was dissolved in DMSO and used at a series of gradient concentrations. Planarians were allowed to regenerate in water supplemented with LY294002 or A66 immediately after amputation until the indicated period of regeneration. For homeostasis experiments, intact animals were maintained in water supplemented with 30 μM LY294002 for 20 days.

RNAi was conducted by oral delivery of bacterially expressed dsRNA as described previously (Newmark et al., 2003; Zeng et al., 2013). Briefly, two fragments against the Class I PI3K accessory and catalytic domains of the Djpi3k gene were introduced into the L4440 vector (BioVector NTCC, China) between two convergent T7 promoters. The RNAi vectors were transformed into an RNaseIII-deficient E. coli strain HT115 (BioVector NTCC, China) to generate dsRNA by IPTG induction (0.4 mM, Amresco, United States). Then animals were fed four rounds with a mixture of the dsRNA-expressing bacteria and liver paste. Animals fed with HT115 bacteria carrying a fragment encoding EGFP served as a negative control. Three days after the last feeding, the RNAi efficiency was examined by whole-mount in situ hybridization (WISH) and qRT-PCR.

Whole-mount in situ hybridization was performed using a slightly modified protocol previously described (Pearson et al., 2009). Digoxigenin-labeled antisense RNA probes were synthesized using the in vitro RNA labeling kit (Roche, Germany). Prior to 5% N-Acetyl Cysteine treatment and 4% paraformaldehyde fixation, planarians were treated with 2% HCl to kill and remove mucus for 5 min. After bleaching and rehydration, specimens were treated with proteinase K (10 mg/mL, Merck, Germany) and fixed in 4% paraformaldehyde at room temperature. Hybridizations with the labeled probe in the concentration of 1 ng/μL were performed at 56°C for 16 h after pre-hybridization. Next, specimens were washed with preheated (to 56°C) gradient solutions. Blocking and incubation of the anti-Digoxigenin-AP antibody (1:5,000, Roche, Germany) were done overnight at 4°C. Following 5-Bromo-4-chloro-3-indolyl phosphate and nitroblue tetrazolium (Roche, Germany) development to detect the Dig-labeled riboprobe.

The planarians were sacrificed in ice-cold 2% HCl for 2–5 min and fixed in 4% paraformaldehyde, then rinsed in 4°C methanol for 1 h and bleached in methanol containing 6% H₂O₂ overnight. After rehydration, specimens were blocked in 0.6% BSA and 0.45% fish gelatin diluted with PBSTx for 1 h at room temperature and incubated sequentially in primary antibody concentrations were used as follows: anti-phospho-Histone H3 (anti-H3P, 1:250, R&D Systems, United States); anti-Acetylated Tubulin (anti-Ac-Tubulin, 1:500, Sigma, United States); anti-SYNORF1 (1:50, Developmental Studies Hybridoma Bank, United States). Secondary antibody concentrations were: goat-anti-mouse IgG (1:200, Millipore, United States), and goat anti-rabbit IgG (Alexa Fluor488) (1:1,000, Abcam, United Kingdom). The signals were photographed and fluorescent images were captured and analyzed with LEICA SP8 confocal microscope and ImageJ software.

A whole-mount TUNEL assay using the planarians was performed as described previously (Pellettieri et al., 2010). After fixation, the planarians were permeabilized in 1% SDS for 20 min and bleached overnight in bleached in methanol containing 6% H₂O₂ overnight. Next, specimens were incubated with the terminal transferase enzyme diluted in label solution (Roche, Germany) for 4 h at 37°C, and then rinsed in PBSTB (PBST with 0.25% BSA), and these animals were incubated in anti–digoxigenin-rhodamine diluted in blocking solution for 4 h at room temperature. DAPI was used to stain nuclei. The signals were photographed and fluorescent images were captured and analyzed with LEICA SP8 confocal microscope and ImageJ software.

Planarian live and WISH images were acquired with Canon EOS 600D and OLYMPUS SZX10 microscopes, respectively. Z-stacks of fluorescently labeled samples were acquired with a LEICA SP8 confocal microscope and processed with the ImageJ package Fiji. For quantification of H3P-positive or TUNEL-positive cells, stacks with equal numbers of z-sections were taken and cells were counted in the indicated areas using ImageJ package Fiji. To quantify cilia density, the fluorescence intensity of the Ac-tubulin signal was measured and normalized to DAPI fluorescence. To assess blastema formation, the length of the regenerated blastema was measured using ImageJ package Fiji and normalized to the whole-body length of the trunk. The ratio of blastema/whole-body length of small molecules treated animals or Djpi3k(RNAi) animals were compared with control animals at each time point.

Values were expressed as mean ± SEM. The statistical differences were calculated using the Student’s t-test in SPSS 21.0 software. Statistical significance is indicated as follows: ^∗P < 0.05, ^∗∗P < 0.01, ^∗∗∗P < 0.001, and ns, not significant.

A planarian PI3K-like gene, Djpi3k, was cloned from Dugesia japonica and contains an ORF of 3,075 bp. The predicted Djpi3k protein is composed of 1,024 amino acid residues and shares conserved functional domains with the catalytic subunit of Class I PI3K, including p85-binding domain, Ras-binding domain, C2 domain, PI3K accessory, and catalytic domains (Supplementary Figure 1). The molecular weight of the predicted Djpi3k protein was estimated to be 256.21 kDa, and the pI was estimated at 4.98.

The catalytic subunits of Class I PI3K includes four highly homologous p110 isoforms (p110α,p110β, p110γ, and p110δ) encoded by genes PIK3CA, PIK3CB, PIK3CG, and PIK3CD, respectively (Zhao and Vogt, 2008). To investigate which subgroup the Djpi3k is classified, 30 genes encoding the p110 catalytic isoforms from 13 representative species were selected for phylogenetic analysis. As shown in Figure 1, all selected genes were mainly categorized into four clusters corresponding to PIK3CA (blue branch), PIK3CB (yellow branch), PIK3CG (green branch), and PIK3CD (red branch), among which Djpi3k (A0A222C681_DUGJA) was classified into the PIK3CA subgroup as well as another planarian ortholog (D2DJU4_SCHMD) from Schmidtea mediterranea. Domain annotations for each sequence were determined using the Pfam database (Mistry et al., 2021), and the predicted domains were consistent with the results from other databases (Supplementary Figure 1).

To characterize the expression pattern of Djpi3k during planarian regeneration, animals were cut anterior of the pharynx and allowed to regenerate for 7 days (Figure 2A), samples at different regenerating time were fixed and subjected to WISH. The specification of the Djpi3k probe was confirmed in intact animals using a Djpi3k sense probe as a negative control and a neoblast marker gene DjpiwiA as a positive control (Supplementary Figures 2A,B). The WISH results showed that Djpi3k was expressed immediately after amputation at the wound surface, suggesting that the gene was in response to injury/tissue absence. High levels of Djpi3k expression were detected in the newly-regenerated blastema and pharynx-forming regions during regeneration. The Djpi3k level decreased gradually as the regeneration became complete (Figure 2B). The dynamic change of Djpi3k expression during planarian regeneration was also analyzed using qRT-PCR analysis during anterior and posterior regeneration (Figure 2C). The qRT-PCR results indicated that Djpi3k levels were upregulated following amputation, peaked at 1 day post amputation (dpa), and then fell to basal level until the regeneration was complete, which was consistent with the results of WISH analysis.

To investigate the role of PI3K signaling in planarian regeneration, we inactivated the PI3K signaling by using LY294002, a selective PI3K inhibitor (Walker et al., 2000). Animals were amputated and exposed to a gradient concentration of LY294002 for 7 days. Treatment with 45 μM LY294002 caused damage to the epidermis followed soon by lysis, while animals survived by treatment with a moderate concentration at 30 μM (Supplementary Figure 3). However, both head and tail regeneration were largely impaired after 30 μM LY294002 treatment (Figure 3A). The LY294002 treated fragments only regrew marginal blastemas at 7 dpa, which were significantly smaller than those of the control animals (Figure 3B). In addition, no eyes were formed in the regenerating tail fragments at 7 dpa when the control animals had regenerated their eyes properly (Figure 3A). A similar inhibitory effect was observed in the animals treated with A66 (80 μM, Supplementary Figures 4A,B), a highly specific and selective p110α inhibitor (Jamieson et al., 2011). The planarian brain is mainly composed of four structurally and functionally different regions including mechanosensory neurons (yellow), lateral branch neurons (blue, nine asterisks), two main lobes (red), and photosensory neurons (green) (Agata and Umesono, 2008). We tested whether planarian brain regeneration was disturbed after treatment of LY294002 using an anti-SYNORF1 antibody that recognizes brain structures. The immunostaining results revealed that PI3K inhibition caused apparent defects in brain regeneration at 7 dpa, mainly manifested as incomplete main lobes, disappeared lateral branch neurons, and the brain commissure of the two lobes not connected in the top (Figure 3C). These data indicate that PI3K signaling is required for appropriate regeneration in planarians.

We next examined the effect of PI3K inhibition on planarian homeostasis, a self-regulating process allowing for tissue maintenance during adult cell turnover. After a 20-day treatment with LY294002, slight ulceration of the epidermis was observed in quite a few animals (12/40), and tissue lysis could be found in several animals (4/40) (Figure 3D). Although more than half of the animals appeared no morphological abnormality after LY294002 treatment, their locomotion was considerably hampered. When exposed to light, the control animals moved quickly away from the light and reached the dark field within 2 min. On the contrary, LY294002 treated animals twisted forward, flipped over, or stayed still during their traveling toward the dark field. No animals reached the dark field until 3 min post light exposure, and two-third of the animals never arrived at the dark field even at 20 min after light exposure (Figure 3E and Supplementary Videos 1, 2). Plane gliding is mediated by synchronous cilia movement on the ventral surface of planarians (Rink et al., 2009). To verify the possibility that LY294002-induced impaired gliding ability was caused by cilia defect, we used whole-mount staining with the anti-Ac-tubulin antibody to visualize the integrity of ciliary epithelial cells in the ventral epithelium. The LY294002 treated animals displayed poor staining in cilia that were severely shortened, whereas the control animals showed dense coverage of cilia (Figures 3F,G). Our results suggest that impaired motility in the Djpi3k inhibition phenotype is most likely due to insufficient density of ventral epithelial cells. Thus, these results indicated that PI3K activation is required for planarian homeostasis, especially cilia maintenance.

RNAi has been widely used to study regeneration-associated genes in planarians (Sánchez Alvarado and Newmark, 1999; Newmark et al., 2003; Rouhana et al., 2013). A bacterial feeding protocol (Newmark et al., 2003; Zeng et al., 2013) was employed in this study to investigate the role of Djpi3k in planarian regeneration and homeostasis. RNAi efficiency was confirmed by WISH and qRT-PCR analysis (Figures 4A,B). Djpi3k(RNAi) animals also exhibited limited regenerative capacity compared with the control animals (Figures 4C,D). Different from the LY294002 treatment, RNAi-mediated PI3K inhibition led to relatively mild phenotypes and a lower proportion of animals with regeneration defects (27/50 vs 46/50 in head regeneration, 29/50 vs 48/50 in tail regeneration). A similar situation occurs in tissue homeostasis, decreased cilia density was observed in Djpi3k(RNAi) animals, nevertheless, the integrity of the ciliated structure was maintained (Figures 4E,F). Taken together, these data suggest that PI3K signaling is required for planarian regeneration and tissue maintenance.

PI3K signaling participates in multiple biological processes associated with cell growth and proliferation (Madsen, 2020). Neoblasts represent the only proliferative cell population in planarians and are therefore the source of new cells for normal tissue turnover, growth, and regeneration (Wagner et al., 2011). We checked the effect of PI3K inhibition on the mitotic activity of neoblasts during planarian regeneration using the anti-H3P antibody, which labels cells in the G2/M phase (Peiris et al., 2012, 2016). In intact planarians, the H3P-positive cells were detected distributed throughout the body except in the head region anterior to the eyes and the pharynx region (Supplementary Figure 2C), which was consistent with the distribution patterns of planarian neoblasts reported in previous studies (Newmark and Sánchez Alvarado, 2000; Salvetti et al., 2000) and further verified the specificity of the H3P staining results in our study. For regeneration studies, the number of the H3P-positive cells in the control animals reached the first peak of mitosis at 6 hpa and the second peak at 3 dpa, which was consistent with previous studies (Wenemoser and Reddien, 2010). However, the number of the H3P-positive cells was dramatically decreased after LY294002 treatment, making the mitotic response vanished at the early stage of planarian regeneration (Figures 5A,B). Knockdown of Djpi3k also reduced the neoblast number, the mitotic response of Djpi3k(RNAi) animals was alleviated but still existed (Supplementary Figures 5A,B). Moreover, the expression pattern of Djmcm2, a neoblast marker (Salvetti et al., 2000), was evaluated before and after PI3K inactivation. The qRT-PCR results showed that the expression levels of Djmcm2 could reflect the mitotic trends as manifested by H3P staining, and were substantially suppressed after LY294002/A66 treatment or Djpi3k knockdown (Figure 5C and Supplementary Figures 4C, 5C). Our results suggest that PI3K signaling is required to maintain the appropriate number of proliferating neoblast during tissue renewal in adult planarians.

Apoptosis, a prominent form of cell death in planarians, facilitates cell turnover during homeostasis and regeneration (Pellettieri et al., 2010). To investigate the possible role of PI3K signaling in apoptotic cell death during planarian regeneration, the spatiotemporal distribution of apoptosis was evaluated by TUNEL staining. The results revealed that the number of apoptotic cells in both the head and tail fragments was substantially decreased throughout the regeneration period after LY294002 treatment (Figure 6A). The system-wide cell death expected at 3 dpa was also suppressed by LY294002 treatment (Figure 6A and left panel of Figure 6B). It was noted that the localized cell death response at 6 hpa in control animals was also largely impaired by LY294002 treatment (Figure 6B, right panel). Similarly, cell death in Djpi3k(RNAi) animals failed to localize at the amputation site as is expected at 6 hpa (Supplementary Figures 6A,B). Nevertheless, Djpi3k(RNAi) animals showed a delayed apoptotic peak at 1 dpa during head and tail regeneration. Moreover, the expression pattern of Djbax, a pro-apoptotic marker gene, was evaluated by qRT-PCR before and after PI3K inhibition, and the results were consistent with the TUNEL staining results (Figure 6C and Supplementary Figures 4D, 6C). These findings indicate that PI3K signaling is essential for controlling wound-induced apoptotic cell death during tissue regeneration. Taken together, we propose Djpi3k functions as a regulator of proliferation and apoptosis during large-scale tissue regeneration in planarians.