Intestine-enriched apolipoprotein b orthologs are required for stem cell progeny differentiation and regeneration in planarians

Lipid metabolism plays an instructive role in regulating stem cell state and differentiation. However, the roles of lipid mobilization and utilization in stem cell-driven regeneration are unclear. Planarian flatworms readily restore missing tissue due to injury-induced activation of pluripotent somatic stem cells called neoblasts. Here, we identify two intestine-enriched orthologs of apolipoprotein b, apob-1 and apob-2, which mediate transport of neutral lipid stores from the intestine to target tissues including neoblasts, and are required for tissue homeostasis and regeneration. Inhibition of apob function by RNAi causes head regression and lysis in uninjured animals, and delays body axis re-establishment and regeneration of multiple organs in amputated fragments. Furthermore, apob RNAi causes expansion of the population of differentiating neoblast progeny and dysregulates expression of genes enriched in differentiating and mature cells in eight major cell type lineages. We conclude that intestine-derived lipids serve as a source of metabolites required for neoblast progeny differentiation. Lipid metabolism regulates stem cell states and differentiation. Here, the authors demonstrate a requirement in planarians for Apolipoprotein B-mediated neutral lipid transport from intestinal stores to stem cells and their progeny during differentiation and whole-body regeneration.

R egeneration requires metabolites and energy for cell proliferation, differentiation, migration, and growth. Currently, specific mechanisms by which various metabolites are produced, transported, and utilized to promote regeneration are not well understood. In animals, neutral lipids (NLs) (e.g., triglycerides and cholesteryl esters) are stored in intracellular lipid droplets, which serve as central organelles for energy and lipid homeostasis 1,2 . Lipid stores can be mobilized either by NL packaging and export in lipoprotein particles (LPs), or by lipolysis and secretion as fatty acids (FAs) [3][4][5] . Mobilization enables transport to other tissues, for example between intestine, liver, and peripheral tissues such as brain and muscle in vertebrates, or between the fat body, intestine, nervous system, and imaginal discs in Drosophila [6][7][8] . Upon lipase-mediated hydrolysis of NLs, the FAs and cholesterol produced serve as building blocks for membrane biosynthesis, substrates for energy via fatty acid beta oxidation, sources of acetyl-CoA and other precursors for chromatin remodeling, and precursors of signaling molecules like eicosanoids and sphingolipids [9][10][11] .
Considerable evidence indicates that disruption of lipid catabolism, biosynthesis, and transport can dysregulate stem cell pluripotency, proliferation, and differentiation [12][13][14] . For example, inhibiting fatty acid oxidation (FAO, the process of converting fatty acids to acetyl-CoA) causes symmetric production of differentiating progeny and stem cell depletion in the mammalian hematopoietic lineage 15 . By contrast, in the adult mouse hippocampus, inhibiting FAO promotes exit from quiescence and proliferation of neural stem/progenitor cells in the adult mouse hippocampus 16 . In Drosophila, compromising lipolysis and/or FAO causes intestinal stem cell necrosis 17 , and loss of germline stem cells 18 . On the other hand, FAO-mediated production of acetyl-CoA (a precursor for histone acetylation) is required for differentiation, but not maintenance, of quiescent hematopoietic stem cells in Drosophila larvae, illustrating the growing recognition of the intersection of lipid metabolism and epigenetic regulation 19 . Perturbing lipid synthesis and delivery also dysregulates stem cell state and differentiation dynamics. For example, inhibiting fatty acid synthase (required for de novo lipogenesis) reduces proliferation by adult mouse neural stem and progenitor cells 20 . Similarly, mutations in hydroxysteroid (17-beta) dehydrogenase 7, a regulator of cholesterol biosynthesis, cause premature differentiation of neural progenitors during development 21 . In culture, depriving human pluripotent stem cells of extrinsic lipids promotes a "naive-to-primed" intermediate state, demonstrating the importance of exogenous lipid availability for regulation of stem cell state 22 .
Because stem and progenitor cell activation, proliferation, and differentiation are central to regeneration, these diverse observations underline the potential importance of lipid metabolism during regenerative growth. Currently, however, evidence that lipid transport and utilization influence regeneration is limited. In zebrafish, leptin b (a hormonal regulator of systemic lipid metabolism) is one of the most upregulated genes during fin and heart regeneration 23 , and apolipoprotein E (a regulator of NL transport in LPs) is upregulated during fin regeneration 24 . In the axolotl Ambystoma mexicanum, genes involved in steroid, cholesterol, and fatty acid metabolism are among the most upregulated at later stages of limb regeneration, when differentiation of cartilage and muscle progenitors occurs 25 . In a primary cell culture model of nerve injury, rat retinal ganglion cells regrow projections after axotomy more efficiently in the presence of glial-derived LPs, reinforcing the role of cholesterol in axon regeneration 26 . In mice, low density lipoprotein receptor deficiency delays liver regeneration and reduces hepatocyte proliferation 27 . Similarly, elevated FA levels (induced by lipoprotein lipase overexpression) causes lipotoxicity and compromises skeletal muscle regeneration 28 , while inhibition of peroxisomal FAO induces differentiation of myogenic satellite cells and muscle hypertrophy during regeneration 29 . During skin wound repair, inhibiting triglyceride lipase-mediated lipolysis by dermal adipocytes compromises recruitment of inflammatory macrophages and adipocyte fateswitching to extracellular-matrix-secreting myofibroblasts 30 .
Although these intriguing observations point to the importance of lipid metabolism, there is scant functional understanding of how lipid transport and utilization influence stem cell regulation during regeneration, particularly in emerging animal models with extensive regenerative capacity. Planarians are freshwater flatworms capable of whole-body regeneration, an ability conferred by pluripotent somatic stem cells called neoblasts that divide and differentiate to replace damaged and lost tissues after amputation 31,32 . Diet-derived NLs are stored in the planarian intestine in lipid droplets, suggesting the intestine is a major lipid storage organ, as in Drosophila and C. elegans 6,[33][34][35] . Planarian intestinal lipids have been proposed to be a source of metabolites during extended fasting, as well as regeneration 34,36,37 . However, mechanisms by which lipid secretion is controlled, and whether delivery to neoblasts or their progeny is functionally required for regeneration, have not been investigated.
In this study, we show that two intestine-enriched apolipoprotein b (apob) orthologs are required for NL transport from the intestine to neoblasts and their progeny, and that ApoB function is required for stem cell progeny differentiation and regeneration in the planarian Schmidtea mediterranea. In mammals and insects, ApoB and ApoB-like proteins mediate trafficking of NLs in LPs from digestive and storage organs to peripheral target tissues 5,38 . Here, we identify adult stem cells and their differentiating progeny as additional target tissues for ApoB-mediated NL transport in a regeneration competent animal, and propose utilizing planarians as models to understand the influence of lipid metabolism on stem cells during regeneration.

Results
Planarian apolipoprotein b orthologs are expressed by intestinal cells. Previously, we demonstrated that knockdown of an intestine-enriched transcription factor, nkx2.2, inhibited neoblast proliferation and formation of the regeneration blastema, the unpigmented mass of tissue produced after amputation 39 . These observations suggested that the intestine could play a nonautonomous role in regulating stem cell dynamics. In an effort to identify intestine-enriched transcripts encoding neoblast regulators, we generated transcriptomes from control and nkx2.2(RNAi) planarians ( Supplementary Fig. 1a-c) 39 . An ortholog of human apolipoprotein b ("apob-1") encoding conserved Vitellogenin, DUF1943, and von Willebrand Factor D domains, was the second-most-downregulated transcript by significance and foldchange, while a paralog, apob-2, was also significantly downregulated ( Fig. 1a; Supplementary Fig. 1d, e; Supplementary Data 1a-c and Data 2a, b). Consistent with single cell expression profiling 40 (Supplementary Fig. 1f), both transcripts were highly enriched in intestinal phagocytes, absorptive cells responsible for digestion, nutrient storage, and metabolite secretion ( Fig. 1b and Supplementary Fig. 1f-h). In addition, both transcripts were weakly expressed in a small number of cells outside the intestine likely to be differentiating neoblast progeny ( Fig. 1b and Supplementary Fig. 1f).
ApoB orthologs in vertebrates and insects regulate LP biogenesis and secretion 5,43 . To test whether ApoB protein was secreted, we made a custom antibody recognizing the N-terminus of Smed-ApoB-1 (Fig. 2b, c). ApoB-1 protein was enriched in the intestine, but also throughout the parenchyma (tissues surrounding other organs, sandwiched between the epidermis and intestine, where neoblasts also reside). This indicated that although apob mRNAs were intestine-enriched (Fig. 1a, b), ApoB protein was robustly secreted and transported to peripheral tissues (Fig. 2c). Expression was dramatically reduced in both regions in apob(RNAi) double knockdown planarians, demonstrating that knockdown effectively reduced ApoB protein levels (Fig. 2c).
ApoB orthologs facilitate NL secretion and transport via LPs 5,38 , while ApoB binding by receptors mediate NL uptake and metabolism by target cells 44 . To test whether planarian ApoB functioned similarly, we first evaluated NL distribution in apob(RNAi) animals. In histological sections, NLs (labeled with Oil Red O) were elevated in both the intestine as well as tissues surrounding the intestine, suggesting that both LP secretion by the intestine and uptake/metabolism by peripheral tissues were compromised (Fig. 2d). Using thin layer chromatography, we also found significant elevation of cholesteryl esters and triglycerides in lipid extracts from apob(RNAi) animals (Fig. 2e). These phenotypes predicted that delivery of NLs to neoblasts and/or their progeny would be compromised by ApoB inhibition. To test this, we quantified NL content in neoblasts and their progeny using a fluorescent NL probe, BODIPY-493/503, by flow cytometry. Three major subpopulations of planarian cells can be distinguished by their DNA content and sensitivity to X-irradiation 45,46 (Fig. 2f-h). The "X1" fraction/gate includes >2C DNA content neoblasts in S/G2/M phase of the cell cycle, while "X2" includes 2C neoblasts in G0/G1 and G0 post-mitotic progeny. "Xins," named for the fact that cells in this gate are insensitive to X-irradiation, consists of later stage progeny and mature differentiated cells 47 . We found a dramatic reduction of fluorescence in both the X1 and X2 fractions in apob(RNAi) animals vs. controls (Fig. 2i, j), indicating that a reduction of NL content in neoblasts and their progeny was caused by ApoB inhibition. Together, these data demonstrated that planarian ApoB proteins were produced by the intestine and were likely secreted as LPs to transport NLs from the intestine to stem cells and their differentiating progeny.
apob and lipoprotein receptor genes are upregulated in regenerating fragments. Amputation induces changes in gene expression during planarian regeneration [48][49][50] . We predicted that if apob and other genes involved in neutral lipid metabolism played functional roles during regeneration, they would be up-or downregulated in amputated tissue fragments. Using quantitative PCR, we found that both apob-1 and apob-2 transcripts trended upwards in tissue fragments during earlier stages of regeneration commonly associated with neoblast proliferation (1-3 days) and differentiation (2-7 days) ( Supplementary Fig. 2a, b) 48,51,52 . Significant upregulation of apob-1 and apob-2 was also observed in previously published RNA-Seq data from a 14-day time course of whole-body planarian regeneration ( Supplementary Fig. 2c) 50 . Consistently, accumulation of neutral lipids was sustained in 3and 6-day apob(RNAi) regenerates ( Supplementary Fig. 2d, e), suggesting that LP trafficking from the intestine was disrupted, as in uninjured animals (Fig. 2d).
In addition, we found that three planarian orthologs of human lipoprotein receptors (which bind apolipoproteins, enabling LP uptake/metabolism) were upregulated in the blastema, and regenerating brain and pharynx ( Supplementary Fig. 3a, b, d). These transcripts were expressed in both piwi-1-mRNA-positive neoblasts as well as piwi-1-negative cells that were likely to be differentiating neoblast progeny in the regenerating pharynx and blastema ( Supplementary Fig. 3d). Two lipoprotein receptor genes (ldlr-1 and ldlr-2) were also differentially expressed in the whole-body regeneration RNA-Seq dataset ( Supplementary  Fig. 3c) 50 . Using the Gene Ontology to mine the same dataset 50 , we identified additional planarian genes predicted to regulate lipid transport, triglyceride, and cholesterol metabolism, and found that many of these transcripts were also up-and downregulated during regeneration (Supplementary Fig. 3e and Supplementary Data 3). Together, these observations indicated that apob-1, apob-2, and other genes involved in neutral lipid metabolism were differentially expressed in response to amputation, consistent with roles during regeneration.  apob paralogs are required for polarity re-establishment and organogenesis during regeneration. To assess functional roles of apob paralogs during regeneration (Fig. 3a), we chose uninjured animals after three dsRNA feedings that exhibited "mild" and "severe" phenotypes ( Fig. 2a), amputated their heads, and assessed blastema morphogenesis, which absolutely requires neoblast proliferation and differentiation [53][54][55] . Controls, apob-1(RNAi)-only, and apob-2(RNAi)-only fragments regenerated normally (Fig. 3b), again suggesting apob paralogs likely functioned redundantly. However, apob(RNAi) double knockdown animals had reduced posterior blastemas (head fragments) and anterior blastemas (trunk fragments) at 8 days post-amputation (Fig. 3b). Blastema size was noticeably smaller in fragments from "severe" ("apob-S") animals compared to "mild" ("apob-M") animals ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-31385-2 ( Fig. 3b), suggesting that neoblasts and/or their progeny were more strongly affected.
Expression of so-called "position control genes" to re-establish axial polarity is an essential early event required for whole-body regeneration 56 . We asked whether expression of anterior (notum) 57 or posterior (wnt11-2) 58 was affected in apob(RNAi) regenerates. Indeed, at three days post-amputation (3 dpa), the number of both posterior wnt11-2-expressing cells (Fig. 3c) and anterior notumexpressing cells (Fig. 3d) was reduced in apob-M and apob-S fragments. By 6 dpa, although most apob(RNAi) fragments had regenerated more of these cells, many fragments (especially apob-S) still had fewer than five cells compared to control fragments (Fig. 3c,  d). These data suggested that apob inhibition delayed, but did not completely abolish, re-establishment of these cells.
apob RNAi also affected regeneration of planarian organs, including the brain, pharynx, and intestine. Although both brain and pharynx regenerated, these organs were smaller than in controls ( Supplementary Fig. 2f, g). This suggested that regeneration was delayed, but not blocked, similar to the phenotype for cells expressing polarity cues. Quantitatively, at 6 dpa, both brain and pharynx were smaller in size, especially in apob-S fragments (Fig. 3e, f). To assess intestine, we analyzed both head and tail regenerates, in which a combination of neoblast-driven new cell production and remodeling of pre-existing, differentiated intestinal branches is required for successful regeneration 51 . Newly regenerated posterior branches were significantly shorter in apob(RNAi) head fragments (Fig. 3g). By contrast, in tail fragments, length of the new anterior intestine was not significantly affected (Fig. 3h). Rather, these branches failed to fuse at the anterior midline, leading to a "split" anterior branch phenotype in~50% of apob(RNAi) tail fragments (Fig. 3h). Together, these results suggested that ApoB reduction delayed regeneration of multiple cell types and organs, and, in the case of the intestine, affected differentiation of new cells as well as collective cell migration or other processes required for remodeling.
apob knockdown causes accumulation of early neoblast progeny. Because apob knockdown disrupted multiple neoblastdriven processes during regeneration, we asked whether defects in neoblast maintenance or proliferation were responsible for the effects of apob RNAi on regeneration. First, we assessed neoblast numbers by FISH using piwi-1, a pan-neoblast marker, and tgs-1, a marker for a more pluripotent subpopulation (which also includes neural specialized neoblasts) 50,53,59 . In both uninjured planarians and 7.5-day regenerates, neoblast numbers and distribution were grossly normal ( Supplementary Fig. 4a, b). We also assessed proliferation using anti-phospho-Histone H3 (Ser10) ("PS10") immunolabeling, which marks cells in late G2 and M phase of the cell cycle 60,61 . In uninjured animals and 2-day head regenerates, the number of mitotic neoblasts increased modestly in apob-M, but not apob-S samples (Fig. 4a, b). In 2-day trunk regenerates, there was a more significant (~50%) increase in apob-M fragments, and a modest increase in apob-S fragments (Fig. 4b). Together, these results suggested that ApoB reduction might cause moderate hyperproliferation, or, alternatively, a modest mitotic delay, without dramatically affecting piwi-1+ or tgs-1+ neoblast numbers. These mild phenotypes also raised the possibility that ApoB reduction might preferentially dysregulate differentiation, rather than proliferation or maintenance of actively cycling neoblasts.
We tested these possibilities quantitatively using live cell flow cytometry. We dissociated uninjured and regenerating planarians, and evaluated the proportions of cells in the X1 and X2 fractions (Fig. 4c). In uninjured planarians, as well as 2-day and 7-day regeneration fragments, the relative X1 neoblast cell fraction was unaffected by apob knockdown (Fig. 4d, e and Supplementary  Fig. 4c), consistent with the moderate effect of apob RNAi on neoblast proliferation (Fig. 4a, b and Supplementary Fig. 4a, b). By contrast, in uninjured planarians as well as 2-and 7-day trunk fragments, the proportion of cells in the X2 gate increased significantly by~20-40% ( Fig. 4f and Supplementary Fig. 4c) in apob(RNAi) animals vs. control. In 2 dpa head fragments, there was also a modest, but statistically insignificant, increase (~15%) in the X2 fraction (Fig. 4f). Because the X2 fraction includes both cycling G1-phase neoblasts and differentiating post-mitotic progeny 47,62 , these results suggested that apob inhibition might cause lengthening of G1 phase of the cell cycle, and/or a delay in differentiation of neoblast progeny, either of which could increase the proportion of cells in X2.
In order to distinguish between these possibilities, we examined the X2 fraction in uninjured planarians 24 h after X-irradiation, which preferentially ablates over 95% radiation-sensitive cycling neoblasts, without affecting many early neoblast progeny 47,50,55 . As expected, the X1 fraction was almost completely eliminated in both control and apob(RNAi) samples (Fig. 4g, h). However, although irradiation reduced the proportion of X2 cells bỹ 48-56% in both control and apob(RNAi) animals ( Fig. 4i), the expansion of the X2 fraction persisted in apob-M (~62.5% increase) and apob-S animals (~42.5% increase) relative to controls (Fig. 4i). This increase in radiation-insensitive cells in X2 strongly suggested that the primary defect in apob knockdown animals was a delay in differentiation of neoblast progeny. Fig. 2 apob orthologs are required for viability and neutral lipid metabolism. a Simultaneous RNAi-mediated knockdown of apob-1 and apob-2 caused mild ("apob-M") (yellow) and severe ("apob-S") (orange) depigmentation, and very severe ("apob-VS") (pink) phenotypes including head regression, ventral curling, and lysis. n = 15 animals per condition, representative of >10 experiments. b Simple Western (capillary-based protein analysis) lane view of extracts from control and knockdown planarians labeled with custom anti-ApoB-1. Representative of one experiment with three biological replicates per condition. Uncropped Wes data are in Source Data. c ApoB-1 protein expression (magenta), sagittal sections. Expression inside (dotted outline) and outside the intestine (arrows) was reduced in apob-1(RNAi);apob-2(RNAi) planarians (right panels) (7 days after the last dsRNA feeding). mAb 6G10 (green) labeled visceral and other muscle fibers. Representative of four independent labeling experiments, n = 1 animal per condition (>8 sections per animal). d Neutral lipids (red) accumulated in the intestine (dashed outlines) and parenchyma (arrows) of "mild" apob(RNAi) planarians (7 days after last dsRNA feeding). Oil Red O labeling, sagittal sections. Representative of two independent experiments, n = 1-2 animals per condition (>10 sections per animal). e Cholesteryl esters (CE) and triglycerides (TG), but not diacylglycerides (DG) or monoglycerides (MG) were significantly elevated in "mild" apob(RNAi) animals. Thin layer chromatography quantification. Student's t test (unpaired, two-tailed); **p = 0.0017 (CE) and *p = 0.0303 (TG). Error bars: mean ± S.D., n = 4 biological replicates, representative of two independent experiments. Uncropped TLC plate is provided in Source Data. Intriguingly, the increase in radiation-resistant X2 cells was transient: by 48 h post-irradiation, the proportion of X2 cells in apob(RNAi) samples was only modestly (but not significantly) elevated (Fig. 4i). This suggested that differentiation was delayed, but not arrested, in apob(RNAi) animals. Specifically, at later time points, progeny had differentiated further, and resided in the Xins gate (rather than the X2 gate) as they achieved a more mature cell state. Alternatively, some early progeny might have undergone apoptosis or necrosis at later time points, reducing the apparent accumulation of neoblast progeny in X2.
ApoB reduction preferentially dysregulates expression of transcripts enriched in differentiating neoblast progeny and mature cell types. Our flow cytometry data suggested that the primary phenotype in uninjured and regenerating apob(RNAi)  Fig. 5d and Supplementary Data 5a, b). Enrichment of the subcategories of acylglycerol, fatty acid, steroid, and glycerolipid metabolism was consistent with known roles of ApoB orthologs, and suggested a possible compensatory gene expression response to apob(RNAi)-induced disruption of NL transport. apob(RNAi) also downregulated transcripts in additional metabolism categories, including gluconeogenesis, glycolysis, pyruvate, and nucleotide metabolism (e.g., NADH and ADP), as well as ion transport, indicating wide-ranging dysregulation of metabolite processing and trafficking, especially in apob-S animals (Supplementary Fig. 5e and Supplementary Data 5c, d). Interestingly, apob inhibition dysregulated many additional, non-metabolism-related transcripts. Upregulated functional categories included differentiation and/or specification of multiple tissues including epidermis, nervous system, and the eye, while downregulated categories included cilium morphogenesis and function (possibly reflecting disruption of ciliated epidermal and/or protonephridial cell numbers or physiology), muscle morphogenesis and function, and extracellular matrix organization ( Supplementary Fig. 5d, e and Supplementary Data 5a-d). Importantly, categories related to cell cycle, mitosis, etc. were not enriched among up-or downregulated transcripts. Together, these data suggested that ApoB reduction affected metabolism, cell/tissue differentiation, and functions of mature cell types, but not processes required specifically for cellular proliferation.
Next, to test whether apob disproportionately dysregulated genes expressed by differentiating progeny and mature cells, we crossreferenced our DE transcript list with recently published bulk transcriptomes from flow-sorted X1, X2, and Xins cell fractions ( Fig. 4c) 50 , as well as cells sorted based on their expression of PIWI-1 protein, a widely used marker whose expression marks states within planarian cell lineages (e.g., high PIWI-1 levels in neoblasts, low PIWI-1 levels in some neoblasts and differentiating progeny, and negligible PIWI-1 levels in mature cells) 50,63,64 . We first identified transcripts that were enriched in each cell fraction to generate a "signature" transcript list for each fraction (Supplementary Fig. 6a, c). We then asked what percent of these signature transcripts were dysregulated in apob-M and apob-S animals ( Fig. 5a, b and Supplementary Fig. 6b, d). apob RNAi dysregulated signature transcripts in all six fractions. However, compared to X1 or "PIWI-HI" signature transcripts, 2-4 times as many transcripts were up-or downregulated in X2 and "PIWI-LO" cell classes, which included both neoblasts and early progeny (Fig. 5a, b). Relative to Xins and PIWI-NEG fractions, there were also more dysregulated transcripts represented in the X2 and PIWI-LO fractions in apob RNAi animals ( Fig. 5a, b). To demonstrate the validity of this approach, we cross-referenced transcriptomes from planarians 24 h after irradiation 50 , and found that dysregulated transcripts overlapped mainly with X1 and PIWI-HI signature transcripts, and were primarily downregulated, consistent with neoblast loss ( Supplementary Fig. 6e, f). Together, this analysis demonstrated that apob knockdown preferentially affected transcripts enriched in differentiating progeny, an observation that is consistent with accumulation of cells in this state in flow cytometry experiments ( Fig. 4d-i).
Despite this complexity, we observed a striking and consistent dysregulation pattern in all eight lineages. In three wellcharacterized lineages (epidermis, intestine, and protonephridia) ( Fig. 5c-i), the percentage of dysregulated "state-enriched" transcripts was lowest for neoblast and early transition state subclusters, but higher for progeny and mature cell states. Similar trends were observed for less characterized lineages such as muscle, pharynx, and cathepsin-positive cells (Supplementary Figs. 8c-j and 9c, d). Validating this approach, irradiation 50 primarily dysregulated neoblast and transition state transcripts (Supplementary Fig. 7a-g), and, to a lesser degree for some lineages, progenyenriched transcripts (Supplementary Figs. 8a, k-n, and 9a, e). Similarly, and as expected, X1-and PIWI-HI-enriched transcripts overlapped with neoblast and transition state transcripts, while X2-/PIWI-LO and Xins/PIWI-NEG transcripts were enriched in progressively later cell state subclusters in each lineage (Supplementary Figs. 7h-k, 8b, o-r and 9b, f). Together, these results provided further evidence that apob inhibition specifically dysregulated gene expression in progeny and mature cell types/ states in all eight lineages, and lent additional support to the interpretation that ApoB was required for differentiation of planarian neoblast progeny.
apob knockdown delays differentiation of neoblast progeny. To determine directly whether differentiation of neoblast progeny in apob(RNAi) animals was delayed, we generated an anti-PIWI-1 antibody to quantify progeny. As described above, PIWI-1 protein is highly expressed by neoblasts, but perdures at lower levels in early neoblast progeny, before becoming undetectable in late progeny and fully differentiated cells 50,63 (Figs. 5b and 6i). Although PIWI-1 perdurance has not been rigorously analyzed in individual planarian lineages, our bioinformatic analysis suggested that signature PIWI-LO transcripts were highly enriched in early progeny of the well-characterized epidermal lineage ( Supplementary Fig. 7i). Accordingly, we analyzed PIWI-1 levels in cells expressing prog-1 mRNA, a widely used marker of early epidermal progeny, one of the largest progeny subpopulations in planarians 40,47,55,65 (Fig. 6a, b). We predicted that ApoB inhibition might lead to greater numbers of prog-1+ cells, and/or a higher proportion of prog-1+ cells expressing PIWI-1. The relative proportion of prog-1+ cells was similar between apob(RNAi) and controls (Fig. 6c), suggesting that lineage specification from pluripotent neoblasts to epidermal lineage-primed progenitor cells to very early progeny was unaffected by ApoB reduction. However, when we counted cells that were both prog-1+ and PIWI-LO, we saw a statistically significant increase in this cell fraction in apob-M animals, and a modest (but insignificant) increase in apob-S animals compared to controls (Fig. 6d). There was also a moderate upward trend in the proportion of prog-1+ and PIWI-HI cells (Fig. 6e). In addition, we observed reductions of the prog-1+ and PIWI-NEG cell fractions in apob(RNAi) animals (significant for apob-M animals) (Fig. 6f), indicating a slower rate of PIWI-1 protein degradation, and further supporting the idea that differentiation of epidermal cells was delayed. We reasoned that slowing early progeny differentiation might also cause an eventual reduction in mature cell types. To test this, we quantified the density of mature cells located in the anterior ventral epidermis of apob(RNAi) animals (Fig. 6g). Although the number of these cells was unaffected in apob-M animals, there was a significant reduction of epidermal cell density in apob-S animals (Fig. 6g). Together, these results directly demonstrated that apob knockdown caused a delay in epidermal progeny differentiation, which eventually led to reduced production of mature epidermal cells as the apob RNAi phenotype became more severe.
As mentioned above, the correlation of PIWI-1 protein levels with specific states in other planarian lineages has not yet been rigorously characterized. Therefore, instead of analyzing additional lineages, which would require identification of markers that overlap with PIWI-1 in each lineage, we examined cell states by flow cytometry, using low PIWI-1 protein levels as a marker of early progeny, enabling their global quantification ( Fig. 6h-k and Supplementary Fig. 10a, b). We detected significant increases in the PIWI-LO cell fraction in both apob-M and apob-S planarians, compared to controls (Fig. 6h, j). This result suggested that ApoB reduction resulted in longer PIWI-1 protein perdurance in neoblast progeny, and provided further direct evidence of a global, multi-lineage delay in early differentiation. Unexpectedly, we also found statistically significant decreases (~15-20%) in the fraction of PIWI-HI cells in apob(RNAi) animals (Fig. 6k). Further analysis of the PIWI-HI subfraction revealed that the percentage of cells with <4C DNA content (i.e., cells in G0/G1 and S) was increased in apob(RNAi) animals, while the G2/M fraction was decreased ( Supplementary  Fig. 10c-f). We observed similar shifting in S-phase and G2/Mphase cell percentages in the X1 (S/G2/M) fraction analyzed earlier (Fig. 4e, Supplementary Fig. 10g-i), despite no detectable overall change in the percentage of cells in this fraction in apob(RNAi) animals (Fig. 4e). The increase in PIWI-HI G0/ G1 cells (Supplementary Fig. 10d) could reflect earlier dysregulation of PIWI-1 protein reduction (i.e., from high to low levels in very early, recently specified progeny), consistent with the upward trend in prog-1+, PIWI-HI cells we observed (Fig. 6e). Additionally, the increase in PIWI-HI G0/G1 and S fractions relative to the G2/M fraction ( Supplementary  Fig. 10d-f) could indicate a delay in cell cycle progression. This modest reduction in the PIWI-HI cell population could contribute to the delay in apob(RNAi) regeneration. However, ApoB inhibition did not dramatically affect the number or location of piwi-1+ or tgs-1+ neoblasts (Fig. 4a, b) or their gene expression (Fig. 5). Furthermore, cell cycle delays or reduced proliferating neoblasts would be expected to cause a decrease in the proportion of PIWI-LO cells, not the increase we observed. Therefore, we concluded that delay in neoblast progeny Fig. 4 apob inhibition causes accumulation of neoblast progeny. a, b PhosphoHistone-H3-S10-positive (pH3-PS10) cells (white/gray) were modestly elevated in apob-M ("mild"), but not apob-S ("severe"), uninjured animals (a) and 2-day regenerates (b). One-way ANOVA with Dunnett's T3 multiple comparisons test. Error bars: mean ± S.D. a n = 17 (control), 23

Discussion
In this study, we discover a non-cell autonomous role for ApoB proteins in the regulation of planarian neoblast progeny differentiation. Our data support a working model in which ApoB transports NLs from the intestine to both neoblasts and their differentiating progeny via LPs (Fig. 7). In the absence of ApoB, neoblast proliferation and maintenance are not substantially impaired, but differentiation is slowed, causing an accumulation of neoblast progeny, compromising production of mature cells, and delaying homeostatic cell renewal and regeneration (Fig. 7).
We suggest that ApoB inhibition disrupts the progression of differentiation after cell fate specification, and not specification itself. In planarians, specification likely begins during S-phase,  because expression of fate-specific transcription factors is significantly higher in S/G2/M neoblasts than in G1 neoblasts 59,66 . Intriguingly, inhibition of other planarian genes required for differentiation (e.g., the transcription factor mex3-1, the extracellular matrix component collagen4-1, and the transcriptional co-activating protein cbp-3) also cause increases in neoblast numbers in vivo 55,67-69 . Furthermore, knockdown of exocyst component 3 (exoc3), a negative regulator of pluripotency whose mammalian homolog Tnfaip3 promotes embryonic stem cell differentiation, causes expansion of the S/G2/M (X1) neoblast fraction 70 . We speculate that these genes may be required for cell fate specification, and that their inhibition shifts neoblast dynamics in favor of renewal divisions that expand the stem cell compartment. By contrast, apob RNAi does not cause accumulation of G2/M neoblasts, but rather increases post-mitotic progeny number, and causes greater dysregulation of transcripts associated with differentiation and mature cell states. Therefore, ApoB is likely required to drive post-specification stages of differentiation, and inhibition delays late commitment steps (i.e., after cell cycle exit), and/or transitions to final mature states (Fig. 7).
Differentiation requires extensive changes in gene expression that are often preceded by genome-wide chromatin remodeling 71,72 . Intriguingly, FA oxidation is a significant source of carbon for acetyl-CoA production and histone acetylation 73 . Consistent with a role for NLs in planarian differentiation, knockdown of exoc3 reduces triglyceride levels, causes expansion of the neoblast population, inhibits organogenesis, and reduces expression of differentiation markers; palmitic acid supplementation rescues these differentiation-associated phenotypes 70 . Furthermore, in the sexually reproducing S. mediterranea planarian biotype, inhibition of nuclear hormone receptor 1 causes NL accumulation and blocks differentiation of gonads and accessory reproductive tissues, a phenotype that is rescued by supplementation with either acetyl-CoA or Acyl-CoA synthetase 74 . Because acetyl-CoA can also enter the citric acid cycle to produce alpha-ketoglutarate, a substrate for histone demethylation 75 , ApoB inhibition could dysregulate epigenetic changes through multiple pathways. Histone acetylases, deacetylases, methyltransferases, and demethylases are conserved in planarians, and their inhibition disrupts stem cell maintenance, differentiation, and regeneration 68,69,[76][77][78][79] . Because apob RNAi results in widespread dysregulation of thousands of transcripts associated with differentiating progeny, it is reasonable to speculate that in planarians, intestinal lipid stores serve as a ready carbon source that is trafficked by ApoB-containing LPs to neoblasts and progeny to support epigenetic modifications required for differentiation.
ApoB depletion may also delay differentiation by other mechanisms. For example, NL-derived fatty acids may be utilized to produce ATP via beta-oxidation and the mitochondrial electron transport chain (ETC) to support energy-dependent processes during differentiation. Consistent with this idea, planarian mitochondrial mass is higher in differentiating progeny, and pharmacological inhibition of the ETC promotes pluripotency and neoblast colony expansion, which may also limit differentiation 80 . In mammals, disrupting the ETC blocks differentiation of cardiomyocytes and mesenchymal stem cells 81,82 , but the ETC is dispensable for differentiation of mammalian epidermal progenitor cells and Drosophila ovarian stem cells 83,84 . Further study will be needed to determine whether LPtransported NLs serve as a significant energy source during planarian differentiation. Additionally, LP-mediated transport of morphogens like Hedgehog or Wnt proteins 85 , whose planarian orthologs play important roles in regulating axial polarity and tissue differentiation 56 , may be affected by apob knockdown. However, we find that ApoB inhibition delays, but does not block or alter regeneration of axial polarity, and we find little evidence of dysregulation of polarity-related transcripts ( Supplementary  Data 4 and 5), suggesting that planarian LPs may not play a major role in planarian morphogen trafficking. Similarly, fat-soluble vitamins known to influence stem cell dynamics are also transported in LPs [86][87][88][89][90][91] . Although characterization of LP cargo may yield additional insights, the dramatic dysregulation of lipid metabolism at the gene expression level, and the minimal disruption of vitamin-related gene expression in apob(RNAi) animals (Supplementary Data 4 and 5), suggest that LP-mediated vitamin transport may not play a significant role in planarian differentiation. Lastly, we find that apob inhibition dysregulated genes associated with muscle differentiation and function (Supplementary Data 5). In planarians, muscle cells not only secrete axial polarity cues, but also serve a fibroblast-like role by secreting most components of the extracellular matrix, whose functions are required to both spatially restrict the stem cell compartment, and modulate proliferation and differentiation 67,92,93 . apob RNAi causes moderate downregulation of most fibrillar collagens, as well as the basement membrane collagen4-1, which promotes differentiation 67 ( Supplementary Data 4 and 5). Thus, ApoB depletion may also delay differentiation indirectly, by compromising the generation and/or function of muscle cells.
The effect of apoB knockdown on regeneration suggests several future directions. First, although the existence of prominent LDs in planarian neoblasts has been known for decades 94 , the roles of this intriguing organelle have not been investigated. We did not assess neoblast LD numbers or size, but apoB knockdown dramatically reduces NL content in both X1 (S/G2/M) and X2 (G0/ G1) neoblast fractions, suggesting that ApoB and LPs may influence LD content and/or function in both neoblasts and progeny. In addition, NL content is lower in G0/G1 cells, suggesting that a primary role of LDs may be to support differentiation. Once thought of as static storage particles, recent work has demonstrated that LDs in animal cells are dynamic, multifunctional organelles that regulate nutrient sensing, cell stress responses, and even intracellular localization of histones, transcription factors, and other proteins 2,10 . Additionally, although emerging data suggest LDs as a potential therapeutic target in cancer stem cells 95,96 , our knowledge of the regulation and functions of LDs in stem cells, especially during regeneration, is limited. Studies in planarians and other regeneration models could further illuminate the roles of this organelle. Second, coordinated metabolic shifts between glycolysis and oxidative phosphorylation may be a widespread aspect of stem cell transitions between quiescence, proliferation, and differentiation, and extrinsic lipids can influence these states [11][12][13][14] . Our RNA-Seq data suggest that transcription of regulators of amino acid metabolism, glycolysis, the tricarboxylic acid cycle, and other metabolic pathways respond to apob(RNAi), suggesting that planarian stem cell metabolism is just as dynamic as in other animals. The fact that apob(RNAi) seems to primarily affect post-mitotic states also suggests that planarian neoblasts might rely primarily on glycolysis for energy and metabolite supply, and shift to lipid metabolism and oxidative phosphorylation during differentiation, as in other systems [97][98][99] . Studies in animals with high regenerative capacity could generate greater insights into whether and how injury can induce metabolic switching. Third, we find that expression of apob-1 and apob-2, as well as numerous other regulators of NL metabolism are dynamically up-and downregulated at the transcript level during regeneration (Supplementary Fig. 3e). This suggests that coordination of lipid metabolism is part of a genome-encoded program of regenerative gene expression. Furthermore, some lipid regulators are up-and downregulated at later regeneration time points ( Supplementary  Fig. 3e). This raises the intriguing possibility that planarians may replenish NLs utilized during regeneration ( Supplementary  Fig. 2d, e), or adapt metabolic networks to reduced lipid levels, especially in the absence of post-regeneration feeding. Identification of additional lipid regulators required at different stages of regeneration, and unraveling which transcription factors, chromatin modifiers, and other factors control their expression are thus additional priorities for future study.
Finally, because apob-1 and apob-2 are downregulated by inhibition of nkx2.2, an intestine-enriched transcription factor also required for regeneration 39 , our results provide a specific example of how the planarian intestine can non-autonomously influence neoblast dynamics. Intriguingly, unlike nkx2.2 RNAi, apob knockdown does not reduce the abundance of phosphoHistone-H3-S10-positive neoblasts, indicating that the proliferative defect in nkx2.2(RNAi) animals is not caused by apob reduction, and additional downstream regulators of proliferation remain to be discovered. Intriguingly, dozens of additional regulators of lipid metabolism and transport of other metabolites are downregulated in nkx2.2(RNAi) animals (Supplementary Data 1b), suggesting additional ways in which the intestine could influence neoblasts and their niche.
In summary, we have identified apolipoprotein B orthologs and neutral lipid metabolism as important regulators of stem cell progeny differentiation and regenerative tissue growth. Since the discovery of lipoproteins a century ago, their roles in lipid transport and disease have been extensively investigated 100 , but functions in stem cell regulation are not nearly as well characterized [101][102][103][104] . Efforts to define functions of LPs and their metabolic derivatives more precisely in planarians and other models will therefore improve our understanding of metabolic requirements of stem cell-driven regeneration. In addition, because lipid metabolism is amenable to pharmacological manipulation 105 , further study may provide new insights relevant to the dual goals of promoting repair of damaged human tissues, and inhibiting growth in pathological contexts like cancer.

Methods
Ethics statement. Anti-ApoB-1 and anti-PIWI-1 antibodies were generated by GenScript USA (Piscataway, NJ), an OLAW, AAALAC, and PHS-approved vendor. GenScript's animal welfare protocols were approved by OMRF IACUC . No other vertebrate organisms were used in this study.
Domain organization and phylogenetic analysis. For ApoB-1, ApoB-2, Ldlr-1, Ldlr-2, and Vldrl-1, protein domains were identified using "HMMSCAN [https:// www.ebi.ac.uk/Tools/hmmer/search/hmmscan]" to search Pfam, TIGRFAM, and Superfamily databases, with Phobius for transmembrane and signal peptide predictions (conditional E-value cutoff of 1e-03) 113 . For the ApoB phylogenetic tree, N-terminal Vitellogenin domains from ApoB and related proteins were aligned in Geneious using MAAFT (default settings), and alignment was manually trimmed to the N-and C-terminal boundaries of human Apo B-100. Phylogenetic tree was generated using "PhyML 3.0 [http://www.atgc-montpellier.fr/phyml/]" 114 , using AIC for automatic selection of the LG substitution model, BioNJ starting tree, NNI for tree topology improvement, and 100 bootstrap replicates. Accession numbers for proteins used in domain diagrams and phylogenetic analysis are included in Supplementary Data 6.
Cryosections were generated after FISH (Fig. 1b) or before FISH (Fig. 6a-f) as previously described 117 . On-slide FISH was similar to whole mount FISH, with the following modifications: NAc treatment was 7 min; after sectioning, all steps were performed in polypropylene LockMailers (Ted Pella 21096); and slides were mounted in Fluoromount G (Southern Biotech 0100-01).
RNAi. dsRNA synthesis (Thermo Fisher Scientific, MEGAscript T7 Transcription Kit, AM1333) and RNAi experiments were conducted as described 41,110 by mixing in vitro-synthesized dsRNA with 4-9 μl of 1:10 food coloring (Durkee):water mix, and 40 μl of 2:1 liver:water homogenate. For nkx2.2 RNAi, animals were fed only once with 1 μg egfp control or nkx2.2 dsRNA; RNA was extracted for RNA-Seq after seven days. For apob RNAi, 2-6 μg control egfp dsRNA or 1-3 μg each apob-1 and apob-2 (for simultaneous RNAi) were mixed with liver and food coloring. Animals were fed five times for initial viability experiments, and 3-5 times for most other experiments. Animals with "mild" (apob-M) and "severe" (apob-S) phenotypes were separated seven days after the last dsRNA feeding prior to fixation, amputation, or flow cytometry of control and apob(RNAi) samples. Non-eating planarians were always removed from the experiment if they refused a second dsRNA feeding one day later.
Anti-ApoB-1 and anti-PIWI-1 antibody generation. Antigen sequences were supplied to GenScript USA (Piscataway, NJ) for antibody production. For ApoB-1, sequence corresponding to the N-terminal Vitellogenin domain (dd_636, nucleotides 45-1919, amino acids 1-625) was utilized to synthesize expression constructs for fusion protein expression and purification. For PIWI-1, the peptide CVRPKEKTENEPEGP (dd_659, nucleotides 1010-1051, amino acids 302-315, overlapping with a previously used peptide 63 ) was used as the antigen. GenScript immunized New Zealand rabbits, affinity purified the antibody, and assessed antibody titer and immunogenicity with both ELISA and Western blot. The predicted N-terminus of ApoB-1 is only 22% identical and 42% similar to ApoB-2. Furthermore, ApoB-1's mass is predicted to be 235 kD less than ApoB-2. Anti-ApoB-1 labeled only a single prominent band by Western analysis (Fig. 2b and Source Data), suggesting the polyclonal antibody is specific for ApoB-1. Similarly, most cells expressing high levels of PIWI-1 were eliminated 24 h after irradiation ( Supplementary Fig. 10a, b), indicating specificity of this antibody for neoblasts and early progeny. Limited quantities of custom antibodies are available upon reasonable request.
Epidermal nuclei labeling. Uninjured planarians were fixed as for FISH (above), then labeled in 10 μg/ml DAPI in PBSTx for four hours at RT. After 4 PBSTx washes, animals were mounted in Fluoromount G. plus 4 ml ultrapure water) for 15 min at RT. Slides were quickly dipped 3-5X in 60% isopropanol to remove excess dye, then rinsed for 1 min in 60% isopropanol, then rinsed 1 min in DI water. Slides were then rinsed in PBS-Tween-20 (0.01%, to prevent drying), then mounted in 90% glycerol/1X PBS, and imaged within 2-3 days.
RNA extraction, library preparation, and RNA sequencing. Uninjured planarians (5-6 animals per biological replicate) were homogenized in Trizol (Thermo Fisher Scientific 15596026) using a motorized Kontes pestle grinder, and RNA was extracted using two chloroform extractions and high-salt precipitation buffer according to the manufacturer's instructions. After precipitation, solutions were transferred to Zymo RNA columns (Zymo RNA Clean & Concentrator 5 kit, R1013) for DNAse treatment and purification, according to manufacturer's instructions. RNA samples were analyzed using Agilent RNA ScreenTape (5067-5576) on an Agilent TapeStation 2200 according to manufacturer's protocol.
For analysis of gene expression in control vs. nkx2.2(RNAi) (three biological replicates each), mRNA was enriched using oligo-dT homopolymer beads, and libraries were generated using the Illumina Truseq Stranded mRNA Library Prep Kit according to the manufacturer's protocol. Final libraries were assayed on the Agilent TapeStation for appropriate size and quantity. Libraries were pooled in equimolar amounts as ascertained by fluorometric analysis, then final pools were absolutely quantified using qPCR on a Roche LightCycler 480 with Kapa Biosystems Illumina Library Quantification Reagents. Paired-end (2 × 150 bp) sequence was generated on an Illumina NovaSeq 6000 instrument. 28 M-43 M reads were generated for each of three biological replicates per condition. For analysis of gene expression in control, apob-M, and apob-S animals, total RNA for six (control) or four (apob-M and apob-S) biological replicates per condition was submitted to GENEWIZ (South Plainfield, NJ) for library generation using NEB NEXT ULTRA library prep (New England Biolabs) and RNA sequencing with standard Illumina adapters. Paired-end (2 × 150 bp) sequence was generated on an Illumina HiSeq 4000 instrument; 19 M-26 M reads were generated for each replicate.
First, all transcripts with counts per million (CPM) < 1 in three samples (nxk2.2(RNAi), Zeng X1/X2/Xins data, and Zeng 24 h irradiation data) or four samples (all others) (e.g., lowly expressed transcripts) were excluded from further analysis. Next, after recalculation of library sizes, samples were normalized using trimmed mean of M-values (TMM) method, followed by calculation of common, trended, and tagwise dispersions. Finally, differentially expressed transcripts were identified using the pairwise exact test (nkx2.2(RNAi) and apob(RNAi) experiments, Zeng 24 h irradiation data) or the generalized linear model (GLM) likelihood ratio test (other Zeng datasets). Expression changes were considered to be significant if the false discovery rate-adjusted p value ("FDR") was <0.05. dd_Smed_v6 expression analyses for Zeng data sets are provided in Supplementary Data 7.
Gene ontology analysis. GO analysis was conducted using BiNGO 125 (v3.0.4, in Cytoscape 3.8.0) using a custom S. mediterranea GO annotation as previously described 110 . RefSeq protein collections used for BLASTX and UniProtKB Biological Process GO terms used for annotation were downloaded for each organism in April 2020.
Hierarchical clustering and heat maps. Hierarchical clustering of transcripts annotated with lipid-metabolism-related GO terms was conducted using EdgeRgenerated log 2 FC values in Cluster 3.0 126 , with Euclidean distance and complete linkage. Heat maps were generated with Java Treeview 127 .
qRT-PCR. Total RNA was extracted from biological triplicates (5-10 fragments per replicate) using Trizol as for RNA-Seq samples. 1 μg total RNA was reverse transcribed using the iScript cDNA Synthesis kit (BioRad 1708890). apob-1 and apob-2 levels were detected using the Fast Start Essential Green DNA master mix (Roche 06924204001) on a Roche LightCycler 96 instrument (SW 1.1 and Light-Cycler 96 Application Version 1.1.0.1320). RNA levels were normalized to the geometric mean of endogenous controls ef-2 and gapdh using the Livak ΔΔCt method 128 . Primer sequences are provided in Supplementary Data 6.
For fixed cell flow cytometry, 4-6 planarians per biological replicate were dissociated as above. Cells were fixed in 2% formaldehyde (EM grade) for 10 min at room temperature (RT), then rinsed twice with 1X PBS. Cells were incubated with permeabilization and blocking buffer (perm/block: 1X PBS, 10% horse serum, 2 mM EDTA (Sigma E5134), 2 mM NaN 3 , 0.3% Triton X100) for 30 min. Each sample was split in half to place into 2 wells of a 96-well V bottom plate (~1-2 million cells per well) (ThermoFisher Scientific, 249662) for incubation with anti-PIWI-1 or normal rabbit IgG (Jackson ImmunoResearch, 011-000-003) at 0.2 µg/ ml in perm/block (O/N, 4°C). Samples were washed twice in perm/block before incubation with goat anti-rabbit-Alexa 488 (Jackson ImmunoResearch 111-545-144) at 1:8000 for 30 min at RT. Cells were washed once with perm/block before incubation with DAPI in perm/block at 3 µg/ml for 10 min at RT. Split samples were then resuspended in half of the volume of 1X PBS with 0.5% BSA, filtered through a 40 µm filter (Falcon 352235), and recombined into flow tubes before flow cytometry as above. After gating for singlet and DAPI positive cells, PIWI-LO and PIWI-HI cell fractions were distinguished against side scatter signal ( Supplementary Fig. 11d-f).
For comparison of dysregulated transcripts in apob-M and apob-S animals with single cell type/state data in ref. 40 , we again used the "merge" function in RStudio to identify the number of transcripts enriched in individual lineage subclusters (Supplementary Table S2) 40 that were also dysregulated by apob RNAi. "% of dysregulated transcripts" was calculated as the number of overlapping transcripts divided by the total number of transcripts in each individual subcluster. In Fig. 5, the "N/TS" (Neoblast/Transition State) designation included subclusters with high piwi-1 mRNA expression thought to be neoblast/progenitor subpopulations in epidermal, intestine, and protonephridia lineages based on the conclusions of Fincher et al. and other published data 45,66,[129][130][131] . Similarly, the "P" (Progeny) and "M" (Mature) designations were based on conclusions from both single cell RNA-Seq data and previous work. For lineages that are less well understood in vivo ( Supplementary Figs. 8 and 9), we designated subclusters/states using both transcript dysregulation in 24 h irradiated animals 50 and piwi-1 mRNA levels in t-SNE plots 40 . "N/TS" subclusters possessed the greatest number of irradiationdysregulated transcripts and the highest piwi-1 expression; "P" subclusters possessed fewer (by proportion) irradiation-dysregulated transcripts and lower piwi-1 expression; and "M" subclusters had the fewest radiation-sensitive transcripts and negligible piwi-1 expression.
Image collection and quantification. Epifluorescent images (FISH samples and cryosections) were collected on a Zeiss AxioObserver.Z1 with Excelitas X-Cite 120 LED Boost illumination and Zen 2.3 Blue (v2.3.64.0). For quantification of pharynx and brain size, organ area and animal area were measured in ImageJ 132 and organto-body size ratios were calculated. For intestine, length of anterior branch and posterior branches were measured in ImageJ. Means of posterior branch length were calculated, and then posterior-to-anterior (head fragments) or anterior-toposterior (tail fragments) length ratio was calculated. For tail fragments with split anterior branch, anterior branch length was measured from the anterior of the pharynx to the tip of the anterior-most primary branch. For tail fragments, the split anterior branch phenotype (failure to fuse at the midline) was scored if there was an obvious gap between anterior branches for at least half the length of the anterior branch. For images of apob-1 and apob-2 FISH on cryosections, z-stacks were collected with an Apotome.2 for generation of maximum orthogonal projections.
For anti-pH3-PS10-labeled samples, z-stacks were collected on a Zeiss AxioObserver.Z1 at 10× magnification, followed by tile stitching, extended depth of focus projection, and background subtraction (PS10 channel only) with a radius of 30. Control and experimental samples to be compared were imaged at identical exposures. For quantification, animal area and PS10+ nuclei were quantified using the Automated Segmentation tools in Zen 2.3 Blue. Briefly, animal area was measured with Gaussian smoothing, no background subtraction or sharpening, Morphology separation, and custom threshold settings that were the same for all samples to be directly compared. PS10+ nuclei number was measured using Lowpass smoothing, Rolling Ball background subtraction, Delineate sharpening, Watersheds separation, with threshold settings and other parameters that were identical for all samples to be directly compared. Number of mitoses per area were calculated for each animal/fragment. Confocal images were collected on a Zeiss LSM 710 (ZEN version 11.0.3.190, 2012-SP2) or LSM 880 (Zen Black version 2.3 SP1 FP3) laser scanning microscope with 10× Plan NeoFluar, 20X Plan Apo, or 40X C-Apo objectives. For orthogonal projections (anti-ApoB1 immunolabeling and ldlr/vldlr FISH), between three and ten z-planes were collected at 1-2X "optimal" section thickness (based on objective NA). For notum and wnt11-2 FISH, full fragment thickness stacks were projected to ensure that all mRNA-positive cells were counted.
Images of Oil-Red-O-stained sections, live animals, live regenerates, and WISH samples were collected on a Zeiss Stemi 508 with an Axiocam 105 color camera (ZEN 2.3 Blue), or a Zeiss Axio Zoom.V16 with an Axiocam 105 camera (ZEN 2012 Blue). In some cases, brightness and/or contrast were adjusted in Adobe Photoshop to improve signal contrast for Figure panels. Original fluorescent intensity was always used for automated quantification (below).
For automated analysis of prog-1 FISH and anti-PIWI-1 colabeling on sections, z-stacks (5 slices at 1 µm interval) were collected on a Zeiss AxioObserver.Z1 with Apotome.2, using a 20× objective, N.A. = 0.8, followed by tile stitching and extended depth of focus projection. Control and experimental samples were imaged at identical exposures. For automated quantification in FIJI (version 2.3.0), DAPI and prog-1 channels were first merged, then Auto Local Threshold (Phansalkar algorithm) was used to generate a cell region mask, followed by Dilation and Watershedding to segment regions. Analyze Particles was used to limit to 4-40 µm diameter regions, and the resulting mask was used to quantify DAPI, prog-1, and PIWI-1 mean fluorescence intensity per region. Cutoff values for DAPI and prog-1 were set using unlabeled samples to exclude false positives due to background noise. Low and high cutoff values for PIWI-1 signal were set to distinguish PIWI-HI, PIWI-LO, and PIWI-NEG cells, using cell proportions from flow cytometry of anti-PIWI-1-labeled cells as a guide. Data generated by FIJI were compiled in R (version 4.1.2).
For ventral epidermal nuclei counting, single epidermal planes were imaged with a Zeiss ApoTome.2 using a 20× objective, N.A. = 0.8. Four to seven regions (50-100 µm 2 ) in the anterior third of the prepharyngeal region were delineated in FIJI, avoiding regions with gaps caused by N-Acetyl-L-Cysteine treatment. Nuclei were counted using the "FindMaxima" function with Prominence > 300. Nuclear density was calculated per animal by adding total number of nuclei and dividing by total area.
Statistics. Detailed data and information regarding statistical testing, n values, and replicate information are included in Source Data, Reporting Summary, and Figure Legends. For experiments with statistical analysis, n values are indicated exactly by the number of data points in figure plots, along with definitions of error bars and p or q values in Legends. All tests were conducted in Prism 9 (GraphPad Software, San Diego, CA). For one-way ANOVA, ordinary ANOVA was performed unless Brown-Forsythe and Bartlett's tests indicated standard deviations were significantly different, then Brown-Forsythe and Welch ANOVA tests were performed. For two-way ANOVA (flow cytometry experiments with irradiation), q values (FDR-adjusted p values) were reported when interaction between RNAi condition ("Genotype") and irradiation was significant (X2 subpopulation in Fig. 4i). Otherwise, p values were derived using one-way ANOVA followed by Tukey's or Dunnett's multiple comparisons test. p values of <0.05 (*), <0.01 (**), <0.001 (***), and <0.0001 (****) were annotated with asterisks in figures.
For statistical testing of overlap between genes dysregulated in apob(RNAi) planarians and other RNA-Seq data sets, the R Package GeneOverlap 133 was used to conduct Fisher's exact test on each comparison. Total number of detected transcripts ("genome size" in GeneOverlap v1.20.0) was determined conservatively by only including transcripts detected in both apob RNA-Seq and bulk 50 or single cell 40 sequencing data. For sc-RNA-Seq, the digital expression matrix in GEO "GSE111764" was normalized in Seurat as in Fincher et al. 40 ; only transcripts with non-zero expression in 0.5% of cells (Fincher TableS1) in each subcluster were considered to be detected. p values <0.05 are indicated with a caret (^) in figures, and provided individually in Source Data.
Replicate and statistical information for RNA-Seq and other bioinformatics experiments are detailed in appropriate Methods sections.